Influence of solvent on the ring-chain hydrolysis equilibrium of anabaseine and syntheses of anabaseine and nicotine ana...

MISSING IMAGE

Material Information

Title:
Influence of solvent on the ring-chain hydrolysis equilibrium of anabaseine and syntheses of anabaseine and nicotine analogues
Physical Description:
xiv, 207 leaves : ill. ; 29 cm.
Language:
English
Creator:
Bloom, Linda B., 1964-
Publication Date:

Subjects

Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1990.
Bibliography:
Includes bibliographical references (leaves 199-206).
Statement of Responsibility:
by Linda B. Bloom.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001638758
notis - AHR3698
oclc - 24160902
System ID:
AA00004119:00001

Full Text











INFLUENCE OF SOLVENT ON THE RING-CHAIN HYDROLYSIS EQUILIBRIUM
OF ANABASEINE AND SYNTHESES OF ANABASEINE AND NICOTINE
ANALOGUES















BY


LINDA B. BLOOM


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


1990













ACKNOWLEDGEMENTS


The author is indebted to Dr. John A. Zoltewicz for his

guidance and patience throughout her graduate studies.

Special thanks are also given to Dr. William R. Kem for his

advice and assistance. The support of the other members of

her supervisory committee, Dr. Merle A. Battiste, Dr. James

A. Deyrup, and Dr. John R. Eyler, is also appreciated. The

author is grateful to Dr. Roy W. King for his training and

technical assistance in NMR spectroscopy.

Special appreciation is given to her husband, David

Bloom, for his support and encouragement.

Financial support from the Chemistry Department and the

Graduate School at the University of Florida and from Taiho

Pharmaceutical Co. Ltd. is gratefully acknowledged.













TABLE OF CONTENTS

Page

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

LIST OF TABLES ........................................... vii

LIST OF FIGURES ............... ............................ ix

ABSTRACT .................................................. xii

CHAPTER

1 INTRODUCTION ........................... ............... 1
General Overview ...................................... 1
Brief Summary of Investigations by Chapter ........... 3

2 QUANTITATIVE DETERMINATION OF THE RING-CHAIN
HYDROLYSIS EQUILIBRIUM CONSTANT FOR ANABASEINE ....... 13
Introduction .... ........ ................. ............ 13
Results .............................................. 14
NMR Measurements of Aqueous Anabaseine
Solutions ................ ........................ 14
Assignment of structures and composition in
aqueous solutions ............................ 15
Calculation of pKa values from pD dependent
chemical shifts ........................... .. 23
The pD-hydrolysis profile and the hydrolysis
equilibrium constant KH ...................... 26
Ultraviolet Spectra of Aqueous Anabaseine
Solutions ............... ......................... 28
Discussion ... .. ... .............. ........... ...... 32
Anabaseine Equilibria ............................. 32
Apparent pKa Value.................................. 35
Conclusions ............ ............. .................. 37

3 COMPUTER SIMULATION OF pH-DEPENDENT RING-CHAIN
EQUILIBRIA FOR OTHER COMPOUNDS OF BIOLOGICAL
INTEREST ............ ................................... 38
Introduction ................................ ........... 38
Results ................ .............................. 40
pD-Hydrolysis Profiles for Myosmine and Analogs ... 40
Myosmine ......................................... 44
1-Methyl-2-(3-pyridyl)-1-pyrrolinium ion ....... 44
l-Methyl-2-phenyl-l-pyrrolinium ion and
2-phenyl-l-pyrroline ......................... 47


iii











Oxidation Product of Spermine and Spermidine ...... 49
Discussion ........................................... 55
Skewing of Observed pKa Values by Concurrent
Ring-Chain Equilibria ........................... 55
Myosmine and Iminium Ion Acidities ................ 57
KH Values for Hydrolysis of Cyclic Iminium Ions.... 61
Conclusions ................................ ........ 62

4 COMPENSATORY CHANGES IN THE INFLUENCE OF COSOLVENTS
ON THE POSITION OF THE RING CHAIN EQUILIBRIUM FOR
ANABASEINE AND N-METHYLANABASEINE .................. 64
Introduction ...... ...................... .............. 64
Results ........................................... 66
Ring-Chain Equilibria. Nomenclature ................ 66
Survey of Solvent Effects ......................... 67
(1) Water .. ............................. ......... 67
(2) DMSO .. ......................... ............. 68
(3) Methanol ........................... ........ 73
Quantitative Studies .............................. 74
(1) Water-methanol ....................... ......... 74
(2) Water-DMSO .................................. 82
Discussion ........................................... 84
Solvent Effects .................................. 84
Steric Effects ................................... 87
Significance for Binding Studies .................. 90

5 PREPARATION OF 1-METHYL-2,3'BIPYRIDINIUM ION ......... 93
Introduction ............................ .......... .... 93
Results ................................................. 98
Syntheses .......................................... 98
The 2-(2-pyridyl)ethyl protecting group ........ 98
The 2-(4-nitrophenyl)ethyl protecting group ..... 100
Meisenheimer-type a Adducts of 2,3'-Bipyridinium
Dications ................ ...................... 102
UV evidence for adducts ......................... 102
NMR evidence for adducts ........................ 105
1H NMR of N-Methylated 2,3'-Bipyridines and
Protonated 2,3'-Bipyridines ............. ...... 112
Discussion ............................................ 114
Syntheses ........................................ 114
Formation of Meisenheimer-Type 0 Adducts of
Diquaternized 2,3'-Bipyridinium Ions ............ 115
1H NMR of N-Methylated 2,3'-Bipyridines and
Protonated 2,3'-Bipyridines ..................... 117

6 SYNTHESIS OF ANABASEINE, ANABASEINE ANALOGUES, AND
BIPYRIDINES ............. ......... ..................... 120
Introduction .................................... ..... 120










Results and Discussion .................. ............. 123
Synthesis of Anabaseine and Cyclic Imine
Analogues ....................................... 123
Preparation of 5-(N,N-Dimethylamino)-1-(3-
pyridyl)-1-pentanone ............................ 127
(S)-N-Methylanabasine ............................. 128
Palladium Catalyzed Cross-Coupling of
Heteroaromatic Molecules ........................ 129

7 NATURE OF BINDING ENVIRONMENT OF SUBSTRATES IN SDS
MICELLES AS REVEALED BY PROTON CHEMICAL SHIFTS AND
LONGITUDINAL RELAXATION TIMES (TI) ................... 132
Introduction ..........................................132
Results .............................................. 137
Chemical Shift Differences ........................ 137
1-Methyl-4,4'-bipyridinium iodide .............. 137
Purine nucleosides ...............................140
Proton Longitudinal Relaxation Rates .............. 142
1-Methyl-4,4'-bipyridinium iodide ............... 143
Purine nucleosides ...............................147
Discussion .............................................150
Binding Environment of 1-Methyl-4,4'-
bipyridinium Iodide ............................. 151
Binding Environments of the Purine Nucleosides .... 154

8 EXPERIMENTAL .......................................... 156
Instrumentation ...................................... 156
Reagents ............................................. 156
Preparations ......................................... 157
Measurements and Calculations ......................... 183
Measurement of Equilibrium Compositions of
Aqueous Anabaseine Solutions .................... 183
Buffers ......................................... 184
1H NMR measurements ............................. 184
UV measurements ................................. 185
Computer program ................................ 185
Measurement of the Compositions of Solutions of
Anabaseine and N-Methylanabaseine in
Nonaqueous Solvent Systems ....................... 186
1H NMR measurements ................ ........... 186
Control experiments to establish that the
hydrolysis reaction is at its equilibrium
position .................................... 187
Titration of methanolic solutions of
anabaseine.HC1 and N-methylanabaseine.Cl
with D20 ..................................... 187
Control experiments to measure the equilibrium
constant for N-methylanabaseine.Cl in dry
methanol. ................................... 190







Page


Reactions of 1,1'-Disubstituted-2,3'-bipyridines
with Base ..................... .................. 191
Methoxide adduct of 1,1'-dimethyl-2,3'-
bipyridinium diiodide in 4/1 DMSO-
d6/methanol-d4 ................................ 192
Suggested methoxide adduct of 1-methyl-l'-
(2-(4-nitrophenyl)ethyl)-2,3'-bipyridinium
diiodide in methanol-d4 and elimination
products ..................................... 193
Proposed hydroxide adduct of 1'-(2-(4-
nitrophenyl)ethyl)-l-methyl-2,3'-
bipyridinium diiodide in DMSO-d6 and
subsequent elimination ....................... 195
Longitudinal Relaxation Times ..................... 195
Preparation of NMR samples .................... 195
Proton T1 measurements .......................... 198

LIST OF REFERENCES ........................................ 199

BIOGRAPHICAL SKETCH ....................................... 207













LIST OF TABLES


2-1. Observed Chemical Shifts for H6 of the Cyclic
Imine Form of Anabaseine in D20 at 22 OC and 0.6 M
Ionic Strength ................ ...................... 24

2-2. Equilibrium Constants and Chemical Shifts for the
Compounds in Figure 2-1 in D20 at 22 OC and 0.6 M
Ionic Strength ...................................... 25

2-3. Observed Chemical Shifts for H2', H4', and H5' of
the Amino Ketone Form of Anabaseine in D20 at
22 OC and 0.6 M Ionic Strength ...................... 26

3-1. Equilibrium Constants Obtained from Analysis of
Literature Data for the Hydrolysis of Cyclic
Imines .............................................. 45

4-1. Composition of Ring-Chain Equilibrium Mixtures in
Various Solvents Determined by 1H NMR at 22 OC...... 69

4-2. Chemical Shifts (6/ppm) for Amino Ketones 4-2 and
4-4 in Nonaqueous Solvents .......................... 70

4-3. Chemical Shifts (8/ppm) for Cyclic Imines 4-1 and
4-3 in Nonaqueous Solvents .......................... 71

5-1. Comparison of Chemical Shifts and Coupling
Constants for 1H NMR Spectra of 5-3 and 5-11 in
4/1 DMSO-d6/CD30D relative to TSP .................. 109

5-2. Chemical Shift Differences Between Protons of 5-3
and 5-11 in 4/1 DMSO-d6/CD30D ....................... 110

5-3. Chemical Shifts of Protonated and N-Methylated
2,3'-Bipyridinium Ions in D20 ....................... 113

6-1. Compounds Prepared by Pd(0) Catalyzed Cross-
coupling of Diethyl(3-pyridyl)borane with
Heteroaryl Halides According to eq 6-3.............. 131


vii









7-1. Chemical Shifts of 1-Methyl-4,4'-bipyridinium
Iodide in D20 with and without SDS and Ni2+ at
25 .0 C ............................................. 137

7-2. Proton Chemical Shift Differences For Solutions of
1-Methyl-4,4'-bipyridinium Iodide in Various Media
Compared with a 30 mM Aqueous Solution .............. 138

7-3. Chemical Shifts of 20 mM Adenosine Solutions in
D20 with and without SDS and Ni2+ at 25.0 oC. ....... 141

7-4. Chemical Shifts of 20 mM 2',3'-
Isopropylideneadenosine Solutions in D20 with and
without SDS and Ni2+ at 25.0 oC. ................... 141

7-5. Proton Chemical Shift Differences For Solutions of
20 mM Ado and 20 mM iAdo in 0.18 M Aqueous SDS
Compared with 20 mM Ado and 20 mM iAdo Aqueous
Solutions ................ .......................... 142

7-6. Proton Longitudinal Relaxation Rate Constants for
1-Methyl-4,4'-bipyridinium Iodide in D20 with and
without Nickel and SDS at 25.0 oC................... 144

7-7. Relative Proton Longitudinal Relaxation Rates for
1-Methyl-4,4'-bipyridinium Iodide in D20 with and
without Nickel and SDS at 25.0 oC. ................. 146

7-8. Proton Longitudinal Relaxation Rate Constants for
20 mM Adenosine Solutions in D20 with and without
Nickel and SDS at 25.0 C. ........................... 148

7-9. Proton Longitudinal Relaxation Rate Constants for
20 mM 2',3'-Isopropylideneadenosine Solutions in
D20 with and without Nickel and SDS at 25.0 oC. ..... 149

7-10. Relative Proton Relaxation Rate Constants for
20 mM Adenosine Solutions in D20 with and without
Nickel and SDS at 25.0 oC ........................... 150

7-11. Relative Proton Relaxation Rate Constants for
20 mM 2',3'-Isopropylideneadenosine Solutions in
D20 with and without Nickel and SDS at 25.0 oC...... 151


viii












LIST OF FIGURES


Figure Pacrge

1-1. Anabaseine equilibria in aqueous solution............ 4

1-2. Myosmine and analogues studied by Brdndange and
coworkers............................................ 6

1-3. Analogues of anabaseine, nicotine and
2,3'-bipyridine .................................... 10

2-1. Model for anabaseine equilibria in aqueous
solution ........................................... 15

