CHEMICAL MODIFICATION OF ALPHA-AMANITIN TO
YIELD DERIVATIVES SUITABLE FOR CONJUGATION
TO PROTEINS VIA REDUCTIVE ALKYLATION
JERALD ERIC MULLERSMAN
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Words cannot fully express the gratitude that I wish to offer to
the many people who have encouraged and assisted me during the execu-
tion of this work. They have provided me with an environment in which
I have greatly enjoyed working and learning.
Many thanks go to Dr. James F. Preston who has shared with me his
delight in the mysteries of nature for many years. He has provided me
with numerous opportunities to test and develop my scientific abili-
ties and has given me patient and insightful counsel on innumerable
I would also like to thank the other members of my committee,
Drs. Hoffmann, Boyle, Small, and Wakeland. Each has earnestly sought
to enhance my education at e ery stage of this work.
In addition, I would like to gratefully recognize the assistance
of many other faculty members. Dr. Lonnie Ingram has helped at every
turn, sharing freely of equipment, advice, and his good nature. Dr.
John Gander provided advice on several matters and instruction in the
use of the CD spectrometer and digital polarimeter. Sandra Bonetti and
Roy King were instrumental in the acquisition of NMR spectra and ele-
mental analysis data, respectively.
I want to thank fellow graduate students, Mike Little, Tony
Romeo, and Dave Dusek, for their friendship and receptive ears.
TABLE OF CONTENTS
ACKNOWLEDGEMENTS. ...................... .... ... ... ..
LIST OF ABBREVIATIONS.... ..... ..............
ABSTRACT.. .. . . .. .. . . . . . . .
CHAPTER I. BACKGROUND AND RATIONALE................
Modified Proteins As Therapeutic Agents...........
Conjugates of Amatoxins with Proteins.............
Toward Optimal Conjugation of Small Molecular Weight
Cytotoxic Agents to Proteins.................
Rationale for the Current Work. ........ ......
CHAPTER II. STUDIES ON THE NATURE OF 6'-0-METHYLALDO-a-AMANITIN.
Introduction .. . . . .. .. . . . . .
Materials and Methods...... .... ... .......
Results.......... . .. .. . . . . . . .
Discussion... . .. . .. . . .. . . . .
CHAPTER III. TRANSFORMATIONS OF 61-0-METHYLALDO-a-AMANITIN
VIA REDUCTIVE AMINATION AND OXIDATION........
Introduction.. ... . . . . . . . . . .
Materials and Methods.... ..... .... .... ...
Results and Discussion.....................
CHAPTER IV. STUDIES ON THE CONJUGATION OF N-ACYLATED AMINO
SUGARS TO BOVINE SERUM ALBUMIN BY MEANS OF
Introduction............ ... ... ...... .....
Materials and Methods.......... ... .......
Results.... . . . . . . . , ,
Discussion... .. . . .. .. .. . . .. .
CHAPTER V. CONJUGATION OF A NOVEL AZO AMANITIN TO BOVINE SERUM
ALBUMIN VIA REDUCTIVE ALKYLATION WITH SODIUM
CYAN080ROHYDRIDE... ........... ......
Introduction...... . .. . . . .. . . . ..
Materials and Methods.....................
Results........... . . . . . . . . . . .
Discussion...... . . . . . . . . . . .
CHAPTER VI. CONCLUSIONS........ ... ...........
BIBLIOGRAPHY... . .. . . . . . . . . .. ..
LIST OF ABBREVIATIONS
BSA.................bovine serum albumin
cinn-HC1........,...trans-cinnamaldehyde-C fumes; TLC spray reagent
CT RNAP II..........calf thymus RNA polymerase II
Fab.................univalent antigen-binding antibody fragment
(Fab')2..........divalent antigen-binding antibody fragment
FAB-MS..............fast atom bombardment mass spectroscopy
HPLC................high-performance liguid chromatography
OMA-X (X =CN, C00H, gly, NH
and pro).......amanitin2dlerivatives defined in the text
PMR................,proton magnetic resonance
RF...............index of TLC mobility; quotient of analyte and
and solvent front migration distances
SMWCA...............small molecular weight cytotoxic agent
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
CHEMICAL MODIFICATION OF ALPHA-AMANITIN TO
YIELD DERIVATIVES FOR CONJUGATION
TO PROTEINS VIA REDUCTIVE ALKYLATION
JERALD ERIC MULLERSMAN
Chairman: James F. Preston, III
Co-chairman: Edward M. Hoffmann
Major Department: Microbiology and Cell Science
A toxic peptide which inhibits RNA polymerase II (RNAP II),
e-amanitin, was chemically modified to generate aldehydic derivatives
which were suitable for reductive coupling to proteins.
