Effects of ionic perturbations and metal ion competition on the binding of a model drug to DNA


Material Information

Effects of ionic perturbations and metal ion competition on the binding of a model drug to DNA
Physical Description:
xvi, 301 leaves : ill. ; 29 cm.
Eisenhardt, Peter Forrest, 1945-
Publication Date:


Subjects / Keywords:
Acridines   ( mesh )
Macromolecular Systems   ( mesh )
DNA   ( mesh )
Ions   ( mesh )
Osmolar Concentration   ( mesh )
Binding, Competitive   ( mesh )
Medicinal Chemistry thesis Ph.D   ( mesh )
Dissertations, Academic -- Medicinal Chemistry -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1977.
Includes bibliographical references (leaves 295-300).
Statement of Responsibility:
by Peter F. Eisenhardt.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000898605
oclc - 25764483
notis - AEK7311
System ID:

This item is only available as the following downloads:

Full Text








To Dearest Joannie
who, more than any other,
deserves credit for any positive
attributes I may possess.


I am indebted to my supervisory committee, Dr. Stephen

G. Schulman, Chairman; Dr. B. Andresen; Dr. M. Battiste;

Dr. C.H. Becker; and Dr. R.H. Hammer for their time and

guidance in the preparation of this manuscript.

To three Professors, I owe special thanks. To Dr.

Schulman for the many hours of his time he has given to

share his prodigious scientific knowledge. To Dr. "C" whose

unaffected concern for the well-being of all with whom he is

associated, combined with his keen intellect, makes him with-

out peer. And, to Dr. John Baxter for his fine example and

unflagging attempts to maintain high standards within the

academic community.

Space permits mentioning only a few of my friends to

whom I'm indebted: Mum, Dad, Clam, Burns, Sue, Flyman and

Ms. Fly, Teem, Merm, Tony, Roy, Lenny-whats-is-name, Edgar,

and Lysa Chancey Smith.

Robert, Ann, Robbie, and Sarah are four of my life's


Thanks, too, to Carolyn Grantham, expert cryptographer,

who deciphered my cruel first drafts so competently without

once betraying her graceful, elegant mien.




ACKNOWLEDGEMENTS ..................................... iii

LIST OF FIGURES...................................... vi

LIST OF TABLES ....................................... x

ABSTRACT............................................... xiv

CHAPTER I INTRODUCTION ............................. 1

Binding of Small Molecules to Biopolymers....... 2

Thermodynamics of Binding Reactions............... 34

Rationale for the Selection of the Model
System and Experimental Design................... 42

CHAPTER II EXPERIMENTAL ............................ 47

Materials and Procedures ........................ 47

Calculations....................................... 55


General Spectral and Titration Characteristics.. 62

Apparent Association Constants for Surface
and Intercalative Binding ....................... 84

Association Constants Based on Activities --
Extended Debye-Hickel Considerations.............. 97

Evaluation of a Simple Competitive Binding
Model ........................................... 102

Thermodynamics of the Binding of 3-amino-
acridinium and 7-aminoquinolinium to DNA........ 105


APPENDIX I FIGURES ............................ 120



APPENDIX II TABLES ............................ 187


REFERENCES............................................. 295

BIOGRAPHICAL SKETCH .................................. 301


Figure Page

1 The structures of selected substituted
acridines and related compounds (Structures
I through XVI) 122

2 General form of Scatchard plots: R/C vs.
R for one and two classes of binding sites 128

3 Absorptiometric titration of 3-amino-
acridinium cation with DNA, cesium salt,
pH 5.90, 25.0C. Background electrolyte:
CsH2PO4, initial concentration, 0.15M 130

4 Absorptiometric titration of 3-amino-
acridinium cation with DNA, cesium salt,
pH 5.90, 25.OOC. Background electrolyte:
Cs H2PO4, initial concentration, 0.0025M 132

5 Absorptiometric titration of 3-amino-
acridinium cation with DNA, magnesium salt,
pH 5.90, 25.00C. Background electrolyte:
Mg(O2CCH3)2, initial concentration, 0.010N 134

6 Absorptiometric titration of 3-amino-
acridinium cation with DNA, magnesium salt,
pH 5.90, 25.0C. Background electrolyte:
Mg(O2CCH3)2, initial concentration, 0.00063N 136

7 Log([BHP]/[BH]) vs. log([Pt/31-[BHP]) with
0.0025M CsH2PO4 and 6.3 x 10- N Mg(02CCH3)2,
as supporting electrolytes, pH 5.90, 25.00C 138

8 Log([BHP]/[BH]) vs. log([Pt/31-.[BHP]) with
0.15M CsH2PO4, 0.010M CsH2P04, and 0.010N
Mg(02CCH3)2 as supporting electrolytes, pH
5.90, 25.00C 140

9 Log([BHP]/[BH]) vs. log([Pt/2 -2[BHP]) with
0.0025M CsH2PO4 and 6.3 x 10 N Mg(02CCH3)2
as supporting electrolytes, pH 5.90, 25.0C 142


Figure Page

10 Log([BHP]/[BH]) vs. log([Pt/21-2[BHP]) with
0.15M CsH2PO4, 0.01M CsH2PO4, and 0.010N
Mg(02CCH3)2 as supporting electrolytes, pH
5.90, 25.0 C 144

11 Log of the apparent association constant,
Ks, for the surface binding of 3-amino-
acridinium cation to DNA vs. t square
root of the ionic strength, I NaH2PO4
and LiH2PO4 as supporting electrolytes, pH
5.90, 25.0C 146
12 Log of the apparent association constant,
Kg, for the surface binding of 3-amino-
acridinium cation to DNA vs. t square
root of the ionic strength, I KH2PO4
and KO2CCH3 as supporting electrolytes,
pH 5.90, 25.0C 148
13 Log of the apparent association constant,
Ks, for the surface binding of 3-amino-
acridinium cation to DNA vs. the square
root of the ionic strength, 11/2. KH2PO4
and RbH2PO4 as supporting electrolytes,
pH 5.90, 25.00C 150

14 Log of the apparent association constant,
Ks, for the surface binding of 3-amino-
acridinium cation to DNA vs. the square
root of the ionic strength, Il/2. CsH2PO4
and (CH3)4NH2PO4 as supporting electrolytes,
pH 5.90, 25.0C 152

15 Log of the apparent association constant,
Kg, for the surface binding of 3-amino-
acridinium cation to DNA vs. the square
root of the ionic strength, 11/2. Mg(O2CCH3)2
and Ca(02CCH3)2 as supporting electrolytes,
pH 5.90, 25.00C 154

16 Log of the apparent association constant,
Kg, for the surface binding of 3-amino-
acridinium cation to DNA vs. the square
root of the ionic strength, 11/2. Sr(O2CCH3)2
and Ba(02CCH3)2 as supporting electrolytes,
pH 5.90, 25.00o 156



Figure Page

17 Log of the apparent association constant,
KI, for the intercalative binding of 3-
aminoacridinium cation to DNA vs. the
square root of the ionic strength, II/2
LiH2PO4 and NaH2PO4 as supporting
electrolytes, pH 5.90, 25.0C 158

18 Log of the apparent association constant,
KI, for the intercalative binding of 3-
aminoacridinium cation to DNA vs. the
square root of the ionic strength, 11/2.
KH2PO4 and KO2CCH3 as supporting
electrolytes, pH 5.90, 25.0C 160

19 Log of the apparent association constant,
KI, for the intercalative binding of 3-
aminoacridinium cation to DNA vs. the
square root of the ionic strength, 11/2.
KH2PO4 and RbH2PO4 as supporting
electrolytes, pH 5.90, 25.0C 162

20 Log of the apparent association constant,
KI, for the intercalative binding of 3-
aminoacridinium cation to DNA vs. the
square root of the ionic strength, 11/2.
CsH2PO4 and (CH3)4NH2PO4 as supporting
electrolytes, pH 5.90, 25.0oC 164

21 Log of the corrected association constant,
Ks, for the surface binding of 3-amino-
acridinium cation to DNA vs. the square
root of the activity of solution, 11/2.
LiH2PO4, NaH2PO4, and KH2PO4 as supporting
electrolytes, pH 5.90, 25.00C 166

22 Log of the corrected association constant,
Kg, for the surface binding of 3-amino-
acridinium cation to DNA vs. the square
root of the activity of solution, 11/2.
RbH2PO4, CsH2PO4, and (CH3)4NH2PO4 as
supporting electrolytes, pH 5.90, 25.OOC 168

23 Log of the corrected association constant,
Kg, for the surface binding of 3-amino-
acridinium cation to DNA vs. the square
root of the activity of solution, 11/2.
Mg(02CCH3)2 and Ca(02CCH3)2 as supporting
electrolytes, pH 5.90, 25.00C 170



Figure Page

24 Log of the corrected association constant,
Kg, for the surface binding of 3-amino-
acridinium cation to DNA vs. the square
root of the activity of solution, 11/2.
Sr(02CCH3)2 and Ba(O2CCH3)2 as supporting
electrolytes, pH 5.90, 25.00C 172

25 Absorptiometric titration of 7-amino-
quinolinium cation with DNA, potassium
salt, pH 5.90, 25.0C. Background
electrolyte: KH2PO4, initial concentra-
tion, 0.0050M 174

26 Log([AHP]/[AH]) vs. log([Pt/3]-[AHP]) for
the binding of 7-aminoquinolinium cation
to DNA at 15.OOC, 25.00C, and 35.OOC, pH
5.90, 0.010M KH2PO4 as supporting electro-
lyte 176

27 Log([AHP]/[AH]) vs. log([Pt/2]-2[AHP]) for
the binding of 7-aminoquinolinium cation
to DNA at 15.0C, 25.0C, and 35.0C, pH
5.90, 0.010M KH2PO4 as supporting electro-
lyte 178

28 Log Kg vs. 1/T( K) for the reaction between
3-aminoacridinium cation and DNA at 15.0OC,
25.00C, and 35.00C, pH 5.90, 0.010M KH2PO4
as supporting electrolyte 180

29 Log Ks vs. 1/T( K) for the reaction between
7-aminoquinolinium cation and DNA at 15.OOC,
25.OOC, and 35.OOC, pH 5.90, 0.010M KH2PO4
as supporting electrolyte 182

30 Log Ks vs. 1/T(K) for the reaction between
3-aminoacridinium cation and DNA at 15.0OC,
25.00C, and 35.OOC, pH 5.90, 0.010M KH2PO4
as supporting electrolyte 184

31 Log Ks vs. 1/T(oK) for the reaction between
7-aminoquinolinium cation and DNA at 15.0OC.
25.00C, and 35.0OC, pH 5.90, 0.010M KH2PO4
as supporting electrolyte 186


Table Page

1 Molar absorptivities of the 3-amino-
acridinium-DNA complex and the 7-amino-
quinolinium-DNA complex in various
supporting electrolytes 188

2 Absorptiometric titration of 3-amino-
acridinium cation with DNA, lithium salt.
Lithium dihydrogen phosphate as supporting
electrolyte 189

3 Absorptiometric titration of 3-amino-
acridinium cation with DNA, sodium salt.
Sodium dihydrogen phosphate as supporting
electrolyte 194

4 Absorptiometric titration of 3-amino-
acridinium cation with DNA, potassium
salt. Potassium dihydrogen phosphate as
supporting electrolyte 198

5 Absorptiometric titration of 3-amino-
acridinium cation with DNA, potassium
salt. Potassium acetate as supporting
electrolyte 202

6 Absorptiometric titration of 3-amino-
acridinium cation with DNA, rubidium
salt. Rubidium dihydrogen phosphate as
supporting electrolyte 207

7 Absorptiometric titration of 3-amino-
acridinium cation with DNA, cesium salt.
Cesium dihydrogen phosphate as supporting
electrolyte 212

8 Absorptiometric titration of 3-amino-
acridinium cation with DNA, tetramethyl-
ammonium salt. Tetramethylammonium
dihydrogen phosphate as supporting
electrolyte 216

LIST OF TABLES (Continued)

Table Page

9 Absorptiometric titration of 3-amino-
acridinium cation with DNA, magnesium salt.
Magnesium acetate as supporting electrolyte 222

10 Absorptiometric titration of 3-amino-
acridinium cation with DNA, calcium salt.
Calcium acetate as supporting electrolyte 226

11 Absorptiometric titration of 3-amino-
acridinium cation with DNA, strontium salt.
Strontium acetate as supporting electrolyte 230

12 Absorptiometric titration of 3-amino-
acridinium cation with DNA, barium salt.
Barium acetate as supporting electrolyte 234

13 Absorptiometric titration of 3-amino-
acridinium cation with DNA, potassium salt,
at 150C 239

14 Absorptiometric titration of 3-amino-
acridinium cation with DNA, potassium salt,
at 35C 241

15 Absorptiometric titration of 7-amino-
quinolinium cation with DNA, potassium salt,
at 150C 243

16 Absorptiometric titration of 7-amino-
quinolinium cation with DNA, potassium salt,
at 250C 245

17 Absorptiometric titration of 7-amino-
.quinolinium cation with DNA, potassium salt,
at 350C 247

18 Calculated values of [BH], [BHP], "Ks", and
related data for the reaction between 3-
aminoacridinium cation and DNA in 0.15M
CsH2 PO4 249

19 Calculated values of [BH], [BHP], Ks, KI, and
related data for the reaction between 3-
aminoacridinium cation and DNA in 0.0025M
CsH2PO4 251

LIST OF TABLES (Continued)

Table Page

20 Calculated values of [BH], [BHP], Ks, "KI",
and related data for the reaction between 3-
aminoacridinium cation and DNA in 0.010M
Mg(O2CCH3)2 253

21 Calculated values of [BH], [BHPI, Ks, "KI",
and related data for the reaction between 3-
aminoacridinium cation and DNA in 6.3 x 10-4N
Mg(O2CCH3)2 255

22 Percent total 3-aminoacridinium bound to DNA
in the presence of various electrolytes after
addition of excess DNA 257

23 Initial and final slopes of plots of log([BHP]/
[BH]) vs. log([Pt/m]-q[BHP]) in various con-
centrations of CsH2PO4 and Mg(O2CCH3)2 258

24 Apparent equilibrium association constants
for surface, Ks, and intercalative, KI,
binding of 3-aminoacridinium to DNA, LiH2PO4
as supporting electrolyte 259
25 Apparent equilibrium association constants
for surface, Ks, and intercalative, KI, bind-
ing of 3-aminoacridinium to DNA, NaH2PO4 as
supporting electrolyte 261

26 Apparent equilibrium association constants
for surface, K and intercalative, K bind-
ing of 3-aminoacridinium to DNA, KH2PO4 as
supporting electrolyte 262

27 Apparent equilibrium association constants
for surface, Ks, and intercalative, KI, bind-
ing of 3-aminoacridinium to DNA, KO2CCH3 as
supporting electrolyte 263
28 Apparent equilibrium association constants
for surface, K and intercalative, K bind-
ing of 3-aminoacridinium to DNA, RbH2 PO as
supporting electrolyte 265

29 Apparent equilibrium association constants
for surface, Ks, and intercalative, KI, bind-
ing of 3-aminoacridinium to DNA, CsH2PO4 as
supporting electrolyte 267


LIST OF TABLES (Continued)

Table Page

30 Apparent equilibrium association constants
for the surface, Ks, and intercalative, KT,
binding of 3-aminoacridinium to DNA, (CH3)4-
NH2PO4 as supporting electrolyte 269

31 Apparent equilibrium association constants,
Ks, for surface binding of 3-amino-
acridinium to DNA, Mg(OCCH )2 and Ca(O2CCH3)2
as supporting electrolytes 271
32 Apparent equilibrium association constants,
K for the surface binding of 3-amino-
acridinium to DNA, Sr(O CCH )2 and Ba(O2CCH3)2
as supporting electrolytes 273

33 Logarithms of the apparent surface binding
association constants for the binding of 3-
aminoacridinium cation to DNA, corrected
for ion activities 275

34 Equilibrium association constants for the
binding of 3-aminoacridinium to DNA, Ks,
and for the binding of alkali and alkaline
earth metal cations to DNA, KM 277

35 Apparent association constants for the
surface, Ks, and intercalative, KI, binding
of 3-aminoacridinium and 7-aminoquinolinium
to DNA at 15.00C, 25.0C, and 35.0C 279
36 Thermodynamic parameters for the surface
and intercalative binding reactions between
3-aminoacridinium and 7-aminoquinolinium
cations to DNA 280


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



Peter F. Eisenhardt

August, 1977

Chairman: Stephen G. Schulman
Major Department: Pharmaceutical Chemistry

Absorptiometric titrations of a model drug compound,

3-aminoacridinium cation, with salts of calf thymus DNA

were carried out at pH 5.90 in a series of solutions in

which the supporting electrolyte was the dihydrogen

phosphate salt of one of the following: lithium, sodium,

potassium, rubidium, cesium, or tetramethylammonium ion.