2-2. 1H NMR of anabaseine in D20 at 22 OC and 0.6 M
ionic strength at (a) pD 6.59 and (b) pD 8.65 ....... 17

2-3. 1H NMR of anabaseine in D20 at 22 OC and 0.6 M
ionic strength at (a) pD 6.59 and (b) pD 4.11 ....... 19

2-4. A titration curve describing the hydrolysis of
anabaseine (2-1) in D20 at 22 OC and 0.6 M ionic
strength .. ..... ................ ............ .......... 22

2-5. A titration curve for anabaseine according to
Figure 2-1 based on ultraviolet absorption data
collected at 238 nm and 25 OC using H20 at 0.15 M
ionic strength ................................ ....... 30

2-6. A plot showing how each of the four components of
the anabaseine equilibrium given in Figure 2-1
varies with pD ...................................... 33

3-1. A titration curve for the hydrolysis of myosmine
(3-1) in D20 .. ...................................... 41

3-2. A titration curve for the hydrolysis of
N-methylmyosmine (3-2) in D20 ....................... 42

3-3. A titration curve for the hydrolysis of 1-methyl-
2-phenyl-l-pyrrolinium ion (3-3) in D20............. 43

3-4. Model for N-methylmyosmine (3-2) equilibria in
aqueous solution .................................... 46








3-5. Model for equilibria of l-methyl-2-phenyl-l-
pyrrolinium ion (3-3) in aqueous solution ........... 48

3-6. Titration curve for the oxidation product of
spermine and spermidine in D20....................... 50

3-7. Model the equilibria associated with aqueous
solutions of the oxidation product of spermine and
spermidine.............................. ............ 51

3-8. The same titration data as in Figure 3-6 but this
time the curve is calculated using eq 3-6 with
Klapp = 3.33 x 10-6 and K2app = 2.67 x 10-6. ......... 53

3-9. Calculated changes in the composition of aqueous
solutions of myosmine (3-1) based on the model in
2-1 as a function of pH ............................. 58

4-1. Observed ratio of concentrations of keto ammonium
ion, 4-2, to iminium ion, 4-1, as a function of
mole fraction of D20 in CD3OD at 22 OC and
uncorrected for "impurity" of water in methanol..... 75

4-2. Observed ratio of concentrations of keto ammonium
ion, 4-4, to iminium ion, 4-3, as a function of
mole fraction of D20 in CD3OD at 22 OC and
uncorrected for "impurity" of water in methanol...... 76

4-3. Linear relationship between In Kmix verses mole
fraction of D20 in CD30D at 22 OC where Kmix =
[4-2]/ ([4-1] [D20]) .. ....................... ..... .... 78

4-4. Linear relationship between In Kmix verses mole
fraction of D20 in CD30D at 22 OC where Kmix =
[4-4]/([4-3] [D20]) .. ............. ................... 79

5-1. Scheme for selective alkylation of 2,3'-bipyridine
using a removable protecting group .................. 90

5-2. Formation of a methoxide adduct of l,1'-dimethyl-
2,3'-bipyridinium ion in methanol-d4................ 108

6-1. Synthesis of 2-aryl substituted 1-piperideines ...... 124

6-2. Synthesis of an open-chain analogue of
anabaseine .......................................... 128

7-1. Hypothetical model for a spherical micelle.......... 133







Page


7-2. Possible model for binding of l-methyl-4,4'-
bipyridinium ion with SDS micelles .................. 152













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


INFLUENCE OF SOLVENT ON THE RING-CHAIN HYDROLYSIS EQUILIBRIUM
OF ANABASEINE AND SYNTHESES OF ANABASEINE AND NICOTINE
ANALOGUES

By

Linda B. Bloom

December, 1990




Chairman: John A. Zoltewicz
Major Department: Chemistry


Anabaseine (3,4,5,6-tetrahydro-2,3'-bipyridine), a

marine toxin produced by some species of Nemertines, is

hydrolyzed in aqueous solution to form an open-chain amino

ketone. This toxin, used by the worms to paralyze their

prey, is active on vertebrate cholinergic nicotinic receptors

as well as on arthropod receptors. In addition to

anabaseine, other species of nemertines produce related

pyridine alkaloids such as 2,3'-bipyridine and nemertelline,

a tetrapyridyl. These natural products and analogues are

useful model compounds for characterizing the binding sites

of nicotinic receptors.

Because of the biological significance of anabaseine and

the lack of both a qualitative and a quantitative


xii








understanding of the hydrolysis equilibrium for cyclic

imines, we performed a systematic investigation of the

hydrolysis equilibrium of anabaseine. The hydrolysis

reaction was first studied in water since biological systems

are largely aqueous. The relative concentrations of amino

ketone and cyclic imine in aqueous solutions over a pD range

of about 2-10 were measured using proton NMR. The resulting

pD-hydrolysis profile was computer fit using a model for the

equilibria to arrive at values for the hydrolysis equilibrium

constant (KH) and dissociation constants (Ka) for the basic

nitrogen atoms. We demonstrated the general utility of this

approach by calculating equilibrium constants from data for

other pH-dependent equilibria which appeared in the

literature.

Because binding sites may be located in hydrophobic

pockets of receptors, the compositions of equilibrium

mixtures of anabaseine and of N-methylanabaseine (1-methyl-

3,4,5,6-tetrahydro-2,3'-bipyridinium ion) were measured using

proton NMR in nonaqueous systems (CD30D and DMSO-d6). A

general survey of solvent effects is presented along with a

quantitative study in water-methanol and water-DMSO mixtures.

The compositions of solutions of anabaseine and of N-

methylanabaseine in water-methanol mixtures can be described

by linear free energy relationships.

Analogues of nicotine, a known agonist on nicotinic

receptors, anabaseine, and 2,3'-bipyridine were prepared for

biological testing.


xiii








Finally, methods which could be used to investigate the

nature of binding environments were explored using a micellar

model system. Proton chemical shifts and longitudinal

relaxation times were used to probe the binding environment

of substrates in SDS micelles.


xiv












CHAPTER 1
INTRODUCTION


General Overview


Anabaseine (1-1, 3,4,5,6-tetrahydro-2,3'-bipyridine) is

present in the venom of some species of nemertines, a phylum

of carnivorous marine worms [1 3]. The venom is used to

paralyze the worms' prey and possibly to repel predators.

Anabaseine is of pharmacological interest because of its

activity as an agonist on cholinergic nicotinic receptors in

the central and peripheral nervous systems. Its activity is

comparable to that of nicotine.

The structure of anabaseine consists of a 1-piperideine

ring attached at its 2-position to the 3-position of a

pyridine ring. Imines are water labile and can be hydrolyzed

to give an amine and ketone or aldehyde. In aqueous solution

anabaseine is hydrolyzed to give an open-chain amino ketone

which is in equilibrium with the cyclic imine (eq 1-1) [4].

Both of these structures contain basic nitrogen sites. Thus,

species which vary in protonation state may be present in an

equilibrium mixture also.




S- N 0 NH.. (1-1)








The possibility of several forms of anabaseine existing

in aqueous solutions raises many interesting questions

concerning its biological activity. Which form is

responsible for nicotinic activity? Why is anabaseine not

toxic to the worm? Perhaps, the worm stores the toxin in an

inactive form and then changes the medium to convert the

toxin to an active form when it stings its prey [3].

Anabaseine is also of biological significance due to its

activity on nicotinic receptors. Various types of nicotinic

receptors which respond differently to known agonists and

antagonists exist within the many parts of the central and

peripheral nervous systems. While the gross structures of

some of these receptors are known, their binding sites have

not been well characterized. Anabaseine and other

structurally related model compounds can be used to help

define the nature of the binding sites of nicotinic

receptors.

While the hydrolysis of acyclic imines has been

extensively studied [5 7], comparatively little work has

been done on the hydrolysis equilibria of cyclic imines.

Because of the biological significance of anabaseine and the

lack of both a qualitative and a quantitative understanding

of the hydrolysis equilibria for cyclic imines in general, we

undertook a systematic investigation of the hydrolysis

equilibrium of anabaseine. Our study began with a

quantitative determination of the equilibrium compositions of

aqueous solutions of anabaseine from which we were able to








derive an equilibrium constant for hydrolysis (KH) and pKa

values. This was followed by a general survey of solvent

effects on the position of the equilibrium and finally a

quantitative study of the system in mixed solvents which

contained water. In addition to the equilibrium studies,

model compounds were prepared for biological testing. These

model compounds were designed with two goals in mind: (1) to

help determine which form of anabaseine was biologically

active and (2) to aid in the characterization of the binding

sites of nicotinic receptors and the determination of those

structural features that would render a molecule active.

Questions concerning the environment of the binding site at

nicotinic receptors sparked our interest in methods that

could be used to investigate binding environments. In order

to explore the possibilities, model studies were done on the

binding of substrates to micelles. We used proton NMR

techniques to gain information about the environments of

different substrates when bound to SDS micelles.


Brief Summary of Investigations by Chapter


Because biological systems are largely aqueous, the

hydrolysis reaction of anabaseine was first investigated in

water to determine how the equilibrium composition changed

with pH. Both proton NMR and UV measurements were made on

aqueous solutions of anabaseine. The results of these

measurements appear in Chapter 2. The NMR measurements over








a pD range of about 2 to 10 showed that anabaseine exists as
a mixture of four possible species depending on the solution
pD: (1) cyclic imine free base (i-i), (2) cyclic iminium ion

(1-2), (3) monoprotonated amino ketone (1-3) where only the

amino group is protonated, and (4) diprotonated amino ketone
(1-4) where both the pyridine and amino nitrogen atoms were

protonated (Figure 1-1).



KH
H20 + .N + NH3

2 N 1



Kai Ka2



N OO NH3
N
N H 1-4

Figure 1-1. Anabaseine equilibria in aqueous solution.

These NMR measurements made possible a quantitative

determination of the effects of pD on the equilibrium

concentrations of each component in heavy water. The NMR

data describing the fraction of cyclic imine over the pH

range of 2-10 were computer simulated using the model for the

equilibria in Figure 1-1 and iterative nonlinear regression

analysis [4]. The computer fit of the data yielded a




5



hydrolysis equilibrium constant (KH) and pKa values for

anabaseine.

Because each UV spectrum is the sum of the spectra of

all the species present in equilibrium, the UV spectra do not

contain as much information as the NMR spectra. Both the

molar absorptivities of each species and the amount of each

present are unknown; this makes analysis more complicated.

The results obtained from UV measurements supported the

results obtained from NMR measurements although a unique fit

of the data was not obtained using the model in Figure 1-1

and the nonlinear regression program.

The method used to simulate the pD-hydrolysis profile

for anabaseine and to obtain equilibrium constants is

general. The utility of this method was demonstrated by

applying it to some other examples of pH-dependent equilibria

which appeared in the literature.

Brandange and coworkers presented data for the pH-

dependent hydrolysis of myosmine, a minor tobacco alkaloid,

and other structurally related imines (Figure 1-2) which were

based on NMR measurements in aqueous solutions [8, 9].

Myosmine is structurally very similar to anabaseine but

contains a 5-membered 1-pyrroline ring in place of the 1-

piperideine ring of anabaseine. A quantitative analysis of

equilibrium constants for these compounds was not given by

Brandange, however. We devised models for these equilibria

and used our computer simulation to fit the data and extract

the appropriate equilibrium constants [4].













Myosmine


Figure 1-2. Myosmine and analogues studied by Brandange and
coworkers.


The computer simulation of pH-dependent equilibria is

not limited to the hydrolysis of cyclic imines. NMR

measurements in aqueous solution were reported in the

literature for a pH-dependent tautomeric equilibrium (eq 1-2)

between a protonated bicyclic aminal, octahydropyrrolo[1,2-

a]pyrimidinium ion (1-5), and an open-chain monocyclic amino

pyrrolinium ion, 3-aminopropyl-l-pyrrolinium ion (1-6) [10].

We were also able to fit these data by computer and derive

equilibrium constants [11]. The results of these fits appear

in Chapter 3.




N ND, N ND2 (1-2)
KT






Our studies of the anabaseine equilibrium also included

nonaqueous systems. Although biological systems are largely

aqueous, the binding site at nicotinic receptors for

anabaseine and other agonists may be located in a hydrophobic

pocket of the proteins that form the receptor.








A hydrophobic environment at the receptor site may have

different effects on the hydrolysis equilibrium for

anabaseine than are observed in aqueous solution. For this

reason, the influence of nonaqueous systems on the

equilibrium composition of anabaseine was investigated.

N-Methylanabaseine, a derivative of anabaseine which

possesses a methyl substituent on the imine nitrogen, was

prepared so that comparisons could be made between the

effects of the substituent on the position of the equilibrium

and the biological activity. Because data on biological

activity were obtained by others, such information is not

reported here. The compositions of anabaseine and N-

methylanabaseine solutions in DMSO-d6 and in methanol-d4 were

measured using proton NMR and compared with those in D20

[12]. These solutions contained at least one equivalent of

water so that a hydrolysis equilibrium could be established.

