Citation
Hydrogen/deuterium fractionation factors of the aqueous ligand of cobalt in Co(H₂O)₆² and Co(II)-substituted carbonic anhydrase

Material Information

Title:
Hydrogen/deuterium fractionation factors of the aqueous ligand of cobalt in Co(H₂O)₆² and Co(II)-substituted carbonic anhydrase
Creator:
Kassebaum, James Web, 1962-
Publication Date:
Language:
English

Subjects

Subjects / Keywords:
Cobalt ( jstor )
Deuterium ( jstor )
Eggshells ( jstor )
Fractionation ( jstor )
Isotope effects ( jstor )
Ligands ( jstor )
pH ( jstor )
Protons ( jstor )
Solvents ( jstor )
Water tables ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright James Web Kassebaum. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
23090432 ( ALEPH )
25072742 ( OCLC )

Downloads

This item has the following downloads:


Full Text









HYDROGEN / DEUTERIUM FRACTIONATION FACTORS OF THE AQUEOUS LIGAND OF COBALT IN Co(H20)62+ AND
Co(II)-SUBSTITUTED CARBONIC ANHYDRASE









By

JAMES WEB KASSEBAUM









A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1988













ACKNOWLEDGEMENTS
I wish to thank Dr. David Silverman for his thoughtfulness and guidance shown to me during the work presented in this dissertation. I have benefitted greatly from his scientific insight into experimental design and his careful attention to excellence in scientific writing. I also wish to thank very much Dr. C. K. Tu and Mr. George Wynns for their help in experimental procedures and for many helpful discussions. I acknowledge the very valuable advice of the members of my committee, Dr. E. Raymond Andrew, Dr. Robert Cohen, Dr. Russell Drago, and Dr. Daniel Purich. Finally, I especially thank Dr. Seymour Koenig at the IBM T. J. Watson Research Center in Yorktown Heights, New York, for allowing me to visit his laboratory and use the field-cycling relaxometer which added greatly to the scope of this work.






















ii













TABLE OF CONTENTS



ACKNOWLEDGEMENTS .......................................................................... ii

ABSTRACT .................................................................................................... v

CHAPTERS

1 INTRODUCTION .................................................................. 1

2 WATER PROTON NMR RELAXATION ENHANCEMENT IN
SOLUTIONS OF Co(H20)62+ AND Ni(H20)62+

Introduction...................................................
Introduction ......................................................................................... 9
Experimental Procedure. ............................................................. 10
R esults................................................................................................. 11
Discussion............................................... ..................................... 16

3 HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF
THE AQUEOUS LIGAND IN Co(H20)62+ AND Ni(H20)62+
20

Introduction...................................................................................... 20
Experimental Procedure. ............................................................. 22
Results ............................................................................................... .. 25
Discussion ..................................................................................... ..... 32

4 HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF
THE AQUEOUS LIGAND OF COBALT IN Co(II)SUBSTITUTED CARBONIC ANHYDRASE ............................. 39

Introduction ............. ........................ 39
Experimental Procedure. ............................................................. 40
Results........................................................ .................................. 42
D iscussion ........................................................................................... 48

5 INTERPRETATION OF SOLVENT HYDROGEN ISOTOPE
EFFECTS ASSOCIATED WITH CARBONIC ANHYDRASE...... 52

Introduction ............................................. ..... ............................ 52


iii








Experimental Procedure........................... ..... .............. 53
R esults ................................................................................................. 5 5
Discussion................................................................................... 58

6 MAGNETIC FIELD AND pH DEPENDENCE OF THE WATER
PROTON NMR RELAXATION ENHANCEMENT IN
SOLUTIONS OF Co(II)-SUBSTITUTED CARBONIC
ANHYDRASE III.............................................................................. 65

Introduction................................................ 65
Experimental Procedure................................ .............. 66
R esults......................................................... ....... .............................. 67
Discussion................................................ ..................................... 67

7 CONCLUSIONS ............................................................................. 73

REFERENCES ............................................................................................... 79

BIOGRAPHICAL SKETCH ............................................................... 83































iv













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

HYDROGEN / DEUTERIUM FRACTIONATION FACTORS OF
THE AQUEOUS LIGAND OF COBALT IN Co(H20)62+ AND
Co(II)-SUBSTITUTED CARBONIC ANHYDRASE By

James Web Kassebaum

December, 1988

Chairman: Dr. David N. Silverman
Major Department: Biochemistry and Molecular Biology
I have measured the hydrogen / deuterium fractionation factor for the

rapidly exchanging aqueous ligands of cobalt in Co(H20)62+ and in three Co(ll)substituted isozymes of carbonic anhydrase. The fractionation factor was determined from NMR relaxation rates at 300 MHz of the protons of water in mixed solutions of H20 and D20 containing these complexes. In each case, the paramagnetic contribution to 1/T2 was greater than to 1 /T1, consistent with a chemical shift mechanism affecting 1/T2. The fractionation factors obtained from T1 for Co(H20)62+ and for the isozymes of Co(ll)-substituted carbonic anhydrase were close to the fractionation factor for bulk water, which is unity. The fractionation factors obtained from T2 were 0.73 0.02 for Co(H20)62+, 0.72 +
0.02 for Co(ll)-substituted carbonic anhydrase 1, 0.77 0.01 for Co(ll)substituted carbonic anhydrase II, and 1.00 0.07 for Co(ll)-substituted carbonic anhydrase III. I concluded that fractionation factors in these cases determined from T1 and T2 measured isotope preferences for different populations of ligand sites. I suggest that since T2 has a large contribution from


V








a chemical shift mechanism, the fractionation factor determined from T2 has a large contribution of the fractionation of inner shell ligands. The fractionation factors determined from T1 are close to unity, the value of the fractionation factor of bulk water, and contain a larger contribution of the fractionation of outer shell water.

The fractionation factor of Co(H20)62+ was used to interpret the solvent hydrogen isotope effects on the formation of complexes of cobalt with the bidentate ligands glycine, N,N-dimethylglycine, and acetylacetone. The contribution of the fractionation factor of the inner shell water in Co(H20)62+ did not account completely for the measured isotope effect, and I suggest that the hydrogen / deuterium fractionation of outer shell water makes a large contribution to the isotope effect on the formation of these complexes.
The solvent hydrogen isotope effect on the equilibrium binding of iodide ion to Co(ll)-substituted carbonic anhydrase I and II was determined and then interpreted using the fractionation factor of the aqueous ligand of cobalt at the active site. This fractionation factor could not account for the measured isotope effect, implying that the hydrogen / deuterium fractionation of additional groups, or water associated with the enzyme, changes on going from the reactant state to the product state when iodide binds to Co(ll)-substituted carbonic anhydrase I and II. Although not yet helpful in interpreting the complex contributions to the isotope effects in the enzymatic catalysis, these hydrogen / deuterium fractionation factors for the water bound to cobalt in carbonic anhydrase can be significantly different from the fractionation factor for solvent water and may be sensitive to the active site environment in these homologous isozymes of carbonic anhydrase.





vi













CHAPTER 1
INTRODUCTION

The effect of an isotopic substitution on rate and equilibrium constants

associated with a chemical reaction can give information about the mechanism of the reaction. One method in making an isotopic substitution is the replacement of D20 for H20 as the solvent, which can affect a reaction in two ways. First if water is a substrate in the reaction, then the rate and equilibrium constants can be affected by the presence of D20. Also, if there are exchangeable hydrogenic sites in the reactants or products, or, in the case of an enzyme-catalyzed reaction, if there are exchangeable hydrogenic sites on the enzyme, then the rate and equilibrium constants can be affected when D20 is the solvent, because deuterium will be present in the exchangeable positions.
The difference in a reaction in H20 and D20 results from differences in the zero point energy of bonds to H or D (Schowen, 1978). In the case of a reaction rate, if the zero point energy difference for a bond to H and to D decreases on going to the transition state, then the activation energy is greater in the case of D, and the rate is lower. This is an example of a normal isotope effect, in which the rate constant when the bond is to H, kH, is greater than the rate constant when the bond is to D, kD. There also exist reactions where kD > kH, which is known as an inverse isotope effect.
The difference in zero point energy can be expressed in terms of an
equilibrium constant for isotope exchange for a particular hydrogenic site. For a single solute species in an isotopically-mixed aqueous solvent, the hydrogen /




1





2


deuterium fractionation factor, (, is the equilibrium constant for this isotope exchange reaction

R(H) + (HOD) -+ R(D) + (HOH)

[R(D)] / [R(H)]

[HOD] / [HOH]

in which a solute hydrogenic species can exchange with a solvent hydrogenic species. The fractionation factor measures the tendency of D to accumulate in the solute relative to the deuterium content of bulk solvent. A fractionation factor greater than unity implies that D will accumulate in the solute position relative to the solvent, and that the bond to hydrogen is stronger in the solute position than that bond in the solvent. For a fractionation factor less than unity, the light isotope of hydrogen will accumulate in the solute position, and the bond to hydrogen is weaker in the solute relative to the solvent.
An isotope effect is a function of the fractionation factors of each hydrogenic site in a reaction. For an isotope effect on a kinetic constant,

all reactant state sites
Il *iR
kHo_ (1-1) kD20 all transition state sites

i n

and for an isotope effect on an equilibrium constant,





3

all reactant state sites

KH20 i (1-2) KD20 all product state sites




where OR, OT, and OP are reactant, transition, and product state hydrogenic site fractionation factors, respectively (Schowen, 1978). Thus it is clear that to understand completely an isotope effect associated with a reaction, the knowledge of the underlying fractionation factors is required.
Values of the fractionation factor for various functional groups are known relative to water as the solvent (Schowen and Schowen, 1982; Kresge et al. 1987), and range from 0.4 for some examples of hydrogen bound to sulfur, to
1.5 for one case of a hydrogen bound to a tertiary amine. The majority of these fractionation factors where measured by one of two methods. One involves using NMR chemical shifts to measure the fractionation factor of a hydrogenic site directly. The technique is based on the rapid exchange of hydrogens between solute and solvent species, so that the observed proton resonance is a weighted average of the chemical shift of the solute and solvent positions. The fractionation factor is determined by measuring the chemical shift at different concentrations of solute in solvents of low and high atom fraction of deuterium. Using this method, Gold and Lowe (1968) determined the fractionation factor of the exchangeable hydrogen of acetic acid to be 0.96 0.02. A variation of this method when there is slow exchange of hydrogens between solute and solvent involves measuring the NMR signal areas in mixed isotopic solvents. In this manner Chiang et al. (1980) have determined the fractionation factor of the exchanging position of 1,8-Bis(dimethylamino)-naphthalene to be 0.90 + 0.01, and Szawelski et al. (1982) have measured the fractionation factor of the





4


hydrogen bound to sulfur in mercaptoethanol to be 0.55 0.02. The second method to determine fractionation factors is to measure the equilibrium constant for a reaction as a function of the atom fraction of deuterium in the solvent, and then using relationships like those that lead to equations 1-1 and 1-2 to calculate the fractionation factor from a fit to the data. This method is then indirect and is unable to separate the contribution from the fractionation at one site from contributions from the medium, which may be a factor. Lowe and Smith (1975) have used this method to determine the fractionation factor of the acidic hydrogen in benzoic acid to be 1.02 0.01. Both of these methods for determining fractionation factors lead to precise values for the simple cases cited above involving one exchangeable hydrogen.
The unifying theme of this work is the study of the hydrogen / deuterium fractionation factor properties of the hydrogenic positions of aqueous ligands of metal ions in simple inorganic complexes and carbonic anhydrase. There is a relative lack of information about the fractionation of the hydrogenic sites of the aqueous ligands of metals compared to that of other functional groups. Using an NMR relaxation method, I have measured the fractionation factor of the aqueous ligand of cobalt in some simple inorganic complexes and in Co(ll)substituted carbonic anhydrase. The goal is to use these fractionation factors in the interpretation of isotope effects on the formation of cobalt complexes and on equilibrium and kinetic constants associated with carbonic anhydrase. The fractionation factor of the aqueous ligand of cobalt in these systems represents the individual contributions of those hydrogenic positions to the overall isotope effect on the reactions.
Solvent hydrogen isotope effects in inorganic reactions are relatively little studied compared to those effects in organic reactions (Kresge et al. 1987). The data for ligand substitution reactions with transition metal complexes consists of





5


a small number of solvent hydrogen isotope effects measured for equilibrium constants and for rate constants of reactions with Cr3+ and Co3+ (Laughton and Robertson, 1969; Gold and Wood, 1982). Kresge et al. (1987) have suggested that large rate and equilibrium differences between reactions of transition metal aquacomplexes in H20 and D20 should be expected because the hydrogen / deuterium fractionation factor of the water molecules in the inner coordination shell would be raised to the twelfth power for a hexaaquacomplex.
Solvent hydrogen isotope effects on reactions catalyzed by carbonic anhydrase are known and have been used to deduce aspects of the catalytic mechanism (Silverman and Vincent, 1983). Carbonic anhydrase is a zinc metalloenzyme which catalyzes the reversible hydration of carbon dioxide to produce bicarbonate and a proton.
CO2+ H20 HCO" +H'
The zinc can be removed and replaced with cobalt to produce an enzyme with very similar catalytic properties (Lindskog, 1983) and the cobalt, because of its unpaired electrons, provides a spectroscopic probe of the enzyme (Bertini and Luchinat, 1983). In particular, the paramagnetic cobalt will enhance the NMR relaxation rate of water protons in solutions of Co(ll)-substituted carbonic anhydrase. The metal ion is at the base of the active site cleft and is coordinated by three histidine ligands with a water molecule as a fourth ligand which can ionize to a hydroxide. The mechanism of CO2 hydration (scheme 11) takes place by direct nucleophilic attack of a metal-bound hydroxide on C02 to produce HC03 (Steiner et al. 1975), with a water molecule then displacing HCO3- to produce a metal-bound water. A proton must then be transferred out of the active site to regenerate the metal-bound hydroxide for the next round of catalysis.





6



0H20

:Co-OH- + C Co-OH +-2 HC step 1




Co--OH2 C- Co-OH" + H* step 2 Scheme 1-1: The metal hydroxide mechanism of carbonic anhydrase.





7


When this reaction is performed in D20 some very interesting isotope effects are observed. The isotope effect on the steady-state parameter kat for both native and Co(ll)-substituted carbonic anhydrase II is large, 3.8 in the case of native zinc-carbonic anhydrase II (Steiner et al. 1975), and 1.8 for Co(Il)substituted carbonic anhydrase II (R. S. Rowlett, unpublished), which means that a primary proton transfer is at least part of the rate limiting step measured by kcat. The ratio kcat/Km contains steps up to and including the first irreversible step in the reaction, and for the catalyzed reaction the first irreversible step is the release of HCO3- from the enzyme. The isotope effect is unity on kat/Km for both native zinc-carbonic anhydrase II (Steiner et al. 1975) and for Co(ll)substituted carbonic anhydrase II (R. S. Rowlett, unpublished), which implies that the rate determining step measured by kcat is separate and distinct from the interconversion of CO2 and HC03. Thus the rate-limiting step measured by kcat is the rate of transfer of a proton from the metal-bound water out of the active site to regenerate the metal-bound hydroxide (step 2 of Scheme 1-1). From this description it is clear that the aqueous ligand of the metal at the active site of carbonic anhydrase is intimately involved in the mechanism of catalysis, and to interpret the isotope effects knowledge of the fractionation factor of the aqueous ligand of the metal is required.
The fractionation factor of the aqueous ligand of a paramagnetic metal can be determined by taking advantage of the paramagnetic contribution to the NMR relaxation rates of water protons in solutions of the metal in which the protons are in fast exchange between the coordination shell of the metal and the bulk solvent. The NMR measurements in this work were performed at a proton resonance frequency of 300 MHz, which is a higher frequency than has been used to study the water proton relaxation enhancement of first row transition metals (Dwek, 1975; Bertini and Luchinat, 1986). Thus the relaxation





8


mechanisms operating in these systems at 300 MHz need to be understood before the relaxation can be used to measure a fractionation factor, and this is discussed in Chapter 2. Chapter 3 includes the determination of the fractionation factor of Co(H20)62+ and Ni(H20)62+ along with measurements of the solvent hydrogen isotope effect on the formation constants of some cobalt complexes to see if the fractionation factor of Co(H20)62+ can account for any of these effects. In Chapter 4 I have measured the fractionation factor of three mammalian isozymes of Co(ll)-substituted carbonic anhydrase using NMR relaxation, and in Chapter 5 I discuss the use of these fractionation factors to interpret solvent hydrogen isotope effects on the equilibrium binding of an inhibitor to carbonic anhydrase and on the catalysis. Finally, the relaxation properties of Co(ll)-substituted carbonic anhydrase III have not previously been reported, and in Chapter 6 I discuss experiments to determine the magnetic field and pH dependence of the water proton relaxation enhancement in solutions of Co(ll)-substituted carbonic anhydrase II1.













CHAPTER 2
WATER PROTON NMR RELAXATION ENHANCEMENT IN
SOLUTIONS OF Co(H20)62+ AND Ni(H20)62+ Introduction

The initial studies of the influence of paramagnetic ions on the NMR relaxation rates of water protons were performed at proton resonance frequencies up to 60 MHz (Bernheim et al., 1959; Morgan and Nolle, 1959). These results were interpreted in terms of the Solomon-Bloembergen equations, which describe the relaxation as the sum of an electron-nuclear dipole-dipole interaction and a scalar interaction (Solomon, 1955; Bloembergen, 1957). At a frequency of 20 MHz the water protons in solutions of Co(H20)62+ were observed by Bernheim et al. (1959) to relax such that T1 = T2, and this relaxation was also observed at 10 MHz for solutions of Ni(H20)62+ (Morgan and Nolle, 1959). This result is consistent with the electron-nuclear dipole-dipole mechanism dominating the relaxation of the water protons, the correlation time for the interaction being the very short electron spin relaxation time. This short relaxation time for the electron in Co(H20)62+ and Ni(H20)62+ precludes any contribution to the relaxation from the scalar mechanism. In the absence of a scalar contribution, 7/6 would be the largest observable ratio for
T1/T2.
I have made measurements of the relaxation of water protons in solutions of Co(H20)62+ and Ni(H20)62+ at a proton resonance frequency of 300 MHz. In each case it was observed that the ratio T/T2 was much larger than unity, consistent with a chemical shift mechanism affecting T2 at this frequency. Because this mechanism will affect T2 to the greatest extent when the protons


9





10


are in the primary hydration shell of the ion, this allowed a qualitative differentiation between the relaxation of inner and outer shell water.

Experimental Procedure
The NMR relaxation times T1 and T2 of water protons were measured at a frequency of 300 MHz on a Nicolet NT-300 spectrometer at 23 oC. T1 was measured by the inversion recovery method according to Freeman et al. (1980) and T2 was measured by the Carr-Purcell-Meiboom-Gill sequence (Meiboom and Gill, 1958). The observed relaxation of water protons in a solution of a paramagnetic ion is the sum of the relaxation due to the paramagnetic ion, 1/Tip (i=1,2), and the relaxation in the absence of the ion, 1/Tio,


1/Ti = 1/Tip + 1/Tio (2-1) The contribution of the paramagnetic ion to the longitudinal relaxation is given by (Luz and Meiboom, 1964)

1/Tip = p'[1/(Tim +tm)] (2-2) and the paramagnetic contribution to the transverse relaxation is given by (Swift and Connick, 1962)

(1/T2m)(1/T2m + 1/tm) + AOm2
1/T2p = p' [ ] (2-3) tm((1/T2m + 1/Tm)2 + A0)m2)

where Tim is the relaxation time of a proton in the hydration shell of the metal and is described by the Solomon-Bloembergen equations, Tm is the lifetime of a proton in the hydration shell, and p' is q[M]/55.5 where q is the hydration number and [M] is the concentration of the metal ion. AOm is the chemical shift difference between a proton in the primary hydration shell and a proton in the





11


free solvent site. If Ao)m2 is small relative to the other terms, equation 2-3 reduces to equation 2-2. For these metal ions, 1/Tio was taken as the relaxation of water containing only buffer, with no added metal ions. Ri, the relaxivity, is defined as the paramagnetic relaxation per millimolar concentration of metal ion.


Ri = (1/Tip)/ ([Metal] x 1000) (2-4)

Solutions of Co(H20)62+ and Ni(H20)62+ for NMR measurements
contained ultra-pure CoC12 or NiSO4 (Aldrich Gold-Label) with 5 x 10-4 M acetic acid buffer at pH = 4.4. This pH was necessary to preclude the formation of multi-nuclear hydrolysis products (Baes and Mesmer, 1976). When EDTA was present no acetic acid was used. Water used for solution preparations was distilled and passed through two ion-exchange resin cartridges (Cole-Palmer 1506-35). D20 (99.8%) was stirred with activated charcoal, the charcoal filtered out, and then the D20 was distilled.

Results
The paramagnetic contribution to the relaxivity R2 of water protons in
solutions of Co2+ and Ni2+ at 300 MHz was greater than that obtained at lower frequencies by previous investigators (Table 2-1). For Co2+, R1 at 300 MHz and 20 MHz were close in value while R2 was much larger at 300 MHz than at 20 MHz. For Ni2+, R1 at 300 MHz was slightly larger than at 10 MHz which is understood to be a result of a field dependent increase in the electron relaxation time (Friedman et al., 1979), but R2 was much larger at 300 MHz than at 10 MHz.
The addition of EDTA to the solutions of CoCI2 and NiSO4 reduced R1
and R2 at 300 MHz (Figures 2-1 and 2-2). When an equimolar amount of EDTA





12

Table 2-1: Paramagnetic contribution to the relaxivities of water protons observed for aqueous solutions of CoCI2 and NiSO4.


frequency R1 R2 TlpfT2p
(MHz) (mM-1 s-1) (mM-1 s-1) Co(H20)62+
20 a 0.18 0.21 1.2 300 b 0.16 2.0 12.5 Ni(H20)62+
10C 0.64 0.71 1.1 300 b 0.95 2.3 2.4

a From Bernheim et al. (1959) for an unbuffered solution of 0.5 M CoCl2 at 25 0C.
b Determined in this work at 23 OC for 10 mM CoC12 or 10 mM NiSO4 which contained 10% D20 by volume and 5 x 10-4 M acetic acid buffer. The pH was
4.4.

c From Morgan and Nolle (1959) for a solution of 10 mM Ni(N03)2 at 27 OC in 0.1 M perchloric acid.





13




















4





0 E I
0 1 2 3 [EDTA]/[Co2+] Figure 2-1: The relaxivity of the protons of water at 300 MHz due to Co2+ in solutions of 10 mM CoCI2 as a function of the molar ratio of EDTA to Co2+ at 23 oC. Solutions contained 50% D20 by volume. R1 (m), R2 (A).






14











3






2

A




_W A
A




0- I
0 1 2 3 [EDTA]/[Ni2+] Figure 2-2: The relaxivity of the protons of water at 300 MHz due to Ni2+ in solutions of 10 mM NiSO4 as a function of the molar ratio of EDTA to Ni2+ at 23 OC. Solutions contained 50% D20 by volume. Ri ( ), R2 (A).





15


was added to solutions of Co2+, R1 was reduced to 47% of the value of R1 in the absence of EDTA, R2 was reduced to 5% of the value of R2 in the absence of EDTA, and Tlp/T2p was 1.2. The remaining paramagnetic relaxivity due to the CoEDTA complex was R1 = 0.075 mM-1 s-1 and R2 = 0.091 mM-1 s-1. When an equimolar amount of EDTA was added to solutions of Ni2+, R1 was reduced to 58% of the value of R1 in the absence of EDTA, R2 was reduced to 34% of the value of R2 in the absence of EDTA, and Tlp/T2p was 1.4. The remaining paramagnetic relaxivity due to the NiEDTA complex was R1 = 0.55 mM-1 s-1 and R2 = 0.78 mM-1 s-1.
The observed change in the chemical shift caused by the paramagnetic metal, Ao, at 300 MHz for a 10 mM solution of CoCI2 was measured to be 158 Hz. The measured change in the chemical shift, Aco, is the difference between the chemical shift of water protons in water containing only buffer and the chemical shift of water protons in a solution of a paramagnetic metal. This was measured with tetramethylsilane as an external reference using coaxial tubes. The influence of this chemical shift on the transverse relaxation of Co(H20)62+ solutions at 20 MHz and 300 MHz was estimated using equation 2-3 (Table 22). In the calculations, I used tm = 7.4 x 10-7 s determined by Swift and Connick (1962) using 170 exchange. Also, I used Tim = T2m as predicted by the Solomon-Bloembergen equations for Co(H20)62+ at 20 MHz and 300 MHz. From the observed T, at 300 MHz, Tim was calculated (from equations 2-1 and 2-2) to be 6.62 x 10-4 s. From the data of Bernheim et al. (1959) Tim = 5.95 x 104 s at 20 MHz. Since Tlm>>tm, the fast exchange limit applies at 20 MHz and at 300 MHz. Acom was calculated from Ao at 300 MHz using Ao = p' Aom which is valid in the fast exchange limit (Swift and Connick, 1962). I determined Awm to be 1.46 x 105 Hz at 300 MHz. Since A0m is directly proportional to the precessional frequency, Aom at 20 MHz is 9.73 x 103 Hz. The results using





16


these values with equation 2-3 are in Table 2-2 and are in good agreement with the experimental results in Table 2-1.
Discussion

The results at 300 MHz of Tlp/T2p much greater than unity for solutions of both Co(H20)62+ and Ni(H20)62+ suggest that a relaxation mechanism different from the dipole-dipole mechanism is present at this frequency. The data in Table 2-1 indicate that this mechanism primarily, if not exclusively, affects the transverse relaxation. I have considered three possible causes of this fielddependent increase in 1/T2p. The scalar contribution to 1/T2m has a correlation time equivalent to the electron relaxation time which is field dependent (Bloembergen and Morgan, 1961). However, I have rejected a contribution from the scalar term because it would require an unreasonably large increase in the electron spin relaxation time at 300 MHz. When the electron relaxation time is less than both the rotational correlation time and tm, a Curie spin mechanism can result (Gueron, 1975). This Curie spin is the thermal average of an electron spin which can become large at high field and can cause an increase in Tlm/T2m. However, I also reject a contribution from this mechanism because the rotational correlation time for a hexaaquacomplex is too short to allow the Curie spin to become effective.
Melamud and Mildvan (1975) observed a field-dependent increase in
1/T2p for water protons in solutions of Co(H20)62+ and concluded that the ACOm2 term in equation 2-3, a chemical shift mechanism, made a major contribution at the higher frequencies. I support this conclusion in two ways. First, I have calculated the magnitude of the chemical shift contribution to the transverse relaxation of Co(H20)62+ solutions at 20 MHz and 300 MHz using equation 2-3. These calculations demonstrate that the observed value of Tlp/T2p at 300 MHz is consistent with a chemical shift mechanism which affects T2p (compare





17

Table 2-2: Calculated paramagnetic relaxivities of water protons in solutions containing Co2+.

frequency Ri R2 Tlp/T2p
(MHz) (mM-1 s-') (mM-1 s-1)
20 0.18 0.19 1.1
300 0.16 1.8 11.2

Note: Calculated using equations 2-2 to 2-4 in the text. The calculations used 'm=7.4 x 10-7 s (Swift and Connick, 1962), the 20 MHz relaxation data from Bernheim et aL (1959), and the 300 MHz relaxation data and chemical shift determined in this work. A further description is in the text.





18


Tables 2-1 and 2-2). The chemical shift does not have an effect at 20 MHz at which frequency AO)m2<<1/T2mTm, 1/T2m2. The large chemical shift at 300 MHz due to the addition of Co2+ is the result of contact interactions which result from the delocalization of unpaired electron spin density. At lower magnetic field strengths this mechanism has been observed to affect T2 for 170 and 19F in cobalt complexes (Swift and Connick, 1962; Eisenstadt, 1969). The effects on the relaxation when EDTA is added to solutions of Co(H20)62+ is the second way I support the chemical shift mechanism. The chemical shift mechanism will affect T2p to the greatest extent when the protons are in the primary hydration shell of the ion because it is a contact interaction. EDTA binds to Co2+ (and Ni2+) occupying all coordination sites and thus excludes water molecules from the inner shell. Thus the addition of EDTA would prevent the water protons from experiencing the chemical shift mechanism. The ratio Tlp/T2p was near unity for water protons at 300 MHz when an equimolar amount of EDTA was added to solutions of Co2+, consistent with the absence of a chemical shift relaxation mechanism and the dominance of a dipole-dipole relaxation mechanism (Figure 2-1).
When EDTA was added to solutions of Co2+, the percent decrease in the relaxivity R2 at 300 MHz was much greater than R1 (Figure 2-1). Since EDTA is excluding water molecules from the inner coordination shell, this suggests that R2 has a much larger contribution from the relaxation of water protons in the inner shell. The residual paramagnetic relaxivity of the cobalt-EDTA complex is most likely due to water protons in an "outer shell" of the complex. This residual relaxivity makes up a much larger part of R1 than R2 observed for Co(H20)62+ solutions at 300 MHz. This suggests that the relaxation of outer shell water contributes to a greater extent to R1 than to R2 in Co(H20)62+.





19


As EDTA was added to solutions of Ni2+, R2 decreased more than R1
(Figure 2-2), though the decrease is not as large as seen in solutions of Co2+. Also, the residual relaxivity of the NiEDTA complex is much larger than in solutions of the CoEDTA complex, and this residual relaxivity makes up a large part of both R1 and R2. These results for Ni2+ suggest that the chemical shift mechanism is responsible for the increase in R2 at 300 MHz, but that R1 and R2 each have a large contribution of outer shell water.













CHAPTER 3
HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF THE
AQUEOUS LIGAND IN Co(H20)62+ AND Ni(H20)62+ Introduction
The hydrogen / deuterium fractionation factor, #, of a metal-bound water is the equilibrium constant for this isotope exchange reaction: M(HOH) + (HOD) M(HOD) + HOH)


[M(HOD)] / [M(HOH)]
(3-1)
[HOD] / [HOH]


This fractionation factor measures the tendency of deuterium to accumulate at the aqueous ligand of the metal relative to the deuterium content of bulk solvent. Fractionation factors are valuable in interpreting the effects of deuterium on kinetic and equilibrium constants because they represent individual contributions to isotope effects measured relative to the common reference of water. The fractionation of hydrogen isotopes in water bound to metal ions has been considered (Schowen and Schowen, 1982); however, there are few reports of the fractionation factor of the aqueous ligands of a metal. Using NMR methods, Melton and Pollack (1969) have measured such a factor for the water ligands of Cr(H20)62+ (0 = 1.00 0.03). They relied on the properties of a paramagnetic metal to enhance the NMR relaxation rate of protons of water exchanging rapidly between bulk solvent and the hydration shell of the metal. By measuring the proton relaxation rate as a function of the deuterium content of solvent, the fractionation factor of the metal-bound site can be determined.


20





21


Using this method at a proton resonance frequency of 300 MHz, I have measured the fractionation factor of the aqueous ligand of cobalt in Co(H20)62+ and of nickel in Ni(H20)62+. In the case of Co(H20)62+ the value of the fractionation factor was dependent upon whether it was determined from T1 or T2; however, the corresponding values of 0 for Ni(H20)62+ were identical when obtained from T1 or T2. The analysis of the relaxation discussed in Chapter 2 can be used to interpret the fractionation factors, and allows a qualitative differentiation between the hydrogen / deuterium fractionation of inner and outer shell water in these complexes.
I have then used the knowledge of the fractionation of Co(H20)62+ to study the solvent hydrogen isotope effects on the formation constants of the reactions of glycine, N,N-dimethylglycine, and acetylacetone with Co2+. An isotope effect on an equilibrium constant is the product of the reactant state hydrogenic site fractionation factors divided by the product of the product state hydrogenic site fractionation factors (equation 1-2). In a reaction of Co2+ with a bidentate ligand L-1 represented as
Co(H20)62+ + L-1 Co(H20)4L+ + 2 H20 (3-2) the solvent hydrogen isotope effect is a function of the fractionation of the hydrogenic positions of Co(H20)62+ as well as Co(H20)4L+ and H20, and the fractionation factor of Co(H20)62+ represents the individual contribution of the hydrogenic positions of Co(H20)62+ to the overall isotope effect. Because the solvent hydrogen isotope effects on the formation constants were measured to be near unity in the case of each of the ligands, I determined that the contributions of the fractionation of the hydrogenic positions of the inner shell water in Co(H20)62+ and in the ligand complexes of cobalt can not account completely for the overall isotope effect. I suggest that the fractionation of outer





22


shell water also makes a large contribution to the isotope effect on these reactions.
Experimental Procedure
NMR Measurements: Measurements at 300 MHz and solutions were the same as in Chapter 2. Some measurements of T1 of water protons in solutions of Co2+ were performed at frequencies between 0.01 and 50 MHz on a field cycling "relaxometer" at the IBM T. J. Watson Research Center, Yorktown Heights, NY (Koenig et al., 1983).
Measurement of the Hydrogen / Deuterium Fractionation Factor: The measurement of the relaxation of water protons in solutions of varying deuterium content and containing paramagnetic metal ions is given by

1/Tip = p'[X]/(1 n + no) (3-3) where n is the atom fraction of deuterium in solvent water, # is the hydrogen / deuterium fractionation factor of the water ligands of the metal, and [X] is the term in brackets in equations 2-2 and 2-3 for i = 1 or 2, respectively. This equation is derived considering the fractionation of isotopes that arises when there is a deuterium content in water. In this case the ratio of protons in the hydration shell of the metal to total protons in solution is p'/(1-n+nO) (Schowen and Schowen, 1982). Equation 3-3 can be rearranged to

p'Tip = [X]-1 n(1-0)[X]-1 (3-4) Thus p'Tip vs n gives a slope of -(1-0)[X]-1 and an intercept of [X]-1, and # = 1 + slope / intercept.





23


Determination of the isotope effects on formation constants: The
formation constants for the reactions of Co2+ with a bidentate ligand, L-1, as in equation 3-2 are defined as

[Co(H20)4L+]
kl = (3-5) [Co(H20)62+] [L-1]
with similar expressions for k2 and k3. Potentiometric titrations were performed in H20 and D20 (99.8%) containing 1 mM CoC12 and 4 mM glycine, N,Ndimethylglycine, or acetylacetone. The values of pH and pD used in these experiments were corrected from pH meter readings. The correction of a pH meter reading in 100% D20 is pD = meter reading + 0.4 (Glasoe and Long, 1960). The formation constants were determined by a non-linear least-squares fit of the data (RS1, BBN Software, Cambridge, MA) to the equation derived by Carlson et al. (1945) describing n, the ratio of the concentration of metal-bound ligand to total concentration of metal, as a function of the concentration of coordinating ligand, [A].

kl [A] + 2klk2[A]2 + 3kl k2k3[A]3
(3-6)
1 + kl [A] + kl k2[A]2 + klk2k3[A]3

The calculation of ii and [A] required knowledge of the acid ionization
constants of the free ligands, which were measured by potentiometric titrations in H20 and D20 (Table 3-1). The negative logarithm of the acid ionization constants, pK, in H20 were all within 0.07 pK units from the literature values (Smith and Martell, 1975). The solvent hydrogen isotope effects for glycine expressed as ApK where ApK = pKD20 pKH20 were within 0.03 ApK units from the values obtained previously by Jencks and Salvesen (1971).





24

Table 3-1: Solvent hydrogen isotope effects on the acid ionization constants of glycine, N,N-dimethylglycine, and acetylacetone.

(pK1)H20 a ApK1 (pK2)H20 b ApK2 glycine 2.37 0.02 0.40 0.04 9.72 0.03 0.56 0.05 N,N-dimethylglycine 1.96 0.03 0.40 0.04 9.81 0.03 0.55 0.05 acetylacetone 9.03 0.10 0.62 + 0.17

Note: Acid ionization constants were determined in H20 and D20 by potentiometric titrations at 230C. Solutions contained 10 mM of either glycine, N,N-dimethylglycine, or acetylacetone. a pK = -log K where K is the acid ionization constant. Standard errors resulted from a non-linear least-squares fit of the titration data. ApK = pKD20 pKH2o.





25


These potentiometric titrations were performed in solutions which contained ultra-pure CoCl2 (Aldrich Gold-Label), and glycine and N,Ndimethylglycine from Sigma and acetylacetone from Aldrich. Water used for solution preparations was distilled and passed through two ion-exchange resin cartridges (Cole-Palmer 1506-35). D20 (99.8%) was stirred with activated charcoal, the charcoal filtered out, and then the D20 was distilled. Glassware was rinsed with a solution of EDTA prior to use.
Results
Fractionation Factors of Co(HO.2g and Ni(HzO ), The relaxation of water protons in solutions of 10 mM CoC12 and 10 mM NiSO4 as a function of the atom fraction of deuterium in solvent water is shown in Figures 3-1 and 3-2. From these data, the hydrogen / deuterium fractionation factor, 0, for Co(H20)62+ and Ni(H20)62+ was obtained (Table 3-2). The value of the fractionation factor was independent of the concentration of metal in the range 0.010 M to 0.025 M. The value of 0 for Co(H20)62+ was dependent on whether it was obtained from T1 or T2; however, the corresponding values of 0 for Ni(H20)62+ were identical when obtained from T1 or T2. In a separate experiment, the fractionation factor from T1 for Co(H20)62+ was determined at frequencies between .01 and 50 MHz on a field cycling "relaxometer" at the IBM T. J. Watson Research Center and found to be virtually field independent (data not shown).
Solvent Hvdroaen Isotope Effects on the Formation Constants of Cobalt Complexes: The formation constants for the reactions of Co2+ with glycine, N,Ndimethylglycine, and acetylacetone were determined in H20 and D20 and are reported as the logarithm of the formation constants (Table 3-3). Representative data of a formation curve (Carlson et al., 1945) for Co2+ and glycine are in Figure 3-3 where if is plotted as a function of the negative logarithm of the





26




7.1 6.7

%. 6.3
a
5.9

5.5
0.0 0.2 0.4 0.6 0.8 1.0
Atom fraction of deuterium in solvent (n)



6.0


5.0

a.
I- 4.0

3.0

0.0 0.2 0.4 0.6 0.8 1.0
Atom fraction of deuterium in solvent (n)

Figure 3-1: The dependence of p'Tip (i = 1,2) at 300 MHz on n, the atom fraction of deuterium in solvent water, for a 10 mM solution of CoC12 at 23 OC. The solutions were buffered at pH 4.4 with 0.5 mM acetic acid. Data are the means and standard errors propagated from NMR measurements.





27









12.0



10.0


0
x 8.0
0.


6.0



4.0
0.0 0.2 0.4 0.6 0.8 1.0

Atom fraction of deuterium in solvent (n) Figure 3-2: The dependence of p'Tip (i = 1,2) at 300 MHz on n, the atom fraction of deuterium in solvent water, for a 10 mM solution of NiSO4 at 23 OC. The solutions were buffered at pH 4.4 with 0.5 mM acetic acid. Data are the means and standard errors propagated from NMR measurements. p'Tlp (0) p'T2p ( ).





28

Table 3-2: Fractionation factors determined from T1 and T2 at 300 MHz for the aqueous ligands of the paramagnetic metal in Co(H20)62+ and Ni(H20)62+.

OT1 4T2
Co(H20)62+ 0.95+.01 0.73+.02 Ni(H20)62+ 0.99.01 0.98.03

Note: Measurements were made at 23 oC for 10 mM CoCl2 or NiSO4 containing
5 x 10-4 M acetic acid buffer at pH 4.4. Data were obtained from the least squares slope and intercept and standard error in a plot such as Fig. 3-1 according to equation 3-4.





29

Table 3-3: Logarithm of the formation constants determined for the reactions of Co2+ with the ligands glycine, N,N-dimethylglycine, and acetylacetone in H20 and D20.

H20 D20 (log k)Do (log k)Hp glycine log k1 5.04 0.04 4.99 0.02 -0.05 0.04 log k2 4.32 0.04 4.23 0.02 -0.08 0.05 N,N-dimethylglycine log k1 4.14 0.04 4.12 0.06 -0.03 0.07 log k2 3.20 0.06 3.18 0.13 -0.01 0.14 acetylacetone log kl 5.30 0.07 5.42 0.10 0.12 0.13 log k2 4.40 0.09 4.61 0.12 0.21 + 0.15

Note: Determined by potentiometric titrations of solutions which contained 1 mM CoCI12 and 4 mM of the ligand and fitting the data to equation 3-6. Measurements were made at 23 oC. See equation 3-5 for the definition of kl.





30









3





2




1





-2.5 3.5 4.5 5.5 p[A]

Figure 3-3: The ratio of the concentration of metal-bound ligand to total concentration of metal, fi, as a function of the negative logarithm of the concentration of free glycine in solution existing as NH2-CH2-COO-1, p[A] (see equation 3-6). The solutions contained 1 mM total CoC00012 and 4 mM total glycine in H20 (0) or D20 (m) and were titrated with NaOH (NaOD). The solid lines are a computer fit to the data using equation 3-6 with the values of the formation constants given in Table 3-3.





31


concentration of coordinating ligand, p[A]. In the region of the formation curves where p[A] = 3.0 the pH of the solutions was greater than 9.0. Near this pH the hydrolysis of the cobalt-bound water to a cobalt-bound hydroxide occurs in Co(H20)62+ (Baes and Mesmer, 1976) and this same hydrolysis may be occurring in the water ligands of the cobalt complexes with the bidentate ligands. The data in the region p[A] = 3.0 was therefore difficult to interpret, and I was unable to obtain precise values for k3. For glycine and acetylacetone in H20 the values of log kl were within 0.1 log unit and the values of log k2 were within 0.3 log unit of the literature values (Smith and Martell, 1975; no literature value was available for N,N-dimethylglycine). The solvent hydrogen isotope effects on these formation constants were all unity within experimental uncertainty (Table 3-3), with the possible exception of k2 for acetylacetone.
For the binding of a ligand L-' to cobalt as in equation 3-2 the solvent hydrogen isotope effect is (see equation 1-2)

(kl)H20 (OCo(H20)62+)12
(3-7)
(kl)Do20 (Co(H20)4(L)+)8(OH20)4

and from this the fractionation factor of the remaining metal-bound waters in Co(H20)4L+ can be calculated by using the measured isotope effect, the measured fractionation factor for Co(H20)62+ from NMR relaxation, and the fractionation factor for H20 which is 1.0 (Schowen and Schowen, 1982). An analogous calculation can be done to determine the fractionation factor of the remaining metal-bound waters in Co(H20)2(L)2. Fractionation factors T1 and 4T2 for Co(H20)62+ have been measured from the NMR relaxation times T1 and T2 of water protons (Table 3-2). The results of these calculations from equation






32


3-7 with glycine, N,N-dimethylglycine, and acetylacetone are in Table 3-4 using (T1 and (T2 for Co(H20)62+.
I have also tried to measure the fractionation factor for these coordinated water molecules in the complexes directly, using the NMR relaxation method used to determine the fractionation factor of Co(H20)62+. However, these solutions of Co2+ and a bidentate ligand L- consist of an equilibrium mixture of Co(H20)62+ Co(H20)4L+ Co(H20)2L2 and CoL3- and the observed paramagnetic relaxation is the sum of the relaxation of each of these four species. This presents difficulties in obtaining the fractionation factor of one species from these measurements.
Discussion

The definition of the hydrogen / deuterium fractionation factor I have given (equation 3-1) assumes the rule of the geometric mean to be valid (Schowen and Schowen, 1982); that is, the fractionation factor of one particular hydrogenic site is independent of the isotopic composition of any other site. I will assume that the rule is valid in my system, and this is supported by the high degree of linearity in plots as in Figures 3-1 and 3-2. A deviation from the rule of the geometric mean would result in a non-linear dependence of the relaxation data on the atom fraction of deuterium in the solvent.
The value of the fractionation factor for the water ligands of cobalt in
Co(H20)62+ was dependent on whether it was determined from T1 or T2, while that of Ni(H20)62+ was independent of T1 or T2 (Table 3-2). These fractionation factors are a weighted value of all the protons that experience the paramagnetic relaxation. An explanation for this difference in (T1, and O(T2 is based on the analysis of the contributions to the relaxation at 300 MHz (see Chapter 2). The paramagnetic contribution to the transverse relaxation of water protons in solutions of Co(H20)62+ has a predominant contribution of water in the inner





33

Table 3-4: Calculated fractionation factors for the metal-bound water in the complexes of cobalt with glycine(gly), N,N-dimethylglycine (DMG), and acetylacetone (acac).

OT1 OT2

Co(H20)62+ a 0.95 0.01 0.73 0.02

(Co(H20)4(gly)+ 0.91 0.02 0.61 0.03 OCo(H20)2(gly)2 0.79 0.04 0.36 0.03

OCo(H20)4(DMG)+ 0.92 0.02 0.62 0.03 OCo(H20)2(DMG)2 0.84 0.08 0.38 0.05

OCo(H20)4(acac)+ 0.96 0.04 0.65 0.04 )Co(H20)2(acac)2 1.04 0.12 0.47 0.07

Note: Calculated from equation 3-7 using the isotope effects on the formation constants of the reaction of Co2+ with glycine, N,N-dimethylglycine, and acetylacetone (Table 3-3) and OT1 and )T2 measured directly for Co(H20)62+. A further description is in the text. a Determined directly from NMR relaxation measurements.





34


coordination shell, suggesting that T2 is measuring more of the fractionation of inner shell water. Since Tip has a larger contribution of outer shell water, OT1 is more of the fractionation of outer shell water. The analysis of the relaxation for Ni(H20)62+ (Chapter 2) suggests that both T1 and T2 have a large contribution of outer shell water, so OT1 and OT2 for Ni(H20)62+ each represent a larger contribution of the fractionation factor of outer shell water, and the influence of the fractionation of the inner shell water is not measured by T1 or OT2

This interpretation of the fractionation factor for Co(H20)62+ is consistent with expectations for the fractionation factor of metal-bound water. The fractionation factor for H30+ is 0.69 (Chiang et al., 1980), presumably due to the positive charge on the oxygen. It has been suggested that a metal-bound water site ought to have a fractionation factor approximated by (0.69)8, with 8 measuring the degree of positive charge transferred from the metal to the oxygen (Schowen and Schowen, 1982). The fractionation factor from T2 for Co(H20)62+ being 0.73 supports this suggestion in resembling the fractionation factor for H3O+. Pure water has by definition a fractionation factor of 1.0. Outer shell water, being much more loosely bound than inner shell water, should be much more like bulk water in its hydrogen / deuterium fractionation properties (More O'Ferrall et al., 1971). The observation of @T1 for Co(H20)62+ and OT and 4T2 for Ni(H20)62+ all being close to unity is consistent with these fractionation factors having a large contribution of the fractionation of outer shell water. The fractionation factor of the inner shell water in Ni(H20)62+ may well be affected by the positive charge of the nickel and have a value closer to 0.69, or it may have a value near unity; however, because of the large contribution of outer shell water to both T1 and T2 in solutions of Ni(H20)62+, the influence of the inner shell water is not large in the fractionation factors determined from T1 and T2.





35


The usefulness of fractionation factors is judged by their ability to aid in

interpretation of solvent hydrogen isotope effects. I have determined the solvent hydrogen isotope effects on the formation of some simple inorganic complexes of Co2+ with the goal to use the knowledge of the fractionation of Co(H20)62+ to interpret those isotope effects. An isotope effect is a function of all the fractionation factors for each hydrogenic site in a reaction (equation 1-2), and for a reaction such as equation 3-2 the fractionation factor of Co(H20)62+ represents the individual contribution of those hydrogenic sites to the overall isotope effect on the reaction.
In considering the solvent hydrogen isotope effect on the formation
constants (Table 3-3), initially I will assume that the isotope effect is made up entirely of the contribution of the isotope preferences of the hydrogenic positions of water in the inner coordination shell of Co(H20)62+, CO(H20)4L+, and Co(H20)2L2. With this assumption there is enough information from the isotope effect and the fractionation factor of Co(H20)62+ to calculate (from equation 3-7) the fractionation factor of the remaining waters in the cobaltbidentate ligand complexes (Table 3-4). While I have done the calculations with equation 3-7 using both OT1 and T2 for Co(H20)62+ (Table 3-2), the values from OT1 may not be meaningful since T1 is measuring more outer shell water, and the number of outer shell waters is not known. One interpretation of these calculations is that since OT2 of CO(H20)62+ measures more of the fractionation of inner shell water in Co(H20)62+, the calculation using OT2 of CO(H20)62+ represents the resulting fractionation factor of the inner shell waters in the cobalt complexes. As ligands are added to Co2+ the calculated fractionation factor of the remaining coordinated waters decreases, and this decrease is the same for each of the three ligands, though glycine and N,N-dimethylglycine coordinate through an amino and carboxylate, and acetylacetone binds through two





36


carbonyls. If the assumption in these calculations is valid, the results of a lower fractionation factor for Co(H20)4L+ means that the O-H bond becomes weaker relative to that bond in Co(H20)62+, and the O-H bond is also weaker in Co(H20)2L2 compared to Co(H20)4L+. This is similar to the effect of added ligands on the lifetime of water molecules in the hydration shell which, for instance, is shorter in Co(H20)4gly+ relative to Co(H20)62+ (Hammes and Steinfeld, 1962), meaning that the hydration shell becomes more loosely bound when ligands are added. Thus in these cobalt complexes with bidentate ligands the metal-oxygen bond of a coordinated water is looser, and the oxygen hydrogen bond appears looser, relative to Co(H20)62+.
However, from the measured solvent hydrogen isotope effects (Table 33) and the calculations of Table 3-4, there are a number of reasons to question the assumption that the only contribution to the isotope effects comes from the hydrogens of inner shell water. First, such strong fractionation as 0.36 (Table 34) is rare (Schowen and Schowen, 1982). Second, the lower fractionation factor for Co(H20)4L+ and Co(H20)2L2 is in contrast to the suggestion that the charge on the cobalt affects the fractionation of the metal-bound water (Schowen and Schowen, 1982). In these complexes, the net charge on the complex is +1 for Co(H20)4L+, and neutral for Co(H20)2L2, since the ligand has a charge of -1 in each case. Thus one would expect the fractionation factor of the waters in the ligand complexes to be closer to unity, according to the expectation that 0 = (0.69)8, where 0.69 is the fractionation factor of H30+ and 8 is the degree of positive charge transferred from the metal to the oxygen of the coordinated water molecule (Schowen and Schowen, 1982). The OT2 for Co(H20)62+ is consistent with this (Table 3-2), and thus one would expect [0 for Co(H20)62+] < [0 for CO(H20)4L+] < [0 for Co(H20)2L2], but this opposite to what was observed (Table 3-4). Finally, based upon the large number of hydrogenic





37


sites in Co(H20)62+, large isotope effects would be expected if the only contributions to the isotope effect came from the fractionation of inner shbll water (Kresge et al., 1987). However, isotope effects of unity were observed for these reactions (Table 3-3).
I suggest that the individual contribution of the fractionation factor of the inner shell waters in Co(H20)62+, Co(H20)4L+, and Co(H20)2L2 to the overall solvent hydrogen isotope effect on the formation constants is small. The fractionation factor of some other hydrogenic sites has a contribution to the isotope effect which is sufficient to cancel the contribution of the inner shell waters so that the overall isotope effect is unity. The most likely choice for this is the fractionation of outer shell water. Other possibilities include, in the case of glycine, the composite fractionation factor of the hydration shell of the carboxylate group, but this would be a small contribution of a fractionation factor value already close to unity (the fractionation factor of the hydration shell of the carboxylate group in the acetate ion is 0.89, Goodall and Long, 1968). The fractionation factor of the hydrogens on the nitrogen in glycine are probably not important since the isotope effect is the same for N,N-dimethylglycine (Table 33).

The importance of outer shell water in ligand substitution reactions is not without precedent. The first step in the proposed mechanism of ligand substitution reactions is the formation of an outer shell complex with the ligand and metal (Eigen and Tamm, 1962), followed by replacement by the ligand of a water molecule in the inner coordination shell of the metal. Hence it is reasonable to assume that the binding of a ligand would affect the structure of outer shell water.
The following example illustrates how the fractionation of outer shell
water could contribute to the isotope effect. Equation 3-7 can be expanded to





38


include the fractionation factor of outer shell water of Co(H20)62+ in the numerator and of Co(H20)4L+ in the denominator. The number of outer shell waters would be larger than the number of inner shell waters, and if the contribution of an outer shell hydrogenic site in equation 3-7 was 1.06, twelve outer shell waters (24 hydrogenic positions) would contribute (1.06)24 = 4.05. This would then mean that the fractionation factor of the inner shell waters in Co(H20)4L+ is 0.74. If the contribution of an outer shell hydrogenic site is 1.10, then the fractionation factor of the inner shell waters in Co(H20)4L+ is 0.83. Thus a contribution of outer shell water slightly larger than unity would, when raised to a large power for the number of hydrogenic positions, cancel the effect of the inner shell waters.
I conclude that the solvent hydrogen isotope effect of unity on the
formation of complexes of Co2+ and glycine, N,N-dimethylglycine, acetylacetone points to the importance of outer shell water in these reactions. Conversely, one can not get information about the effect of added ligands on the fractionation properties of the inner shell waters from the solvent hydrogen isotope effect on the formation constants. This work shows that the solvent hydrogen isotope effect is the complex contribution of the hydrogen / deuterium fractionation of several different groups or species of water.