Titrations were also conducted in solutions of the acetate

salts of magnesium, calcium, strontium, and barium. The

ionic strengths of the solutions were varied between 0.25

and 0.00063(M) for each alkali metal electrolyte and between

0.036 and 0.00010 for the alkaline earth electrolytes.

Apparent equilibrium association constants for the surface


binding and for the intercalative binding of 3-amino-

acridinium cation to DNA were computed for each concen-

tration of each alkali metal electrolyte solution and

are reported. Apparent constants for the surface binding

of 3-aminoacridinium in the presence of alkaline earths

are also presented. Values for the intercalative binding

mode in the presence of group IIA metal cations could not

be determined.

The extent of binding of the model drug to DNA was

dependent on the nature and concentration of the electrolyte.

Essentially complete binding was attainable for all concen-

trations of all alkali metals while as little as 70% binding

was possible in the presence of 0.010N magnesium. Spectral

characteristics of the drug-DNA complex were also dependent

upon the nature of the background electrolyte and are dis-


The reaction systems did not behave in accordance with

limiting Debye-HUckel relationships over the range of ionic

strength investigated. After applying extended Debye-

Hiickel corrections, the relationships between the surface-

binding equilibrium association constants and the ionic

strength indicated that specific-ion competition between

metal and drug cations, for DNA phosphate sites, was signifi-

cant. A simple 1:1 competitive model was evaluated.

At ionic strengths less than about 0.010, experimental

precision was poor, presumably due to alterations in the

conformation of DNA at low ionicities, suggesting that


evaluations of interactions between small molecules and

DNA be conducted in solutions of ionic strength in excess

of 0.010.

Thermodynamic parameters for surface and intercalative

binding of 3-aminoacridinium cation and its benzolog, 7-

aminoquinolinium, were evaluated from equilibrium association

constants at 15.00C, 25.0C, and 35.00C. The standard

enthalpy change, AHO, for the 3-aminoacridinium binding

was less negative than AHo for the 7-aminoquinolinium process,

while the standard entropy change, AS0, for the former was

more positive than for the latter. Thus, the greater degree

of disordering resulting from the relaxation of the solvent

around the tricyclic 3-aminoacridinium cation as it binds to

DNA appears responsible for the more negative standard free

energy change for the 3-aminoacridinium binding process.

Values of AS0 for the intercalative binding of the two drugs

suggest that significant ordering of the polymer helix occurs

upon intercalation.




The interaction of small molecules with DNA and

biopolymers, in general, has been the subject of intense

investigation during the past several decades. Interest

in this area has been due, in large part, to the early

realization that such interactions may result in muta-

genesis (1) and carcinogenesis (2), especially when the

small molecules are cationic aromatics such as substituted

acridines. Numerous publications deal with various aspects

of the interactions of cationic aromatics with DNA, some

of which will be briefly reviewed here. The specific

systems to be considered in this work include 3-amino-

acridine(I). Hence,-a summary of representative and

pertinent research involving aminoacridines is included.

Moreover, we have investigated the effects of metal ions

on the interactions of 3-aminoacridine with DNA and,

therefore, shall review the past research efforts in this

area. It is emphasized that this review is, by no means,

Roman numerals in parentheses after chemical names refer to
their listings in Figure 1 of appendix I.

exhaustive,but rather, is intended to provide perspective

and an introduction to the work that is included in this


Binding of Small Molecules to Biopolymers

Probably the first acknowledgement of a significant

interaction between a member of the acridine chemical

family and a biological system was reported in 1913 by

Ehrlich and Benda (3) when they reported their observation

on the antibacterial activity of 3,6-diamino-10-methyl-

acridinium chloride(II) against Trypanosonoma brucei.

Subsequently, evidence accumulated that the antibacterial

activity of substituted aminoacridines is dependent on

their capacity to compete with hydronium ion for a vital

anionic site in a bacterium (4-6, 7: Ch. 2) and that

antibacterial activity and mammalian toxicity were not

directly related (8,9). The antibacterial action of

proflavine(III) is reported to be due to its ability to

inhibit DNA-dependent RNA synthesis (10) which suggests

that DNA may contain the anionic site. However, it is

inadvisable, at this time, to make a direct correlation

since proflavine also binds strongly to a variety of other

cellular structures, any one of which may contain the

vital anionic site (11).

Numerous amino-substituted acridines are known to be

carcinogenic while few simple acridines are cancer inducing

(12) which may be due to the former's greater ability to

intercalate into DNA (13). It is worthy of note that

benzacridines(IV) and dibenzacridines(V), first exten-

sively studied in 1935, are nearly all highly carcino-

genic (2). Two possible modes of action have been

proposed. The benzacridine may be metabolically activated

to a cationic form which then reacts as an electrophile

toward nucleophilic centers of DNA, RNA, and protein (14),

or the unmetabolized species may interact physically with

the DNA (15), perhaps by intercalating into the double

stranded helix.

Mutagenesis, presumably arising from the intercalation

of acridines with nucleic acids, was first reported by

DeMars in 1953 (1 ) in relation to his work on proflavine

and T2 bacteriophage (Escherichia coli). The mutagenic

activity of proflavine does not involve its integral

incorporation into the newly synthesized polynucleotide

chain but rather, results from the acridine's capacity to

cause deletions and insertions of nucleotides into the

DNA polynucleotidq (16,17). In fact, proflavines' capacity,

in this regard, was used in 1961 to discover the triplet

nature of the genetic code.

In searching for effective anticancer and antitumor

agents, advantage is taken of the fact that compounds

which may interfere with cell replication at the molecular

level may have medicinal' utility. As a result, a large

number of acridine derivatives have been investigated, only

to find that, while a select few do have antineoplastic

properties, their associated toxicities to the host make

them less desirable than other available compounds. On

the other hand, aminoacridines have enjoyed success as

antimalarial agents and are still in limited use today.

Investigations were begun in the 1920's for a substitute

for quinine, as an antimalarial, which led scientists

first to methylene blue, a phenothiazine dye, then to

pamaquine, both of which had low effectiveness against

acute stages of malaria. Structural similarities in the

above two compounds, with substituted acridines, eventually

led investigators to quinacrine(VI) which became the drug

of choice throughout World War II. Later, the more effec-

tive and less toxic chloroquine(VII) was discovered.

Presently, considerable synthetic research is devoted to

acridine-derived antimalarials to supplant chloroquine

and to combat chloroquine-resistant P. falciparum --

discovered in 1961 (18). To date, the most promising

compounds of this type have a diamine side chain in the 9-

position(VIII). Structure-activity relationships for a

variety of quinacrine analogues having side chain variants,

terminal amine variants, and aromatic substituent variants,

as well as quinine related acridine antimalarials, are

tabulated by Henry (19).

In light of the broad biological significance of the

binding of aminoacridines to DNA and other biopolymers,

in vitro and in vivo studies aimed at understanding the

exact nature of the interactions have been undertaken.

The following paragraphs will briefly review several of

the more important articles germane to our investigation.

Those dealing with the modes of binding will be outlined

first, followed by studies of the effects of the ionic

medium (including possibly competing ions), with a review

of papers involving theoretical discussions, third.

An excellent comprehensive article on the interactions

of aminoacridines with nucleic acids and the methods by

which they may be studied was published in 1956 by A.R.

Peacocke and J.N.H. Skerrett (20). Another, in 1968, by

Blake and Peacocke (21) considered developments in the

field after 1956.

Aminoacridines have long been known to bind to

double-stranded DNA via two distinct modes -- termed type

I and type II binding. Type I is the stronger and was

first proposed by Lerman (22), to be an internal or

intercalative process. Insertion of a molecular ion is

accomplished by transient rotation of base pairs allowing

the flat, lipophilic, aromatic portion of the molecule to

become situated between them. The charged portion of the

molecule was considered to remain external to the helix.

H.J. Li and D.M. Crothers (23) corroborated this model

when they presented indirect evidence that the base pairs

remain perpendicular to the helix axis upon intercalation

of the molecule. Another model for type I binding was

advanced by N.J. Pritchard, A. Blake, and A.R. Peacocke (24)

in which the small molecule is intercalated between adjacent

bases on the same polynucleotide chain. In both of these

models, the cationic heteroatom is associated with an

indeterminate number of anionic phosphate moieties along

the DNA backbone. The viscosity of heat-denatured DNA

is essentially unaffected by binding of acridines (25)

whereas the viscosity of native double-stranded DNA is

increased. An increase in viscosity is indicative of a

lengthening of the polymeric chain such as would result

from intercalation between base pairs or between adjacent

bases. Since the single-stranded DNA undergoes no vis-

cosity change, the Lerman model appears more plausible.

Some specific geometrical parameters of substituted

acridine-DNA intercalative complexes were established by

G.R. Kelly and T. Kurucsev (26) through the evaluation of

linear dichroic spectra of stretched DNA films. An advan-

tage of films over conventional methods involving aqueous

solutions is that they do not require large ratios of drug

to DNA phosphate. Taking advantage of the dichroism of

two mutually perpendicular polarized transitions of the

9-aminoacridine nucleus, the workers were able to assess

the angles of tilt and twist of the plane of the dye

relative to the polymer helix. The long axis of acri-

flavine(II) was essentially perpendicular to the helix

axis (85 + 5o), proflavine(III) was tilted slightly (77 +

2 1/20), while 9-aminoacridine(IX) was tilted significantly

(67 + 30).

Type II, surface, or external binding has come to

be generally accepted as an association of the small

molecules with the exterior surface of the DNA polyanion.

This interaction is, consequently, more strongly affected

by alterations of the solvent system than is type I

binding. Li and Crothers (23) determined, on the basis

of temperature-jump kinetic data, that proflavine binds

both internally and externally to calf thymus DNA and

only about 7% of the total amount bound binds externally

in the presence of 0.2M Na+ (sodium phosphate sodium

nitrate). Decreasing the sodium ion concentration to

0.02M results in about 30% of the total being externally


A careful study of the intercalative binding of a

series of aminoacridines was conducted by Drummond and

co-workers (27). Binding decreases in affinity in the

order: acranil(X)>neomonacrin(XI)>atebrin(VI)>9-amino-

acridine(IX)>proflavine(III) 9-amino-l,2,3,4-tetra-

hydroacridine(XII). This sequence shows that side chain,

and ring, substituents are not predominant in influencing

binding affinity. The effect of greatly decreasing the

planar area of 9-aminoacridine by hydrogenation is

evident upon comparing its position with that of XII. The

authors discuss various geometries of binding and conclude

it is not necessary that there be exact and complete

intercalation for type I (strong) binding. Rather, a

modified Lerman model in which the charged heterocyclic

ring nitrogen interacts with a DNA phosphate while the

rings only partially interact with the bases is more in

accord with observed behavior.

It has been shown that the degree of intercalation

may be dependent on the base composition of the DNA.

Heterogeneity of intercalative sites was demonstrated

by J.C. Thomes, G. Weill, and M. Daune (28) using

fluorescence quenching techniques. They demonstrated

that between 2% and 3% of the base pairs form sites where

proflavine is strongly bound while the remainder have

binding constants 3 to 4 times weaker. Proflavine

fluorescence is relatively unaffected in the former case,

but is totally quenched in the latter. The stronger

sites correspond to adenine-thymine rich regions of the

DNA indicating proflavine's specificity for the base

pair. Similarly, Ramstein and Leng (29) showed that the

location of bound proflavine, within the DNA, was dependent

upon the base composition. The absorptiometric and fluori-

metric titrations of DNA and partially methylated DNA

( 18%, mainly on N7 of the guanine residues) indicated

acridine was intercalated in both the native and methylated

sites. However, greater apparent equilibrium constants for

the methylated DNA than those for the native type were ob-

served at two ionic strengths: 14.25 x 104 M- and 43.5 x

104M-1 in 0.1M NaCl and 40 x 104 M1 and 83.5 x 10M-1 in

0.01M NaCl, respectively. These differences may be due to

an increase in the distance between base pairs arising from

the presence of the methyl groups which may, in turn, result

in the shift in the equilibrium concentrations. S.

Georghiou (30) obtained much the same results as Ramstein

and Leng based on the fluorescence decay of proflavine

when it was bound to DNA at a phosphate to drug ratio of

420. Gabbay and co-workers (31) invoked the "Bookmark"

model of intercalation to postulate ten possible distinct

sites of binding of N-substituted-N-methylphenanthrolinium

cations(XIII) to nucleic acids. They also investigated

the topography of some nucleic acids in solution using

PMR (32).

Binding models for the interactions of acridine

orange(XIV) and proflavine(III) with DNA at ionic strengths

of 0.002M, 0.020M, and 0.20M (tetramethylammonium caco-

dylate buffer, pH 6.5, 22C) were proposed based on

thermodynamic, spectroscopic,and hydrodynamic properties

of the systems (33). The methods precluded consideration

of surface binding. Binding isotherms were generated from

equilibrium association constants which were, in turn,

obtained from theoretical mass action relationships.