Results of these measurements are reported in Chapter 4. The

amounts of amino ketone and cyclic imine observed in the

nonaqueous systems did not change dramatically from those

observed in aqueous systems in spite of the fact that the

concentration of water present was dramatically decreased.

The equilibrium compositions of anabaseine and

N-methylanabaseine were measured in water-DMSO and water-

methanol mixtures also. As the amount of water was increased

in the mixed solvents, the equilibria shifted to favor ketone

as expected from mass action considerations. The equilibrium

compositions of both compounds in methanol-water mixtures can









be quantitatively described by a linear free energy

relationship. These results are discussed in Chapter 4.

The methyl substituent of N-methylanabaseine shifts the

position of the equilibrium, largely for steric reasons, so

that more of the amino ketone is present in these aqueous

solutions than in solutions of anabaseine. These results

stand in marked contrast to those pertaining to the effect of

methyl substitution on the position of equilibrium of

myosmine and its N-methyl derivative. Here methyl

substitution leads to a decrease in the amount of amino

ketone.

In addition to anabaseine, other species of nemertine

worms produce other alkaloids. Two such compounds are 2,3'-

bipyridine and nemertelline, a tetrapyridine [13]. Another

compound which has not yet been identified gives a mass

spectrum which is consistent with a methyl substituted

bipyridine [Kem, unpublished results]. While 2,3'-bipyridine

is not very active in vertebrates, it is quite active in

crustaceans. Nemertelline does not appear to be active in

either vertebrates or arthropods and may have some other

function such as a chemical attractant or repellant. A

general method for the preparation of bi- and polypyridines

was desired both for identification of natural products and

to produce greater quantities of these compounds for

biological testing.

Many methods of coupling aryl and alkenyl halides with

aryl stannanes, boranes and boronic acids using a Pd (0)








catalyst have been reported [14 19]. Palladium catalyzed

cross-coupling of pyridyl halides with pyridyl boranes was

used to prepare substituted bipyridines and this approach is

discussed in Chapter 6. Bipyridines with different

substituents can easily be prepared using this method

allowing comparisons to be made between substitution patterns

and biological activity.


NI N
N





K N

N N

2,3'-Bipyridine Nemertelline


The pharmacological properties of these natural products

and derivatives render them useful for the study of nicotinic

receptors. Comparisons of biological activities with

differences in the structures and the properties of these

compounds will aid in the characterization of the nature of

the binding sites of nicotinic receptors within different

parts of the nervous system. Characterization of arthropod

receptors may aid in the development of insecticides and

other forms of pest control. With this in mind, structural








analogs of anabaseine, nicotine, and 2,3'-bipyridine were

prepared. These compounds are shown in Figure 1-3 and their

syntheses are discussed in Chapters 5 and 6.


*NH(CH3)2


Dimethylamino Analogue of Anabaseine


N-Methylanabaseine


1 -Methyl-2,3'-bipyridinium Ion


(S)-N-Methylanabasine Monocation


2-(3-Pyddyl)pyrimidine


5-Methyl-2,3'-bipyhdine


6-Methyl-2,3'-bipyridine


3-Chloro-2,3'-bipyndine


Figure 1-3. Analogues of anabaseine, nicotine and 2,3'-
bipyridine.


In the course of preparing l-methyl-2,3'-bipyridinium

ion, interesting differences in the proton NMR spectra of

mono- and di-N-methylated 2,3'-bipyridines were observed.

These differences provide information about the angles








between the planes of the rings in these bipyridinium ions.

The chemistry of some diquaternized 2,3'-bipyridines also

proved interesting. Meisenheimer-type 0 complexes were

observed on reaction of methoxide ion with these diquaternary

salts. These results and the synthesis of 1-methyl-2,3'-

bipyridinium ion are presented in Chapter 5.

In our studies of anabaseine, we became interested in

methods that could be used to determine the type of

environment of a binding site. Because biological systems

are so complex, we explored some possibilities in model

systems as a first step. We chose to investigate the

environment of compounds which are solubilized by micelles.

Micelles serve as primitive models for membranes and lipid

bilayers and are composed of surfactants which have a polar

head group and a nonpolar tail [20, 21]. In aqueous

solution, micelles are formed when surfactants aggregate so

that the hydrophobic portions of the molecules interact with

each other and the hydrophilic portions of the molecule

interact with water. Aqueous micellar solutions are able to

solubilize hydrophobic molecules. For this reason, micelles

are useful in reaction media and for many other applications.

Although micelles have been extensively studied many

questions remained to be answered. There is still much

disagreement over their exact structure, the degree of water

penetration, and the structure of substrates when bound

within micelles [20, 22 26]. We used proton NMR techniques

to investigate the environment of substrates when they bind




12



to micelles. The substrates studied include adenosine,

2',3'-isopropylidene adenosine, and 1-methyl-4,4'-

bipyridinium iodide. Micellar solutions of sodium

dodecylsulfate (SDS) were used because the properties of SDS

have been well characterized. The results of these studies

appear in Chapter 7.

Finally, a detailed description of all experimental

procedures is given in Chapter 8.












CHAPTER 2
QUANTITATIVE DETERMINATION OF THE RING-CHAIN HYDROLYSIS
EQUILIBRIUM CONSTANT FOR ANABASEINE


Introduction


As discussed in Chapter 1, the structure of anabaseine

consists of a 1-piperideine ring attached at the 2-position

to a 3-pyridyl ring. The cyclic imine is water labile and is

hydrolyzed in aqueous solution to produce an open-chain amino

ketone.

The hydrolysis reaction in aqueous solution has been

investigated using proton NMR and UV spectroscopy. The

components present in aqueous solution over a pH range of

2-10 have been identified and for the first time the various

equilibria associated with the hydrolysis have been described

quantitatively [4]. Figure 2-1 shows the components (2-1 -

2-4) which exist in equilibrium; the relative amounts of

each are dependent on pH.

Synthetic anabaseine used in this study was synthesized1

by a previously reported method. In Chapter 6, an improved

synthesis is discussed.








1 Prepared by Dr. William R. Kem.









Results



NMR Measurements of Aqueous Anabaseine Solutions


The hydrolytic equilibrium for anabaseine in D20 was

examined with the aid of proton NMR at 22 1 OC at an ionic

strength of 0.6 0.08 M. The hydrolysis reaction was

complete in the 5-10 minutes it took to prepare the sample

and record the spectrum. Found were the cyclic imine 2-1 and

its conjugate acid 2-2 and the hydrolysis product, the open-

chain amino ketone 2-3 as well as its conjugate acid 2-4

(Figure 2-1). Examples of spectra at high, neutral and low

pD's appear in Figures 2-2 and 2-3. These spectra illustrate

the changes in the ratios of the two forms as evidenced by

the changes in relative intensities of the peaks and the

changes in the protonation states as evidenced by the changes

in chemical shifts.

The samples were stable; after standing in a

refrigerator for 9 months there was no evidence of change in

the NMR spectrum. Some yellowing did occur, however. While

the tetrahydropyridine ring in 2-1 may undergo dimerization

and cyclic trimerization [27, 28], our solutions evidently

were too dilute (0.008-0.018 M) for this to be an important

side-reaction.










D20 + N N H IND
DD0 + 0 ND3


2-2 2-3
N N


Kai Ka2

4 3
3 5 2 4
S4'5 +

6'2'
SN N
62 D6 2-4
N D+



Figure 2-1. Model for anabaseine equilibria in aqueous
solution.


Assignment of structures and composition in aqueous solutions

Both the high and low field portions of the spectra were

used to assign structures. The high field NMR signals were

attributed to cyclic and open-chain compounds using the

chemical shifts of the methylene protons next to the nitrogen

atom. Assignments rest on (a) the relative chemical shifts

of these signals and (b) the influence of pD on their

positions. For equivalent states of protonation, an sp2

hybridized nitrogen atom is expected to be more deshielding

than one that is sp3 hybridized. Moreover, the amino group

of the open-chain ketone of 2-3 is likely to remain largely

as its conjugate acid because of its basicity and hence be

invariant in shift over the pD range of our studies (2-10).

But the position of the signal due to the less basic imine




















4




4




4-)


4-
U)

u,
-d















0
-4





C4






0







0
(N

-H











-H








U)







0
LO




















0-4







44
-H.





17








































tn L








w CD








F N C"



















rr .0












0)




04








4-)



4-)




4-)
0
ci













C;









0.4
04
4-





0
0









(C











U)
0




4-4

0






04I
-H



-H

(1

(0
(0




4-4



0







1-Q

-H.






19





















a n m










































co 13DD
in i n
















Nd N








2-1 will change as the pD of the medium is varied and the

ratio of base and its conjugate acid changes. Therefore, the

signal at 8 3.07 0.01 which remained constant was assigned

to the open chain amino ketone and that which varied from

6 3.72 to 3.89 to the cyclic imine (Figure 2-2). The ratio

of the two forms was easily obtained by integration of these

signals. Other signals associated with these two forms can

also be assigned based on their relative intensities.

Shifts due to the CH2CH2 unit of the open and ring-

closed forms overlapped and could not be used to determine

structures. The protons next to the carbonyl group and

adjacent to the imine carbon were replaced by deuterium. For

example, at high pD little of these signals remained by the

time the first spectrum was obtained. Even an acidic sample

(pD 3.64), where isotope exchange was much slower,

demonstrated deuterium incorporation over a period of a day

at room temperature.

These structural assignments also are supported by a

consideration of the chemical shifts at low field associated

with the pyridine ring. For example, the signal for H-2 of

the pyridine ring of one component (imine) moves to lower

field (8 8.75-8.95) over the pD range 9.57 to 6.59 and then

remains constant. However, H-2 for the other component

(ketone) was essentially constant at 8 9.10 from pD 9.57 to

6.24 and then gradually moved to lower field as the solution

was made more acidic, reaching a final value of 6 9.34 at pD

2.20 (Figures 2-2 and 2-3).








The variation in the shifts of the pyridine protons at

high pD can be attributed to changes in protonation of the

cyclic imine while movement of the signals at low pD can be

associated with changes in protonation of the pyridine of the

acyclic ketone. The imine is more basic than the pyridine

and its protonation will influence the chemical shifts of the

attached pyridine ring. Since the nitrogen atom in the

pyridine ring of the ketone is less basic than the imine, it

will only undergo protonation and associated deshielding in

more acidic media. The pyridine ring of the imine is likely

to be less basic than that of the ketone due to the proximity

and electron withdrawing effects of the conjugated iminium

ion. Significant protonation of this ring is not observed.

Figure 2-4 graphically reports our findings concerning the

influence of solution acidity on the equilibrium composition

of anabaseine during hydrolysis. The data are given in terms

of the fraction of the total amount of substrate present as

cyclic imine rather than as a ratio of two forms. This

fraction varies from 0 to 1 and therefore is more easily

understood. At high acidity (pD 2) nearly all the substrate

is present in the open-chain form. As the solution is made

more basic, the concentration of the imine increases and the

amino ketone decreases. From about pD 5 to 7, the

composition of the mixture is approximately constant. There

is a sharp increase in the amount of cyclic imine as the

solution is made still more alkaline until finally only

cyclic imine is present at about pD 9.5.






















CMj
CU



0 6
0 0.6

aC


O

0*

0.4
0
ca 0
o


Cu
L.
LL

0.2








0.0- ,* I,
3 5 7 9 11



pD




Figure 2-4. A titration curve describing the hydrolysis of
anabaseine (2-1) in D20 at 22 oC and 0.6 M ionic strength. The
dark circles are the experimental values expressing the
fractional amount of the total mixture existing as 2-1 and 2-2.
The solid line was calculated using eq 2-3 and the values for
the equilibrium constants in Table 2-2.








Calculation of pK_ values from pD dependent chemical shifts

Those proton signals that show chemical shift changes

with pD provide information about pKa values for the basic

nitrogen sites. The observed chemical shift (Bobs) at any pD

is the population weighted average of the chemical shifts of

the unprotonated (IN) and protonated (6NH) forms which can be

expressed by eq 2-1.


Sobs = 8N X K+ + NH X [D+] (2-1)
Ka + [D+] Ka + [D+]



The fractional amount of free base is given by the expression

Ka/(Ka + [D+]) and the fraction of protonated substrate by

[D+]/(Ka + [D+]).

Values of the dissociation constant (Ka) for the

conjugate acid, 2-2, of the imine nitrogen of 2-1 (Kal) and

the chemical shifts 8N and INH were obtained by a nonlinear,

iterative regression analysis using eq 2-1 to fit the

observed chemical shifts associated with the signals of the

methylene protons a to the nitrogen atom. Observed chemical

shifts are given in Table 2-1 and values obtained from the

computer fit are recorded in Table 2-2.