CHAPTER 4
HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF THE
AQUEOUS LIGAND OF COBALT IN Co(II)-SUBSTITUTED CARBONIC ANHYDRASE
Introduction
Carbonic anhydrase is a zinc metalloenzyme which catalyzes the
reversible hydration of carbon dioxide to produce bicarbonate and a proton.
CO2 + H2 HCO+ H+
The aqueous ligand of the metal at the active site of carbonic anhydrase is believed to have a role in the catalytic pathway. Direct nucleophilic attack by the metal-bound hydroxide on CO2 is a likely step in the production of bicarbonate (Pocker and Sarkanen, 1978; Lindskog, 1983; Silverman and Lindskog, 1988), and an intramolecular proton transfer involving the metalbound water is proposed to be a rate-limiting step in catalysis by isozyme II (Steiner et al., 1975). The solvent hydrogen isotope effect associated with the turnover number for CO2 hydration catalyzed by native zinc-carbonic anhydrase II is 3.8 (Steiner et al, 1975). To interpret fully the solvent hydrogen isotope effects observed for catalysis by carbonic anhydrase requires knowledge of the fractionation factor of the aqueous ligand of the metal.
The fractionation factor measures the tendency of deuterium to
accumulate at the aqueous ligand of the metal relative to the deuterium content of bulk solvent. There are few reports of the fractionation factor of the aqueous ligands of a metal at the active site of an enzyme. Using NMR methods, Silverman (1981) has measured such a factor for the water ligands of cobalt in cobalt(ll)-substituted carbonic anhydrase II ( -= 1.05 0.17) by taking


39





40


advantage of the properties of a paramagnetic metal to enhance the relaxation rate of water exchanging rapidly between bulk solvent and the hydration shell of the metal and measuring the proton relaxation rate as a function of the deuterium content of solvent. Silverman measured that fractionation factor using T1 at a proton resonance frequency of 100 MHz. Because of the unique relaxation properties of water protons in solutions of Co(H20)62+ at 300 MHz (Chapter 2), I have determined the fractionation factor of the aqueous ligand of cobalt in three mammalian isozymes of Co(ll)-substituted carbonic anhydrase from T1 and T2 at 300 MHz. In each case T1 > T2, implying that the same relaxation mechanisms occur for water protons in solutions of Co(ll)-substituted carbonic anhydrase at 300 MHz as in solutions of Co(H20)62+. The values of the fractionation factor determined from T1 for each isozyme of Co(ll)-substituted carbonic anhydrase were close to the fractionation factor for bulk water, which is unity. Except in the case of Co(ll)-substituted carbonic anhydrase III, the fractionation factors obtained from T2 were less than unity and close to the fractionation factor for H30+ which is 0.69. I conclude that fractionation factors in these cases determined from T1 and T2 measured isotope preferences for different populations of ligand sites. I suggest that since T2 has a large contribution from a paramagnetic chemical shift, the fractionation factors determined from T2 have a large contribution of the hydrogen / deuterium fractionation of the inner shell ligand. The fractionation factors determined from T1 are close to unity, the value in bulk water, and contain a larger contribution of the fractionation of outer shell water.
Experimental Procedure
Enzymes: Human carbonic anhydrase I was purified from human red
blood cells by an affinity chromatography method (Kalifah et al., 1977). Bovine carbonic anhydrase II was obtained from Sigma Chemical and used after





41


extensive dialysis to remove paramagnetic impurities. Bovine carbonic anhydrase III was obtained from bovine flank steak by gel filtration and anionexchange chromatography (Tu et al., 1986). The concentration of carbonic anhydrase I, II, and III was estimated at 280 nm using e -= 4.7 x 104 M-1 cm-1 (Nyman and Lindskog, 1964), 5.4 x 104 M-1 cm-1 (Nyman and Lindskog, 1964), and 6.4 x 104 M-1 cm-1 (Engberg and Lindskog, 1984), respectively.
The apoenzymes of carbonic anhydrase I and II were prepared
according to the procedure of Hunt et al. (1977) by dialysis against dipicolinic acid. Co(ll) substituted carbonic anhydrase I and II were prepared by the addition of 1.1 equivalents of CoCI2 followed by dialysis against several changes of large volumes of deionized water. The apoenzyme of carbonic anhydrase III was prepared by addition of the chelator 2-carboxy-1,10phenanthroline to a solution of enzyme followed by dialysis against excess CoC12 as described by Engberg and Lindskog (1984).
NMR Measurements: Measurements at 300 MHz were the same as in Chapter 2. In the cases of Co(ll)-substituted carbonic anhydrase I and II, 1/T was taken as the relaxation of solutions of the Co(ll)-substituted enzyme in the presence of a molar excess of the inhibitor acetazolamide, which is known to displace the aqueous ligand of the metal at the active site. Enough acetazolamide was present to bind to 99.9% of the active sites. Alternatively for Co(ll)-isozymes I and II and solely for Co(ll)-substituted carbonic anhydrase III, 1/Tio was taken as the relaxation of solutions of the native zinc enzyme, which has no paramagnetic contribution. These two methods were equivalent at a frequency of 300 MHz for Co(ll)-substituted carbonic anhydrase I and II. Solutions of enzyme for relaxation measurements were buffered with 50 mM Hepes [4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid] at pH 8.5. At this pH the relaxivity of solutions of Co(ll)-substituted carbonic anhydrase I and II is





42


greatest and in a region that is independent of pH (Wells et al., 1979). (See Chapter 6 for a discussion of the pH and magnetic field dependence of the relaxation of water protons in solutions of Co(ll)-substituted carbonic anhydrase II.) The values of pH reported here are uncorrected pH meter readings. The correction of a pH meter reading in 100% D20 (pD = meter reading + 0.4) is approximately offset by the change in ionization state of the buffer in D20 (pKD2
- pKH20 = 0.5 0.1 for almost all acids with pK values between 3 and 10, Schowen, 1978). Water used for solution preparations was distilled and passed through two ion-exchange resin cartridges (Cole-Palmer 1506-35). D20 (99.8%) was stirred with activated charcoal, the charcoal filtered out, and then the D20 was distilled.
Measurement of the Hydrogen / Deuterium Fractionation Factor: The hydrogen / deuterium fractionation factor was determined by measuring the relaxation of water protons in solutions of Co(ll)-substituted carbonic anhydrase as a function of the atom fraction of deuterium in solvent water as described in Chapter 3.
Results
The substitution of cobalt for zinc at the active site of carbonic anhydrase provides a paramagnetic center which affects the relaxation of water protons in solutions of carbonic anhydrase. The paramagnetic relaxivity of these water protons is greater for R2 than for R1 for each isozyme of carbonic anhydrase (Table 4-1).
The relaxation of water protons in solutions of the three isozymes of Co(ll)-substituted carbonic anhydrase as a function of the atom fraction of deuterium in solvent water is shown in Figures 4-1, 4-2, and 4-3. The hydrogen / deuterium fractionation factors calculated from equation 3-4 using T1 were near unity for the three isozymes (Table 4-2). However, the fractionation factors





43

Table 4-1: Paramagnetic contribution to the relaxivities of water protons observed for aqueous solutions of Co(ll)-substituted carbonic anhydrase I, II, and III.


frequency R1 R2 Tlp/T2p (MHz) (mM-1s-1) (mM-1s-1)
Co(ll)-carbonic anhydrase I 300 0.28 2.3 8.2 Co(ll)-carbonic anhydrase II 300 0.26 1.4 5.3 Co(ll)-carbonic anhydrase III 300 0.14 2.2 16 Note: Measurements were made at 23 oC for solutions of enzyme buffered at pH 8.5 by 50 mM Hepes. Solutions contained 20% D20 by volume.





44




3.8


S3.5


3.2
C.
2.9


2.6 *
0.0 0.2 0.4 0.6 0.8 1.0
Atom fraction of deuterium in solvent (n)

4.2

W 3.9 3.6
x
0. 3.3

a. 3.0

2.7
0.0 0.2 0.4 0.6 0.8 1.0
Atom fraction of deuterium in solvent (n)


Figure 4-1: The dependence of p'Tip (i=1,2) at 300 MHz on n, the atom fraction of deuterium in solvent water, for 1.5 mM solutions of Co(ll)-substituted carbonic anhydrase I. Solutions contained 50 mM Hepes buffer at pH 8.5 and 230C. Data are the means and standard errors of three measurements.





45



4.1

3.8
Un
0

%-X 3.5
CL
3.2


2.9
0.0 0.2 0.4 0.6 0.8 1.0
Atom fraction of deuterium in solvent (n)



7.0


6.5
0

%X 6.0

I
5.5


5.0
0.0 0.2 0.4 0.6 0.8 1.0
Atom fraction of deuterium in solvent (n)

Figure 4-2: The dependence of p'Tip (i=1,2) at 300 MHz on n, the atom fraction of deuterium in solvent water, for 2.0 mM solutions of Co(ll)-substituted carbonic anhydrase II. Solutions contained 50 mM Hepes buffer at pH 8.5 and 230C. Data are the means and standard errors of three measurements.





46




8.0


m 7.0

S6.0


-- 5.0


4.0 ,
0.0 0.2 0.4 0.6 0.8 1.0
Atom fraction of deuterium in solvent (n)


8.0


6.0




-- 2.00.0 ,
0.0 0.2 0.4 0.6 0.8 1.0
Atom fraction of deuterium in solvent (n)

Figure 4-3: The dependence of p'Tip (i=1,2) at 300 MHz on n, the atom fraction of deuterium in solvent water, for 1.5 mM solutions of Co(ll)-substituted carbonic anhydrase III. Solutions contained 50 mM Hepes buffer at pH 8.5 and 230C. Data are the means and standard errors of three measurements.





47

Table 4-2: Hydrogen/deuterium fractionation factors determined from T1 and T2 at 300 MHz for the aqueous ligand of the paramagnetic metal in three mammalian isozymes of Co(ll)-substituted carbonic anhydrase.


OT1 OT2
Co(ll)-carbonic anhydrase I 1.10.02 0.72.02 Co(ll)-carbonic anhydrase II 0.95.02 0.77.01 Co(ll)-carbonic anhydrase III 1.03.06 1.00.07

Note: Data were obtained from the least squares slope and intercept and standard error in a plot such as Figure 4-1 according to equation 3-4. Measurements were made at 23 OC for solutions of enzyme buffered at pH 8.5 by 50 mM Hepes.





48


calculated using T2 were significantly less than unity for isozymes I and II, but near unity for isozyme III (Table 4-2).

Discussion
The relaxation behavior of water protons at 300 MHz and pH 8.5 in
solutions of the Co(ll)-substituted carbonic anhydrases showed Tlp/T2p much greater than unity (Table 4-1). This was the same as observed for the water protons in solutions of cobalt ions (see Chapter 2), and the interpretation in the case of Co(H20)62+ can be applied to Co(ll)-substituted carbonic anhydrase. At 300 MHz, there is a field dependent increase in the chemical shift difference between a proton in the metal-bound site and the free solvent site. This causes an increase in the transverse relaxation of the protons as they experience the two environments (equation 2-3). Because this relaxation mechanism has as its source a paramagnetic contact shift, the transverse relaxation has a larger contribution of the ligand in the inner coordination shell of the cobalt, while the longitudinal relaxation has a larger contribution of water in the outer coordination shell. Thus the fractionation factor measured from T2 represents more of the fractionation of the inner shell aqueous ligand of cobalt in Co(ll)substituted carbonic anhydrase, while the fractionation factor measured from T1 is a greater contribution of the fractionation of outer shell water. A similar situation in which the outer shell relaxation has a larger contribution to R, than to R2 has been reported by Kushnir and Navon (1984) for water protons in solution with Mn(ll)-substituted carbonic anhydrase II.
These relaxation measurements were made at pH 8.5, a pH for which the relaxation rate of water protons in solutions of Co(ll)-substituted carbonic anhydrase I and II is greatest (Wells et al., 1979). (See Chapter 6 for an investigation of this property in solutions of Co(ll)-substituted carbonic anhydrase II.) This pH is above the value of the pK for the ionization of the





49


active site group as determined in many experiments (see Table 5-1 and Engberg and Lindskog, 1984), especially spectrophotometric titrations and activity measurements, and the ligand of cobalt is believed to be a hydroxide ion at this pH. This presents some uncertainty as to which protons are actually being relaxed at pH 8.5 since a tetracoordinate cobalt-bound hydroxide ion would not be expected to exchange rapidly with water in solution. This issue has been discussed by Koenig et al. (1983) who proposed that ligand exchange with solvent involved a pentacoordinate intermediate having both OH- and H20 as ligands. Proton exchange can be rapid between these two ligands so that the departing H20 can contain the oxygen of the initially-bound OH-. On the basis of 180-exchange kinetics, Tu and Silverman (1985) suggested that the NMR relaxation observed at pH 8.5 is that of a water molecule hydrogen bonded to the cobalt-bound hydroxide yet close enough to the metal to be strongly relaxed. Yet another possibility is the rapid exchange of the hydrogen of the metal-bound hydroxide without exchange of the oxygen. Because of these unresolved issues, I cannot state specifically which of the hydrogens near the metal are the main contributors to the observed fractionation factors.
The fractionation factors measured from T1 for each isozyme of Co(ll)substituted carbonic anhydrase were close to unity (Table 4-2), like 4T1 for Co(H20)62+ (Table 3-2), consistent with these values having a large contribution from the fractionation of outer shell water. This is reasonable from distance considerations, as it can be calculated using a 1/r6 dependence that the paramagnetic contribution to the longitudinal relaxation rate for a cobalt-bound hydroxide with a cobalt-proton distance of 2.8 A is about the same as the paramagnetic contribution of the protons of three water molecules at an average distance of 3.8 A. Also, it is known from crystallographic data that there





50


are a large number of water molecules within the active site cavity. Specifically, in human carbonic anhydrase II there are an estimated 9 water molecules between histidine 64 and the zinc, a distance of about 6 A from the x-ray coordinates (Erickson et al., 1986). The fractionation factors measured from T2, )T2, were near 0.72 for both Co(ll)-substituted isozymes I and II, but 1.0 for isozyme III (Table 4-2), and reflect a larger contribution of the hydrogen / deuterium fractionation of the inner shell aqueous ligand. It is difficult to account for these differences, however, and though these isozymes have very similar amino acid sequences, they have unique active site and kinetic properties (Silverman and Vincent, 1983; Silverman and Lindskog, 1988). Isozyme II has the highest CO2 hydration activity, with a turnover number of 106 s-1. There is a histidine in position 64 which protrudes into the active site cavity of isozyme II. Isozyme I has a turnover number of 2 x 10s5 s-1, and has His-64 as well as two other histidines, 67 and 200, in the active site cavity, and it is known that histidine 200 interacts strongly with the ligands of zinc (Kalifah, 1977). The active site cavity of isozyme I is smaller compared with that of isozyme II. Isozyme III has a turnover number of 104 s-1 and has a significant amount of positive charge in the active site, with a lysine in position 64 and an arginine in position 67. This positive charge may influence the ionization properties of the metal bound water in carbonic anhydrase III, for which the pK of the metalbound water is less than 5.5, compared with isozyme I and II which have a pK around 7 (Tu et al., 1983; Engberg and Lindskog, 1984). This difference in charge in the active site may also affect the hydrogen / deuterium fractionation properties of the ligand of the metal in Co(ll)-substituted carbonic anhydrase Ill. Although isozymes I and II have very similar amino acid sequences, the significant differences in their structure, residues in the active site cleft, and catalytic activities (Pocker and Sarkanen, 1978; Lindskog, 1983; Silverman and





51


Lindskog, 1988) apparently do not have a significant effect on the fractionation factors which were nearly identical for isozymes I and II.












CHAPTER 5
INTERPRETATION OF SOLVENT HYDROGEN ISOTOPE EFFECTS ASSOCIATED WITH CARBONIC ANHYDRASE
Introduction
A solvent hydrogen isotope effect on a reaction is made up of individual contributions to the overall isotope effect from each hydrogenic site in the reaction (Schowen, 1978). This contribution is measured by the hydrogen / deuterium fractionation factor of each particular hydrogenic site. In considering reactions of carbonic anhydrase, the fractionation factor of the aqueous ligand of the metal represents a reactant state contribution that can be used to interpret equilibrium solvent hydrogen isotope effects which are a function of reactant and product state site fractionation factors, and kinetic isotope effects which are a function of reactant and transition state site fractionation factors.
Halide ions are inhibitors of carbonic anhydrase which bind to the metal and displace the water ligand of the metal (Bertini et aL, 1982). I have determined the solvent hydrogen isotope effect on the equilibrium binding of iodide ion to Co(ll)-substituted carbonic anhydrase I and II and interpreted this effect using the fractionation factor of the aqueous ligand of cobalt in Co(ll)substituted carbonic anhydrase measured from NMR relaxation rates. The results suggest that the isotope effect on I- binding binding is the complex contribution of the fractionation of more than one hydrogenic position, because the consideration of the fractionation factor of the aqueous ligand of cobalt by itself does not allow a complete interpretation of the isotope effect.
The fractionation factor of the aqueous ligand of cobalt in Co(ll)substituted carbonic anhydrase can also be used to interpret kinetic isotope


52





53


effects, and because a kinetic isotope effect includes the contribution of transition state hydrogenic sites, this allows a prediction of the transition state structure. In the case of Co(ll)-substituted carbonic anhydrase II the fractionation factor measured from NMR leads to a predicted transition state structure that includes a proton transfer to a water molecule in the active site cavity. This is consistent with the large isotope effect on kat (Steiner et al., 1975) and the nonlinear dependence of kcat on the atom fraction of deuterium in solvent water (Venkatasubban and Silverman, 1980).
Experimental Procedure
The equilibrium dissociation constants describing the binding of iodide ion to Co(ll)-substituted carbonic anhydrase were determined from visible absorption measurements (Lindskog, 1966). The pH dependence of the absorbance at 640 nm in solutions of Co(ll)-substituted carbonic anhydrase I and II was measured on a Beckman DU-7 spectrophotometer. Stock solutions of enzyme at low pH contained 12.5 mM Hepes and 25 mM Mops (4morpholinepropanesulfonic acid) buffer, and at high pH 0.5 M of triethylenediamine. The stock solutions had equal concentrations of enzyme. The titrations were performed by adding small increments of the high pH enzyme solution to the low pH enzyme solution. In this way the enzyme concentration stayed the same though the volume increased slightly. These titrations were done in H20 and D20 in the presence and absence of iodide ion which has been shown to bind to the metal site in carbonic anhydrase II (Brown et al., 1977).
In the absence of anion, the data were fit (RS1, BBN Software, Cambridge, MA) to a scheme involving 4 micro equilibrium constants (Simonsson and Lindskog, 1982; Bertini et al., 1985). In this scheme, only 3 of the 4 microconstants are independent, that is K1K3 = K2K4 (Scheme 5-1). When





54




+HCoOH3 K3


+HECoOH2 ECoOH

1 4

ECoOH2
2






Scheme 5-1: A diagram of two significant ionizations at and near the active site of Co(ll)-substituted carbonic anhydrase II. One is the ionization of metal-bound water and the second is suggested to be the ionization of His-64 (Simonsson and Lindskog, 1982), the imidazole ring of which is about 6 A from the metal. For Co(ll)-substituted carbonic anhydrase I the second ionization could be either that of His-64 or His-200 (Whitney and Brandt, 1976).





55


anion was present, Scheme 5-1 was expanded to include the binding of the inhibitor, and it was assumed that the anion binds only to species 1 and 2 (see Tibell et al. (1984) for an estimate of the accuracy of this assumption). Thus two separate binding constants were determined. K1I1 is the equilibrium dissociation constant of the anion complex with species 1 of Scheme 5-1, and K21 is the equilibrium dissociation constant of the anion complex with species 2.

Results
In the absence of anion, the visible pH titration data fit Scheme 5-1 very well (Table 5-1). The constants K1, K2, and K3 were determined from a fit to the data, and then K4 was calculated from K1, K2, and K3. The 4 microconstants determined in H20 for Co(ll)-substituted carbonic anhydrase II agreed very well with those obtained by Simonsson and Lindskog (1982), and the values determined in H20 for Co(ll)-substituted carbonic anhydrase I also agreed fairly well with those obtained by Bertini et al. (1985). In the presence of iodide the pH dependence of the absorbance at 640 nm was shifted to higher pH. Iodide binds with greater affinity to the diprotonated species 1 in Scheme 5-1 than to the monoprotonated species 2 (Table 5-2). This is similar to the iodide inhibition results of Simonsson and Lindskog (1982) for native zinc carbonic anhydrase II determined from activity measurements, and the results of Bertini et al. (1985) for nitrate inhibition of Co(ll)-substituted carbonic anhydrase II determined from spectrophotometric titrations. The change in solvent from H20 to D20 has no measurable effect on K1l- and K21" for Co(ll)-substituted carbonic anhydrase I, but for Co(ll)-substituted carbonic anhydrase II the change in solvent from H20 to D20 results in a decrease in these equilibrium constants (Table 5-2).





56

Table 5-1: Solvent hydrogen isotope effects on the acid-base microequilibrium constants observed for Co(ll)-substituted carbonic anhydrase I and II.

Co(ll)-carbonic anhydrase I Co(ll)-carbonic anhydrase II
value in H20 ApKa value in H20 ApKa
pK1 7.55 0.05 0.54 0.06 5.65 0.04 0.42 0.05 pK2 7.58 0.15 0.45 0.19 5.73 0.07 0.37 0.10 pK3 8.11 0.15 0.45 0.19 6.95 + 0.08 0.38 + 0.12 pK4 8.09 0.22 0.54 0.28 6.87 0.12 0.43 0.17 Note: The microconstants were determined from the variation of the absorbance at 640 nm with pH at 230C and fit to Scheme 5-1. The solutions were buffered with Mops, Hepes and triethylenediamine with no other added ions as described in the experimental procedure. a ApK = (pKi)D20 (pKi)H2O





57

Table 5-2: Solvent hydrogen isotope effects on the equilibrium dissociation constants of iodide ion with Co(ll)-substituted carbonic anhydrase I and II.

Co(ll)-carbonic anhydrase I Co(ll)-carbonic anhydrase II

value in H20 ApKI- a value in H20 ApKI- a pK11I 2.95 0.10 -0.09 0.17 3.48 0.13 0.47 0.19 pK21I 2.35 0.06 -0.07 0.11 2.79 0.02 0.27 0.02 Note: The dissociation constants were determined from the pH dependence of the absorbance at 640 nm in the presence of 10 mM I- with conditions as in Table 5-1. K1- is the dissociation constant of I- binding to species 1 of Scheme 5-1, and K21I is the dissociation constant of I- binding to species 2 of Scheme 51.
a ApK = (pKi-)D20 (pKi')H2o





58


Discussion
I have discussed in Chapter 4 the fractionation factor for the
exchangeable hydrogen at the active site of Co(ll)-substituted carbonic anhydrase in its high pH form. The first application of the fractionation factors determined in Chapter 4 is to calculate the fractionation factor of the active site in its low pH form. Since the paramagnetic contribution to the relaxivity decreases to a very small value below pH 6.0 (Wells et al., 1979), except at very low ionic strength (Bertini et aL, 1978; Bertini et al., 1980), the fractionation factor cannot be measured directly from the relaxation rates at low pH. Instead, this was calculated from the fractionation factor of the high pH form and the solvent hydrogen isotope effect on the ionization constant of the active site, metal-bound water.
The visible pH titration data of isozymes I and II indicate that more than one ionization at or near the active site governs the absorbance at 640 nm (Bertini et al., 1980). Previous investigators (Simonsson and Lindskog, 1982; Bertini et al., 1985) have attributed this pH dependence to the ionization of two groups, the metal-bound water and another group near the active site which may be His-64 for isozyme II and probably His-64 or His-200 in isozyme I (Kalifah, 1977; Whitney and Brandt, 1976). I have measured the pH dependence of the absorbance at 640 nm for isozymes I and II in H20 and D20 and fit the data according to Scheme 5-1 to determine the solvent hydrogen isotope effect on the microequilibrium constants (Table 5-1). In Scheme 5-1, K1 and K4 are the constants for the metal-bound water, and K2 and K3 are the constants for the other active site group. Since the solvent hydrogen isotope effects on these equilibrium constants are all similar, within experimental uncertainty, it appears that the protonation state of one of the groups does not affect the hydrogen / deuterium fractionation properties of the other (consistent





59


with the conclusion that active site properties do not seem to influence the fractionation factors of the aqueous ligand of cobalt in Co(ll)-substituted carbonic anhydrase I and II, see Chapter 4). Then the calculation to determine the fractionation factor of the water ligand of the metal can be performed neglecting the ionization state of the second ionizable group.
The isotope effect on an equilibrium constant is the product of all of the reactant state fractionation factors divided by the product of all of the product state fractionation factors (equation 1-2). For this ionization at the active site of Co(ll)-substituted carbonic anhydrase 'Co'OH2 +H20 'CoOH + H30+ the solvent hydrogen isotope effect can be written
2 2
KHo (OCo-OH2) ( H203


From this measured isotope effect, the measured fractionation factor for the cobalt-bound hydroxide, the fractionation factor for H30+ which has been measured to be 0.69 (Chiang, 1980), and the fractionation factor for H20 which is 1.0, the fractionation factor for the exchangeable hydrogens at the active site in its low pH form can be calculated and are presented in Table 5-3. I have done the calculations of Table 5-3 using both OT, and T2 measured for the high pH form of the aqueous ligand of cobalt in Co(ll)-substituted carbonic anhydrase I and II. However, the calculation from T1 may not be meaningful, since it contains a large contribution of outer shell water. The value calculated using OT2 represents more closely the fractionation factor value of the inner shell cobalt-bound water.





60

Table 5-3: Values of the hydrogen / deuterium fractionation factor for the low pH forms of Co(ll)-substituted carbonic anhydrase.

OT1 OT2
Co(ll)-substituted carbonic anhydrase 1 1.12 0.05 0.90 0.04 Co(ll)-substituted carbonic anhydrase II 0.90 0.03 0.81 0.03 Note: The 0 for the low pH form was calculated using the isotope effect on the ionization constant of the cobalt-bound water and the 4 for the high pH form measured directly from NMR and given in Table 4-2.





61


To apply the fractionation factors to an equilibrium solvent hydrogen isotope effect associated with the enzyme, I have studied at equilibrium the binding of I- to Co(ll)-substituted carbonic anhydrase I and II. I- is known to bind to the metal in Co(ll)-substituted carbonic anhydrase I and II displacing the coordinated water molecule (Brown et aL, 1977). Also, I- binds fairly tightly (Table 5-2) and its size is comparable to that of a water molecule. Thus the isotope effect on the binding of I- to Co(ll)-substituted carbonic anhydrase may be simple enough to be accounted for by the fractionation factor of the metalbound water.
The solvent hydrogen isotope effect on the binding of the iodide ion to the active site of carbonic anhydrase is given by
-Co-I + H20 CoOH2 I


(K 10 (H20)2

(Ki )D20 (OCo-OH2)

where (Kil-)H20 is the equilibrium dissociation constant describing the binding of I- to species i of Scheme 5-1 in H20. Because it is known that 4H20 = 1.0, with knowledge of )Co-OH2 it should be possible to estimate an isotope effect. Using OT2 for )Co-OH2 from Table 5-3, I thus predict that ApKI- = 0.09 0.04 for Co(ll)isozyme I, and 0.18 0.03 for Co(ll)-isozyme II. The predicted value for isozyme I is in rough agreement with the measured value (Table 5-2), and the predicted value for isozyme II is slightly below the measured value. However, the predicted value for isozyme I is less than the predicted value for isozyme II, which is the same as in the experimental case. I have considered other interactions which may be taken into account for I- binding. One of these is the composite fractionation factor of the solvation shell of I- (4 = 0.59, Voice, 1974).





62


Including this in the denominator of the above equation for the isotope effect increases the predicted value, which is then closer to the experimental value for Co(ll)-isozyme II, but farther from the experimental value for Co(ll)-isozyme I. This implies that I have incomplete knowledge of the fractionation properties of each hydrogenic site that is affected when iodide ion binds to the enzyme. However, to include the fractionation factors of any other groups in the equation requires much speculation about the fractionation factors near the active site. Thus it is likely that the isotope effect must be described by the complex contribution of the fractionation factors of additional groups or water associated with the enzyme, and that the fractionation factor of the aqueous ligand of the metal is by itself not sufficient to account for the isotope effect on I- binding.
The discussion of the solvent hydrogen isotope effect on iodide binding suggests that the fractionation factor can be used to interpret at least qualitatively isotope effects associated with carbonic anhydrase. It is also possible to apply the fractionation factors for the aqueous ligand of the metal in Co(ll)-substituted carbonic anhydrase in the interpretation of isotope effects on kinetic constants associated with the enzyme to see if similar information can be obtained. A kinetic isotope effect is the product of the reactant state fractionation factors divided by the product of the transition state fractionation factors (equation 1-1). The isotope effect on kcat for the CO2 hydration activity of Co(ll)-substituted carbonic anhydrase II is 1.8 (R. S. Rowlett, unpublished), and the rate limiting step measured by kcat is the intramolecular transfer of a proton from the metal-bound water to regenerate the metal-bound hydroxide for the next round of catalysis (Steiner et al., 1975). This proton transfer from the metal-bound water is thought to go through a water-bridge to histidine-64 in the active site cleft, from which the proton can then be released to solvent





63


(Venkatasubban and Silverman, 1980). For this individual step in the catalysis, the reaction is

Co-OH2 + H20 + His Co-OH + H20 + His-H+


and the isotope effect is
kH20 ((Co-OH2) 2 (H20 )2

kD20 all transition state sites

iT

The hydrogenic sites of water have a fractionation factor of 1.0, and the hydrogenic sites of the cobalt-bound water have a fractionation factor of 0.81 (Table 5-3). Using these values and the isotope effect in the above equation, the product of the transition state fractionation factors can be calculated to be
0.36. In the transition state, the proton remaining on the metal-bound oxygen is in transition between a metal bound water which has a fractionation factor of
0.81 (Table 5-3), and a metal-bound hydroxide which has a fractionation factor of 0.77 (Table 4-2), so I will arbitrarily use a value of 0.79 as the fractionation factor of this hydrogenic site in the transition state. Therefore, the product of the fractionation factors of the other hydrogenic sites in the transition state is
0.36/0.79 = 0.46. These other hydrogenic sites in the transition state are the positions associated with the water bridge between the metal-bound water and the histidine in the active site cleft.





64




-Co-OH2 0 N 7NH




-Co-OH "H ..... 'NH

This calculation for the fractionation factor of the water bridge in the transition state suggests that the protons of the water bridge have a fractionation factor less than unity, meaning they have hydronium ion character, consistent with positive charge transferred to the oxygen during the proton transfer process. This calculation of the fractionation factor of the transition state suggests that the proton transferred from the cobalt-bound water is in between the cobalt-bound water and the water bridge in the transition state. As in the case of iodide binding to Co(ll)-substituted carbonic anhydrase, there may be other hydrogenic sites in the transition state which contribute to the overall isotope effect. However, the fractionation factors of the aqueous ligand of the metal have led to this qualitative model of the transition state, which is consistent with both the large isotope effect on kat and the results of Venkatasubban and Silverman (1980) for the catalysis of C02 hydration in H20 / D20 mixtures.












CHAPTER 6
MAGNETIC FIELD AND pH DEPENDENCE OF THE WATER
PROTON NMR RELAXATION ENHANCEMENT IN SOLUTIONS
OF Co(II)-SUBSTITUTED CARBONIC ANHYDRASE III Introduction
Carbonic anhydrase III is a mammalian isozyme of carbonic anhydrase found primarily in skeletal muscle. It is a less efficient catalyst of CO2 hydration compared with isozymes I and II, which are found in red blood cells (Lindskog, 1983). Another distinctive feature of carbonic anhydrase III is the pH independence of its catalytic activity in the range pH 6-9, in which carbonic anhydrase II has an inflection at pH 7 and maximal activity at high pH (Tu et al., 1983; Kararli and Silverman, 1985; Tu et al., 1986). It is also apparent that the zinc is bound more tightly in carbonic anhydrase III than in isozymes I and II (Engberg and Lindskog, 1984).
The replacement of cobalt for zinc in carbonic anhydrase I and II has
proved useful to study the properties of the metal at the active site (Bertini and Luchinat, 1983). The catalytic and visible spectral properties of Co(ll)substituted carbonic anhydrase III have been investigated (Engberg and Lindskog, 1984). I report here the NMR relaxation enhancement of water protons in solutions of Co(ll)-substituted carbonic anhydrase III. I found, in contrast to Co(ll)-substituted carbonic anhydrase I and II, no pH dependence of the NMR relaxation in the range of pH 6-9. Also, the frequency dependence of the paramagnetic contribution to the relaxation shows an increase with higher frequency, as opposed to the dispersion to lower relaxation rates observed in solutions of Co(ll)-substituted carbonic anhydrase I and II.


65





66


Experimental Procedure
The purification of bovine carbonic anhydrase III and preparation of Co(ll)-substituted carbonic anhydrase III was performed as described in Chapter 4. The total concentration of carbonic anhydrase III was estimated at 280 nm by using e = 6.4 x 104 M-1 cm-1 (Engberg and Lindskog, 1984). The concentration of Co(ll)-substituted carbonic anhydrase III was estimated at 640 nm by using the extinction coefficients determined by Engberg and Lindskog (1984). Zinc analysis of the native enzyme yielded 1.03 0.05 zinc per molecule of enzyme. Cobalt analysis of the Co(ll)-substituted enzyme yielded
0.33 0.04 cobalt per molecule of enzyme, with the remainder apoenzyme. The fraction of the sample that was apoenzyme was restored to full activity by the addition of zinc.
The longitudinal relaxation rates of the protons of solvent water in solutions of Co(ll)-substituted and native Zn-carbonic anhydrase III were measured using a field cycling "relaxometer" (Koenig et al., 1983) at the IBM T. J. Watson Research Center in Yorktown Heights, New York. The rates were measured at 50C at pH 6.0 (50 mM Mes/NaOH buffer), pH 7.0 (50 mM Mops/NaOH buffer), pH 8.0 (50 mM Hepes/NaOH buffer), and pH 9.0 (50 mM Ches/NaOH buffer) in solutions containing 1 mM of enzyme. The pH 6 samples were also measured at 250C. The samples were in equilibrium with atmospheric CO2. The total relaxivity, Rt, is defined as in equation 6-1

Rt = (1/T 1/T10)/N (6-1) where 1/T1 is the observed relaxation rate of the protein solution, 1/Tlo is the relaxation rate of the buffer, and N is the millimolar concentration of protein. At each pH the relaxivity was determined as a function of frequency over the range





67


0.01 to 50 MHz. The paramagnetic relaxivity, Rp, is Rt determined for a solution of Co(ll)-substituted carbonic anhydrase III minus Rt for a solution of native Zncarbonic anhydrase III.

Results
The frequency dependence of the total relaxivity was similar to the NMR dispersion profiles of other native and Co(ll)-substituted carbonic anhydrases (Fabry et al., 1970; Wells et al., 1979) (Figure 6-1). The total relaxivity is greatest at low frequency, then begins to decrease at about 0.1 MHz. The paramagnetic relaxivity, however, showed a small value at low frequency, then at about 0.5 MHz Rp began to increase to a maximal value at about 5 MHz, and then began to decrease again (Figure 6-2). The total relaxivities for both the Co(ll)-substituted and native Zn-carbonic anhydrase III showed a possible increase with pH at 1 MHz, but remained constant with pH at 40 MHz (Figure 63). This may be the same effect reported by Wells et al. (1979), who observed that the diamagnetic component of the relaxivity in solutions of Co(ll)-substituted and native zinc carbonic anhydrase II increases with pH at low frequency. When only the paramagnetic relaxivity was considered (the difference between Rt of Co(ll)-substituted carbonic anhydrase III and Rt of native zinc carbonic anhydrase III shown in Figure 6-3), there was no pH dependence, especially when the data at pH 7 in Figure 6-3 are excluded.

Discussion
This work is the first study of NMR relaxation enhancement of water
protons in solutions of Co(ll)-substituted carbonic anhydrase III. Previous work with isozyme III has shown that the visible spectrum of Co(ll)-substituted carbonic anhydrase III is very similar to the high pH spectral forms of Co(ll)substituted carbonic anhydrase I and II; however, for Co(ll)-substituted carbonic anhydrase III there is no pH dependence of the spectrum in the range pH 6-9





68












1.4

l) Q3

10 A A A
Ea Aa

.x 0.6 n


-A


0.2

-0.2 2 111, 1 fill, 0
10. 10-2 10-1 10o 101 102 103 frequency (MHz)

Figure 6-1: The total relaxivity of Co(ll)-substituted carbonic anhydrase III and native zinc carbonic anhydrase III as a function of the proton resonance frequency. The solutions contained 1 mM enzyme at pH 6.0 (50 mM Mes / NaOH), pH 8.0 (50 mM Hepes / NaOH), or pH 9.0 (50 mM Ches / NaOH). pH
6.0 Co(lI)-substituted carbonic anhydrase III at 50C ( a), pH 6.0 native zinc carbonic anhydrase III at 500C ( A), pH 9.0 Co(ll)-substituted carbonic anhydrase III at 50C (o), pH 9.0 native zinc carbonic anhydrase III at 50C ( 0), pH 8.0 Co(ll)-substituted carbonic anhydrase III at 230C (0 ), pH 8.0 native zinc carbonic anhydrase III at 230C (e).





69











-- 0.3


0.2


0.1 a 0.0cu 0.1


a. -0.2
10.3 10-2 10-1 100 101 102 frequency (MHz) Figure 6-2: The paramagnetic relaxivity of Co(ll)-substituted carbonic anhydrase III as a function of the proton resonance frequency. Experimental conditions as in Figure 6-1. pH 6.0 ( ), pH 9.0 ( ).





70











1.0


0.8


0.6


0.4


0.2
I-g

0.0
5.0 6.0 7.0 8.0 9.0 10.0
pH

Figure 6-3: The total relaxivity of Co(ll)-substituted carbonic anhydrase III and native zinc carbonic anhydrase III as a function of pH. Experimental conditions as in Figure 6-1. Co(ll)-substituted carbonic anhydrase III measured at 1 MHz
(A ), native zinc carbonic anhydrase III measured at 1 MHz (A ), Co(ll)substituted carbonic anhydrase III measured at 40 MHz (o), native zinc carbonic anhydrase III measured at 40 MHz ( m).





71


(Engberg and Lindskog, 1984). Also, the CO2 hydration activity of the cat isozyme III and the bovine isozyme III is independent in the pH range 6-8.5 (Tu et al., 1983; Engberg et al., 1985). These data imply that the activity controlling group in Co(ll)-substituted and native Zn-carbonic anhydrase III, the aqueous ligand of the metal, is predominantly in the hydroxide form in the pH range 6-9, and that the pK for the ionization of the metal-bound water to the metal-bound hydroxide is well below 6.
NMR relaxation enhancement studies with Co(ll)-substituted carbonic anhydrase I and II have shown a large pH dependence of the paramagnetic relaxation rate in the pH range 6-9, with an inflection at pH 7 and maximal relaxation at high pH (Fabry et al., 1970; Wells et al., 1979). This correlates with the pH dependence of both the visible spectrum and CO2 hydration activity (Bertini et al., 1982; Silverman and Vincent, 1983). Thus the NMR relaxation enhancement seems to correlate with the ionization state of the aqueous ligand of the metal in Co(ll)-substituted carbonic anhydrase I and II.
However, it has been shown that S042- present in solution to maintain ionic strength affects the pH dependence of both the visible spectrum and the paramagnetic relaxation of Co(ll)-substituted carbonic anhydrase I and II (Bertini et al., 1978; Bertini et al.,1980; Simonsson and Lindskog, 1982). When S042- and other anions are absent in either unbuffered solutions or solutions containing only sulfonic acid buffers, the visible spectrums of Co(ll)-substituted carbonic anhydrase I and II still show a large pH dependence, but there is practically no pH dependence on the paramagnetic contribution to the longitudinal relaxation rate of water protons in solutions Co(ll)-substituted carbonic anhydrase II (Bertini et al., 1978), and a small dependence for Co(ll)substituted carbonic anhydrase I (Bertini et al., 1980). The data in Figure 6-3 are from solutions containing no anions other than those of the sulfonic acid





72


buffer. Under these conditions, the relaxation of water protons in solutions of Co(ll)-substituted carbonic anhydrase II would be independent of pH in the range 6-9, and this is what was observed for the paramagnetic relaxivity of solutions of Co(ll)-substituted carbonic anhydrase III (Figure 6-3).
The reason for the small value of Rp at low frequency (Figure 6-2) is not clear. One possibility is that the lifetime of a proton in the hydration shell of the cobalt is too long to allow for a large paramagnetic contribution (Fabry et al., 1970; Tu et al., 1985). This water off rate has a maximal value of 104 S-1 in the case of native carbonic anhydrase III (Tu et aL, 1983), but is 10s s-1 for Co(ll)substituted carbonic anhydrase II (Tu et al., 1985). This rate would be expected to be even slower at 50C, but when the pH 6 sample was measured at 25 OC, there was a larger paramagnetic relaxivity at low frequency (data not shown). Another possibility is that in this isozyme the cobalt is in a pentacoordinate or hexacoordinate environment (Engberg and Lindskog, 1984), which would have less ability to relax due to a shorter electron relaxation time (Koenig et al., 1983). Whatever the reason, the paramagnetic contribution at low frequency is small.
The increase in Rp with higher frequency (Figure 2) is unlike that for any other Co(ll)-substituted carbonic anhydrase which show a decrease in Rp with higher frequency (Wells et al., 1979), but is not unlike that observed for Mn, Cu, and Ni proteins (Koenig and Brown, 1973; Bertini and Luchinat, 1986). While the relaxation behavior of these systems is not completely understood, it is clear that the field dependence of the electron spin relaxation is a factor. Thus, the increase in the electron relaxation time may be responsible for the increase in Rp at intermediate frequency. The decrease at higher frequency may be due to the o(sTc dispersion predicted by the Solomon-Bloembergen equations (Bertini and Luchinat, 1986).












CHAPTER 7
CONCLUSIONS
This work shows the utility of the NMR relaxation method to measure the hydrogen / deuterium fractionation properties of the aqueous ligands of cobalt in some specific cases. In particular, the characterization of the relaxation mechanisms at 300 MHz has allowed the detection of the fractionation of hydrogen and deuterium of two populations of ligand sites associated with cobalt, one having a significant contribution of the inner shell aqueous ligand of the metal and the other apparently dominated by outer shell water. The fractionation factor for the inner shell ligand can differ significantly from that for solvent water, and in the case of Co(H20)62+, 4T2 is near the fractionation factor of H3O+, implying that the water ligands of cobalt in Co(H20)62+ are influenced by the positive charge of the cobalt. The fractionation factors of Co(ll)substituted carbonic anhydrase represent the first direct measurement of this parameter at the active site of an enzyme.
The conclusions of the contributions of inner and outer shell water to the relaxation times T1 and T2 result from the effect of the chemical shift relaxation mechanism at 300 MHz. This conclusion is supported by calculations and experimental evidence with EDTA for Co(H20)62+. In this hexaaquacomplex the only interactions are those of the metal and water, and the water can be in the inner shell metal-bound site, an outer shell site, or the bulk solvent site. There are no other effects of the medium in this system. In the case of Co(ll)substituted carbonic anhydrase, there are many other interactions which may occur to affect the fractionation of the hydrogenic positions of the aqueous



73





74


ligand of the metal. These interactions include influences of charge by residues near the active site, and the effects of the hydrogen bond network that exists in the active site cavity. This hydrogen bond network includes water in the active site cavity and specific residues from the enzyme, as well as the aqueous ligand of the metal. Thus it is clear that outer shell water in Co(ll)-substituted carbonic anhydrase would have some different properties than outer shell water in Co(H20)62+. The relaxation rates at 300 MHz of water protons in solution of Co(ll)-substituted carbonic anhydrase showed T1 much larger than T2, as in Co(H20)62+, and this indicates that the chemical shift mechanism affects T2 in solutions of Co(ll)-substituted carbonic anhydrase. However there is no other support for this from calculations or other experiments, as in the case of Co(H20)62+, except that the fractionation factor determined from T1 and T2 are also different for Co(ll)-substituted carbonic anhydrase. This inference of the chemical shift mechanism affecting T2 in solutions of CO(II)-substituted carbonic anhydrase is the only extrapolation made from Co(H20)62+ to Co(ll)-substituted carbonic anhydrase, which are very different systems in many other respects. It is important to note that the fractionation factors for Co(ll)-substituted carbonic anhydrase are measured directly, and thus are independent of any other interpretation regarding Co(H20)62+, except that of the contribution of inner shell water to T2 and of outer shell water to T1. Thus the fractionation factors for Co(ll)-substituted carbonic anhydrase include the effects of the interactions near the active site.
One very unique opportunity to understand the basis for the differences in the fractionation factors between the isozymes of carbonic anhydrase exists in the study of site-directed mutants of the enzyme. It is known that the pK for the ionization of the metal-bound water in carbonic anhydrase III is lower than that ionization in carbonic anhydrase I and II. This difference is most likely due





75


to the positively charged residues near the active site in carbonic anhydrase ill. The fractionation factor for the metal-bound hydroxide in carbonic anhydrase Ill was greater than that fractionation factor in carbonic anhydrase I and II, and this may also be due to the positively charged residues in carbonic anhydrase III. This could be tested, for instance, by replacing the histidine in position-64 of carbonic anhydrase I and II with a lysine, which occupies position-64 in carbonic anhydrase III, and then measuring the fractionation factor using the cobalt-substituted enzyme. One would predict that the fractionation factor of Co(ll)-substituted carbonic anhydrase I and II with lysine in position-64 would be closer to unity, like that in Co(ll)-substituted carbonic anhydrase III.
The value of the fractionation factor for Co(H2O)62+ measured directly from NMR represents a reactant state fractionation factor that can be used to interpret solvent hydrogen isotope effects associated with Co(H20)62+. The solvent hydrogen isotope effects on the formation of complexes of cobalt with some bidentate ligands suggest that the fractionation factor of some other hydrogenic positions must contribute to the overall isotope effect, most likely the positions of outer shell water in these complexes, because the isotope effect could not be accounted for by considering only the fractionation factor of Co(H20)62+. This may be because of the effect of the bidentate ligands on the structure of outer shell water in these complexes. It could be that the fractionation factor of the outer shell water in a cobalt-bidentate ligand complex is different than that fractionation factor in Co(H20)62+. Also, it may be that the number of outer shell waters that experience the paramagnetic relaxation enhancement by the cobalt is less in the cobalt-bidentate ligand complex than in Co(H20)62+. However, it is clear that there would be a large number of outer shell waters associated with complexes, and that a small effect on the fractionation of the outer shell water in these complexes on going from the





76


reactant to the product state, when raised to a large power for the number of hydrogenic positions in the outer shell water, could have a large contribution to the overall isotope effect.
The values of the fractionation factor of the aqueous ligand of the metal at the active site of Co(ll)-substituted carbonic anhydrase can also be used to interpret solvent hydrogen isotope effects on reactions associated with carbonic anhydrase. The solvent hydrogen isotope effect on the equilibrium binding of iodide ion to Co(ll)-substituted carbonic anhydrase suggests that there must be other hydrogenic positions whose isotope preference changes on going from the reactant to product states, because the consideration of the fractionation factor for the aqueous ligand of cobalt in Co(ll)-substituted carbonic anhydrase by itself does not allow a complete interpretation of the isotope effect. However, it is clear from the results that the hydrogenic positions of the metal-bound water at the active site do make a large contribution to the overall isotope effect. This means that the other hydrogenic positions that contribute to the isotope effect probably have a fractionation factor just slightly different from unity, but the large number of these positions would then give a significant contribution to the overall isotope effect. The solvent hydrogen isotope effects for other anions binding to Co(ll)-substituted carbonic anhydrase could be determined, and even for other inhibitors of the enzyme that are known to bind to the active site. This may give more information about these other hydrogenic positions. For instance, it may be that the fractionation factor of the hydration shell of the anion is important, and this could be determined from the results with other monovalent anions. Also, bulkier groups could be used, such as the sulfonamide inhibitors, and these may displace more outer shell water within the active site cavity. The isotope effect on the binding of a sulfonamide





77


inhibitor may then give some insight into the contribution of these outer shell hydrogenic positions to the overall isotope effect.
To account for the solvent hydrogen isotope effects on a rate constant is even more complex because knowledge of the fractionation factor values for the transition state are required. The model for the transition state of the ratelimiting proton transfer step of the CO02 hydration reaction, developed from the fractionation factor of the aqueous ligand of the cobalt in Co(ll)-substituted carbonic anhydrase II and the isotope effect on kcat, is consistent with other experimental evidence for the structure of the transition state. Other kinetic isotope effects on reactions of carbonic anhydrase may then also be used with the fractionation factor to get information about the mechanisms of these reactions. For instance, the solvent hydrogen isotope effects on the binding of anions could be measured, and then interpreted with the fractionation factors of the reactant state. This may show whether the rate-limiting step of the binding is the initial entry of the anion into the active site cavity, or it may be that the final binding of the anion to the metal is the rate-limiting step. This work represents an important first step in the accurate interpretation of the solvent hydrogen isotope effects in the catalysis by carbonic anhydrase.
Perhaps the most significant achievement in this work is the development of the NMR relaxation method to measure directly the hydrogen / deuterium fractionation factor of the hydrogenic positions of water coordinated to metals. This method could also be used with other transition metals and in other types of inorganic complexes, and in other metalloenzymes. Also, the conclusions of the contributions of inner and outer shell water to the relaxation times T1 and T2 could also be studied further by, for instance, determining the effects of ionic strength and temperature and the relaxation and the fractionation factors





78

determined from T1 and T2. It is hoped that the work presented in this dissertation will be the catalyst for other studies in this field.













Referee nces

Baes, N. F.; Mesmer, R. E. The Hydrolysis of Cations; Wiley: New York, 1976. Bernheim, R. A.; Brown, T. H.; Gutowsky, H. S.; Woessner, D. E. J. Chem Phys. 1959, 30, 950-956.

Bertini, I.; Canti, G.; Luchinat, C.; Scozzafava, A. J. Am Chem Soc. 1978, 100, 4873-4877.

Bertini, I.; Dei, A.; Luchinat, C.; Monnanni, R. Inorg. Chem. 1985, 24, 301-303. Bertini, I.; Luchinat, C. Acc. Chem. Res. 1983, 16, 272-279. Bertini, I.; Luchinat, C. NMR of Paramagnetic Species in Biological Systems; Benjamin-Cummings: Menlo Park, CA, 1986. Bertini, I.; Luchinat, C.; Scozzafava, A.; Inorg. Chem. Acta 1980, 46, 85-89. Bertini, I.; Luchinat, C.; Scozzafava, A. Struct. Bonding (Berlin) 1982, 48, 45-92. Bloembergen, N. J. Chem Phys. 1957, 27, 572-573. Bloembergen, N.; Morgan, L. O. J. Chem. Phys. 1961, 34, 842. Brown, G. S.; Navon, G.; Shulman, R. G. Proc. Natl. Acad. Sci. USA 1977, 74, 1794-1797.

Carlson, G. A.; McReynolds, J. P.; Verhoek, F. H. J. Am. Chem. Soc. 1945, 67, 1334-1339.

Chiang, Y.; Kresge, A. J.; More O'Ferrall, R. A. J. C. S. Perkin 111980, 18321839.

Dwek, R. A. NMR in Biochemistry; Oxford press: Oxford, 1975. Eigen, M.; Tamm, K. Z. Elektrochem. 1962, 66, 93-121. Eisenstadt, M. J. Chem Phys. 1969, 51, 4421-4432. Engberg, P.; Lindskog, S. FEBS Lett. 1984, 170, 326.




79





80

Engberg, P.; Millqvist, E.; Pohl, G.; Lindskog, S. Arch. Biochem. Biophys. 1985, 241, 628-638.

Eriksson, E. A.; Jones, T. A.; Liljas, A. In Zinc Enzymes; Bertini, I.; Luchinat, C.; Maret, W.; Zeppezauer, M. Eds.; Birkhauser: Boston, 1986; 317-328. Fabry, M. E.; Koenig, S. H.; Schillinger, W. E. J. Biol. Chem. 1970, 245, 42564262.

Freeman, R.; Kempsell, S. P.; Levitt, J. J. Magn. Reson. 1980, 138, 453-479. Friedman, H. L.; Holz, M.; Hertz, H. G. J. Chem Phys. 1979, 70, 3369. Glasoe, P. K.; Long, P. A. J. Phys. Chem. 1960, 64, 188-191. Gold, V.; Lowe, B. M. J. Chem. Soc. (A) 1968, 1923-1932. Gold, V.; Wood, D. L. J. Chem. Soc., Dalton Trans. 1982, 2452-2461. Goodall, D. M.; Long, F. A. J. Am. Chem. Soc. 1968, 90, 238. Gueron, M. J. Magn. Reson. 1975, 19, 58. Hammes, G. C.; Steinfeld, J. I. J. Am. Chem. Soc. 1962, 84, 4639-4643. Hunt, J. B.; Rhee, M. J.; Storm, C. B. Anal. Biochem. 1977, 79, 614-617. Jencks, W. P.; Salvesen, K. J. Am Chem Soc. 1971, 93, 4433. Kalifah, R. Biochemistry 1977, 16, 2236-2240. Kalifah, R, G.; Strader, D. J.; Bryant, S. H.; Gibson, S. M. Biochemistry 1977, 16, 2241-2247.

Kararli, T.; Silverman, D. N. J. Biol. Chem. 1985, 260, 3484-3489. Koenig, S. H.; Brown, R. D. Ann. N. Y. Acad. Sci. 1973, 222, 752-763.

Koenig, S. H.; Brown, R. D.; Bertini, I.; Luchinat, C. Biophys. J. 1983, 41, 179187.

Kresge, A. J.; More O'Ferrall, R. A.; Powell, M. F. In Isotopes in Organic Chemistry, Volume 7; Buncel, E.; Lee, C. Eds.; Elsevier: New York, 1987; 177273.

Laughton, P. M.; Robertson, R. E. In Solute-Solvent Interactions; Coetzee, J. F.; Ritchie, C. D. Eds.; Marcel Dekker: New York, 1969; 399-538. Lindskog, S. Biochemistry 1966, 5, 2641-2646.





81


Lindskog, S. in Zinc Enzymes; Spiro, T. G. Ed.; Wiley: New York; 1983; 78-121. Lowe, B. M.; Smith, D. G. J. Chem. Soc., Faraday Trans. 11975, 71, 389-397. Luz, Z.; Meiboom, S. J. Chem. Phys. 1964, 40, 2686-2692. Meiboom, S.; Gill, D. Rev. Sci Instrum. 1958, 29, 688. Melamud, E.; Mildvan, A. S. J. Biol. Chem. 1975, 250, 8193-8201. Melton, B. F.; Pollack, V. L. J. Phys. Chem. 1969, 73, 3669-3679. More O'Ferrall, R. A.; Koeppl, G. W.; Kresge, A. J. Am. Chem. Soc. 1971, 93, 920.

Morgan, L. O.; Nolle, A. W. J. Chem Phys. 1959, 31, 365. Nyman, P. O.; Lindskog, S. Biochim. Biophys. Acta 1964, 85, 141-151. Pocker, Y.; Sarkanen, S. Adv. Enzymol. Relat. Areas Mol. Biol. 1978, 47, 149274.

Schowen, K. B. J. In Transition States in Biochemical Processes; Grandour, R. D.; Schowen, R. L. Eds.; Plenum: New York, 1978; 225-284. Schowen, K. B.; Schowen, R. L. In Methods in Enzymology, Volume 87; Purich, D. L. Ed.; Academic: New York, 1982; 551-606. Silverman, D. N. J. Am. Chem. Soc. 1981, 103, 6242-6243. Silverman, D. N.; Lindskog, S. Acc. Chem. Res. 1988, 21, 30-36. Silverman, D. N.; Vincent, S. H. CRC Crit. Rev. Biochem. 1983, 14, 207-255. Simonsson, I.; Lindskog, S. Eur. J. Biochem. 1982, 123, 29-36. Smith, R. M.; Martell, A. E. Critical Stability Constants; Plenum: New York, 1975. Solomon, I. Phys. Rev. 1955, 99, 559-565. Steiner, H.; Jonsson, B.; Lindskog, S. Eur. J. Biochem. 1975, 59, 253-259. Swift, T. J.; Connick, R. E. J. Chem. Phys. 1962, 37, 307-320; J. Chem. Phys.
1964, 41, 2553-2554.

Szawelski, R. J.; Wharton, C. W.; White, S. Biochem. Soc. Trans. 1982, 10, 232233.





82

Tibell, L.; Forsman, C.; Simonsson, I.; Lindskog, S. Biochim. Biophys. Acta 1984, 789, 302-310.

Tu, C. K.; Sanyal, G.; Wynns, G. C.; Silverman, D. N. J. Biol. Chem. 1983, 258, 8867-8871.

Tu, C. K.; Silverman, D. N. Biochemistry 1985, 24, 5881-5887. Tu, C. K.; Thomas, H. G.; Wynns, G. C.; Silverman, D. N. J. Biol. Chem. 1986,
261, 10100.

Venkatasubban, K. S.; Silverman, D. N. Biochemistry 1980, 19, 4984. Voice, P.J. J. C. S. Faraday 11974, 70, 498-505. Wells, J. W.; Kandel, S. I.; Koenig, S. H. Biochemistry 1979, 18, 1989-1995. Whitney, P. L.; Brandt, H. J. Biol. Chem. 1976, 251, 3862-3867.





83






BIOGRAPHICAL SKETCH
James W. Kassebaum, the son of Amanda L. Schoonmaker and Wilbur F. Kassebaum, was born March 6, 1962, in Mount Vernon, Illinois. James grew up in St. Louis, Missouri, and in 1980 graduated first in his class at Parkway South Senior High School. He then attended Purdue University, West Lafayette, Indiana, and in 1984 graduated with Honors with the degree of Bachelor of Science in chemistry. The summer of 1984 was a very eventful one, when James was wed to Shari L. Gunderman and then entered graduate school at the University of Florida in the Department of Biochemistry and Molecular Biology and studied under the supervision of Dr. David Silverman. James completed the requirements for the Ph.D. degree in 1988, and began his professional career at the Monsanto Company in St. Louis.