Additional information, useful in developing binding

models, was afforded by Goswami, Das, and Das Gupta (34)

who measured the decrease in the static dielectric con-

stant of a proflavine-DNA complex while decreasing the

DNA phosphate/proflavine (P/D) ratio. The authors con-

firmed that: the bound dye neutralizes the charge on the

DNA polyanion; at high P/D values, intercalation predomi-

nates and; that when P/D becomes small (10), a cooperative

increase in the electrostatic binding of phosphate groups

brings about a rapid decrease in the total surface charge

of the polymer. In brief, at P/D values between 10 and

100 most of the cations are intercalated with a concomi-

tant elongation of the polyanionic DNA helix, whereas

below P/D of 10, surface binding is predominant. In the

latter case, the spatial extension of the helix is de-

creased due to charge neutralization with no accompanying

expansion due to intercalation. These results indicate

that alterations in the dielectric constant of a medium

containing such a system would manifest dissimilar effects

on the two equilibrium association constants.

The specific effects of cations, other than the dye

species, on the reaction between DNA and a dye is an

integral part of this investigation. Therefore, a brief

consideration of articles dealing with the binding of

metal cations to DNA and related topics is in order. A

tabulation of sites and thermodynamic quantities associated

with proton and metal ion interactions with DNA, RNA, and

their constituent bases, nucleosides, and nucleotides has

been published by Izatt et al., (35).. H. Sigel and D.B.

McCormick (36) provide an introduction to the fundamental

theories concerning metal-ligand interactions in biological

systems, including discussions of the Irving-Williams

series, competition of metals and protons for sites, the

formation of ternary complexes, and effects of solvent

polarity and ionic strength. The Pearson theory (37) of

hard and soft acids and bases is applied to metals and

polymers in relation to the biological roles of Na K ,

Ca ++, and Mg and to the carcinogenicity of heavy metals


The relative effectiveness of ions in altering the

conformation of macromolecules via specific ion inter-

actions is termed the Hofmeister, or lyotropic, effect

(39). Generalizations based on observations of simple,

ideal polymers may be confidently extrapolated to more

complex polymers since it has been shown that the

Hofmeister effect of a given ion is remarkably independent

of the nature of the macromolecule. Moreover, while

changes in the entropies of polymer segments are affected

by both anions and cations, as are changes in the partial

molal volume, the partial molal internal energy of a segment

is dependent only on cations. Considering the above,

M.J. Hey, J.M. Clough, and D.J. Taylor (39) reported that

the ion binding strength with protein is (in decreasing

order of strength) NH4
This series may be useful in separating solvent effects

from ion binding effects where the above cations are

present along with 3-aminoacridinium cation and DNA.

Stabilizing effects of various ions on four biopoly-

mers, including DNA, were evaluated by Hippel and Schleich

(40). The effects were discussed in terms of free energy

of transfer for model compounds, activity coefficient

variations, and correlations between ion effects on

macromolecules and on water structure. The relative

effectiveness of the ions studied, in increasing the

stability of the native configuration of DNA, paralleled

the Hofmeister series: (CH3)4N >K >Na >Li and Cl Br >

CH3COO >C104 >CNS It was found for a series of

tetraalkylammonium salts, the greater the length of

the alkyl chains, the lower the stabilization of the

native form of DNA.

Hen and Strauss (41) employed equilibrium dialysis,

dilatometry, and viscometry to assess the counterion

binding of a series of cations with poly(vinylsulfonate).

The Hofmeister series was maintained (in order of de-

creasing binding affinity): Ag >K >Na >H ,Li >(CH )4N

and Ba >Mg All of the cations except H and Li

are capable of cross-linking the polyanion through

simultaneous site binding at two sulfonate groups.

The specific interactions of univalent and di-

valent cations with calf thymus DNA can be correlated

with their respective unhydrated ionic radii (42,43)., /

The order of most weakly bound to most strongly bound\
+ + + + ++ ++ ++
is: (CH3)4N
binding parameters with DNA with polyphosphate as polyanion,

the authors showed that the sites on the biopolymer were

the phosphates rather than bases or base pairs. These

results were in agreement with those of other workers


Conductometric titrations of DNA with divalent

cations were first performed by J. Shack, R.J. Jenkins,

and J.M. Thompsett (46) who found a sharp end point cor-

responding to 0.8 equilvalents of Mg and Ca per mole

of polymer phosphate. Similar titration behavior was seen

for poly A and poly U by Felsenfeld and Huang (47) who

developed an improved binding model which showed that the

apparent stoichiometry of 0.8 was in error and that the

true equivalence point corresponded to 1.0 eq of Mg or

Ca++ per mole of DNA phosphate. The authors also disproved

the previously held belief (48) that dications are more

strongly bound to purines than pyrimidines in a poly-


It has been reported (49) that there is binding to

base pairs of the DNA helix as well as to phosphates. The

investigators reported that the effects of adding Mg to

heat-denatured calf thymus DNA was to produce an absorption

spectrum similar to that of denatured DNA alone, except

for reduced absorbance. However, when Mg was added to

native DNA, which was then denatured, a spectral shift to

longer wavelength occurred. The authors concluded, on

these grounds, that the metal was binding only to phos-

phates in the former case while in the latter, there was

interaction with the aromatic nitrogens of the bases. The

bonding between Mg and the nitrogens is presumably not

able to occur unless the hydrogen bonds between the bases

are broken in the presence of the metal.

Mathematical relationships incorporating changes in

activities of ions in solution and deviations in activity

coefficients were developed by Lyons and Kotin (45) to

qualitatively distinguish between specific (site) binding

and non-specific (diffuse electrostatic) binding of

metals to polyanions. The rule of additivity of the

activities of countercations of solutions of polyelectro-

lytes and simple salts was re-examined. It was found

that, under certain conditions, there is significant

exchange between the counterions associated with the

polyanion and those of the bulk solution. A decrease in

the activity coefficient of sodium ion, upon dilution of

the system, was observed, indicating increased binding at

lower concentrations. In the case of magnesium the results

were not as simple but suggested that at high magnesium

ion concentrations, specific site binding was predominant

while at low concentrations, non-specific binding obtained.

The differences between Na and Mg may be due, in part,

to the ability of the latter to bridge between two adjacent

anionic sites on the polymer whereas the monovalent cation

cannot. Note that the distance between adjacent phosphates
along the DNA backbone is approximately 7 A. C.Sander and

and O.P. Ts'o (50) believe that the binding of Mg++ to DNA,

RNA, poly A, poly A.U, poly I, poly I.C, and denatured DNA

can be rationalized using a linear Scatchard relationship

videe infra). Their specific site model was applicable for

up to about 70% saturation.

The degree of binding of acridine orange to native

and denatured DNA was determined by equilibrium dialysis

in 0.1M and 0.001M NaCI at 200C (51). Three successive

stages of binding were observed: the first corresponding

to intercalation of monomers; the second, formation of

bound dimers; and the third, external binding of aggre-

gates. The appearance of the three modes of binding

compared to two for most aminoacridines is attributed

to acridine orange's proclivity to dimerize. The dialysis

data was augmented with absorption and fluorescence

spectra, electric dichroism, electric birefringence, and

circular dichroic spectra.

Scruggs and Ross (42) found that the intrinsic

viscosity of three types of DNA (salmon sperm, calf thymus,

and T4 phage) decreases in the presence of univalent cations

as the ionic strength increases, attaining separate limiting

values for each. Prior to their work, the apparent insen-

sitivity of the viscosity of DNA to variations in ionic

strength was considered anomalous because polyelectrolyte

viscosities are usually highly dependent on salt concentra-

tion. The workers asserted that previous invariance was

due to the presence of trace amounts of basic proteins,

polypeptides, and polyamines and that DNA was, in fact,

similar to all other polyelectrolytes.

S. Bram (52) investigated the secondary structure of

DNA in solutions of varying concentrations of Li Na ,

Rb and Cs using X-ray diffraction spectroscopy.

Differences between scattering patterns by sodium DNA

in solution and those expected from theoretical calcu-

lations based on an unperturbed B form of DNA were due

to structural changes in the B form arising from the

polymer's interaction with the species in the medium.

This is in concert with others' (53-55) conclusions that

the number of base pairs per turn of the DNA helix is

a function of the ionic environment. For cesium DNA

the extrapolated lengths for 10 nucleotide pairs are
o o
43.5 A at zero ionic strength and 34.6 A at finite CsCl
concentration, whereas in NaCl, the values are 33.5 A
and 32.0 A, respectively. (A novel DNA conformation

model which differs markedly from the classical Watson-

Crick B form is proposed by Stig Erlander (56) who argues

that the differences between the values given above cannot

be rationalized using the Watson-Crick model.) A theo-

retical treatment of cooperative binding was advanced by

G. Schwarz (57) in which nearest neighbors along a linear

polymer interact with one another. Two types of intrinsic

internal binding are assumed: (1) that of an isolated

ligand (nucleation); (2) that of a ligand to a site

immediately adjacent to one already occupied. The model

also includes considerations for dimer formation and

competitive (non-cooperative) binding at adjacent sites

by unrelated ligands. Matrix methods of calculating

species concentrations under equilibrium conditions are


An understanding of the effects of ionic strength

on the binding of small molecules to DNA requires at

least some knowledge of the charge distribution on the

surface of the polyanion. Electrophoresis and membrane

equilibrium experiments allow the determination of the

apparent fractional charge per DNA phosphorous, i, which

is an index of net charge (58). The value of i increases

from 0.250 to 0.34 on going from a solution containing

0.005m NaCl to one having 0.05m NaCl, showing that the

net charge of the polyanion increases with increasing

ionic strength. Ross (58) considered the above information

in light of Gorin's model of a long rigid cylinder (59)

and applied it to DNA, which is intermediate between a

rigid cylinder and a random-coil polyelectrolyte. Ross'

data substantiates Gorin's assumption that a rigid cylinder

(and, therefore, DNA) obeys Debye-Hiickel approximations.

Manning (60) postulated that, as the ionic strength

approaches zero, the effective charge of a polyelectrolyte

is maintained at a critical constant by condensation of

counterions along its surface as long as the formal charge

does not exceed a critical value. His predictions were

borne out by independent investigators (33) who found

that at low ionic strength (0.002M) the electrophoretic

mobility of DNA was unaffected by bound acridine orange.

Lerman's model for intercalation of small molecules

into DNA satisfies many experimental observations but

leaves some uncertainties unresolved arising from

consideration of the energetic involved (61). One of

the most perplexing problems is the strong dependence

of the intercalative mode of binding on the ionic strength

(albeit not as dependent as surface binding). Also, the

existence of a maximum degree of association as a function

of ionic strength requires explanation. Attempts have

been made to resolve these problems by invoking "stacking

energies" rationalized in terms of short range, ionic

strength independent interactions between DNA base pairs

and the aromatic portions of the acridines. These ratio-

nalizations, however, are incompatible with the fact that

aromatic hydrocarbons intercalatively bind only very weakly.

M. Gilbert and P. Claverie (61) point out that the solution

energies of the intercalating compounds parallel their

binding affinities and, therefore, may be at least as

important as the stacking energies. Moreover, the limiting

of intercalation processes long before all available sites

have been occupied mandates a limitation mechanism inde-

pendent of the heterogeneity of sites. The critical effect

of transferring positive charges from solvent "cavities,"

having dimensions on the order of the small molecules',

into large cavities interior to the lipophilic DNA must

be accounted for in a model of the dye-DNA complex -- as

it is in their model.

T. Herskovits (62) studied the relationship between

electrolyte concentration and solvent denaturation transi-

tions in various media. Both increasing chain length and

increasing hydrocarbon content increase the effectiveness

of the denaturant. It was concluded that these observations

demonstrate the importance of hydrophobic forces in main-

taining DNA's aqueous configuration. In most cases, the

denaturation was reversible with added electrolyte.

It is Bradley's opinion (63) that by using a combina-

tion of statistical and quantum mechanical methods, it

should be possible to relate all absorption, emission,

equilibrium constant, and optical rotation observations

in a unified way. With this aspiration in mind, he

presented a statistical model of dye-polymer binding as

a step toward that goal. He assumed that specific polymer

sites were in a linear array and dye molecules were bound

to them by simple electrostatic forces, with possible

enhancement by neighboring sites. Furthermore, the sites

were sufficiently close to allow bound dye-bound dye

interactions with concomitant changes in their absorption,

fluorescence, and phosphorescence spectra. The distribution

of N consecutive sites occupied by dye molecules, in their

varying states of aggregation, can be computed theoretically

in terms of either the ground state free energy of inter-

action of a pair of neighboring dye molecules, AF, or in

terms of a stacking coefficient, K, related to AF,

K = exp(-AF/kT) (1-1)

The AF and K terms may be equated to an experimentally

obtainable ratio of P sites per D dye species,

P/D = (1-F1/2) -1(1+(K-1)F) (1-2)

Values of K range from 1.25 for DNA to 826 for poly-

phosphate (K = 6.2 for denatured DNA). There is

evidence that the value of K is directly related to

the conformation of the polymer. The fact that K is

equal to 1.25 and not to 1.00 for DNA -- as it would be

for a completely random site occupancy -- indicates some

non-randomness which could arise from dye-dye interaction

along the polymer. A simplified quantum mechanical

treatment is presented to allow computation of the magni-

tudes of the relationships between the dye molecules of

filled units with empty sites.

Kinetics of intercalation of aminoacridines into DNA

were first studied in depth by Li and Crothers (23) in 1969.

Employing data from temperature-jump relaxation methods,

they proposed a relatively simple two-step process for

intercalative binding. It is worth noting that static

experiments will not allow a separation of these two steps

of the intercalative mode. The data fit the mechanism

k12 23 P
P + DNA k (P) out (P) (1-3)
kout <-k--- in
k21 32
where P is proflavine and (P)out and (P) in are surface

bound and intercalatively bound dye, respectively. The

overall insertion reaction occurs in the millisecond time

range and is first order from the (P)out state. The two-

step hypothesis is substantiated by comparing the large

relative thermodynamic and spectral changes of the

exterior binding with the small changes for the interior-

bound species that occur upon glucosylation of the DNA

videe infra). Two limiting mechanisms are advanced and

alternate mechanistic pathways discussed.

Passero, Gabbay, and others (64) used a simplistic

model which included competitive effects of metal ions

in solution to arrive at apparent association constants

of two reporter molecules (XIIIa, XIIIb) to DNA. Electro-

static potentials and interactions between near neighbors

were ignored, as were any forms of surface binding by the

reporter molecules. The overall binding process was

expressed as two separate equilibria

R + P --- RP (1-4)


M + P -> MP (1-5)

where R, M, and P are free reporter, metal, and DNA

phosphate, respectively, and RP and MP the corresponding

bound species. An apparent binding constant, K R may be

defined relative to the metal ion concentration

K R = [RP] / [R] [P] (1-6)

Finally, assuming the binding of R or M to the polyanion

results in only one phosphate being affected per R or P,

the apparent association constant may be expressed in terms

of association constants for reporter binding, K R, and

metal binding, KM

K R = KR / (1 + KM [M+]) (1-7)

The authors maintain that their assumptions are valid

by virtue of the fact that they obtain reasonably

constant values of KR and KM under a variety of conditions.