Similarly, the observed chemical shifts of each of the

pyridine protons in the open-chain amino ketone is a weighted

average of the chemical shifts of the protonated and

unprotonated forms. From a consideration of the pKa values

of model compounds such as 2,3'-bipyridine (pKa's 1.5, 4.4) [29]








Table 2-1. Observed Chemical Shifts for H6 of the Cyclic
Imine Form of Anabaseine in D20 at 22 oC and 0.6 M Ionic
Strength.



pD Chemical Shift / ppma

8.65 3.738

8.20 3.769

7.93 3.819

7.77 3.804

7.07 3.868

6.88 3.883

6.59 3.893


aTSP was used as an internal standard.


we assume that protonation of the pyridine ring conjugated

with the positively charged iminium ion will not be important

over the pD range where the pyridine ring of the ketone is

protonated. Therefore, the dissociation constant of the

pyridinium ion in the cyclic imine was neglected and the

changes in the chemical shifts of the pyridine protons at low

pD were assigned to the open-chain ketone 2-4. Its Ka2 was

calculated by the nonlinear regression technique using the

data for the H2', H4' and H5' protons separately and eq 2-1.

Observed shifts for the pyridine protons appear in Table 2-3

and results from the computer fit are given in Table 2-2 and

are within the usual agreement found for an NMR method [30].

Our pKa value of 3.36 (3.86-0.50) corrected for the solvent






































0
4J-







Cd
(4






I-

C)

Q)c






4-o
C
a) ca
t04-4 0



S0 0T


a) r-4
>Od

> 0 m-
-1 4-) >


Ca) 4
CU)

r0 0

>0a)



-' t >
o u <
Z 10







Table 2-3. Observed Chemical Shifts for H2', H4', and H5' of
the Amino Ketone Form of Anabaseine in D20 at 22 C and 0.6 M
Ionic Strength.


Chemical Shifts / ppma
pD H2' H4' H5'

4.47 9.177 8.610 7.810

4.37 9.171 8.623 7.818

4.11 9.199 8.705 7.889

3.64 9.275 overlap 8.062

3.0 9.312 9.044 8.183

2.2 9.342 9.120 8.248


aTSP was used as an internal standard.


isotope effect (0.50) is not unlike that for 3-acetylpyridine

(pKa 3.2, H20 [31]), a reasonable model.

The pD-hydrolysis profile and the hydrolysis equilibrium
constant KE

That the system is rapidly reversible and hence at

equilibrium was demonstrated by taking a reaction mixture,

changing its pD and redetermining the substrate ratio. Thus,

following observation of the hydrolysis ratio of a sample

having pD 4.37, the pD was increased to 9.22 by the addition

of carbonate ion and the new ratio of open-chain to cyclic

substrate was determined two hours later. All the substrate

now existed as cyclic imine as was found for a fresh sample

of similar pD made directly from anabaseine. Similarly, a

change was made in the opposite direction; a spectrum was

recorded immediately after an alkaline sample was acidified








from pD 8.20 to 4.82. Again the ratio was the same as that

obtained on a fresh sample of similar pD.

The hydrolysis equilibrium is defined by the reaction

2-2 + D20 -> 2-3 where KH = [2-3 ]/[2-2]. However, NMR area

measurements provide the total amounts of each of the two

components, i.e., (2-3 + 2-4) and (2-1 + 2-2) and not just

2-3 and 2-2 alone. Therefore, the sum must be corrected for

the amount of the unwanted component to obtain the value for

KH. This is accomplished easily by eq 2-2 with the aid of

the appropriate Ka values along with the concentration of

acid obtained from pD measurements. The fraction of amino

ketone where the pyridine is unprotonated is given by

Ka2/([D+]+Ka2) the fraction of imine where the imine nitrogen

is protonated is given by [D+]/([D+]+Kai)


[open-chain]tot Ka2/([D+] + Ka2)
KH [ring-closed]tot [D+]/([D+] + Kal) (2-2)



Prior to computer fitting, eq 2-2 was rewritten to express

the fractional amount of total substrate present as the

cyclic imine (free base 2-1 and its conjugate acid 2-2)

rather than as the ratio of [2-31/[2-21 (eq 2-3).


fraction of Ka2(Kal + [D+])
cyclic imine Ka2(Kal + [D+]) + KH[D+](Ka2 + [D+]) (2-3)



The latter fraction (eq 2-3) ranges in value from 0 to 1

whereas the other tends to a very large value at low pD owing








to the small amount of 2-2 thereby causing computational

problems. Equation 2-3 was then solved with the aid of a

microcomputer program using nonlinear regression analysis.

From the data in Figure 2-4 it was possible to obtain the

three relevant equilibrium constants Kal, Ka2 and KH to

describe quantitatively the changes in the concentrations of

the four species 2-1 2-4 as a function of pD. Table 2-2

lists the calculated values and Figure 2-4 shows the

calculated curve through the data points.

Worthwhile is a comparison of the Ka values obtained by

using the variation in chemical shifts, eq 2-1, and by the

overall fitting technique based on eqs 2-2 and 2-3. There is

satisfactory agreement, Table 2-2. The values for Kal overlap

within one standard deviation for the two approaches while

the value for Ka2 derived from a consideration of all the data

in Figure 2-4 is 25% smaller than the average found by the

chemical shift method, reasonable agreement for NMR analysis.

Our results are self-consistent.

The value of 1.21 for KH shows that almost equal amounts

of the two monocationic species 2-2 and 2-3 are present with

the open-chain ketone being favored slightly.


Ultraviolet Spectra of Aqueous Anabaseine Solutions


Another approach using ultraviolet absorption spectra

was taken to examine the hydrolysis reaction. Spectral

changes were used to follow the reaction, this time in H20 at









25 OC and at a lower salt concentration, 0.15 M. Again

spectra were taken as a function of solution acidity and 238

nm was selected as the analytical wave length because of the

large and complex absorbance changes. Figure 2-5 shows how

the absorption varies with acidity over the pH interval 1 to

12.

Attempts to obtain equilibrium constants were more

difficult and unsatisfactory because unlike the NMR method

individual species could not be observed. Our nonlinear

regression program was employed to estimate constants from eq

2-4 where [2-1 2-4] denotes the total amount of substrate,

Figure 2-1, represents the molar absorptivity, and F

indicates the fraction of the total amount of substrate

present in solution as one of the four forms.



ABSobs =

(E2-1F2-1 + E2-2F2-2 + E2-3F2-3 + E2-4F2-4)- [2-1 2-4] (2-4)


F1 KalKa2
KalKa2 + Ka2[H+] + KHKa2[H+] + KH[H+]2



Equation 2-5 serves as an example of how this fraction may be

expressed in terms of the appropriate equilibrium constants

and the [H+]. The fractional amount of the total substrate

present as 2-1 is indicated. The other fractions may be

written by taking the same denominator and replacing the

























E *--






O
C

CO

01



W 0.4
-Q







0.2







0.0i -i I i i
0 2 4 6 8 10 12


pH




Figure 2-5. A titration curve for anabaseine according to
Figure 2-1 based on ultraviolet absorption data collected at
238 nm and 25 oC using H20 at 0.15 M ionic strength. The
filled circles are experimental absorbances and the curve was
calculated using eq 2-4 and 2-5 and the constants in Table
2-2.








numerator with one of the three remaining terms in the

denominator. Thus, the second term when placed in the

numerator gives F2-2, the third F2-3, and the fourth F2-4.

Additional unknowns, the molar absorptivities or values, are

not present in the NMR equation but are found in eq 2-4.

Estimates of these were made as follows. The value at high

pH (10-12) was assumed to be due to free imine, 2-1, and that

at low pH (1.1) dication, 2-4. The NMR ratios near

neutrality where both 2-2 and 2-3 exist almost exclusively

were considered in an attempt to dissect the observed

absorbance and obtain estimated E values for 2-2 and 2-3. A

unique fit to eq 2-4 was not found because of the uncertainty

in these values. For example, when the initial estimates for

2-2 and 2-3 were varied by about a factor of two from those

reported in the Experimental Section, the value of KH had a

very large uncertainty and so these estimates were rejected.

The final values seem to be reasonable. The points in Figure

2-4 are the experimental values and the line is that

calculated by the equilibrium constants given in Table 2-2,

the molar absorptivities in the Experimental Section and eq

2-4, showing a satisfactory fit.









Discussion



Anabaseine Equilibria


The rapidly equilibrating set of four components,

Figure 2-1, resulting from the hydrolysis of anabaseine is

described quantitatively for the first time. The titration

curve, Figure 2-4, obtained for this complex mixture may be

expressed in another way, Figure 2-6, that shows how the

concentrations of each one of the four components varies with

solution acidity. Figure 2-6 was constructed using [D+] and

fractions defined by eq 2-5 and its counterparts where the

numerator for each fraction changes as explained in the

Results. Open-chain dication 2-4 predominates at low pD and

cyclic free base 2-1 is the major component at high pD. Over

a wide acidity range near neutrality and under physiological

conditions the two monocations 2-2 and 2-3 are essentially

the only two materials present in solution, acyclic ketone

slightly predominating. At pD 7 for example, there is 8% of

2-1, 42% of 2-2 and 50% of 2-3. This composition will be

different in H20. Making the reasonable assumptions that KH

will not change significantly but that the two relevant pKa

values will decrease by 0.5 due to the solvent isotope effect

gives rise to a new equilibrium distribution: 2-1, 21%, 2--2,

36% and 2-3, 43%. There is a major increase in the amount of

2-1 on going to H20.








1.0




0.8




0.6

0
2-


S0.4- 2-2
0



0.2




0.0 -
2 3 4 5 6 7 8 9 10
pD

Figure 2-6. A plot showing how each of the four components
of the anabaseine equilibrium given in Figure 2-1 varies with
pD.


The KH term describing the hydrolysis equilibrium in

terms of the two monocationic constituents may be written in

an alternate form, as an equilibrium between the cyclic free

base 2-1 and the open-chain free base 2-5, where 2-5 is the

conjugate base of acid 2-3 (eq 2-6). This new constant is

given by the expression Kh = (KH x Ka3)/Kal where Ka3 equals

[D+] [2-5]/[2-3]. Using 10.5 as an estimate of pKa3 (a

reasonable estimate based on calculated values for similar

amines in Chapter 3), the value for the alkyl ammonium ion,

provides a value of 3 x 10-3 for Kh, indicating that only 0.3%







of 2-5 exists in the presence of 2-1 at high pD. Although Kh

expresses the hydrolytic equilibrium in terms of neutral
rather than charged substrates, we prefer KH. As Figure 2-6

shows, KH pertains to the predominant constituents of the

equilibrium under most acidity conditions whereas Kh does

not.




S N K N1 0 ND2

N N


I IKa (2-6)



| .Y 0 ND3
N



The pKa values derived from samples in D20 and in H20,

Table 2-2, differ by 0.5 to 1. Neglecting the small
variation in temperature between our two studies, pKa values

for pyridinium ions in heavy water are expected to be some

0.4 to 0.5 higher [32, 33], reflecting the greater acidity of
D30+ over H30+. The difference in the ionic strength (0.15 vs

0.6 M) also leads to a small increase in pKa value. A large

displacement in the position of the titration curve along the

axis expressing solution acidity for a related substance,

myosmine, which contains a 1-pyrroline in place of the








1-piperideine ring of anabaseine, has also been reported when

the solvent was changed from heavy to light water. The

hydrolysis equilibrium for myosmine is considered in Chapter

3. The main result of the spectrophotometric analysis is

confirmation of the value of KH.


Apparent pK. Value


A pKa value of 6.7 already has been reported for

anabaseine based on a titration with aqueous HC1 in H20

containing 5% methylcellosolve at 25 OC [34]. Our work shows

that this value cannot be for the pure substance but rather

for the mixture given in Figure 2-1. That is, during the

titration hydrolysis of 2-1 occurs and this produces the more

basic open-chain, aliphatic amine 2-5. The Ka value is

skewed to include the more basic substance that will be

present as its conjugate acid 2-3. The reported dissociation

constant is an apparent Ka value (Kapp) given by Kapp = [H +]

[2-1]/([2-2] + [2-3]) and not Kal as desired.2 This apparent

dissociation constant is related to the true dissociation

constant Kai as follows: Kapp = Kai/(l + KH) which on a

logarithmic scale is pKapp = pKal + 0.34, including the

2 Careful consideration of the relevant equilibria lead to a
more complex equation. At high pD, Kapp = [D] ([2-1] + [2-5])
/([2-2] + [2-3]) that may be transformed to Kapp =
Kai(1/(1+KH)) + Ka3(KH/(1+KH)). The latter equation has the
same form as that of NMR eq 2-1: in both cases the observed
value is a population weighted average of two constants, in
one case the limiting chemical shifts and in the other the Ka
values. Because the amount of 2-5 is so small under our
conditions, the second term of the Kapp equation may
confidently be neglected.








logarithmic value for (1 + KH). In this case the correction

for the presence of the amine component is not very large

because KH is not very large. Moreover, this analysis is

supported by our titration curve, Figure 2-4, where careful

examination of the inflection point at high pD shows that

this point does not have the value for pKal of 7.8 but rather

for PKapp of 8.1 as expected. Correcting our pKa value of

7.8 for solvent isotope and salt effects by as much as 0.7

gives a value of 7.1 for H20. This is only in fair agreement

with the corrected value of 6.4 that previous workers

obtained by titration.