I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy.


David N. Silverman, Chair Professor of Pharmacology and Therapeutics

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy.


E. Raymond Andrew Graduate Research Professor of Physics


I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy.



Robert J/Cohen
Associate Professor of Biochemistry and Molecular Biology


I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doc r f Philosophy.



Russell S. Drago Graduate Research Professor of Chemistry









I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philos.


Dniel L. Purich
Professor of Biochemistry and
Molecular Biology

This dissertation was submitted to the Graduate Faculty of the College of Medicine and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy.


December, 1988 __ .-__ '_ _.zMan, College of Medicine



Dean, Grduate School




Full Text
29
Table 3-3: Logarithm of the formation constants determined for the reactions of
Co2+ with the ligands glycine, A/,A/-dimethylglycine, and acetylacetone in H2O
and D2O.
h2o
D20
(log kJop-ilogkjHjO
glycine
log ki
5.04 0.04
4.99 0.02
-0.05 0.04
log k2
4.32 0.04
4.23 0.02
-0.08 0.05
A/,A/-dimethylglycine
log ki
4.14 0.04
4.12 0.06
-0.03 0.07
log k2
3.20 0.06
3.18 0.13
-0.01 0.14
acetylacetone
log ki
5.30 0.07
5.42 0.10
0.12 0.13
log k2
4.40 0.09
4.61 0.12
0.21 0.15
Note: Determined by potentiometric titrations of solutions which contained 1
mM CoCI2 and 4 mM of the ligand and fitting the data to equation 3-6.
Measurements were made at 23 C. See equation 3-5 for the definition of ki.


50
are a large number of water molecules within the active site cavity. Specifically,
in human carbonic anhydrase II there are an estimated 9 water molecules
between histidine 64 and the zinc, a distance of about 6 from the x-ray
coordinates (Erickson et al., 1986). The fractionation factors measured from T2,
isozyme III (Table 4-2), and reflect a larger contribution of the hydrogen /
deuterium fractionation of the inner shell aqueous ligand. It is difficult to
account for these differences, however, and though these isozymes have very
similar amino acid sequences, they have unique active site and kinetic
properties (Silverman and Vincent, 1983; Silverman and Lindskog, 1988).
Isozyme II has the highest C02 hydration activity, with a turnover number of 106
S'1. There is a histidine in position 64 which protrudes into the active site cavity
of isozyme II. Isozyme I has a turnover number of 2 x 105 S'1, and has His-64 as
well as two other histidines, 67 and 200, in the active site cavity, and it is known
that histidine 200 interacts strongly with the ligands of zinc (Kalifah, 1977). The
active site cavity of isozyme I is smaller compared with that of isozyme II.
Isozyme III has a turnover number of 104 s*1 and has a significant amount of
positive charge in the active site, with a lysine in position 64 and an arginine in
position 67. This positive charge may influence the ionization properties of the
metal bound water in carbonic anhydrase III, for which the pK of the metal-
bound water is less than 5.5, compared with isozyme I and II which have a pK
around 7 (Tu etai, 1983; Engberg and Lindskog, 1984). This difference in
charge in the active site may also affect the hydrogen / deuterium fractionation
properties of the ligand of the metal in Co(ll)-substituted carbonic anhydrase III.
Although isozymes I and II have very similar amino acid sequences, the
significant differences in their structure, residues in the active site cleft, and
catalytic activities (Pocker and Sarkanen, 1978; Lindskog, 1983; Silverman and


ligand of the metal. These interactions Include influences of charge by residues
near the active site, and the effects of the hydrogen bond network that exists in
the active site cavity. This hydrogen bond network includes water in the active
site cavity and specific residues from the enzyme, as well as the aqueous ligand
of the metal. Thus it is clear that outer shell water in Co(ll)-substituted carbonic
anhydrase would have some different properties than outer shell water in
Co(H20)62+. The relaxation rates at 300 MHz of water protons in solution of
Co(ll)-substituted carbonic anhydrase showed T-i much larger than T2, as in
Co(H20)62+, and this indicates that the chemical shift mechanism affects T2 in
solutions of Co(ll)-substituted carbonic anhydrase. However there is no other
support for this from calculations or other experiments, as in the case of
Co(H20)62+, except that the fractionation factor determined from T-i and T2 are
also different for Co(ll)-substituted carbonic anhydrase. This inference of the
chemical shift mechanism affecting T2 in solutions of Co(ll)-substituted carbonic
anhydrase is the only extrapolation made from Co(H20)62+ to Co(ll)-substituted
carbonic anhydrase, which are very different systems in many other respects. It
is important to note that the fractionation factors for Co(ll)-substituted carbonic
anhydrase are measured directly, and thus are independent of any other
interpretation regarding Co(H20)62+, except that of the contribution of inner shell
water to T2 and of outer shell water to T^ Thus the fractionation factors for
Co(ll)-substituted carbonic anhydrase include the effects of the interactions near
the active site.
One very unique opportunity to understand the basis for the differences
in the fractionation factors between the isozymes of carbonic anhydrase exists
in the study of site-directed mutants of the enzyme. It is known that the pK for
the ionization of the metal-bound water in carbonic anhydrase III is lower than
that ionization in carbonic anhydrase I and II. This difference is most likely due


28
Table 3-2: Fractionation factors determined from T-i and T2 at 300 MHz for the
aqueous ligands of the paramagnetic metal in Co(H20)62+ and Ni(H20)62+.
Co(H20)62+
Ni(H20)62+
^ <>t2
0.95.01 0.73.02
0.99.01 0.98.03
Note: Measurements were made at 23 C for 10 mM CoCI2 or NSO4 containing
5 x 10'4 M acetic acid buffer at pH 4.4. Data were obtained from the least
squares slope and intercept and standard error in a plot such as Fig. 3-1
according to equation 3-4.


2 4
Table 3-1: Solvent hydrogen isotope effects on the acid ionization constants of
glycine, A/,A/-dimethylglycine, and acetylacetone.
(pKi)H2Oa
ApK!
(pK2)H20 b
ApK2
glycine
2.37 0.02
0.40 0.04
9.72 0.03
0.56 0.05
A/,/\/-dimethylglycine
1.96 0.03
0.40 0.04
9.81 0.03
0.55 0.05
acetylacetone
9.03 0.10
0.62 0.17
-
-
Note: Acid ionization constants were determined in H2O and D2O by
potentiometric titrations at 23C. Solutions contained 10 mM of either glycine,
A/,A/-dimethylglycine, or acetylacetone.
a pK = -log K where K is the acid ionization constant. Standard errors resulted
from a non-linear least-squares fit of the titration data. ApK = pKo2o pKh2o-


58
Discussion
I have discussed in Chapter 4 the fractionation factor for the
exchangeable hydrogen at the active site of Co(ll)-substituted carbonic
anhydrase in its high pH form. The first application of the fractionation factors
determined in Chapter 4 is to calculate the fractionation factor of the active site
in its low pH form. Since the paramagnetic contribution to the relaxivity
decreases to a very small value below pH 6.0 (Wells et ai, 1979), except at very
low ionic strength (Bertini eta!., 1978; Bertini et ai, 1980), the fractionation
factor cannot be measured directly from the relaxation rates at low pH. Instead,
this was calculated from the fractionation factor of the high pH form and the
solvent hydrogen isotope effect on the ionization constant of the active site,
metal-bound water.
The visible pH titration data of isozymes I and II indicate that more than
one ionization at or near the active site governs the absorbance at 640 nm
(Bertini et ai, 1980). Previous investigators (Simonsson and Lindskog, 1982;
Bertini et ai, 1985) have attributed this pH dependence to the ionization of two
groups, the metal-bound water and another group near the active site which
may be His-64 for isozyme II and probably His-64 or His-200 in isozyme I
(Kalifah, 1977; Whitney and Brandt, 1976). I have measured the pH
dependence of the absorbance at 640 nm for isozymes I and II in H2O and D2O
and fit the data according to Scheme 5-1 to determine the solvent hydrogen
isotope effect on the microequilibrium constants (Table 5-1). In Scheme 5-1, K-i
and K4 are the constants for the metal-bound water, and 2 and K3 are the
constants for the other active site group. Since the solvent hydrogen isotope
effects on these equilibrium constants are all similar, within experimental
uncertainty, it appears that the protonation state of one of the groups does not
affect the hydrogen / deuterium fractionation properties of the other (consistent


34
coordination shell, suggesting that (j)j2 is measuring more of the fractionation of
inner shell water. Since T-ip has a larger contribution of outer shell water, (¡jj-i is
more of the fractionation of outer shell water. The analysis of the relaxation for
Ni(H20)62+ (Chapter 2) suggests that both Ti and T2 have a large contribution of
outer shell water, so contribution of the fractionation factor of outer shell water, and the influence of
the fractionation of the inner shell water is not measured by or (j)j2.
This interpretation of the fractionation factor for Co(H20)62+ is consistent
with expectations for the fractionation factor of metal-bound water. The
fractionation factor for H30+ is 0.69 (Chiang et al., 1980), presumably due to the
positive charge on the oxygen. It has been suggested that a metal-bound water
site ought to have a fractionation factor approximated by (0.69)8, with 5
measuring the degree of positive charge transferred from the metal to the
oxygen (Schowen and Schowen, 1982). The fractionation factor from T2 for
Co(H20)62+ being 0.73 supports this suggestion in resembling the fractionation
factor for H30. Pure water has by definition a fractionation factor of 1.0. Outer
shell water, being much more loosely bound than inner shell water, should be
much more like bulk water in its hydrogen / deuterium fractionation properties
(More O'Ferrall et al., 1971). The observation of <|)t1 for Co(H20)62+ and (j>r1
and ()>t2 for Ni(H20)62+ all being close to unity is consistent with these
fractionation factors having a large contribution of the fractionation of outer shell
water. The fractionation factor of the inner shell water in Ni(H20)62+ may well be
affected by the positive charge of the nickel and have a value closer to 0.69, or it
may have a value near unity; however, because of the large contribution of
outer shell water to both Ti and T2 in solutions of Ni(H20)62+, the influence of
the inner shell water is not large in the fractionation factors determined from
and T2.


3 2
3-7 with glycine, A/,A/-dimethylglycine, and acetylacetone are in Table 3-4 using
(|)Ti and (¡)t2 for Co(H20)62+.
I have also tried to measure the fractionation factor for these coordinated
water molecules in the complexes directly, using the NMR relaxation method
used to determine the fractionation factor of Co(H20)62+. However, these
solutions of Co2+ and a bidentate ligand L consist of an equilibrium mixture of
Co(H20)62+ Co(H20)4L+ Co(H20)2L2 and CoL3 and the observed
paramagnetic relaxation is the sum of the relaxation of each of these four
species. This presents difficulties in obtaining the fractionation factor of one
species from these measurements.
Discussion
The definition of the hydrogen / deuterium fractionation factor I have
given (equation 3-1) assumes the rule of the geometric mean to be valid
(Schowen and Schowen, 1982); that is, the fractionation factor of one particular
hydrogenic site is independent of the isotopic composition of any other site. I
will assume that the rule is valid in my system, and this is supported by the high
degree of linearity in plots as in Figures 3-1 and 3-2. A deviation from the rule
of the geometric mean would result in a non-linear dependence of the
relaxation data on the atom fraction of deuterium in the solvent.
The value of the fractionation factor for the water ligands of cobalt in
Co(H20)62+ was dependent on whether it was determined from Ti or T2, while
that of Ni(H20)62+ was independent of or T2 (Table 3-2). These fractionation
factors are a weighted value of all the protons that experience the paramagnetic
relaxation. An explanation for this difference in (})t1 and (|)t2 is based on the
analysis of the contributions to the relaxation at 300 MHz (see Chapter 2). The
paramagnetic contribution to the transverse relaxation of water protons in
solutions of Co(H20)62+ has a predominant contribution of water in the inner


8
mechanisms operating in these systems at 300 MHz need to be understood
before the relaxation can be used to measure a fractionation factor, and this is
discussed in Chapter 2. Chapter 3 includes the determination of the
fractionation factor of Co(H20)62+ and Ni(H20)62+ along with measurements of
the solvent hydrogen isotope effect on the formation constants of some cobalt
complexes to see if the fractionation factor of Co(H20)62+ can account for any of
these effects. In Chapter 4 I have measured the fractionation factor of three
mammalian isozymes of Co(ll)-substituted carbonic anhydrase using NMR
relaxation, and in Chapter 5 I discuss the use of these fractionation factors to
interpret solvent hydrogen isotope effects on the equilibrium binding of an
inhibitor to carbonic anhydrase and on the catalysis. Finally, the relaxation
properties of Co(ll)-substituted carbonic anhydrase III have not previously been
reported, and in Chapter 6 I discuss experiments to determine the magnetic
field and pH dependence of the water proton relaxation enhancement in
solutions of Co(ll)-substituted carbonic anhydrase III.


22
shell water also makes a large contribution to the Isotope effect on these
reactions.
Experimental Procedure
NMR Measurements: Measurements at 300 MHz and solutions were the
same as in Chapter 2. Some measurements of Ti of water protons in solutions
of Co2+ were performed at frequencies between 0.01 and 50 MHz on a field
cycling "relaxometer" at the IBM T. J. Watson Research Center, Yorktown
Heights, NY (Koenig etal., 1983).
Measurement of the Hydrogen / Deuterium Fractionation Factor: The
measurement of the relaxation of water protons in solutions of varying
deuterium content and containing paramagnetic metal ions is given by
1/T¡p = p'[X]/(1 n + n<)>) (3-3)
where n is the atom fraction of deuterium in solvent water, (j) is the hydrogen /
deuterium fractionation factor of the water ligands of the metal, and [X] is the
term in brackets in equations 2-2 and 2-3 for i = 1 or 2, respectively. This
equation is derived considering the fractionation of isotopes that arises when
there is a deuterium content in water. In this case the ratio of protons in the
hydration shell of the metal to total protons in solution is p'/(1-n+n<})) (Schowen
and Schowen, 1982). Equation 3-3 can be rearranged to
P'Tjp = [X]1 n(1- Thus pTjp vs n gives a slope of -(1-)[X]-1 and an intercept of [X]*1, and (j) = 1 +
slope / intercept.


Experimental Procedure 53
Results 55
Discussion 58
6 MAGNETIC FIELD AND pH DEPENDENCE OF THE WATER
PROTON NMR RELAXATION ENHANCEMENT IN
SOLUTIONS OF Co(ll)-SUBSTITUTED CARBONIC
ANHYDRASE III 65
Introduction 65
Experimental Procedure 66
Results 67
Discussion 67
7 CONCLUSIONS 73
REFERENCES 79
BIOGRAPHICAL SKETCH 83
i v


4
hydrogen bound to sulfur in mercaptoethanol to be 0.55 0.02. The second
method to determine fractionation factors is to measure the equilibrium constant
for a reaction as a function of the atom fraction of deuterium in the solvent, and
then using relationships like those that lead to equations 1-1 and 1-2 to
calculate the fractionation factor from a fit to the data. This method is then
indirect and is unable to separate the contribution from the fractionation at one
site from contributions from the medium, which may be a factor. Lowe and
Smith (1975) have used this method to determine the fractionation factor of the
acidic hydrogen in benzoic acid to be 1.02 0.01. Both of these methods for
determining fractionation factors lead to precise values for the simple cases
cited above involving one exchangeable hydrogen.
The unifying theme of this work is the study of the hydrogen / deuterium
fractionation factor properties of the hydrogenic positions of aqueous ligands of
metal ions in simple inorganic complexes and carbonic anhydrase. There is a
relative lack of information about the fractionation of the hydrogenic sites of the
aqueous ligands of metals compared to that of other functional groups. Using
an NMR relaxation method, I have measured the fractionation factor of the
aqueous ligand of cobalt in some simple inorganic complexes and in Co(ll)-
substituted carbonic anhydrase. The goal is to use these fractionation factors in
the interpretation of isotope effects on the formation of cobalt complexes and on
equilibrium and kinetic constants associated with carbonic anhydrase. The
fractionation factor of the aqueous ligand of cobalt in these systems represents
the individual contributions of those hydrogenic positions to the overall isotope
effect on the reactions.
Solvent hydrogen isotope effects in inorganic reactions are relatively little
studied compared to those effects in organic reactions (Kresge et al. 1987). The
data for ligand substitution reactions with transition metal complexes consists of


80
Engberg, P.; Millqvist, E.; Pohl, G.; Lindskog, S. Arch. Biochem. Biophys. 1985,
241, 628-638.
Eriksson, E. A.; Jones, T. A.; Liljas, A. In Zinc Enzymes; Bertini, I.; Luchinat, C.;
Maret, W.; Zeppezauer, M. Eds.; Birkhauser: Boston, 1986; 317-328.
Fabry, M. E.; Koenig, S. H.; Schillinger, W. E. J. Biol. Chem. 1970, 245, 4256-
4262.
Freeman, R.; Kempsell, S. P.; Levitt, J. J. Magn. Reson. 1980, 138, 453-479.
Friedman, H. L; Holz, M.; Hertz, H. G. J. Chem Phys. 1979, 70, 3369.
Glasoe, P. K.; Long, P. A. J. Phys. Chem. 1960, 64, 188-191.
Gold, V.; Lowe, B. M. J. Chem. Soc. (A) 1968, 1923-1932.
Gold, V.; Wood, D. L. J. Chem. Soc., Dalton Trans. 1982, 2452-2461.
Goodall, D. M.; Long, F. A. J. Am. Chem. Soc. 1968, 90, 238.
Gueron, M. J. Magn. Reson. 1975, 19, 58.
Hammes, G. C.; Steinfeld, J. I. J. Am. Chem. Soc. 1962, 84, 4639-4643.
Hunt, J. B.; Rhee, M. J.; Storm, C. B. Anal. Biochem. 1977, 79, 614-617.
Jencks, W. P.; Salvesen, K. J. Am Chem Soc. 1971, 93, 4433.
Kalifah, R. Biochemistry 1977, 16, 2236-2240.
Kalifah, R, G.; Strader, D. J.; Bryant, S. H.; Gibson, S. M. Biochemistry 1977, 16,
2241-2247.
Kararli, T.; Silverman, D. N. J. Biol. Chem. 1985, 260, 3484-3489.
Koenig, S. H.; Brown, R. D. Ann. N. Y. Acad. Sci. 1973, 222, 752-763.
Koenig, S. H.; Brown, R. D.; Bertini, I.; Luchinat, C. Biophys. J. 1983, 41, 179-
187.
Kresge, A. J.; More O'Ferrall, R. A.; Powell, M. F. In Isotopes in Organic
Chemistry, Volume 7; Buncel, E.; Lee, C. Eds.; Elsevier: New York, 1987; 177-
273.
Laughton, P. M.; Robertson, R. E. In Solute-Solvent Interactions4, Coetzee, J. F.;
Ritchie, C. D. Eds.; Marcel Dekker: New York, 1969; 399-538.
Lindskog, S. Biochemistry 1966, 5, 2641-2646.


27
Atom fraction of deuterium in solvent (n)
Figure 3-2: The dependence of pTjp (i = 1,2) at 300 MHz on n, the atom fraction
of deuterium in solvent water, for a 10 mM solution of NSO4 at 23 C. The
solutions were buffered at pH 4.4 with 0.5 mM acetic acid. Data are the means
and standard errors propagated from NMR measurements. p'Tip ( ) p'T2P ( ).


\\/
6
Co-OH-
Scheme 1-1:
+
O
II
c
II
o
^Co-OH2
+ HC03
step 1
^Co-OH2
^Co-OH' + H+ step 2
The metal hydroxide mechanism of carbonic anhydrase.


concentration of coordinating ligand, p[A]. In the region of the formation curves
where p[A] = 3.0 the pH of the solutions was greater than 9.0. Near this pH the
hydrolysis of the cobalt-bound water to a cobalt-bound hydroxide occurs in
Co(H20)62+ (Baes and Mesmer, 1976) and this same hydrolysis may be
occurring in the water ligands of the cobalt complexes with the bidentate
ligands. The data in the region p[A] = 3.0 was therefore difficult to interpret, and
I was unable to obtain precise values for k3. For glycine and acetylacetone in
H20 the values of log ki were within 0.1 log unit and the values of log k2 were
within 0.3 log unit of the literature values (Smith and Martell, 1975; no literature
value was available for A/,A/-dimethylglycine). The solvent hydrogen isotope
effects on these formation constants were all unity within experimental
uncertainty (Table 3-3), with the possible exception of k2 for acetylacetone.
For the binding of a ligand L'1 to cobalt as in equation 3-2 the solvent
hydrogen isotope effect is (see equation 1-2)
(ki)H2o (0Co(H2O)62+)12
= (3-7)
(ki)D2o (Co(H20)4(L)+)8(<|>H20)4
and from this the fractionation factor of the remaining metal-bound waters in
Co(H20)4L+ can be calculated by using the measured isotope effect, the
measured fractionation factor for Co(H20)62+ from NMR relaxation, and the
fractionation factor for H20 which is 1.0 (Schowen and Schowen, 1982). An
analogous calculation can be done to determine the fractionation factor of the
remaining metal-bound waters in Co(H20)2(L)2. Fractionation factors (j)ji and
^t2 fof Co(H20)62+ have been measured from the NMR relaxation times Ti and
T2 of water protons (Table 3-2). The results of these calculations from equation


HYDROGEN / DEUTERIUM FRACTIONATION FACTORS OF
THE AQUEOUS LIGAND OF COBALT IN Co(H20)62+ AND
Co(ll)-SUBSTITUTED CARBONIC ANHYDRASE
By
JAMES WEB KASSEBAUM
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1988

ACKNOWLEDGEMENTS
I wish to thank Dr. David Silverman for his thoughtfulness and guidance
shown to me during the work presented in this dissertation. I have benefitted
greatly from his scientific insight into experimental design and his careful
attention to excellence in scientific writing. I also wish to thank very much Dr. C.
K. Tu and Mr. George Wynns for their help in experimental procedures and for
many helpful discussions. I acknowledge the very valuable advice of the
members of my committee, Dr. E. Raymond Andrew, Dr. Robert Cohen, Dr.
Russell Drago, and Dr. Daniel Purich. Finally, I especially thank Dr. Seymour
Koenig at the IBM T. J. Watson Research Center in Yorktown Heights, New
York, for allowing me to visit his laboratory and use the field-cycling relaxometer
which added greatly to the scope of this work.
ii

TABLE OF CONTENTS
page
ACKNOWLEDGEMENTS ii
ABSTRACT v
CHAPTERS
1 INTRODUCTION 1
2 WATER PROTON NMR RELAXATION ENHANCEMENT IN
SOLUTIONS OF Co(H20)62+ AND Ni(H20)62+
9
Introduction 9
Experimental Procedure 10
Results 11
Discussion 16
3 HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF
THE AQUEOUS LIGAND IN Co(H20)62+ AND Ni(H20)62+
20
Introduction 20
Experimental Procedure 22
Results 25
Discussion 32
4 HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF
THE AQUEOUS LIGAND OF COBALT IN Co(ll)-
SUBSTITUTED CARBONIC ANHYDRASE 39
Introduction 39
Experimental Procedure 40
Results 42
Discussion 48
5 INTERPRETATION OF SOLVENT HYDROGEN ISOTOPE
EFFECTS ASSOCIATED WITH CARBONIC ANHYDRASE 52
Introduction 52
in

Experimental Procedure 53
Results 55
Discussion 58
6 MAGNETIC FIELD AND pH DEPENDENCE OF THE WATER
PROTON NMR RELAXATION ENHANCEMENT IN
SOLUTIONS OF Co(ll)-SUBSTITUTED CARBONIC
ANHYDRASE III 65
Introduction 65
Experimental Procedure 66
Results 67
Discussion 67
7 CONCLUSIONS 73
REFERENCES 79
BIOGRAPHICAL SKETCH 83
i v

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
HYDROGEN / DEUTERIUM FRACTIONATION FACTORS OF
THE AQUEOUS LIGAND OF COBALT IN Co(H20)62+ AND
Co(ll)-SUBSTITUTED CARBONIC ANHYDRASE
By
James Web Kassebaum
December, 1988
Chairman: Dr. David N. Silverman
Major Department: Biochemistry and Molecular Biology
I have measured the hydrogen / deuterium fractionation factor for the
rapidly exchanging aqueous ligands of cobalt in Co(H20)62+ and in three Co(ll)-
substituted isozymes of carbonic anhydrase. The fractionation factor was
determined from NMR relaxation rates at 300 MHz of the protons of water in
mixed solutions of H20 and D20 containing these complexes. In each case, the
paramagnetic contribution to 1/T2 was greater than to 1/Ti, consistent with a
chemical shift mechanism affecting 1/T2. The fractionation factors obtained from
Ti for Co(H20)62+ and for the isozymes of Co(ll)-substituted carbonic anhydrase
were close to the fractionation factor for bulk water, which is unity. The
fractionation factors obtained from T2 were 0.73 0.02 for Co(H20)62+, 0.72
0.02 for Co(ll)-substituted carbonic anhydrase I, 0.77 0.01 for Co(ll)-
substituted carbonic anhydrase II, and 1.00 0.07 for Co(ll)-substituted
carbonic anhydrase III. I concluded that fractionation factors in these cases
determined from Ti and T2 measured isotope preferences for different
populations of ligand sites. I suggest that since T2 has a large contribution from
v

a chemical shift mechanism, the fractionation factor determined from T2 has a
large contribution of the fractionation of inner shell ligands. The fractionation
factors determined from T-i are close to unity, the value of the fractionation factor
of bulk water, and contain a larger contribution of the fractionation of outer shell
water.
The fractionation factor of Co(H20)62+ was used to interpret the solvent
hydrogen isotope effects on the formation of complexes of cobalt with the
bidentate ligands glycine, A/,A/-dimethylglycine, and acetylacetone. The
contribution of the fractionation factor of the inner shell water in Co(H20)62+ did
not account completely for the measured isotope effect, and I suggest that the
hydrogen / deuterium fractionation of outer shell water makes a large
contribution to the isotope effect on the formation of these complexes.
The solvent hydrogen isotope effect on the equilibrium binding of iodide
ion to Co(ll)-substituted carbonic anhydrase I and II was determined and then
interpreted using the fractionation factor of the aqueous ligand of cobalt at the
active site. This fractionation factor could not account for the measured isotope
effect, implying that the hydrogen / deuterium fractionation of additional groups,
or water associated with the enzyme, changes on going from the reactant state
to the product state when iodide binds to Co(ll)-substituted carbonic anhydrase I
and II. Although not yet helpful in interpreting the complex contributions to the
isotope effects in the enzymatic catalysis, these hydrogen / deuterium
fractionation factors for the water bound to cobalt in carbonic anhydrase can be
significantly different from the fractionation factor for solvent water and may be
sensitive to the active site environment in these homologous isozymes of
carbonic anhydrase.
VI

CHAPTER 1
INTRODUCTION
The effect of an isotopic substitution on rate and equilibrium constants
associated with a chemical reaction can give information about the mechanism
of the reaction. One method in making an isotopic substitution is the
replacement of D2O for H2O as the solvent, which can affect a reaction in two
ways. First if water is a substrate in the reaction, then the rate and equilibrium
constants can be affected by the presence of D2O. Also, if there are
exchangeable hydrogenic sites in the reactants or products, or, in the case of an
enzyme-catalyzed reaction, if there are exchangeable hydrogenic sites on the
enzyme, then the rate and equilibrium constants can be affected when D2O is
the solvent, because deuterium will be present in the exchangeable positions.
The difference in a reaction in H20 and D2O results from differences in
the zero point energy of bonds to H or D (Schowen, 1978). In the case of a
reaction rate, if the zero point energy difference for a bond to H and to D
decreases on going to the transition state, then the activation energy is greater
in the case of D, and the rate is lower. This is an example of a normal isotope
effect, in which the rate constant when the bond is to H, kn, is greater than the
rate constant when the bond is to D, kp. There also exist reactions where ko >
kn, which is known as an inverse isotope effect.
The difference in zero point energy can be expressed in terms of an
equilibrium constant for isotope exchange for a particular hydrogenic site. For a
single solute species in an isotopically-mixed aqueous solvent, the hydrogen /
1

2
deuterium fractionation factor, (¡), is the equilibrium constant for this isotope
exchange reaction
R(H) + (HOD) -> R(D) + (HOH)
[R(D)] / [R(H)]
=
[HOD] / [HOH]
in which a solute hydrogenic species can exchange with a solvent hydrogenic
species. The fractionation factor measures the tendency of D to accumulate in
the solute relative to the deuterium content of bulk solvent. A fractionation factor
greater than unity implies that D will accumulate in the solute position relative to
the solvent, and that the bond to hydrogen is stronger in the solute position than
that bond in the solvent. For a fractionation factor less than unity, the light
isotope of hydrogen will accumulate in the solute position, and the bond to
hydrogen is weaker in the solute relative to the solvent.
An isotope effect is a function of the fractionation factors of each
hydrogenic site in a reaction. For an isotope effect on a kinetic constant,
all reactant state sites
n all transition state sites
(Mi
n ^
and for an isotope effect on an equilibrium constant,

3
all reactant state sites
n all product state sites
(1-2)
where (j)R, (j)T, and §p are reactant, transition, and product state hydrogenic site
fractionation factors, respectively (Schowen, 1978). Thus it is clear that to
understand completely an isotope effect associated with a reaction, the
knowledge of the underlying fractionation factors is required.
Values of the fractionation factor for various functional groups are known
relative to water as the solvent (Schowen and Schowen, 1982; Kresge et at.
1987), and range from 0.4 for some examples of hydrogen bound to sulfur, to
1.5 for one case of a hydrogen bound to a tertiary amine. The majority of these
fractionation factors where measured by one of two methods. One involves
using NMR chemical shifts to measure the fractionation factor of a hydrogenic
site directly. The technique is based on the rapid exchange of hydrogens
between solute and solvent species, so that the observed proton resonance is a
weighted average of the chemical shift of the solute and solvent positions. The
fractionation factor is determined by measuring the chemical shift at different
concentrations of solute in solvents of low and high atom fraction of deuterium.
Using this method, Gold and Lowe (1968) determined the fractionation factor of
the exchangeable hydrogen of acetic acid to be 0.96 0.02. A variation of this
method when there is slow exchange of hydrogens between solute and solvent
involves measuring the NMR signal areas in mixed isotopic solvents. In this
manner Chiang et at. (1980) have determined the fractionation factor of the
exchanging position of 1,8-Bis(dimethylamino)-naphthalene to be 0.90 0.01,
and Szawelski et ai (1982) have measured the fractionation factor of the

4
hydrogen bound to sulfur in mercaptoethanol to be 0.55 0.02. The second
method to determine fractionation factors is to measure the equilibrium constant
for a reaction as a function of the atom fraction of deuterium in the solvent, and
then using relationships like those that lead to equations 1-1 and 1-2 to
calculate the fractionation factor from a fit to the data. This method is then
indirect and is unable to separate the contribution from the fractionation at one
site from contributions from the medium, which may be a factor. Lowe and
Smith (1975) have used this method to determine the fractionation factor of the
acidic hydrogen in benzoic acid to be 1.02 0.01. Both of these methods for
determining fractionation factors lead to precise values for the simple cases
cited above involving one exchangeable hydrogen.
The unifying theme of this work is the study of the hydrogen / deuterium
fractionation factor properties of the hydrogenic positions of aqueous ligands of
metal ions in simple inorganic complexes and carbonic anhydrase. There is a
relative lack of information about the fractionation of the hydrogenic sites of the
aqueous ligands of metals compared to that of other functional groups. Using
an NMR relaxation method, I have measured the fractionation factor of the
aqueous ligand of cobalt in some simple inorganic complexes and in Co(ll)-
substituted carbonic anhydrase. The goal is to use these fractionation factors in
the interpretation of isotope effects on the formation of cobalt complexes and on
equilibrium and kinetic constants associated with carbonic anhydrase. The
fractionation factor of the aqueous ligand of cobalt in these systems represents
the individual contributions of those hydrogenic positions to the overall isotope
effect on the reactions.
Solvent hydrogen isotope effects in inorganic reactions are relatively little
studied compared to those effects in organic reactions (Kresge et al. 1987). The
data for ligand substitution reactions with transition metal complexes consists of

5
a small number of solvent hydrogen isotope effects measured for equilibrium
constants and for rate constants of reactions with Cr3+ and Co3+ (Laughton and
Robertson, 1969; Gold and Wood, 1982). Kresge et al. (1987) have suggested
that large rate and equilibrium differences between reactions of transition metal
aquacomplexes in H2O and D20 should be expected because the hydrogen /
deuterium fractionation factor of the water molecules in the inner coordination
shell would be raised to the twelfth power for a hexaaquacomplex.
Solvent hydrogen isotope effects on reactions catalyzed by carbonic
anhydrase are known and have been used to deduce aspects of the catalytic
mechanism (Silverman and Vincent, 1983). Carbonic anhydrase is a zinc
metalloenzyme which catalyzes the reversible hydration of carbon dioxide to
produce bicarbonate and a proton.
C02 + H20 ~ HCOg+H+
The zinc can be removed and replaced with cobalt to produce an enzyme with
very similar catalytic properties (Lindskog, 1983) and the cobalt, because of its
unpaired electrons, provides a spectroscopic probe of the enzyme (Bertini and
Luchinat, 1983). In particular, the paramagnetic cobalt will enhance the NMR
relaxation rate of water protons in solutions of Co(ll)-substituted carbonic
anhydrase. The metal ion is at the base of the active site cleft and is
coordinated by three histidine ligands with a water molecule as a fourth ligand
which can ionize to a hydroxide. The mechanism of C02 hydration (scheme 1-
1) takes place by direct nucleophilic attack of a metal-bound hydroxide on C02
to produce HCO3' (Steiner et al. 1975), with a water molecule then displacing
HCCV to produce a metal-bound water. A proton must then be transferred out
of the active site to regenerate the metal-bound hydroxide for the next round of
catalysis.

\\/
6
Co-OH-
Scheme 1-1:
+
O
II
c
II
o
^Co-OH2
+ HC03
step 1
^Co-OH2
^Co-OH' + H+ step 2
The metal hydroxide mechanism of carbonic anhydrase.

7
When this reaction is performed in D20 some very interesting isotope
effects are observed. The isotope effect on the steady-state parameter kcat for
both native and Co(ll)-substituted carbonic anhydrase II is large, 3.8 in the case
of native zinc-carbonic anhydrase II (Steiner et al. 1975), and 1.8 for Co(ll)-
substituted carbonic anhydrase II (R. S. Rowlett, unpublished), which means
that a primary proton transfer is at least part of the rate limiting step measured
by kcat- The ratio kcat/Km contains steps up to and including the first irreversible
step in the reaction, and for the catalyzed reaction the first irreversible step is
the release of HCO3' from the enzyme. The isotope effect is unity on kcat/Km for
both native zinc-carbonic anhydrase II (Steiner et al. 1975) and for Co(ll)-
substituted carbonic anhydrase II (R. S. Rowlett, unpublished), which implies
that the rate determining step measured by kcat is separate and distinct from the
interconversion of C02 and HCO3'. Thus the rate-limiting step measured by kcat
is the rate of transfer of a proton from the metal-bound water out of the active
site to regenerate the metal-bound hydroxide (step 2 of Scheme 1-1). From this
description it is clear that the aqueous ligand of the metal at the active site of
carbonic anhydrase is intimately involved in the mechanism of catalysis, and to
interpret the isotope effects knowledge of the fractionation factor of the aqueous
ligand of the metal is required.
The fractionation factor of the aqueous ligand of a paramagnetic metal
can be determined by taking advantage of the paramagnetic contribution to the
NMR relaxation rates of water protons in solutions of the metal in which the
protons are in fast exchange between the coordination shell of the metal and
the bulk solvent. The NMR measurements in this work were performed at a
proton resonance frequency of 300 MHz, which is a higher frequency than has
been used to study the water proton relaxation enhancement of first row
transition metals (Dwek, 1975; Bertini and Luchinat, 1986). Thus the relaxation

8
mechanisms operating in these systems at 300 MHz need to be understood
before the relaxation can be used to measure a fractionation factor, and this is
discussed in Chapter 2. Chapter 3 includes the determination of the
fractionation factor of Co(H20)62+ and Ni(H20)62+ along with measurements of
the solvent hydrogen isotope effect on the formation constants of some cobalt
complexes to see if the fractionation factor of Co(H20)62+ can account for any of
these effects. In Chapter 4 I have measured the fractionation factor of three
mammalian isozymes of Co(ll)-substituted carbonic anhydrase using NMR
relaxation, and in Chapter 5 I discuss the use of these fractionation factors to
interpret solvent hydrogen isotope effects on the equilibrium binding of an
inhibitor to carbonic anhydrase and on the catalysis. Finally, the relaxation
properties of Co(ll)-substituted carbonic anhydrase III have not previously been
reported, and in Chapter 6 I discuss experiments to determine the magnetic
field and pH dependence of the water proton relaxation enhancement in
solutions of Co(ll)-substituted carbonic anhydrase III.

CHAPTER 2
WATER PROTON NMR RELAXATION ENHANCEMENT IN
SOLUTIONS OF Co(H20)62+ AND Ni(H20)62+
Introduction
The initial studies of the influence of paramagnetic ions on the NMR
relaxation rates of water protons were performed at proton resonance
frequencies up to 60 MHz (Bernheim et al., 1959; Morgan and Nolle, 1959).
These results were interpreted in terms of the Solomon-Bloembergen
equations, which describe the relaxation as the sum of an electron-nuclear
dipole-dipole interaction and a scalar interaction (Solomon, 1955;
Bloembergen, 1957). At a frequency of 20 MHz the water protons in solutions of
Co(H20)62+ were observed by Bernheim et al. (1959) to relax such that T1 = T2,
and this relaxation was also observed at 10 MHz for solutions of Ni(H20)62+
(Morgan and Nolle, 1959). This result is consistent with the electron-nuclear
dipole-dipole mechanism dominating the relaxation of the water protons, the
correlation time for the interaction being the very short electron spin relaxation
time. This short relaxation time for the electron in Co(H20)62+ and Ni(H20)62+
precludes any contribution to the relaxation from the scalar mechanism. In the
absence of a scalar contribution, 7/6 would be the largest observable ratio for
Ti/T2.
I have made measurements of the relaxation of water protons in solutions
of Co(H20)62+ and Ni(H20)62+ at a proton resonance frequency of 300 MHz. In
each case it was observed that the ratio T-|/T2 was much larger than unity,
consistent with a chemical shift mechanism affecting T2 at this frequency.
Because this mechanism will affect T2 to the greatest extent when the protons
9

10
are in the primary hydration shell of the ion, this allowed a qualitative
differentiation between the relaxation of inner and outer shell water.
Experimental Procedure
The NMR relaxation times T1 and T2 of water protons were measured at a
frequency of 300 MHz on a Nicolet NT-300 spectrometer at 23 C. T1 was
measured by the inversion recovery method according to Freeman et al. (1980)
and T2 was measured by the Carr-Purcell-Meiboom-Gill sequence (Meiboom
and Gill, 1958). The observed relaxation of water protons in a solution of a
paramagnetic ion is the sum of the relaxation due to the paramagnetic ion, 1/Tip
(i=1,2), and the relaxation in the absence of the ion, 1/Tio,
1/T¡ = 1/T¡p +1/Tb (2-1)
The contribution of the paramagnetic ion to the longitudinal relaxation is given
by (Luz and Meiboom, 1964)
1/Tip- p[1/(Tim +xm)] (2-2)
and the paramagnetic contribution to the transverse relaxation is given by (Swift
and Connick, 1962)
0/"l"2m)(1/T2m + 1/Tm) + A(Om2
1/T2p = p [ ] (2-3)
Tm((1/T2m + 1/Xm)2 + ACOm2)
where Tim is the relaxation time of a proton in the hydration shell of the metal
and is described by the Solomon-Bloembergen equations, Tm is the lifetime of a
proton in the hydration shell, and p' is q[M]/55.5 where q is the hydration
number and [M] is the concentration of the metal ion. Acom is the chemical shift
difference between a proton in the primary hydration shell and a proton in the

free solvent site. If Acom2 is small relative to the other terms, equation 2-3
reduces to equation 2-2. For these metal ions, 1/T¡0 was taken as the relaxation
of water containing only buffer, with no added metal ions. R¡, the relaxivity, is
defined as the paramagnetic relaxation per millimolar concentration of metal
ion.
Ri = (1/Tjp)/ ([Metal] x 1000) (2-4)
Solutions of Co(H20)62+ and Ni(H20)62+ for NMR measurements
contained ultra-pure C0CI2 or NSO4 (Aldrich Gold-Label) with 5 x 10-4 M acetic
acid buffer at pH = 4.4. This pH was necessary to preclude the formation of
multi-nuclear hydrolysis products (Baes and Mesmer, 1976). When EDTA was
present no acetic acid was used. Water used for solution preparations was
distilled and passed through two ion-exchange resin cartridges (Cole-Palmer
1506-35). D20 (99.8%) was stirred with activated charcoal, the charcoal filtered
out, and then the D20 was distilled.
Results
The paramagnetic contribution to the relaxivity R2 of water protons in
solutions of Co2+ and Ni2+ at 300 MHz was greater than that obtained at lower
frequencies by previous investigators (Table 2-1). For Co2+, R1 at 300 MHz and
20 MHz were close in value while R2 was much larger at 300 MHz than at 20
MHz. For Ni2+, R1 at 300 MHz was slightly larger than at 10 MHz which is
understood to be a result of a field dependent increase in the electron
relaxation time (Friedman et al., 1979), but R2 was much larger at 300 MHz than
at 10 MHz.
The addition of EDTA to the solutions of C0CI2 and N1SO4 reduced R1
and R2 at 300 MHz (Figures 2-1 and 2-2). When an equimolar amount of EDTA

12
Table 2-1: Paramagnetic contribution to the relaxivities of water protons
observed for aqueous solutions of C0CI2 and NSO4.
frequency
(MHz)
R1
(mM'1 s'1)
r2
(mM-1 s'1)
Tip/T2P
Co(H20)62+
(0
0
0.18
0.21
1.2
300 b
0.16
2.0
12.5
Ni(H20)62+
10c
0.64
0.71
1.1
300 b
0.95
2.3
2.4
a From Bernheim et al. (1959) for an unbuffered solution of 0.5 M C0CI2 at 25
C.
b Determined in this work at 23 C for 10 mM CoCI2 or 10 mM NiS04 which
contained 10% D2O by volume and 5 x 10'4 M acetic acid buffer. The pH was
4.4.
cFrom Morgan and Nolle (1959) for a solution of 10 mM Ni(N03)2at 27 C in 0.1
M perchloric acid.

1 3
Figure 2-1: The relaxivity of the protons of water at 300 MHz due to Co2+ in
solutions of 10 mM CoCI2 as a function of the molar ratio of EDTA to Co2+ at 23
C. Solutions contained 50% D2O by volume. R1 (), R2 (a).

Relaxivity (mM*1 s-1)
14
[EDTA]/[Ni2+]
Figure 2-2: The relaxivity of the protons of water at 300 MHz due to Ni2+ in
solutions of 10 mM NiS04 as a function of the molar ratio of EDTA to Ni2+ at 23
C. Solutions contained 50% D2O by volume. R1 (), R2 (a).

15
was added to solutions of Co2+, Ri was reduced to 47% of the value of R-i in the
absence of EDTA, R2 was reduced to 5% of the value of R2 in the absence of
EDTA, and Tip/T2P was 1.2. The remaining paramagnetic relaxivity due to the
CoEDTA complex was R-| = 0.075 mM'1 S'1 and R2 = 0.091 mM'1 s*1. When an
equimolar amount of EDTA was added to solutions of Ni2+, R-i was reduced to
58% of the value of Ri in the absence of EDTA, R2 was reduced to 34% of the
value of R2 in the absence of EDTA, and Tip/T2p was 1.4. The remaining
paramagnetic relaxivity due to the NiEDTA complex was Ri = 0.55 mM-1 s-1 and
R2 = 0.78 mM-1 sr\
The observed change in the chemical shift caused by the paramagnetic
metal, Aco, at 300 MHz for a 10 mM solution of C0CI2 was measured to be 158
Hz. The measured change in the chemical shift, Aco, is the difference between
the chemical shift of water protons in water containing only buffer and the
chemical shift of water protons in a solution of a paramagnetic metal. This was
measured with tetramethylsilane as an external reference using coaxial tubes.
The influence of this chemical shift on the transverse relaxation of Co(H20)62+
solutions at 20 MHz and 300 MHz was estimated using equation 2-3 (Table 2-
2). In the calculations, I used xm = 7.4 x 107 s determined by Swift and Connick
(1962) using 170 exchange. Also, I used T1m = T2m as predicted by the
Solomon-Bloembergen equations for Co(H20)62+ at 20 MHz and 300 MHz.
From the observed T1 at 300 MHz, T1m was calculated (from equations 2-1 and
2-2) to be 6.62 x 10'4 s. From the data of Bernheim et al. (1959) T-im = 5.95 x 10-
4 s at 20 MHz. Since Timxm, the fast exchange limit applies at 20 MHz and at
300 MHz. Acom was calculated from Aco at 300 MHz using Aco = p' Acom
which is valid in the fast exchange limit (Swift and Connick, 1962). I determined
Acom to be 1.46 x 105 Hz at 300 MHz. Since Acom is directly proportional to the
precessional frequency, Acom at 20 MHz is 9.73 x 103 Hz. The results using

these values with equation 2-3 are in Table 2-2 and are in good agreement with
the experimental results in Table 2-1.
Discussion
The results at 300 MHz of T1p/T2P much greater than unity for solutions of
both Co(H20)62+ and Ni(H20)62+ suggest that a relaxation mechanism different
from the dipole-dipole mechanism is present at this frequency. The data in
Table 2-1 indicate that this mechanism primarily, if not exclusively, affects the
transverse relaxation. I have considered three possible causes of this field-
dependent increase in 1/T2p. The scalar contribution to 1/T2m has a correlation
time equivalent to the electron relaxation time which is field dependent
(Bloembergen and Morgan, 1961). However, I have rejected a contribution
from the scalar term because it would require an unreasonably large increase
in the electron spin relaxation time at 300 MHz. When the electron relaxation
time is less than both the rotational correlation time and xm, a Curie spin
mechanism can result (Gueron, 1975). This Curie spin is the thermal average
of an electron spin which can become large at high field and can cause an
increase in T-im/T2m. However, I also reject a contribution from this mechanism
because the rotational correlation time for a hexaaquacomplex is too short to
allow the Curie spin to become effective.
Melamud and Mildvan (1975) observed a field-dependent increase in
1/T2p for water protons in solutions of Co(H20)62+ and concluded that the Acom2
term in equation 2-3, a chemical shift mechanism, made a major contribution at
the higher frequencies. I support this conclusion in two ways. First, I have
calculated the magnitude of the chemical shift contribution to the transverse
relaxation of Co(H20)62+ solutions at 20 MHz and 300 MHz using equation 2-3.
These calculations demonstrate that the observed value of Tip/T2p at 300 MHz
is consistent with a chemical shift mechanism which affects T2p (compare

Table 2-2: Calculated paramagnetic relaxivities of water protons in solutions
containing Co2+.
frequency
Ri
r2
Tip/T2P
(MHz)
(mM1 S'1)
(mM'1 s'1)
20
0.18
0.19
1.1
300
0.16
1.8
11.2
Note: Calculated using equations 2-2 to 2-4 in the text. The calculations used
Tm=7.4 x 10'7 s (Swift and Connick, 1962), the 20 MHz relaxation data from
Bernheim et at. (1959), and the 300 MHz relaxation data and chemical shift
determined in this work. A further description is in the text.

18
Tables 2-1 and 2-2). The chemical shift does not have an effect at 20 MHz at
which frequency Acom21/T2mTm, 1AT2m2. The large chemical shift at 300 MHz
due to the addition of Co2+ is the result of contact interactions which result from
the delocalization of unpaired electron spin density. At lower magnetic field
strengths this mechanism has been observed to affect T2 for 170 and 19F in
cobalt complexes (Swift and Connick, 1962; Eisenstadt, 1969). The effects on
the relaxation when EDTA is added to solutions of Co(H20)62+ is the second
way I support the chemical shift mechanism. The chemical shift mechanism will
affect T2p to the greatest extent when the protons are in the primary hydration
shell of the ion because it is a contact interaction. EDTA binds to Co2+ (and
Ni2+) occupying all coordination sites and thus excludes water molecules from
the inner shell. Thus the addition of EDTA would prevent the water protons from
experiencing the chemical shift mechanism. The ratio Tip/T2p was near unity for
water protons at 300 MHz when an equimolar amount of EDTA was added to
solutions of Co2+, consistent with the absence of a chemical shift relaxation
mechanism and the dominance of a dipole-dipole relaxation mechanism
(Figure 2-1).
When EDTA was added to solutions of Co2+, the percent decrease in the
relaxivity R2 at 300 MHz was much greater than (Figure 2-1). Since EDTA is
excluding water molecules from the inner coordination shell, this suggests that
R2 has a much larger contribution from the relaxation of water protons in the
inner shell. The residual paramagnetic relaxivity of the cobalt-EDTA complex is
most likely due to water protons in an "outer shell" of the complex. This residual
relaxivity makes up a much larger part of R1 than R2 observed for Co(H20)62+
solutions at 300 MHz. This suggests that the relaxation of outer shell water
contributes to a greater extent to Rt than to R2 in Co(H20)62+.

As EDTA was added to solutions of Ni2+, R2 decreased more than R1
(Figure 2-2), though the decrease is not as large as seen in solutions of Co2+.
Also, the residual relaxivity of the NiEDTA complex is much larger than in
solutions of the CoEDTA complex, and this residual relaxivity makes up a large
part of both R^ and R2. These results for Ni2+ suggest that the chemical shift
mechanism is responsible for the increase in R2 at 300 MHz, but that Ri and R2
each have a large contribution of outer shell water.

Q|_|AP1"£P 0
HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF THE
AQUEOUS LIGAND IN Co(H20)62+ AND Ni(H20)62+
Introduction
The hydrogen / deuterium fractionation factor, <{), of a metal-bound water
is the equilibrium constant for this isotope exchange reaction:
M(HOH) + (HOD) > M(HOD) + {HOH)
[M(HOD)] / [M(HOH)]
(3-1)
[HOD] / [HOH]
This fractionation factor measures the tendency of deuterium to accumulate at
the aqueous ligand of the metal relative to the deuterium content of bulk solvent.
Fractionation factors are valuable in interpreting the effects of deuterium on
kinetic and equilibrium constants because they represent individual
contributions to isotope effects measured relative to the common reference of
water. The fractionation of hydrogen isotopes in water bound to metal ions has
been considered (Schowen and Schowen, 1982); however, there are few
reports of the fractionation factor of the aqueous ligands of a metal. Using NMR
methods, Melton and Pollack (1969) have measured such a factor for the water
ligands of Cr(H20)62+ ((j) = 1.00 0.03). They relied on the properties of a
paramagnetic metal to enhance the NMR relaxation rate of protons of water
exchanging rapidly between bulk solvent and the hydration shell of the metal.
By measuring the proton relaxation rate as a function of the deuterium content
of solvent, the fractionation factor of the metal-bound site can be determined.
20

21
Using this method at a proton resonance frequency of 300 MHz, I have
measured the fractionation factor of the aqueous ligand of cobalt in Co(H20)62+
and of nickel in Ni(H20)62+. In the case of Co(H20)62+ the value of the
fractionation factor was dependent upon whether it was determined from T-i or
T2; however, the corresponding values of (j) for Ni(H20)62+ were identical when
obtained from T-| or T2. The analysis of the relaxation discussed in Chapter 2
can be used to interpret the fractionation factors, and allows a qualitative
differentiation between the hydrogen / deuterium fractionation of inner and outer
shell water in these complexes.
I have then used the knowledge of the fractionation of Co(H20)62+ to
study the solvent hydrogen isotope effects on the formation constants of the
reactions of glycine, A/,A/-dimethylglycine, and acetylacetone with Co2+. An
isotope effect on an equilibrium constant is the product of the reactant state
hydrogenic site fractionation factors divided by the product of the product state
hydrogenic site fractionation factors (equation 1-2). In a reaction of Co2+ with a
bidentate ligand L*1 represented as
Co(H20)62+ + L-1 Co(H20)4L+ + 2 H20 (3-2)
the solvent hydrogen isotope effect is a function of the fractionation of the
hydrogenic positions of Co(H20)62+ as well as Co(H20)4L+ and H20, and the
fractionation factor of Co(H20)62+ represents the individual contribution of the
hydrogenic positions of Co(H20)62+ to the overall isotope effect. Because the
solvent hydrogen isotope effects on the formation constants were measured to
be near unity in the case of each of the ligands, I determined that the
contributions of the fractionation of the hydrogenic positions of the inner shell
water in Co(H20)62+ and in the ligand complexes of cobalt can not account
completely for the overall isotope effect. I suggest that the fractionation of outer

22
shell water also makes a large contribution to the Isotope effect on these
reactions.
Experimental Procedure
NMR Measurements: Measurements at 300 MHz and solutions were the
same as in Chapter 2. Some measurements of Ti of water protons in solutions
of Co2+ were performed at frequencies between 0.01 and 50 MHz on a field
cycling "relaxometer" at the IBM T. J. Watson Research Center, Yorktown
Heights, NY (Koenig etal., 1983).
Measurement of the Hydrogen / Deuterium Fractionation Factor: The
measurement of the relaxation of water protons in solutions of varying
deuterium content and containing paramagnetic metal ions is given by
1/T¡p = p'[X]/(1 n + n<)>) (3-3)
where n is the atom fraction of deuterium in solvent water, (j) is the hydrogen /
deuterium fractionation factor of the water ligands of the metal, and [X] is the
term in brackets in equations 2-2 and 2-3 for i = 1 or 2, respectively. This
equation is derived considering the fractionation of isotopes that arises when
there is a deuterium content in water. In this case the ratio of protons in the
hydration shell of the metal to total protons in solution is p'/(1-n+n<})) (Schowen
and Schowen, 1982). Equation 3-3 can be rearranged to
P'Tjp = [X]1 n(1- Thus pTjp vs n gives a slope of -(1-)[X]-1 and an intercept of [X]*1, and (j) = 1 +
slope / intercept.

23
Determination of the isotope effects on formation constants: The
formation constants for the reactions of Co2+ with a bidentate ligand, L-1, as in
equation 3-2 are defined as
[Co(H20)4L+]
ki = (3-5)
[Co(H20)62+] [L-1]
with similar expressions for k2 and (<3. Potentiometric titrations were performed
in H20 and D20 (99.8%) containing 1 mM CoCI2 and 4 mM glycine, N,N-
dimethylglycine, or acetylacetone. The values of pH and pD used in these
experiments were corrected from pH meter readings. The correction of a pH
meter reading in 100% D20 is pD = meter reading + 0.4 (Glasoe and Long,
1960). The formation constants were determined by a non-linear least-squares
fit of the data (RS1, BBN Software, Cambridge, MA) to the equation derived by
Carlson et al. (1945) describing , the ratio of the concentration of metal-bound
ligand to total concentration of metal, as a function of the concentration of
coordinating ligand, [A].
ki[A] + 2k-i k2[A]2 + 3kik2k3[A]3
= (3-6)
1 + k-|[A] + k^A]2 + k1k2k3[A]3
The calculation of and [A] required knowledge of the acid ionization
constants of the free ligands, which were measured by potentiometric titrations
in H20 and D20 (Table 3-1). The negative logarithm of the acid ionization
constants, pK, in H20 were all within 0.07 pK units from the literature values
(Smith and Martell, 1975). The solvent hydrogen isotope effects for glycine
expressed as ApK where ApK = pKD2o pKh2o were within 0.03 ApK units from
the values obtained previously by Jencks and Salvesen (1971).