However, if the reporter molecules are intercalatively

bound, as the authors assert (and they are probably

correct) then it is not valid to compare the reporter

and metal binding because two entirely different types

of sites are being occupied. The highly hydrophilic,

symmetrically charged metal cations are almost certainly

surface bound. Thus, the constancy of the results are

probably fortuitous or, at best, important deviations

are being masked by experimental error.

Relaxation kinetics of ethidium bromide(XV) binding

to DNA revealed three types of bound species at equilib-

rium (65). Aside from the conventional surface-bound and

intercalated species ascribed to type II and type I binding,

a third species was found which involved an ethidium bromide

transiently attached to sites of two separate DNA strands.

The bimolecular rate constant for transferring an ethidium

bromide from one DNA phosphate to another is 3 to 6 times

larger than the rate of intercalation of free molecules.

The more rapid bimolecular mechanism is dependent on the

small molecule's ability to bind to two sites simultaneously.

In the case of ethidium bromide, the tricyclic region of the

molecule provides one site while the phenyl moiety is

believed to provide the second, less favorable, site.

Divalent-cation-specific electrodes were used to

study the binding of Mg to DNA (50) in a solvent
containing 5 x 10-3 M pH 7 phosphate buffer. Results

were treated in terms of site-binding to phosphates in

a multiple equilibrium process, employing a Scatchard

treatment, which showed linearity over a range of 30%

to 70% saturation of sites. Approximations to the

Scatchard method yielded an apparent intrinsic binding

constant of 6 x 103 (AG = 5.2 Kcal/mole). The use of

cation-specific electrodes mandated DNA phosphate concen-

trations in excess of 5 x 10-3M, causing aggregation of

DNA and precipitation of MgDNA. Thus, the values of the

constants may include significant error.

Three methods for evaluating equilibrium association

constants for the binding of small molecules to DNA will

be outlined. They are, the classical and most commonly

employed Scatchard method, a modified Scatchard model,

and a mass-action approach which will be used for this

investigation. Scatchard's original treatment was designed

for binding of small molecules to proteins. In Scatchard's


If the various groups on a protein molecule
act independently, we can apply the Law of
Mass Action as though each group were on a
separate molecule and the strength of binding
can be expressed as the constant for each
group. Often a single constant will express
the behavior of several groups. (66:660)

If the groups do act independently of one another, changes

in free energy for the reaction of the protein with

small molecules is made up of the statistical entropy

terms for v plus a term proportional to V. When the

total number of sites (or groups of sites) is large, the

calculations become tedious -- if the total number is

not known the calculations cannot be performed. Such

a situation does arise if one attempts to apply the

equations without modification to nucleic acid binding.

Since DNA polyanions commonly have molecular weights

in excess of one million and may contain as many as 106

phosphates (67) Peacocke and Skerrett (20) redefined the

parameters of the Scatchard equations to allow their

application to nucleic acids. Their assignments and an

abbreviated derivation of the Scatchard equations are as


Let R represent the amount of small molecules, D,

bound per mole of nucleic acid phosphorus (generally a

fraction), and C represent the molar concentration of

free D. Consider a number, n., of P classes of binding

sites for D on the DNA polyanion. If each class has

associated with it an intrinsic binding constant K., then

we may write for any general case

J=P n.K.C
R = E 1 1 (1-8)
J=l 1 + KjC

This may be simplified when all sites are identical

R/C = Kn Kr


If there are two distinct types of binding sites, I

and II, then

n K C nii K C
R = 1 + K C + 1 + KIC (1-10)

Figures 2a and 2b are representative plots of R/C vs C

for one and two types of sites, respectively. Curvature

may result in the case of only one type of site from

variations in the electrostatic free energy term which

is dependent on R. Such non-linearity can become critical

when the intercalating species bear opposite charges but

may be partially obviated by replacing KI with the term

K I exp (AG0 /(RT) (1-11)

where K I is independent of R, and AG r is an electro-

static free energy term dependent only on the total effec-

tive potential. The case in which there are two types of

binding sites is of primary interest in this work. The

stronger, intercalative binding corresponds to type I of

figure 2b whereas surface binding is represented by

region II. With two types of sites, it is sometimes

possible to discern distinct linear portions in the plots

of R/C vs R as suggested in the idealized figure. Only if

the magnitude of one of the constants is about 10 to 100

times greater than the other, can reliable values of the

individual association constants be obtained by extrapola-

tion of the data in each of the linear regions. Even then,

potentially erroneous assumptions regarding changes in

the electrostatic free energy must be considered. More-

over, as previously stated, the proposed model does not

allow for any form of interaction between the two types

of binding sites. This is not generally serious in

protein binding where the sites are often noninteracting

but can be a problem in the case of the nucleic acids.

Factors such as cooperative binding and near-neighbor

exclusion will introduce sigmoidal characteristics into

the plots of R/C vs R (21). Another shortcoming of the

Scatchard method is that extrapolation of the legs of

the hyperbolic plot to obtain slopes and intercepts

often places an inordinately high degree of reliance

on the data at the extrema of titrations. It is at the

extrema that the data, regardless of how it was obtained,

will be the least precise.

For aminoacridines, intercalative binding pre-

dominates up to a value of R <0.20 corresponding to 4 or

more phosphates per bound cation. Surface binding, on

the other hand, is predominant for values of R between

approximately 0.5 and 1 for compounds which do not dimerize

and up to 2 or more for those that are known to aggregate

as free species at moderate concentrations.

Numerous investigators have suggested improvements

of the basic Scatchard method as modified by Peacocke and

Skerrett. Included are corrections for electrostatic

effects (27,31,61,63), competition of metallic ions in

solution (45,46,67), near-neighbor effects (23,50,57,

63,65), small molecule aggregation (51,57), and configura-

tional effects due to solvent and ionic factors (23,25,2.8,

39,44,52,62,68). One of these improved models, proposed

by Armstrong, Kurucsev, and Strauss (33) incorporates

some novel features including: (1) the total concentration

of intercalation sites available is fixed, a priori, to

include every slot between successive DNA base pairs; (2)

an intercalated dye inhibits intercalation at its two

immediately adjacent slots; and (3) inclusion of a term

to allow for an intercalated monomer forming a dimer with

a free monomer resulting in a spectroscopically distinct

species. The authors found (3) became important when the

degree of binding, 8, exceeded about 0.2 mole of dye per

mole of DNA phosphate (dyes were acridine orange and pro-

flavine). Their expressions for the association constants,

based upon theoretical mass action relationships are, for


2(081-2) (1-281)
K1 = 2 (1-12)

and for surface binding

K2 = ( ) (1-13)
2 CM(81- 2

where 81 = the number of intercalated molecules per DNA


82 = the number of externally bound dimers per

DNA phosphate.

CM = the molar concentration of free dye monomer.

K1 = the apparent binding constant for the monomer

species intercalating.

K2 = the apparent binding constant for surface


These relationships imply that the electrostatic potential

of the DNA is unaffected by binding. Values of 81 and 82

may be spectroscopically evaluated knowing the molar

absorptivities of the bound monomer, eM, and the bound

dimer, ED,

T ( M M + CDD/ (1-14)

where M = 1-82, 8D = 282, and eT is the total "molar

absorptivity" of all bound species. The model was tested

by determining the binding of acridine orange and pro-

flavine to DNA (pH 6.5, 22C) at ionic strengths of 0.002m,

0.020m, and 0.200m. The above method is an improvement

over the Scatchard procedure but still demands a priori

selection of an exact mode of reaction. That is, though

the overall stoichiometries are experimentally evaluated,

the thermodynamic manner in which the free species react

is not defined. For example, in the intercalative case,

where 4 phosphates are involved for each 3-aminoacridinium

binding site, are the 4 phosphates to be considered as 4

separate entities, as 2 separately interacting moieties,

each containing 2 phosphates, or as one entity containing

all 4 phosphates? In each case, the overall stoichiometry

is 4 to 1, but the equilibrium expression of the first

includes a fourth power term in DNA phosphate concentration,

the second would have a squared phosphate concentration

term, while the third would be linear in phosphate.

To resolve this problem, Capomacchia (69) has proposed

a mass-action approach based on best-fit analyses of

spectroscopically derived concentrations of free and bound

species. Assuming that the binding process obeys the Law

of Mass Action and total concentrations of all species are

sufficiently low that aggregation of like species is in-

significant, we may write

BH + qSu N % BH(Su) (1-15)

where BH is the free cationic form of the small molecule,

q is the number of DNA binding sites that complex one

small molecule and Su denotes the unbound DNA binding

sites. Each site, Su, acts as one distinct entity in

the equilibrium expression and may contain any number

(1,2,3,. m) of phosphates. Note that throughout this

text binding sites will be discussed in terms of DNA

phosphates, simply for convenience, because the total

concentration of DNA phosphate is easily obtained ex-

perimentally. This terminology does not imply, for

instance, that phosphates are the binding sites for

intercalatively bound drug. The association constant,

on a molar basis, for the process is

K = [BHP]/[BH] [Su]q (1-16)

where [BHP] is the molar concentration of bound species,

[BH] is the concentration of free small molecule, and

[Su] is the concentration of the uncomplexed DNA sites.

Unbound site concentration may be determined from the

total site concentration, St, and bound site concentration,


Su = St Sc (1-17)

Now, since

Sc/q = [BHP] (1-18)

and St = [Pt/m] (1-19)

where Pt is total DNA phosphate concentration, we may


Su = [Pt/m]- q[BHP] (1-20)

Substituting equation (1-20) into (1-16) yields

K = [BHP] (1-21)
[BH] ([Pt/m]- q[BHP])q

Taking the log of both sides of (1-21) and rearranging,

we obtain

log ( [BH] = log K + qlog ([Pt/m]- q[BHP])q (1-22)

Molar concentrations of free and bound small molecule

species may be determined from any of a variety of

experimental techniques. The overall stoichiometry of

the process may be obtained from a Scatchard treatment or

a Job's plot. Knowing the overall stoichiometry, initial

values of m and q may be selected which satisfy the simple

relationship mq = n, where n is the total number of DNA

phosphates per bound small molecule. A rapid determination

of the most probable combinations of values may be made by

plotting log ([Pt/m]- q[BHP]) vs.log ([BHP]/[BH]). That

pair which yields the most consistent straight line having

a slope of q may then be used to compute values of K for a

series of ratios of [BHP] to [BH]. Again, the pair which

provides the most consistent calculated equilibrium associa-

tion constants for all data points, corresponding to either

the surface or intercalative mode, are taken as the exponents

in the binding expression. For interactions between 3-

aminoacridinium cation and calf thymus DNA, best-fit values

of m and q are 3 and 1,respectively for surface binding and

2 and 2 for the intercalative mode (69). To summarize, the

method of Capomacchia allows for the determination of

equilibrium association constants based on mechanistic

considerations. Furthermore, by computing values of K from

data in the midregion of a titration, undue reliance on the

least trustworthy data at the extrema of a titration is


Many workers have been satisfied with reporting associ-
ation constants calculated from molar quantities rather than

from the more correct activities of the various species.

Doing so implies two things: activities of the various

species in the expression are equal to molarities, and

the molarities are equal to molalities. It is unlikely

that approximating molalities with molarities introduces

significant error throughout the usual concentration

ranges employed. But activities, on the other hand,

may differ markedly from molarities. For this reason,

evaluation, of association constants in terms of activities

is recommended. Molality is related to activity by

a. = a.m. (1-23)
1 i 1

where a. is the activity, a. is the activity coefficient,

and m. is the molality of species i. Note that the
activity equals the molality only when a = 1. For a

process in solution

A-+ nB ABn(n+l) (1-24)

the equilibrium constant is rigourously defined as

a AB- (n+l) aAB MAB
K = n = n+ n n (1-25)
(A +) (AB-)n (a +MA+) (a-MB -)

It is extremely difficult to determine individual activities

since their is no way of separating effects due to positive

ions from those due to the accompanying negative ions. We
may, however, define a mean activity, a-, which is the

geometric mean of the individual activities

A! = (A+-A )1/2 (1-26)


+ + 1/2 (1-27)

For ideal dilute solutions, the mean ionic activity

coefficient will be equal to 1. If deviations from

ideality are caused entirely by electrical interactions

of point charges, it may be shown that, for aqueous

solutions at 25 C (70)

log aX- = -.509Z +Z_ll/2 (1-28)

where Z+ and Z_ are the charges of interacting species

and I is the total ionic strength of the system

I = 1/2 E .m. Z (1-29)

Equation (1-28) is an expression of the Debye-Hiickel

limiting law for activity coefficients and is applicable

only to dilute solutions containing point charges. As the

ionic strength becomes large (e.g., >0.01m) and/or the

ions become less ideal (become larger, more polarizable,

etc.) the relationship becomes increasingly approximate.

Under these conditions, an extended Debye-Hickel equation

may be invoked which introduces a correction for the finite

sizes of ions:

+ -0.509/Z ZJI1/2
log +a- =
1 + BdII/2 (1-30)

B is a constant for a given solvent and includes such

terms as solvent viscosity, dielectric, and temperature

while d is an average effective diameter of the ions. A

factor not considered in equation (1-30) is the effect of

ions on solvent molecules and solvent structure. This

may become especially important for solutions containing

large polyanions such as DNA.

Lyons and Kotin (44,45) report a method of assessing

values of a. and a independently of one another. They

also determined a- for Na and Mg when binding to DNA

and three synthetic polyanions. Their somewhat tedious

procedure involving equilibrium dialysis allowed them to

distinguish site-specific and non-specific forms of surface

binding. The activity coefficient of Na+ was shown to

decrease with increasing dilution of metal ion in the

presence of DNA suggesting an increased degree of binding.

Thermodynamics of Binding Reactions

A second topic of this work is concerned with the

thermodynamic parameters associated with the binding of

small molecules to DNA. Hence, a brief review of several

pertinent articles is presented here.

M. Gilbert and P. Claverie (61) measured total energies

for an intercalation process and attributed them to three

types of electrostatic interactions. They were (1) the

attractions between the bound cations and the DNA phosphates,

(2) the repulsive forces between cations, and (3) solvent

response to the electric field generated by the charged

species. At high ionic strength the phosphates are

relatively neutralized causing (1) to become insignificant

and, thereby, causing a decrease in both the strength and

extent of binding. Some limiting of the binding process

occurs even at very low ionicities because cation-cation

repulsions are still important with respect to intercalated

species. The third type of interaction serves to reduce

all forms of electrostatic energies and will be most ef-

fective for those species most intimately in contact with

the solvent. The authors have shown that phosphates and

completed acridines are more solvated than free acridines

so that, as the ionic strength is raised, their stabilizing

interactions are more reduced than are acridine-acridine

interactions. In that the forces arising from the latter

are destabilizing, the result of the countervailing trends

is a net destabilization of the complex with increasing

ionic strength. The model used here can also be applied

to other cationic small molecules, including protons.

Similar results were reported for the binding of proflavine

to DNA via fluorimetry (71). It was found that the electro-

static contribution to the total free energy decreased with

increasing ionic strength due to screening of the potential

by an ionic atmosphere.