While our assumption concerning the absence of

protonation of the pyridine ring of the cyclic imine is not

strictly correct, little error should be introduced into our

value of the Ka for the pyridine ring of the ketone because

the degree of protonation is likely to be small. Inclusion

of diprotonated imine, 2-6, would give an apparent PKapp =

[D+] ([2-2]+[2-3])/([2-4j]+[2-6]) If [2-6] is small relative

to [2-4], it can be neglected to give Kapp = [D+] ([2-2]+[2-3])

/[2-4] which is equal to Kapp = Ka2(1+KH)/KH. Converting to a

logarithmic scale gives pKapp = pKa2 + 0.26. Substituting in

our calculated value of 3.99 for PKa2 gives pKapp = 3.7 which

nicely reproduces the inflection point on our titration curve

at low pD, Figure 2-4, showing that the contribution from 2-6

can confidently be neglected. Moreover, our data at very low

pD are sparce and attempts to include the additional Ka value









into the computer fit had no effect on the values for the

other constants.


Conclusions


The composition of aqueous solutions of 2-1 is described

quantitatively for the first time. The state of protonation

as well as the degree of conversion to the open-chain

constituents is presented in Figures 2-4 and 2-6 as a

function of the acidity of aqueous solutions. These findings

make it possible to develop a strategy to ascertain the

active form responsible for the pharmacological action of

2-1.












CHAPTER 3
COMPUTER SIMULATION OF pH-DEPENDENT RING-CHAIN EQUILIBRIA FOR
OTHER COMPOUNDS OF BIOLOGICAL INTEREST


Introduction


Encouraged by our success with anabaseine, pH-dependent

ring-chain equilibria for other biologically significant

molecules were simulated using the appropriate models with

the iterative nonlinear regression program. Other compounds

studied include the tobacco alkaloid myosmine (3-1)1 [8, 35]

which contains a 5-memebered 1-pyrroline ring instead of the

1-piperideine ring of anabaseine, its cationic N-methylated

derivative 3-2, and a derivative containing an N-methylated

pyrrolinium ring and a phenyl ring in place of the 3-pyridyl

ring 3-3. In aqueous solutions, myosmine and its derivatives

are also hydrolyzed to produce mixtures of cyclic imine and

open-chain amino ketone forms. The pD-hydrolysis profiles

for these compounds have been reported in the literature and

are constructed based on 1H NMR data [8, 9]. No attempt was

made previously to describe quantitatively these titration

curves and calculate equilibrium constants. These data were

used along with the nonlinear regression program and models



1 The hydrolysis product of myosmine is named poikiline [35].
Myosmine is also known as 3-(3,4-dihydro-2i-pyrrol-5-yl)-
pyridine, 3-(l-pyrrolin-2-yl)pyridine, and as 2-(3-pyridyl)-
1-pyrroline.









for the hydrolysis equilibria to obtain Ka and KH values for

the species involved [4].

A fourth compound studied was the oxidation product of

spermine and spermidine. Spermine2 and spermidine3 are two of

several aliphatic polyamines important in the control of

proliferative processes [36]. They may be oxidatively

degraded either chemically or enzymatically [37]. The

structure of the product from the enzymatic reaction depends

on both the nature of the polyamine and the enzyme [38], in

some instances being the same as that from the nonenzymatic

reaction [37]. One of these products has a structure that

has been controversial. It formally arises from the

cyclization of a diamino aldehyde. The early assignment of a

monocyclic 2-pyrroline or enamine structure 3-5 provides a

compound with properties that are not entirely consistent

with some of those reported for the oxidation product and so

3-6 was proposed [39]. The latter is a racemic, bicyclic

aminal having been formed by the spontaneous cyclization of

the same diamino aldehyde precursor.

More recently, a family of rapidly interconverting

structures including 3-6, 3-7 and 3-8 has been suggested for

the oxidized material based on a careful consideration of 1H

and 13C NMR spectra of a synthetic sample in a series of

aqueous solutions of varying acidity (Figure 3-7) [10].

Bicyclic ion 3-7 consists of two rapidly interconverting


2 N,N'-Bis(3-aminopropyl)-1,4-butanediamine.
3 N-(3-Aminopropyl)-1,4-butanediamine.








conjugate acids of secondary 3-7a and tertiary 3-7b amines.

Its monocyclic conjugate acid 3-8 is a 3-ammoniopropyl-l-

pyrrolinium dication. This system includes a tautomeric

ring-chain equilibrium between 3-7a and 3-9 rather than a

hydrolytic equilibrium as in the previous systems. The pH-

dependent ring-chain equilibrium was simulated using the

model in Figure 3-7 and the published NMR data to obtain

equilibrium constants [11].


Results



pD-Hydrolysis Profiles for Myosmine and Analogs


Figures 3-1, 3-2, and 3-3 show how the composition of

solutions 3-1, 3-2, and 3-3, respectively, in D20 change with

pD. Again, the points in the figures are experimental

values, estimated from the published figures and the curves

are those calculated from our computer simulations.




) N N+

X X/ CH,

,1,X=N -2,X=N
4,X = CH 3-, X = CH



The published data for 3-1 [8], 3-2 [9] and 3-3 [9] had not

been subjected to a computer analysis and so equilibrium




41





1.0-
0






0.8





E
o 0.6



o 0


S0.4
.0-

U-




0
0.2



0



0.0* I *
2 4 6 8 10



PD


Figure 3-1. A titration curve for the hydrolysis of myosmine
(3-1) in D20. The open circles are experimental values
representing the fraction of imine and its conjugate acid
taken from ref 8. The curve was calculated based on eq 2-3
and the equilibrium constants obtained from the computer fit.

























- 0.4-
E


E



0
0.


0
-0
o


o 0.2

LL
o



0.1



0 0
0
0
0.0 ,I *
0 2 4 6 8 10 12



pD




Figure 3-2. A titration curve for the hydrolysis of N-
methylmyosmine (3-2) in D20. The open circles are
experimental values representing the fraction of iminium ion
taken from ref 9. The curve was calculated based on eq 3-1
and the equilibrium constants obtained from the computer fit.




43







1.0



o o



0.8



0

CI
E 0.6-
0
"

0
C
0
5 0.4
U-




0
0.2-







0.0-
0 2 4 6 8 10 12 14

pD



Figure 3-3. A titration curve for the hydrolysis of 1-
methyl-2-phenyl-l-pyrrolinium ion (3-3) in D20. The open
circles are experimental values representing the fraction of
iminium ion taken from ref 9. The curve was calculated based
on eq 3-2 and the equilibrium constants obtained from the
computer fit.








constants describing the titration curves had not been

defined. Our analysis allows these values to be extracted

from the data for the first time.

Myosmine

The pD-hydrolysis profile for myosmine was simulated

using the same equation (eq 2-3) and model (Figure 2-1) as

those used for anabaseine. The calculated equilibrium

constants appear in Table 3-1 and the curve through the data

points in Figure 3-1 was calculated using eq 2-3 and the

constants.

1-Methyl-2-(3-pyridyl)-l-pyrrolinium ion

The hydrolysis profile for l-methyl-2-(3-pyridyl)-1-

pyrrolinium ion (3-2) is different in shape from those for

2-1 and 3-1 in that at high pD values the uncharged open-

chain amino ketone predominates rather than neutral, cyclic

compound as in the cases of 2-1 and 3-1. However, with 3-2

the cyclic form is constrained to be a cation due to

quaternization instead of protonation of the nitrogen atom

and so the more reactive cyclic iminium ion hydrolyzes to the

amino ketone at high pD. Because of the presence of large

amounts of this open-chain material at high pD, the scheme in

Figure 2-1 does not correctly describe the hydrolysis of 3-2.

Kai is not relevant and Ka3 must be added to reflect

conversion of the open-chain alkylammonium ion to its

conjugate base (Figure 3-4). Again, KH is defined as the

ratio of the ketoammonium ion to the cyclic iminium ion.














a Oa\ 0i


0
4-4



a



4.-
so

1-1
4-4





0



U







44

4-4



,4-4
-H
-H(













U)
S.I





0,




U

0
-e























-4
-H

-4
.0












i-l-
.II

-r-1












E-r


'-


-r-1




ao



-O
4-4
0



H

r 0





0 1
CO


04



0
4r-1J






a




0
0 c




4J C






ua
4-4>
Cv


3 4-1
cr
<0 j3













1Ka2


D20 + N 'I)K H /H IND CH.
CHK
N H3 N



I lKa3

0 NDCH3
N


Figure 3-4. Model for N-methylmyosmine (3-2) equilibria in
aqueous solution.

Hence, eq 2-2 which was used for anabaseine and myosmine must
be modified; the concentration ratio of open-chain to cyclic
substrate must now be multiplied by the fraction [D+]Ka2 /
([D+]2 + [D+]Ka2 + Ka2Ka3) instead of the fraction given

therein to reflect the amount of open-chain material present
as the monocation. Again, the equation was rearranged and
written in terms of fraction of free imine to avoid
computational problems (eq 3-1). This consideration produces
a satisfactory fit (Figure 3-2).

Fraction of Ka2[D+l
Iminium Ion Ka2[D+] + KH([D+]2 + Ka2[D+] + Ka2Ka3)








The hydrolysis profile for 3-2 could be made even more

complex because the cyclic iminium ion is the conjugate acid

of an enamine and could be deprotonated to the enamine at

high pD values [40]. However, after a careful search the

enamine was not detected [9] and so it was not included in

our scheme. Redefining Ka3 as a dissociation constant to

reflect the formation of enamine conjugate base with the same

value for Ka3 rather than to yield acyclic amine base would

have no influence on the shape of the titration curve.

In the most acidic solutions for 3-2 there are some

deviating points (Figure 3-2). It is not clear whether these

points reflect the incursion of a new conjugate acid of

substrate or simply represent NMR errors in estimating small

amounts of minor component.

l-Methyl-2-phenyl-l-pyrrolinium ion and 2-phenyl-l-pyrroline

Two other reports concerning the equilibrium hydrolysis

of 1-pyrrolines are of interest because they provide

additional insight into the influence of structure on the

value of KH. l-Methyl-2-phenyl-l-pyrrolinium ion (3-3) and

2-phenyl-l-pyrroline (3A4) have a phenyl substituent in place

of the pyridyl ring of myosmine, the former also an N-methyl

group. The titration curve for 3-3 resembles that for 3-2 at

high pD in that open-chain material is favored while acidic

solutions show no variation in composition because of the

lack of a second basic group, the pyridine nitrogen (Figure

3-3) [9]. Again, we were able to fit the reported titration








curve for 3-3 based on the model in Figure 3-5 and using eq

3-2 with the computer program, equilibrium constants are

listed in Table 3-1.


Fraction of [D+]
=i (3-2)
Iminium Ion [D+] + KH([D+] + Ka3)



Although an enamine equilibrium could be incorporated into

the hydrolysis model, we find this to be unnecessary. Even

if the situation is made more complex by the inclusion of a

pKa value for an enamine (12.7), both pKa3 and KH are

essentially unchanged, now being 10.2 and 0.083,

respectively.




+N KH O ND0CH3 Ka3O NDCH3






Figure 3-5. Model for equilibria of l-methyl-2-phenyl-l-
pyrrolinium ion (3-3) in aqueous solution.


The titration curve was not presented for 3-4 but from

the reported equilibrium composition [8] we were able to

estimate a value for KH, Table 3-1.









Oxidation Product of Spermine and Spermidine


A titration curve based on proton chemical shifts of a

synthetic sample of the oxidation product dissolved in D20

was reported and the data points are reproduced in

Figure 3-6. According to this report [10] bicyclic 3-6

converts to monocyclic 3-8 on acidification. The first

analysis starts with the pair of equilibrating bicyclic

monodeuteronated cations 3-7a and 3-7b, said to be the major

forms in neutral water (Figure 3-7) [10]. They are in

equilibrium with monocyclic, tautomeric monocation 3-9. This

amine, expected to be more basic than the aminal because it

resembles a simple aliphatic amine with a remotely situated

electron-withdrawing group, accepts the deuteron, trapping

the substrate in its monocyclic form to give the observed

product 3-8. According to this model the fraction (F) of the

total amount of substrate present as imine is given by eq 3-3

which assumes that in the acidity interval in question there

are only three significant forms present, 3-7, 3-8 and 3-9

[41]. Kal is the dissociation constant for 3-8 going to 3-9

while KT denotes the ratio of ring-chain tautomers [42] [3-

17a/[3-9] and Kt gives the ratio of acids [3-7b]/[3-7a].