2 4
Table 3-1: Solvent hydrogen isotope effects on the acid ionization constants of
glycine, A/,A/-dimethylglycine, and acetylacetone.
(pKi)H2Oa
ApK!
(pK2)H20 b
ApK2
glycine
2.37 0.02
0.40 0.04
9.72 0.03
0.56 0.05
A/,/\/-dimethylglycine
1.96 0.03
0.40 0.04
9.81 0.03
0.55 0.05
acetylacetone
9.03 0.10
0.62 0.17
-
-
Note: Acid ionization constants were determined in H2O and D2O by
potentiometric titrations at 23C. Solutions contained 10 mM of either glycine,
A/,A/-dimethylglycine, or acetylacetone.
a pK = -log K where K is the acid ionization constant. Standard errors resulted
from a non-linear least-squares fit of the titration data. ApK = pKo2o pKh2o-

25
These potentiometric titrations were performed in solutions which
contained ultra-pure C0CI2 (Aldrich Gold-Label), and glycine and N,N-
dimethylglycine from Sigma and acetylacetone from Aldrich. Water used for
solution preparations was distilled and passed through two ion-exchange resin
cartridges (Cole-Palmer 1506-35). D20 (99.8%) was stirred with activated
charcoal, the charcoal filtered out, and then the D20 was distilled. Glassware
was rinsed with a solution of EDTA prior to use.
Results
Fractionation Factors of Co(HQ)s2 and Ni(HzO)g2; The relaxation of
water protons in solutions of 10 mM CoCI2 and 10 mM NiS04 as a function of
the atom fraction of deuterium in solvent water is shown in Figures 3-1 and 3-2.
From these data, the hydrogen / deuterium fractionation factor, 0, for
Co(H20)62+ and Ni(H20)62+ was obtained (Table 3-2). The value of the
fractionation factor was independent of the concentration of metal in the range
0.010 M to 0.025 M. The value of (j) for Co(H20)62+ was dependent on whether
it was obtained from T-i or T2; however, the corresponding values of (}) for
Ni(H20)62+ were identical when obtained from T1 orT2. In a separate
experiment, the fractionation factor from T1 for Co(H20)62+ was determined at
frequencies between .01 and 50 MHz on a field cycling "re laxo meter" at the IBM
T. J. Watson Research Center and found to be virtually field independent (data
not shown).
Solvent Hydrogen Isotope Effects on the Formation Constants of Cobalt
Complexes: The formation constants for the reactions of Co2+ with glycine, N,N-
dimethylglycine, and acetylacetone were determined in H20 and D20 and are
reported as the logarithm of the formation constants (Table 3-3). Representative
data of a formation curve (Carlson etal., 1945) for Co2+ and glycine are in
Figure 3-3 where is plotted as a function of the negative logarithm of the

26
Atom fraction of deuterium in solvent (n)
Atom fraction of deuterium in solvent (n)
Figure 3-1: The dependence of p'T¡p (i = 1,2) at 300 MHz on n, the atom fraction
of deuterium in solvent water, for a 10 mM solution of C0CI2 at 23 C. The
solutions were buffered at pH 4.4 with 0.5 mM acetic acid. Data are the means
and standard errors propagated from NMR measurements.

27
Atom fraction of deuterium in solvent (n)
Figure 3-2: The dependence of pTjp (i = 1,2) at 300 MHz on n, the atom fraction
of deuterium in solvent water, for a 10 mM solution of NSO4 at 23 C. The
solutions were buffered at pH 4.4 with 0.5 mM acetic acid. Data are the means
and standard errors propagated from NMR measurements. p'Tip ( ) p'T2P ( ).

28
Table 3-2: Fractionation factors determined from T-i and T2 at 300 MHz for the
aqueous ligands of the paramagnetic metal in Co(H20)62+ and Ni(H20)62+.
Co(H20)62+
Ni(H20)62+
^ <>t2
0.95.01 0.73.02
0.99.01 0.98.03
Note: Measurements were made at 23 C for 10 mM CoCI2 or NSO4 containing
5 x 10'4 M acetic acid buffer at pH 4.4. Data were obtained from the least
squares slope and intercept and standard error in a plot such as Fig. 3-1
according to equation 3-4.

29
Table 3-3: Logarithm of the formation constants determined for the reactions of
Co2+ with the ligands glycine, A/,A/-dimethylglycine, and acetylacetone in H2O
and D2O.
h2o
D20
(log kJop-ilogkjHjO
glycine
log ki
5.04 0.04
4.99 0.02
-0.05 0.04
log k2
4.32 0.04
4.23 0.02
-0.08 0.05
A/,A/-dimethylglycine
log ki
4.14 0.04
4.12 0.06
-0.03 0.07
log k2
3.20 0.06
3.18 0.13
-0.01 0.14
acetylacetone
log ki
5.30 0.07
5.42 0.10
0.12 0.13
log k2
4.40 0.09
4.61 0.12
0.21 0.15
Note: Determined by potentiometric titrations of solutions which contained 1
mM CoCI2 and 4 mM of the ligand and fitting the data to equation 3-6.
Measurements were made at 23 C. See equation 3-5 for the definition of ki.

30
Figure 3-3: The ratio of the concentration of metal-bound ligand to total
concentration of metal, , as a function of the negative logarithm of the
concentration of free glycine in solution existing as NH2-CH2-COO'1, p[A] (see
equation 3-6). The solutions contained 1 mM total C0CI2 and 4 mM total glycine
in H2O ( ) or D2O ( ) and were titrated with NaOH (NaOD). The solid lines are
a computer fit to the data using equation 3-6 with the values of the formation
constants given in Table 3-3.

concentration of coordinating ligand, p[A]. In the region of the formation curves
where p[A] = 3.0 the pH of the solutions was greater than 9.0. Near this pH the
hydrolysis of the cobalt-bound water to a cobalt-bound hydroxide occurs in
Co(H20)62+ (Baes and Mesmer, 1976) and this same hydrolysis may be
occurring in the water ligands of the cobalt complexes with the bidentate
ligands. The data in the region p[A] = 3.0 was therefore difficult to interpret, and
I was unable to obtain precise values for k3. For glycine and acetylacetone in
H20 the values of log ki were within 0.1 log unit and the values of log k2 were
within 0.3 log unit of the literature values (Smith and Martell, 1975; no literature
value was available for A/,A/-dimethylglycine). The solvent hydrogen isotope
effects on these formation constants were all unity within experimental
uncertainty (Table 3-3), with the possible exception of k2 for acetylacetone.
For the binding of a ligand L'1 to cobalt as in equation 3-2 the solvent
hydrogen isotope effect is (see equation 1-2)
(ki)H2o (0Co(H2O)62+)12
= (3-7)
(ki)D2o (Co(H20)4(L)+)8(<|>H20)4
and from this the fractionation factor of the remaining metal-bound waters in
Co(H20)4L+ can be calculated by using the measured isotope effect, the
measured fractionation factor for Co(H20)62+ from NMR relaxation, and the
fractionation factor for H20 which is 1.0 (Schowen and Schowen, 1982). An
analogous calculation can be done to determine the fractionation factor of the
remaining metal-bound waters in Co(H20)2(L)2. Fractionation factors (j)ji and
^t2 fof Co(H20)62+ have been measured from the NMR relaxation times Ti and
T2 of water protons (Table 3-2). The results of these calculations from equation

3 2
3-7 with glycine, A/,A/-dimethylglycine, and acetylacetone are in Table 3-4 using
(|)Ti and (¡)t2 for Co(H20)62+.
I have also tried to measure the fractionation factor for these coordinated
water molecules in the complexes directly, using the NMR relaxation method
used to determine the fractionation factor of Co(H20)62+. However, these
solutions of Co2+ and a bidentate ligand L consist of an equilibrium mixture of
Co(H20)62+ Co(H20)4L+ Co(H20)2L2 and CoL3 and the observed
paramagnetic relaxation is the sum of the relaxation of each of these four
species. This presents difficulties in obtaining the fractionation factor of one
species from these measurements.
Discussion
The definition of the hydrogen / deuterium fractionation factor I have
given (equation 3-1) assumes the rule of the geometric mean to be valid
(Schowen and Schowen, 1982); that is, the fractionation factor of one particular
hydrogenic site is independent of the isotopic composition of any other site. I
will assume that the rule is valid in my system, and this is supported by the high
degree of linearity in plots as in Figures 3-1 and 3-2. A deviation from the rule
of the geometric mean would result in a non-linear dependence of the
relaxation data on the atom fraction of deuterium in the solvent.
The value of the fractionation factor for the water ligands of cobalt in
Co(H20)62+ was dependent on whether it was determined from Ti or T2, while
that of Ni(H20)62+ was independent of or T2 (Table 3-2). These fractionation
factors are a weighted value of all the protons that experience the paramagnetic
relaxation. An explanation for this difference in (})t1 and (|)t2 is based on the
analysis of the contributions to the relaxation at 300 MHz (see Chapter 2). The
paramagnetic contribution to the transverse relaxation of water protons in
solutions of Co(H20)62+ has a predominant contribution of water in the inner

3 3
Table 3-4: Calculated fractionation factors for the metal-bound water in the
complexes of cobalt with glycine(gly), A/,A/-dimethylglycine (DMG), and
acetylacetone (acac).
0T1
<>T2
Co(H20)62+ a
0.95 0.01
0.73 0.02
<¡>Co(H20)4(gly)+
0Co(H2O)2(gly)2
0.91 0.02
0.79 0.04
0.61 0.03
0.36 0.03
Co(H20)4(DMG)+
<}>Co(H20)2(DMG)2
0.92 0.02
0.84 0.08
0.62 0.03
0.38 0.05
(t)Co(H20)4(acac)+
0Co(H2O)2(acac)2
0.96 0.04
1.04 0.12
0.65 0.04
0.47 0.07
Note: Calculated from equation 3-7 using the isotope effects on the formation
constants of the reaction of Co2+ with glycine, A/./V-dimethylglycine, and
acetylacetone (Table 3-3) and further description is in the text.
a Determined directly from NMR relaxation measurements.

34
coordination shell, suggesting that (j)j2 is measuring more of the fractionation of
inner shell water. Since T-ip has a larger contribution of outer shell water, (¡jj-i is
more of the fractionation of outer shell water. The analysis of the relaxation for
Ni(H20)62+ (Chapter 2) suggests that both Ti and T2 have a large contribution of
outer shell water, so contribution of the fractionation factor of outer shell water, and the influence of
the fractionation of the inner shell water is not measured by or (j)j2.
This interpretation of the fractionation factor for Co(H20)62+ is consistent
with expectations for the fractionation factor of metal-bound water. The
fractionation factor for H30+ is 0.69 (Chiang et al., 1980), presumably due to the
positive charge on the oxygen. It has been suggested that a metal-bound water
site ought to have a fractionation factor approximated by (0.69)8, with 5
measuring the degree of positive charge transferred from the metal to the
oxygen (Schowen and Schowen, 1982). The fractionation factor from T2 for
Co(H20)62+ being 0.73 supports this suggestion in resembling the fractionation
factor for H30. Pure water has by definition a fractionation factor of 1.0. Outer
shell water, being much more loosely bound than inner shell water, should be
much more like bulk water in its hydrogen / deuterium fractionation properties
(More O'Ferrall et al., 1971). The observation of <|)t1 for Co(H20)62+ and (j>r1
and ()>t2 for Ni(H20)62+ all being close to unity is consistent with these
fractionation factors having a large contribution of the fractionation of outer shell
water. The fractionation factor of the inner shell water in Ni(H20)62+ may well be
affected by the positive charge of the nickel and have a value closer to 0.69, or it
may have a value near unity; however, because of the large contribution of
outer shell water to both Ti and T2 in solutions of Ni(H20)62+, the influence of
the inner shell water is not large in the fractionation factors determined from
and T2.

35
The usefulness of fractionation factors is judged by their ability to aid in
interpretation of solvent hydrogen isotope effects. I have determined the solvent
hydrogen isotope effects on the formation of some simple inorganic complexes
of Co2+ with the goal to use the knowledge of the fractionation of Co(H20)62+ to
interpret those isotope effects. An isotope effect is a function of all the
fractionation factors for each hydrogenic site in a reaction (equation 1-2), and
for a reaction such as equation 3-2 the fractionation factor of Co(H20)62+
represents the individual contribution of those hydrogenic sites to the overall
isotope effect on the reaction.
In considering the solvent hydrogen isotope effect on the formation
constants (Table 3-3), initially I will assume that the isotope effect is made up
entirely of the contribution of the isotope preferences of the hydrogenic
positions of water in the inner coordination shell of Co(H20)62+, Co(H20)4L+,
and Co(H20)2L2. With this assumption there is enough information from the
isotope effect and the fractionation factor of Co(H20)62+ to calculate (from
equation 3-7) the fractionation factor of the remaining waters in the cobalt-
bidentate ligand complexes (Table 3-4). While I have done the calculations
with equation 3-7 using both <()t1 and (J)j2 for Co(H20)62+ (Table 3-2), the values
from (J)^ may not be meaningful since Ti is measuring more outer shell water,
and the number of outer shell waters is not known. One interpretation of these
calculations is that since t2 of Co(H20)62+ measures more of the fractionation
of inner shell water in Co(H20)62+, the calculation using ^ of Co(H20)62+
represents the resulting fractionation factor of the inner shell waters in the cobalt
complexes. As ligands are added to Co2+ the calculated fractionation factor of
the remaining coordinated waters decreases, and this decrease is the same for
each of the three ligands, though glycine and A/,A/-dimethylglycine coordinate
through an amino and carboxylate, and acetylacetone binds through two

36
carbonyls. If the assumption in these calculations is valid, the results of a lower
fractionation factor for Co(H20)4L+ means that the O-H bond becomes weaker
relative to that bond in Co(H20)62+, and the O-H bond is also weaker in
Co(H20)2L2 compared to Co(H20)4L+. This is similar to the effect of added
ligands on the lifetime of water molecules in the hydration shell which, for
instance, is shorter in Co(H20)4gly+ relative to Co(H20)62+ (Hammes and
Steinfeld, 1962), meaning that the hydration shell becomes more loosely bound
when ligands are added. Thus in these cobalt complexes with bidentate
ligands the metal-oxygen bond of a coordinated water is looser, and the oxygen
hydrogen bond appears looser, relative to Co(H20)62+.
However, from the measured solvent hydrogen isotope effects (Table 3-
3) and the calculations of Table 3-4, there are a number of reasons to question
the assumption that the only contribution to the isotope effects comes from the
hydrogens of inner shell water. First, such strong fractionation as 0.36 (Table 3-
4) is rare (Schowen and Schowen, 1982). Second, the lower fractionation
factor for Co(H20)4L+ and Co(H20)2L2 is in contrast to the suggestion that the
charge on the cobalt affects the fractionation of the metal-bound water
(Schowen and Schowen, 1982). In these complexes, the net charge on the
complex is +1 for Co(H20)4L+, and neutral for Co(H20)2L2, since the ligand has
a charge of -1 in each case. Thus one would expect the fractionation factor of
the waters in the ligand complexes to be closer to unity, according to the
expectation that <|) = (0.69)5, where 0.69 is the fractionation factor of H30+ and 8
is the degree of positive charge transferred from the metal to the oxygen of the
coordinated water molecule (Schowen and Schowen, 1982). The c()t2 for
Co(H20)62+ is consistent with this (Table 3-2), and thus one would expect [(f) for
Co(H20)62+] < [(f) for Co(H20)4L+] < [(f) for Co(H20)2L2], but this opposite to what
was observed (Table 3-4). Finally, based upon the large number of hydrogenic

37
sites in Co(H20)62+, large isotope effects would be expected if the only
contributions to the isotope effect came from the fractionation of inner shfell
water (Kresge etal., 1987). However, isotope effects of unity were observed for
these reactions (Table 3-3).
I suggest that the individual contribution of the fractionation factor of the
inner shell waters in Co(H20)62+, Co(H20)4L+, and Co(H20)2L2 to the overall
solvent hydrogen isotope effect on the formation constants is small. The
fractionation factor of some other hydrogenic sites has a contribution to the
isotope effect which is sufficient to cancel the contribution of the inner shell
waters so that the overall isotope effect is unity. The most likely choice for this is
the fractionation of outer shell water. Other possibilities include, in the case of
glycine, the composite fractionation factor of the hydration shell of the
carboxylate group, but this would be a small contribution of a fractionation factor
value already close to unity (the fractionation factor of the hydration shell of the
carboxylate group in the acetate ion is 0.89, Goodall and Long, 1968). The
fractionation factor of the hydrogens on the nitrogen in glycine are probably not
important since the isotope effect is the same for A/,/V-dimethylglycine (Table 3-
3).
The importance of outer shell water in ligand substitution reactions is not
without precedent. The first step in the proposed mechanism of ligand
substitution reactions is the formation of an outer shell complex with the ligand
and metal (Eigen and Tamm, 1962), followed by replacement by the ligand of a
water molecule in the inner coordination shell of the metal. Hence it is
reasonable to assume that the binding of a ligand would affect the structure of
outer shell water.
The following example illustrates how the fractionation of outer shell
water could contribute to the isotope effect. Equation 3-7 can be expanded to

38
include the fractionation factor of outer shell water of Co(H20)62+ in the
numerator and of Co(H20)4L+ in the denominator. The number of outer shell
waters would be larger than the number of inner shell waters, and if the
contribution of an outer shell hydrogenic site in equation 3-7 was 1.06, twelve
outer shell waters (24 hydrogenic positions) would contribute (1.06)24 = 4.05.
This would then mean that the fractionation factor of the inner shell waters in
Co(H20)4l_+ is 0.74. If the contribution of an outer shell hydrogenic site is 1.10,
then the fractionation factor of the inner shell waters in Co(H20)4L+ is 0.83.
Thus a contribution of outer shell water slightly larger than unity would, when
raised to a large power for the number of hydrogenic positions, cancel the effect
of the inner shell waters.
I conclude that the solvent hydrogen isotope effect of unity on the
formation of complexes of Co2+ and glycine, A/,A/-dimethylglycine, acetylacetone
points to the importance of outer shell water in these reactions. Conversely,
one can not get information about the effect of added ligands on the
fractionation properties of the inner shell waters from the solvent hydrogen
isotope effect on the formation constants. This work shows that the solvent
hydrogen isotope effect is the complex contribution of the hydrogen / deuterium
fractionation of several different groups or species of water.

CHAPTER 4
HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF THE
AQUEOUS LIGAND OF COBALT IN Co(ll)-SUBSTITUTED
CARBONIC ANHYDRASE
Introduction
Carbonic anhydrase is a zinc metalloenzyme which catalyzes the
reversible hydration of carbon dioxide to produce bicarbonate and a proton.
CO2 + H2O ~- HC£>+ H+
The aqueous ligand of the metal at the active site of carbonic anhydrase is
believed to have a role in the catalytic pathway. Direct nucleophilic attack by
the metal-bound hydroxide on CO2 is a likely step in the production of
bicarbonate (Pocker and Sarkanen, 1978; Lindskog, 1983; Silverman and
Lindskog, 1988), and an intramolecular proton transfer involving the metal-
bound water is proposed to be a rate-limiting step in catalysis by isozyme II
(Steiner etai, 1975). The solvent hydrogen isotope effect associated with the
turnover number for CO2 hydration catalyzed by native zinc-carbonic anhydrase
II is 3.8 (Steiner et a/., 1975). To interpret fully the solvent hydrogen isotope
effects observed for catalysis by carbonic anhydrase requires knowledge of the
fractionation factor of the aqueous ligand of the metal.
The fractionation factor measures the tendency of deuterium to
accumulate at the aqueous ligand of the metal relative to the deuterium content
of bulk solvent. There are few reports of the fractionation factor of the aqueous
ligands of a metal at the active site of an enzyme. Using NMR methods,
Silverman (1981) has measured such a factor for the water ligands of cobalt in
cobalt(ll)-substituted carbonic anhydrase II ((}> = 1.05 0.17) by taking
3 9

40
advantage of the properties of a paramagnetic metal to enhance the relaxation
rate of water exchanging rapidly between bulk solvent and the hydration shell of
the metal and measuring the proton relaxation rate as a function of the
deuterium content of solvent. Silverman measured that fractionation factor
using T-i at a proton resonance frequency of 100 MHz. Because of the unique
relaxation properties of water protons in solutions of Co(H20)62+ at 300 MHz
(Chapter 2), I have determined the fractionation factor of the aqueous ligand of
cobalt in three mammalian isozymes of Co(ll)-substituted carbonic anhydrase
from T-\ and T2 at 300 MHz. In each case T1 > T2( implying that the same
relaxation mechanisms occur for water protons in solutions of Co(ll)-substituted
carbonic anhydrase at 300 MHz as in solutions of Co(H20)62+. The values of
the fractionation factor determined from T-i for each isozyme of Co(ll)-substituted
carbonic anhydrase were close to the fractionation factor for bulk water, which is
unity. Except in the case of Co(ll)-substituted carbonic anhydrase III, the
fractionation factors obtained from T2 were less than unity and close to the
fractionation factor for H30+ which is 0.69. I conclude that fractionation factors in
these cases determined from T-i and T2 measured isotope preferences for
different populations of ligand sites. I suggest that since T2 has a large
contribution from a paramagnetic chemical shift, the fractionation factors
determined from T2 have a large contribution of the hydrogen / deuterium
fractionation of the inner shell ligand. The fractionation factors determined from
T-i are close to unity, the value in bulk water, and contain a larger contribution of
the fractionation of outer shell water.
Experimental Procedure
Enzymes: Human carbonic anhydrase I was purified from human red
blood cells by an affinity chromatography method (Kalifah et al., 1977). Bovine
carbonic anhydrase II was obtained from Sigma Chemical and used after

4 1
extensive dialysis to remove paramagnetic impurities. Bovine carbonic
anhydrase III was obtained from bovine flank steak by gel filtration and anion-
exchange chromatography (Tu et al., 1986). The concentration of carbonic
anhydrase I, II, and III was estimated at 280 nm using 8 = 4.7 x 104 M'1 cm-1
(Nyman and Lindskog, 1964), 5.4 x 104 M1 cm*1 (Nyman and Lindskog, 1964),
and 6.4 x 104 M-1 cm-1 (Engberg and Lindskog, 1984), respectively.
The apoenzymes of carbonic anhydrase I and II were prepared
according to the procedure of Hunt et al. (1977) by dialysis against dipicolinic
acid. Co(ll) substituted carbonic anhydrase I and II were prepared by the
addition of 1.1 equivalents of C0CI2 followed by dialysis against several
changes of large volumes of deionized water. The apoenzyme of carbonic
anhydrase III was prepared by addition of the chelator 2-carboxy-1,10-
phenanthroline to a solution of enzyme followed by dialysis against excess
C0CI2 as described by Engberg and Lindskog (1984).
NMR Measurements: Measurements at 300 MHz were the same as in
Chapter 2. In the cases of Co(ll)-substituted carbonic anhydrase I and II, 1/T¡0
was taken as the relaxation of solutions of the Co(ll)-substituted enzyme in the
presence of a molar excess of the inhibitor acetazolamide, which is known to
displace the aqueous ligand of the metal at the active site. Enough
acetazolamide was present to bind to 99.9% of the active sites. Alternatively for
Co(ll)-isozymes I and II and solely for Co(ll)-substituted carbonic anhydrase III,
1/Tj0 was taken as the relaxation of solutions of the native zinc enzyme, which
has no paramagnetic contribution. These two methods were equivalent at a
frequency of 300 MHz for Co(ll)-substituted carbonic anhydrase I and II.
Solutions of enzyme for relaxation measurements were buffered with 50 mM
Hepes [4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid] at pH 8.5. At this
pH the relaxivity of solutions of Co(ll)-substituted carbonic anhydrase I and II is

greatest and in a region that is independent of pH (Wells et al., 1979). (See
Chapter 6 for a discussion of the pH and magnetic field dependence of the
relaxation of water protons in solutions of Co(ll)-substituted carbonic anhydrase
III.) The values of pH reported here are uncorrected pH meter readings. The
correction of a pH meter reading in 100% D2O (pD = meter reading + 0.4) is
approximately offset by the change in ionization state of the buffer in D2O (pKq2o
- pKh2o = 0.5 0.1 for almost all acids with pK values between 3 and 10,
Schowen, 1978). Water used for solution preparations was distilled and
passed through two ion-exchange resin cartridges (Cole-Palmer 1506-35).
D20 (99.8%) was stirred with activated charcoal, the charcoal filtered out, and
then the D2O was distilled.
Measurement of the Hydrogen / Deuterium Fractionation Factor; The
hydrogen / deuterium fractionation factor was determined by measuring the
relaxation of water protons in solutions of Co(ll)-substituted carbonic anhydrase
as a function of the atom fraction of deuterium in solvent water as described in
Chapter 3.
Results
The substitution of cobalt for zinc at the active site of carbonic anhydrase
provides a paramagnetic center which affects the relaxation of water protons in
solutions of carbonic anhydrase. The paramagnetic relaxivity of these water
protons is greater for R2 than for R1 for each isozyme of carbonic anhydrase
(Table 4-1).
The relaxation of water protons in solutions of the three isozymes of
Co(ll)-substituted carbonic anhydrase as a function of the atom fraction of
deuterium in solvent water is shown in Figures 4-1,4-2, and 4-3. The hydrogen
/ deuterium fractionation factors calculated from equation 3-4 using T-\ were
near unity for the three isozymes (Table 4-2). However, the fractionation factors

Table 4-1: Paramagnetic contribution to the relaxivities of water protons
observed for aqueous solutions of Co(ll)-substituted carbonic anhydrase I, II,
and III.
frequency
(MHz)
R1
(mM'V1)
r2
(mM-'s-1)
T1p/T2p
Co(ll)-carbonic anhydrase I
300
0.28
2.3
8.2
Co(ll)-carbonic anhydrase II
300
0.26
1.4
5.3
Co(ll)-carbonic anhydrase III
300
0.14
2.2
16
Note: Measurements were made at 23 C for solutions of enzyme buffered at
pH 8.5 by 50 mM Hepes. Solutions contained 20% D2O by volume.

44
Atom fraction of deuterium in solvent (n)
Figure 4-1: The dependence of p'Tip (i=1,2) at 300 MHz on n, the atom fraction
of deuterium in solvent water, for 1.5 mM solutions of Co(ll)-substituted carbonic
anhydrase I. Solutions contained 50 mM Hepes buffer at pH 8.5 and 23C.
Data are the means and standard errors of three measurements.

4.1
Atom fraction of deuterium in solvent (n)
Figure 4-2: The dependence of p'Tip (i=1,2) at 300 MHz on n, the atom fraction
of deuterium in solvent water, for 2.0 mM solutions of Co(ll)-substituted carbonic
anhydrase II. Solutions contained 50 mM Hepes buffer at pH 8.5 and 23C.
Data are the means and standard errors of three measurements.

46
Atom fraction of deuterium in solvent (n)
Atom fraction of deuterium in solvent (n)
Figure 4-3: The dependence of pTip (i=1,2) at 300 MHz on n, the atom fraction
of deuterium in solvent water, for 1.5 mM solutions of Co(ll)-substituted carbonic
anhydrase III. Solutions contained 50 mM Hepes buffer at pH 8.5 and 23C.
Data are the means and standard errors of three measurements.

Table 4-2: Hydrogen/deuterium fractionation factors determined from Ti and T2
at 300 MHz for the aqueous ligand of the paramagnetic metal in three
mammalian isozymes of Co(ll)-substituted carbonic anhydrase.
T1
$T2
Co(ll)-carbonic anhydrase I
1.10.02
0.72.02
Co(ll)-carbonic anhydrase II
0.95.02
0.77.01
Co(ll)-carbonic anhydrase III
1.03.06
1.00.07
Note: Data were obtained from the least squares slope and intercept and
standard error in a plot such as Figure 4-1 according to equation 3-4.
Measurements were made at 23 C for solutions of enzyme buffered at pH 8.5
by 50 mM Hepes.

48
calculated using T2 were significantly less than unity for isozymes I and II, but
near unity for isozyme III (Table 4-2).
Discussion
The relaxation behavior of water protons at 300 MHz and pH 8.5 in
solutions of the Co(ll)-substituted carbonic anhydrases showed Tip/T2P much
greater than unity (Table 4-1). This was the same as observed for the water
protons in solutions of cobalt ions (see Chapter 2), and the interpretation in the
case of Co(H20)62+ can be applied to Co(ll)-substituted carbonic anhydrase. At
300 MHz, there is a field dependent increase in the chemical shift difference
between a proton in the metal-bound site and the free solvent site. This causes
an increase in the transverse relaxation of the protons as they experience the
two environments (equation 2-3). Because this relaxation mechanism has as its
source a paramagnetic contact shift, the transverse relaxation has a larger
contribution of the ligand in the inner coordination shell of the cobalt, while the
longitudinal relaxation has a larger contribution of water in the outer
coordination shell. Thus the fractionation factor measured from T2 represents
more of the fractionation of the inner shell aqueous ligand of cobalt in Co(ll)-
substituted carbonic anhydrase, while the fractionation factor measured from T-i
is a greater contribution of the fractionation of outer shell water. A similar
situation in which the outer shell relaxation has a larger contribution to than
to R2 has been reported by Kushnir and Navon (1984) for water protons in
solution with Mn(ll)-substituted carbonic anhydrase II.
These relaxation measurements were made at pH 8.5, a pH for which the
relaxation rate of water protons in solutions of Co(ll)-substituted carbonic
anhydrase I and II is greatest (Wells etal., 1979). (See Chapter 6 for an
investigation of this property in solutions of Co(ll)-substituted carbonic
anhydrase III.) This pH is above the value of the pK for the ionization of the

active site group as determined in many experiments (see Table 5-1 and
Engberg and Lindskog, 1984), especially spectrophotometric titrations and
activity measurements, and the ligand of cobalt is believed to be a hydroxide
ion at this pH. This presents some uncertainty as to which protons are actually
being relaxed at pH 8.5 since a tetracoordinate cobalt-bound hydroxide ion
would not be expected to exchange rapidly with water in solution. This issue
has been discussed by Koenig et al. (1983) who proposed that ligand
exchange with solvent involved a pentacoordinate intermediate having both
OH' and H20 as ligands. Proton exchange can be rapid between these two
ligands so that the departing H20 can contain the oxygen of the initially-bound
OH*. On the basis of 180-exchange kinetics, Tu and Silverman (1985)
suggested that the NMR relaxation observed at pH 8.5 is that of a water
molecule hydrogen bonded to the cobalt-bound hydroxide yet close enough to
the metal to be strongly relaxed. Yet another possibility is the rapid exchange of
the hydrogen of the metal-bound hydroxide without exchange of the oxygen.
Because of these unresolved issues, I cannot state specifically which of the
hydrogens near the metal are the main contributors to the observed
fractionation factors.
The fractionation factors measured from Ti for each isozyme of Co(ll)-
substituted carbonic anhydrase were close to unity (Table 4-2), like (})t1 for
Co(H20)62+ (Table 3-2), consistent with these values having a large contribution
from the fractionation of outer shell water. This is reasonable from distance
considerations, as it can be calculated using a 1/r6 dependence that the
paramagnetic contribution to the longitudinal relaxation rate for a cobalt-bound
hydroxide with a cobalt-proton distance of 2.8 is about the same as the
paramagnetic contribution of the protons of three water molecules at an
average distance of 3.8 . Also, it is known from crystallographic data that there

50
are a large number of water molecules within the active site cavity. Specifically,
in human carbonic anhydrase II there are an estimated 9 water molecules
between histidine 64 and the zinc, a distance of about 6 from the x-ray
coordinates (Erickson et al., 1986). The fractionation factors measured from T2,
isozyme III (Table 4-2), and reflect a larger contribution of the hydrogen /
deuterium fractionation of the inner shell aqueous ligand. It is difficult to
account for these differences, however, and though these isozymes have very
similar amino acid sequences, they have unique active site and kinetic
properties (Silverman and Vincent, 1983; Silverman and Lindskog, 1988).
Isozyme II has the highest C02 hydration activity, with a turnover number of 106
S'1. There is a histidine in position 64 which protrudes into the active site cavity
of isozyme II. Isozyme I has a turnover number of 2 x 105 S'1, and has His-64 as
well as two other histidines, 67 and 200, in the active site cavity, and it is known
that histidine 200 interacts strongly with the ligands of zinc (Kalifah, 1977). The
active site cavity of isozyme I is smaller compared with that of isozyme II.
Isozyme III has a turnover number of 104 s*1 and has a significant amount of
positive charge in the active site, with a lysine in position 64 and an arginine in
position 67. This positive charge may influence the ionization properties of the
metal bound water in carbonic anhydrase III, for which the pK of the metal-
bound water is less than 5.5, compared with isozyme I and II which have a pK
around 7 (Tu etai, 1983; Engberg and Lindskog, 1984). This difference in
charge in the active site may also affect the hydrogen / deuterium fractionation
properties of the ligand of the metal in Co(ll)-substituted carbonic anhydrase III.
Although isozymes I and II have very similar amino acid sequences, the
significant differences in their structure, residues in the active site cleft, and
catalytic activities (Pocker and Sarkanen, 1978; Lindskog, 1983; Silverman and

Undskog, 1988) apparently do not have a significant effect on the fractionation
factors which were nearly identical for isozymes I and II.

CHAPTER 5
INTERPRETATION OF SOLVENT HYDROGEN ISOTOPE
EFFECTS ASSOCIATED WITH CARBONIC ANHYDRASE
Introduction
A solvent hydrogen isotope effect on a reaction is made up of individual
contributions to the overall isotope effect from each hydrogenic site in the
reaction (Schowen, 1978). This contribution is measured by the hydrogen /
deuterium fractionation factor of each particular hydrogenic site. In considering
reactions of carbonic anhydrase, the fractionation factor of the aqueous ligand
of the metal represents a reactant state contribution that can be used to interpret
equilibrium solvent hydrogen isotope effects which are a function of reactant
and product state site fractionation factors, and kinetic isotope effects which are
a function of reactant and transition state site fractionation factors.
Halide ions are inhibitors of carbonic anhydrase which bind to the metal
and displace the water ligand of the metal (Bertini et al., 1982). I have
determined the solvent hydrogen isotope effect on the equilibrium binding of
iodide ion to Co(ll)-substituted carbonic anhydrase I and II and interpreted this
effect using the fractionation factor of the aqueous ligand of cobalt in Co(ll)-
substituted carbonic anhydrase measured from NMR relaxation rates. The
results suggest that the isotope effect on T binding binding is the complex
contribution of the fractionation of more than one hydrogenic position, because
the consideration of the fractionation factor of the aqueous ligand of cobalt by
itself does not allow a complete interpretation of the isotope effect.
The fractionation factor of the aqueous ligand of cobalt in Co(ll)-
substituted carbonic anhydrase can also be used to interpret kinetic isotope

53
effects, and because a kinetic isotope effect includes the contribution of
transition state hydrogenic sites, this allows a prediction of the transition state
structure. In the case of Co(ll)-substituted carbonic anhydrase II the
fractionation factor measured from NMR leads to a predicted transition state
structure that includes a proton transfer to a water molecule in the active site
cavity. This is consistent with the large isotope effect on kcat (Steiner et al.,
1975) and the nonlinear dependence of kcat on the atom fraction of deuterium in
solvent water (Venkatasubban and Silverman, 1980).
Experimental Procedure
The equilibrium dissociation constants describing the binding of iodide
ion to Co(ll)-substituted carbonic anhydrase were determined from visible
absorption measurements (Lindskog, 1966). The pH dependence of the
absorbance at 640 nm in solutions of Co(ll)-substituted carbonic anhydrase I
and II was measured on a Beckman DU-7 spectrophotometer. Stock solutions
of enzyme at low pH contained 12.5 mM Hepes and 25 mM Mops (4-
morpholinepropanesulfonic acid) buffer, and at high pH 0.5 M of
triethylenediamine. The stock solutions had equal concentrations of enzyme.
The titrations were performed by adding small increments of the high pH
enzyme solution to the low pH enzyme solution. In this way the enzyme
concentration stayed the same though the volume increased slightly. These
titrations were done in H2O and D20 in the presence and absence of iodide ion
which has been shown to bind to the metal site in carbonic anhydrase II (Brown
et al., 1977).
In the absence of anion, the data were fit (RS1, BBN Software,
Cambridge, MA) to a scheme involving 4 micro equilibrium constants
(Simonsson and Lindskog, 1982; Bertini et al., 1985). In this scheme, only 3 of
the 4 microconstants are independent, that is KtK3 = K2K4 (Scheme 5-1). When

54
+HCoOH'
+HECoOH2 ECoOH"
ECoOH2
2
Scheme 5-1: A diagram of two significant ionizations at and near the active site
of Co(ll)-substituted carbonic anhydrase II. One is the ionization of metal-bound
water and the second is suggested to be the ionization of His-64 (Simonsson
and Lindskog, 1982), the imidazole ring of which is about 6 from the metal.
For Co(ll)-substituted carbonic anhydrase I the second ionization could be
either that of His-64 or His-200 (Whitney and Brandt, 1976).

55
anion was present, Scheme 5-1 was expanded to include the binding of the
inhibitor, and it was assumed that the anion binds only to species 1 and 2 (see
Tibell etal. (1984) for an estimate of the accuracy of this assumption). Thus two
separate binding constants were determined. K-i1' is the equilibrium
dissociation constant of the anion complex with species 1 of Scheme 5-1, and
K2i_ is the equilibrium dissociation constant of the anion complex with species 2.
Results
In the absence of anion, the visible pH titration data fit Scheme 5-1 very
well (Table 5-1). The constants K-i, K2, and K3 were determined from a fit to the
data, and then K4 was calculated from K-i, K2, and K3. The 4 microconstants
determined in H20 for Co(ll)-substituted carbonic anhydrase II agreed very well
with those obtained by Simonsson and Lindskog (1982), and the values
determined in H20 for Co(ll)-substituted carbonic anhydrase I also agreed fairly
well with those obtained by Bertini et al. (1985). In the presence of iodide the
pH dependence of the absorbance at 640 nm was shifted to higher pH. Iodide
binds with greater affinity to the diprotonated species 1 in Scheme 5-1 than to
the monoprotonated species 2 (Table 5-2). This is similar to the iodide
inhibition results of Simonsson and Lindskog (1982) for native zinc carbonic
anhydrase II determined from activity measurements, and the results of Bertini
et al. (1985) for nitrate inhibition of Co(ll)-substituted carbonic anhydrase II
determined from spectrophotometric titrations. The change in solvent from H20
to D20 has no measurable effect on K^'and K2I_ for Co(ll)-substituted carbonic
anhydrase I, but for Co(ll)-substituted carbonic anhydrase II the change in
solvent from H20 to D20 results in a decrease in these equilibrium constants
(Table 5-2).

Table 5-1: Solvent hydrogen isotope effects on the acid-base microequilibrium
constants observed for Co(ll)-substituted carbonic anhydrase I and II.
o
jD
1
o
0>
cr
o
13
o'
anhydrase I
Co(ll)-carbonic anhydrase II
value in H2O
ApKa
value in H2O
ApKa
pKi
7.55 0.05
0.54 0.06
5.65 0.04
0.42 0.05
pK2
7.58 + 0.15
0.45 0.19
5.73 0.07
0.37 0.10
pK3
8.11 0.15
0.45 0.19
6.95 0.08
0.38 0.12
pK4
8.09 0.22
0.54 0.28
6.87 0.12
0.43 0.17
Note: The microconstants were determined from the variation of the
absorbance at 640 nm with pH at 23C and fit to Scheme 5-1. The solutions
were buffered with Mops, Hepes and triethylenediamme with no other added
ions as described in the experimental procedure.
a ApK = (pK¡)d2o (pK¡)h2o

57
Table 5-2: Solvent hydrogen isotope effects on the equilibrium dissociation
constants of iodide ion with Co(ll)-substituted carbonic anhydrase I and II.
Co(ll)-carbonic anhydrase I
value in H2O ApK1'3
Co(ll)-carbonic anhydrase II
value in H2O ApK1'3
3.48 0.13 0.47 0.19
pK-i1' 2.95 0.10 -0.09 0.17
pK2z' 2.35 0.06 -0.07 0.11
2.79 0.02 0.27 0.02
Note: The dissociation constants were determined from the pH dependence of
the absorbance at 640 nm in the presence of 10 mM T with conditions as in
Table 5-1. K11' is the dissociation constant of T binding to species 1 of Scheme
5-1, and K21' is the dissociation constant of I- binding to species 2 of Scheme 5-
1.
a
ApK = (pKj^OgO (pKj^HgO

58
Discussion
I have discussed in Chapter 4 the fractionation factor for the
exchangeable hydrogen at the active site of Co(ll)-substituted carbonic
anhydrase in its high pH form. The first application of the fractionation factors
determined in Chapter 4 is to calculate the fractionation factor of the active site
in its low pH form. Since the paramagnetic contribution to the relaxivity
decreases to a very small value below pH 6.0 (Wells et ai, 1979), except at very
low ionic strength (Bertini eta!., 1978; Bertini et ai, 1980), the fractionation
factor cannot be measured directly from the relaxation rates at low pH. Instead,
this was calculated from the fractionation factor of the high pH form and the
solvent hydrogen isotope effect on the ionization constant of the active site,
metal-bound water.
The visible pH titration data of isozymes I and II indicate that more than
one ionization at or near the active site governs the absorbance at 640 nm
(Bertini et ai, 1980). Previous investigators (Simonsson and Lindskog, 1982;
Bertini et ai, 1985) have attributed this pH dependence to the ionization of two
groups, the metal-bound water and another group near the active site which
may be His-64 for isozyme II and probably His-64 or His-200 in isozyme I
(Kalifah, 1977; Whitney and Brandt, 1976). I have measured the pH
dependence of the absorbance at 640 nm for isozymes I and II in H2O and D2O
and fit the data according to Scheme 5-1 to determine the solvent hydrogen
isotope effect on the microequilibrium constants (Table 5-1). In Scheme 5-1, K-i
and K4 are the constants for the metal-bound water, and 2 and K3 are the
constants for the other active site group. Since the solvent hydrogen isotope
effects on these equilibrium constants are all similar, within experimental
uncertainty, it appears that the protonation state of one of the groups does not
affect the hydrogen / deuterium fractionation properties of the other (consistent

5 9
with the conclusion that active site properties do not seem to influence the
fractionation factors of the aqueous ligand of cobalt in Co(ll)-substituted
carbonic anhydrase I and II, see Chapter 4). Then the calculation to determine
the fractionation factor of the water ligand of the metal can be performed
neglecting the ionization state of the second ionizable group.
The isotope effect on an equilibrium constant is the product of all of the
reactant state fractionation factors divided by the product of all of the product
state fractionation factors (equation 1-2). For this ionization at the active site of
Co(ll)-substituted carbonic anhydrase
^Co OH2 +H20 3;Co OH +H30+
the solvent hydrogen isotope effect can be written
Khp ^Co-OHj^HgO)2
2
KD20 (^Co-OH )
From this measured isotope effect, the measured fractionation factor for the
cobalt-bound hydroxide, the fractionation factor for H3 measured to be 0.69 (Chiang, 1980), and the fractionation factor for H2O which
is 1.0, the fractionation factor for the exchangeable hydrogens at the active site
in its low pH form can be calculated and are presented in Table 5-3. I have
done the calculations of Table 5-3 using both T2 measured for the
high pH form of the aqueous ligand of cobalt in Co(ll)-substituted carbonic
anhydrase I and II. However, the calculation from T-i may not be meaningful,
since it contains a large contribution of outer shell water. The value calculated
using shell cobalt-bound water.

60
Table 5-3: Values of the hydrogen / deuterium fractionation factor for the low pH
forms of Co(ll)-substituted carbonic anhydrase.
4>ti
t2
Co(ll)-substituted carbonic anhydrase 1
1.12 0.05
0.90 0.04
Co(ll)-substituted carbonic anhydrase II
0.90 0.03
0.81 0.03
Note: The § for the low pH form was calculated using the isotope effect on the
ionization constant of the cobalt-bound water and the $ for the high pH form
measured directly from NMR and given in Table 4-2.

To apply the fractionation factors to an equilibrium solvent hydrogen
isotope effect associated with the enzyme, I have studied at equilibrium the
binding of I" to Co(ll)-substituted carbonic anhydrase I and II. I' is known to
bind to the metal in Co(ll)-substituted carbonic anhydrase I and II displacing the
coordinated water molecule (Brown et ai, 1977). Also, I" binds fairly tightly
(Table 5-2) and its size is comparable to that of a water molecule. Thus the
isotope effect on the binding of I to Co(ll)-substituted carbonic anhydrase may
be simple enough to be accounted for by the fractionation factor of the metal-
bound water.
The solvent hydrogen isotope effect on the binding of the iodide ion to
the active site of carbonic anhydrase is given by
^Co-I + H20 ^Co-OHI '
(fco-OH/
where (K¡I')h2o is the equilibrium dissociation constant describing the binding of
I- to species i of Scheme 5-1 in H20. Because it is known that (|)h2o = 1-0, with
knowledge of (¡>t2 for Co-OH2 from Table 5-3,1 thus predict that ApK1' = 0.09 0.04 for Co(ll)-
isozyme I, and 0.18 0.03 for Co(ll)-isozyme II. The predicted value for isozyme
I is in rough agreement with the measured value (Table 5-2), and the predicted
value for isozyme II is slightly below the measured value. However, the
predicted value for isozyme I is less than the predicted value for isozyme II,
which is the same as in the experimental case. I have considered other
interactions which may be taken into account for r binding. One of these is the
composite fractionation factor of the solvation shell of T (0 = 0.59, Voice, 1974).

6 2
Including this in the denominator of the above equation for the isotope effect
increases the predicted value, which is then closer to the experimental value for
Co(ll)-isozyme II, but farther from the experimental value for Co(ll)-isozyme I.
This implies that I have incomplete knowledge of the fractionation properties of
each hydrogenic site that is affected when iodide ion binds to the enzyme.
However, to include the fractionation factors of any other groups in the equation
requires much speculation about the fractionation factors near the active site.
Thus it is likely that the isotope effect must be described by the complex
contribution of the fractionation factors of additional groups or water associated
with the enzyme, and that the fractionation factor of the aqueous ligand of the
metal is by itself not sufficient to account for the isotope effect on I* binding.
The discussion of the solvent hydrogen isotope effect on iodide binding
suggests that the fractionation factor can be used to interpret at least
qualitatively isotope effects associated with carbonic anhydrase. It is also
possible to apply the fractionation factors for the aqueous ligand of the metal in
Co(ll)-substituted carbonic anhydrase in the interpretation of isotope effects on
kinetic constants associated with the enzyme to see if similar information can be
obtained. A kinetic isotope effect is the product of the reactant state
fractionation factors divided by the product of the transition state fractionation
factors (equation 1-1). The isotope effect on kcat for the CO2 hydration activity of
Co(ll)-substituted carbonic anhydrase II is 1.8 (R. S. Rowlett, unpublished), and
the rate limiting step measured by kca, is the intramolecular transfer of a proton
from the metal-bound water to regenerate the metal-bound hydroxide for the
next round of catalysis (Steiner et at., 1975). This proton transfer from the
metal-bound water is thought to go through a water-bridge to histidine-64 in the
active site cleft, from which the proton can then be released to solvent

63
(Venkatasubban and Silverman, 1980). For this individual step in the catalysis,
the reaction is
Co-OH2 + H20 + His Co-OH + H20 + His-H+
and the isotope effect is
kH20 ^Co-OH2) ^h2o)
u
D20 all transition state sites
n i
The hydrogenic sites of water have a fractionation factor of 1.0, and the
hydrogenic sites of the cobalt-bound water have a fractionation factor of 0.81
(Table 5-3). Using these values and the isotope effect in the above equation,
the product of the transition state fractionation factors can be calculated to be
0.36. In the transition state, the proton remaining on the metal-bound oxygen is
in transition between a metal bound water which has a fractionation factor of
0.81 (Table 5-3), and a metal-bound hydroxide which has a fractionation factor
of 0.77 (Table 4-2), so I will arbitrarily use a value of 0.79 as the fractionation
factor of this hydrogenic site in the transition state. Therefore, the product of the
fractionation factors of the other hydrogenic sites in the transition state is
0.36/0.79 = 0.46. These other hydrogenic sites in the transition state are the
positions associated with the water bridge between the metal-bound water and
the histidine in the active site cleft.

64
n\
%/NH
r=\
V/NH
This calculation for the fractionation factor of the water bridge in the transition
state suggests that the protons of the water bridge have a fractionation factor
less than unity, meaning they have hydronium ion character, consistent with
positive charge transferred to the oxygen during the proton transfer process.
This calculation of the fractionation factor of the transition state suggests that the
proton transferred from the cobalt-bound water is in between the cobalt-bound
water and the water bridge in the transition state. As in the case of iodide
binding to Co(ll)-substituted carbonic anhydrase, there may be other
hydrogenic sites in the transition state which contribute to the overall isotope
effect. However, the fractionation factors of the aqueous ligand of the metal
have led to this qualitative model of the transition state, which is consistent with
both the large isotope effect on kcat and the results of Venkatasubban and
Silverman (1980) for the catalysis of CO2 hydration in H20 / D20 mixtures.
X H
Co-OH2 0
H
X ,H'"Ox
-Co-<3 'H1
/ H

CHAPTER 6
MAGNETIC FIELD AND pH DEPENDENCE OF THE WATER
PROTON NMR RELAXATION ENHANCEMENT IN SOLUTIONS
OF Co(ll)-SUBSTITUTED CARBONIC ANHYDRASE III
Introduction
Carbonic anhydrase III is a mammalian isozyme of carbonic anhydrase
found primarily in skeletal muscle. It is a less efficient catalyst of CO2 hydration
compared with isozymes I and II, which are found in red blood cells (Lindskog,
1983). Another distinctive feature of carbonic anhydrase III is the pH
independence of its catalytic activity in the range pH 6-9, in which carbonic
anhydrase II has an inflection at pH 7 and maximal activity at high pH (Tu etal.,
1983; Kararli and Silverman, 1985; Tu etal., 1986). It is also apparent that the
zinc is bound more tightly in carbonic anhydrase III than in isozymes I and II
(Engberg and Lindskog, 1984).
The replacement of cobalt for zinc in carbonic anhydrase I and II has
proved useful to study the properties of the metal at the active site (Bertini and
Luchinat, 1983). The catalytic and visible spectral properties of Co(ll)-
substituted carbonic anhydrase III have been investigated (Engberg and
Lindskog, 1984). I report here the NMR relaxation enhancement of water
protons in solutions of Co(ll)-substituted carbonic anhydrase III. I found, in
contrast to Co(ll)-substituted carbonic anhydrase I and II, no pH dependence of
the NMR relaxation in the range of pH 6-9. Also, the frequency dependence of
the paramagnetic contribution to the relaxation shows an increase with higher
frequency, as opposed to the dispersion to lower relaxation rates observed in
solutions of Co(ll)-substituted carbonic anhydrase I and II.
65

66
Experimental Procedure
The purification of bovine carbonic anhydrase III and preparation of
Co(ll)-substituted carbonic anhydrase III was performed as described in
Chapter 4. The total concentration of carbonic anhydrase III was estimated at
280 nm by using e = 6.4 x 104 M'1 crrr1 (Engberg and Lindskog, 1984). The
concentration of Co(ll)-substituted carbonic anhydrase III was estimated at 640
nm by using the extinction coefficients determined by Engberg and Lindskog
(1984). Zinc analysis of the native enzyme yielded 1.03 0.05 zinc per
molecule of enzyme. Cobalt analysis of the Co(ll)-substituted enzyme yielded
0.33 0.04 cobalt per molecule of enzyme, with the remainder apoenzyme.
The fraction of the sample that was apoenzyme was restored to full activity by
the addition of zinc.
The longitudinal relaxation rates of the protons of solvent water in
solutions of Co(ll)-substituted and native Zn-carbonic anhydrase III were
measured using afield cycling "relaxometer" (Koenig et ai, 1983) at the IBM T.
J. Watson Research Center in Yorktown Heights, New York. The rates were
measured at 5C at pH 6.0 (50 mM Mes/NaOH buffer), pH 7.0 (50 mM
Mops/NaOH buffer), pH 8.0 (50 mM Hepes/NaOH buffer), and pH 9.0 (50 mM
Ches/NaOH buffer) in solutions containing 1 mM of enzyme. The pH 6 samples
were also measured at 25C. The samples were in equilibrium with
atmospheric CO2. The total relaxivity, Rt, is defined as in equation 6-1
Rt = (1/Ti 1/Tio)/N (6-1)
where 1/Ti is the observed relaxation rate of the protein solution, 1 /T1o is the
relaxation rate of the buffer, and N is the millimolar concentration of protein. At
each pH the relaxivity was determined as a function of frequency over the range

6 7
0.01 to 50 MHz. The paramagnetic relaxivity, Rp, is Rt determined for a solution
of Co(ll)-substituted carbonic anhydrase III minus R, for a solution of native Zn-
carbonic anhydrase III.
Re?UltS
The frequency dependence of the total relaxivity was similar to the NMR
dispersion profiles of other native and Co(ll)-substituted carbonic anhydrases
(Fabry etal., 1970; Wells etal., 1979) (Figure 6-1). The total relaxivity is
greatest at low frequency, then begins to decrease at about 0.1 MHz. The
paramagnetic relaxivity, however, showed a small value at low frequency, then
at about 0.5 MHz Rp began to increase to a maximal value at about 5 MHz, and
then began to decrease again (Figure 6-2). The total relaxivities for both the
Co(ll)-substituted and native Zn-carbonic anhydrase III showed a possible
increase with pH at 1 MHz, but remained constant with pH at 40 MHz (Figure 6-
3). This may be the same effect reported by Wells etal. (1979), who observed
that the diamagnetic component of the relaxivity in solutions of Co(ll)-substituted
and native zinc carbonic anhydrase II increases with pH at low frequency.
When only the paramagnetic relaxivity was considered (the difference between
Rt of Co(ll)-substituted carbonic anhydrase III and Rt of native zinc carbonic
anhydrase III shown in Figure 6-3), there was no pH dependence, especially
when the data at pH 7 in Figure 6-3 are excluded.
Discussion
This work is the first study of NMR relaxation enhancement of water
protons in solutions of Co(ll)-substituted carbonic anhydrase III. Previous work
with isozyme III has shown that the visible spectrum of Co(ll)-substituted
carbonic anhydrase III is very similar to the high pH spectral forms of Co(ll)-
substituted carbonic anhydrase I and II; however, for Co(ll)-substituted carbonic
anhydrase III there is no pH dependence of the spectrum in the range pH 6-9

68
Figure 6-1: The total relaxivity of Co(ll)-substituted carbonic anhydrase III and
native zinc carbonic anhydrase III as a function of the proton resonance
frequency. The solutions contained 1 mM enzyme at pH 6.0 (50 mM Mes /
NaOH), pH 8.0 (50 mM Hepes / NaOH), or pH 9.0 (50 mM Ches / NaOH). pH
6.0 Co(ll)-substituted carbonic anhydrase III at 5C ( a), pH 6.0 native zinc
carbonic anhydrase III at 5C ( ), pH 9.0 Co(ll)-substituted carbonic anhydrase
III at 5C (), pH 9.0 native zinc carbonic anhydrase III at 5C ( ), pH 8.0
Co(ll)-substituted carbonic anhydrase III at 23C (o ), pH 8.0 native zinc
carbonic anhydrase III at 23C ( ).

Paramagnetic relaxivity (mM'1 s'1)
69
Figure 6-2: The paramagnetic relaxivity of Co(ll)-substituted carbonic
anhydrase III as a function of the proton resonance frequency. Experimental
conditions as in Figure 6-1. pH 6.0 ( A), pH 9.0 ().