Thermodynamic parameters accompanying the intercalation

of calf thymus DNA were evaluated by Chambron, Daune, and Sadron

(72) using equilibrium dialysis. Temperatures ranged from

0C to 70C and ionic strengths of 0.01M, 0.1M, and 1.OM

were maintained using acetate buffer at pH 5.9 in the

presence of EDTA. The number of sites available for

binding decreased upon increasing either the temperature

or the ionic strength. The enthalpy of reaction decreased

with decreasing temperature; AH= -19 Kcal/mole at 70C

and -4.5 Kcal/mole at 0C (I = 0.01). This was ascribed

to a thermally reversible change of state of DNA which

occurs at approximately 400C. Total standard free

energy was considered to be a sum of three terms. The

first of these, AGO, characterizes the dipole-induced
dipole interaction between base pairs and cations while

the second, AG0, is the electrostatic free energy. The

third, AG, is the solvation free energy corresponding

to desolvation of reactants and solvation of the bound

complex. At high ionic strengths, AGO and AGO are very
s e
small, hence, the total free energy of 6.4 Kcal/mole is

due to AG0. The authors' value of ASO= 11 Kcal/mole

degree is in agreement with theoretical predictions.

Thermodynamic energies of reaction between proflavine

and two forms of DNA determined from temperature-jump

relaxation methods are summarized below (23).

12 21 23 32 12 23 12 23
thymus 4 14 16 14 -9.8 2.0 -5.4 3.2

T2 13 17 -3.4 -3.9 1.6 -2.8

E. is the activation energy for transformation from state i

to j (see equation (1-13) and AH?. and ASO. are the thermo-
ij 1J
dynamic enthalpy and entropy changes for the same reaction.

Measurements were done over the temperature range 10 C to

25C, 0.20M Na+ and pH 6.9. Of particular note here is

that the enthalpy change for outside binding is much more

strongly affected by glucosylation of the DNA than it is

for the intercalated form (T2 DNA is glucosylated). For

example, AH20 changes from -9.8 to -3.4 Kcal/mole dye on

going from calf thymus to T2 DNA but for the overall inter-

calation process (AH 2 + AH23 = AH ) the change is only

from -7.8 to -7.3 Kcal/mole dye. It is unlikely that small

differences in base composition between the two forms of

DNA can account for the large differences in the thermo-

dynamic values of the binding reaction.

Entropy and enthalpy changes during reaction of

ethidium bromide with calf thymus DNA were computed from

temperature-jump relaxation data by Bresloff and Crothers

(65) over the temperature range 15 C to 320C in 1.OM Na+.

Values for the formation of a surface bound species were:

AHO = 1.1 Kcal/mole and ASo(19C) = 5.6 Kcal/mole, while

for one of a pair of intercalatively bound species: AH =

-7.8 Kcal/mole and TASo(190C) = -2.3 Kcal/mole. Values

for a second intercalated form could not be obtained due

to the rapidity of its binding process.

Bradley (63) determined the ground state free energy

of intercalation of a pair of neighboring acridine orange

molecules bound to DNA to be -0.065 Kcal/mole dye. By

comparing this to the spectral shift accompanying aggrega-

tion (504 nm to 464 nm) corresponding to 4.8 Kcal/mole dye

separation in energy between the ground and excited states,

he concluded that the spectral shift results from an

increase in energy of the excited state, rather than a

decrease in the ground state.

Enthalpy and entropy changes, AH and AS respectively,

of reactions may be determined from the relationship be-

tween the equilibrium association constants for the processes

at various temperatures and the temperature as set forth

by the van't Hoff relationship.

-AH 1
log K 2 1 + Constant (1-31)

This important equation will be derived below (73).

For ideal systems, any change in the total free

energy, G, can be said to be due to changes in either

temperature or pressure, since the state function, G,

can be defined in terms of pressure, volume, and temperature.

dG = (-G) dP + (aG) dT (1-32)

We define

G = H TS (1-33)


H = E + PV (1-34)

where H is the enthalpy, S is the entropy, and E is the

internal energy for the system under consideration. Recall

that, the enthalpy changes are commonly termed "heats of

reaction" for processes, entropy may be used as a measure

of the degree of order or disorder, and that the internal

energy is unavailable for work of any form. The change in

the free energy, dG, may now be expressed as

dG = dE + PdV + VdP TdS SdT (1-35)

For a reversible process in which no work other than

expansion occurs,

PdV = dw (1-36)


dE dq + dw = 0 (1-37)

The terms q and w refer to the heat absorbed, and work

done, by the system respectively. Equation (1-37) is a

direct consequence of the First Law of Thermodynamics and

states, in effect, that the total change in the internal

energy of a system is the sum (or difference) of the amount

of heat absorbed, dq, by the system and the amount of work

done, dw, by the system. Notice that heat and work are not

functions of state (which are conventionally assigned upper

case symbols). Equation (1-35) now becomes

dG = VdP SdT


which upon equating with equation (1-32) yields

(3G) = -S (1-39)


(DG) = V (1-40)

Equation (1-39) suggests an experimentally tractable

relationship between free energy (via K) and temperature,

when pressure is held constant. Moreover, we may extend

the relationship to include the changes in free energy as

a function of temperature for a process involving reactants

going to products. We may now equate (1-38) with equation

(1-32), rearrange, and solve for the partial differentials

to obtain

(3AG/3T)p = -AS (1-41)

where AG = (G products G reactants) and S = (S products

S reactants). The entropy term of equation (1-41) may be

eliminated by recalling that AS = (-AH + AG)/T and substi-


(3G/aT)p = (-AH + AG)/T (1-42)

Assume, henceforth, that the pressure will be held constant.

Rearranging (1-42) gives

dAG/dT AG/T = -AH/T (1-43)

Equation (1-43) may be expressed as

T d (AG/T) = -AH/T (1-44)

since, upon differentiation, it would become

T T(dAG/dT)-AG = dAG/dT AG (1-45)

Equation (1-45) may be written in terms of standard states,

and, in that

AG = -RT In K (1-46)

it can be stated as

d In K/dT = AH/RT2 (1-47)

which is equivalent to

d log K -AH
d lo- + Constant (1-48)
d(l/T) 2.303R

Assuming that AHO is temperature independent, equation (1-48)

can be integrated to yield equation (1-3.1). The above

relationship may be applied to reactions in solution whose

equilibrium association constants are based on the activities

of the species involved (see equations (1-23) through (1-30)].

Plots of log K vs.l/T should yield straight lines having

slopes equal to -AH /2.303R. If the relationship between

log K and l/T is not linear then the enthalpy of binding is

not constant with respect to temperature. For systems

containing DNA, enthalpies of binding determined from van't

Hoff plots are often not constant with temperature, but

decrease with increased temperature, presumably due to

changes in DNA's conformation (72). As a result, it should

not be surprising that binding curves obtained spectro-

photometrically and by dialysis equilibrium can agree at

one temperature and disagree at another (21).

Rationale for the Selection of the Model System
and Experimental Design

As this study is concerned with the effects of com-

peting metal ions and of alterations of the ionic strength

on the binding of small molecules to biopolymers, the choice

of DNA as the model polymer is evident. The selection of

3-aminoacridine as the small molecule is based on several

factors, two of which are especially important. The first

criterion that the small molecule must meet is that it bind

to DNA. To do so, it is necessary that the species exist

as a cation since neutral compounds do not bind to DNA --

or if they do, the binding is too weak for our purposes.

(It should be evident that electrostatic repulsive forces

prevent anions from interacting with DNA.) Furthermore,

the compound must be positively charged in solutions whose

pH is between 4 and 11 so that denaturation of the polymer

is minimal (74). Hence, if the compound is acidic its pKa

should be at least 5. The pKa of 3-aminoacridine is 8.04

(7:Ch 4) meaning it is essentially totally protonated in a

solution of pH 6. Moreover, 3-aminoacridine allows con-

venient evaluation of both of the principal modes of binding

of small molecules to DNA, under a given set of conditions,

in a single titration. That is, the range of the ratio of

total drug to total DNA concentration over which the two

modes of binding can be experimentally evaluated is about

0.1 to 10. Obviously, it is mandatory that each type of

binding occur in separate regions of the titration.

The second major criterion that the model compound

must satisfy is that it allow an expedient means of measur-

ing the extent of binding to the polyanion. Of the various

methods of determining equilibrium constants one of the best

is absorptiometric spectroscopy. The 3-aminoacridine cation

is particularly amenable to this method for the following

reasons. (1) It absorbs strongly in the visible region of

the spectrum as the free cation having molar absorptivities

of 1.413 x 10 Ml cm-1 at 365 nm and 1.259 x 10 MlO cm at

454 nm (75). Such magnitudes allow good sensitivity at low

concentrations of the compound. The molar absorptivities

of the bound drug in this region are 8.494 x 104 4Mcm-1 at

368 nm and 9.071 x 104 M cm-1 at 463 nm. The large dif-

ference between the absorbances of the band maxima of the

free and bound drug provides good sensitivity in following

a titration. The longest wavelength absorption maximum

for DNA is at approximately 260 nm meaning that there is

little spectroscopic interference from the polymer. (2)

Both the free and bound species have two distinct absorption

envelopes in the visible region corresponding to Lb and

L bands [Platt's nomenclature (76); a and para bands,
respectively, in the Clar nomenclature (77)]. These arise

from transitions from the 1A ground state to the 1Lb

(=365 nm), and to the 1La (=460 nm) excited states. The

transition moment of the former is parallel to the long

axis of the acridine ring system while the moment of the

latter is orthogonal to the 1Lb and parallel to the short

axis of the molecule. By observing shifts in the two

maxima, along with any changes in vibronic character,

during the course of a titration, one may be able to

deduce binding geometries (78). The 3-aminoacridine

molecule also fluoresces in its free cationic form and,

less intensely, as the bound species, thereby providing

another spectroscopic method with which to study its

interaction with DNA.

The more thoroughly studied aminoacridines, proflavine

and acridine orange, meet the criteria mentioned above in

regard to 3-aminoacridine. Both of them, however, begin to

aggregate at concentrations as low as 10-5M, thereby

introducing undesireable complications. Also, the mono-

substituted 3-aminoacridine, being a simpler molecule than

either proflavine or acridine orange, is preferable as a

model compound. It is hoped that information gained using

the simple compound may be applied to more complex systems.

Many references to the effects of ionic strength on the

binding of small molecules to biopolymers can be found in

the literature. Unfortunately, the majority are ancillary

to other studies and are not sufficiently complete in them-

selves to provide much specific information. Too, even for

a given molecule binding to a given polymer, the wide

variety of experimental variables under which the separate

investigations were done precludes any valid comparisons of

results. The present work was undertaken as a result of

this lack of a comprehensive investigation. Our pharmaceu-

tical interest led us specifically to DNA as the polyanion,

though other, synthetic polymers may have provided a less

complex system.

Some of the immediate questions which come to mind and

which may be answered by an investigation of the effects of

changing the ionic constitution of a drug-DNA solution are:

are the effects predictable on the basis of electrostatic

theory? That is, may the effects be explained in terms of

the simple Debye-Hickel limiting law at low ionic strengths

and by extended Debye-Hickel equations at higher ionicities?

If the above questions cannot be answered affirmatively,

might deviations at low ionic strengths be due to the

polymeric nature of the anion? Are deviations at high

ionic strengths due to specific ion interactions and, if so,

can they be rationalized on the grounds of simple competition

for binding sites or must a less simplified, mixed complex

model be introduced? Other questions are: will altering the

ionic medium affect surface and intercalative binding in the

same manner? If not, how can the differences be reconciled?

How will changing the ionic radii, ionic charge, and

charge density of the metallic ions affect the equilibria?

The last question is particularly germane in in vivo

biological binding studies because of the wide variations

in concentrations of Na K Mg and Ca in a living

organism (79).

To try to answer these questions, equilibrium associa-

tion constants for the surface and intercalative binding

of 3-aminoacridinium cation to DNA in a variety of ionic

media were determined. Background electrolytes included

concentrations between 0.15M and 6.3 x 10-4M of the phosphate

salts of Li Na K Rb Cs and (CH 3) Also con-

sidered were the acetates of the alkaline earths: Mg Ca ,

Sr ++, and Ba++ at concentrations ranging from 0.025N to 1 x

10-5 N. The effect of H 2PO4 as opposed to CH3COO- as the

counteranion of the supporting electrolyte was evaluated.

The pH for all of the titrations was 5.90 and the tempera-

ture was 25.00C. A minimum of two titrations were done for

each set of conditions.

Thermodynamic parameters were evaluated for systems

containing 3-aminoacridine and DNA with a background

electrolyte of 0.010M KH2PO4 at pH 5.90. Similar investiga-

tions were conducted using 7-aminoacridinium cation(XVI) in

place of 3-aminoacridinium. By studying the binding of the

linear benzalog of 3-aminoacridine and comparing the thermo-

dynamic parameters of the two, the relative importance of

enthalpy and entropy contributions to overall free energies

may be assessed.



Materials and Procedures

All chemicals were, unless otherwise stated, either

reagent or analytical grade. Water was multiply distilled

in a tin-lined still (Barnstead, Sybron Corp.). It was

redistilled until its specific conductance was less than

1.0 megaohm cm-1 at 25.0C. Stock solutions of supporting

electrolytes were prepared within three weeks of use and

were refrigerated during storage to minimize biological

growth. Approximately 100 to 200 ml of a 0.15M solution

of each electrolyte were prepared as outlined below and

titrated with 0.72M H3PO4, IM CH3CO2H, or IM metal hydroxide

solution to pH 5.9, in all cases. Concentrations of the

hydroxide solutions of the various cations were determined

by titration with standardized 1.OOON sulfuric acid.

A stock solution of tetramethylammonium phosphate was

made by titrating 13.16 ml of 1.14M tetramethylammonium

hydroxide (Mallinckrodt Organic Reagent, 10% in water,

lot: 15) in 70 ml of water with 0.72M phosphoric acid

(Fisher, lot: 755459) and adjusting the final volume to

100 ml. Lithium phosphate stock solution was prepared by

titration of 0.476M lithium hydroxide with 0.72M phosphoric

acid. The hydroxide solution was made by reacting stoi-

chiometric amounts of anhydrous lithium sulfate (PCR, Inc.,

Gainesville, Florida, lot: 10465) and barium hydroxide

heptahydrate (Mallinckrodt, lot: RTZ). Sodium and potas-

sium phosphate stock solutions were prepared using sodium

phosphate monohydrate (Mallinckrodt, lot: ABK) and mono-

basic potassium phosphate (Fisher Certified, Reagent, lot:

20175), respectively. The acid level of each solution was

adjusted with 0.72M phosphoric acid to pH 5.9. The stock

solution of rubidium phosphate was prepared by titrating

an aliquot of 0.43M rubidium hydroxide (Pfaltz and Bauer,

no lot number) with 0.72M phosphoric acid. Similarly,

cesium phosphate was prepared from 0.201M cesium hydroxide

(Pfaltz and Bauer, no lot number). Stock solutions of

potassium acetate and tetramethylammonium acetate were

prepared by titrating standardized aliquots of the respec-

tive hydroxides with 1M acetic acid (Baker Chemical, lot:

41397). Magnesium acetate solutions were made from

magnesium acetate tetrahydrate (Mallinckrodt, control:

NSB). Solid calcium acetate was made by dissolving calcium

carbonate (Mallinckrodt, no lot number) in excess acetic

acid. The calcium acetate was recrystallized twice from

aqueous ethanol, washed repeatedly with 95% ethanol, and

dried for twenty-four hours at 110C to yield the mono-

hydrate of the salt. The sample was found to contain 97.3%

Ca(C2H302).H20, by dissolving a known amount of the acetate

in water, precipitating the calcium as calcium carbonate

and titrating the carbonate with hydrochloric acid.