Because Kal is small relative to the concentration of acid

[D+] in eq 3-3 it may be neglected as indicated in eq 3-4

where the term KalKT(1+Kt) is a constant equal to Kapp.















0.95



0


0.75


o
0



E 0.55


o
0
o I


L- 0.35







0.15

o


O0 0

-0.05 0 0-i
0 2 4 6 8



pD



Figure 3-6. Titration curve for the oxidation product of
spermine and spermidine in D20. The open circles are
experimental values taken from ref 10 and the solid line was
calculated using eq 3-4 and Kapp = 5.47 x 10-6.








N ND,


Kt
if


N ND
KT


.0
N ND3







Q ND2


Figure 3-7. Model the equilibria associated with aqueous
solutions of the oxidation product of spermine and
spermidine.

Kapp is the apparent Ka value as given by the half-titration
point.


[31-] + [3-9]
F
L[37] + [3-8] + [3-91


F = [D+
[D+] + Kapp


[D+] + Kal
[D+] + Kal + KalKT(l + Kt)


(3-4)


The experimental data as taken from the reported titration
curve [10) have been fit using a nonlinear regression
microcomputer program and eq 3-4. The fit shown by the solid
line in Figure 3-6 is satisfactory except that our calculated
values for the fraction of imine appear to be slightly too
large at a pD of about 6. If this analysis is accepted, then
the Kapp value of 5.47 x 10-6 (pKapp 5.26) with a standard
deviation of 0.56 x 10-6 may be converted to its two


N ND


(3-3)


Ka








associated constants using an estimate of 9.68 for pKal [43]

from a model, the dideuteronated form of 1,3-diaminopropane.4

The value for KT(1+Kt) becomes 2.6 x 104. Since tertiary

alkylammonium ions are likely to be slightly stronger acids

than their corresponding secondary relatives the term (1+Kt)

is expected to be about 1 in value and so KT is approximately

2.6 x 104 showing that only a very small amount of the open-

chain amino tautomer is present along with the deuteronated

cyclic aminal.

The titration curve can also be simulated using a more

complex expression starting with aminal 3-6 as its free base.

In this simulation two hydrons are involved in the overall

conversion to monocyclic dication 3-8. As shown in Figure

3-8 the fit in the pD 6 region is much better. According to

this model the fraction (F) of total substrate present as the

monocyclic imine 3-8 and 3-9 is given by eq 3-5 where the

previously employed K values are as represented earlier and

Ka2 stands for the combined dissociation constant for the

mixture of 3-7a and 3-7b going to 3-6 (Figure 3-7). Again,

since Kal is small relative to [D+] over the pD range of the

titration, the term [D+]Kal can be neglected and the constants

combined into two where Klapp equals Kal(KT(1+Kt)) and K2app

equals Ka2/(1+Kt), eq 3-6.





4 The reported pKa of 8.88 at 25 OC (H20) is statistically
corrected for two equivalent acidic sites (0.30) and a
solvent isotope effect of 0.50[44] to give the value of 9.68.















0.95
0



o

0.75





C I
E 0.55

0
C

CO

U- 0.35






0.15






-0.05 I I *
0 2 4 6 8



pD


Figure 3-8. The same titration data as in Figure 3-6 but
this time the curve is calculated using eq 3-6 with Klapp
3.33 x 10-6 and K2app = 2.67 x 10-6.









F [[ + [D+]Kal + [D+]2
F = (3-5)
[D+]Kal + [D+]2 + [D+]KalKT(1 + Kt) + KalKa2KT(l + Kt)


[D+]2
F (3-6)
[D+]2 + [D+]Klapp + KlappK2app (3-6)



Reduced eq 3-6 has the same form as that for the

titration of a diprotic acid in which the fraction of the

total amount of acid in the final diprotonated state is

expressed. On computer fitting both values turn out to be

about the same, 5.48 and 5.57, for pKlapp and pK2appr

respectively.

Unfortunately, this second analysis is flawed. (1) At

the start of the titration the substrate was said to exist as

a mixture of aminal monocations [10]. (2) Our estimate from

the computer fit for the pKa value (5.57) of the conjugate

acid of 3-6 is not reasonable if K2app = Ka2/(1+Kt) or

approximately Ka2. The calculated pKa value for Ka2 is much

too low. The pKa (H20) of the conjugate acid of the model

aminal 2-isopropyl-1,3-diethylimidazolidine is 8.42 at 35 OC

[45]. (3) The standard deviation of Ka2 is an unsatisfactory

57% of the equilbrium value. The first scheme based on a

single degree of deuteronation is most likely the better

model.5

5 Dideuteronated aminal is not likely to be present in
significant amounts [41, 45]. Our superficial analysis based
on this dication can be made to fit the experimental data
when a step for the direct conversion of this ion to 3-8 is
included. Both ring cleavage and proton transfer then are
required, perhaps by the participation of a water bridge.









Enamine 3-5 must be only one of several rapidly

interconverting structures, its presence being inferred by

the observation of hydrogen-deuterium exchange at the beta

position of the enamine [10]. The minor contribution of 3-5

also is consistent with the high basicity of enamines [40];

it is expected to be an unimportant contributing structure

except in highly alkaline solutions.






Discussion



Skewing of Observed pK_ Values by Concurrent Ring-Chain
Equilibria


The presence of concurrent tautomeric and acid/base

equilibria during the titration of the oxidation product of

spermine and spermidine gives an apparent pKa which is skewed

relative to the true value. Taking the pD value at the

midpoint of the titration curve (Figure 3-6) as being equal

to a pKa value provides an apparent pKa (pKapp) of 5.2. Is

this value consistent with the structural assignments or does

it invalidate them?

On first consideration this would seem to be a pKa value

for the mixture of secondary and tertiary amino groups in the

conjugate acid of 3-7, a dication. Comparison of this pKa

value with that of a model compound initially seems to be

consistent. Aminals are hydrolytically labile [41] and so we








make comparison not with a desired gem-diamine as in 3-7 but

rather with a vic-diamine, as its dicationic acid, e.g., the

diprotonated form of 1,2-diaminoethane which has as its first

pKa (H20) a value of 7.4 [46]. The diprotonated form of

aminal 3-7 would be expected to have a pKa value much lower

than this owing to the larger acidifying effect of the more

closely situated ammonio group [41, 45] in apparent agreement

with the assignment.

This superficial analysis is not correct, however. The

reported titration curve not only applies to a deuteronation

step but also to a ring cleavage reaction, a ring-chain

tautomeric equilibrium between aminal 3-7a deuteronated at

its secondary amino group and its monocyclic aminoalkyl

iminium ion 3-9. This equilibrium provides a major

perturbation on the true pKa value. Kapp then is the product

of an equilibrium constant for a tautomeric step (KT) and an

equilibrium constant (Kai) for a weak acid in which the

conjugate base of this acid is disfavored by the prior ring-

chain equilibrium. The large KT term skews the apparent

acidity constant of the ammonium ion making it large. The

apparent acidity constant is not a value for dissociation

alone as a simple consideration of the data might first

suggest.

Consideration of the pH-dependent hydrolysis profile of

3-3 yields the same result. The inflection point of the

titration curve gives an apparent pKa value of 11.4. This

value is skewed relative to the true values due to








simultaneous hydrolytic equilibrium. The observed inflection

point (PKapp) of 11.4 in the reported titration curve for 3-3

is nicely reproduced by considering the pKa of the open-chain

amine and KH, i.e., pKapp = pKa3 log(KH /(KH + 1)) = 11.4

where pKa3 = 10.3 and KH = 0.085. A similar result was found

for anabaseine and was discussed in Chapter 2. Thus, care

must be taken when interpreting titration curves for

substances which undergo ring-chain interconversion as the

apparent pKa values may not reflect the true pKa values but

instead represent mixtures of equilibrium constants.


Myosmine and Iminium Ion Acidities


Because of the expected interest in the heretofore

undescribed variation in the composition of reaction mixtures

of 3-1 by those engaged in pharmacological studies, a plot is

given in Figure 3-9 showing how the fractional amounts of its

four components change with pH. These curves were

constructed using the equilibrium constants in Table 3-1

after making a reduction of 0.5 in pKa values to reflect the

change from D20 to H20. Equations such as that given by

eq 2-5 describe how the fractional amount of a given

component varies with pH.

The titration curve for 3-1 is similar in shape to that

of 2-1 and their KH values also are alike, 1.9 vs 1.2,

respectively. However, comparison of the Kal values for the

two substances produces a major surprise. The pKa value of
















O



U
a)






0.20


0.0
C 0.4-

C0.0







1 2 3 4 5 6 7 8
pH

Figure 3-9. Calculated changes in the composition of aqueous
solutions of myosmine (3-1) based on the model in Figure 2-1
as a function of pH. The pKa values in Table 3-1 were
decreased by 0.5 to correct for solvent isotope effect.
Curves A and B refer to the fraction of amino ketone dication
and monocation, respectively. Curves C and D refer to the
fraction of protonated cyclic iminium ion and the imine free
base, respectively.



5.9 for the smaller ring is very much less than the value of

7.7 for the larger ring. This lower value serves to change

markedly the populations of the various components. For

example at pD 7 there is 80% of 3-1, 7% of its conjugate acid

and 13% of monoprotonated, ring-opened amino ketone while for

anabaseine there is 8% of 2-1, 42% of its conjugate acid and

50% of monoprotonated amino ketone. In H20 these values are









likely to be 93%, 2% and 5%, respectively, for myosmine,

Figure 3-9. Much more of the cyclic free base is present

compared with 2-1 under the same conditions where 21% imine

free base, 36% cyclic iminium ion, and 43% mono-protonated

keto ammonium ion are present.

The remarkably large difference in pKa values for the 5-

and 6-membered iminium ions stands in marked contrast to the

insignificant difference in pKa values for the corresponding

saturated, secondary amines, pyrrolidine (pKa 11.13) [47] and

piperidine (pKa 11.07) [48]. This large variation certainly

is not due to the different ionic strengths used in the two

studies, the former not being reported but is likely to be

less than our own.

Two other reports support our claim that the pKa values

of 5- and 6-membered iminium ion acids are very different.

(1) A large difference in pKa values (H20) had been reported

much earlier for anabaseine (6.7) and myosmine (5.5) from

classical titrations [34]. (2) Another old study on similar,

simpler structures also has gone largely unrecognized,

possibly because the initial structures were assigned their

incorrect tautomeric enamine forms. We know today that

enamines of secondary amines really exist as imines so that

the correct forms of 3-10 and 3-11 are the tautomers [49, 50]

3-12 and 3-13, respectively (eqs 3-7 and 3-8). The pKa value

for the conjugate acid of 3-12 is 9.55 [51] and for 3-13 it

is 7.91 [52]. Again there is a large difference and again

the acid with the 5-membered ring is stronger.








The early workers recognized but denied the possibility of

hydrolysis during their titrations of 3-12 and 3-13.

Considering the effect of the 2-methyl group on the value of

KH as discussed below, this assumption may be largely

correct. Even if the reported pKa values are not for a pure

substance but are for an equilibrium mixture that is somewhat

skewed by the presence of more basic amine, the difference is

real and highly significant. The amount by which the

reported pKa (pKapp) differs from the true value must be about

the same for both the conjugate acids of 3-12 and 3-13, since

PKapp = pKa + log(l + KH).




a C I (3-7)
N CH CH
H
-10 3-12




14(3-8)
N CH3 N CH3
H

3-11 3-13



The greater acidity of the iminium ion with the

5-membered ring over that of the 6-membered ring is likely

due to the difference in energy between the free bases. The

greater s-orbital character of the lone pair associated with

the smaller interior bond angle of the 5-membered ring

provides a larger stabilization thereby making the nitrogen








atom less basic. While the direction of the stabilization is

expected, the large magnitude is a surprise and raises the

question of whether other cyclic imines of varying ring size

show similar hybridization effects.


K Values for Hydrolysis of Cyclic Iminium Ions


Consideration of all the KH values, Tables 2-1 and 3-1,

for the cyclic imines as a function of substituents suggests

a trend. As the group at the imino carbon atom is made less

electron-withdrawing in the order 3-pyridyl and phenyl there

is less hydrolysis. That is, cyclic imine is favored more as

the substituent is made more electron-donating, the same kind

of electronic effect as found earlier [9]. Similarly,

addition of a methyl substituent to the imine nitrogen atom

causes the cyclic form to increase in abundance. These

limited data should be considered with caution but they

suggest that in the absence of steric effects electron-

donating groups preferentially stabilize the protonated

iminium ion more than the open-chain ketone. But the

combined effect of both changes is modest, being only a

factor of 22. This conclusion also suggests that the old pKa

values reported for the conjugate acids of 3-12 [51] and 3-13

[52] may be skewed only a little by the presence of open-

chain hydrolysis product because only a little may be

present.









Conclusions


Much information can be gained from titration curves for

materials which undergo pH-dependent ring-chain equilibria.