70
1.0
0.8
0.6
0.4
0.2
0.0
5.0 6.0 7.0 8.0 9.0 10.0
pH
Figure 6-3: The total relaxivity of Co(ll)-substituted carbonic anhydrase III and
native zinc carbonic anhydrase III as a function of pH. Experimental conditions
as in Figure 6-1. Co(ll)-substituted carbonic anhydrase III measured at 1 MHz
( a ), native zinc carbonic anhydrase III measured at 1 MHz ( ), Co(ll)-
substituted carbonic anhydrase III measured at 40 MHz ( ), native zinc
carbonic anhydrase III measured at 40 MHz ( ).

(Engberg and Lindskog, 1984). Also, the CO2 hydration activity of the cat
isozyme III and the bovine isozyme III is independent in the pH range 6-8.5 (Tu
etal., 1983; Engberg etal., 1985). These data imply that the activity controlling
group in Co(ll)-substituted and native Zn-carbonic anhydrase III, the aqueous
ligand of the metal, is predominantly in the hydroxide form in the pH range 6-9,
and that the pK for the ionization of the metal-bound water to the metal-bound
hydroxide is well below 6.
NMR relaxation enhancement studies with Co(ll)-substituted carbonic
anhydrase I and II have shown a large pH dependence of the paramagnetic
relaxation rate in the pH range 6-9, with an inflection at pH 7 and maximal
relaxation at high pH (Fabry etal., 1970; Wells etai, 1979). This correlates with
the pH dependence of both the visible spectrum and CO2 hydration activity
(Bertini etal., 1982; Silverman and Vincent, 1983). Thus the NMR relaxation
enhancement seems to correlate with the ionization state of the aqueous ligand
of the metal in Co(ll)-substituted carbonic anhydrase I and II.
However, it has been shown that SO42- present in solution to maintain
ionic strength affects the pH dependence of both the visible spectrum and the
paramagnetic relaxation of Co(ll)-substituted carbonic anhydrase I and II
(Bertini et ai, 1978; Bertini et a/.,1980; Simonsson and Lindskog, 1982). When
S042- and other anions are absent in either unbuffered solutions or solutions
containing only sulfonic acid buffers, the visible spectrums of Co(ll)-substituted
carbonic anhydrase I and II still show a large pH dependence, but there is
practically no pH dependence on the paramagnetic contribution to the
longitudinal relaxation rate of water protons in solutions Co(ll)-substituted
carbonic anhydrase II (Bertini etal., 1978), and a small dependence for Co(ll)-
substituted carbonic anhydrase I (Bertini etal., 1980). The data in Figure 6-3
are from solutions containing no anions other than those of the sulfonic acid

72
buffer. Under these conditions, the relaxation of water protons in solutions of
Co(ll)-substituted carbonic anhydrase II would be independent of pH in the
range 6-9, and this is what was observed for the paramagnetic relaxivity of
solutions of Co(ll)-substituted carbonic anhydrase III (Figure 6-3).
The reason for the small value of Rp at low frequency (Figure 6-2) is not
clear. One possibility is that the lifetime of a proton in the hydration shell of the
cobalt is too long to allow for a large paramagnetic contribution (Fabry et al.,
1970; Tu et al., 1985). This water off rate has a maximal value of 104 S1 in the
case of native carbonic anhydrase III (Tu et al., 1983), but is 105 S'1 for Co(ll)-
substituted carbonic anhydrase II (Tu et al., 1985). This rate would be expected
to be even slower at 5C, but when the pH 6 sample was measured at 25 C,
there was a larger paramagnetic relaxivity at low frequency (data not shown).
Another possibility is that in this isozyme the cobalt is in a pentacoordinate or
hexacoordinate environment (Engberg and Lindskog, 1984), which would have
less ability to relax due to a shorter electron relaxation time (Koenig et al.,
1983). Whatever the reason, the paramagnetic contribution at low frequency is
small.
The increase in Rp with higher frequency (Figure 2) is unlike that for any
other Co(ll)-substituted carbonic anhydrase which show a decrease in Rp with
higher frequency (Wells et al., 1979), but is not unlike that observed for Mn, Cu,
and Ni proteins (Koenig and Brown, 1973; Bertini and Luchinat, 1986). While
the relaxation behavior of these systems is not completely understood, it is clear
that the field dependence of the electron spin relaxation is a factor. Thus, the
increase in the electron relaxation time may be responsible for the increase in
Rp at intermediate frequency. The decrease at higher frequency may be due to
the (Osxc dispersion predicted by the Solomon-Bloembergen equations (Bertini
and Luchinat, 1986).

CHAPTER 7
CONCLUSIONS
This work shows the utility of the NMR relaxation method to measure the
hydrogen / deuterium fractionation properties of the aqueous ligands of cobalt
in some specific cases. In particular, the characterization of the relaxation
mechanisms at 300 MHz has allowed the detection of the fractionation of
hydrogen and deuterium of two populations of ligand sites associated with
cobalt, one having a significant contribution of the inner shell aqueous ligand of
the metal and the other apparently dominated by outer shell water. The
fractionation factor for the inner shell ligand can differ significantly from that for
solvent water, and in the case of Co(H20)62+, 4)t2 is near the fractionation factor
of H30+, implying that the water ligands of cobalt in Co(H20)62+ are influenced
by the positive charge of the cobalt. The fractionation factors of Co(ll)-
substituted carbonic anhydrase represent the first direct measurement of this
parameter at the active site of an enzyme.
The conclusions of the contributions of inner and outer shell water to the
relaxation times Ti and T2 result from the effect of the chemical shift relaxation
mechanism at 300 MHz. This conclusion is supported by calculations and
experimental evidence with EDTA for Co(H20)62+. In this hexaaquacomplex the
only interactions are those of the metal and water, and the water can be in the
inner shell metal-bound site, an outer shell site, or the bulk solvent site. There
are no other effects of the medium in this system. In the case of Co(ll)-
substituted carbonic anhydrase, there are many other interactions which may
occur to affect the fractionation of the hydrogenic positions of the aqueous
73

ligand of the metal. These interactions Include influences of charge by residues
near the active site, and the effects of the hydrogen bond network that exists in
the active site cavity. This hydrogen bond network includes water in the active
site cavity and specific residues from the enzyme, as well as the aqueous ligand
of the metal. Thus it is clear that outer shell water in Co(ll)-substituted carbonic
anhydrase would have some different properties than outer shell water in
Co(H20)62+. The relaxation rates at 300 MHz of water protons in solution of
Co(ll)-substituted carbonic anhydrase showed T-i much larger than T2, as in
Co(H20)62+, and this indicates that the chemical shift mechanism affects T2 in
solutions of Co(ll)-substituted carbonic anhydrase. However there is no other
support for this from calculations or other experiments, as in the case of
Co(H20)62+, except that the fractionation factor determined from T-i and T2 are
also different for Co(ll)-substituted carbonic anhydrase. This inference of the
chemical shift mechanism affecting T2 in solutions of Co(ll)-substituted carbonic
anhydrase is the only extrapolation made from Co(H20)62+ to Co(ll)-substituted
carbonic anhydrase, which are very different systems in many other respects. It
is important to note that the fractionation factors for Co(ll)-substituted carbonic
anhydrase are measured directly, and thus are independent of any other
interpretation regarding Co(H20)62+, except that of the contribution of inner shell
water to T2 and of outer shell water to T^ Thus the fractionation factors for
Co(ll)-substituted carbonic anhydrase include the effects of the interactions near
the active site.
One very unique opportunity to understand the basis for the differences
in the fractionation factors between the isozymes of carbonic anhydrase exists
in the study of site-directed mutants of the enzyme. It is known that the pK for
the ionization of the metal-bound water in carbonic anhydrase III is lower than
that ionization in carbonic anhydrase I and II. This difference is most likely due

to the positively charged residues near the active site in carbonic anhydrase III.
The fractionation factor for the metal-bound hydroxide in carbonic anhydrase III
was greater than that fractionation factor in carbonic anhydrase I and II, and this
may also be due to the positively charged residues in carbonic anhydrase III.
This could be tested, for instance, by replacing the histidine in position-64 of
carbonic anhydrase I and II with a lysine, which occupies position-64 in
carbonic anhydrase III, and then measuring the fractionation factor using the
cobalt-substituted enzyme. One would predict that the fractionation factor of
Co(ll)-substituted carbonic anhydrase I and II with lysine in position-64 would
be closer to unity, like that in Co(ll)-substituted carbonic anhydrase III.
The value of the fractionation factor for Co(H20)62+ measured directly
from NMR represents a reactant state fractionation factor that can be used to
interpret solvent hydrogen isotope effects associated with Co(H20)62+. The
solvent hydrogen isotope effects on the formation of complexes of cobalt with
some bidentate ligands suggest that the fractionation factor of some other
hydrogenic positions must contribute to the overall isotope effect, most likely the
positions of outer shell water in these complexes, because the isotope effect
could not be accounted for by considering only the fractionation factor of
Co(H20)62+. This may be because of the effect of the bidentate ligands on the
structure of outer shell water in these complexes. It could be that the
fractionation factor of the outer shell water in a cobalt-bidentate ligand complex
is different than that fractionation factor in Co(H20)62+. Also, it may be that the
number of outer shell waters that experience the paramagnetic relaxation
enhancement by the cobalt is less in the cobalt-bidentate ligand complex than
in Co(H20)62+. However, it is clear that there would be a large number of outer
shell waters associated with complexes, and that a small effect on the
fractionation of the outer shell water in these complexes on going from the

reactant to the product state, when raised to a large power for the number of
hydrogenic positions in the outer shell water, could have a large contribution to
the overall isotope effect.
The values of the fractionation factor of the aqueous ligand of the metal at
the active site of Co(ll)-substituted carbonic anhydrase can also be used to
interpret solvent hydrogen isotope effects on reactions associated with carbonic
anhydrase. The solvent hydrogen isotope effect on the equilibrium binding of
iodide ion to Co(ll)-substituted carbonic anhydrase suggests that there must be
other hydrogenic positions whose isotope preference changes on going from
the reactant to product states, because the consideration of the fractionation
factor for the aqueous ligand of cobalt in Co(ll)-substituted carbonic anhydrase
by itself does not allow a complete interpretation of the isotope effect. However,
it is clear from the results that the hydrogenic positions of the metal-bound water
at the active site do make a large contribution to the overall isotope effect. This
means that the other hydrogenic positions that contribute to the isotope effect
probably have a fractionation factor just slightly different from unity, but the large
number of these positions would then give a significant contribution to the
overall isotope effect. The solvent hydrogen isotope effects for other anions
binding to Co(ll)-substituted carbonic anhydrase could be determined, and
even for other inhibitors of the enzyme that are known to bind to the active site.
This may give more information about these other hydrogenic positions. For
instance, it may be that the fractionation factor of the hydration shell of the anion
is important, and this could be determined from the results with other
monovalent anions. Also, bulkier groups could be used, such as the
sulfonamide inhibitors, and these may displace more outer shell water within
the active site cavity. The isotope effect on the binding of a sulfonamide

77
inhibitor may then give some insight into the contribution of these outer shell
hydrogenic positions to the overall isotope effect.
To account for the solvent hydrogen isotope effects on a rate constant is
even more complex because knowledge of the fractionation factor values for the
transition state are required. The model for the transition state of the rate-
limiting proton transfer step of the CO2 hydration reaction, developed from the
fractionation factor of the aqueous ligand of the cobalt in Co(ll)-substituted
carbonic anhydrase II and the isotope effect on kcat) is consistent with other
experimental evidence for the structure of the transition state. Other kinetic
isotope effects on reactions of carbonic anhydrase may then also be used with
the fractionation factor to get information about the mechanisms of these
reactions. For instance, the solvent hydrogen isotope effects on the binding of
anions could be measured, and then interpreted with the fractionation factors of
the reactant state. This may show whether the rate-limiting step of the binding is
the initial entry of the anion into the active site cavity, or it may be that the final
binding of the anion to the metal is the rate-limiting step. This work represents
an important first step in the accurate interpretation of the solvent hydrogen
isotope effects in the catalysis by carbonic anhydrase.
Perhaps the most significant achievement in this work is the development
of the NMR relaxation method to measure directly the hydrogen / deuterium
fractionation factor of the hydrogenic positions of water coordinated to metals.
This method could also be used with other transition metals and in other types
of inorganic complexes, and in other metalloenzymes. Also, the conclusions of
the contributions of inner and outer shell water to the relaxation times and T2
could also be studied further by, for instance, determining the effects of ionic
strength and temperature and the relaxation and the fractionation factors

78
determined from T-i and T2. It is hoped that the work presented in this
dissertation will be the catalyst for other studies in this field.

References
Baes, N. F.; Mesmer, R. E. The Hydrolysis of Cations; Wiley: New York, 1976.
Bernheim, R. A.; Brown, T. H.; Gutowsky, H. S.; Woessner, D. E. J. Chem Phys.
1959, 30, 950-956.
Bertini, I.; Canti, G.; Luchinat, C.; Scozzafava, A. J. Am Chem Soc. 1978, 100,
4873-4877.
Bertini, I.; Dei, A.; Luchinat, C.; Monnanni, R. Inorg. Chem. 1985, 24, 301-303.
Bertini, I.; Luchinat, C. Acc. Chem. Res. 1983, 16, 272-279.
Bertini, I.; Luchinat, C. NMR of Paramagnetic Species in Biological Systems;
Benjamin-Cummings: Menlo Park, CA, 1986.
Bertini, I.; Luchinat, C.; Scozzafava, A.; Inorg. Chem. Acta 1980, 46, 85-89.
Bertini, I.; Luchinat, C.; Scozzafava, A. Struct. Bonding (Berlin) 1982, 48, 45-92.
Bloembergen, N. J. Chem Phys. 1957, 27, 572-573.
Bloembergen, N.; Morgan, L. O. J. Chem. Phys. 1961, 34, 842.
Brown, G. S.; Navon, G.; Shulman, R. G. Proc. Natl. Acad. Sci. USA 1977, 74,
1794-1797.
Carlson, G. A.; McReynolds, J. P.; Verhoek, F. H. J. Am. Chem. Soc. 1945, 67,
1334-1339.
Chiang, Y.; Kresge, A. J.; More O'Ferrall, R. A. J. C. S. Perkin I11980, 1832-
1839.
Dwek, R. A. NMR in Biochemistry; Oxford press: Oxford, 1975.
Eigen, M.; Tamm, K. Z Elektrochem. 1962, 66, 93-121.
Eisenstadt, M. J. Chem Phys. 1969, 51, 4421-4432.
Engberg, P.; Lindskog, S. FEBS Lett. 1984, 170, 326.
79

80
Engberg, P.; Millqvist, E.; Pohl, G.; Lindskog, S. Arch. Biochem. Biophys. 1985,
241, 628-638.
Eriksson, E. A.; Jones, T. A.; Liljas, A. In Zinc Enzymes; Bertini, I.; Luchinat, C.;
Maret, W.; Zeppezauer, M. Eds.; Birkhauser: Boston, 1986; 317-328.
Fabry, M. E.; Koenig, S. H.; Schillinger, W. E. J. Biol. Chem. 1970, 245, 4256-
4262.
Freeman, R.; Kempsell, S. P.; Levitt, J. J. Magn. Reson. 1980, 138, 453-479.
Friedman, H. L; Holz, M.; Hertz, H. G. J. Chem Phys. 1979, 70, 3369.
Glasoe, P. K.; Long, P. A. J. Phys. Chem. 1960, 64, 188-191.
Gold, V.; Lowe, B. M. J. Chem. Soc. (A) 1968, 1923-1932.
Gold, V.; Wood, D. L. J. Chem. Soc., Dalton Trans. 1982, 2452-2461.
Goodall, D. M.; Long, F. A. J. Am. Chem. Soc. 1968, 90, 238.
Gueron, M. J. Magn. Reson. 1975, 19, 58.
Hammes, G. C.; Steinfeld, J. I. J. Am. Chem. Soc. 1962, 84, 4639-4643.
Hunt, J. B.; Rhee, M. J.; Storm, C. B. Anal. Biochem. 1977, 79, 614-617.
Jencks, W. P.; Salvesen, K. J. Am Chem Soc. 1971, 93, 4433.
Kalifah, R. Biochemistry 1977, 16, 2236-2240.
Kalifah, R, G.; Strader, D. J.; Bryant, S. H.; Gibson, S. M. Biochemistry 1977, 16,
2241-2247.
Kararli, T.; Silverman, D. N. J. Biol. Chem. 1985, 260, 3484-3489.
Koenig, S. H.; Brown, R. D. Ann. N. Y. Acad. Sci. 1973, 222, 752-763.
Koenig, S. H.; Brown, R. D.; Bertini, I.; Luchinat, C. Biophys. J. 1983, 41, 179-
187.
Kresge, A. J.; More O'Ferrall, R. A.; Powell, M. F. In Isotopes in Organic
Chemistry, Volume 7; Buncel, E.; Lee, C. Eds.; Elsevier: New York, 1987; 177-
273.
Laughton, P. M.; Robertson, R. E. In Solute-Solvent Interactions4, Coetzee, J. F.;
Ritchie, C. D. Eds.; Marcel Dekker: New York, 1969; 399-538.
Lindskog, S. Biochemistry 1966, 5, 2641-2646.

Undskog, S. in Zinc Enzymes; Spiro, T. G. Ed.; Wiley: New York; 1983; 78-121.
Lowe, B. M.; Smith, D. G. J. Chem. Soc., Faraday Trans. /1975, 71, 389-397.
Luz, Z.; Meiboom, S. J. Chem. Phys. 1964, 40, 2686-2692.
Meiboom, S.; Gill, D. Rev. Sci Instrum. 1958, 29, 688.
Melamud, E.; Mildvan, A. S. J. Biol. Chem. 1975, 250, 8193-8201.
Melton, B. F.; Pollack, V. L. J. Phys. Chem. 1969, 73, 3669-3679.
More O'Ferrall, R. A.; Koeppl, G. W.; Kresge, A. J. Am. Chem. Soc. 1971,93, 9-
20.
Morgan, L. O.; Nolle, A. W. J. Chem Phys. 1959, 31, 365.
Nyman, P. O.; Lindskog, S. Biochim. Biophys. Acta 1964, 85, 141-151.
Pocker, Y.; Sarkanen, S. Adv. Enzymol. Relat. Areas Mol. Biol. 1978, 47,149-
274.
Schowen, K. B. J. In Transition States in Biochemical Processes] Grandour, R.
D.; Schowen, R. L. Eds.; Plenum: New York, 1978; 225-284.
Schowen, K. B.; Schowen, R. L. In Methods in Enzymology, Volume 87; Purich,
D. L. Ed.; Academic: New York, 1982; 551-606.
Silverman, D. N. J. Am. Chem. Soc. 1981, 103, 6242-6243.
Silverman, D. N.; Lindskog, S. Acc. Chem. Res. 1988, 21, 30-36.
Silverman, D. N.; Vincent, S. H. CRC Crit. Rev. Biochem. 1983, 14, 207-255.
Simonsson, I.; Lindskog, S. Eur. J. Biochem. 1982, 123, 29-36.
Smith, R. M.; Martell, A. E. Critical Stability Constants; Plenum: New York, 1975.
Solomon, I. Phys. Rev. 1955, 99, 559-565.
Steiner, H.; Jonsson, B.; Lindskog, S. Eur. J. Biochem. 1975, 59, 253-259.
Swift, T. J.; Connick, R. E. J. Chem. Phys. 1962, 37, 307-320; J. Chem. Phys.
1964, 41, 2553-2554.
Szawelski, R. J.; Wharton, C. W.; White, S. Biochem. Soc. Trans. 1982, 10, 232-
233.

82
Tibell, L; Forsman, C.; Simonsson, I.; Lindskog, S. Biochim. Biophys. Acta
1984, 789, 302-310.
Tu, C. K.; Sanyal, G.; Wynns, G. C.; Silverman, D. N. J. Biol. Chem. 1983, 258,
8867-8871.
Tu, C. K.; Silverman, D. N. Biochemistry !985, 24, 5881-5887.
Tu, C. K.; Thomas, H. G.; Wynns, G. C.; Silverman, D. N. J. Biol. Chem. 1986,
261, 10100.
Venkatasubban, K. S.; Silverman, D. N. Biochemistry!980, 19, 4984.
Voice, P.J. J. C. S. Faraday /1974, 70, 498-505.
Wells, J. W.; Kandel, S. I.; Koenig, S. H. Biochemistry 1979, 18, 1989-1995.
Whitney, P. L; Brandt, H. J. Biol. Chem. 1976, 251, 3862-3867.

83
BIOGRAPHICAL SKETCH
James W. Kassebaum, the son of Amanda L. Schoonmaker and Wilbur
F. Kassebaum, was born March 6, 1962, in Mount Vernon, Illinois. James grew
up in St. Louis, Missouri, and in 1980 graduated first in his class at Parkway
South Senior High School. He then attended Purdue University, West
Lafayette, Indiana, and in 1984 graduated with Honors with the degree of
Bachelor of Science in chemistry. The summer of 1984 was a very eventful
one, when James was wed to Shari L. Gunderman and then entered graduate
school at the University of Florida in the Department of Biochemistry and
Molecular Biology and studied under the supervision of Dr. David Silverman.
James completed the requirements for the Ph.D. degree in 1988, and began his
professional career at the Monsanto Company in St. Louis.

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
/'Jxju Jc(lu
David N. Silverman, Chair
Professor of Pharmacology and
Therapeutics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
CLwcfotS}
E. Raymond Andrew
Graduate Research Professor of
Physics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
Robert J/Cohen
Associate Professor of
Biochemistry and Molecular
Biology
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Docjdf^f Philosophy.
Russell S. Drago
Graduate Research Professor of
Chemistry

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosopfiy^
fin
DanTel L. Purich
Professor of Biochemistry and
Molecular Biology
This dissertation was submitted to the Graduate Faculty of the College of
Medicine and to the Graduate School and was accepted as partial fulfillment of
the requirements for the degree of Doctor of Philosophy.
December, 1988
Dean, Graduate School



Q|_|AP1"£P 0
HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF THE
AQUEOUS LIGAND IN Co(H20)62+ AND Ni(H20)62+
Introduction
The hydrogen / deuterium fractionation factor, <{), of a metal-bound water
is the equilibrium constant for this isotope exchange reaction:
M(HOH) + (HOD) > M(HOD) + {HOH)
[M(HOD)] / [M(HOH)]
(3-1)
[HOD] / [HOH]
This fractionation factor measures the tendency of deuterium to accumulate at
the aqueous ligand of the metal relative to the deuterium content of bulk solvent.
Fractionation factors are valuable in interpreting the effects of deuterium on
kinetic and equilibrium constants because they represent individual
contributions to isotope effects measured relative to the common reference of
water. The fractionation of hydrogen isotopes in water bound to metal ions has
been considered (Schowen and Schowen, 1982); however, there are few
reports of the fractionation factor of the aqueous ligands of a metal. Using NMR
methods, Melton and Pollack (1969) have measured such a factor for the water
ligands of Cr(H20)62+ ((j) = 1.00 0.03). They relied on the properties of a
paramagnetic metal to enhance the NMR relaxation rate of protons of water
exchanging rapidly between bulk solvent and the hydration shell of the metal.
By measuring the proton relaxation rate as a function of the deuterium content
of solvent, the fractionation factor of the metal-bound site can be determined.
20


48
calculated using T2 were significantly less than unity for isozymes I and II, but
near unity for isozyme III (Table 4-2).
Discussion
The relaxation behavior of water protons at 300 MHz and pH 8.5 in
solutions of the Co(ll)-substituted carbonic anhydrases showed Tip/T2P much
greater than unity (Table 4-1). This was the same as observed for the water
protons in solutions of cobalt ions (see Chapter 2), and the interpretation in the
case of Co(H20)62+ can be applied to Co(ll)-substituted carbonic anhydrase. At
300 MHz, there is a field dependent increase in the chemical shift difference
between a proton in the metal-bound site and the free solvent site. This causes
an increase in the transverse relaxation of the protons as they experience the
two environments (equation 2-3). Because this relaxation mechanism has as its
source a paramagnetic contact shift, the transverse relaxation has a larger
contribution of the ligand in the inner coordination shell of the cobalt, while the
longitudinal relaxation has a larger contribution of water in the outer
coordination shell. Thus the fractionation factor measured from T2 represents
more of the fractionation of the inner shell aqueous ligand of cobalt in Co(ll)-
substituted carbonic anhydrase, while the fractionation factor measured from T-i
is a greater contribution of the fractionation of outer shell water. A similar
situation in which the outer shell relaxation has a larger contribution to than
to R2 has been reported by Kushnir and Navon (1984) for water protons in
solution with Mn(ll)-substituted carbonic anhydrase II.
These relaxation measurements were made at pH 8.5, a pH for which the
relaxation rate of water protons in solutions of Co(ll)-substituted carbonic
anhydrase I and II is greatest (Wells etal., 1979). (See Chapter 6 for an
investigation of this property in solutions of Co(ll)-substituted carbonic
anhydrase III.) This pH is above the value of the pK for the ionization of the


greatest and in a region that is independent of pH (Wells et al., 1979). (See
Chapter 6 for a discussion of the pH and magnetic field dependence of the
relaxation of water protons in solutions of Co(ll)-substituted carbonic anhydrase
III.) The values of pH reported here are uncorrected pH meter readings. The
correction of a pH meter reading in 100% D2O (pD = meter reading + 0.4) is
approximately offset by the change in ionization state of the buffer in D2O (pKq2o
- pKh2o = 0.5 0.1 for almost all acids with pK values between 3 and 10,
Schowen, 1978). Water used for solution preparations was distilled and
passed through two ion-exchange resin cartridges (Cole-Palmer 1506-35).
D20 (99.8%) was stirred with activated charcoal, the charcoal filtered out, and
then the D2O was distilled.
Measurement of the Hydrogen / Deuterium Fractionation Factor; The
hydrogen / deuterium fractionation factor was determined by measuring the
relaxation of water protons in solutions of Co(ll)-substituted carbonic anhydrase
as a function of the atom fraction of deuterium in solvent water as described in
Chapter 3.
Results
The substitution of cobalt for zinc at the active site of carbonic anhydrase
provides a paramagnetic center which affects the relaxation of water protons in
solutions of carbonic anhydrase. The paramagnetic relaxivity of these water
protons is greater for R2 than for R1 for each isozyme of carbonic anhydrase
(Table 4-1).
The relaxation of water protons in solutions of the three isozymes of
Co(ll)-substituted carbonic anhydrase as a function of the atom fraction of
deuterium in solvent water is shown in Figures 4-1,4-2, and 4-3. The hydrogen
/ deuterium fractionation factors calculated from equation 3-4 using T-\ were
near unity for the three isozymes (Table 4-2). However, the fractionation factors


77
inhibitor may then give some insight into the contribution of these outer shell
hydrogenic positions to the overall isotope effect.
To account for the solvent hydrogen isotope effects on a rate constant is
even more complex because knowledge of the fractionation factor values for the
transition state are required. The model for the transition state of the rate-
limiting proton transfer step of the CO2 hydration reaction, developed from the
fractionation factor of the aqueous ligand of the cobalt in Co(ll)-substituted
carbonic anhydrase II and the isotope effect on kcat) is consistent with other
experimental evidence for the structure of the transition state. Other kinetic
isotope effects on reactions of carbonic anhydrase may then also be used with
the fractionation factor to get information about the mechanisms of these
reactions. For instance, the solvent hydrogen isotope effects on the binding of
anions could be measured, and then interpreted with the fractionation factors of
the reactant state. This may show whether the rate-limiting step of the binding is
the initial entry of the anion into the active site cavity, or it may be that the final
binding of the anion to the metal is the rate-limiting step. This work represents
an important first step in the accurate interpretation of the solvent hydrogen
isotope effects in the catalysis by carbonic anhydrase.
Perhaps the most significant achievement in this work is the development
of the NMR relaxation method to measure directly the hydrogen / deuterium
fractionation factor of the hydrogenic positions of water coordinated to metals.
This method could also be used with other transition metals and in other types
of inorganic complexes, and in other metalloenzymes. Also, the conclusions of
the contributions of inner and outer shell water to the relaxation times and T2
could also be studied further by, for instance, determining the effects of ionic
strength and temperature and the relaxation and the fractionation factors


60
Table 5-3: Values of the hydrogen / deuterium fractionation factor for the low pH
forms of Co(ll)-substituted carbonic anhydrase.
4>ti
t2
Co(ll)-substituted carbonic anhydrase 1
1.12 0.05
0.90 0.04
Co(ll)-substituted carbonic anhydrase II
0.90 0.03
0.81 0.03
Note: The § for the low pH form was calculated using the isotope effect on the
ionization constant of the cobalt-bound water and the $ for the high pH form
measured directly from NMR and given in Table 4-2.


46
Atom fraction of deuterium in solvent (n)
Atom fraction of deuterium in solvent (n)
Figure 4-3: The dependence of pTip (i=1,2) at 300 MHz on n, the atom fraction
of deuterium in solvent water, for 1.5 mM solutions of Co(ll)-substituted carbonic
anhydrase III. Solutions contained 50 mM Hepes buffer at pH 8.5 and 23C.
Data are the means and standard errors of three measurements.



PAGE 1

HYDROGEN / DEUTERIUM FRACTIONATION FACTORS OF THE AQUEOUS LIGAND OF COBALT IN Co(H20)62+ AND Co(ll)-SUBSTITUTED CARBONIC ANHYDRASE By JAMES WEB KASSEBAUM A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1988

PAGE 2

ACKNOWLEDGEMENTS I wish to thank Dr. David Silverman for his thoughtfulness and guidance shown to me during the work presented in this dissertation. I have benefitted greatly from his scientific insight into experimental design and his careful attention to excellence in scientific writing. I also wish to thank very much Dr. C. K. Tu and Mr. George Wynns for their help in experimental procedures and for many helpful discussions. I acknowledge the very valuable advice of the members of my committee, Dr. E. Raymond Andrew, Dr. Robert Cohen, Dr. Russell Drago, and Dr. Daniel Purich. Finally, I especially thank Dr. Seymour Koenig at the IBM T. J. Watson Research Center in Yorktown Heights, New York, for allowing me to visit his laboratory and use the field-cycling relaxometer which added greatly to the scope of this work. ii

PAGE 3

TABLE OF CONTENTS page ACKNOWLEDGEMENTS ii ABSTRACT v CHAPTERS 1 INTRODUCTION 1 2 WATER PROTON NMR RELAXATION ENHANCEMENT IN SOLUTIONS OF Co(H20)62+ AND Ni(H20)62+ 9 Introduction 9 Experimental Procedure 1 o Results 1 1 Discussion 1 6 3 HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF THE AQUEOUS LIGAND IN Co(H20)62+ AND Ni(H20)62+ 20 Introduction 20 Experimental Procedure 22 Results 1. 25 Discussion 32 4 HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF THE AQUEOUS LIGAND OF COBALT IN Co(ll)SUBSTITUTED CARBONIC ANHYDRASE 39 Introduction 39 Experimental Procedure 40 Results 42 Discussion "3.3.1... 48 5 INTERPRETATION OF SOLVENT HYDROGEN ISOTOPE EFFECTS ASSOCIATED WITH CARBONIC ANHYDRASE 52 Introduction 52 iii

PAGE 4

Experimental Procedure 53 Results 55 Discussion 58 6 MAGNETIC FIELD AND pH DEPENDENCE OF THE WATER PROTON NMR RELAXATION ENHANCEMENT IN SOLUTIONS OF Co(ll)-SUBSTITUTED CARBONIC ANHYDRASE III 65 Introduction 65 Experimental Procedure 66 Results 67 Discussion 67 7 CONCLUSIONS 73 REFERENCES 79 BIOGRAPHICAL SKETCH 83 iv

PAGE 5

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy HYDROGEN / DEUTERIUM FRACTIONATION FACTORS OF THE AQUEOUS LIGAND OF COBALT IN Co(H20)62+ AND Co(ll)-SUBSTITUTED CARBONIC ANHYDRASE By James Web Kassebaum December, 1988 Chairman: Dr. David N. Silverman Major Department: Biochemistry and Molecular Biology I have measured the hydrogen / deuterium fractionation factor for the rapidly exchanging aqueous ligands of cobalt in Co(H20)62+ and in three Co(ll)substituted isozymes of carbonic anhydrase. The fractionation factor was determined from NMR relaxation rates at 300 MHz of the protons of water in mixed solutions of H2O and D2O containing these complexes. In each case, the paramagnetic contribution to I/T2 was greater than to I/T1, consistent with a chemical shift mechanism affecting I/T2. The fractionation factors obtained from Ti for Co(H20)62+ and for the isozymes of Co(ll)-substituted carbonic anhydrase were close to the fractionation factor for bulk water, which is unity. The fractionation factors obtained from T2 were 0.73 0.02 for Co(H20)62+, 0.72 0.02 for Co(ll)-substituted carbonic anhydrase I, 0.77 0.01 for Co(ll)substituted carbonic anhydrase II, and 1.00 0.07 for Co(ll)-substituted carbonic anhydrase III. I concluded that fractionation factors in these cases determined from Ti and T2 measured isotope preferences for different populations of ligand sites. I suggest that since T2 has a large contribution from V

PAGE 6

a chemical shift mechanism, the fractionation factor determined from T2 has a large contribution of the fractionation of inner shell ligands. The fractionation factors determined from Ti are close to unity, the value of the fractionation factor of bulk water, and contain a larger contribution of the fractionation of outer shell water. The fractionation factor of Co(H20)62+ was used to interpret the solvent hydrogen isotope effects on the formation of complexes of cobalt with the bidentate ligands glycine, A/,A/-dimethylglycine, and acetylacetone. The contribution of the fractionation factor of the inner shell water in Co(H20)62+ did not account completely for the measured isotope effect, and I suggest that the hydrogen / deuterium fractionation of outer shell water makes a large contribution to the isotope effect on the formation of these complexes. The solvent hydrogen isotope effect on the equilibrium binding of iodide ion to Co(ll)-substituted carbonic anhydrase I and II was determined and then interpreted using the fractionation factor of the aqueous ligand of cobalt at the active site. This fractionation factor could not account for the measured isotope effect, implying that the hydrogen / deuterium fractionation of additional groups, or water associated with the enzyme, changes on going from the reactant state to the product state when iodide binds to Co(ll)-substituted carbonic anhydrase I and II. Although not yet helpful in interpreting the complex contributions to the isotope effects in the enzymatic catalysis, these hydrogen / deuterium fractionation factors for the water bound to cobalt in carbonic anhydrase can be significantly different from the fractionation factor for solvent water and may be sensitive to the active site environment in these homologous isozymes of carbonic anhydrase. vi

PAGE 7

CHAPTER 1 INTRODUCTION The effect of an isotopic substitution on rate and equilibrium constants associated with a chemical reaction can give information about the mechanism of the reaction. One method in making an isotopic substitution is the replacement of D2O for H2O as the solvent, which can affect a reaction in two ways. First if water is a substrate in the reaction, then the rate and equilibrium constants can be affected by the presence of D2O. Also, if there are exchangeable hydrogenic sites in the reactants or products, or, in the case of an enzyme-catalyzed reaction, if there are exchangeable hydrogenic sites on the enzyme, then the rate and equilibrium constants can be affected when D2O is the solvent, because deuterium will be present in the exchangeable positions. The difference in a reaction in H2O and D2O results from differences in the zero point energy of bonds to H or D (Schowen, 1 978). In the case of a reaction rate, if the zero point energy difference for a bond to H and to D decreases on going to the transition state, then the activation energy is greater in the case of D, and the rate is lower. This is an example of a normal isotope effect, in which the rate constant when the bond is to H, kn, is greater than the rate constant when the bond is to D, ko. There also exist reactions where ko > kH, which is known as an inverse isotope effect. The difference in zero point energy can be expressed in terms of an equilibrium constant for isotope exchange for a particular hydrogenic site. For a single solute species in an isotopically-mixed aqueous solvent, the hydrogen / 1

PAGE 8

2 deuterium fractionation factor, (|), is the equilibrium constant for this isotope exchange reaction R(H) + (HOD) R(D) + (HOH) [R(D)] / [R(H)] (|, = [HOD] / [HOH] in which a solute hydrogenic species can exchange with a solvent hydrogenic species. The fractionation factor measures the tendency of D to accumulate in the solute relative to the deuterium content of bulk solvent. A fractionation factor greater than unity implies that D will accumulate in the solute position relative to the solvent, and that the bond to hydrogen is stronger in the solute position than that bond in the solvent. For a fractionation factor less than unity, the light isotope of hydrogen will accumulate in the solute position, and the bond to hydrogen is weaker in the solute relative to the solvent. An isotope effect is a function of the fractionation factors of each hydrogenic site in a reaction. For an isotope effect on a kinetic constant, all reactant state sites n (1-1) D2O all transition state sites n d.,^ and for an isotope effect on an equilibrium constant,

PAGE 9

3 all reactant state sites n (Di" 1^20 all product state sites n *r i where and are reactant, transition, and product state hydrogenic site fractionation factors, respectively (Schowen, 1978). Thus it is clear that to understand completely an isotope effect associated with a reaction, the knowledge of the underlying fractionation factors is required. Values of the fractionation factor for various functional groups are known relative to water as the solvent (Schowen and Schowen, 1982; Kresge et al. 1987), and range from 0.4 for some examples of hydrogen bound to sulfur, to 1 .5 for one case of a hydrogen bound to a tertiary amine. The majority of these fractionation factors where measured by one of two methods. One involves using NMR chemical shifts to measure the fractionation factor of a hydrogenic site directly. The technique is based on the rapid exchange of hydrogens between solute and solvent species, so that the observed proton resonance is a weighted average of the chemical shift of the solute and solvent positions. The fractionation factor is determined by measuring the chemical shift at different concentrations of solute in solvents of low and high atom fraction of deuterium. Using this method, Gold and Lowe (1968) determined the fractionation factor of the exchangeable hydrogen of acetic acid to be 0.96 0.02. A variation of this method when there is slow exchange of hydrogens between solute and solvent involves measuring the NMR signal areas in mixed isotopic solvents. In this manner Chiang et al. (1980) have determined the fractionation factor of the exchanging position of 1,8-Bis(dimethylamino)-naphthalene to be 0.90 0.01, and Szawelski et al. (1982) have measured the fractionation factor of the

PAGE 10

4 hydrogen bound to sulfur in mercaptoethanol to be 0.55 0.02. The second method to determine fractionation factors is to measure the equilibrium constant for a reaction as a function of the atom fraction of deuterium in the solvent, and then using relationships like those that lead to equations 1-1 and 1-2 to calculate the fractionation factor from a fit to the data. This method is then indirect and is unable to separate the contribution from the fractionation at one site from contributions from the medium, which may be a factor. Lowe and Smith (1975) have used this method to determine the fractionation factor of the acidic hydrogen In benzoic acid to be 1 .02 0.01 Both of these methods for determining fractionation factors lead to precise values for the simple cases cited above involving one exchangeable hydrogen. The unifying theme of this work is the study of the hydrogen / deuterium fractionation factor properties of the hydrogenic positions of aqueous ligands of metal ions in simple inorganic complexes and carbonic anhydrase. There is a relative lack of information about the fractionation of the hydrogenic sites of the aqueous ligands of metals compared to that of other functional groups. Using an NMR relaxation method, I have measured the fractionation factor of the aqueous ligand of cobalt in some simple inorganic complexes and in Co(ll)substituted carbonic anhydrase. The goal is to use these fractionation factors in the interpretation of isotope effects on the formation of cobalt complexes and on equilibrium and kinetic constants associated with carbonic anhydrase. The fractionation factor of the aqueous ligand of cobalt in these systems represents the individual contributions of those hydrogenic positions to the overall isotope effect on the reactions. Solvent hydrogen isotope effects in inorganic reactions are relatively little studied compared to those effects in organic reactions (Kresge etal. 1987). The data for ligand substitution reactions with transition metal complexes consists of

PAGE 11

5 a small number of solvent hydrogen isotope effects measured for equilibrium constants and for rate constants of reactions with Cr3+ and Co3+ (Laughton and Robertson, 1969; Gold and Wood, 1982). Kresge etal. (1987) have suggested that large rate and equilibrium differences between reactions of transition metal aquacomplexes in H2O and D2O should be expected because the hydrogen / deuterium fractionation factor of the water molecules in the inner coordination shell would be raised to the twelfth power for a hexaaquacomplex. Solvent hydrogen isotope effects on reactions catalyzed by cari^onic anhydrase are known and have been used to deduce aspects of the catalytic mechanism (Silverman and Vincent, 1983). CartDonic anhydrase is a zinc metalloenzyme which catalyzes the reversible hydration of carbon dioxide to produce bicarbonate and a proton. CO2 + H2O HCO3 +H* The zinc can be removed and replaced with cobalt to produce an enzyme with very similar catalytic properties (Lindskog, 1983) and the cobalt, because of its unpaired electrons, provides a spectroscopic probe of the enzyme (Bertini and Luchinat, 1983). In particular, the paramagnetic cobalt will enhance the NMR relaxation rate of water protons in solutions of Co(ll)-substituted carbonic anhydrase. The metal ion is at the base of the active site cleft and is coordinated by three histidine ligands with a water molecule as a fourth ligand which can ionize to a hydroxide. The mechanism of CO2 hydration (scheme 11) takes place by direct nucleophilic attack of a metal-bound hydroxide on CO2 to produce HCO3(Steiner etal. 1975), with a water molecule then displacing HCO3to produce a metal-bound water. A proton must then be transferred out of the active site to regenerate the metal-bound hydroxide for the next round of catalysis.

PAGE 12

^oHDHg ^:;p=^ ^Co-OH+ H* step 2 Scheme 1-1 : The metal hydroxide mechanism of carbonic anhydrase.

PAGE 13

7 When this reaction is performed in D2O some very interesting isotope effects are observed. The isotope effect on the steady-state parameter kcat for both native and Co(ll)-substituted carbonic anhydrase II is large, 3.8 in the case of native zinc-carbonic anhydrase II (Steiner etal. 1975), and 1.8 for Co(ll)substituted carbonic anhydrase II (R. S. Rowlett, unpublished), which means that a primary proton transfer is at least part of the rate limiting step measured by kcatThe ratio kcat/Km contains steps up to and including the first irreversible step in the reaction, and for the catalyzed reaction the first irreversible step is the release of HCO2from the enzyme. The isotope effect is unity on kcat/Km for both native zinc-carbonic anhydrase II (Steiner etal. 1975) and for Co(ll)substituted carbonic anhydrase II (R. S. Rowlett, unpublished), which implies that the rate determining step measured by kcat is separate and distinct from the interconversion of CO2 and HCO3-. Thus the rate-limiting step measured by kcat is the rate of transfer of a proton from the metal-bound water out of the active site to regenerate the metal-bound hydroxide (step 2 of Scheme 1-1). From this description it is clear that the aqueous ligand of the metal at the active site of carbonic anhydrase is intimately involved in the mechanism of catalysis, and to interpret the isotope effects knowledge of the fractionation factor of the aqueous ligand of the metal is required. The fractionation factor of the aqueous ligand of a paramagnetic metal can be determined by taking advantage of the paramagnetic contribution to the NMR relaxation rates of water protons in solutions of the metal in which the protons are in fast exchange between the coordination shell of the metal and the bulk solvent. The NMR measurements in this work were performed at a proton resonance frequency of 300 MHz, which is a higher frequency than has been used to study the water proton relaxation enhancement of first row transition metals (Dwek, 1975; Bertini and Luchinat, 1986). Thus the relaxation

PAGE 14

8 mechanisms operating in these systems at 300 MHz need to be understood before the relaxation can be used to measure a fractionation factor, and this is discussed in Chapter 2. Chapter 3 includes the determination of the fractionation factor of Co(H20)62+ and Ni(H20)62+ along with measurements of the solvent hydrogen isotope effect on the formation constants of some cobalt complexes to see if the fractionation factor of Co(H20)62+ can account for any of these effects. In Chapter 4 I have measured the fractionation factor of three mammalian isozymes of Co(II)-substituted carbonic anhydrase using NMR relaxation, and in Chapter 5 I discuss the use of these fractionation factors to interpret solvent hydrogen isotope effects on the equilibrium binding of an inhibitor to carbonic anhydrase and on the catalysis. Finally, the relaxation properties of Co(ll)-substituted carbonic anhydrase III have not previously been reported, and in Chapter 6 I discuss experiments to determine the magnetic field and pH dependence of the water proton relaxation enhancement in solutions of Co(ll)-substituted carbonic anhydrase III.

PAGE 15

CHAPTER 2 WATER PROTON NMR RELAXATION ENHANCEMENT IN SOLUTIONS OF Co(H20)62+ AND Ni(H20)62+ Introduction The initial studies of the influence of paramagnetic ions on the NMR relaxation rates of water protons were performed at proton resonance frequencies up to 60 MHz (Bernheim et al., 1959; Morgan and Nolle, 1959). These results were interpreted in terms of the Solomon-Bloembergen equations, which describe the relaxation as the sum of an electron-nuclear dipole-dipole interaction and a scalar interaction (Solomon, 1955; Bloembergen, 1957). At a frequency of 20 MHz the water protons in solutions of Co(H20)62+ were observed by Bernheim etal. (1959) to relax such that Ti = T2, and this relaxation was also observed at 10 MHz for solutions of Ni(H20)62+ (Morgan and Nolle, 1959). This result is consistent with the electron-nuclear dipole-dipole mechanism dominating the relaxation of the water protons, the correlation time for the interaction being the very short electron spin relaxation time. This short relaxation time for the electron in Co(H20)62+ and Ni(H20)62+ precludes any contribution to the relaxation from the scalar mechanism. In the absence of a scalar contribution, 7/6 would be the largest observable ratio for T1/T2. I have made measurements of the relaxation of water protons in solutions of Co(H20)62+ and Ni(H20)62+ at a proton resonance frequency of 300 MHz. In each case it was observed that the ratio T1/T2 was much larger than unity, consistent with a chemical shift mechanism affecting T2 at this frequency. Because this mechanism will affect T2 to the greatest extent when the protons 9

PAGE 16

10 are in the primary hydration shell of the ion, this allowed a qualitative differentiation between the relaxation of inner and outer shell water. Experimental Procedure The NMR relaxation times Ti and T2 of water protons were measured at a frequency of 300 MHz on a Nicolet NT-300 spectrometer at 23 C. Ti was measured by the inversion recovery method according to Freeman etal. (1980) and T2 was measured by the Carr-Purcell-Meiboom-Gill sequence (Meiboom and Gill, 1958). The observed relaxation of water protons in a solution of a paramagnetic ion is the sum of the relaxation due to the paramagnetic ion, l/Tjp (i=1 ,2), and the relaxation in the absence of the ion, 1/Tjo, 1/Ti = 1/Tip + 1/Tio (2-1) The contribution of the paramagnetic ion to the longitudinal relaxation is given by (Luz and Meiboom, 1964) 1^1p=P'[1/(Tim+Tm)] (2-2) and the paramagnetic contribution to the transverse relaxation is given by (Swift and Connick, 1962) (1/T2m)(1/T2m + 1/Xm) + AC0m2 1 /T2p = p' [ ] (2-3) Tm((1/T2m + 1/Xm)2 + AC0m2) where Tjm is the relaxation time of a proton in the hydration shell of the metal and is described by the Solomon-Bloembergen equations, Tm is the lifetime of a proton in the hydration shell, and p' is q[M]/55.5 where q is the hydration number and [M] is the concentration of the metal ion. AcOm is the chemical shift difference between a proton in the primary hydration shell and a proton in the

PAGE 17

1 1 free solvent site. If AcOm^ is small relative to the other terms, equation 2-3 reduces to equation 2-2. For these metal ions, 1/Tjo was taken as the relaxation of water containing only buffer, with no added metal ions. Rj, the relaxivity, is defined as the paramagnetic relaxation per millimolar concentration of metal ion. Ri = (1/Tip)/ ([Metal] x 1 000) (2-4) Solutions of Co(H20)62+ and Ni(H20)62+ for NMR measurements contained ultra-pure C0CI2 or NISOa (Aldrich Gold-Label) with 5 x 10-^ M acetic acid buffer at pH = 4.4. This pH was necessary to preclude the formation of multi-nuclear hydrolysis products (Baes and Mesmer, 1976). When EDTA was present no acetic acid was used. Water used for solution preparations was distilled and passed through two ion-exchange resin cartridges (Cole-Palmer 1506-35). D2O (99.8%) was stirred with activated charcoal, the charcoal filtered out, and then the D2O was distilled. Results The paramagnetic contribution to the relaxivity R2 of water protons in solutions of Co2+ and Ni2+ at 300 MHz was greater than that obtained at lower frequencies by previous investigators (Table 2-1 ). For Co2+, R^ at 300 MHz and 20 MHz were close in value while R2 was much larger at 300 MHz than at 20 MHz. For Ni2+, R^ at 300 MHz was slightly larger than at 1 0 MHz which is understood to be a result of a field dependent increase in the electron relaxation time (Friedman et al.. 1979), but R2 was much larger at 300 MHz than at 10 MHz. The addition of EDTA to the solutions of C0CI2 and NiS04 reduced Ri and R2 at 300 MHz (Figures 2-1 and 2-2). When an equimolar amount of EDTA

PAGE 18

12 Table 2-1 : Paramagnetic contribution to the relaxivities of water protons observed for aqueous solutions of C0CI2 and NiS04. frequency (MHz) Ri {mM-1 s-1) R2 (mM-1 s-i) Tip/T2p C0(H20)62+ 20 a 0.18 0.21 1.2 300 b 0.16 2.0 12.5 Ni(H20)62+ IOC 0.64 0.71 1.1 300 b 0.95 2.3 2.4 a From Bernheim etal. (1959) for an unbuffered solution of 0.5 M CoCb at 25 C. b Determined in this work at 23 C for 10 mM C0CI2 or 10 mM NiS04 which contained 10% D2O by volume and 5 x 10-^ M acetic acid buffer. The pH was 4.4. c From Morgan and Nolle (1959) for a solution of 10 mM Ni(N03)2 at 27 C in 0.1 M perchloric acid.

PAGE 19

13 CO > x DC [EDTA]/[Co2+] Figure 2-1 : The relaxivity of the protons of water at 300 MHz due to Co2+ in solutions of 10 mM CoCi2 as a function of the molar ratio of EDTA to Co2+ at 23 C. Solutions contained 50% D2O by volume. Ri (), R2 (a).

PAGE 20

14 3 0 H 1 1 0 12 3 [EDTA]/[Ni2+] Figure 2-2: The relaxivity of the protons of water at 300 MHz due to Ni2+ in solutions of 1 0 mM NiS04 as a function of the molar ratio of EDTA to Ni2+ at 23 C. Solutions contained 50% D2O by volume. Ri ( ), R2 ( a ).

PAGE 21

15 was added to solutions of Co2+, was reduced to 47% of the value of Ri in the absence of EDTA, R2 was reduced to 5% of the value of R2 in the absence of EDTA, and Tip/T2p was 1 .2. The remaining paramagnetic relaxivity due to the CoEDTA complex was Ri = 0.075 mM-i S""" and R2 = 0.091 mM-i s-\ When an equimolar amount of EDTA was added to solutions of Ni2+, R^ was reduced to 58% of the value of Ri in the absence of EDTA, R2 was reduced to 34% of the value of R2 in the absence of EDTA, and Tip/T2p was 1.4. The remaining paramagnetic relaxivity due to the NiEDTA complex was Ri = 0.55 mM-"" s-"" and R2 = 0.78mM-"' s-i. The observed change in the chemical shift caused by the paramagnetic metal, A(0, at 300 MHz for a 1 0 mM solution of C0CI2 was measured to be 1 58 Hz. The measured change in the chemical shift, Aco, is the difference between the chemical shift of water protons in water containing only buffer and the chemical shift of water protons in a solution of a paramagnetic metal. This was measured with tetramethyisilane as an external reference using coaxial tubes. The influence of this chemical shift on the transverse relaxation of Co(H20)62+ solutions at 20 MHz and 300 MHz was estimated using equation 2-3 (Table 22). In the calculations, I used Xm = 7.4 x 10-7 s determined by Swift and Connick (1962) using 1^0 exchange. Also, I used Tim = T2m as predicted by the Solomon-Bloembergen equations for Co(H20)62+ at 20 MHz and 300 MHz. From the observed Ti at 300 MHz, Tim was calculated (from equations 2-1 and 2-2) to be 6.62 x 1 0-^ s. From the data of Bernheim et al. (1 959) Ti ^ = 5.95 x 1 04 s at 20 MHz. Since TimTm, the fast exchange limit applies at 20 MHz and at 300 MHz. AcOm was calculated from Aco at 300 MHz using Aco = p' AcOm which is valid in the fast exchange limit (Swift and Connick, 1962). I determined AcOm to be 1 .46 X 105 Hz at 300 MHz. Since AcOm is directly proportional to the precessional frequency, Aco^ at 20 MHz is 9.73 x 103 Hz. The results using

PAGE 22

16 these values with equation 2-3 are in Table 2-2 and are in good agreement with the experimental results in Table 2-1. Discussion The results at 300 MHz of Tip/T2p much greater than unity for solutions of both Co(H20)62+ and Ni(H20)62+ suggest that a relaxation mechanism different from the dipole-dipole mechanism is present at this frequency. The data in Table 2-1 indicate that this mechanism primarily, if not exclusively, affects the transverse relaxation. I have considered three possible causes of this fielddependent increase in 1/T2p. The scalar contribution to 1/T2m has a correlation time equivalent to the electron relaxation time which is field dependent (Bloembergen and Morgan, 1961). However, I have rejected a contribution from the scalar term because it would require an unreasonably large increase in the electron spin relaxation time at 300 MHz. When the electron relaxation time is less than both the rotational correlation time and x^, a Curie spin mechanism can result (Gueron, 1975). This Curie spin is the thermal average of an electron spin which can become large at high field and can cause an increase in Tim/T2mHowever, I also reject a contribution from this mechanism because the rotational correlation time for a hexaaquacomplex is too short to allow the Curie spin to become effective. Melamud and Mildvan (1975) observed a field-dependent increase in 1/T2p for water protons in solutions of Co(H20)62+ and concluded that the AcOm^ term in equation 2-3, a chemical shift mechanism, made a major contribution at the higher frequencies. I support this conclusion in two ways. First, I have calculated the magnitude of the chemical shift contribution to the transverse relaxation of Co(H20)62+ solutions at 20 MHz and 300 MHz using equation 2-3. These calculations demonstrate that the observed value of Tip/T2p at 300 MHz is consistent with a chemical shift mechanism which affects T2p (compare

PAGE 23

17 Table 2-2: Calculated paramagnetic relaxivities of water protons in solutions containing Co2+. frequency Ri R2 Tip/Tap (MHz) (mM-1 s-1) (mM-1 s'l) 20 0.18 0.19 1.1 300 0.16 1.8 11.2 Note: Calculated using equations 2-2 to 2-4 in the text. The calculations used Tm=7.4 X 10-7 s (Swift and Connick, 1962), the 20 MHz relaxation data from Bernhelm etal. (1959), and the 300 MHz relaxation data and chemical shift determined in this work. A further description is in the text.

PAGE 24

18 Tables 2-1 and 2-2). The chemical shift does not have an effect at 20 MHz at which frequency AcOm21/T2mXm. 1^2m2. The large chemical shift at 300 MHz due to the addition of Co2+ is the result of contact interactions which result from the delocalization of unpaired electron spin density. At lower magnetic field strengths this mechanism has been observed to affect Jz for i^o and in cobalt complexes (Swift and Connick, 1962; Eisenstadt, 1969). The effects on the relaxation when EDTA is added to solutions of Co(H20)62+ is the second way I support the chemical shift mechanism. The chemical shift mechanism will affect T2p to the greatest extent when the protons are in the primary hydration shell of the ion because it is a contact interaction. EDTA binds to Co2+ (and Ni2+) occupying all coordination sites and thus excludes water molecules from the inner shell. Thus the addition of EDTA would prevent the water protons from experiencing the chemical shift rriechanism. The ratio Tip/T2p was near unity for water protons at 300 MHz when an equimolar amount of EDTA was added to solutions of Co2+, consistent with the absence of a chemical shift relaxation mechanism and the dominance of a dipole-dipole relaxation mechanism (Figure 2-1 ). When EDTA was added to solutions of Co2+, the percent decrease in the relaxivity R2 at 300 MHz was much greater than Ri (Figure 2-1). Since EDTA is excluding water molecules from the inner coordination shell, this suggests that R2 has a much larger contribution from the relaxation of water protons in the inner shell. The residual paramagnetic relaxivity of the cobalt-EDTA complex is most likely due to water protons in an "outer shell" of the complex. This residual relaxivity makes up a much larger part of Ri than R2 observed for Co(H20)62+ solutions at 300 MHz. This suggests that the relaxation of outer shell water contributes to a greater extent to Ri than to R2 in Co(H20)62+.