Strontium acetate was prepared from its carbonate (Matheson,

Coleman & Bell, lot: 19) and assayed in a manner similar

to that for calcium acetate. It was determined to contain

95.0% Sr(C2H302). A recrystallized sample of barium

acetate was dried for twenty-four hours at 170 C and

assayed by forming barium carbonate. The solid carbonate

was reacted with excess 1.00N sulfuric acid, carbonic acid

removed by gentle heating, and the excess sulfuric acid

determined by titration with standard sodium hydroxide

solution. The assay of the barium acetate was 99.6%


Solutions of the supporting electrolytes to be used

as solvents were made from the stock solutions. Aliquots

of the stock solutions were diluted to within 90% of their

calculated final volumes and their acidities adjusted to

pH 5.90 + 0.05, if necessary, using the appropriate metal

hydroxide, phosphoric acid, or acetic acid. They were

then diluted to their exact desired final volumes and their

ionic strengths computed, taking into account any additional

ions added to adjust the acid levels.

A commercially available sodium salt of DNA was

dialyzed against a solution of the desired metal phosphate

or acetate to obtain the DNA salts having the same cation as

that of the supporting electrolyte. Solutions containing

approximately 0.01 mole of DNA phosphate per liter were made

by placing 3.6 mg of DNA, sodium salt (Calbiochem calf

thymus DNA, A grade, lot: 900007: 8.04% P; 12.21% N;

moisture, 14%; E = 192) per milliliter of solvent
into twenty milliliter volumetric flasks. Between 5 and

20 ml of solvent were then added. The solvent was a 0.01M

solution (pH 5.90) of the metal phosphate or acetate. The

flask and its contents were placed on a vertical rotating

mixer (Scientific Products model 150-V) operating at 6 rpm

for about one hour, after which the sample was placed in

an ultrasonic mixer for two to three minutes. Rotation

and sonication were repeated for a total of at least five

hours or well beyond the time at which the moderately

viscous sample appeared homogeneous. The sodium DNA solu-

tion was then transferred to dialysis tubing (Union Carbide

Films Packaging Division, 7.3 mm dia.). The tubing and its

contents were immersed in a volume of 0.01M solution of the

desired salt which was at least ten times greater than the

volume of DNA solution being dialyzed. The external solu-

tion was agitated for two hours and then replaced. A

total of six volumes of external solution were used for

each dialysis procedure. The dialyzed DNA was again mixed

for one hour on the rotating mixer and submitted to ultra-

sonication for two to three minutes. Then, the sample was
divided among Pyrex culture tubes (6x50 mm) each

receiving approximately one half milliliter of sample.

These were immediately stored at -5 C.

Determinations of the concentrations of DNA after

dissolution and after dialysis yielded values which were

identical, within experimental error and also within + 5%

of the calculated concentration based on the manufacturer's

data. The agreement of these data were taken as evidence

that any degradation of the biopolymer during the pro-

cedures was minimal.

The efficiency of the dialysis method, used here, to

replace the sodium cations of the original DNA salt with

the cations of the external solution was evaluated by

atomic absorption spectroscopy. A Perkin-Elmer model 290B

atomic absorption spectrophotometer outfitted with a combina-

tion sodium/potassium lamp was used to measure the residual

concentration of sodium ion in a sample of sodium DNA which

had been dialyzed against a potassium phosphate solution.

Instrument readings were taken using the contents of the

dialysis bag and compared to those obtained for standard

solutions in which the ratio of potassium to sodium ions

was approximately one thousand. Note that in this instance

high relative concentrations of potassium ion must be included

in the standard solution. Less than 0.1% of the original

sodium remained within the dialysis bag when the above

procedure was employed.

Crystalline 3-aminoacridine was prepared by Dr. Timothy

Roy using a standard procedure (80) except for additional

purification of the product by vacuum sublimation. Thin

layer chromatographic analysis of the purified product

using a variety of solvents and solid supports failed to

indicate the presence of any impurities. The spectra of

solutions of 3-aminoacridine at pH 11.0 and at pH 2.5 had

the same relative absorptivities as those reported by

Albert (75). Reported (pH 11.0) xnm (log e): 237 (4.46),
262 (4.83), 321 (3.35), 337 (3.65), 353 (3.92), 410 (3.79).

(pH 2.5) 233 (4.62), 274 (4.65), 349 (4.03), 365 (4.15),

454 (4.10). The melting point agreed with that reported

(81): 2180C, uncorr., 223.50C, corr.
Stock solutions of 1 x 10-3 M 3-aminoacridine were

prepared by dissolving 2 mg of the solid in 7 ml of absolute

ethanol. The solutions were stored at -5C in glass. One

hundred microliter aliquots of the ethanolic solutions were

used for each titration. Periodic tests of the stock

solutions using thin layer chromatography showed no observ-

able degradation during the time periods in which the samples

were being employed.

Absorptiometric titrations were conducted by successive

additions of microliter aliquots of DNA stock solution to

10 ml solutions of 3-aminoacridinium cation. After each

addition, the absorbance spectrum between 550 nm and 320 nm

of the solution was recorded. To prepare the solutions, 10 ml

of supporting electrolyte at a given concentration were

measured in a 10 ml volumetric flask. This volume was de-

termined when the liquid was at the same temperature as it

was to be during the titrations (e.g., 15.0, .25.0, or 35.0 C).

Immediately prior to a titration, 0.1 ml of stock 3-amino-

acridine in ethanol was added to the flask, the contents

mixed and then transferred to a 4.00 cm quartz cell (Bolab

Inc., Derry, New Hampshire). The cell and its contents

were placed in the thermostated compartment of the

spectrophotometer (Beckman model DB-GT) and allowed to

become thermally equilibrated. The temperature of the

instrument's cell compartment was maintained at 15.0 +

0.2C, 25.0 + 0.050C or 35.0 + 0.20C by circulating thermo-

stated water (Aminco refrigerated water bath) through its

external jacket.

The spectrum of the sample was recorded after which

an aliquot of stock DNA solution was added. Syringes (Uni-

metrics, Inc.) were employed to dispense volumes of DNA

solution ranging from 4 pl to 100 pl while volumes greater

than 100 Ul (generally added toward the ends of the

titrations) were measured with a 100 microliter micro-

pipette (Centaur Chemical, Danbury, Connecticut). The

individual culture tube of the stock DNA solution from

which aliquots were being withdrawn for a given titration

was maintained at 25C, regardless of the temperature at

which titration was being conducted. After adding the DNA

to the 3-aminoacridinium solution in the cell, the solution

was mixed by withdrawing about 5 ml of the cell's contents

into a serological pipet, then rapidly forcing them back

into the cell. This was done three or four times after

each addition of DNA.

Concentrations of the stock DNA solutions were

determined by measuring the absorbance of a solution

containing 50 ml of stock DNA in 10.00 ml of 0.010M

supporting electrolyte at pH 5.90 (1.00 cm pathlength).

The concentrations of at least ten samples, each taken

from separate tubes, were calculated using the molar

absorptivity at 260 nm of 1.413 x 10 4M-cm-1 (75). All

glassware was cleaned by immersion into concentrated

sulfuric acid saturated with potassium dichromate. Spec-

trophotometer cells were rinsed with water followed by

ethanol and were stored in 50% concentrated hydrochloric

acid in ethanol to prevent accumulative absorption of the

dye or of DNA to their surfaces.

Solid 7-aminoquinoline was a gift from Dr. D. Jackson,

Texas Tech. University, Lubbock, Texas, and was determined

to be at least 98% pure according to molar absorptivity

values (82-84). Reported (pH 12) Xnm (log e): 240 (4.56),
275s (3.58), 335 (3.68), 346 (3.64). (pH 2.0): 260 (4.44),

285s (3.57), 392 (3.90). The reported pKa of the compound

is 6.65 (84). A stock 3.4 x 10-3M solution in absolute

ethanol was prepared and 30 Pi aliquots added to 8.00 ml

of 0.10M KH2PO4 buffer solution at pH 5.9. The 7-amino-

quinolinium solutions were titrated at 15.0 + 0.20C, 25.0 +

0.050C, and 35.0 + 0.20C using the same equipment and pro-

cedures as for the 3-aminoacridinium titrations, vide



The molar absorptivity of the 3-aminoacridinium-DNA

complex, EBHP, at 368 nm was calculated from absorbance

spectra of solutions in which essentially all of the

compound was in the bound form. Complete binding (> 99%

of all drug present) was assumed when at least three suc-

cessive additions of excess DNA to the solution resulted

in no change in the total absorbance, corrected for

dilution. It was further presumed that, under these

conditions, all absorbance at 368 nm was due to the complex

and no 3-aminoacridine existed, either as the neutral or

monocationic, unbound form. Also, only solutions containing

monovalent metals (and tertamethylammonium cation) at low

ionicities were used since, for other systems, the large

amounts of DNA needed to drive the equilibrium toward

products caused serious light scattering and, in some

instances, precipitation of DNA salts. Calculated values

of EBHP' in the presence of each type of electrolyte are

presented in Table 1. The averaged value of 8494 M cm

was used for E BHP in the presence of all supporting elec-

trolytes at all concentrations. A similar procedure was

used to evaluate the molar absorptivity of the 7-amino-

quinolinium-DNA complex, AHP,' at its band maximum (405 nm).

At pH 5.9 the ratio of neutral to protonated 7-amino-

quinoline is 0.178. It was assumed that only the cationic

form of the drug binds to DNA. The averaged molar absorp-

tivity of the 7-aminoquinolinium-DNA complex was evaluated

using three simultaneous equations and was found to be

5332 M 1cm-1 at 405 nm.

Concentrations of free 3-aminoacridinium cation,

[BH], and bound monocation, [BHP], were calculated from

absorptiometric data assuming simple additivity of the

absorbances of the two species at the analytical wave-


A = BH[BH] + E BHP[BHP] (2-1)

and considering the mass balance expression

C = [BH] + [BHP]



A = the total absorbance of the solution at a given


EBH = the molar absorptivity of the free species,

1.413 x 10 M cm1.

BHP = the molar absorptivity of the bound complex,

8.494 x 103 M cm1 .

A = pathlength of light in the cell.

Ct = the total concentration of free and bound species.

[BH] = the molar concentration of free 3-aminoacridinium


[BHP] = the molar concentration of bound 3-aminoacridinium-

DNA complex.

Solving equation (2-2) for [BH], substituting into equation

(2-1), and rearranging yields

A-e C L
[BHP] -= BHB t (2-3)

Before any DNA has been added, the total absorbance of

the solution will be due entirely to free BH and, since

BH at this point will be equal to Ct, the total absorbance

before any titration, Ao, may be expressed as

A = E BHCt (2-4)

Similarly, when all of the compound has been bound, the

final absorbance, A is equal to

A = HP C (2-5)

and for any point in the titration, where the total absor-

bance is At,

[BHP] = [(At-A )/(A -A )]Ct (2-6)


[BH] = [1-(A -A )/(A -A )]Ct (2-7)

Since an appreciable amount of the total 7-amino-

quinoline exists in the solution in the neutral form at

pH 5.9, the equations representing the protolytic equilibrium,

[H+] [A]
K [H[A] (2-8)
a [AH]

the conservation of mass for 7-aminoquinoline,

Ct = [A] + [AH] + [AHP],


and the photometric absorbance at any point in the

titration of the small molecule with DNA,

At = A[A] + AH[AH] + AHP[AHP]


must be employed to determine the concentrations of the

species involved in the binding. Concentrations of drug

species are, explicitly,

(EA---a +
A[H ]

SAH+) C t -At(l + K/[H + )

a +-
A [H+] AH EAHP(1 +a/[H+]


A[H + AH -

EAHP(1 + a/[H ])}

(At AHP C t) K[H +
t AHP t a[H ]


a a +
A + AH

- eAHP (1+ aH)}

[AHP] =

[AH] =



[A] =

It was also useful to directly calculate the ratio,

[AHP] (EA a + E AH+) Ct A(1 + Ka/[H+])
] ----- =(2-14)



Ka = dissociation constant for the drug or probe,

2.24 x 10-7 M.

[H ] = hydronium ion concentration, 1.26 x 10-6 M.

EAHP = molar absorptivity of the bound monocation at

the specified analytical wavelength, 5.332 x

103 M-cm- 1 at 420 nm.

AH = molar absorptivity of the free monocation at

the specified analytical wavelength, 7.943 x

103 M- cm-1 at 420 nm.

EA = molar absorptivity of the free base at the

specified analytical wavelength, 3.47 x 102 M-

cm1 at 392 nm.

L = absorption cell path, 4.00 cm.

At = the absorbance at any point during the titration

at the specified analytical wavelength.

Ct = total concentration of drug or probe.

For each point in a titration the total absorbance was

determined along with appropriate constants which were

substituted into the necessary relationships to yield the

concentrations of free and bound species. Raw data for

all titrations are presented in Tables 2 through 17. Note

that volume corrections for the addition of DNA solution

to the sample were made in all cases.

The equilibrium association constants for the

surface and intercalative modes of binding were calculated

for each point in the absorptiometric titrations of the

small molecules with DNA, using the relationships developed

by A.C. Capomacchia and S.G. Schulman (85).

Ks = [BHP]/[BH]([Pt/3]-[BHP]) (2-15)


KI = [BHP]/[BH] ([Pt/2]-2[BHPI)2 (2-16)

where Ks and K are the apparent association constants

for the surface and intercalative modes, respectively, and

Pt is the total molar DNA phosphate concentration. The

equations for the binding of 7-aminoquinolinium to DNA

are the same except [AHP] and [AH] are substituted for

[BHP] and [BH] respectively. The extended Debye-Hickel equation

(1-28) was used to calculate activity coefficients at the

various ionic strengths, for charged species involved in

the association equilibrium.

Apparent association constants were corrected for

deviations of molar concentration from activity, by

K' = K n (2-17)

where K' is the corrected apparent association constant,

aBHP' aBH' and a are activity coefficients for bound

species, free species, and DNA phosphate, respectively.

The value of n is 1 for surface binding and 2 for inter-

calative binding. For the binding of 7-aminoquinolinium

to DNA, the subscripts BHP and BH are replaced with AHP

and AH corresponding to bound and free 7-aminoquinolinium,

respectively. These, and other calculations, were done

using a Litton-Monroe model Beta 326 Scientist computer

equipped with a tape cassette drive (Litton-Monroe,

Orange, New Jersey). The simultaneous solution of (2-8),

(2-9), and (2-10) to obtain concentrations of A, AH, and

AHP was accomplished using a program supplied by the

computer manufacturer. Programs for specific calculations

used here were designed and are presented in Appendix III.