Construction of these curves is relatively simple and only

requires measurements of the relative amounts of the species

in equilibrium over the pH range where a change in the

composition occurs. With the appropriate model system, the

pH-profile can be simulated and relevant equilibrium

constants can be extracted.

By varying the pH of physiological experiments with

cyclic imines from about 6 to 8 the following predictions may

be made with the important assumptions that pH changes in

solution are mirrored at the receptor site and that

variations in receptor properties due to pH may be corrected

by comparison with, say, carbamylcholine. The concentration

of anabaseine free base over this interval will increase by a

large factor of about 30 while the concentrations of its

monocationic cyclic and acyclic ions will decrease by a

modest factor of approximately 3. For myosmine over this

same pH interval the cyclic free base concentration will

increase only by about 2-fold and the concentrations of the

cations will fall by about a large factor of 50.

Fortunately, the observations for these two substances

reinforce each other because there is a major change in

composition of different species as the pH is varied, the








neutral form of anabaseine and the monoprotonated entities

for myosmine. It therefore should be possible from the

increase or decrease in bioactivity to state whether it is

the neutral or cationic form that is responsible for

promoting activity at the receptor site. However, because

both the amounts and pH dependence of the monocations are so

similar, it will be necessary to employ non-interconverting

model compounds to distinguish between them6 [53]. In this

way it should be possible for the first time to identify both

the structure and the state of protonation of these

interesting, old neurotoxins when bound to the active site7

[54].

The identity of the chemically produced oxidation

product from spermine and spermidine and of the plant

polyamine oxidase product from spermine now has been firmly

established. Under physiological conditions the major form

is 3-7. The titration data are consistent with the proposed

structures.












6 Previous analyses of nicotine action upon the neuromuscular
acetylcholine receptor as a function of the external pH have
provided strong evidence that the monocationic species is
much more active than the neutral form [53].
7 A comparison of the activity of 2-1 and 3-1 has been made
with insects, 3-1 being less reactive [54].













CHAPTER 4
COMPENSATORY CHANGES IN THE INFLUENCE OF COSOLVENTS ON THE
POSITION OF THE RING CHAIN EQUILIBRIUM FOR ANABASEINE AND
N-METHYLANABASEINE


Introduction


Because the nicotinic receptor for cholinergic agents

may be in a hydrophobic environment [55 57], studies were

designed to determine what influence a hydrophobic

environment might have on the equilibrium ratio for

anabaseine (4-1), its derivatives and analogues. The

positions of the ring-chain equilibria of 4-1 with 4-2 and of

its N-methyl derivative 4-3 (1-methyl-3,4,5,6-tetrahydro-

2,3'-bipyridinium ion) interconverting with 4-4 were measured

in aqueous and nonaqueous solvents and their mixtures by

1H NMR [12]. First, a general survey of the solvent effects

of water, DMSO, and methanol on the equilibrium compositions

of solutions of 4-1 and 4-3 is presented. This is followed

by a quantitative study of the changes in the equilibrium

compositions in water-methanol and water-DMSO. The

compositions of solutions of 4-1 and A4- in water-methanol

mixtures can be described by a linear free energy

relationship.








Analogues 4-3 and 4-5 (l-methyl-6-phenyl-2,3,4,5-

tetrahydropyridinium ion [58]) were prepared so that the

effects of methyl substitution on the hydrolysis of the

6-membered imine rings could be compared with those of the

5-membered cyclic imines discussed in Chapter 3. Analogue

4-5 has a phenyl substituent in place of the 3-pyridyl ring

and hydrolyzes to A4-. N-Methylated 4-5 was included to

further substantiate the influence of a steric effect in the

N-methyl compounds on the position of the ring-chain

equilibrium.


3
3 5 2 4

N+ O N 0 NH2G
6' ,2 6' 72'
x x

4A1. X=N, G=H 4-2, X=N, G=H
4-3, X=N, G=CH3 44, X=N, G=CH3
4, X=CH,G=CH3 4&, X=CH, G=CH3



Claisen-type syntheses of the known compounds 4-1 [59],

4-3 [58], and 4-5 [58] gave the desired substrates and will

be discussed in Chapter 6. The nonaqueous model solvents

include two that are moderately polar, amphiprotic CD30D and

the more basic dimethyl sulfoxide-d6, a strong hydrogen bond

acceptor [60 62].













Ring-Chain Equilibria and Nomenclature


A question arises as to how to name the solid compounds

used as starting materials. Elemental analyses indicate they

contain one equivalent of water in the solid state; the

associated structures are compatible either with a ketone or

with an imine containing a stoichiometric amount of free

water in the lattice (water of hydration). Which structure

is correct and therefore which name should be designated?

This is a classical question that could be answered by "magic

angle" solid state NMR but it was not addressed because we

are interested in the solution chemistry. Further confusion

arises about how to indicate the degree of protonation of

N-methylated substrates in the solid state because, for

example, the conjugate acid of ketone 4-4 is diprotic (a

dihydrohalide) while that for iminium ion 4-3, its

equilibrium component, is monoprotic (a monohydrohalide).

Equation 4-1 schematically illustrates the stoichiometry for

the hydrolysis reaction, showing the equilibrium between the

cyclic and acyclic forms where G is H or CH3.




S+ D20 (4-1)
R"' aO NH2G R N
G








We have elected to name our starting materials using the

common name for the cyclic imine as have others [58] and

therefore designate its associated degree of protonation.

The systematic name for the open-chain amino-ketone is not

employed in spite of the possibility that in the solid state

the predominant form is acyclic. Thus, we prefer the common

name N-methylanabaseine cation (4-3) over the cumbersome name

for its hydrolysis product, 5-methylammonio-l-(3-pyridyl)-l-

pentanone (4-4). Hence, salts of 4-3 will be referred to

schematically as 4-3.Cl for that cyclic chloride having an

unprotonated pyridine ring while 4-3.C1.HC1 denotes its

conjugate acid, the pyridinium salt. Similarly, the

dihydrobromide of 4-1 is designated 4-1.2HBr and the

monohydrobromide 4-1.HBr.


Survey of Solvent Effects


Preliminary studies first show the scope of solvent

changes on the position of the equilibrium [63]. Then a

quantitative investigation with mixed water-methanol solvents

along with a more limited study of water-dimethyl sulfoxide

mixtures is described.

(1) Water

N-methylated 4-3.Cl.HC1 gave a solution with a pD of 3.0

where the material exists largely as the dication. The only

form present was the conjugate acid of open-chain amino

ketone 4-4. A solution of 4-3.Cl gave a pD of 6.0 and two








separate solutions of 4-3.C1.HC1 made basic with carbonate

gave pD values of 9.2 and 9.6. In these three cases

approximately 6-7% of the cyclic iminium ion 4-3 appeared,

Table 4-1. The ratio of open-chain keto-ammonium ion to

cyclic iminium cation under neutral conditions thus is about

14. .The assignment of structures was made by comparison with

the proton NMR shifts we previously observed for the non-

methylated derivatives 4-1 and 4-2 under similar conditions

[Chapter 2].

Anabaseine dihydrobromide exists largely as the amino

ketone while 4-1.HBr has nearly equal amounts of the two

forms, the ratio of the open-chain to cyclic structures being

about 16 and 1.3, respectively. Phenyl derivative 4-5 exists

largely as open-chain protonated amino ketone; there is 3.5

times more ketone than iminium ion, Table 4-1.

(2) DMSO

NMR spectra of 4-1 and 4-3 either as their dihydrohalide

or monohydrohalide salts were recorded using DMSO-d6 over a

concentration range of 0.04 to 0.16 M. The added solids

contain one equivalent of water as indicated by elemental

analyses and so an equilibrium exchange of water is possible

in the absence of any added water. Chemical shifts for the

amino ketones, 4-2 and 4-4, in DMSO-d6 and CD30D appear in

Table 4-2 and chemical shifts for cyclic imines, 4-1 and 4-3,

and enamine, 4-7, appear in Table 4-3.








Table 4-1. Composition of Ring-Chain Equilibrium Mixtures in
Various Solvents Determined by 1H NMR at 220 C.


Percent
Compound M Solvent Ketone Imine Other


4-1.2HBr




4-1.HBr



4-1.HC1



4-3.C1.HC1


4-5.Br


0.085
0.071
0.071

0.080
0.079
0.075

0.080
0.074
0.071

0.17
0.053
0.039
0.036
0.068
0.083

0.13
0.043
0.082
0.16
0.078
0.087
0.080


D20
DMSO-d6
CD30D
CD30D

D20
CD30D
CD30D/D20

DMSO-d6
DMSO/D20
DMSO/D20

D20
D20/Na2CO3
D20/Na2CO3
DMSO-d6
CD30D
CD3OD

D20
DMSO-d6
DMSO-d6
DMSO-d6
DMSO/D20
CD30D
CD30D/D20


D20


94
100
72
32

57
trace
27

10
76
78

100
94
93
97
65
86


93.5
60
72
78
100
34
77

78


0
0
9a
26a,b


43
100
73


6.5
21
23
16
trace
66
23

22


0
19h
5h
6h
tracec,h
0
Oc

0


unknown structure.
bSame sample as above but after 10 days.
c2.6 M D20.
d4.1 M D20.
e5.9 M D20.
fpD=9.62, 70 mM Na2C03, 0.5 M NaC1.
gpD=9.20, 42 mM Na2C03, 0.5 M NaCl.
hEnamine.














I *n

CM1


I u

N(


r-
~~9~N
.Q .Q ,Q '-


(N (N







- -q






0 0
o o






(V) C4

a;o


o 'O.








CD 00I


-1 (N
.-4 .-4

c a c


I -






r-
o o

m mO




T oo
oo oo

r-{ i-


I r-






a\ o







t- co
0( 0



CM u


ra,
~r~ o


3:








O
0


0
O








-41
U-)
0U


0)


<'


C


-04 0
S(0








00
OO.0
.-I <-


CSC
ON 1I
1 I 1I V I
I q1 r Iq' 1 ^












a,












m

I'-
r-4



00
ScO
N N









N N






(Y)
C
a, a,


0 0 l






0O O
'-4 0O'
o. a


I U I <3
(*') ro)


0 C>
0 0


,N oN

o tr


M 1-4

( (N


S) Lo
Un


i)

a,
00


00 t.0
t.0 CN













00 OD
roo o


Sa)
rI



c X

Oa)
0 o me
4-1 > 0

OaOn




oa 4 no


i-l r-^
u n c u
a w .-i CQ a ,y -^
* a u a S u r~l
r-l a- fS] ^ W ^1
u ^ ^ ^ u ^ ^ ^

^ ^ T ^ m 4' 'T
f sr








Results summarized in Table 4-1 show (i) dications exist

almost entirely as their open-chain keto ammonium ions, (ii)

monocations give more of the cyclic iminium ions, iminium ion

being the major product from A4- and a minor product from 4-

3. In addition, the enamine 4-7 (1-methyl-1,4,5,6-

tetrahydro-2,3'-bipyridine) is clearly detectable from the

monohydrohalide of 4-3. Formation of the enamine under such

mild conditions is remarkable. By contrast, in aqueous

solution the addition of base and a high pH is required for

enamine formation [9].


3 5



6' r2' CH3

4-7



An authentic sample of the enamine derivative 4-7 was

generated by extracting with chloroform an aqueous solution

of 4-3 made alkaline with sodium carbonate. Its NMR spectrum

contains among other signals a triplet due to the alkene

proton at 5.03 ppm as proof of its structure.

Varying the concentrations of 4-3.Cl causes the product

ratio to change. More dilute solutions give rise to lower

concentrations of ketone and consequently more iminium ion

and more of its conjugate base, the enamine 4-7. The ratio

of ketone to the total amount of iminium ion and enamine

decreases from 3.5 to 2.6 to 1.5 on diluting substrate.








Both the presence of enamine and its increasing contribution

with dilution is readily understandable. In an equilibrium

such as that given by eq 1 where unequal numbers of particles

are involved, dilution favors that side having the larger

number of particles, in this case the iminium ion and water.

On dilution the solution becomes less acidic as the keto

ammonium ion acid is converted to the weaker acid water.

This decrease in medium acidity causes more of the N-methyl

iminium ion to dissociate to its conjugate base the enamine

4-7. The halide counter ion although substantially more

basic in DMSO than in water is not likely to serve as the

active base for deprotonation. Instead it is the more

abundant solvent. Adding 5 volume percent D20 to 4-3.Cl

shifts the equilibrium almost completely to ketone as

expected from mass action considerations.

(3) Methanol

Samples about 0.08 M in CD30D show the following

characteristics, Table 4-1: (i) dications exist largely as

open-chain ketones but substantial amounts of cyclic iminium

ions are present as well, (ii) monocations exist mostly as

the iminium ion, more being favored from 4-1 than from 4-3,

(iii) adding 5 volume percent D20 to the monocations causes

more ketone to form and (iv) both dications give <10% of an

unknown material which may be a cyclic hemiaminal [64] or an

acyclic ketal. The structure was not identifiable because








some of the high field peaks were overlapped by the major

components.