PAGE 25

1 9 As EDTA was added to solutions of Ni2+, R2 decreased more than Ri (Figure 2-2), though the decrease is not as large as seen in solutions of Co2+. Also, the residual relaxivity of the NiEDTA complex is much larger than in solutions of the CoEDTA complex, and this residual relaxivity makes up a large part of both Ri and R2. These results for Ni2+ suggest that the chemical shift mechanism is responsible for the increase in R2 at 300 MHz, but that Ri and R2 each have a large contribution of outer shell water.

PAGE 26

CHAPTER 3 HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF THE AQUEOUS LIGAND IN Co(H20)62+ AND Ni(H20)62+ Introduction The hydrogen / deuterium fractionation factor, (|), of a metal-bound water is the equilibrium constant for this isotope exchange reaction: M(HOH) + (HOD) M(HOD)+{HOH) [M(HOD)] / [M(HOH)] ({): (3-1) [HOD] / [HOH] This fractionation factor measures the tendency of deuterium to accumulate at the aqueous ligand of the metal relative to the deuterium content of bulk solvent. Fractionation factors are valuable in interpreting the effects of deuterium on kinetic and equilibrium constants because they represent individual contributions to isotope effects measured relative to the common reference of water. The fractionation of hydrogen isotopes in water bound to metal ions has been considered (Schowen and Schowen, 1982); however, there are few reports of the fractionation factor of the aqueous ligands of a metal. Using NMR methods, Melton and Pollack (1969) have measured such a factor for the water ligands of Cr(H20)62+ ((]) = 1 .00 0.03). They relied on the properties of a paramagnetic metal to enhance the NMR relaxation rate of protons of water exchanging rapidly between bulk solvent and the hydration shell of the metal. By measuring the proton relaxation rate as a function of the deuterium content of solvent, the fractionation factor of the metal-bound site can be determined. 20

PAGE 27

21 Using this metliod at a proton resonance frequency of 300 MHz, I have measured the fractionation factor of the aqueous ligand of cobalt in Co(H20)62+ and of nickel in Ni(H20)62+. In the case of Co(H20)62+ the value of the fractionation factor was dependent upon whether it was determined from Ti or T2; however, the corresponding values of (j) for Ni(H20)62+ were identical when obtained from Ti or T2. The analysis of the relaxation discussed in Chapter 2 can be used to interpret the fractionation factors, and allows a qualitative differentiation between the hydrogen / deuterium fractionation of inner and outer shell water in these complexes. I have then used the knowledge of the fractionation of Co(H20)62+ to study the solvent hydrogen isotope effects on the formation constants of the reactions of glycine, A/,A/-dimethylglycine, and acetylacetone with Co2+. An isotope effect on an equilibrium constant is the product of the reactant state hydrogenic site fractionation factors divided by the product of the product state hydrogenic site fractionation factors (equation 1-2). In a reaction of Co2+ with a bidentate ligand L-i represented as Co(H20)62+ + L-1 — > Co(H20)4L^ + 2 H2O (3-2) the solvent hydrogen isotope effect is a function of the fractionation of the hydrogenic positions of Co(H20)62+ as well as Co(H20)4L-' and H2O, and the fractionation factor of Co(H20)62+ represents the individual contribution of the hydrogenic positions of Co(H20)62+ to the overall isotope effect. Because the solvent hydrogen isotope effects on the formation constants were measured to be near unity in the case of each of the ligands, I determined that the contributions of the fractionation of the hydrogenic positions of the inner shell water in Co(H20)62+ and in the ligand complexes of cobalt can not account completely for the overall isotope effect. I suggest that the fractionation of outer

PAGE 28

22 shell water also makes a large contribution to the isotope effect on these reactions. Experimental Procedure NMR Measurements: Measurements at 300 MHz and solutions were the same as in Chapter 2. Some measurements of Ti of water protons in solutions of Co2+ were performed at frequencies between 0.01 and 50 MHz on a field cycling "relaxometer" at the IBM T. J. Watson Research Center, Yorktown Heights, NY (Koenig etal., 1983). Measurement of the Hydrogen / Deuterium Fractionation Factor: The measurement of the relaxation of water protons in solutions of varying deuterium content and containing paramagnetic metal ions is given by 1/Tip = p'[X]/(1 n + n(t)) (3-3) where n is the atom fraction of deuterium in solvent water, ^ is the hydrogen / deuterium fractionation factor of the water ligands of the metal, and [X] is the term in brackets in equations 2-2 and 2-3 for i = 1 or 2, respectively. This equation is derived considering the fractionation of isotopes that arises when there is a deuterium content in water. In this case the ratio of protons in the hydration shell of the metal to total protons in solution is pV(1-n+n(t)) (Schowen and Schowen, 1982). Equation 3-3 can be rearranged to pTip = [X]-i -n(1-(j))[X]-i (3-4) Thus pTip vs n gives a slope of -(1-(1))[X]-"' and an intercept of [X]-1, and (j) = 1 + slope / intercept.

PAGE 29

23 Determination of the isotope effects on formation constants: The formation constants for the reactions of Co2+ with a bidentate ligand, L-"", as in equation 3-2 are defined as [Co(H20)4L1 ki = (3-5) [Co(H20)62i [L-1] with similar expressions for k2 and ka. Potentiometric titrations were performed in H2O and D2O (99.8%) containing 1 mM C0CI2 and 4 mlVl glycine, N,Ndimethylglycine, or acetylacetone. The values of pH and pD used in these experiments were corrected from pH meter readings. The correction of a pH meter reading in 100% D2O is pD = meter reading + 0.4 (Glasoe and Long, 1960). The formation constants were determined by a non-linear least-squares fit of the data (RS1 BBN Software, Cambridge, MA) to the equation derived by Carlson etal. (1945) describing n, the ratio of the concentration of metal-bound ligand to total concentration of metal, as a function of the concentration of coordinating ligand, [A]. ki[A] + 2kik2[A]2 + 3kik2k3[A]3 n= — (3-6) 1 + ki[A] + kik2[A]2 + kik2k3[A]3 The calculation of n and [A] required knowledge of the acid ionization constants of the free ligands, which were measured by potentiometric titrations in H2O and D2O (Table 3-1). The negative logarithm of the acid ionization constants, pK, in H2O were all within 0.07 pK units from the literature values (Smith and Martell, 1975). The solvent hydrogen isotope effects for glycine expressed as ApK where ApK = pKd20 PKH20 were within 0.03 ApK units from the values obtained previously by Jencks and Salvesen (1971).

PAGE 30

24 Table 3-1 : Solvent hydrogen isotope effects on the acid ionization constants of glycine, A/,A/-dimethylglycine, and acetylacetone. (PK1)H20^ ApKi (PK2)H20'' ApK2 glycine 2.37 0.02 0.40 0.04 9.72 0.03 0.56 0.05 A/,A/-dimethylgIycine 1.96 0.03 0.40 0.04 9.81 0.03 0.55 0.05 acetylacetone 9.03 0.10 0.62 0.17 Note: Acid ionization constants were determined in H2O and D2O by potentiometric titrations at 23C. Solutions contained 10 mM of either glycine, A/,A/-dimethylglycine, or acetylacetone. 3 pK = -log K where K is the acid ionization constant. Standard errors resulted from a non-linear least-squares fit of the titration data. ApK = pKogO pKRaO-

PAGE 31

25 These potentiometric titrations were performed in solutions wliich contained ultra-pure C0CI2 (Aldrich Gold-Label), and glycine and N,Ndimethylglycine from Sigma and acetylacetone from Aldrich. Water used for solution preparations was distilled and passed through two ion-exchange resin cartridges (Cole-Palmer 1506-35). D2O (99.8%) was stirred with activated charcoal, the charcoal filtered out, and then the D2O was distilled. Glassware was rinsed with a solution of EDTA prior to use. Results Fractionation Factors of Co(H20)^2 and Nimo0^5 2: The relaxation of water protons in solutions of 10 mM C0CI2 and 10 mM NiS04 as a function of the atom fraction of deuterium in solvent water is shown in Figures 3-1 and 3-2. From these data, the hydrogen / deuterium fractionation factor, (j), for Co(H20)62+ and Ni(H20)62+ was obtained (Table 3-2). The value of the fractionation factor was independent of the concentration of metal in the range 0.010 M to 0.025 M. The value of (() for Co(H20)62+ was dependent on whether it was obtained from Ti or T2; however, the corresponding values of ^ for Ni(H20)62+ were identical when obtained from Ti or T2. In a separate experiment, the fractionation factor from Ti for Co(H20)62+ was determined at frequencies between .01 and 50 MHz on a field cycling "relaxometer" at the IBM T. J. Watson Research Center and found to be virtually field independent (data not shown). Solvent Hvdroqen Isotope Eff ects on the Formation Constants of Cobalt Qomplgxeg; The formation constants for the reactions of Co2+ with glycine, N,Ndimethylglycine, and acetylacetone were determined in H2O and D2O and are reported as the logarithm of the formation constants (Table 3-3). Representative data of a formation curve (Carlson etal., 1945) for Co2+ and glycine are in Figure 3-3 where n is plotted as a function of the negative logarithm of the

PAGE 32

26 Figure 3-1 : The dependence of pTjp (i = 1 ,2) at 300 MHz on n, the atom fraction of deuterium in solvent water, for a 10 mM solution of C0CI2 at 23 C. The solutions were buffered at pH 4.4 with 0.5 mM acetic acid. Data are the means and standard errors propagated from NMR measurements.

PAGE 33

27 12.0 Atom fraction of deuterium in solvent (n) Figure 3-2: The dependence of pTjp (i = 1 ,2) at 300 MHz on n, the atom fraction of deuterium in solvent water, for a 10 mM solution of NiS04 at 23 C. The solutions were buffered at pH 4.4 with 0.5 mM acetic acid. Data are the means and standard errors propagated from NMR measurements. p'Tip ( ) pT2p ( ).

PAGE 34

28 Table 3-2: Fractionation factors determined from Ti and T2 at 300 MHz for the aqueous ligands of the paramagnetic metal in Co{H20)62'*" and Ni(H20)62"^. Co{H20)62+ 0.95+.01 0.73.02 Ni(H20)62+ 0.99.01 0.98.03 Note: Measurements were made at 23 C for 10 mM C0CI2 or NiS04 containing 5 X lO-'^ M acetic acid buffer at pH 4.4. Data were obtained from the least squares slope and intercept and standard error in a plot such as Fig. 3-1 according to equation 3-4.

PAGE 35

29 Table 3-3: Logarithm of the formation constants determined for the reactions of Co2+ with the iigands glycine, A/,A/-dimethy!glycine, and acetylacetone in H2O and D2O. H2O D2O (logk)D20-(logk)H20 glycine logki 5.04 0.04 4.99 0.02 -0.05 0.04 logk2 4.32 0.04 4.23 0.02 -0.08 0.05 A/,A/-dimethylglycine logki 4.14 0.04 4.12 0.06 -0.03 0.07 logk2 3.20 0.06 3.18 0.13 -0.01 0.14 acetylacetone log ki 5.30 0.07 5.42 0.10 0.12 0.13 logk2 4.40 0.09 4.61 0.12 0.21 0.15 Note: Determined by potentiometric titrations of solutions which contained 1 mM C0CI2 and 4 mM of the ligand and fitting the data to equation 3-6. Measurements were made at 23 C. See equation 3-5 for the definition of ki.

PAGE 36

30 0 T 1 1 1 2.5 3.5 4.5 5.5 p[A] Figure 3-3: The ratio of the concentration of metal-bound ligand to total concentration of nnetal, n, as a function of the negative logarithm of the concentration of free glycine in solution existing as NH2-CH2-COO-'', p[A] (see equation 3-6). The solutions contained 1 mM total C0CI2 and 4 mM total glycine in H2O ( ) or D2O { ) and were titrated with NaOH (NaOD). The solid lines are a computer fit to the data using equation 3-6 with the values of the formation constants given in Table 3-3.

PAGE 37

3 1 concentration of coordinating ligand, p[A]. In the region of the formation curves where p[A] = 3.0 the pH of the solutions was greater than 9.0. Near this pH the hydrolysis of the cobalt-bound water to a cobalt-bound hydroxide occurs in Co(H20)62+ (Baes and Mesmer, 1976) and this same hydrolysis may be occurring in the water ligands of the cobalt complexes with the bidentate ligands. The data in the region p[A] = 3.0 was therefore difficult to interpret, and I was unable to obtain precise values for k^. For glycine and acetylacetone in H2O the values of log ki were within 0.1 log unit and the values of log k2 were within 0.3 log unit of the literature values (Smith and Martell, 1975; no literature value was available for A/,A/-dimethylglycine). The solvent hydrogen isotope effects on these formation constants were all unity within experimental uncertainty (Table 3-3), with the possible exception of k2 for acetylacetone. For the binding of a ligand L-i to cobalt as in equation 3-2 the solvent hydrogen isotope effect is (see equation 1 -2) (ki)H20 ((t)Co(H20)62+)^2 = (3-7) (ki)D20 ((t)Co(H20)4(L)-^)^(())H20)^ and from this the fractionation factor of the remaining metal-bound waters in Co(H20)4L+ can be calculated by using the measured isotope effect, the measured fractionation factor for Co(H20)62+ from NMR relaxation, and the fractionation factor for H2O which is 1.0 (Schowen and Schowen, 1982). An analogous calculation can be done to determine the fractionation factor of the remaining metal-bound waters in Co(H20)2(L)2. Fractionation factors (t)Ti and (t)T2 for Co(H20)62+ have been measured from the NMR relaxation times Ti and T2 of water protons (Table 3-2). The results of these calculations from equation

PAGE 38

32 3-7 with glycine, /\/,A/-dimethylglycine, and acetylacetone are in Table 3-4 using (j)Ti and (t)T2 for Co(H20)62+. I have also tried to measure the fractionation factor for these coordinated water molecules in the complexes directly, using the NMR relaxation method used to determine the fractionation factor of Co(H20)62+. However, these solutions of Co2+ and a bidentate ligand L" consist of an equilibrium mixture of Co(H20)62+ Co(H20)4L+ Co(H20)2L2 and C0L3and the observed paramagnetic relaxation is the sum of the relaxation of each of these four species. This presents difficulties in obtaining the fractionation factor of one species from these measurements. Discussion The definition of the hydrogen / deuterium fractionation factor I have given (equation 3-1) assumes the rule of the geometric mean to be valid (Schowen and Schowen, 1982); that is, the fractionation factor of one particular hydrogenic site is independent of the isotopic composition of any other site. I will assume that the rule is valid in my system, and this is supported by the high degree of linearity in plots as in Figures 3-1 and 3-2. A deviation from the rule of the geometric mean would result in a non-linear dependence of the relaxation data on the atom fraction of deuterium in the solvent. The value of the fractionation factor for the water ligands of cobalt in Co(H20)62+ was dependent on whether it was determined from Ti orT2, while that of Ni(H20)62+ was independent of Ti or T2 (Table 3-2). These fractionation factors are a weighted value of all the protons that experience the paramagnetic relaxation. An explanation for this difference in (^j^ and (j)T2 is based on the analysis of the contributions to the relaxation at 300 MHz (see Chapter 2). The paramagnetic contribution to the transverse relaxation of water protons in solutions of Co(H20)62+ has a predominant contribution of water in the inner

PAGE 39

33 Table 3-4: Calculated fractionation factors for the metal-bound water in the complexes of cobalt with glycine(gly), A/,A/-dimethylglycine (DMG), and acetylacetone (acac). Ti
PAGE 40

34 coordination shell, suggesting that (|)t2 is measuring more of the fractionation of inner shell water. Since Tip has a larger contribution of outer shell water, is more of the fractionation of outer shell water. The analysis of the relaxation for Ni(H20)62+ (Chapter 2) suggests that both Ti and T2 have a large contribution of outer shell water, so and for Ni{H20)62+ each represent a larger contribution of the fractionation factor of outer shell water, and the influence of the fractionation of the inner shell water is not measured by or (j)T2. This interpretation of the fractionation factor for Co(H20)62+ is consistent with expectations for the fractionation factor of metal-bound water. The fractionation factor for H3O+ is 0.69 (Chiang et ai, 1980), presumably due to the positive charge on the oxygen. It has been suggested that a metal-bound water site ought to have a fractionation factor approximated by (0.69)5, with 5 measuring the degree of positive charge transferred from the metal to the oxygen (Schowen and Schowen, 1982). The fractionation factor from T2 for Co(H20)62+ being 0.73 supports this suggestion in resembling the fractionation factor for H3O. Pure water has by definition a fractionation factor of 1 .0. Outer shell water, being much more loosely bound than inner shell water, should be much more like bulk water in its hydrogen / deuterium fractionation properties (More O'Ferrall etai, 1971). The observation of ())ti for Co(H20)62+ and (s^j^ and (t)T2 for Ni(H20)62+ all being close to unity is consistent with these fractionation factors having a large contribution of the fractionation of outer shell water. The fractionation factor of the inner shell water in Ni(H20)62+ may well be affected by the positive charge of the nickel and have a value closer to 0.69, or it may have a value near unity; however, because of the large contribution of outer shell water to both Ti and T2 in solutions of Ni(H20)62+, the influence of the inner shell water is not large in the fractionation factors determined from Ti and T2.

PAGE 41

35 The usefulness of fractionation factors is judged by their ability to aid in interpretation of solvent hydrogen isotope effects. I have determined the solvent hydrogen isotope effects on the formation of some simple inorganic complexes of Co2+ with the goal to use the knowledge of the fractionation of Co(H20)62+ to interpret those isotope effects. An isotope effect is a function of all the fractionation factors for each hydrogenic site in a reaction (equation 1-2), and for a reaction such as equation 3-2 the fractionation factor of Co(H20)62+ represents the Individual contribution of those hydrogenic sites to the overall isotope effect on the reaction. In considering the solvent hydrogen isotope effect on the formation constants (Table 3-3), initially I will assume that the isotope effect is made up entirely of the contribution of the isotope preferences of the hydrogenic positions of water in the Inner coordination shell of Co(H20)62+, Co(H20)4L+, and Co(H20)2L2. With this assumption there is enough information from the isotope effect and the fractionation factor of Co(H20)62+ to calculate (from equation 3-7) the fractionation factor of the remaining waters in the cobaltbidentate ligand complexes (Table 3-4). While I have done the calculations with equation 3-7 using both and for Co(H20)62+ (Table 3-2), the values from ())Ti may not be meaningful since Ti is measuring more outer shell water, and the number of outer shell waters is not known. One interpretation of these calculations is that since (|>t2 of Co(H20)62+ measures more of the fractionation of inner shell water in Co(H20)62+, the calculation using of Co(H20)62+ represents the resulting fractionation factor of the inner shell waters in the cobalt complexes. As ligands are added to Co2+ the calculated fractionation factor of the remaining coordinated waters decreases, and this decrease is the same for each of the three ligands, though glycine and A/,A/-dimethylglycine coordinate through an amino and carboxylate, and acetylacetone binds through two

PAGE 42

36 I carbonyls. If the assumption in these calculations is valid, the results of a lower fractionation factor for Co(H20)4L+ means that the 0-H bond becomes weaker relative to that bond in Co(H20)62+, and the 0-H bond is also weaker in Co(H20)2L2 compared to Co(H20)4L+. This is similar to the effect of added ligands on the lifetime of water molecules in the hydration shell which, for instance, is shorter in Co(H20)4gly+ relative to Co(H20)62+ (Hammes and Steinfeld, 1962), meaning that the hydration shell becomes more loosely bound when ligands are added. Thus in these cobalt complexes with bidentate ligands the metal-oxygen bond of a coordinated water is looser, and the oxygen hydrogen bond appears looser, relative to Co(H20)62+. However, from the measured solvent hydrogen isotope effects (Table 33) and the calculations of Table 3-4, there are a number of reasons to question the assumption that the only contribution to the isotope effects comes from the hydrogens of inner shell water. First, such strong fractionation as 0.36 (Table 34) is rare (Schowen and Schowen, 1982). Second, the lower fractionation factor for Co(H20)4L+ and Co(H20)2L2 is In contrast to the suggestion that the charge on the cobalt affects the fractionation of the metal-bound water (Schowen and Schowen, 1982). In these complexes, the net charge on the complex is +1 for Co(H20)4L+, and neutral for Co(H20)2L2, since the ligand has a charge of -1 In each case. Thus one would expect the fractionation factor of the waters in the ligand complexes to be closer to unity, according to the expectation that (j) = (0.69)5, where 0.69 is the fractionation factor of H3O+ and 5 is the degree of positive charge transferred from the metal to the oxygen of the coordinated water molecule (Schowen and Schowen, 1982). The (j)T2 for Co(H20)62+ is consistent with this (Table 3-2), and thus one would expect [([) for Co(H20)62+] < [(t) for Co(H20)4L+] < for Co(H20)2L2], but this opposite to what was observed (Table 3-4). Finally, based upon the large number of hydrogenic

PAGE 43

37 sites in Co(H20)62+, large Isotope effects would be expected if the only contributions to the isotope effect came from the fractionation of inner shfell water (Kresge et al., 1987). However, isotope effects of unity were observed for these reactions (Table 3-3). I suggest that the individual contribution of the fractionation factor of the inner shell waters in Co(H20)62+, Co(H20)4L+, and Co(H20)2L2 to the overall solvent hydrogen isotope effect on the formation constants is small. The fractionation factor of some other hydrogenic sites has a contribution to the isotope effect which is sufficient to cancel the contribution of the inner shell waters so that the overall isotope effect is unity. The most likely choice for this is the fractionation of outer shell water. Other possibilities include, in the case of glycine, the composite fractionation factor of the hydration shell of the carboxylate group, but this would be a small contribution of a fractionation factor value already close to unity (the fractionation factor of the hydration shell of the carboxylate group in the acetate ion is 0.89, Goodall and Long, 1968). The fractionation factor of the hydrogens on the nitrogen in glycine are probably not important since the isotope effect is the same for A/,A/-dimethylglycine (Table 33). The importance of outer shell water in ligand substitution reactions is not without precedent. The first step in the proposed mechanism of ligand substitution reactions is the formation of an outer shell complex with the ligand and metal (Eigen and Tamm, 1962), followed by replacement by the ligand of a water molecule in the inner coordination shell of the metal. Hence it is reasonable to assume that the binding of a ligand would affect the stmcture of outer shell water. The following example illustrates how the fractionation of outer shell water could contribute to the isotope effect. Equation 3-7 can be expanded to

PAGE 44

38 include the fractionation factor of outer shell water of Co(H20)62+ in the numerator and of Co(H20)4L+ in the denominator. The number of outer shell waters would be larger than the number of inner shell waters, and if the contribution of an outer shell hydrogenic site in equation 3-7 was 1 .06, twelve outer shell waters (24 hydrogenic positions) would contribute (1 .06)2^ = 4.05. This would then mean that the fractionation factor of the inner shell waters in Co(H20)4L+ is 0.74. If the contribution of an outer shell hydrogenic site is 1 .1 0, then the fractionation factor of the inner shell waters in Co(H20)4L+ is 0.83. Thus a contribution of outer shell water slightly larger than unity would, when raised to a large power for the number of hydrogenic positions, cancel the effect of the inner shell waters. I conclude that the solvent hydrogen isotope effect of unity on the formation of complexes of Co2+ and glycine, A/,A/-dimethylglycine, acetylacetone points to the importance of outer shell water in these reactions. Conversely, one can not get information about the effect of added ligands on the fractionation properties of the inner shell waters from the solvent hydrogen isotope effect on the formation constants. This work shows that the solvent hydrogen isotope effect is the complex contribution of the hydrogen / deuterium fractionation of several different groups or species of water.

PAGE 45

CHAPTER 4 HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF THE AQUEOUS LIGAND OF COBALT IN Co(ll)-SUBSTITUTED CARBONIC ANHYDRASE Introduction Carbonic anhydrase is a zinc metalloenzyme which catalyzes the reversible hydration of carbon dioxide to produce bicarbonate and a proton. CO2 + H2O HCP+ H^ The aqueous ligand of the metal at the active site of carbonic anhydrase is believed to have a role in the catalytic pathway. Direct nucleophilic attack by the metal-bound hydroxide on CO2 is a likely step in the production of bicarbonate (Pocker and Sarkanen, 1978; Lindskog, 1983; Silverman and Lindskog, 1988), and an intramolecular proton transfer involving the metalbound water is proposed to be a rate-limiting step in catalysis by isozyme II (Steiner etal., 1975). The solvent hydrogen isotope effect associated with the turnover number for CO2 hydration catalyzed by native zinc-carbonic anhydrase II is 3.8 (Steiner etal., 1975). To interpret fully the solvent hydrogen isotope effects observed for catalysis by carbonic anhydrase requires knowledge of the fractionation factor of the aqueous ligand of the metal. The fractionation factor measures the tendency of deuterium to accumulate at the aqueous ligand of the metal relative to the deuterium content of bulk solvent. There are few reports of the fractionation factor of the aqueous ligands of a metal at the active site of an enzyme. Using NMR methods, Silverman (1981) has measured such a factor for the water ligands of cobalt in cobalt(ll)-substituted carbonic anhydrase II ((j) = 1.05 0.17) by taking 39

PAGE 46

40 advantage of the properties of a paramagnetic metal to enhance the relaxation rate of water exchanging rapidly between bulk solvent and the hydration shell of the metal and measuring the proton relaxation rate as a function of the deuterium content of solvent. Silverman measured that fractionation factor using Ti at a proton resonance frequency of 100 MHz. Because of the unique relaxation properties of water protons in solutions of Co(H20)62+ at 300 MHz (Chapter 2), I have determined the fractionation factor of the aqueous ligand of cobalt in three mammalian isozymes of Co(ll)-substituted carbonic anhydrase from Ti and T2 at 300 MHz. In each case Ti > T2, implying that the same relaxation mechanisms occur for water protons in solutions of Co(ll)-substituted carbonic anhydrase at 300 MHz as in solutions of Co(H20)62+. The values of the fractionation factor determined from Ti for each isozyme of Co(ll)-substituted carbonic anhydrase were close to the fractionation factor for bulk water, which is unity. Except in the case of Co(ll)-substituted carbonic anhydrase III, the fractionation factors obtained from T2 were less than unity and close to the fractionation factor for H3O+ which is 0.69. I conclude that fractionation factors in these cases determined from Ti and T2 measured isotope preferences for different populations of ligand sites. I suggest that since T2 has a large contribution from a paramagnetic chemical shift, the fractionation factors determined from T2 have a large contribution of the hydrogen / deuterium fractionation of the inner shell ligand. The fractionation factors determined from Ti are close to unity, the value in bulk water, and contain a larger contribution of the fractionation of outer shell water. Experimental Procedure Enzvmes: Human carbonic anhydrase I was purified from human red blood cells by an affinity chromatography method (Kalifah etal., 1977). Bovine carbonic anhydrase II was obtained from Sigma Chemical and used after

PAGE 47

4 1 extensive dialysis to remove paramagnetic impurities. Bovine carbonic anhydrase III was obtained from bovine flank steak by gel filtration and anionexchange chromatography (Tu et al., 1986). The concentration of carbonic anhydrase I, II, and III was estimated at 280 nm using e = 4.7 x 10^ M-^ cm-"" (Nyman and Lindskog, 1964), 5.4 x 10^* M-i cm-i (Nyman and Lindskog, 1964), and 6.4 x 10"^ M-i cm-i (Engberg and Lindskog, 1984), respectively. The apoenzymes of carbonic anhydrase I and II were prepared according to the procedure of Hunt etal. (1977) by dialysis against dipicolinic acid. Co(ll) substituted carbonic anhydrase I and II were prepared by the addition of 1.1 equivalents of C0CI2 followed by dialysis against several changes of large volumes of deionized water. The apoenzyme of carbonic anhydrase III was prepared by addition of the chelator 2-carboxy-1 ,10phenanthroline to a solution of enzyme followed by dialysis against excess C0CI2 as described by Engberg and Lindskog (1984). NMR Measurements: Measurements at 300 MHz were the same as in Chapter 2. In the cases of Co(ll)-substituted carbonic anhydrase I and II, l/Tjo was taken as the relaxation of solutions of the Co(ll)-substituted enzyme in the presence of a molar excess of the inhibitor acetazolamide, which is known to displace the aqueous ligand of the metal at the active site. Enough acetazolamide was present to bind to 99.9% of the active sites. Alternatively for Co(ll)-isozymes I and II and solely for Co(ll)-substituted carbonic anhydrase III, 1/Tjo was taken as the relaxation of solutions of the native zinc enzyme, which has no paramagnetic contribution. These two methods were equivalent at a frequency of 300 MHz for Co(ll)-substituted carbonic anhydrase I and II. Solutions of enzyme for relaxation measurements were buffered with 50 mM Hepes [4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid] at pH 8.5. At this pH the relaxivity of solutions of Co(ll)-substituted carbonic anhydrase I and II is

PAGE 48

greatest and in a region that is independent of pH (Wells et al., 1979). (See Chapter 6 for a discussion of the pH and magnetic field dependence of the relaxation of water protons in solutions of Co(ll)-substituted carbonic anhydrase III.) The values of pH reported here are uncorrected pH meter readings. The correction of a pH meter reading in 100% D2O (pD = meter reading + 0.4) is approximately offset by the change in ionization state of the buffer in D2O (pKogO PKhzO = 0.5 0.1 for almost all acids with pK values between 3 and 10, Schowen, 1978). Water used for solution preparations was distilled and passed through two ion-exchange resin cartridges (Cole-Palmer 1506-35). D2O (99.8%) was stirred with activated charcoal, the charcoal filtered out, and then the D2O was distilled. Measurement of the Hvd roaen / Deuterium Fractionation Factor: The hydrogen / deuterium fractionation factor was determined by measuring the relaxation of water protons in solutions of Co(ll)-substituted carbonic anhydrase as a function of the atom fraction of deuterium in solvent water as described in Chapter 3. Results The substitution of cobalt for zinc at the active site of carbonic anhydrase provides a paramagnetic center which affects the relaxation of water protons in solutions of carbonic anhydrase. The paramagnetic relaxivity of these water protons is greater for R2 than for Ri for each isozyme of carbonic anhydrase (Table 4-1). The relaxation of water protons in solutions of the three isozymes of Co(ll)-substituted carbonic anhydrase as a function of the atom fraction of deuterium in solvent water is shown in Figures 4-1 4-2, and 4-3. The hydrogen / deuterium fractionation factors calculated from equation 3-4 using Ti were near unity for the three isozymes (Table 4-2). However, the fractionation factors

PAGE 49

43 Table 4-1 : Paramagnetic contribution to the relaxivities of water protons observed for aqueous solutions of Co(ll)-substitutecl carbonic anhydrase I, II, and III. frequency (MHz) Ri (mM-is-i) R2 (mM-is-'') Tif/T2p Co(ll)-carbonic anhydrase 1 300 0.28 2.3 8.2 Co(ll)-carbonic anhydrase 11 300 0.26 1.4 5.3 Co(ll)-carbonic anhydrase III 300 0.14 2.2 16 Note: fvleasurements were made at 23 C for solutions of enzyme buffered at pH 8.5 by 50 mM Hepes. Solutions contained 20% D2O by volume.

PAGE 50

44 3.8 2.6 H 1 1 1 1 0.0 0.2 0.4 0.6 0.8 1.0 Atom fraction of deuterium in solvent (n) 4.2 2.7 H r— 1 1 1 i 1 0.0 0.2 0.4 0.6 0.8 1.0 Atom fraction of deuterium in solvent (n) Figure 4-1 : The dependence of pTip (i=1 ,2) at 300 MHz on n, the atom fraction of deuterium in solvent water, for 1.5 mM solutions of Co(ll)-substituted carbonic anhydrase I. Solutions contained 50 mM Hepes buffer at pH 8.5 and 23C. Data are the means and standard errors of three measurements.

PAGE 51

4 5 Figure 4-2: The dependence of pTjp (i=1 ,2) at 300 MHz on n, the atom fraction of deuterium in solvent water, for 2.0 mM solutions of Co(ll)-substituted carbonic anhydrase II. Solutions contained 50 mM Hepes buffer at pH 8.5 and 23C. Data are the means and standard errors of three measurements.

PAGE 52

46 (A to o Q. 0.0 0.2 0.4 0.6 0.8 1.0 Atom fraction of deuterium in solvent (n) 0.0 0.2 0.4 0.6 0.8 1.0 Atom fraction of deuterium in solvent (n) Figure 4-3: The dependence of pTjp (i=1 ,2) at 300 MHz on n, the atom fraction of deuterium in solvent water, for 1.5 mM solutions of Co(ll)-substituted carbonic anhydrase III. Solutions contained 50 mM Hepes buffer at pH 8.5 and 23C. Data are the means and standard errors of three measurements.

PAGE 53

4 7 Table 4-2: Hydrogen/deuterium fractionation factors determined from Ti and T2 at 300 MHz for the aqueous ligand of the paramagnetic metal in three mammalian isozymes of Co(ll)-substituted carbonic anhydrase. Co(ll)-carbonic anhydrase I Co(ll)-carbonic anhydrase II Co(II)-carbonic anhydrase III Ti <^J2 1.10.02 0.72.02 0.95.02 0.77.01 1.03.06 1.00.07 Note: Data were obtained from the least squares slope and intercept and standard error in a plot such as Figure 4-1 according to equation 3-4. Measurements were made at 23 C for solutions of enzyme buffered at pH 8.5 by 50 mM Hepes.

PAGE 54

48 calculated using T2 were significantly less than unity for isozymes I and II, but near unity for isozyme III (Table 4-2). Discussion The relaxation behavior of water protons at 300 MHz and pH 8.5 in solutions of the Co(II)-substituted carbonic anhydrases showed Tip/T2p much greater than unity (Table 4-1). This was the same as observed for the water protons in solutions of cobalt ions (see Chapter 2), and the interpretation in the case of Co(H20)62+ can be applied to Co(ll)-substituted carbonic anhydrase. At 300 MHz, there is a field dependent increase in the chemical shift difference between a proton in the metal-bound site and the free solvent site. This causes an increase in the transverse relaxation of the protons as they experience the two environments (equation 2-3). Because this relaxation mechanism has as its source a paramagnetic contact shift, the transverse relaxation has a larger contribution of the ligand in the inner coordination shell of the cobalt, while the longitudinal relaxation has a larger contribution of water in the outer coordination shell. Thus the fractionation factor measured from T2 represents more of the fractionation of the inner shell aqueous ligand of cobalt in Co(ll)substituted carbonic anhydrase, while the fractionation factor measured from Ti is a greater contribution of the fractionation of outer shell water. A similar situation in which the outer shell relaxation has a larger contribution to Ri than to R2 has been reported by Kushnir and Navon (1984) for water protons in solution with Mn(ll)-substituted carbonic anhydrase II. These relaxation measurements were made at pH 8.5, a pH for which the relaxation rate of water protons in solutions of Co(ll)-substituted carbonic anhydrase I and II is greatest (Wells et al., 1979). (See Chapter 6 for an investigation of this property in solutions of Co(ll)-substituted carbonic anhydrase III.) This pH is above the value of the pK for the ionization of the

PAGE 55

49 active site group as determined in many experiments (see Table 5-1 and Engberg and Lindskog, 1984), especially spectrophotometric titrations and activity measurements, and the ligand of cobalt is believed to be a hydroxide ion at this pH. This presents some uncertainty as to which protons are actually being relaxed at pH 8.5 since a tetracoordinate cobalt-bound hydroxide ion would not be expected to exchange rapidly with water in solution. This issue has been discussed by Koenig et al. (1983) who proposed that ligand exchange with solvent involved a pentacoordinate intermediate having both OH" and H2O as ligands. Proton exchange can be rapid between these two ligands so that the departing H2O can contain the oxygen of the initially-bound OH". On the basis of ""SO-exchange kinetics, Tu and Silverman (1985) suggested that the NMR relaxation observed at pH 8.5 is that of a water molecule hydrogen bonded to the cobalt-bound hydroxide yet close enough to the metal to be strongly relaxed. Yet another possibility is the rapid exchange of the hydrogen of the metal-bound hydroxide without exchange of the oxygen. Because of these unresolved issues, I cannot state specifically which of the hydrogens near the metal are the main contributors to the observed fractionation factors. The fractionation factors measured from Ti for each Isozyme of Co(ll)substituted carbonic anhydrase were close to unity (Table 4-2), like for Co(H20)62+ (Table 3-2), consistent with these values having a large contribution from the fractionation of outer shell water. This is reasonable from distance considerations, as it can be calculated using a l/r^ dependence that the paramagnetic contribution to the longitudinal relaxation rate for a cobalt-bound hydroxide with a cobalt-proton distance of 2.8 A is about the same as the paramagnetic contribution of the protons of three water molecules at an average distance of 3.8 A. Also, it is known from crystallographic data that there

PAGE 56

50 are a large number of water molecules within the active site cavity. Specifically, in human carbonic anhydrase II there are an estimated 9 water molecules between histidine 64 and the zinc, a distance of about 6 A from the x-ray coordinates (Erickson et al., 1986). The fractionation factors measured from T2, (t)T2. were near 0.72 for both Co(ll)-substituted isozymes I and II, but 1 .0 for isozyme III (Table 4-2), and reflect a larger contribution of the hydrogen / deuterium fractionation of the inner shell aqueous ligand. It is difficult to account for these differences, however, and though these isozymes have very similar amino acid sequences, they have unique active site and kinetic properties (Silverman and Vincent, 1983; Silverman and Lindskog, 1988). Isozyme II has the highest CO2 hydration activity, with a turnover number of 10^ s-i. There is a histidine in position 64 which protrudes into the active site cavity of isozyme II. Isozyme I has a turnover number of 2 x 10^ s-\ and has His-64 as well as two other histidines, 67 and 200, in the active site cavity, and it is known that histidine 200 interacts strongly with the ligands of zinc (Kalifah, 1977). The active site cavity of isozyme I is smaller compared with that of isozyme II. Isozyme III has a turnover number of lO^s-"" and has a significant amount of positive charge in the active site, with a lysine in position 64 and an arginine in position 67. This positive charge may influence the ionization properties of the metal bound water in carbonic anhydrase III, for which the pK of the metalbound water is less than 5.5, compared with isozyme I and II which have a pK around 7 (Tu etal.. 1983; Engberg and Lindskog, 1984). This difference in charge in the active site may also affect the hydrogen / deuterium fractionation properties of the ligand of the metal in Co(ll)-substituted carbonic anhydrase III. Although isozymes I and II have very similar amino acid sequences, the significant differences in their structure, residues in the active site cleft, and catalytic activities (Pocker and Sarkanen, 1978; Lindskog, 1983; Silverman and

PAGE 57

5 1 Lindskog, 1988) apparently do not have a significant effect on the fractionation factors which were nearly identical for isozymes I and II.

PAGE 58

CHAPTER 5 INTERPRETATION OF SOLVENT HYDROGEN ISOTOPE EFFECTS ASSOCIATED WITH CARBONIC ANHYDRASE introduction A solvent hydrogen isotope effect on a reaction is made up of individual contributions to the overall isotope effect from each hydrogenic site in the reaction (Schowen, 1978). This contribution is measured by the hydrogen / deuterium fractionation factor of each particular hydrogenic site. In considering reactions of carbonic anhydrase, the fractionation factor of the aqueous iigand of the metal represents a reactant state contribution that can be used to interpret equilibrium solvent hydrogen isotope effects which are a function of reactant and product state site fractionation factors, and kinetic isotope effects which are a function of reactant and transition state site fractionation factors. Halide ions are inhibitors of carbonic anhydrase which bind to the metal and displace the water Iigand of the metal (Bertini etal., 1982). I have determined the solvent hydrogen isotope effect on the equilibrium binding of iodide ion to Co(ll)-substituted carbonic anhydrase I and II and interpreted this effect using the fractionation factor of the aqueous Iigand of cobalt in Co(ll)substituted carbonic anhydrase measured from NMR relaxation rates. The results suggest that the isotope effect on r binding binding is the complex contribution of the fractionation of more than one hydrogenic position, because the consideration of the fractionation factor of the aqueous Iigand of cobalt by itself does not allow a complete interpretation of the isotope effect. The fractionation factor of the aqueous Iigand of cobalt in Co(ll)substituted carbonic anhydrase can also be used to interpret kinetic isotope 52

PAGE 59

53 effects, and because a kinetic isotope effect includes the contribution of transition state hydrogenic sites, this allows a prediction of the transition state structure. In the case of Co(ll)-substituted carbonic anhydrase II the fractionation factor measured from NMR leads to a predicted transition state structure that includes a proton transfer to a water molecule in the active site cavity. This is consistent with the large isotope effect on kcat (Steiner et al., 1975) and the nonlinear dependence of kcat on the atom fraction of deuterium in solvent water (Venkatasubban and Silverman, 1980). Experimental Procedure The equilibrium dissociation constants describing the binding of iodide ion to Co(ll)-substituted carbonic anhydrase were determined from visible absorption measurements (Lindskog, 1966). The pH dependence of the absorbance at 640 nm in solutions of Co(ll)-substituted carbonic anhydrase I and II was measured on a Beckman DU-7 spectrophotometer. Stock solutions of enzyme at low pH contained 12.5 mM Hepes and 25 mM Mops (4morpholinepropanesulfonic acid) buffer, and at high pH 0.5 M of triethylenediamine. The stock solutions had equal concentrations of enzyme. The titrations were performed by adding small increments of the high pH enzyme solution to the low pH enzyme solution. In this way the enzyme concentration stayed the same though the volume increased slightly. These titrations were done in H2O and D2O in the presence and absence of iodide ion which has been shown to bind to the metal site in carbonic anhydrase II (Brown etal.. 1977). In the absence of anion, the data were fit (RSI BBN Software, Cambridge, MA) to a scheme involving 4 micro equilibrium constants (Simonsson and Lindskog, 1982; Bertini etal., 1985). In this scheme, only 3 of the 4 microconstants are independent, that is K1K3 = K2K4 (Scheme 5-1). When

PAGE 60

54 "HCoOH" *<1 ^ 3 ^ K. ""HECoOHs 1 Scheme 5-1 : A diagram of two significant ionizations at and near the active site of Co(ll)-substituted carbonic anhydrase II. One is the ionization of metal-bound water and the second is suggested to be the ionization of His-64 (Simonsson and Lindskog, 1982), the imidazole ring of which is about 6 A from the metal. For Co(ll)-substituted carbonic anhydrase I the second ionization could be either that of His-64 or His-200 (Whitney and Brandt, 1976).

PAGE 61

anion was present, Scheme 5-1 was expanded to include the binding of the inhibitor, and it was assumed that the anion binds only to species 1 and 2 (see Tibell etal. (1984) for an estimate of the accuracy of this assumption). Thus two separate binding constants were determined. Ki^' is the equilibrium dissociation constant of the anion complex with species 1 of Scheme 5-1 and is the equilibrium dissociation constant of the anion complex with species 2. In the absence of anion, the visible pH titration data fit Scheme 5-1 very well (Table 5-1). The constants Ki, K2, and K3 were determined from a fit to the data, and then K4 was calculated from Ki K2, and K3. The 4 microconstants determined in H2O for Co(li)-substituted carbonic anhydrase II agreed very well with those obtained by Simonsson and Lindskog (1982), and the values determined in H2O for Co(ll)-substituted carbonic anhydrase I also agreed fairly well with those obtained by Bertini etal. (1985). In the presence of iodide the pH dependence of the absorbance at 640 nm was shifted to higher pH. Iodide binds with greater affinity to the diprotonated species 1 in Scheme 5-1 than to the monoprotonated species 2 (Table 5-2). This is similar to the iodide inhibition results of Simonsson and Lindskog (1982) for native zinc carbonic anhydrase II determined from activity measurements, and the results of Bertini etal. (1985) for nitrate inhibition of Co(ll)-substituted carbonic anhydrase II determined from spectrophotometric titrations. The change in solvent from H2O to D2O has no measurable effect on Ki^'and K2I" for Co(ll)-substituted carbonic anhydrase I, but for Co(ll)-substituted carbonic anhydrase II the change in solvent from H2O to D2O results in a decrease in these equilibrium constants (Table 5-2).

PAGE 62

56 Table 5-1 : Solvent hydrogen isotope effects on the acid-base microequilibrium constants observed for Co(ll)-substituted carbonic anhydrase I and II. Co(II)-carbonic anhydrase I Co(ll)-carbonic anhydrase II value in H2O ApKa value in H2O ApKa pKi 7.55 0.05 0.54 0.06 5.65 0.04 0.42 0.05 PK2 7.58 0.15 0.45 0.19 5.73 0.07 0.37 0.10 PK3 8.11 0.15 0.45 0.19 6.95 0.08 0.38 0.12 PK4 8.09 0.22 0.54 0.28 6.87 0.12 0.43 0.1 7 Note: The microconstants were determined from the variation of the absorbance at 640 nm with pH at 23C and fit to Scheme 5-1 The solutions were buffered with Mops, Hepes and triethylenediamine with no other added ions as described in the experimental procedure. ApK = (pKi)D20-(pKi)H20

PAGE 63

57 Table 5-2: Solvent hydrogen isotope effects on the equilibrium dissociation constants of iodide ion with Co(ll)-substituted carbonic anhydrase I and II. Co(ll)-carbonic anhydrase 1 Co(ll)-carbonic anhydrase II value in H2O ApK^' ^ value in H2O ApKi-a pKii2.95 0.10 -0.09 0.1 7 3.48 0.13 0.47 0.19 2.35 0.06 -0.07 0.11 2.79 0.02 0.27 0.02 Note: The dissociation constants were determined from the pH dependence of the absorbance at 640 nm in the presence of 10 mM with conditions as in Table 5-1. Ki^" is the dissociation constant of Ibinding to species 1 of Scheme 5-1 and K2^' is the dissociation constant of Ibinding to species 2 of Scheme 51. aApK = (pKii-)D20-(pKii-)H20

PAGE 64

58 Discussion I have discussed in Chapter 4 the fractionation factor for the exchangeable hydrogen at the active site of Co(ll)-substituted carbonic anhydrase in its high pH form. The first application of the fractionation factors determined in Chapter 4 is to calculate the fractionation factor of the active site in its low pH form. Since the paramagnetic contribution to the relaxivity decreases to a very small value below pH 6.0 (Wells et al., 1 979), except at very low ionic strength (Bertini etal., 1978; Bertini etal., 1980), the fractionation factor cannot be measured directly from the relaxation rates at low pH. Instead, this was calculated from the fractionation factor of the high pH form and the solvent hydrogen isotope effect on the ionization constant of the active site, metal-bound water. The visible pH titration data of isozymes I and II indicate that more than one ionization at or near the active site governs the absorbance at 640 nm (Bertini etal., 1980). Previous investigators (Simonsson and Lindskog, 1982; Bertini et al., 1985) have attributed this pH dependence to the Ionization of two groups, the metal-bound water and another group near the active site which may be His-64 for isozyme II and probably His-64 or His-200 in isozyme I (Kalifah, 1977; Whitney and Brandt, 1976). I have measured the pH dependence of the absorbance at 640 nm for isozymes I and II in H2O and D2O and fit the data according to Scheme 5-1 to determine the solvent hydrogen isotope effect on the microequilibrium constants (Table 5-1). In Scheme 5-1 Ki and K4 are the constants for the metal-bound water, and K2 and K3 are the constants for the other active site group. Since the solvent hydrogen isotope effects on these equilibrium constants are all similar, within experimental uncertainty, it appears that the protonation state of one of the groups does not affect the hydrogen / deuterium fractionation properties of the other (consistent

PAGE 65

with the conclusion that active site properties do not seem to influence the fractionation factors of the aqueous ligand of cobalt in Co(ll)-substituted carbonic anhydrase I and II, see Chapter 4). Then the calculation to determine the fractionation factor of the water ligand of the metal can be performed neglecting the ionization state of the second ionizable group. The isotope effect on an equilibrium constant is the product of all of the reactant state fractionation factors divided by the product of all of the product state fractionation factors (equation 1-2). For this ionization at the active site of Co(II)-substituted carbonic anhydrase ^0 OH2 + H2O ^CoOH + HaO^ the solvent hydrogen isotope effect can be written Khp (^co-oh,)^(<1^H2o)^ '^P.o (0CO-OH ) (^l^KjO*) From this measured isotope effect, the measured fractionation factor for the cobalt-bound hydroxide, the fractionation factor for HaO"*" which has been measured to be 0.69 (Chiang, 1980), and the fractionation factor for H2O which is 1 .0, the fractionation factor for the exchangeable hydrogens at the active site in its low pH form can be calculated and are presented in Table 5-3. I have done the calculations of Table 5-3 using both and <})t2 measured for the high pH form of the aqueous ligand of cobalt in Co(ll)-substituted carbonic anhydrase I and II. However, the calculation from Ti may not be meaningful, since it contains a large contribution of outer shell water. The value calculated using represents more closely the fractionation factor value of the inner shell cobalt-bound water.

PAGE 66

60 Table 5-3: Values of the hydrogen / deuterium fractionation factor for the low pH forms of Co(ll)-substituted carbonic anhydrase. Co(ll)-substituted carbonic anhydrase 1 1.12 0.05 0.90 0.04 Co(ll)-substituted carbonic anhydrase II 0.90 0.03 0.81 0.03 Note: The ^ for the low pH form was calculated using the isotope effect on the ionization constant of the cobalt-bound water and the ^ for the high pH form measured directly from NMR and given in Table 4-2.

PAGE 67

6 1 To apply the fractionation factors to an equilibrium solvent hydrogen isotope effect associated with the enzyme, I have studied at equilibrium the binding of I" to Co(ll)-substituted carbonic anhydrase I and 11. I" is known to bind to the metal in Co(ll)-substituted carbonic anhydrase I and II displacing the coordinated water molecule (Brown et al., 1977). Also, I" binds fairly tightly (Table 5-2) and its size is comparable to that of a water molecule. Thus the isotope effect on the binding of I" to Co(ll)-substituted carbonic anhydrase may be simple enough to be accounted for by the fractionation factor of the metalbound water. The solvent hydrogen isotope effect on the binding of the iodide ion to the active site of carbonic anhydrase is given by 7C0-I + H2O ^CaOH2+ I (t-Co-OH,)' where (Kii")H20 is the equilibrium dissociation constant describing the binding of Ito species i of Scheme 5-1 in H2O. Because it is known that ^^20 = 1 -0, with knowledge of (t)co-OH2 should be possible to estimate an isotope effect. Using Co-OH2 ^rom Table 5-3, I thus predict that ApKi" = 0.09 0.04 for Co(ll)isozyme I, and 0.18 0.03 for Co(ll)-isozyme II. The predicted value for isozyme I is in rough agreement with the measured value (Table 5-2), and the predicted value for isozyme II is slightly below the measured value. However, the predicted value for isozyme I is less than the predicted value for isozyme II, which is the same as in the experimental case. I have considered other interactions which may be taken into account for r binding. One of these is the composite fractionation factor of the solvation shell of I" ((]) = 0.59, Voice, 1974).

PAGE 68

62 Including this in the denominator of the above equation for the isotope effect increases the predicted value, which is then closer to the experimental value for Co(ll)-isozyme II, but farther from the experimental value for Co(ll)-isozyme I. This implies that I have incomplete knowledge of the fractionation properties of each hydrogenic site that is affected when iodide ion binds to the enzyme. However, to include the fractionation factors of any other groups in the equation requires much speculation about the fractionation factors near the active site. Thus it is likely that the isotope effect must be described by the complex contribution of the fractionation factors of additional groups or water associated with the enzyme, and that the fractionation factor of the aqueous ligand of the metal is by itself not sufficient to account for the isotope effect on Ibinding. The discussion of the solvent hydrogen isotope effect on iodide binding suggests that the fractionation factor can be used to interpret at least qualitatively isotope effects associated with carbonic anhydrase. It is also possible to apply the fractionation factors for the aqueous ligand of the metal in Co(ll)-substituted carbonic anhydrase in the interpretation of isotope effects on kinetic constants associated with the enzyme to see if similar information can be obtained. A kinetic isotope effect is the product of the reactant state fractionation factors divided by the product of the transition state fractionation factors (equation 1-1). The isotope effect on kcat for the CO2 hydration activity of Co(ll)-substituted carbonic anhydrase II is 1.8 (R. S. Rowlett, unpublished), and the rate limiting step measured by kcat is the intramolecular transfer of a proton from the metal-bound water to regenerate the metal-bound hydroxide for the next round of catalysis (Steiner et ai, 1975). This proton transfer from the metal-bound water is thought to go through a water-bridge to histidine-64 in the active site cleft, from which the proton can then be released to solvent

PAGE 69

63 (Venkatasubban and Silverman, 1980). For this individual step in the catalysis, the reaction is C0-OH2 + H2O + His Co-OH + H2O + His-H* and the isotope effect is \o (Co-OH2)^(^H2o)^ __ D2O all transition state sites n (Di^ I The hydrogenic sites of water have a fractionation factor of 1 .0, and the hydrogenic sites of the cobalt-bound water have a fractionation factor of 0.81 (Table 5-3). Using these values and the isotope effect in the above equation, the product of the transition state fractionation factors can be calculated to be 0.36. In the transition state, the proton remaining on the metal-bound oxygen is in transition between a metal bound water which has a fractionation factor of 0.81 (Table 5-3), and a metal-bound hydroxide which has a fractionation factor of 0.77 (Table 4-2), so I will arbitrarily use a value of 0.79 as the fractionation factor of this hydrogenic site in the transition state. Therefore, the product of the fractionation factors of the other hydrogenic sites in the transition state is 0.36/0.79 = 0.46. These other hydrogenic sites in the transition state are the positions associated with the water bridge between the metal-bound water and the histidine in the active site cleft.

PAGE 70

64 -Co-OK m N^NH / o. t H H -Co-O' NH H This calculation for the fractionation factor of the water bridge in the transition state suggests that the protons of the water bridge have a fractionation factor less than unity, meaning they have hydronium ion character, consistent with positive charge transferred to the oxygen during the proton transfer process. This calculation of the fractionation factor of the transition state suggests that the proton transferred from the cobalt-bound water is in between the cobalt-bound water and the water bridge in the transition state. As in the case of iodide binding to Co(ll)-substituted carbonic anhydrase, there may be other hydrogenic sites in the transition state which contribute to the overall isotope effect. However, the fractionation factors of the aqueous ligand of the metal have led to this qualitative model of the transition state, which is consistent with both the large isotope effect on kcat and the results of Venkatasubban and Silverman (1980) for the catalysis of CO2 hydration in H2O / D2O mixtures.