Thermodynamic parameters for the binding of 3-amino-

acridinium and 7-aminoquinolinium to DNA were evaluated

using equation (1-31). Values of log K at 15C, 25C, and

35C for surface and for intercalative binding of each

species were determined. These values were then plotted

vs. l/T (K). The slopes of the lines, equal to AH/2.303R,

were used to calculate the standard entalpies of binding.

Values for the standard entropies of the reactions were

evaluated using

AS0 = (AHo AG)/T




General Spectral and Titration Characteristics

A number of interesting spectral features and

titration characteristics are common to all of the

systems investigated. Space limitations preclude inclu-

sion of all of the absorptiometric titration spectra,

though four which are representative are presented

(Figures 3 through 6). These are spectra of titrations

done in the presence of the background cation having the

lowest charge density, Cs and the greatest charge density,

Mg at the highest and lowest concentrations of each.

As a rule, the lower the charge and charge density, of

the background cation, the less effect it appears to have

on the two modes of binding. The highest concentration of

the smallest divalent cation represents the other extreme,

as it severely inhibits binding processes. The rest of

the systems are intermediate between these two. In each

of the figures, only 7 curves are drawn for the sake of

clarity, though usually between 18 and 22 spectra were

recorded for any given titration. For the curves committed

from figures 3 through 6, the absorbances at X362nm of the

Lb bands may be obtained from the appropriate tables in

Appendix II.

In all cases, addition of DNA to solutions of 3-

aminoacridinium cation resulted in decreases (hypochromisms)

in the intensities of both the Lb and L bands. It is,
b a
noteworthy that even small amounts of DNA (< 1 mole DNA

per mole of drug) added to 3-aminoacridinium solutions

produced only hypochromic shifts. However, addition of

small amounts of DNA to other compounds such as 7-amino-

quinolinium causes an initial increase in the total intensity

(hyperchromism) of the drug's spectrum. Hypochromism sub-

sequently occurs when DNA in excess of about 1 mole per mole

of drug is added. Presently, it is not known what factors

are responsible for the initial hyperchromism of some species'

spectra or why some exhibit initial hyperchromism while

others exhibit only hypochromism throughout their whole

titration, though significant variations in binding behavior

appear highly probable.

It may be seen from Figures 3 through 6 that addition

of increasing increments of DNA to 3-aminoacridinium results

in red (bathochromic) shifts of both bands of the drug's

spectrum. The bathochromism of the 1L "-- A and 1L <-- 'A
b b
bands are a result of decreases in the energy separations

between the ground and excited states. This may be due to

a lowering of the IL states, a raising of the ground. state,

or a combination of both. If binding of small molecules to

DNA occurs as the reaction proceeds, then it is reasonable

to hypothesize that the cationic drug is moving from a

highly polar, high dielectric medium (water) to a less

polar, more lipophilic solvent such as the region in, and

around, the DNA helix. An environmental change of this

type would tend to favor a decrease in charge separation

within a molecule and, thus, favor n --go i transitions

in which the formal positive charge of the heterocyclic

ring nitrogen is decreased. Such a case would mean that

a lowering of the energies of the excited states is the

predominant factor in causing the red shift upon binding

if, indeed, such "solvent effect" arguments are valid when

considering binding processes. Caution must be exercised

in this regard as Capomacchia (69) has found that, for a

number of compounds, spectral shifts upon binding to DNA

are not in qualitative agreement with shifts observed when

the compounds were transferred from polar aqueous solvents

to less polar organic solvents. Maximum shifts of the two

bands are from 362 nm to 368 nm for the Lb band and from

454 nm to 463 nm for the 1L band. These differences (6 nm
and 9 nm, respectively) correspond to decreases in the

transition energies of 450 cm-1 (1.29 Kcal/mole dye) and

428 cm (1.23 Kcal/mole dye). The similar magnitudes show

that, with excess DNA at low total ionic strength, the two

transitions are equally affected from an energetic stand-

point in the presence of excess DNA. However, a large

number of titrations have substantiated the observation

that the 1Lb moves only slightly, from 362 nm to = 364 nm,

over most of the titration and then abruptly shifts from

= 364 nm to 368 nm while the shift of the 1L band is

smooth throughout the whole titration. The abrupt shift

of the 1Lb takes place at approximately the point in the

titration where any additional DNA results in no further

decrease in 1Lb band intensity. In fact, this phenomena

has been accepted as evidence that all of the protonated

3-aminoacridine has been bound. Moreover, the rapid shift

of the higher energy band always occurs significantly

beyond the point in the titration where very little change

is noted in band intensity or position of the maximum of

the longer wavelength Lb band. These observations suggest

that the environment appears to affect both transitions

equally in the presence of excess DNA, but they are affected

differently during the course of a titration. A cursory

consideration of the data may lead one to postulate that

the abrupt shift is due to intercalation of the molecule.

This is plausible since internal binding would affect the

1Lb more dramatically than the 1La, especially if the charged

ring nitrogen remained exterior to the helix. Further

consideration, however, reveals substantial evidence, both

in this work and in the literature, which shows that inter-

calation occurs much earlier in the titration and that it is

much less abrupt a process than the bathochromic shift

considered here would require. The abrupt change might

arise from some form of near-neighbor interaction of bound

drug molecules (63). During most of the titration, where

the total DNA phosphate to total 3-aminoacridine ratio,

P/D, is low, the bound molecules can interact with one

another. But as the ratio increases, bound species become

isolated from each other to the point of no longer being

able to interact. The hypochromic shift is not incon-

sistent with this proposal as both red and blue shifts

are possible upon exciton formation. Another factor to

be considered is that as the DNA concentration is increased

aggregation of the DNA, itself, may begin to occur. At

some critical level, the bound dye may experience a marked

decrease in the hydrophilicity of its environment, perhaps

as a result of the neighboring polymers excluding water

between themselves. Under these circumstances, red shifts

would be expected, though why they would be seen only for

the Lb and not for the 1La transition is unanswered.

The onset of the rapid red shift appears to be depen-

dent on the total ionic strength of the medium as well as

the type of cations in solution. In the presence of

large, diffusely charged, monovalent cesium a distinct

bathochromism occurred, even in the highest concentration

(0.15M) at the moderate P/D ratio of 65. On the other

hand, 6.3 x 10-4 N Mg++ was sufficient to prevent the shift

to 368 nm even when phosphate is in excess of drug by a

factor of 65. In 0.0025M CsH2PO4, a P/D ratio of only 14

results in the spectral shift (in fact, it probably occurs

at about P/D = 11). Charge density of the countercations

influenced the point at which the abrupt red shift was seen

in a manner parallel to the degree of completion of a

titration for a given P/D ratio. The overall spectral

behavior of the systems in the presence of the alkaline

earths was the same as for the alkalies except the

maximum concentrations of M++ that allowed at least 90%

binding at P/D < 100 was much lower than for alkali

metals. Concentrations of group IIa cations in excess

of about 0.02N resulted in precipitates forming in the

samples. These were, presumably, the metal salts of DNA

and appeared long before sufficient polymer was added to

appreciably bind the drug.

Throughout this discussion, the absorbances at the

band maxima have been assumed to be directly related to

the areas under the curves. Since it is the areas, and

not peak heights at the maxima which are true measures of

transition probabilities, care must be taken in equating

peak heights with relative transition probabilities. If

the overall geometry of the 1Lb band were to change during

the course of a titration, peak heights would not be valid

representations of the progress of the titration. In our

studies, inspection of the bands showed that their geome-

tries were constant during the titrations, hence, we feel

approximating peak heights with peak areas was valid.

In addition, we have assumed that the 0-0 vibronic

band maximum of the 1Lb A transition is coincident

with the maximum of the total transition envelope. This

may be misleading if the maximum of the envelope arises

from the sums of the intensities of vibronic transitions

which are close enough together to appreciably overlap

at the maximum wavelength. If the vibronic energies are

altered in dissimilar ways during a titration, then the

maximal envelope intensity will reflect this dissimilar

change along with any real shift in the 0-0 band. It is

possible, therefore, that shifts of the Lb envelope

maximum may be due to either a disproportionate change in

the magnitudes of the band maxima of the 0-0 and 0-1 vibronic

bands or a bona fide, pure 0-0 shift.

Variations in the spectral characteristics of the 3-

aminoacridinium-DNA systems in the presence of metal phos-

phates and metal acetates were not due to the different

counteranions, as evidenced by the results of titrations

using KH2PO4 and KO2 CCH3 as background electrolytes. Re-

sults obtained with the two buffer ions were the same in

all respects except at very low ionic strengths. Deviations

in this region are probably due to lack of sufficient pH

control or alterations in the conformation of DNA possibly

arising from mild denaturation.

The ILb envelope of the free 3-aminoacridinium cation

consists of two distinct vibronic bands having maxima at

348 nm and 362 nm, corresponding to the 0-1 and 0-0 vibronic

transitions, respectively. As a titration proceeds, the fine

structure of the envelope is lost as a result of the 348 nm

peak becoming increasingly less pronounced, to the point of

becoming a mere perturbation of the overall envelope. It is

difficult to determine the exact position of the shoulder

when the titration is nearly complete, but it appears that

its shift is essentially the same as that of the 362 nm

band. In cases where the abrupt 364 nm to 368 nm shift

of the v vibronic band of the Lb occurs, there is an

accompanying shift of the shoulder maximum from = 350 nm to

about 354 nm with a slight, but definite increase in the

resolution of the two bands. The degree of fine structure

of a compound's spectrum provides qualitative information

regarding the environment of the species. Loss of fine

structure may be rationalized on the basis of increasing

the degrees of vibrational freedom of the molecule. If,

upon binding, the energy separation between the v = 0 and

v = 1 vibrational levels of the ground state were to be

reduced, a coalescing of the two peaks would be anticipated.

Such a situation may arise when a molecule moves from an

environment in which its vibrational motions are restricted

to one in which they are less so. For this to maintain in

our systems, the bound species would have to have greater

vibrational freedom than the wholly water-solvated free

cation. This is feasible considering the strong degree of

interaction between water molecules and any dipole in their

midst compared to the lipophilic, noninteracting "solvent"

of the surface and interior of DNA.

The increase in fine structure which accompanies the

dramatic red shift of the 0-0 band of the 1Lb envelope at

high DNA concentrations is not well understood but serves

to corroborate evidence for a dramatic environmental change

of bound drug at these very high levels.

To summarize, titrations of 3-aminoacridinium cation

with DNA salts in the presence of monovalent and divalent

metal ions (including tetramethylammonium ion) of varying

concentrations result in hypochromism only. In all cases,

both the 1a and Lb envelopes of the free drug's spectrum

red shift, at least to some extent, during titration --

the 1L from 454 nm to 463 nm and the 1L from 362 nm to

about 364 nm. In the presence of all concentrations of

supporting electrolytes having large, monovalent cations,

a further, sharp red shift from 364 nm to 368 nm takes place

when a P/D ratio of 10 to 15 is reached. The shift may be

effected in the presence of smaller alkalimetal ions at

P/D ratios approaching 70. Regardless of the concentration

or charge density of divalent cations, the dramatic 1Lb

shift did not occur. Loss of vibronic fine structure of

the 1Lb envelope was observed in all cases as the titrations

progressed, up to the point at which the dramatic batho-

chromic shift occurred. Concomitant with the rapid shift

was a reemergence of the 0-1 vibronic band.

Using relationships developed in the Experimental

section, concentrations of free and bound 3-aminoacridinium

cation were computed from the total absorbances of solutions

which contained varying amounts of DNA. Data for all of

the titrations are not included here, though some, which

include the extremes of experimental conditions, are presented.

Tables 18 through 21 contain tabulations of [BH] and

[BHP] along with log ([BHP]/[BH]) values for reactions in

0.15M and 0.0025M CsH2PO4 and in 0.010N and 6.3 x 10-4N

Mg(O2CCH3)2. In lower concentrations of alkali metals

and tetramethylammonium ion, the addition of as little

as 5 pl of 9 x 10-3 M DNA phosphate solution results in a

significant degree of binding of the drug. For instance,

nearly 10% of the total 3-aminoacridinium present in

0.0025N CsH2PO4 is bound after addition of only 5.0 pl of

DNA. Since volumes of DNA less than 5 pl cannot be

measured with much reliability due to its viscosity, the

ratio of bound to free drug of 0.1 should be considered

as the effective minimum for titrations of this sort. When

lower ratios were possible, such as in solutions containing

divalent cations or > 0.010M monocations, random experi-

mental error was sufficiently high to make the data unreliable.

As a result, our studies indicate that absorptiometric

titrations of small molecules with DNA, under conditions

similar to ours, are probably invalid whenever the ratio of

bound to total drug is less than 0.1. It should be recalled

that our use of 4.00 cm cells maximized sensitivity, thereby

allowing minimal increments of DNA solution to be used.

Smaller amounts of DNA could be added in 5 pl increments by

employing a more dilute stock solution, but the errors arising

from other sources would likely be unimproved. At the other

titration extreme, ratios of bound to free drug in excess of

about 0.95 also result in unacceptable randomness in most


cases. Our procedures are designed to minimize the impact

of these sources of imprecision on the calculated association


The maximum extent of total binding which could be

achieved varied with the character and concentration of

the supporting electrolyte. Table 22 lists the percent

of total drug bound to DNA after the final increment of

DNA was added in titrations done in various media. The

P/D values listed are the maximum that could be obtained

due to solubility and pH-control limitations. No effort

was made to determine how the drug was bound or what per-

centage of the total was surface bound compared to inter-

calatively bound. As discussed previously, for those titra-

tions in which an abrupt bathochromic shift of the ILb band

was observed, the binding was considered to be complete

(100 + 2%). The averaged molar absorptivity, BHP, determined

from these titrations was used to compute concentrations of

free and bound species for all systems. In light of the

experimental error incurred in evaluating EBHP and the

simple fact that thermodynamic equilibrium demands that

there cannot be total binding, percentages greater than

about 96% are taken as indicative of complete binding. From

Table 22, it may be seen that at least 97% of the total 3-

aminoacridinium present is bound in solutions of < 0.010M

MH2PO4. To attain essentially complete binding in these

systems only about 10 to 20 times as much DNA phosphate as

total drug need be present (see P/D column of Table 22). As

expected, larger and larger ratios of DNA to drug are

necessary for complete binding as the concentration of

a given electrolyte is increased. Similarly, it becomes

increasingly more difficult to shift the equilibrium wholly

toward bound product in the order (of electrolyte cations):

(CH 3)4N < Cs
required for 97% binding in 0.0025N Cs +; 17-fold was needed

for 97% binding in 0.010M Cs; while, even with a 65-fold

excess of DNA, only 90% of the drug was bound in 0.15M

CsH2PO4. It was relatively easy to totally bind the 3-

aminoacridinium cation in low concentrations of monovalent

cations but very difficult to attain greater than 90%

binding in even the lowest concentrations of the divalent

electrolytes. In 6.3 x 10-4 N Ba++ (the dication having

the lowest charge density of those investigated) only 90%

of the drug was bound at a P/D ratio of 47 and, for Mg++

at the same concentration, 90% was bound at P/D = 65.