Ouantitative Studies


(1) Water-methanol

Serial dilution experiments were performed with 4-1.HC1

and with 4-3.Cl in the mixed solvent. To a methanolic

solution of each substrate was added a measured amount of D20

and the NMR spectrum of the mixture was recorded to provide

the equilibrium composition. As expected from Table 4-1 the

addition of water causes the amount of acyclic ketone to

increase. This increase occurred rapidly at first with small

additions of water and then more slowly. The smooth increase

in the ratio of the concentrations of acyclic keto ammonium

ion 4-2 to cyclic iminium 4-1 is shown in Figure 4-1 and

similarly for the ratio of 4-4 to 4-3 in Figure 4-2 as the

mole fraction of water increases to about 0.45.

A linear free energy relationship (LFER) quantitatively

describes the change in the ratio of the two forms of the two

substrates as the composition of the solvent is varied and

provides the two desired equilibrium constants for the pure

solvents water and methanol, eq 4-2.



In Kmix = FD In KD + (1 FD) In KM (4-2)



In Kmix = FD In (KD/KM) + In KM (4-3)




























0.6


0.0 0.1 0.2 0.3 0.4


mole fraction of D20



Figure 4-1. Observed ratio of concentrations of keto
ammonium ion, 4-2, to iminium ion, 4-1, as a function of mole
fraction of D20 in CD30D at 22 OC and uncorrected for
"impurity" of water in methanol. The cross denotes the
result of a separate determination.









25






20






15 0
0





10






5-

+




0 a I I I
0.0 0.1 0.2 0.3 0.4 0.5


Mole Fraction of D20


Figure 4-2. Observed ratio of concentrations of keto
ammonium ion, 4-4, to iminium ion, 4-3, as a function of mole
fraction of D20 in CD30D at 22 OC and uncorrected for
"impurity" of water in methanol. Open and filled circles are
due to two separate serial dilution experiments. The cross
represents a separate determination.








In the LFER given by eq 4-2, Kmix represents the observed

equilibrium constant for the mixed solvent, KD denotes the

equilibrium constant when pure D20 is the medium, KM is that

for pure CD30D, FD is the mole fraction of deuteriated water

and (1 FD) is the mole fraction of CD30D in the mixture.

The equilibrium constant is defined as [ammonio

ketone]/([iminium ion] [D20]) or K = [AK]/[Im] [D20]tot where

[D20]tot denotes the total amount of water from all sources as

explained below. The LFER actually was applied using

rearranged eq 4-3 and the outcome is presented graphically in

Figures 4-3 and 4-4 for 4-1 and 4-3, respectively. An

equation similar to eq 4-2 has been reported, for example,

for rates of reactions in mixed solvents but in these cases,

unlike our own, one of the solvents was not a reactant [65].

Calculation of the mole fraction FD and the

concentration of water in a mixture requires some care

because two minor corrections are needed. They are based on

the following considerations: (i) a correction needs to be

applied to reflect the release of water into the medium when

the ketone is converted to iminium ion, eq 4-1 and (ii) the

commercial sample of CD30D usually was not dried and the

initial amount of water "impurity" in the methanol was not

determined independently. Both corrections are made easily.

In the first the observed equilibrium quantities of ketone

and imine provide a measure of the amount of water liberated

when ketone cyclizes. In the second the unknown quantity of

"impurity" water is treated as an adjustable parameter












-0.75







-1.25






E
y -1.75







-2.25







-2.75 *,-
0.0 0.1 0.2 0.3 0.4 0.5



mole fraction of D20


Figure 4-3. Linear relationship between In Kmix verses mole
fraction of D20 in CD30D at 22 oC where Kmix = [4-2]/ ([4-
1] [D20]). The filled circles represent the actual data
points uncorrected for the amount of "impurity" water present
in the solvent while the open squares include this
adventitious water calculated according to eqs 8-3 and 8-4.
The slope is -2.02 and the intercept, KM, is -1.65.








1.4






1.2



0



1.0

X
E


0
0.8






0.6
0



0

0.4
0.0 0.1 0.2 0.3 0.4



Mole Fraction of D20


Figure 4-4. Linear relationship between In Kmix verses mole
fraction of D20 in CD3OD at 22 oC where Kmix = [4-4]/
([4-3][D20]). The slope is -1.90 and the intercept, In KM, is
1.20. Open and filled circles represent two separate serial
dilution experiments while the arrow denotes the value of the
equilibrium constant obtained using solvent dried with
molecular sieves.








thereby avoiding the need for tedious drying and

determination of the actual water concentration.

Determination of the amount of water impurity in the solvent

is achieved by a computer fitting of all the experimental

data based on eq 4-3 as described in the Experimental Section

[Chapter 8]. These are minor corrections and only materially

affect Kmix when the measured amount of added water is small

but they provide gratifyingly improved fits of the data.

The LFER given by eq 4-3 first was rearranged and then

the data were fit using a nonlinear regression microcomputer

program that yielded the desired equilibrium constants as

well as the the amount of impurity water. The exact form of

the equation is given in the Experimental Section

[Chapter 8]. Figure 4-3 shows a typical set of data points

with and without the correction for water impurity; there is

only one value that changes substantially.

Having a measure of the total water concentration allows

the construction of a plot of eq 4-3 to show clearly the

influence of water on the apparent equilibrium constant.

Accordingly, the intercept gives KM while the product of the

antilog of the slope and the antilog of the intercept

provides KD. The KD and KM values for 4-1 are 0.025 and

0.19 M-1 respectively (correlation coefficient, r, for eq 4-3

is 0.997) and for 4-3 they are 0.49 and 3.3 M-1, respectively

(r, 0.984). For 4-1 the value of KD determined in earlier

[Chapter 2] using purely aqueous solutions is 0.022 M-1

(KH/55.3 M), which is in very good agreement with the present








value. There is more scatter in the data for 4-3, Figure 4-

4. The ratio of ketone to imine is so large that the

uncertainty in the measurement of the minor component is

sizable. Thus, for example, the KD value for 4-3 from the

LFER is 0.49 M-1 while the experimental value measured

directly is 0.26 M-1 (93.5/(6.5 x 55.3)). However, in the

former case this corresponds to a composition of 3.6% imine,

the latter to 6.5%, two values the same within the

experimental uncertainty of our measurements. The computer

fit indicates that the amount of water initially present in

the methanol is 0.13 and 0.036 M for 4-1 and A-3,

respectively. Contributing to the quality of the fit is the

expectation that the pKa values for the nitrogen acids will

be similar in the two solvents [66, 67] and that no

significant change in the fraction of protonated material

occurs in the mixtures. Only small variations in the

chemical shifts of the substrates were observed in support of

this suggestion.

The two linear relationships given in Figures 4-3 and 4-

4 may be interrelated through their common axis, mole

fraction water. Thus, In(Kmix)4-1 = 1.061n(Kmix)4-3 2.93

showing that both 4-1 and 4-3 respond in very nearly the same

way to additions of water to methanol.

As a check on our computational approach to determine

and to correct for the amount of adventitious water present

in "dry" methanol two experiments were carried out following

drying of the solvent with 3A sieves. Only when elaborate









attempts were employed in the second of these to exclude

moisture, including drying of the glassware, was it possible

to obtain an equilibrium constant KM having essentially the

same value (3.6 M-1) as that obtained from the LFER

(3.3 M-) Moreover, in this case enamine 4-7 also was

detected in about 6% yield. Clearly the additional effort

required to dry the methanol is unnecessary; trace amounts of

water may easily be established using our serial dilution

technique and computer fitting of the data as given by eqs 8-

3 and 8-4 in the Experimental Section.

In addition some other new samples were prepared

independently in the usual way without prior drying and the

resultant data applied to the curves in the figures in order

to check that the curves indeed are reproducible. Figures

4-1 and 4-2 show the new points fit within the uncertainty of

the measurements and therefore our results are verifiable.

However, a generalization should be kept in mind: as the

ratio of components changes from one in value, the

uncertainty in the value of the equilibrium constant

increases [30] and this accounts for the greater scatter in

Figure 4-4.

(2) Water-DMSO

Serial dilution experiments were performed with 4-1.HC1

by adding D20 to samples in DMSO-d6. Analysis of the three

data points, Table 4-1, produced an LFER based on the mole

fraction of added water. The value of KDMSO representing the








equilibrium constant in pure DMSO is 1.5 M-1. However, the

derived value of KD for pure water solvent is 0.078 M-1,

substantially higher than that observed directly for

measurements made with pure water and also that obtained from

methanol-water mixtures. Moreover, in contrast to the

spectra obtained using methanol-water samples which did not

show significant changes in chemical shift on addition of

water, the NMR spectra of the DMSO-water samples contained

large changes in shifts for the imine but not for the ketone.

Pyridine signals usually moved upfield suggesting that the

mixture became less acidic. However, the methylene protons

adjacent to the imino nitrogen moved in the opposite

direction consistent with increased protonation. As these

shifts give inconsistent data about protonation, we made no

attempt to interpret them and do not regard the LFER as

particularly meaningful.

An estimate was made of the value of KDMSO for 4-3.Cl

based on the assumption that all the water from the three

entries in Table 4-1 come from substrate when it forms imine

and enamine. The derived equilibrium constant has the value

15013 M-1 and represents an upper limit because any water

present as an impurity was not considered in the derivation.









Discussion



Solvent Effects


Three different solvents were chosen to study the

influence of environment on the position of the ring-chain

equilibrium involving the transfer of water to cyclic iminium

ion to form acyclic keto ammonium ion, eq 4-1. These include

water as the reference and hydroxylic but less polar methanol

and polar, protophilic dimethyl sulfoxide. For the

positively charged nitrogen acids examined, only small

changes in their pKa values are expected when dissolved in

the chosen solvents. DMSO, being more basic than water, will

bring about more dissociation than water [60, 62, 68]. The

pKa values for the compounds in methanol and water are likely

to be similar [66, 67].

No attempt was made to dry the commercial,

perdeuteriated solvents. Moreover, up to one equivalent of

free water may be present in the non-aqueous solvents as

provided by the heterocyclic salts themselves based on their

composition. This stoichiometric water is bound in the amino

ketone and is liberated as free water on conversion to imine,

eq 4-1. As our quantitative treatment of the data provided

by methanol-water mixtures shows, eq 4-3, drying is

unnecessary for a systematic study because the amount of








adventitious water can be determined on computer fitting of

the data.

Dramatic shifts in the position of the ring-chain

equilibrium do occur both as a function of the composition of

the starting material as well as of the solvent. While the

dihydrobromide of 4-1 exists almost completely as the acyclic

ketone in water and in DMSO, it is almost entirely in the

cyclic iminium structure when present in methanol as the

monohydrobromide. A similar but less extensive shift is

found with 4-3.C1.HC1 which is virtually all ketone in both

water and DMSO while 4-3.Cl is about 65% iminium ion in

methanol, Table 4-1.

Major conclusions derived from the contents of Table 4-1

and the figures are as follows. (1) Ketone is favored in

most of the trials listed in Table 4-1. That is, the

equilibrium given by eq 4-1 need not be shifted largely to

the left favoring iminium ion according to the law of mass

action in spite of the concentration of water being low in

"anhydrous" methanol and DMSO. (2) Generally the least

amount of cyclic iminium ion is present in pure water, (3)

more iminium ion is present in methanol than in

dimethylsulfoxide and (4) more iminium ion is present in the

less acidic solutions, especially when the starting compound

is the monohydrohalide rather than the dihydrohalide salt.

The latter conclusion is in keeping with our prior

observations for purely aqueous solutions [Chapter 2].

Interestingly, although iminium ion and especially neutral








imine are favored by basic conditions, the most basic

solvent, DMSO, does not provide the most iminium ion from any

salt, showing that the basicity of the medium is not the only

relevant factor. Solvation, perhaps by means of H-bonding,

especially to stabilize the alkyl ammonium cation, which has

a larger number of acidic protons than the iminium ion,

should have an important influence on the position of the

equilibrium and DMSO is an especially effective H-bond

acceptor [68]. (5) The values of KD for 4-1.HBr and for

4-3.Cl are 0.025 and 0.49 M-1, respectively, and for KM they

are 0.19 and 3.3 M-1, respectively, as derived from Figures

4-3 and 4-4.

The changes in the values of the equilibrium constants

can be understood in terms of the activity coefficients for

substrate transfer from one solvent to another. According to

this, the concentration terms cancel in our ratio of

equilibrium constants leaving a ratio of solvent activity

coefficients [69, 70] eq 4-4, where D.M represents the solvent

activity coefficient for transfer from D20 (D) to CD30D (M)



KD/KM = D AK /(DYMIm DMD) (4-4)



of the component designated by the subscript. Activity

coefficients have been reported for H20 in CH30H based on mole

fraction with unit mole fraction as the standard state

[71, 72]. These values indicate little deviation from

ideality. Therefore, our water term appears to account for