PAGE 71

CHAPTER 6 MAGNETIC FIELD AND pH DEPENDENCE OF THE WATER PROTON NMR RELAXATION ENHANCEMENT IN SOLUTIONS OF Co(II)-SUBSTITUTED CARBONIC ANHYDRASE III Introduction Carbonic anhydrase III is a mammalian isozyme of carbonic anhydrase found primarily in skeletal muscle. It is a less efficient catalyst of CO2 hydration compared with isozymes I and II, which are found in red blood cells (Lindskog, 1983). Another distinctive feature of carbonic anhydrase III is the pH independence of its catalytic activity in the range pH 6-9, in which carbonic anhydrase II has an inflection at pH 7 and maximal activity at high pH (Tu etal., 1983; Kararii and Silverman, 1985; Tu etal., 1986). It is also apparent that the zinc is bound more tightly in carbonic anhydrase III than in isozymes I and II (Engberg and Lindskog, 1984). The replacement of cobalt for zinc in carbonic anhydrase I and II has proved useful to study the properties of the metal at the active site (Bertini and Luchinat, 1983). The catalytic and visible spectral properties of Co(ll)substituted carbonic anhydrase III have been investigated (Engberg and Lindskog, 1984). I report here the NMR relaxation enhancement of water protons in solutions of Co(II)-substituted carbonic anhydrase III. I found, in contrast to Co(ll)-substituted carbonic anhydrase I and II, no pH dependence of the NMR relaxation in the range of pH 6-9. Also, the frequency dependence of the paramagnetic contribution to the relaxation shows an increase with higher frequency, as opposed to the dispersion to lower relaxation rates observed in solutions of Co(II)-substituted carbonic anhydrase I and II. 65

PAGE 72

66 Experimental Procedure The purification of bovine carbonic anhydrase III and preparation of Co(ll)-substituted carbonic anhydrase III was performed as described in Chapter 4. The total concentration of carbonic anhydrase III was estimated at 280 nm by using e = 6.4 x 10^* M-i cm-i (Engberg and Lindskog, 1984). The concentration of Co(ll)-substituted carbonic anhydrase III was estimated at 640 nm by using the extinction coefficients determined by Engberg and Lindskog (1 984). Zinc analysis of the native enzyme yielded 1 .03 0.05 zinc per molecule of enzyme. Cobalt analysis of the Co(ll)-substituted enzyme yielded 0.33 0.04 cobalt per molecule of enzyme, with the remainder apoenzyme. The fraction of the sample that was apoenzyme was restored to full activity by the addition of zinc. The longitudinal relaxation rates of the protons of solvent water in solutions of Co(ll)-substituted and native Zn-carbonic anhydrase III were measured using a field cycling "relaxometer (Koenig etal., 1983) at the IBM T. J. Watson Research Center in Yorktown Heights, New York. The rates were measured at 5C at pH 6.0 (50 mM Mes/NaOH buffer), pH 7.0 (50 mM Mops/NaOH buffer), pH 8.0 (50 mM Hepes/NaOH buffer), and pH 9.0 (50 mM Ches/NaOH buffer) in solutions containing 1 mM of enzyme. The pH 6 samples were also measured at 25C. The samples were in equilibrium with atmospheric CO2. The total reiaxivity, Rt, is defined as in equation 6-1 Rt = (1/Ti-1/Tio)/N (6-1) where I/T1 is the observed relaxation rate of the protein solution, I/T10 is the relaxation rate of the buffer, and N is the millimolar concentration of protein. At each pH the reiaxivity was determined as a function of frequency over the range

PAGE 73

67 0.01 to 50 MHz. The paramagnetic relaxivity, Rp, is Rt determined for a solution of Co(ll)-substituted carbonic anhydrase III minus Rt for a solution of native Zncarbonic anhydrase III. Results The frequency dependence of the total relaxivity was similar to the NMR dispersion profiles of other native and Co(ll)-substituted carbonic anhydrases (Fabry etal., 1970; Wells etal., 1979) (Figure 6-1). The total relaxivity is greatest at low frequency, then begins to decrease at about 0.1 MHz. The paramagnetic relaxivity, however, showed a small value at low frequency, then at about 0.5 MHz Rp began to increase to a maximal value at about 5 MHz, and then began to decrease again (Figure 6-2). The total reiaxivities for both the Co(ll)-substituted and native Zn-carbonic anhydrase III showed a possible increase with pH at 1 MHz, but remained constant with pH at 40 MHz (Figure 63). This may be the same effect reported by Wells etal. (1979), who observed that the diamagnetic component of the relaxivity in solutions of Co(ll)-substituted and native zinc carbonic anhydrase II increases with pH at low frequency. When only the paramagnetic relaxivity was considered (the difference between Rt of Co(ll)-substituted carbonic anhydrase III and Rt of native zinc carbonic anhydrase III shown in Figure 6-3), there was no pH dependence, especially when the data at pH 7 in Figure 6-3 are excluded. Discussion This work is the first study of NMR relaxation enhancement of water protons in solutions of Co(ll)-substituted carbonic anhydrase III. Previous work with isozyme III has shown that the visible spectrum of Co(ll)-substituted carbonic anhydrase III is very similar to the high pH spectral forms of Co(ll)substituted carbonic anhydrase I and II; however, for Co(ll)-substituted carbonic anhydrase III there is no pH dependence of the spectrum in the range pH 6-9

PAGE 74

68 cq 0.210" 10 frequency (MHz) 10 10 Figure 6-1 : The total relaxivity of Co(II)-substituted carbonic anhydrase III and native zinc carbonic anhydrase III as a function of the proton resonance frequency. The solutions contained 1 mM enzyme at pH 6.0 (50 mM Mes / NaOH), pH 8.0 (50 mM Hepes / NaOH), or pH 9.0 (50 mM Ches / NaOH). pH 6.0 Co(ll)-substituted carbonic anhydrase III at 5C ( a), pH 6.0 native zinc carbonic anhydrase III at 5C ( A ), pH 9.0 Co(ll)-substituted carbonic anhydrase III at 5C (), pH 9.0 native zinc carbonic anhydrase III at 5C ( ), pH 8.0 Co(ll)-substituted carbonic anhydrase III at 23C (o ), pH 8.0 native zinc carbonic anhydrase III at 23C ( • ).

PAGE 75

69 10 10 -2 10-' 10 frequency (MHz) 10 10 Figure 6-2: The paramagnetic relaxivity of Co(ll)-substituted carbonic anfiydrase III as a function of the proton resonance frequency. Experimental conditions as in Figure 6-1 pH 6.0 ( A ), pH 9.0 ( ).

PAGE 76

70 (0 0.8 X (0 o 9.0 10.0 PH Figure 6-3: The total relaxivity of Co(ll)-substitutecl carbonic anhydrase 111 and native zinc carbonic anhydrase III as a function of pH. Experimental conditions as in Figure 6-1. Co(ll)-substituted carbonic anhydrase III measured at 1 MHz ( A ), native zinc carbonic anhydrase III measured at 1 MHz ( A ), Co(ll)substituted carbonic anhydrase III measured at 40 MHz ( ), native zinc carbonic anhydrase III measured at 40 MHz ( ).

PAGE 77

7 1 (Engberg and Lindskog, 1984). Also, the CO2 hydration activity of the cat isozyme III and the bovine isozyme III is independent in the pH range 6-8.5 (Tu etal., 1983; Engberg etal., 1985). These data imply that the activity controlling group in Co(ll)-substituted and native Zn-carbonic anhydrase III, the aqueous ligand of the metal, is predominantly in the hydroxide form in the pH range 6-9, and that the pK for the ionization of the metal-bound water to the metal-bound hydroxide is well below 6. NMR relaxation enhancement studies with Co(ll)-substituted carbonic anhydrase I and II have shown a large pH dependence of the paramagnetic relaxation rate in the pH range 6-9, with an inflection at pH 7 and maximal relaxation at high pH (Fabry etal., 1970; Wells etal., 1979). This correlates with the pH dependence of both the visible spectmm and CO2 hydration activity (Bertini etal., 1982; Silverman and Vincent, 1983). Thus the NMR relaxation enhancement seems to correlate with the ionization state of the aqueous ligand of the metal in Co(ll)-substituted carbonic anhydrase I and II. However, it has been shown that SO42present in solution to maintain ionic strength affects the pH dependence of both the visible spectrum and the paramagnetic relaxation of Co(ll)-substituted carbonic anhydrase I and II (Bertini etal., 1978; Bertini ef a/.,1980; Simonsson and Lindskog, 1982). When SO42and other anions are absent in either unbuffered solutions or solutions containing only sulfonic acid buffers, the visible spectrums of Co(ll)-substituted carbonic anhydrase I and II still show a large pH dependence, but there is practically no pH dependence on the paramagnetic contribution to the longitudinal relaxation rate of water protons in solutions Co(ll)-substituted carbonic anhydrase II (Bertini etal., 1978), and a small dependence for Co(ll)substituted carbonic anhydrase I (Bertini etal., 1980). The data in Figure 6-3 are from solutions containing no anions other than those of the sulfonic acid

PAGE 78

72 buffer. Under these conditions, the relaxation of water protons in solutions of Co(ll)-substituted carbonic anhydrase II would be independent of pH in the range 6-9, and this is what was observed for the paramagnetic relaxivity of solutions of Co(ll)-substituted carbonic anhydrase III (Figure 6-3). The reason for the small value of Rp at low frequency (Figure 6-2) is not clear. One possibility is that the lifetime of a proton in the hydration shell of the cobalt is too long to allow for a large paramagnetic contribution (Fabry et al., 1970; Tu etal., 1985). This water off rate has a maximal value of 10"^ s-"" in the case of native carbonic anhydrase III (Tu et al., 1983), but is 10^ s-i for Co(ll)substituted carbonic anhydrase II (Tu etal., 1985). This rate would be expected to be even slower at 5C, but when the pH 6 sample was measured at 25 C, there was a larger paramagnetic relaxivity at low frequency (data not shown). Another possibility is that in this isozyme the cobalt is in a pentacoordinate or hexacoordinate environment (Engberg and Lindskog, 1984), which would have less ability to relax due to a shorter electron relaxation time (Koenig etal., 1983). Whatever the reason, the paramagnetic contribution at low frequency is small. The increase in Rp with higher frequency (Figure 2) is unlike that for any other Co(II)-substituted carbonic anhydrase which show a decrease in Rp with higher frequency (Wells etal., 1979), but is not unlike that observed for Mn, Cu, and Ni proteins (Koenig and Brown, 1973; Bertini and Luchinat, 1986). While the relaxation behavior of these systems is not completely understood, it is clear that the field dependence of the electron spin relaxation is a factor. Thus, the increase in the electron relaxation time may be responsible for the increase in Rp at intermediate frequency. The decrease at higher frequency may be due to the (OsXc dispersion predicted by the Solomon-Bloembergen equations (Bertini and Luchinat, 1986).

PAGE 79

CHAPTER 7 CONCLUSIONS This work shows the utility of the NMR relaxation method to measure the hydrogen / deuterium fractionation properties of the aqueous ligands of cobalt in some specific cases. In particular, the characterization of the relaxation mechanisms at 300 MHz has allowed the detection of the fractionation of hydrogen and deuterium of two populations of ligand sites associated with cobalt, one having a significant contribution of the inner shell aqueous ligand of the metal and the other apparently dominated by outer shell water. The fractionation factor for the inner shell ligand can differ significantly from that for solvent water, and in the case of Co(H20)62+, (})t2 is near the fractionation factor of H3O+, implying that the water ligands of cobalt in Co(H20)62+ are influenced by the positive charge of the cobalt. The fractionation factors of Co(ll)substituted carbonic anhydrase represent the first direct measurement of this parameter at the active site of an enzyme. The conclusions of the contributions of Inner and outer shell water to the relaxation times Ti and T2 result from the effect of the chemical shift relaxation mechanism at 300 MHz. This conclusion is supported by calculations and experimental evidence with EDTA for Co(H20)62+. In this hexaaquacomplex the only interactions are those of the metal and water, and the water can be in the inner shell metal-bound site, an outer shell site, or the bulk solvent site. There are no other effects of the medium in this system. In the case of Co(ll)substituted carbonic anhydrase, there are many other interactions which may occur to affect the fractionation of the hydrogenic positions of the aqueous 73

PAGE 80

ligand of the metal. These interactions include influences of charge by residues near the active site, and the effects of the hydrogen bond network that exists in the active site cavity. This hydrogen bond network includes water in the active site cavity and specific residues from the enzyme, as well as the aqueous ligand of the metal. Thus it is clear that outer shell water in Co(ll)-substituted carbonic anhydrase would have some different properties than outer shell water in Co(H20)62+. The relaxation rates at 300 MHz of water protons in solution of Co(ll)-substituted carbonic anhydrase showed Ti much larger than T2, as in Co(H20)62+, and this indicates that the chemical shift mechanism affects T2 in solutions of Co(ll)-substituted carbonic anhydrase. However there is no other support for this from calculations or other experiments, as in the case of Co(H20)62+, except that the fractionation factor determined from Ti and T2 are also different for Co(ll)-substituted carbonic anhydrase. This inference of the chemical shift mechanism affecting T2 in solutions of Co(ll)-substituted carbonic anhydrase is the only extrapolation made from Co(H20)62+ to Co(ll)-substituted carbonic anhydrase, which are very different systems in many other respects. It Is important to note that the fractionation factors for Co(ll)-substituted carbonic anhydrase are measured directly, and thus are independent of any other interpretation regarding Co(H20)62+, except that of the contribution of inner shell water to T2 and of outer shell water to Ti. Thus the fractionation factors for Co(ll)-substituted carbonic anhydrase include the effects of the interactions near the active site. One very unique opportunity to understand the basis for the differences in the fractionation factors between the isozymes of carbonic anhydrase exists in the study of site-directed mutants of the enzyme. It is known that the pK for the ionization of the metal-bound water in carbonic anhydrase III is lower than that ionization in carbonic anhydrase I and II. This difference is most likely due

PAGE 81

75 to the positively cliarged residues near the active site in carbonic anhydrase III. The fractionation factor for the metal-bound hydroxide in carbonic anhydrase III was greater than that fractionation factor in carbonic anhydrase I and II, and this may also be due to the positively charged residues in carbonic anhydrase III. This could be tested, for instance, by replacing the histidine in position-64 of carbonic anhydrase I and II with a lysine, which occupies position-64 in carbonic anhydrase III, and then measuring the fractionation factor using the cobalt-substituted enzyme. One would predict that the fractionation factor of Co{ll)-substituted carbonic anhydrase I and II with lysine in position-64 would be closer to unity, like that in Co(ll)-substituted carbonic anhydrase III. The value of the fractionation factor for Co(H20)62+ measured directly from NMR represents a reactant state fractionation factor that can be used to interpret solvent hydrogen isotope effects associated with Co(H20)62+. The solvent hydrogen isotope effects on the formation of complexes of cobalt with some bidentate ligands suggest that the fractionation factor of some other hydrogenic positions must contribute to the overall isotope effect, most likely the positions of outer shell water in these complexes, because the isotope effect could not be accounted for by considering only the fractionation factor of Co(H20)62+. This may be because of the effect of the bidentate ligands on the structure of outer shell water in these complexes. It could be that the fractionation factor of the outer shell water in a cobalt-bidentate ligand complex is different than that fractionation factor in Co(H20)62+. Also, it may be that the number of outer shell waters that experience the paramagnetic relaxation enhancement by the cobalt is less in the cobalt-bidentate ligand complex than in Co(H20)62+. However, it is clear that there would be a large number of outer shell waters associated with complexes, and that a small effect on the fractionation of the outer shell water in these complexes on going from the

PAGE 82

76 reactant to the product state, when raised to a large power for the number of hydrogenic positions in the outer shell water, could have a large contribution to the overall isotope effect. The values of the fractionation factor of the aqueous ligand of the metal at the active site of Co(ll)-substituted carbonic anhydrase can also be used to interpret solvent hydrogen isotope effects on reactions associated with carbonic anhydrase. The solvent hydrogen isotope effect on the equilibrium binding of iodide ion to Co(ll)-substituted carbonic anhydrase suggests that there must be other hydrogenic positions whose isotope preference changes on going from the reactant to product states, because the consideration of the fractionation factor for the aqueous ligand of cobalt in Co(ll)-substituted carbonic anhydrase by itself does not allow a complete interpretation of the isotope effect. However, it is clear from the results that the hydrogenic positions of the metal-bound water at the active site do make a large contribution to the overall isotope effect. This means that the other hydrogenic positions that contribute to the isotope effect probably have a fractionation factor just slightly different from unity, but the large number of these positions would then give a significant contribution to the overall isotope effect. The solvent hydrogen isotope effects for other anions binding to Co(ll)-substituted carbonic anhydrase could be determined, and even for other inhibitors of the enzyme that are known to bind to the active site. This may give more information about these other hydrogenic positions. For instance, it may be that the fractionation factor of the hydration shell of the anion is important, and this could be determined from the results with other monovalent anions. Also, bulkier groups could be used, such as the sulfonamide inhibitors, and these may displace more outer shell water within the active site cavity. The isotope effect on the binding of a sulfonamide

PAGE 83

77 inhibitor may then give some insight into the contribution of these outer shell hydrogenic positions to the overall isotope effect. To account for the solvent hydrogen isotope effects on a rate constant is even more complex because knowledge of the fractionation factor values for the transition state are required. The model for the transition state of the ratelimiting proton transfer step of the CO2 hydration reaction, developed from the fractionation factor of the aqueous ligand of the cobalt in Co(ll)-substituted carbonic anhydrase II and the isotope effect on kcat. is consistent with other experimental evidence for the structure of the transition state. Other kinetic isotope effects on reactions of carbonic anhydrase may then also be used with the fractionation factor to get information about the mechanisms of these reactions. For instance, the solvent hydrogen isotope effects on the binding of anions could be measured, and then interpreted with the fractionation factors of the reactant state. This may show whether the rate-limiting step of the binding is the initial entry of the anion into the active site cavity, or it may be that the final binding of the anion to the metal is the rate-limiting step. This work represents an important first step in the accurate interpretation of the solvent hydrogen isotope effects in the catalysis by carbonic anhydrase. Perhaps the most significant achievement in this work is the development of the NMR relaxation method to measure directly the hydrogen / deuterium fractionation factor of the hydrogenic positions of water coordinated to metals. This method could also be used with other transition metals and in other types of inorganic complexes, and in other metalloenzymes. Also, the conclusions of the contributions of inner and outer shell water to the relaxation times Ti and T2 could also be studied further by, for instance, determining the effects of ionic strength and temperature and the relaxation and the fractionation factors

PAGE 84

78 determined from Ti and T2. It is hoped that the worl< presented in this dissertation will be the catalyst for other studies in this field.

PAGE 85

References Baes, N. F.; Mesmer, R. E. The Hydrolysis of Cations; W\\ey : New York, 1976. Bernheim, R. A.; Brown, T. H.; Gutowsky, H. S.; Woessner, D. E. J. Cfiem Phys. 1959, 30, 950-956. Bertini, I.; Canti, G.; Luchinat, C; Scozzafava, A. J. Am Ciiem Soc. 1978, 100, 4873-4877. Bertini, I.; Dei, A.; Luchinat, C; Monnanni, R. Inorg. Chem. 1985, 24, 301-303. Bertini, I.; Luchinat, C. Acc. Chem. Res. 1983, 16, 272-279. Bertini, I.; Luchinat, C. NMR of Paramagnetic Species in Biological Systems; Benjamin-Cummings: Menio Park, CA, 1986. Bertini, I.; Luchinat, C; Scozzafava, A.; Inorg. Chem. Acta ^980, 46, 85-89. Bertini, I.; Luchinat, C; Scozzafava, A. Struct. Bonding (Berlin) 1982, 48, 45-92. Bloembergen, N. J. Chem Phys. 1957, 27, 572-573. Bloembergen, N.; Morgan, L. O. J. Chem. Phys. 1961, 34, 842. Brown, G. S.; Navon. G.; Shulman, R. G. Proc. Natl. Acad. Sci. USA 1977, 74, 1794-1797. Carlson, G. A.; McReynolds, J. P.; Verhoek, F. H. J. Am. Chem. Soc. 1945, 67, 1334-1339. Chiang, Y.; Kresge, A. J.; More O'Ferrall, R. A. J. C. S. Perkin //1980, 18321839. Dwek, R. A. NMR in Biochemistry; Oxford press: Oxford, 1975. Eigen, M.; Tamm, K. Z Elektrochem. 1962, 66, 93-121. Eisenstadt, M. J. Chem Phys. 1969, 51, 4421-4432. Engberg, P.; Lindskog, S. FEBS Lett. 1984, 170, 326. 79

PAGE 86

80 Engberg, P.; Millqvist, E.; Pohl, G.; Lindskog, S. Arch. Biochem. Biophys. 1985, 241, 628-638. Eriksson, E. A.; Jones, T. A.; Liljas, A. In Zinc Enzymes; Bertini, I.; Luchinat, C; Maret, W.; Zeppezauer, M. Eds.; Birkhauser: Boston, 1986; 317-328. Fabry, M. E.; Koenig, S. H.; Schillinger, W. E. J. Biol. Chem. 1970, 245, 42564262. Freeman, R.; Kempsell, S. P.; Levitt, J. J. Magn. Reson. 1980, 138, 453-479. Friedman, H. L; Holz, M.; Hertz, H. G. J. Cfiem Phys. 1979, 70, 3369. Glasoe, P. K.; Long, P. A. J. Phys. Chem. 1960, 64, 188-191. Gold, v.; Lowe, B. M. J. Chem. Soc. (A) 1968, 1923-1932. Gold, v.; Wood, D. L. J. Chem. Soc, Dalton Trans. 1982, 2452-2461. Goodall, D. M.; Long, F. A. J. Am. Chem. Soc. 1968, 90, 238. Gueron, M. J. Magn. Reson. 1975, 19, 58. Hammes, G. C; Steinfeld, J. I. J. Am. Chem. Soc. 1962, 84, 4639-4643. Hunt, J. B.; Rhee, M. J.; Storm, C. B. Anal. Biochem. 1977, 79, 614-617. Jencks, W. P.; Salvesen, K. J. Am Chem Soc. 1971, 93, 4433. Kalifah, R. Biochemistry ^977, 16, 2236-2240. Kalifah, R, G.; Strader, D. J.; Bryant, S. H.; Gibson, S. M. Biochemistry ^977 16, 2241-2247. Kararii, T.; Silverman, D. N. J. Biol. Chem. 1985, 260, 3484-3489. Koenig, S. H.; Brown, R. D. Ann. N. Y. Acad. Sci. 1973, 222, 752-763. Koenig, S. H.; Brown, R. D.; Bertini, I.; Luchinat, C. Biophys. J. 1983, 41, 179187. Kresge, A. J.; More O'Ferrall, R. A.; Powell, M. F. In Isotopes in Organic Chemistry, Volume 7; Buncel, E.; Lee, C. Eds.; Elsevier: New York, 1987; 177273. Laughton, P. M.; Robertson, R. E. In Solute-Solvent Interactions; Coetzee, J. F.; Ritchie, C. D. Eds.; Marcel Dekker: New York, 1969; 399-538. Lindskog, S. Biochemistry ^966, 5, 2641-2646.

PAGE 87

81 Lindskog, S. in Zinc Enzymes; Spiro, T. G. Ed.; Wiley: New York; 1983; 78-121. Lowe, B. M.; Smith, D. G. J. Chem. Soc, Faraday Trans. /1975, 71, 389-397. Luz, Z.; Meiboom, S. J. Chem. Phys. 1964, 40, 2686-2692. Meiboom, S.; Gill, D. Rev. Sci Instrum. 1958, 29, 688. Melamud, E.; Mildvan, A. S. J. Biol. Chem. 1975, 250, 8193-8201. Melton, B. F.; Pollack, V. L. J. Phys. Chem. 1969, 73, 3669-3679. More O'Ferrall, R. A.; KoeppI, G. W.; Kresge, A. J. Am. Chem. Soc. 1971, 93, 920. Morgan, L. O.; Nolle, A. W. J. Chem Phys. 1959, 31, 365. Nyman, P. 0.; Lindskog, S. Biochim. Biophys. >4cfa1964, 85, 141-151. Pocker, Y.; Sarkanen, S. Adv. Enzymol. Relat. Areas Mol. Biol. 1978, 47, 149274. Schowen, K. B. J. In Transition States in Biochemical Processes; Grandour, R. D.; Schowen, R. L. Eds.; Plenum: New York, 1978; 225-284. Schowen, K. B.; Schowen, R. L. In Methods in Enzymology, Volume 87; Purich, D. L. Ed.; Academic: New York, 1982; 551-606. Silverman, D. N. J. Am. Chem. Soc. 1981, 103, 6242-6243. Silverman, D. N.; Lindskog, S. Acc. Chem. Res. 1988, 21, 30-36. Silverman, D. N.; Vincent, S. H. CRC Cht. Rev. Biochem. 1983, 14, 207-255. Simonsson, I.; Lindskog, S. Eur. J. Biochem. 1982, 123, 29-36. Smith, R. M.; Martell, A. E. Critical Stability Constants; Plenum: New York, 1975. Solomon, I. Phys. Rev. 1955, 99, 559-565. Steiner, H.; Jonsson, B.; Lindskog, S. Eur. J. Biochem. 1975, 59, 253-259. Swift, T. J.; Connick, R. E. J. Chem. Phys. 1962, 37, 307-320; J. Chem. Phys 1964, 41, 2553-2554. Szawelski, R. J.; Wharton, C. W.; White, S. Biochem. Soc. Trans. 1982, 10 232233.

PAGE 88

82 Tibell, L.; Forsman, C; Simonsson, I.; Lindskog, S. Biochim. Biophys. Acta 1984, 789, 302-310. Tu, C. K.; Sanyal, G.; Wynns, G. C; Silverman, D. N. J. Biol. Chem. 1983, 258, 8867-8871. Tu, C. K.; Silverman, D. N. Biochemistry -{985, 24, 5881-5887. Tu, C. K.; Thomas, H. G.; Wynns, G. C; Silverman, D. N. J. Biol. Chem. 1986, 261, 10100. Venkatasubban, K. S.; Silverman, D. N. Biochemistry 1980, 19, 4984. Voice, P.J. J. C. S. Faraday 1197^, 70, 498-505. Wells, J. W.; Kandel, S. 1.; Koenig, S. H. Biochemistry 1979, 18, 1989-1995. Whitney, P. L; Brandt, H. J. Biol. Chem. 1976, 251, 3862-3867.

PAGE 89

83 BIOGRAPHICAL SKETCH James W. Kassebaum, the son of Amanda L. Schoonmaker and Wilbur F. Kassebaum, was bom March 6, 1962, in Mount Vernon, Illinois. James grew up in St. Louis, Missouri, and in 1980 graduated first in his class at Parkway South Senior High School. He then attended Purdue University, West Lafayette, Indiana, and in 1984 graduated with Honors with the degree of Bachelor of Science in chemistry. The summer of 1984 was a very eventful one, when James was wed to Shari L. Gunderman and then entered graduate school at the University of Florida in the Department of Biochemistry and Molecular Biology and studied under the supervision of Dr. David Silverman. James completed the requirements for the Ph.D. degree in 1988, and began his professional career at the Monsanto Company in St. Louis.

PAGE 90

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. David N. Silverman, Chair Professor of Pharmacology and Therapeutics 1 certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. E. Raymond Andrew Graduate Research Professor of Physics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Robert J/Cohen Associate Professor of Biochemistry and Molecular Biology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doc^fl>f Philosophy. Russell S. Drago Graduate Research Professor of Chemistry

PAGE 91

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosc Daniel L. Purich Professor of Biochemistry and Molecular Biology This dissertation was submitted to the Graduate Faculty of the College of Medicine and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. December, 1988


free solvent site. If Acom2 is small relative to the other terms, equation 2-3
reduces to equation 2-2. For these metal ions, 1/T¡0 was taken as the relaxation
of water containing only buffer, with no added metal ions. R¡, the relaxivity, is
defined as the paramagnetic relaxation per millimolar concentration of metal
ion.
Ri = (1/Tjp)/ ([Metal] x 1000) (2-4)
Solutions of Co(H20)62+ and Ni(H20)62+ for NMR measurements
contained ultra-pure C0CI2 or NSO4 (Aldrich Gold-Label) with 5 x 10-4 M acetic
acid buffer at pH = 4.4. This pH was necessary to preclude the formation of
multi-nuclear hydrolysis products (Baes and Mesmer, 1976). When EDTA was
present no acetic acid was used. Water used for solution preparations was
distilled and passed through two ion-exchange resin cartridges (Cole-Palmer
1506-35). D20 (99.8%) was stirred with activated charcoal, the charcoal filtered
out, and then the D20 was distilled.
Results
The paramagnetic contribution to the relaxivity R2 of water protons in
solutions of Co2+ and Ni2+ at 300 MHz was greater than that obtained at lower
frequencies by previous investigators (Table 2-1). For Co2+, R1 at 300 MHz and
20 MHz were close in value while R2 was much larger at 300 MHz than at 20
MHz. For Ni2+, R1 at 300 MHz was slightly larger than at 10 MHz which is
understood to be a result of a field dependent increase in the electron
relaxation time (Friedman et al., 1979), but R2 was much larger at 300 MHz than
at 10 MHz.
The addition of EDTA to the solutions of C0CI2 and N1SO4 reduced R1
and R2 at 300 MHz (Figures 2-1 and 2-2). When an equimolar amount of EDTA


reactant to the product state, when raised to a large power for the number of
hydrogenic positions in the outer shell water, could have a large contribution to
the overall isotope effect.
The values of the fractionation factor of the aqueous ligand of the metal at
the active site of Co(ll)-substituted carbonic anhydrase can also be used to
interpret solvent hydrogen isotope effects on reactions associated with carbonic
anhydrase. The solvent hydrogen isotope effect on the equilibrium binding of
iodide ion to Co(ll)-substituted carbonic anhydrase suggests that there must be
other hydrogenic positions whose isotope preference changes on going from
the reactant to product states, because the consideration of the fractionation
factor for the aqueous ligand of cobalt in Co(ll)-substituted carbonic anhydrase
by itself does not allow a complete interpretation of the isotope effect. However,
it is clear from the results that the hydrogenic positions of the metal-bound water
at the active site do make a large contribution to the overall isotope effect. This
means that the other hydrogenic positions that contribute to the isotope effect
probably have a fractionation factor just slightly different from unity, but the large
number of these positions would then give a significant contribution to the
overall isotope effect. The solvent hydrogen isotope effects for other anions
binding to Co(ll)-substituted carbonic anhydrase could be determined, and
even for other inhibitors of the enzyme that are known to bind to the active site.
This may give more information about these other hydrogenic positions. For
instance, it may be that the fractionation factor of the hydration shell of the anion
is important, and this could be determined from the results with other
monovalent anions. Also, bulkier groups could be used, such as the
sulfonamide inhibitors, and these may displace more outer shell water within
the active site cavity. The isotope effect on the binding of a sulfonamide


53
effects, and because a kinetic isotope effect includes the contribution of
transition state hydrogenic sites, this allows a prediction of the transition state
structure. In the case of Co(ll)-substituted carbonic anhydrase II the
fractionation factor measured from NMR leads to a predicted transition state
structure that includes a proton transfer to a water molecule in the active site
cavity. This is consistent with the large isotope effect on kcat (Steiner et al.,
1975) and the nonlinear dependence of kcat on the atom fraction of deuterium in
solvent water (Venkatasubban and Silverman, 1980).
Experimental Procedure
The equilibrium dissociation constants describing the binding of iodide
ion to Co(ll)-substituted carbonic anhydrase were determined from visible
absorption measurements (Lindskog, 1966). The pH dependence of the
absorbance at 640 nm in solutions of Co(ll)-substituted carbonic anhydrase I
and II was measured on a Beckman DU-7 spectrophotometer. Stock solutions
of enzyme at low pH contained 12.5 mM Hepes and 25 mM Mops (4-
morpholinepropanesulfonic acid) buffer, and at high pH 0.5 M of
triethylenediamine. The stock solutions had equal concentrations of enzyme.
The titrations were performed by adding small increments of the high pH
enzyme solution to the low pH enzyme solution. In this way the enzyme
concentration stayed the same though the volume increased slightly. These
titrations were done in H2O and D20 in the presence and absence of iodide ion
which has been shown to bind to the metal site in carbonic anhydrase II (Brown
et al., 1977).
In the absence of anion, the data were fit (RS1, BBN Software,
Cambridge, MA) to a scheme involving 4 micro equilibrium constants
(Simonsson and Lindskog, 1982; Bertini et al., 1985). In this scheme, only 3 of
the 4 microconstants are independent, that is KtK3 = K2K4 (Scheme 5-1). When


Table 4-2: Hydrogen/deuterium fractionation factors determined from Ti and T2
at 300 MHz for the aqueous ligand of the paramagnetic metal in three
mammalian isozymes of Co(ll)-substituted carbonic anhydrase.
T1
$T2
Co(ll)-carbonic anhydrase I
1.10.02
0.72.02
Co(ll)-carbonic anhydrase II
0.95.02
0.77.01
Co(ll)-carbonic anhydrase III
1.03.06
1.00.07
Note: Data were obtained from the least squares slope and intercept and
standard error in a plot such as Figure 4-1 according to equation 3-4.
Measurements were made at 23 C for solutions of enzyme buffered at pH 8.5
by 50 mM Hepes.


3 3
Table 3-4: Calculated fractionation factors for the metal-bound water in the
complexes of cobalt with glycine(gly), A/,A/-dimethylglycine (DMG), and
acetylacetone (acac).
0T1
<>T2
Co(H20)62+ a
0.95 0.01
0.73 0.02
<¡>Co(H20)4(gly)+
0Co(H2O)2(gly)2
0.91 0.02
0.79 0.04
0.61 0.03
0.36 0.03
Co(H20)4(DMG)+
<}>Co(H20)2(DMG)2
0.92 0.02
0.84 0.08
0.62 0.03
0.38 0.05
(t)Co(H20)4(acac)+
0Co(H2O)2(acac)2
0.96 0.04
1.04 0.12
0.65 0.04
0.47 0.07
Note: Calculated from equation 3-7 using the isotope effects on the formation
constants of the reaction of Co2+ with glycine, A/./V-dimethylglycine, and
acetylacetone (Table 3-3) and further description is in the text.
a Determined directly from NMR relaxation measurements.


ACKNOWLEDGEMENTS
I wish to thank Dr. David Silverman for his thoughtfulness and guidance
shown to me during the work presented in this dissertation. I have benefitted
greatly from his scientific insight into experimental design and his careful
attention to excellence in scientific writing. I also wish to thank very much Dr. C.
K. Tu and Mr. George Wynns for their help in experimental procedures and for
many helpful discussions. I acknowledge the very valuable advice of the
members of my committee, Dr. E. Raymond Andrew, Dr. Robert Cohen, Dr.
Russell Drago, and Dr. Daniel Purich. Finally, I especially thank Dr. Seymour
Koenig at the IBM T. J. Watson Research Center in Yorktown Heights, New
York, for allowing me to visit his laboratory and use the field-cycling relaxometer
which added greatly to the scope of this work.
ii


6 2
Including this in the denominator of the above equation for the isotope effect
increases the predicted value, which is then closer to the experimental value for
Co(ll)-isozyme II, but farther from the experimental value for Co(ll)-isozyme I.
This implies that I have incomplete knowledge of the fractionation properties of
each hydrogenic site that is affected when iodide ion binds to the enzyme.
However, to include the fractionation factors of any other groups in the equation
requires much speculation about the fractionation factors near the active site.
Thus it is likely that the isotope effect must be described by the complex
contribution of the fractionation factors of additional groups or water associated
with the enzyme, and that the fractionation factor of the aqueous ligand of the
metal is by itself not sufficient to account for the isotope effect on I* binding.
The discussion of the solvent hydrogen isotope effect on iodide binding
suggests that the fractionation factor can be used to interpret at least
qualitatively isotope effects associated with carbonic anhydrase. It is also
possible to apply the fractionation factors for the aqueous ligand of the metal in
Co(ll)-substituted carbonic anhydrase in the interpretation of isotope effects on
kinetic constants associated with the enzyme to see if similar information can be
obtained. A kinetic isotope effect is the product of the reactant state
fractionation factors divided by the product of the transition state fractionation
factors (equation 1-1). The isotope effect on kcat for the CO2 hydration activity of
Co(ll)-substituted carbonic anhydrase II is 1.8 (R. S. Rowlett, unpublished), and
the rate limiting step measured by kca, is the intramolecular transfer of a proton
from the metal-bound water to regenerate the metal-bound hydroxide for the
next round of catalysis (Steiner et at., 1975). This proton transfer from the
metal-bound water is thought to go through a water-bridge to histidine-64 in the
active site cleft, from which the proton can then be released to solvent


TABLE OF CONTENTS
page
ACKNOWLEDGEMENTS ii
ABSTRACT v
CHAPTERS
1 INTRODUCTION 1
2 WATER PROTON NMR RELAXATION ENHANCEMENT IN
SOLUTIONS OF Co(H20)62+ AND Ni(H20)62+
9
Introduction 9
Experimental Procedure 10
Results 11
Discussion 16
3 HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF
THE AQUEOUS LIGAND IN Co(H20)62+ AND Ni(H20)62+
20
Introduction 20
Experimental Procedure 22
Results 25
Discussion 32
4 HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF
THE AQUEOUS LIGAND OF COBALT IN Co(ll)-
SUBSTITUTED CARBONIC ANHYDRASE 39
Introduction 39
Experimental Procedure 40
Results 42
Discussion 48
5 INTERPRETATION OF SOLVENT HYDROGEN ISOTOPE
EFFECTS ASSOCIATED WITH CARBONIC ANHYDRASE 52
Introduction 52
in


Table 2-2: Calculated paramagnetic relaxivities of water protons in solutions
containing Co2+.
frequency
Ri
r2
Tip/T2P
(MHz)
(mM1 S'1)
(mM'1 s'1)
20
0.18
0.19
1.1
300
0.16
1.8
11.2
Note: Calculated using equations 2-2 to 2-4 in the text. The calculations used
Tm=7.4 x 10'7 s (Swift and Connick, 1962), the 20 MHz relaxation data from
Bernheim et at. (1959), and the 300 MHz relaxation data and chemical shift
determined in this work. A further description is in the text.


78
determined from T-i and T2. It is hoped that the work presented in this
dissertation will be the catalyst for other studies in this field.


I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosopfiy^
fin
DanTel L. Purich
Professor of Biochemistry and
Molecular Biology
This dissertation was submitted to the Graduate Faculty of the College of
Medicine and to the Graduate School and was accepted as partial fulfillment of
the requirements for the degree of Doctor of Philosophy.
December, 1988
Dean, Graduate School


Paramagnetic relaxivity (mM'1 s'1)
69
Figure 6-2: The paramagnetic relaxivity of Co(ll)-substituted carbonic
anhydrase III as a function of the proton resonance frequency. Experimental
conditions as in Figure 6-1. pH 6.0 ( A), pH 9.0 ().


55
anion was present, Scheme 5-1 was expanded to include the binding of the
inhibitor, and it was assumed that the anion binds only to species 1 and 2 (see
Tibell etal. (1984) for an estimate of the accuracy of this assumption). Thus two
separate binding constants were determined. K-i1' is the equilibrium
dissociation constant of the anion complex with species 1 of Scheme 5-1, and
K2i_ is the equilibrium dissociation constant of the anion complex with species 2.
Results
In the absence of anion, the visible pH titration data fit Scheme 5-1 very
well (Table 5-1). The constants K-i, K2, and K3 were determined from a fit to the
data, and then K4 was calculated from K-i, K2, and K3. The 4 microconstants
determined in H20 for Co(ll)-substituted carbonic anhydrase II agreed very well
with those obtained by Simonsson and Lindskog (1982), and the values
determined in H20 for Co(ll)-substituted carbonic anhydrase I also agreed fairly
well with those obtained by Bertini et al. (1985). In the presence of iodide the
pH dependence of the absorbance at 640 nm was shifted to higher pH. Iodide
binds with greater affinity to the diprotonated species 1 in Scheme 5-1 than to
the monoprotonated species 2 (Table 5-2). This is similar to the iodide
inhibition results of Simonsson and Lindskog (1982) for native zinc carbonic
anhydrase II determined from activity measurements, and the results of Bertini
et al. (1985) for nitrate inhibition of Co(ll)-substituted carbonic anhydrase II
determined from spectrophotometric titrations. The change in solvent from H20
to D20 has no measurable effect on K^'and K2I_ for Co(ll)-substituted carbonic
anhydrase I, but for Co(ll)-substituted carbonic anhydrase II the change in
solvent from H20 to D20 results in a decrease in these equilibrium constants
(Table 5-2).


57
Table 5-2: Solvent hydrogen isotope effects on the equilibrium dissociation
constants of iodide ion with Co(ll)-substituted carbonic anhydrase I and II.
Co(ll)-carbonic anhydrase I
value in H2O ApK1'3
Co(ll)-carbonic anhydrase II
value in H2O ApK1'3
3.48 0.13 0.47 0.19
pK-i1' 2.95 0.10 -0.09 0.17
pK2z' 2.35 0.06 -0.07 0.11
2.79 0.02 0.27 0.02
Note: The dissociation constants were determined from the pH dependence of
the absorbance at 640 nm in the presence of 10 mM T with conditions as in
Table 5-1. K11' is the dissociation constant of T binding to species 1 of Scheme
5-1, and K21' is the dissociation constant of I- binding to species 2 of Scheme 5-
1.
a
ApK = (pKj^OgO (pKj^HgO


35
The usefulness of fractionation factors is judged by their ability to aid in
interpretation of solvent hydrogen isotope effects. I have determined the solvent
hydrogen isotope effects on the formation of some simple inorganic complexes
of Co2+ with the goal to use the knowledge of the fractionation of Co(H20)62+ to
interpret those isotope effects. An isotope effect is a function of all the
fractionation factors for each hydrogenic site in a reaction (equation 1-2), and
for a reaction such as equation 3-2 the fractionation factor of Co(H20)62+
represents the individual contribution of those hydrogenic sites to the overall
isotope effect on the reaction.
In considering the solvent hydrogen isotope effect on the formation
constants (Table 3-3), initially I will assume that the isotope effect is made up
entirely of the contribution of the isotope preferences of the hydrogenic
positions of water in the inner coordination shell of Co(H20)62+, Co(H20)4L+,
and Co(H20)2L2. With this assumption there is enough information from the
isotope effect and the fractionation factor of Co(H20)62+ to calculate (from
equation 3-7) the fractionation factor of the remaining waters in the cobalt-
bidentate ligand complexes (Table 3-4). While I have done the calculations
with equation 3-7 using both <()t1 and (J)j2 for Co(H20)62+ (Table 3-2), the values
from (J)^ may not be meaningful since Ti is measuring more outer shell water,
and the number of outer shell waters is not known. One interpretation of these
calculations is that since t2 of Co(H20)62+ measures more of the fractionation
of inner shell water in Co(H20)62+, the calculation using ^ of Co(H20)62+
represents the resulting fractionation factor of the inner shell waters in the cobalt
complexes. As ligands are added to Co2+ the calculated fractionation factor of
the remaining coordinated waters decreases, and this decrease is the same for
each of the three ligands, though glycine and A/,A/-dimethylglycine coordinate
through an amino and carboxylate, and acetylacetone binds through two


HYDROGEN / DEUTERIUM FRACTIONATION FACTORS OF
THE AQUEOUS LIGAND OF COBALT IN Co(H20)62+ AND
Co(ll)-SUBSTITUTED CARBONIC ANHYDRASE
By
JAMES WEB KASSEBAUM
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1988


15
was added to solutions of Co2+, Ri was reduced to 47% of the value of R-i in the
absence of EDTA, R2 was reduced to 5% of the value of R2 in the absence of
EDTA, and Tip/T2P was 1.2. The remaining paramagnetic relaxivity due to the
CoEDTA complex was R-| = 0.075 mM'1 S'1 and R2 = 0.091 mM'1 s*1. When an
equimolar amount of EDTA was added to solutions of Ni2+, R-i was reduced to
58% of the value of Ri in the absence of EDTA, R2 was reduced to 34% of the
value of R2 in the absence of EDTA, and Tip/T2p was 1.4. The remaining
paramagnetic relaxivity due to the NiEDTA complex was Ri = 0.55 mM-1 s-1 and
R2 = 0.78 mM-1 sr\
The observed change in the chemical shift caused by the paramagnetic
metal, Aco, at 300 MHz for a 10 mM solution of C0CI2 was measured to be 158
Hz. The measured change in the chemical shift, Aco, is the difference between
the chemical shift of water protons in water containing only buffer and the
chemical shift of water protons in a solution of a paramagnetic metal. This was
measured with tetramethylsilane as an external reference using coaxial tubes.
The influence of this chemical shift on the transverse relaxation of Co(H20)62+
solutions at 20 MHz and 300 MHz was estimated using equation 2-3 (Table 2-
2). In the calculations, I used xm = 7.4 x 107 s determined by Swift and Connick
(1962) using 170 exchange. Also, I used T1m = T2m as predicted by the
Solomon-Bloembergen equations for Co(H20)62+ at 20 MHz and 300 MHz.
From the observed T1 at 300 MHz, T1m was calculated (from equations 2-1 and
2-2) to be 6.62 x 10'4 s. From the data of Bernheim et al. (1959) T-im = 5.95 x 10-
4 s at 20 MHz. Since Timxm, the fast exchange limit applies at 20 MHz and at
300 MHz. Acom was calculated from Aco at 300 MHz using Aco = p' Acom
which is valid in the fast exchange limit (Swift and Connick, 1962). I determined
Acom to be 1.46 x 105 Hz at 300 MHz. Since Acom is directly proportional to the
precessional frequency, Acom at 20 MHz is 9.73 x 103 Hz. The results using


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EY3ZDI4SJ_8NMRIN INGEST_TIME 2014-12-03T19:48:43Z PACKAGE AA00026457_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


CHAPTER 4
HYDROGEN / DEUTERIUM FRACTIONATION FACTOR OF THE
AQUEOUS LIGAND OF COBALT IN Co(ll)-SUBSTITUTED
CARBONIC ANHYDRASE
Introduction
Carbonic anhydrase is a zinc metalloenzyme which catalyzes the
reversible hydration of carbon dioxide to produce bicarbonate and a proton.
CO2 + H2O ~- HC£>+ H+
The aqueous ligand of the metal at the active site of carbonic anhydrase is
believed to have a role in the catalytic pathway. Direct nucleophilic attack by
the metal-bound hydroxide on CO2 is a likely step in the production of
bicarbonate (Pocker and Sarkanen, 1978; Lindskog, 1983; Silverman and
Lindskog, 1988), and an intramolecular proton transfer involving the metal-
bound water is proposed to be a rate-limiting step in catalysis by isozyme II
(Steiner etai, 1975). The solvent hydrogen isotope effect associated with the
turnover number for CO2 hydration catalyzed by native zinc-carbonic anhydrase
II is 3.8 (Steiner et a/., 1975). To interpret fully the solvent hydrogen isotope
effects observed for catalysis by carbonic anhydrase requires knowledge of the
fractionation factor of the aqueous ligand of the metal.
The fractionation factor measures the tendency of deuterium to
accumulate at the aqueous ligand of the metal relative to the deuterium content
of bulk solvent. There are few reports of the fractionation factor of the aqueous
ligands of a metal at the active site of an enzyme. Using NMR methods,
Silverman (1981) has measured such a factor for the water ligands of cobalt in
cobalt(ll)-substituted carbonic anhydrase II ((}> = 1.05 0.17) by taking
3 9


54
+HCoOH'
+HECoOH2 ECoOH"
ECoOH2
2
Scheme 5-1: A diagram of two significant ionizations at and near the active site
of Co(ll)-substituted carbonic anhydrase II. One is the ionization of metal-bound
water and the second is suggested to be the ionization of His-64 (Simonsson
and Lindskog, 1982), the imidazole ring of which is about 6 from the metal.
For Co(ll)-substituted carbonic anhydrase I the second ionization could be
either that of His-64 or His-200 (Whitney and Brandt, 1976).


CHAPTER 6
MAGNETIC FIELD AND pH DEPENDENCE OF THE WATER
PROTON NMR RELAXATION ENHANCEMENT IN SOLUTIONS
OF Co(ll)-SUBSTITUTED CARBONIC ANHYDRASE III
Introduction
Carbonic anhydrase III is a mammalian isozyme of carbonic anhydrase
found primarily in skeletal muscle. It is a less efficient catalyst of CO2 hydration
compared with isozymes I and II, which are found in red blood cells (Lindskog,
1983). Another distinctive feature of carbonic anhydrase III is the pH
independence of its catalytic activity in the range pH 6-9, in which carbonic
anhydrase II has an inflection at pH 7 and maximal activity at high pH (Tu etal.,
1983; Kararli and Silverman, 1985; Tu etal., 1986). It is also apparent that the
zinc is bound more tightly in carbonic anhydrase III than in isozymes I and II
(Engberg and Lindskog, 1984).
The replacement of cobalt for zinc in carbonic anhydrase I and II has
proved useful to study the properties of the metal at the active site (Bertini and
Luchinat, 1983). The catalytic and visible spectral properties of Co(ll)-
substituted carbonic anhydrase III have been investigated (Engberg and
Lindskog, 1984). I report here the NMR relaxation enhancement of water
protons in solutions of Co(ll)-substituted carbonic anhydrase III. I found, in
contrast to Co(ll)-substituted carbonic anhydrase I and II, no pH dependence of
the NMR relaxation in the range of pH 6-9. Also, the frequency dependence of
the paramagnetic contribution to the relaxation shows an increase with higher
frequency, as opposed to the dispersion to lower relaxation rates observed in
solutions of Co(ll)-substituted carbonic anhydrase I and II.
65


23
Determination of the isotope effects on formation constants: The
formation constants for the reactions of Co2+ with a bidentate ligand, L-1, as in
equation 3-2 are defined as
[Co(H20)4L+]
ki = (3-5)
[Co(H20)62+] [L-1]
with similar expressions for k2 and (<3. Potentiometric titrations were performed
in H20 and D20 (99.8%) containing 1 mM CoCI2 and 4 mM glycine, N,N-
dimethylglycine, or acetylacetone. The values of pH and pD used in these
experiments were corrected from pH meter readings. The correction of a pH
meter reading in 100% D20 is pD = meter reading + 0.4 (Glasoe and Long,
1960). The formation constants were determined by a non-linear least-squares
fit of the data (RS1, BBN Software, Cambridge, MA) to the equation derived by
Carlson et al. (1945) describing , the ratio of the concentration of metal-bound
ligand to total concentration of metal, as a function of the concentration of
coordinating ligand, [A].
ki[A] + 2k-i k2[A]2 + 3kik2k3[A]3
= (3-6)
1 + k-|[A] + k^A]2 + k1k2k3[A]3
The calculation of and [A] required knowledge of the acid ionization
constants of the free ligands, which were measured by potentiometric titrations
in H20 and D20 (Table 3-1). The negative logarithm of the acid ionization
constants, pK, in H20 were all within 0.07 pK units from the literature values
(Smith and Martell, 1975). The solvent hydrogen isotope effects for glycine
expressed as ApK where ApK = pKD2o pKh2o were within 0.03 ApK units from
the values obtained previously by Jencks and Salvesen (1971).


7
When this reaction is performed in D20 some very interesting isotope
effects are observed. The isotope effect on the steady-state parameter kcat for
both native and Co(ll)-substituted carbonic anhydrase II is large, 3.8 in the case
of native zinc-carbonic anhydrase II (Steiner et al. 1975), and 1.8 for Co(ll)-
substituted carbonic anhydrase II (R. S. Rowlett, unpublished), which means
that a primary proton transfer is at least part of the rate limiting step measured
by kcat- The ratio kcat/Km contains steps up to and including the first irreversible
step in the reaction, and for the catalyzed reaction the first irreversible step is
the release of HCO3' from the enzyme. The isotope effect is unity on kcat/Km for
both native zinc-carbonic anhydrase II (Steiner et al. 1975) and for Co(ll)-
substituted carbonic anhydrase II (R. S. Rowlett, unpublished), which implies
that the rate determining step measured by kcat is separate and distinct from the
interconversion of C02 and HCO3'. Thus the rate-limiting step measured by kcat
is the rate of transfer of a proton from the metal-bound water out of the active
site to regenerate the metal-bound hydroxide (step 2 of Scheme 1-1). From this
description it is clear that the aqueous ligand of the metal at the active site of
carbonic anhydrase is intimately involved in the mechanism of catalysis, and to
interpret the isotope effects knowledge of the fractionation factor of the aqueous
ligand of the metal is required.
The fractionation factor of the aqueous ligand of a paramagnetic metal
can be determined by taking advantage of the paramagnetic contribution to the
NMR relaxation rates of water protons in solutions of the metal in which the
protons are in fast exchange between the coordination shell of the metal and
the bulk solvent. The NMR measurements in this work were performed at a
proton resonance frequency of 300 MHz, which is a higher frequency than has
been used to study the water proton relaxation enhancement of first row
transition metals (Dwek, 1975; Bertini and Luchinat, 1986). Thus the relaxation


4.1
Atom fraction of deuterium in solvent (n)
Figure 4-2: The dependence of p'Tip (i=1,2) at 300 MHz on n, the atom fraction
of deuterium in solvent water, for 2.0 mM solutions of Co(ll)-substituted carbonic
anhydrase II. Solutions contained 50 mM Hepes buffer at pH 8.5 and 23C.
Data are the means and standard errors of three measurements.


active site group as determined in many experiments (see Table 5-1 and
Engberg and Lindskog, 1984), especially spectrophotometric titrations and
activity measurements, and the ligand of cobalt is believed to be a hydroxide
ion at this pH. This presents some uncertainty as to which protons are actually
being relaxed at pH 8.5 since a tetracoordinate cobalt-bound hydroxide ion
would not be expected to exchange rapidly with water in solution. This issue
has been discussed by Koenig et al. (1983) who proposed that ligand
exchange with solvent involved a pentacoordinate intermediate having both
OH' and H20 as ligands. Proton exchange can be rapid between these two
ligands so that the departing H20 can contain the oxygen of the initially-bound
OH*. On the basis of 180-exchange kinetics, Tu and Silverman (1985)
suggested that the NMR relaxation observed at pH 8.5 is that of a water
molecule hydrogen bonded to the cobalt-bound hydroxide yet close enough to
the metal to be strongly relaxed. Yet another possibility is the rapid exchange of
the hydrogen of the metal-bound hydroxide without exchange of the oxygen.
Because of these unresolved issues, I cannot state specifically which of the
hydrogens near the metal are the main contributors to the observed
fractionation factors.
The fractionation factors measured from Ti for each isozyme of Co(ll)-
substituted carbonic anhydrase were close to unity (Table 4-2), like (})t1 for
Co(H20)62+ (Table 3-2), consistent with these values having a large contribution
from the fractionation of outer shell water. This is reasonable from distance
considerations, as it can be calculated using a 1/r6 dependence that the
paramagnetic contribution to the longitudinal relaxation rate for a cobalt-bound
hydroxide with a cobalt-proton distance of 2.8 is about the same as the
paramagnetic contribution of the protons of three water molecules at an
average distance of 3.8 . Also, it is known from crystallographic data that there


66
Experimental Procedure
The purification of bovine carbonic anhydrase III and preparation of
Co(ll)-substituted carbonic anhydrase III was performed as described in
Chapter 4. The total concentration of carbonic anhydrase III was estimated at
280 nm by using e = 6.4 x 104 M'1 crrr1 (Engberg and Lindskog, 1984). The
concentration of Co(ll)-substituted carbonic anhydrase III was estimated at 640
nm by using the extinction coefficients determined by Engberg and Lindskog
(1984). Zinc analysis of the native enzyme yielded 1.03 0.05 zinc per
molecule of enzyme. Cobalt analysis of the Co(ll)-substituted enzyme yielded
0.33 0.04 cobalt per molecule of enzyme, with the remainder apoenzyme.
The fraction of the sample that was apoenzyme was restored to full activity by
the addition of zinc.
The longitudinal relaxation rates of the protons of solvent water in
solutions of Co(ll)-substituted and native Zn-carbonic anhydrase III were
measured using afield cycling "relaxometer" (Koenig et ai, 1983) at the IBM T.
J. Watson Research Center in Yorktown Heights, New York. The rates were
measured at 5C at pH 6.0 (50 mM Mes/NaOH buffer), pH 7.0 (50 mM
Mops/NaOH buffer), pH 8.0 (50 mM Hepes/NaOH buffer), and pH 9.0 (50 mM
Ches/NaOH buffer) in solutions containing 1 mM of enzyme. The pH 6 samples
were also measured at 25C. The samples were in equilibrium with
atmospheric CO2. The total relaxivity, Rt, is defined as in equation 6-1
Rt = (1/Ti 1/Tio)/N (6-1)
where 1/Ti is the observed relaxation rate of the protein solution, 1 /T1o is the
relaxation rate of the buffer, and N is the millimolar concentration of protein. At
each pH the relaxivity was determined as a function of frequency over the range


Table 5-1: Solvent hydrogen isotope effects on the acid-base microequilibrium
constants observed for Co(ll)-substituted carbonic anhydrase I and II.
o
jD
1
o
0>
cr
o
13
o'
anhydrase I
Co(ll)-carbonic anhydrase II
value in H2O
ApKa
value in H2O
ApKa
pKi
7.55 0.05
0.54 0.06
5.65 0.04
0.42 0.05
pK2
7.58 + 0.15
0.45 0.19
5.73 0.07
0.37 0.10
pK3
8.11 0.15
0.45 0.19
6.95 0.08
0.38 0.12
pK4
8.09 0.22
0.54 0.28
6.87 0.12
0.43 0.17
Note: The microconstants were determined from the variation of the
absorbance at 640 nm with pH at 23C and fit to Scheme 5-1. The solutions
were buffered with Mops, Hepes and triethylenediamme with no other added
ions as described in the experimental procedure.
a ApK = (pK¡)d2o (pK¡)h2o


21
Using this method at a proton resonance frequency of 300 MHz, I have
measured the fractionation factor of the aqueous ligand of cobalt in Co(H20)62+
and of nickel in Ni(H20)62+. In the case of Co(H20)62+ the value of the
fractionation factor was dependent upon whether it was determined from T-i or
T2; however, the corresponding values of (j) for Ni(H20)62+ were identical when
obtained from T-| or T2. The analysis of the relaxation discussed in Chapter 2
can be used to interpret the fractionation factors, and allows a qualitative
differentiation between the hydrogen / deuterium fractionation of inner and outer
shell water in these complexes.
I have then used the knowledge of the fractionation of Co(H20)62+ to
study the solvent hydrogen isotope effects on the formation constants of the
reactions of glycine, A/,A/-dimethylglycine, and acetylacetone with Co2+. An
isotope effect on an equilibrium constant is the product of the reactant state
hydrogenic site fractionation factors divided by the product of the product state
hydrogenic site fractionation factors (equation 1-2). In a reaction of Co2+ with a
bidentate ligand L*1 represented as
Co(H20)62+ + L-1 Co(H20)4L+ + 2 H20 (3-2)
the solvent hydrogen isotope effect is a function of the fractionation of the
hydrogenic positions of Co(H20)62+ as well as Co(H20)4L+ and H20, and the
fractionation factor of Co(H20)62+ represents the individual contribution of the
hydrogenic positions of Co(H20)62+ to the overall isotope effect. Because the
solvent hydrogen isotope effects on the formation constants were measured to
be near unity in the case of each of the ligands, I determined that the
contributions of the fractionation of the hydrogenic positions of the inner shell
water in Co(H20)62+ and in the ligand complexes of cobalt can not account
completely for the overall isotope effect. I suggest that the fractionation of outer


68
Figure 6-1: The total relaxivity of Co(ll)-substituted carbonic anhydrase III and
native zinc carbonic anhydrase III as a function of the proton resonance
frequency. The solutions contained 1 mM enzyme at pH 6.0 (50 mM Mes /
NaOH), pH 8.0 (50 mM Hepes / NaOH), or pH 9.0 (50 mM Ches / NaOH). pH
6.0 Co(ll)-substituted carbonic anhydrase III at 5C ( a), pH 6.0 native zinc
carbonic anhydrase III at 5C ( ), pH 9.0 Co(ll)-substituted carbonic anhydrase
III at 5C (), pH 9.0 native zinc carbonic anhydrase III at 5C ( ), pH 8.0
Co(ll)-substituted carbonic anhydrase III at 23C (o ), pH 8.0 native zinc
carbonic anhydrase III at 23C ( ).