Similarly, in 0.010N Ca++ and 0.010N Mg++ the maximum

degrees of binding were 77% and 70%, respectively, with

P/D ratios of 103 and 104. The choices of the upper limits

of total phosphate concentrations reported here were de-

pendent on the solubility of species in the concentrated

electrolytes and on pH considerations for the dilute systems.

Visible precipitation or erratic spectral behavior were

evident in solutions having P/D values greater than those

listed in Table 22 for barium and magnesium. The maximum

concentration of the divalent electrolytes was set at 0.010N,

since obvious precipitation occurred in solutions of

0.025N M2+ well before the titrations were complete (i.e.,

at 30% to 50% binding). Acetates of the alkaline earths

were used instead of their phosphates due to the latter's

very low solubilities. Despite dialysis of DNA solutions

against pH 5.90 buffer, their pH levels were still ap-

parently close to 7. Thus, addition of moderate volumes

of DNA to the very poor acetate buffer systems caused

excessive increases in the pH levels of the dilute buffer

solutions. Consequently, acetate solutions having more
than approximately 5 x 10-4 M DNA phosphate had pH values

considerably in excess of 5.90.

Upon perusing the second column of Table 22, it becomes

quickly evident that the extent of binding is not predomi-

nately and directly related to ionic strength. Total ionic

strengths were computed for the systems having the P/D

values listed using equation (1-29) and Program 4 of

Appendix III. Notice that while the ionic strengths of

0.0025M CsH2PO4 and 6.3 x 10-4N Ba(O2CCH3)2 are the same,

the binding is complete in the former and only 90% complete

in the latter. This shows that the effect of the formal

charge on the cations of the electrolyte is greater than

predicted on the basis of the simple electrostatic effects

incorporated in the limiting Debye-Hickel equation. That

the contribution of excess DNA to the total ionic strength

is not responsible for these large differences may be seen

by comparing 0.010N barium and 0.010N magnesium solutions

which both contain the same amount of DNA and have, there-

fore, the same total ionic strengths. The extent of

binding in the former is 77% while in the latter it is

only 70%.

Rapid estimations of the values of m and q of equation

(1-21) may be obtained by selecting reasonable terms such

that mq = n (n is total phosphate per drug). It has been

shown that (85) in a medium containing 0.002M to 0.005M

KH2PO4 (pH 5.9) the binding of 3-aminoacridinium cation to

DNA is best expressed by

BH+ + (P04)3 BH(PO4)2- (3-1)

for the surface mode, and

BH+ + 2(PO4)2 BH((PO )2 (3-2)

for the intercalative mode. In equation (3-1), three DNA

phosphates act as a single moiety, reacting with the small

molecular cation to form the surface-bound species. In

equation (3-2), 2 phosphates act as a single reacting

entity with two of these, in turn, interacting with one

cation to form the intercalated complex. The values of m

and q are 3 and 1, respectively, for the first case and 2

and 2 for the second. Substituting these into equation (1-21)


Ks = [BHP]/[BH] ([Pt/3]-[BHP])) (3-3)


KI = [BHP]/([BH] ([Pt/2]-2[BHP])2) (3-4)

Ks and K are the apparent association constants for the

surface-bound and intercalatively-bound drug-DNA complexes,

respectively. These equations were tested to certify their

validity under our experimental conditions. Recall that

plots of log([BHP]/[BH]) vs. log([P t/]-q[BHP]) yield lines

whose slopes are equal to q and whose intercepts are equal

to log K. Tables 18 through 21 include calculated values

of log([BHP]/[BH]), log([Pt/3]-[BHP]), and log([Pt/2]-2[BHP])

for representative titrations. Figures 7 and 8 are plots

of log([BHP]/[BH]) vs. log([P t/3]-[BHP]) for the reactions

in the presence of 0.0025M Cs+ and 6.3 x 10-4N Mg++; and in

0.15M Cs 0.010N Mg ++, and 0.010M Cs respectively.

Figures 9 and 10 are plots of log([BHP]/[BH]) vs. log([Pt/2]-

2[BHP]) for the same systems. Three general features are

particularly significant. One, the shapes and slopes of

the curves are essentially the same whether log([Pt/3]-[BHP])

or log([Pt/2]-2[BHP]) is plotted against log([BHP]/[BH]).

(Calculated slopes are tabulated in Table 23.) Only in the

latter portions of the titrations done in the more dilute

alkali metal electrolytes does this rule begin to break down.

Two, except for 0.15M CsH2PO4, all plots of the data in-

volving the presence of monovalent cations exhibit definite

curvature -- the lower the concentration, and further down

a series of the periodic chart, the greater the curvature.

Conversely, all concentrations of all M2+ salts yield

essentially straight lines over the whole titration

region. Three, in those cases in which there is no

curvature throughout the titration range, the slopes of

the lines are all nearly 1 and parallel the initial

regions of the other titrations.

That the values of q evaluated from both types of

plots are much the same in most media is evidence of the

validity of this technique. It appears that going from

q = 1 to q = 2 or 3 does not substantially affect the

values of log([Pt/m]-q[BHP]) except at the extrema of the

titrations. With our procedure, however, data very early

and very late in a titration need not be relied upon when

estimating lines, unlike, for instance, the Scatchard

method. For cases where the background electrolyte is

especially weak (i.e., low concentrations of cations having

low charge density) differences in the slopes of the two

plots become appreciable for the later portions of the

titrations. For example, the final slopes where 0.010M

Cs was employed as the background electrolyte are 1.68 and

1.88 for q = 2 and q = 1, respectively, and for 0.0025M Cs ,

the values are 1.71 and 2.16. Several explanations may be

advanced for this lack of agreement, including the fact

that the relatively short overall range of the amount of

DNA needed to complete the titrations magnifies the error

inherent at the extrema. Also, the high degree of curvature

makes the determination of the two separate straight line

regions more subjective. Perturbations of the reaction

system which are significant only in low ionic strength

media (e.g., mild polymer denaturation) would be expected

to affect both types of plots in the same manner. Re-

gardless of the sources of the discrepancies, the pro-

cedure is still useful, as both plots yield the same

integer values of q. Final selection of m and q is made

on the basis of the constancy of the computed values of

the equilibrium association constant videe infra).

The presence of at least two distinct linear regions

of the plots for lower concentrations of alkali metals

strongly indicates two binding stoichiometries. Numerous

investigators, as outlined in the Introduction, have shown

the presence of two types of binding processes which likely

account for the different values of q. What has not been

previously reported is the apparent change in the ratio of

phosphate to 3-aminoacridinium cation on going from one

type of binding to the other as evidenced by these plots.

Also, as the ionic strength is increased, the extent of

intercalative binding apparently decreases. This is mani-

fested in Figures 7 through 10 by a progressive trend

toward a wholly linear plot as the ionic strength increases.

That is, compare the shapes of the curves for systems con-

taining 0.0025M, 0.010M, and 0.15M CsH2PO4. Similarly, only
one stoichiometry is evident from the plots for M -containing

systems, which are linear throughout their entire titration

regions. Slopes of 1 + .1 for all of these, including that
for the reaction in the presence of 1 x 10-5N Ba(02CCH3),
for the reaction in the presence of 1 x 10 N Ba(O 2CCH 3)2'

indicates that only surface binding occurs in the

presence of the divalent metal cations over the titration

range investigated.

Previously, it was brought out that, in the presence

of low concentrations of monovalent cations as background

electrolytes, plots of log([BHP]/[BH]) vs. log([P t/m]-q[BHP])

exhibit two regions of linearity. During the early part of

a titration the slope of the lines is approximately 1 while

in the later region the slope is about 2. As the ionic

strength is increased, or as the charge density of the

cation of the electrolyte becomes greater, the slope of

the later region decreases, eventually becoming equal to

about 1. Slopes of the initial portions are essentially

unaffected by alterations in the supporting media in which

the titrations are conducted.

The ramifications of these relationships may be explored

by first defining the binding interaction in terms of con-

ventional metal-ligand chemistry. We shall analogize the

cationic 3-aminoacridinium to the metal center and the DNA

phosphates to ligands which bind to the center. For clarity,

we shall further define two types of ligands, having indepen-

dent and variable affinities for the cation centers; a weakly

bound ligand (corresponding to surface binding) which forms

a 1:1 complex, and a more strongly bound ligand (corresponding

to intercalative binding) which forms a 1:2 metal:ligand

complex. These definitions are based on the values of q for

surface-bound and intercalatively-bound ligand types being 1

and 2, respectively. Finally, we shall assume that the

ligands compete for metal binding sites such that if a

cation is bound in the surface mode, then it is unavailable

for intercalative binding, and vice versa.

During the initial stages of a titration there is a

large excess of 3-aminoacridinium over DNA phosphate or,

by analogy, an excess of metal over ligand. In such a

situation, there would be minimal competition for metal

sites and both forms of binding would be expected to occur

to their fullest extent, based on their respective association

constants. We may view the titration of the cation with DNA

ligand as a decrease in the metal center concentration as

the process proceeds. Addition of aliquots of DNA serves

to limit the number of sites available for binding, causing

the more weakly bound surface species to be affected first.

At this stage, the strongly bound intercalative species is

relatively unaffected. This argument presupposes that the

binding affinity of the intercalative mode is substantially

greater than that of the surface type. Hence, the first

region of the titration reflects surface binding of 3-

aminoacridinium to DNA. As the titration progresses, the

relative number of available binding sites becomes limited

to such an extent that the equilibrium involving the inter-

calated complex is the one predominantly affected. Now the

plot of log([BHP]/[BH]) vs. log([Pt/m]-q[BHP]) is a reflection

of equilibrium concentrations governed by the stronger mode

which, in turn, results in the slope of the line being 2

instead of 1. Note that the difference between the

relative binding affinities of the two modes must be

sufficiently great to allow one to predominate over the

other in the initial and final portions of the titrations.

On the other hand, if they are too disparate, only one

would be seen throughout the experimentally accessible

titration range. Increasing the ionic strength of a

reaction medium causes electrostatic contributions to the

total free energy to become less exothermic for a process in

which oppositely charged moieties coalesce. Based on this,

decreases in the apparent association constants for both

surface and intercalative binding would be anticipated. It

is assumed, at this stage of the argument, that changes in

electrostatic free energies are the predominant affects

accompanying alterations in ionic strength. Extrapolation

of the straight lines of Figures 7 through 10 to the vertical

axis shows that, as the ionic strength is increased, both

Ks and K1 do become less positive in agreement with the above

prediction. However, such extrapolations may be misleading

in the intercalative cases. Recall that, upon altering the

ionic strength, the slopes of the initial, surface binding

regions of the plots are essentially unchanged and that they

form a family of lines. This constancy of slope indicates

that there are no changes in the surface binding reaction

(aside from equilibrium shifts) as the ionic medium is

altered. However, two factors contribute to the decrease in

the intercept values of intercalative plots: (1) lower values

of log([BHP]/[BH]) for any given point along the abcissa

as ionic strength is increased, and (2) the decrease in

the slopes of the lines as the ionic strength is increased.

Two explanations for the change in slope may be advanced.

The value of q may be indeed changing from 2 to 1, for the

intercalation reaction, as the ionic strength is increased.

If this is true, then either the overall stoichiometry is

changing such that mq # 4 or, the stoichiometry is remaining

constant and m is becoming equal to 4 while q is decreasing

to 1. A more likely explanation can be advanced on the basis

of relative changes of K and K If K and KI do not change

in the same manner as the ionic strength is varied, then

what we may be observing is an apparent change in q for

intercalative binding when, in reality, the apparent change

is due to an overlapping of surface binding manifestations

with intercalative binding. Assume that the values of m and

q remain unchanged for each type of binding, then, the real

slopes of the two line segments will remain constant and

their point of intersection will be dependent on the ratio

of Ks to K For example, in one extreme case there is

essentially no intercalative binding, Ks/K is very large,

and only the line having slope 1 will be observed. At the

other extreme, Ks/K would be very small when only inter-

calative binding occurs and a single line having slope 2

would be seen. If the system is perturbed in a manner which

caused Ks to decrease more rapidly than KI the point of

intersection of the two lines will reflect the change in

K s/K In plots such as Figures 7 through 10, this would

be manifested as a shift toward the right, or to less

negative log([BHP]/[BH]) values. The total number of

sites which are involved in surface and intercalative

binding are, of course, directly related to the magnitudes

of the equilibrium constants. In other words, an increase

in Ks/KI indicates that a greater percentage of the total

occupied sites are occupied by surface-bound species. The

ratio of Ks to KI is seen to increase for the binding

reactions between 3-aminoacridinium and DNA in the presence

of alkali metal electrolytes as the ionic strength is raised'.

The following data are representative of Ks/KI x 105 at

ionic strengths of 0.0025, 0.0050, 0.025, 0.050, and 0.10(M),

respectively. With (CH3)4NH2PO4 as supporting electrolyte:

0.44, 0.56, 1.28, 1.83, and 2.9; whereas with LiH2PO4 as

supporting electrolyte; 0.43, 10.7, 1.2, 5.0, and 10.7,
respectively. Excepting the value of 10.7 x 10-5 for 0.005

LiH2PO4,the ratios show that as the ionic strength is

increased, the total percentage of bound drug which is

surface-bound increases. Moreover, for any given ionic

strength, the ratio is greater when tetramethylammonium

dihydrogen phosphate is the electrolyte than when the

lithium salt is present. It is obvious from the decreases

in both K and K that increasing the dielectric of the
s I
medium does not enhance surface binding relative to inter-

calative binding. Rather, K does not decrease as rapidly

as K The change in Ks/KI may result from specific ion

competition of the electrolyte cations with the drug

cations. As the ionic strength increases, the concen-

tration of cations which may directly compete for the DNA

phosphate ligands also increases. Moreover, an increase

in the effectiveness of competition by the metals for the

ligands would be expected as the charge density on the

metal increases. Therefore, specific ion interactions may

strongly affect surface binding. It is important to note

that no evidence from the experiments conducted with the

alkali and alkaline earth metals, as counterions, indicated

any alterations in surface binding stoichiometry resulting

from changes in the medium (aside from the magnitude of Ks).

That is, all initial slopes of figures such as Figure 7 are

equal to 1. Apparent changes in q for the intercalative

region are not likely to be due to intercalation of metal

cations because intercalation of the symmetrically charged,

highly hydrophilic metal cations is improbable. Specific

ion interference occurring at the surface may, however,

affect the overall intercalation of the drug cations.

Apparent Association Constants for
Surface and Intercalative Binding

Tables 24 through 32 include the apparent association

constants for surface binding, Ks, of 3-aminoacridinium

cation to DNA in the presence of various concentrations of

background electrolytes. For a given titration, values of

Ks were computed at each data point as discussed in the

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EX6FVDMW8_Y3NQFS INGEST_TIME 2012-02-29T18:02:58Z PACKAGE AA00009127_00001