38
include the fractionation factor of outer shell water of Co(H20)62+ in the
numerator and of Co(H20)4L+ in the denominator. The number of outer shell
waters would be larger than the number of inner shell waters, and if the
contribution of an outer shell hydrogenic site in equation 3-7 was 1.06, twelve
outer shell waters (24 hydrogenic positions) would contribute (1.06)24 = 4.05.
This would then mean that the fractionation factor of the inner shell waters in
Co(H20)4l_+ is 0.74. If the contribution of an outer shell hydrogenic site is 1.10,
then the fractionation factor of the inner shell waters in Co(H20)4L+ is 0.83.
Thus a contribution of outer shell water slightly larger than unity would, when
raised to a large power for the number of hydrogenic positions, cancel the effect
of the inner shell waters.
I conclude that the solvent hydrogen isotope effect of unity on the
formation of complexes of Co2+ and glycine, A/,A/-dimethylglycine, acetylacetone
points to the importance of outer shell water in these reactions. Conversely,
one can not get information about the effect of added ligands on the
fractionation properties of the inner shell waters from the solvent hydrogen
isotope effect on the formation constants. This work shows that the solvent
hydrogen isotope effect is the complex contribution of the hydrogen / deuterium
fractionation of several different groups or species of water.


1 3
Figure 2-1: The relaxivity of the protons of water at 300 MHz due to Co2+ in
solutions of 10 mM CoCI2 as a function of the molar ratio of EDTA to Co2+ at 23
C. Solutions contained 50% D2O by volume. R1 (), R2 (a).


to the positively charged residues near the active site in carbonic anhydrase III.
The fractionation factor for the metal-bound hydroxide in carbonic anhydrase III
was greater than that fractionation factor in carbonic anhydrase I and II, and this
may also be due to the positively charged residues in carbonic anhydrase III.
This could be tested, for instance, by replacing the histidine in position-64 of
carbonic anhydrase I and II with a lysine, which occupies position-64 in
carbonic anhydrase III, and then measuring the fractionation factor using the
cobalt-substituted enzyme. One would predict that the fractionation factor of
Co(ll)-substituted carbonic anhydrase I and II with lysine in position-64 would
be closer to unity, like that in Co(ll)-substituted carbonic anhydrase III.
The value of the fractionation factor for Co(H20)62+ measured directly
from NMR represents a reactant state fractionation factor that can be used to
interpret solvent hydrogen isotope effects associated with Co(H20)62+. The
solvent hydrogen isotope effects on the formation of complexes of cobalt with
some bidentate ligands suggest that the fractionation factor of some other
hydrogenic positions must contribute to the overall isotope effect, most likely the
positions of outer shell water in these complexes, because the isotope effect
could not be accounted for by considering only the fractionation factor of
Co(H20)62+. This may be because of the effect of the bidentate ligands on the
structure of outer shell water in these complexes. It could be that the
fractionation factor of the outer shell water in a cobalt-bidentate ligand complex
is different than that fractionation factor in Co(H20)62+. Also, it may be that the
number of outer shell waters that experience the paramagnetic relaxation
enhancement by the cobalt is less in the cobalt-bidentate ligand complex than
in Co(H20)62+. However, it is clear that there would be a large number of outer
shell waters associated with complexes, and that a small effect on the
fractionation of the outer shell water in these complexes on going from the


12
Table 2-1: Paramagnetic contribution to the relaxivities of water protons
observed for aqueous solutions of C0CI2 and NSO4.
frequency
(MHz)
R1
(mM'1 s'1)
r2
(mM-1 s'1)
Tip/T2P
Co(H20)62+
(0
0
0.18
0.21
1.2
300 b
0.16
2.0
12.5
Ni(H20)62+
10c
0.64
0.71
1.1
300 b
0.95
2.3
2.4
a From Bernheim et al. (1959) for an unbuffered solution of 0.5 M C0CI2 at 25
C.
b Determined in this work at 23 C for 10 mM CoCI2 or 10 mM NiS04 which
contained 10% D2O by volume and 5 x 10'4 M acetic acid buffer. The pH was
4.4.
cFrom Morgan and Nolle (1959) for a solution of 10 mM Ni(N03)2at 27 C in 0.1
M perchloric acid.


2
deuterium fractionation factor, (¡), is the equilibrium constant for this isotope
exchange reaction
R(H) + (HOD) -> R(D) + (HOH)
[R(D)] / [R(H)]
=
[HOD] / [HOH]
in which a solute hydrogenic species can exchange with a solvent hydrogenic
species. The fractionation factor measures the tendency of D to accumulate in
the solute relative to the deuterium content of bulk solvent. A fractionation factor
greater than unity implies that D will accumulate in the solute position relative to
the solvent, and that the bond to hydrogen is stronger in the solute position than
that bond in the solvent. For a fractionation factor less than unity, the light
isotope of hydrogen will accumulate in the solute position, and the bond to
hydrogen is weaker in the solute relative to the solvent.
An isotope effect is a function of the fractionation factors of each
hydrogenic site in a reaction. For an isotope effect on a kinetic constant,
all reactant state sites
n all transition state sites
(Mi
n ^
and for an isotope effect on an equilibrium constant,


26
Atom fraction of deuterium in solvent (n)
Atom fraction of deuterium in solvent (n)
Figure 3-1: The dependence of p'T¡p (i = 1,2) at 300 MHz on n, the atom fraction
of deuterium in solvent water, for a 10 mM solution of C0CI2 at 23 C. The
solutions were buffered at pH 4.4 with 0.5 mM acetic acid. Data are the means
and standard errors propagated from NMR measurements.


CHAPTER 7
CONCLUSIONS
This work shows the utility of the NMR relaxation method to measure the
hydrogen / deuterium fractionation properties of the aqueous ligands of cobalt
in some specific cases. In particular, the characterization of the relaxation
mechanisms at 300 MHz has allowed the detection of the fractionation of
hydrogen and deuterium of two populations of ligand sites associated with
cobalt, one having a significant contribution of the inner shell aqueous ligand of
the metal and the other apparently dominated by outer shell water. The
fractionation factor for the inner shell ligand can differ significantly from that for
solvent water, and in the case of Co(H20)62+, 4)t2 is near the fractionation factor
of H30+, implying that the water ligands of cobalt in Co(H20)62+ are influenced
by the positive charge of the cobalt. The fractionation factors of Co(ll)-
substituted carbonic anhydrase represent the first direct measurement of this
parameter at the active site of an enzyme.
The conclusions of the contributions of inner and outer shell water to the
relaxation times Ti and T2 result from the effect of the chemical shift relaxation
mechanism at 300 MHz. This conclusion is supported by calculations and
experimental evidence with EDTA for Co(H20)62+. In this hexaaquacomplex the
only interactions are those of the metal and water, and the water can be in the
inner shell metal-bound site, an outer shell site, or the bulk solvent site. There
are no other effects of the medium in this system. In the case of Co(ll)-
substituted carbonic anhydrase, there are many other interactions which may
occur to affect the fractionation of the hydrogenic positions of the aqueous
73


63
(Venkatasubban and Silverman, 1980). For this individual step in the catalysis,
the reaction is
Co-OH2 + H20 + His Co-OH + H20 + His-H+
and the isotope effect is
kH20 ^Co-OH2) ^h2o)
u
D20 all transition state sites
n i
The hydrogenic sites of water have a fractionation factor of 1.0, and the
hydrogenic sites of the cobalt-bound water have a fractionation factor of 0.81
(Table 5-3). Using these values and the isotope effect in the above equation,
the product of the transition state fractionation factors can be calculated to be
0.36. In the transition state, the proton remaining on the metal-bound oxygen is
in transition between a metal bound water which has a fractionation factor of
0.81 (Table 5-3), and a metal-bound hydroxide which has a fractionation factor
of 0.77 (Table 4-2), so I will arbitrarily use a value of 0.79 as the fractionation
factor of this hydrogenic site in the transition state. Therefore, the product of the
fractionation factors of the other hydrogenic sites in the transition state is
0.36/0.79 = 0.46. These other hydrogenic sites in the transition state are the
positions associated with the water bridge between the metal-bound water and
the histidine in the active site cleft.


82
Tibell, L; Forsman, C.; Simonsson, I.; Lindskog, S. Biochim. Biophys. Acta
1984, 789, 302-310.
Tu, C. K.; Sanyal, G.; Wynns, G. C.; Silverman, D. N. J. Biol. Chem. 1983, 258,
8867-8871.
Tu, C. K.; Silverman, D. N. Biochemistry !985, 24, 5881-5887.
Tu, C. K.; Thomas, H. G.; Wynns, G. C.; Silverman, D. N. J. Biol. Chem. 1986,
261, 10100.
Venkatasubban, K. S.; Silverman, D. N. Biochemistry!980, 19, 4984.
Voice, P.J. J. C. S. Faraday /1974, 70, 498-505.
Wells, J. W.; Kandel, S. I.; Koenig, S. H. Biochemistry 1979, 18, 1989-1995.
Whitney, P. L; Brandt, H. J. Biol. Chem. 1976, 251, 3862-3867.


I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
/'Jxju Jc(lu
David N. Silverman, Chair
Professor of Pharmacology and
Therapeutics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
CLwcfotS}
E. Raymond Andrew
Graduate Research Professor of
Physics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
Robert J/Cohen
Associate Professor of
Biochemistry and Molecular
Biology
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Docjdf^f Philosophy.
Russell S. Drago
Graduate Research Professor of
Chemistry


References
Baes, N. F.; Mesmer, R. E. The Hydrolysis of Cations; Wiley: New York, 1976.
Bernheim, R. A.; Brown, T. H.; Gutowsky, H. S.; Woessner, D. E. J. Chem Phys.
1959, 30, 950-956.
Bertini, I.; Canti, G.; Luchinat, C.; Scozzafava, A. J. Am Chem Soc. 1978, 100,
4873-4877.
Bertini, I.; Dei, A.; Luchinat, C.; Monnanni, R. Inorg. Chem. 1985, 24, 301-303.
Bertini, I.; Luchinat, C. Acc. Chem. Res. 1983, 16, 272-279.
Bertini, I.; Luchinat, C. NMR of Paramagnetic Species in Biological Systems;
Benjamin-Cummings: Menlo Park, CA, 1986.
Bertini, I.; Luchinat, C.; Scozzafava, A.; Inorg. Chem. Acta 1980, 46, 85-89.
Bertini, I.; Luchinat, C.; Scozzafava, A. Struct. Bonding (Berlin) 1982, 48, 45-92.
Bloembergen, N. J. Chem Phys. 1957, 27, 572-573.
Bloembergen, N.; Morgan, L. O. J. Chem. Phys. 1961, 34, 842.
Brown, G. S.; Navon, G.; Shulman, R. G. Proc. Natl. Acad. Sci. USA 1977, 74,
1794-1797.
Carlson, G. A.; McReynolds, J. P.; Verhoek, F. H. J. Am. Chem. Soc. 1945, 67,
1334-1339.
Chiang, Y.; Kresge, A. J.; More O'Ferrall, R. A. J. C. S. Perkin I11980, 1832-
1839.
Dwek, R. A. NMR in Biochemistry; Oxford press: Oxford, 1975.
Eigen, M.; Tamm, K. Z Elektrochem. 1962, 66, 93-121.
Eisenstadt, M. J. Chem Phys. 1969, 51, 4421-4432.
Engberg, P.; Lindskog, S. FEBS Lett. 1984, 170, 326.
79


72
buffer. Under these conditions, the relaxation of water protons in solutions of
Co(ll)-substituted carbonic anhydrase II would be independent of pH in the
range 6-9, and this is what was observed for the paramagnetic relaxivity of
solutions of Co(ll)-substituted carbonic anhydrase III (Figure 6-3).
The reason for the small value of Rp at low frequency (Figure 6-2) is not
clear. One possibility is that the lifetime of a proton in the hydration shell of the
cobalt is too long to allow for a large paramagnetic contribution (Fabry et al.,
1970; Tu et al., 1985). This water off rate has a maximal value of 104 S1 in the
case of native carbonic anhydrase III (Tu et al., 1983), but is 105 S'1 for Co(ll)-
substituted carbonic anhydrase II (Tu et al., 1985). This rate would be expected
to be even slower at 5C, but when the pH 6 sample was measured at 25 C,
there was a larger paramagnetic relaxivity at low frequency (data not shown).
Another possibility is that in this isozyme the cobalt is in a pentacoordinate or
hexacoordinate environment (Engberg and Lindskog, 1984), which would have
less ability to relax due to a shorter electron relaxation time (Koenig et al.,
1983). Whatever the reason, the paramagnetic contribution at low frequency is
small.
The increase in Rp with higher frequency (Figure 2) is unlike that for any
other Co(ll)-substituted carbonic anhydrase which show a decrease in Rp with
higher frequency (Wells et al., 1979), but is not unlike that observed for Mn, Cu,
and Ni proteins (Koenig and Brown, 1973; Bertini and Luchinat, 1986). While
the relaxation behavior of these systems is not completely understood, it is clear
that the field dependence of the electron spin relaxation is a factor. Thus, the
increase in the electron relaxation time may be responsible for the increase in
Rp at intermediate frequency. The decrease at higher frequency may be due to
the (Osxc dispersion predicted by the Solomon-Bloembergen equations (Bertini
and Luchinat, 1986).


10
are in the primary hydration shell of the ion, this allowed a qualitative
differentiation between the relaxation of inner and outer shell water.
Experimental Procedure
The NMR relaxation times T1 and T2 of water protons were measured at a
frequency of 300 MHz on a Nicolet NT-300 spectrometer at 23 C. T1 was
measured by the inversion recovery method according to Freeman et al. (1980)
and T2 was measured by the Carr-Purcell-Meiboom-Gill sequence (Meiboom
and Gill, 1958). The observed relaxation of water protons in a solution of a
paramagnetic ion is the sum of the relaxation due to the paramagnetic ion, 1/Tip
(i=1,2), and the relaxation in the absence of the ion, 1/Tio,
1/T¡ = 1/T¡p +1/Tb (2-1)
The contribution of the paramagnetic ion to the longitudinal relaxation is given
by (Luz and Meiboom, 1964)
1/Tip- p[1/(Tim +xm)] (2-2)
and the paramagnetic contribution to the transverse relaxation is given by (Swift
and Connick, 1962)
0/"l"2m)(1/T2m + 1/Tm) + A(Om2
1/T2p = p [ ] (2-3)
Tm((1/T2m + 1/Xm)2 + ACOm2)
where Tim is the relaxation time of a proton in the hydration shell of the metal
and is described by the Solomon-Bloembergen equations, Tm is the lifetime of a
proton in the hydration shell, and p' is q[M]/55.5 where q is the hydration
number and [M] is the concentration of the metal ion. Acom is the chemical shift
difference between a proton in the primary hydration shell and a proton in the


83
BIOGRAPHICAL SKETCH
James W. Kassebaum, the son of Amanda L. Schoonmaker and Wilbur
F. Kassebaum, was born March 6, 1962, in Mount Vernon, Illinois. James grew
up in St. Louis, Missouri, and in 1980 graduated first in his class at Parkway
South Senior High School. He then attended Purdue University, West
Lafayette, Indiana, and in 1984 graduated with Honors with the degree of
Bachelor of Science in chemistry. The summer of 1984 was a very eventful
one, when James was wed to Shari L. Gunderman and then entered graduate
school at the University of Florida in the Department of Biochemistry and
Molecular Biology and studied under the supervision of Dr. David Silverman.
James completed the requirements for the Ph.D. degree in 1988, and began his
professional career at the Monsanto Company in St. Louis.


25
These potentiometric titrations were performed in solutions which
contained ultra-pure C0CI2 (Aldrich Gold-Label), and glycine and N,N-
dimethylglycine from Sigma and acetylacetone from Aldrich. Water used for
solution preparations was distilled and passed through two ion-exchange resin
cartridges (Cole-Palmer 1506-35). D20 (99.8%) was stirred with activated
charcoal, the charcoal filtered out, and then the D20 was distilled. Glassware
was rinsed with a solution of EDTA prior to use.
Results
Fractionation Factors of Co(HQ)s2 and Ni(HzO)g2; The relaxation of
water protons in solutions of 10 mM CoCI2 and 10 mM NiS04 as a function of
the atom fraction of deuterium in solvent water is shown in Figures 3-1 and 3-2.
From these data, the hydrogen / deuterium fractionation factor, 0, for
Co(H20)62+ and Ni(H20)62+ was obtained (Table 3-2). The value of the
fractionation factor was independent of the concentration of metal in the range
0.010 M to 0.025 M. The value of (j) for Co(H20)62+ was dependent on whether
it was obtained from T-i or T2; however, the corresponding values of (}) for
Ni(H20)62+ were identical when obtained from T1 orT2. In a separate
experiment, the fractionation factor from T1 for Co(H20)62+ was determined at
frequencies between .01 and 50 MHz on a field cycling "re laxo meter" at the IBM
T. J. Watson Research Center and found to be virtually field independent (data
not shown).
Solvent Hydrogen Isotope Effects on the Formation Constants of Cobalt
Complexes: The formation constants for the reactions of Co2+ with glycine, N,N-
dimethylglycine, and acetylacetone were determined in H20 and D20 and are
reported as the logarithm of the formation constants (Table 3-3). Representative
data of a formation curve (Carlson etal., 1945) for Co2+ and glycine are in
Figure 3-3 where is plotted as a function of the negative logarithm of the


37
sites in Co(H20)62+, large isotope effects would be expected if the only
contributions to the isotope effect came from the fractionation of inner shfell
water (Kresge etal., 1987). However, isotope effects of unity were observed for
these reactions (Table 3-3).
I suggest that the individual contribution of the fractionation factor of the
inner shell waters in Co(H20)62+, Co(H20)4L+, and Co(H20)2L2 to the overall
solvent hydrogen isotope effect on the formation constants is small. The
fractionation factor of some other hydrogenic sites has a contribution to the
isotope effect which is sufficient to cancel the contribution of the inner shell
waters so that the overall isotope effect is unity. The most likely choice for this is
the fractionation of outer shell water. Other possibilities include, in the case of
glycine, the composite fractionation factor of the hydration shell of the
carboxylate group, but this would be a small contribution of a fractionation factor
value already close to unity (the fractionation factor of the hydration shell of the
carboxylate group in the acetate ion is 0.89, Goodall and Long, 1968). The
fractionation factor of the hydrogens on the nitrogen in glycine are probably not
important since the isotope effect is the same for A/,/V-dimethylglycine (Table 3-
3).
The importance of outer shell water in ligand substitution reactions is not
without precedent. The first step in the proposed mechanism of ligand
substitution reactions is the formation of an outer shell complex with the ligand
and metal (Eigen and Tamm, 1962), followed by replacement by the ligand of a
water molecule in the inner coordination shell of the metal. Hence it is
reasonable to assume that the binding of a ligand would affect the structure of
outer shell water.
The following example illustrates how the fractionation of outer shell
water could contribute to the isotope effect. Equation 3-7 can be expanded to


To apply the fractionation factors to an equilibrium solvent hydrogen
isotope effect associated with the enzyme, I have studied at equilibrium the
binding of I" to Co(ll)-substituted carbonic anhydrase I and II. I' is known to
bind to the metal in Co(ll)-substituted carbonic anhydrase I and II displacing the
coordinated water molecule (Brown et ai, 1977). Also, I" binds fairly tightly
(Table 5-2) and its size is comparable to that of a water molecule. Thus the
isotope effect on the binding of I to Co(ll)-substituted carbonic anhydrase may
be simple enough to be accounted for by the fractionation factor of the metal-
bound water.
The solvent hydrogen isotope effect on the binding of the iodide ion to
the active site of carbonic anhydrase is given by
^Co-I + H20 ^Co-OHI '
(fco-OH/
where (K¡I')h2o is the equilibrium dissociation constant describing the binding of
I- to species i of Scheme 5-1 in H20. Because it is known that (|)h2o = 1-0, with
knowledge of (¡>t2 for Co-OH2 from Table 5-3,1 thus predict that ApK1' = 0.09 0.04 for Co(ll)-
isozyme I, and 0.18 0.03 for Co(ll)-isozyme II. The predicted value for isozyme
I is in rough agreement with the measured value (Table 5-2), and the predicted
value for isozyme II is slightly below the measured value. However, the
predicted value for isozyme I is less than the predicted value for isozyme II,
which is the same as in the experimental case. I have considered other
interactions which may be taken into account for r binding. One of these is the
composite fractionation factor of the solvation shell of T (0 = 0.59, Voice, 1974).


CHAPTER 1
INTRODUCTION
The effect of an isotopic substitution on rate and equilibrium constants
associated with a chemical reaction can give information about the mechanism
of the reaction. One method in making an isotopic substitution is the
replacement of D2O for H2O as the solvent, which can affect a reaction in two
ways. First if water is a substrate in the reaction, then the rate and equilibrium
constants can be affected by the presence of D2O. Also, if there are
exchangeable hydrogenic sites in the reactants or products, or, in the case of an
enzyme-catalyzed reaction, if there are exchangeable hydrogenic sites on the
enzyme, then the rate and equilibrium constants can be affected when D2O is
the solvent, because deuterium will be present in the exchangeable positions.
The difference in a reaction in H20 and D2O results from differences in
the zero point energy of bonds to H or D (Schowen, 1978). In the case of a
reaction rate, if the zero point energy difference for a bond to H and to D
decreases on going to the transition state, then the activation energy is greater
in the case of D, and the rate is lower. This is an example of a normal isotope
effect, in which the rate constant when the bond is to H, kn, is greater than the
rate constant when the bond is to D, kp. There also exist reactions where ko >
kn, which is known as an inverse isotope effect.
The difference in zero point energy can be expressed in terms of an
equilibrium constant for isotope exchange for a particular hydrogenic site. For a
single solute species in an isotopically-mixed aqueous solvent, the hydrogen /
1


44
Atom fraction of deuterium in solvent (n)
Figure 4-1: The dependence of p'Tip (i=1,2) at 300 MHz on n, the atom fraction
of deuterium in solvent water, for 1.5 mM solutions of Co(ll)-substituted carbonic
anhydrase I. Solutions contained 50 mM Hepes buffer at pH 8.5 and 23C.
Data are the means and standard errors of three measurements.


70
1.0
0.8
0.6
0.4
0.2
0.0
5.0 6.0 7.0 8.0 9.0 10.0
pH
Figure 6-3: The total relaxivity of Co(ll)-substituted carbonic anhydrase III and
native zinc carbonic anhydrase III as a function of pH. Experimental conditions
as in Figure 6-1. Co(ll)-substituted carbonic anhydrase III measured at 1 MHz
( a ), native zinc carbonic anhydrase III measured at 1 MHz ( ), Co(ll)-
substituted carbonic anhydrase III measured at 40 MHz ( ), native zinc
carbonic anhydrase III measured at 40 MHz ( ).


18
Tables 2-1 and 2-2). The chemical shift does not have an effect at 20 MHz at
which frequency Acom21/T2mTm, 1AT2m2. The large chemical shift at 300 MHz
due to the addition of Co2+ is the result of contact interactions which result from
the delocalization of unpaired electron spin density. At lower magnetic field
strengths this mechanism has been observed to affect T2 for 170 and 19F in
cobalt complexes (Swift and Connick, 1962; Eisenstadt, 1969). The effects on
the relaxation when EDTA is added to solutions of Co(H20)62+ is the second
way I support the chemical shift mechanism. The chemical shift mechanism will
affect T2p to the greatest extent when the protons are in the primary hydration
shell of the ion because it is a contact interaction. EDTA binds to Co2+ (and
Ni2+) occupying all coordination sites and thus excludes water molecules from
the inner shell. Thus the addition of EDTA would prevent the water protons from
experiencing the chemical shift mechanism. The ratio Tip/T2p was near unity for
water protons at 300 MHz when an equimolar amount of EDTA was added to
solutions of Co2+, consistent with the absence of a chemical shift relaxation
mechanism and the dominance of a dipole-dipole relaxation mechanism
(Figure 2-1).
When EDTA was added to solutions of Co2+, the percent decrease in the
relaxivity R2 at 300 MHz was much greater than (Figure 2-1). Since EDTA is
excluding water molecules from the inner coordination shell, this suggests that
R2 has a much larger contribution from the relaxation of water protons in the
inner shell. The residual paramagnetic relaxivity of the cobalt-EDTA complex is
most likely due to water protons in an "outer shell" of the complex. This residual
relaxivity makes up a much larger part of R1 than R2 observed for Co(H20)62+
solutions at 300 MHz. This suggests that the relaxation of outer shell water
contributes to a greater extent to Rt than to R2 in Co(H20)62+.


a chemical shift mechanism, the fractionation factor determined from T2 has a
large contribution of the fractionation of inner shell ligands. The fractionation
factors determined from T-i are close to unity, the value of the fractionation factor
of bulk water, and contain a larger contribution of the fractionation of outer shell
water.
The fractionation factor of Co(H20)62+ was used to interpret the solvent
hydrogen isotope effects on the formation of complexes of cobalt with the
bidentate ligands glycine, A/,A/-dimethylglycine, and acetylacetone. The
contribution of the fractionation factor of the inner shell water in Co(H20)62+ did
not account completely for the measured isotope effect, and I suggest that the
hydrogen / deuterium fractionation of outer shell water makes a large
contribution to the isotope effect on the formation of these complexes.
The solvent hydrogen isotope effect on the equilibrium binding of iodide
ion to Co(ll)-substituted carbonic anhydrase I and II was determined and then
interpreted using the fractionation factor of the aqueous ligand of cobalt at the
active site. This fractionation factor could not account for the measured isotope
effect, implying that the hydrogen / deuterium fractionation of additional groups,
or water associated with the enzyme, changes on going from the reactant state
to the product state when iodide binds to Co(ll)-substituted carbonic anhydrase I
and II. Although not yet helpful in interpreting the complex contributions to the
isotope effects in the enzymatic catalysis, these hydrogen / deuterium
fractionation factors for the water bound to cobalt in carbonic anhydrase can be
significantly different from the fractionation factor for solvent water and may be
sensitive to the active site environment in these homologous isozymes of
carbonic anhydrase.
VI


CHAPTER 5
INTERPRETATION OF SOLVENT HYDROGEN ISOTOPE
EFFECTS ASSOCIATED WITH CARBONIC ANHYDRASE
Introduction
A solvent hydrogen isotope effect on a reaction is made up of individual
contributions to the overall isotope effect from each hydrogenic site in the
reaction (Schowen, 1978). This contribution is measured by the hydrogen /
deuterium fractionation factor of each particular hydrogenic site. In considering
reactions of carbonic anhydrase, the fractionation factor of the aqueous ligand
of the metal represents a reactant state contribution that can be used to interpret
equilibrium solvent hydrogen isotope effects which are a function of reactant
and product state site fractionation factors, and kinetic isotope effects which are
a function of reactant and transition state site fractionation factors.
Halide ions are inhibitors of carbonic anhydrase which bind to the metal
and displace the water ligand of the metal (Bertini et al., 1982). I have
determined the solvent hydrogen isotope effect on the equilibrium binding of
iodide ion to Co(ll)-substituted carbonic anhydrase I and II and interpreted this
effect using the fractionation factor of the aqueous ligand of cobalt in Co(ll)-
substituted carbonic anhydrase measured from NMR relaxation rates. The
results suggest that the isotope effect on T binding binding is the complex
contribution of the fractionation of more than one hydrogenic position, because
the consideration of the fractionation factor of the aqueous ligand of cobalt by
itself does not allow a complete interpretation of the isotope effect.
The fractionation factor of the aqueous ligand of cobalt in Co(ll)-
substituted carbonic anhydrase can also be used to interpret kinetic isotope


As EDTA was added to solutions of Ni2+, R2 decreased more than R1
(Figure 2-2), though the decrease is not as large as seen in solutions of Co2+.
Also, the residual relaxivity of the NiEDTA complex is much larger than in
solutions of the CoEDTA complex, and this residual relaxivity makes up a large
part of both R^ and R2. These results for Ni2+ suggest that the chemical shift
mechanism is responsible for the increase in R2 at 300 MHz, but that Ri and R2
each have a large contribution of outer shell water.


3
all reactant state sites
n all product state sites
(1-2)
where (j)R, (j)T, and §p are reactant, transition, and product state hydrogenic site
fractionation factors, respectively (Schowen, 1978). Thus it is clear that to
understand completely an isotope effect associated with a reaction, the
knowledge of the underlying fractionation factors is required.
Values of the fractionation factor for various functional groups are known
relative to water as the solvent (Schowen and Schowen, 1982; Kresge et at.
1987), and range from 0.4 for some examples of hydrogen bound to sulfur, to
1.5 for one case of a hydrogen bound to a tertiary amine. The majority of these
fractionation factors where measured by one of two methods. One involves
using NMR chemical shifts to measure the fractionation factor of a hydrogenic
site directly. The technique is based on the rapid exchange of hydrogens
between solute and solvent species, so that the observed proton resonance is a
weighted average of the chemical shift of the solute and solvent positions. The
fractionation factor is determined by measuring the chemical shift at different
concentrations of solute in solvents of low and high atom fraction of deuterium.
Using this method, Gold and Lowe (1968) determined the fractionation factor of
the exchangeable hydrogen of acetic acid to be 0.96 0.02. A variation of this
method when there is slow exchange of hydrogens between solute and solvent
involves measuring the NMR signal areas in mixed isotopic solvents. In this
manner Chiang et at. (1980) have determined the fractionation factor of the
exchanging position of 1,8-Bis(dimethylamino)-naphthalene to be 0.90 0.01,
and Szawelski et ai (1982) have measured the fractionation factor of the


(Engberg and Lindskog, 1984). Also, the CO2 hydration activity of the cat
isozyme III and the bovine isozyme III is independent in the pH range 6-8.5 (Tu
etal., 1983; Engberg etal., 1985). These data imply that the activity controlling
group in Co(ll)-substituted and native Zn-carbonic anhydrase III, the aqueous
ligand of the metal, is predominantly in the hydroxide form in the pH range 6-9,
and that the pK for the ionization of the metal-bound water to the metal-bound
hydroxide is well below 6.
NMR relaxation enhancement studies with Co(ll)-substituted carbonic
anhydrase I and II have shown a large pH dependence of the paramagnetic
relaxation rate in the pH range 6-9, with an inflection at pH 7 and maximal
relaxation at high pH (Fabry etal., 1970; Wells etai, 1979). This correlates with
the pH dependence of both the visible spectrum and CO2 hydration activity
(Bertini etal., 1982; Silverman and Vincent, 1983). Thus the NMR relaxation
enhancement seems to correlate with the ionization state of the aqueous ligand
of the metal in Co(ll)-substituted carbonic anhydrase I and II.
However, it has been shown that SO42- present in solution to maintain
ionic strength affects the pH dependence of both the visible spectrum and the
paramagnetic relaxation of Co(ll)-substituted carbonic anhydrase I and II
(Bertini et ai, 1978; Bertini et a/.,1980; Simonsson and Lindskog, 1982). When
S042- and other anions are absent in either unbuffered solutions or solutions
containing only sulfonic acid buffers, the visible spectrums of Co(ll)-substituted
carbonic anhydrase I and II still show a large pH dependence, but there is
practically no pH dependence on the paramagnetic contribution to the
longitudinal relaxation rate of water protons in solutions Co(ll)-substituted
carbonic anhydrase II (Bertini etal., 1978), and a small dependence for Co(ll)-
substituted carbonic anhydrase I (Bertini etal., 1980). The data in Figure 6-3
are from solutions containing no anions other than those of the sulfonic acid


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
HYDROGEN / DEUTERIUM FRACTIONATION FACTORS OF
THE AQUEOUS LIGAND OF COBALT IN Co(H20)62+ AND
Co(ll)-SUBSTITUTED CARBONIC ANHYDRASE
By
James Web Kassebaum
December, 1988
Chairman: Dr. David N. Silverman
Major Department: Biochemistry and Molecular Biology
I have measured the hydrogen / deuterium fractionation factor for the
rapidly exchanging aqueous ligands of cobalt in Co(H20)62+ and in three Co(ll)-
substituted isozymes of carbonic anhydrase. The fractionation factor was
determined from NMR relaxation rates at 300 MHz of the protons of water in
mixed solutions of H20 and D20 containing these complexes. In each case, the
paramagnetic contribution to 1/T2 was greater than to 1/Ti, consistent with a
chemical shift mechanism affecting 1/T2. The fractionation factors obtained from
Ti for Co(H20)62+ and for the isozymes of Co(ll)-substituted carbonic anhydrase
were close to the fractionation factor for bulk water, which is unity. The
fractionation factors obtained from T2 were 0.73 0.02 for Co(H20)62+, 0.72
0.02 for Co(ll)-substituted carbonic anhydrase I, 0.77 0.01 for Co(ll)-
substituted carbonic anhydrase II, and 1.00 0.07 for Co(ll)-substituted
carbonic anhydrase III. I concluded that fractionation factors in these cases
determined from Ti and T2 measured isotope preferences for different
populations of ligand sites. I suggest that since T2 has a large contribution from
v


36
carbonyls. If the assumption in these calculations is valid, the results of a lower
fractionation factor for Co(H20)4L+ means that the O-H bond becomes weaker
relative to that bond in Co(H20)62+, and the O-H bond is also weaker in
Co(H20)2L2 compared to Co(H20)4L+. This is similar to the effect of added
ligands on the lifetime of water molecules in the hydration shell which, for
instance, is shorter in Co(H20)4gly+ relative to Co(H20)62+ (Hammes and
Steinfeld, 1962), meaning that the hydration shell becomes more loosely bound
when ligands are added. Thus in these cobalt complexes with bidentate
ligands the metal-oxygen bond of a coordinated water is looser, and the oxygen
hydrogen bond appears looser, relative to Co(H20)62+.
However, from the measured solvent hydrogen isotope effects (Table 3-
3) and the calculations of Table 3-4, there are a number of reasons to question
the assumption that the only contribution to the isotope effects comes from the
hydrogens of inner shell water. First, such strong fractionation as 0.36 (Table 3-
4) is rare (Schowen and Schowen, 1982). Second, the lower fractionation
factor for Co(H20)4L+ and Co(H20)2L2 is in contrast to the suggestion that the
charge on the cobalt affects the fractionation of the metal-bound water
(Schowen and Schowen, 1982). In these complexes, the net charge on the
complex is +1 for Co(H20)4L+, and neutral for Co(H20)2L2, since the ligand has
a charge of -1 in each case. Thus one would expect the fractionation factor of
the waters in the ligand complexes to be closer to unity, according to the
expectation that <|) = (0.69)5, where 0.69 is the fractionation factor of H30+ and 8
is the degree of positive charge transferred from the metal to the oxygen of the
coordinated water molecule (Schowen and Schowen, 1982). The c()t2 for
Co(H20)62+ is consistent with this (Table 3-2), and thus one would expect [(f) for
Co(H20)62+] < [(f) for Co(H20)4L+] < [(f) for Co(H20)2L2], but this opposite to what
was observed (Table 3-4). Finally, based upon the large number of hydrogenic


6 7
0.01 to 50 MHz. The paramagnetic relaxivity, Rp, is Rt determined for a solution
of Co(ll)-substituted carbonic anhydrase III minus R, for a solution of native Zn-
carbonic anhydrase III.
Re?UltS
The frequency dependence of the total relaxivity was similar to the NMR
dispersion profiles of other native and Co(ll)-substituted carbonic anhydrases
(Fabry etal., 1970; Wells etal., 1979) (Figure 6-1). The total relaxivity is
greatest at low frequency, then begins to decrease at about 0.1 MHz. The
paramagnetic relaxivity, however, showed a small value at low frequency, then
at about 0.5 MHz Rp began to increase to a maximal value at about 5 MHz, and
then began to decrease again (Figure 6-2). The total relaxivities for both the
Co(ll)-substituted and native Zn-carbonic anhydrase III showed a possible
increase with pH at 1 MHz, but remained constant with pH at 40 MHz (Figure 6-
3). This may be the same effect reported by Wells etal. (1979), who observed
that the diamagnetic component of the relaxivity in solutions of Co(ll)-substituted
and native zinc carbonic anhydrase II increases with pH at low frequency.
When only the paramagnetic relaxivity was considered (the difference between
Rt of Co(ll)-substituted carbonic anhydrase III and Rt of native zinc carbonic
anhydrase III shown in Figure 6-3), there was no pH dependence, especially
when the data at pH 7 in Figure 6-3 are excluded.
Discussion
This work is the first study of NMR relaxation enhancement of water
protons in solutions of Co(ll)-substituted carbonic anhydrase III. Previous work
with isozyme III has shown that the visible spectrum of Co(ll)-substituted
carbonic anhydrase III is very similar to the high pH spectral forms of Co(ll)-
substituted carbonic anhydrase I and II; however, for Co(ll)-substituted carbonic
anhydrase III there is no pH dependence of the spectrum in the range pH 6-9


Undskog, S. in Zinc Enzymes; Spiro, T. G. Ed.; Wiley: New York; 1983; 78-121.
Lowe, B. M.; Smith, D. G. J. Chem. Soc., Faraday Trans. /1975, 71, 389-397.
Luz, Z.; Meiboom, S. J. Chem. Phys. 1964, 40, 2686-2692.
Meiboom, S.; Gill, D. Rev. Sci Instrum. 1958, 29, 688.
Melamud, E.; Mildvan, A. S. J. Biol. Chem. 1975, 250, 8193-8201.
Melton, B. F.; Pollack, V. L. J. Phys. Chem. 1969, 73, 3669-3679.
More O'Ferrall, R. A.; Koeppl, G. W.; Kresge, A. J. Am. Chem. Soc. 1971,93, 9-
20.
Morgan, L. O.; Nolle, A. W. J. Chem Phys. 1959, 31, 365.
Nyman, P. O.; Lindskog, S. Biochim. Biophys. Acta 1964, 85, 141-151.
Pocker, Y.; Sarkanen, S. Adv. Enzymol. Relat. Areas Mol. Biol. 1978, 47,149-
274.
Schowen, K. B. J. In Transition States in Biochemical Processes] Grandour, R.
D.; Schowen, R. L. Eds.; Plenum: New York, 1978; 225-284.
Schowen, K. B.; Schowen, R. L. In Methods in Enzymology, Volume 87; Purich,
D. L. Ed.; Academic: New York, 1982; 551-606.
Silverman, D. N. J. Am. Chem. Soc. 1981, 103, 6242-6243.
Silverman, D. N.; Lindskog, S. Acc. Chem. Res. 1988, 21, 30-36.
Silverman, D. N.; Vincent, S. H. CRC Crit. Rev. Biochem. 1983, 14, 207-255.
Simonsson, I.; Lindskog, S. Eur. J. Biochem. 1982, 123, 29-36.
Smith, R. M.; Martell, A. E. Critical Stability Constants; Plenum: New York, 1975.
Solomon, I. Phys. Rev. 1955, 99, 559-565.
Steiner, H.; Jonsson, B.; Lindskog, S. Eur. J. Biochem. 1975, 59, 253-259.
Swift, T. J.; Connick, R. E. J. Chem. Phys. 1962, 37, 307-320; J. Chem. Phys.
1964, 41, 2553-2554.
Szawelski, R. J.; Wharton, C. W.; White, S. Biochem. Soc. Trans. 1982, 10, 232-
233.


30
Figure 3-3: The ratio of the concentration of metal-bound ligand to total
concentration of metal, , as a function of the negative logarithm of the
concentration of free glycine in solution existing as NH2-CH2-COO'1, p[A] (see
equation 3-6). The solutions contained 1 mM total C0CI2 and 4 mM total glycine
in H2O ( ) or D2O ( ) and were titrated with NaOH (NaOD). The solid lines are
a computer fit to the data using equation 3-6 with the values of the formation
constants given in Table 3-3.


40
advantage of the properties of a paramagnetic metal to enhance the relaxation
rate of water exchanging rapidly between bulk solvent and the hydration shell of
the metal and measuring the proton relaxation rate as a function of the
deuterium content of solvent. Silverman measured that fractionation factor
using T-i at a proton resonance frequency of 100 MHz. Because of the unique
relaxation properties of water protons in solutions of Co(H20)62+ at 300 MHz
(Chapter 2), I have determined the fractionation factor of the aqueous ligand of
cobalt in three mammalian isozymes of Co(ll)-substituted carbonic anhydrase
from T-\ and T2 at 300 MHz. In each case T1 > T2( implying that the same
relaxation mechanisms occur for water protons in solutions of Co(ll)-substituted
carbonic anhydrase at 300 MHz as in solutions of Co(H20)62+. The values of
the fractionation factor determined from T-i for each isozyme of Co(ll)-substituted
carbonic anhydrase were close to the fractionation factor for bulk water, which is
unity. Except in the case of Co(ll)-substituted carbonic anhydrase III, the
fractionation factors obtained from T2 were less than unity and close to the
fractionation factor for H30+ which is 0.69. I conclude that fractionation factors in
these cases determined from T-i and T2 measured isotope preferences for
different populations of ligand sites. I suggest that since T2 has a large
contribution from a paramagnetic chemical shift, the fractionation factors
determined from T2 have a large contribution of the hydrogen / deuterium
fractionation of the inner shell ligand. The fractionation factors determined from
T-i are close to unity, the value in bulk water, and contain a larger contribution of
the fractionation of outer shell water.
Experimental Procedure
Enzymes: Human carbonic anhydrase I was purified from human red
blood cells by an affinity chromatography method (Kalifah et al., 1977). Bovine
carbonic anhydrase II was obtained from Sigma Chemical and used after


CHAPTER 2
WATER PROTON NMR RELAXATION ENHANCEMENT IN
SOLUTIONS OF Co(H20)62+ AND Ni(H20)62+
Introduction
The initial studies of the influence of paramagnetic ions on the NMR
relaxation rates of water protons were performed at proton resonance
frequencies up to 60 MHz (Bernheim et al., 1959; Morgan and Nolle, 1959).
These results were interpreted in terms of the Solomon-Bloembergen
equations, which describe the relaxation as the sum of an electron-nuclear
dipole-dipole interaction and a scalar interaction (Solomon, 1955;
Bloembergen, 1957). At a frequency of 20 MHz the water protons in solutions of
Co(H20)62+ were observed by Bernheim et al. (1959) to relax such that T1 = T2,
and this relaxation was also observed at 10 MHz for solutions of Ni(H20)62+
(Morgan and Nolle, 1959). This result is consistent with the electron-nuclear
dipole-dipole mechanism dominating the relaxation of the water protons, the
correlation time for the interaction being the very short electron spin relaxation
time. This short relaxation time for the electron in Co(H20)62+ and Ni(H20)62+
precludes any contribution to the relaxation from the scalar mechanism. In the
absence of a scalar contribution, 7/6 would be the largest observable ratio for
Ti/T2.
I have made measurements of the relaxation of water protons in solutions
of Co(H20)62+ and Ni(H20)62+ at a proton resonance frequency of 300 MHz. In
each case it was observed that the ratio T-|/T2 was much larger than unity,
consistent with a chemical shift mechanism affecting T2 at this frequency.
Because this mechanism will affect T2 to the greatest extent when the protons
9


64
n\
%/NH
r=\
V/NH
This calculation for the fractionation factor of the water bridge in the transition
state suggests that the protons of the water bridge have a fractionation factor
less than unity, meaning they have hydronium ion character, consistent with
positive charge transferred to the oxygen during the proton transfer process.
This calculation of the fractionation factor of the transition state suggests that the
proton transferred from the cobalt-bound water is in between the cobalt-bound
water and the water bridge in the transition state. As in the case of iodide
binding to Co(ll)-substituted carbonic anhydrase, there may be other
hydrogenic sites in the transition state which contribute to the overall isotope
effect. However, the fractionation factors of the aqueous ligand of the metal
have led to this qualitative model of the transition state, which is consistent with
both the large isotope effect on kcat and the results of Venkatasubban and
Silverman (1980) for the catalysis of CO2 hydration in H20 / D20 mixtures.
X H
Co-OH2 0
H
X ,H'"Ox
-Co-<3 'H1
/ H


Table 4-1: Paramagnetic contribution to the relaxivities of water protons
observed for aqueous solutions of Co(ll)-substituted carbonic anhydrase I, II,
and III.
frequency
(MHz)
R1
(mM'V1)
r2
(mM-'s-1)
T1p/T2p
Co(ll)-carbonic anhydrase I
300
0.28
2.3
8.2
Co(ll)-carbonic anhydrase II
300
0.26
1.4
5.3
Co(ll)-carbonic anhydrase III
300
0.14
2.2
16
Note: Measurements were made at 23 C for solutions of enzyme buffered at
pH 8.5 by 50 mM Hepes. Solutions contained 20% D2O by volume.


Relaxivity (mM*1 s-1)
14
[EDTA]/[Ni2+]
Figure 2-2: The relaxivity of the protons of water at 300 MHz due to Ni2+ in
solutions of 10 mM NiS04 as a function of the molar ratio of EDTA to Ni2+ at 23
C. Solutions contained 50% D2O by volume. R1 (), R2 (a).


5
a small number of solvent hydrogen isotope effects measured for equilibrium
constants and for rate constants of reactions with Cr3+ and Co3+ (Laughton and
Robertson, 1969; Gold and Wood, 1982). Kresge et al. (1987) have suggested
that large rate and equilibrium differences between reactions of transition metal
aquacomplexes in H2O and D20 should be expected because the hydrogen /
deuterium fractionation factor of the water molecules in the inner coordination
shell would be raised to the twelfth power for a hexaaquacomplex.
Solvent hydrogen isotope effects on reactions catalyzed by carbonic
anhydrase are known and have been used to deduce aspects of the catalytic
mechanism (Silverman and Vincent, 1983). Carbonic anhydrase is a zinc
metalloenzyme which catalyzes the reversible hydration of carbon dioxide to
produce bicarbonate and a proton.
C02 + H20 ~ HCOg+H+
The zinc can be removed and replaced with cobalt to produce an enzyme with
very similar catalytic properties (Lindskog, 1983) and the cobalt, because of its
unpaired electrons, provides a spectroscopic probe of the enzyme (Bertini and
Luchinat, 1983). In particular, the paramagnetic cobalt will enhance the NMR
relaxation rate of water protons in solutions of Co(ll)-substituted carbonic
anhydrase. The metal ion is at the base of the active site cleft and is
coordinated by three histidine ligands with a water molecule as a fourth ligand
which can ionize to a hydroxide. The mechanism of C02 hydration (scheme 1-
1) takes place by direct nucleophilic attack of a metal-bound hydroxide on C02
to produce HCO3' (Steiner et al. 1975), with a water molecule then displacing
HCCV to produce a metal-bound water. A proton must then be transferred out
of the active site to regenerate the metal-bound hydroxide for the next round of
catalysis.


Undskog, 1988) apparently do not have a significant effect on the fractionation
factors which were nearly identical for isozymes I and II.


these values with equation 2-3 are in Table 2-2 and are in good agreement with
the experimental results in Table 2-1.
Discussion
The results at 300 MHz of T1p/T2P much greater than unity for solutions of
both Co(H20)62+ and Ni(H20)62+ suggest that a relaxation mechanism different
from the dipole-dipole mechanism is present at this frequency. The data in
Table 2-1 indicate that this mechanism primarily, if not exclusively, affects the
transverse relaxation. I have considered three possible causes of this field-
dependent increase in 1/T2p. The scalar contribution to 1/T2m has a correlation
time equivalent to the electron relaxation time which is field dependent
(Bloembergen and Morgan, 1961). However, I have rejected a contribution
from the scalar term because it would require an unreasonably large increase
in the electron spin relaxation time at 300 MHz. When the electron relaxation
time is less than both the rotational correlation time and xm, a Curie spin
mechanism can result (Gueron, 1975). This Curie spin is the thermal average
of an electron spin which can become large at high field and can cause an
increase in T-im/T2m. However, I also reject a contribution from this mechanism
because the rotational correlation time for a hexaaquacomplex is too short to
allow the Curie spin to become effective.
Melamud and Mildvan (1975) observed a field-dependent increase in
1/T2p for water protons in solutions of Co(H20)62+ and concluded that the Acom2
term in equation 2-3, a chemical shift mechanism, made a major contribution at
the higher frequencies. I support this conclusion in two ways. First, I have
calculated the magnitude of the chemical shift contribution to the transverse
relaxation of Co(H20)62+ solutions at 20 MHz and 300 MHz using equation 2-3.
These calculations demonstrate that the observed value of Tip/T2p at 300 MHz
is consistent with a chemical shift mechanism which affects T2p (compare


4 1
extensive dialysis to remove paramagnetic impurities. Bovine carbonic
anhydrase III was obtained from bovine flank steak by gel filtration and anion-
exchange chromatography (Tu et al., 1986). The concentration of carbonic
anhydrase I, II, and III was estimated at 280 nm using 8 = 4.7 x 104 M'1 cm-1
(Nyman and Lindskog, 1964), 5.4 x 104 M1 cm*1 (Nyman and Lindskog, 1964),
and 6.4 x 104 M-1 cm-1 (Engberg and Lindskog, 1984), respectively.
The apoenzymes of carbonic anhydrase I and II were prepared
according to the procedure of Hunt et al. (1977) by dialysis against dipicolinic
acid. Co(ll) substituted carbonic anhydrase I and II were prepared by the
addition of 1.1 equivalents of C0CI2 followed by dialysis against several
changes of large volumes of deionized water. The apoenzyme of carbonic
anhydrase III was prepared by addition of the chelator 2-carboxy-1,10-
phenanthroline to a solution of enzyme followed by dialysis against excess
C0CI2 as described by Engberg and Lindskog (1984).
NMR Measurements: Measurements at 300 MHz were the same as in
Chapter 2. In the cases of Co(ll)-substituted carbonic anhydrase I and II, 1/T¡0
was taken as the relaxation of solutions of the Co(ll)-substituted enzyme in the
presence of a molar excess of the inhibitor acetazolamide, which is known to
displace the aqueous ligand of the metal at the active site. Enough
acetazolamide was present to bind to 99.9% of the active sites. Alternatively for
Co(ll)-isozymes I and II and solely for Co(ll)-substituted carbonic anhydrase III,
1/Tj0 was taken as the relaxation of solutions of the native zinc enzyme, which
has no paramagnetic contribution. These two methods were equivalent at a
frequency of 300 MHz for Co(ll)-substituted carbonic anhydrase I and II.
Solutions of enzyme for relaxation measurements were buffered with 50 mM
Hepes [4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid] at pH 8.5. At this
pH the relaxivity of solutions of Co(ll)-substituted carbonic anhydrase I and II is


5 9
with the conclusion that active site properties do not seem to influence the
fractionation factors of the aqueous ligand of cobalt in Co(ll)-substituted
carbonic anhydrase I and II, see Chapter 4). Then the calculation to determine
the fractionation factor of the water ligand of the metal can be performed
neglecting the ionization state of the second ionizable group.
The isotope effect on an equilibrium constant is the product of all of the
reactant state fractionation factors divided by the product of all of the product
state fractionation factors (equation 1-2). For this ionization at the active site of
Co(ll)-substituted carbonic anhydrase
^Co OH2 +H20 3;Co OH +H30+
the solvent hydrogen isotope effect can be written
Khp ^Co-OHj^HgO)2
2
KD20 (^Co-OH )
From this measured isotope effect, the measured fractionation factor for the
cobalt-bound hydroxide, the fractionation factor for H3 measured to be 0.69 (Chiang, 1980), and the fractionation factor for H2O which
is 1.0, the fractionation factor for the exchangeable hydrogens at the active site
in its low pH form can be calculated and are presented in Table 5-3. I have
done the calculations of Table 5-3 using both T2 measured for the
high pH form of the aqueous ligand of cobalt in Co(ll)-substituted carbonic
anhydrase I and II. However, the calculation from T-i may not be meaningful,
since it contains a large contribution of outer shell water. The value calculated
using shell cobalt-bound water.