Periodate oxidation of 6'-0-methyl -a-amanitin (0MA) at neutral pH
generated a mixture of two amanitin aldehydes which underwent ready
interconversion. These two forms of 6'-0-methylaldo-a-amanitin (0MAA)
were reduced to the corresponding alcohol with sodium borohydride, but
were inert to treatment with sodium chlorite at pH 2.0 or 3.0. A
chemically distinct form of OMAA was produced by periodate oxidation
at pH 2.0. Spectral data, in combination with the finding that this
compound is easily oxidized to the corresponding carboxyl derivative
by sodium chlorite at pH 2.0, indicated that this latter form of OMAA
contains a free aldehyde group.
The reductive amination of OMAA with several different amines
proved to be very slow, even when high reactant concentrations were
used. The products of reaction with ammonium acetate, glycine, and
L-proline were all relatively poor inhibitors of calf thymus RNAP II
with K 's of 1.7 X 10-7 M, 2.5 X 10-7 M, and 7 X 10-6 M, respec-
tively. Conversion of the aldehyde moiety of OMAA to carboxyl and
nitrile groups yielded derivatives with K.'s of 1 X 10-7Man
3 X 10- M, respectively. These data, in conjunction with those
previously published, suggest that introduction of ionizable groups
into this part of the amanitin molecule substantially diminishes its
An aromatic amine attached to an amino sugar via a dipeptide
linker was synthesized and coupled to the 7-position of the hydroxy-
tryptophan residue of (r-amanitin to yield a novel azo amanitin
(ABGG-GLU). Marked effects of borate, temperature, cyanoborohydride
concentration, nature of the sugar residue, and pH upon the rate of
conjugation of a model compound to bovine serum albumin (BSA) were
delineated. Conjugation of ABGG-GLU itself to BSA was complicated by
side reactions. Several combinations of reaction conditions were
examined to identify ways to lower the rate of loss of ABGG-GLU to
these side reactions. Raising the concentrations of BSA and ABGG-GLU
provided the most effective means of achieving this goal. ABGG-GLU-BSA
conjugates demonstrated Ki's of approximately 10-7 M, which are
comparable to those obtained by others for amanitin-BSA conjugates.
The potential utility of this conjugation method is discussed.
BACKGROUND AND RATIONALE
Modified Proteins as Therapeutic Agents
Study of the chemical modification of proteins began largely as
an approach to understanding the role that particular chemical moie-
ties play in the structural integrity and function of proteins. Selec-
tive modification is still applied in this way and is an area of in-
vestigation in which significant strides continue to be made. In addi-
tion to its continued use in research activities, protein modification
is currently used to produce a large variety of products. For example,
proteins labeled with various enzymes and dyes are being used as very
important scientific and diagnostic tools.
A relatively new area which has been under development is the
application of modified proteins as therapeutic agents. The approaches
to using modified proteins as therapeutic agents are legion (Sezaki
and Hashida, 1984; Zaharko et al., 1979). Some simply aim at utilizing
a protein as a convenient polymeric support to which a drug can be
attached with the hope of a favorable alteration of the drug's pharma-
cokinetic properties (Hurwitz et al., 1980), but considerable atten-
tion has been focused more recently on the use of proteins which bind
to specific structures in order to help a therapeutic agent to "home
in" on the cell or tissue which is the desired target.
This latter avenue has been most extensively studied with the
hope of developing better ways to treat neoplastic conditions, since
current therapies for many cancers are associated with high levels of
morbidity and mortality and often have relatively low efficacy. Anti-
bodies that recognize tumor-associated antigens (Zalcberg and
McKenzie, 1985) have come to be the preferred target-specific carriers
of therapeutic agents (Ghose et al., 1983; 01snes and Pih1, 1982;
Thorpe et al., 1982), especially since the hybridoma method (Kijhler
and Milstein, 1975) has made antibodies of a single specificity avail-
able in almost unlimited quantities.
Conjugates of Antibodies with Protein Toxins
The development of modified antibodies as cancer therapeutic
agents has proceeded primarily along two lines, conjugates made with
enzymes and those made with a variety of small molecular weight cyto-
toxic agents (SMWCAs). The enzymatic conjugates have been prepared
mainly with toxins which inactivate ribosomes and are called immuno-
toxins (ITs) (Blythman et al., 1981). The ribosome-inactivating pro-
teins which have been employed to prepare ITs (reviewed in Stirpe and
Barbieri, 1986) include ricin (Jansen et al., 1982; Raso, 1982;
Vitetta et al., 1982), diphtheria toxin (Moolten et al., 1982),
gelonin, mordeccin, abrin, pokeweed antiviral protein, and saporin
(Thorpe et al., 1985a).
Although many ITS have shown potent and specific toxicity toward
cells in vitro, application of these conju ates to the treatment of
tumor-bearing animals has generally yielded discouraging results. As a
consequence, clinical application of ITs has evolved most rapidly only
in the few circumstances where they can be used in vitro. This has
mainly been limited to freeing bone marrow of unwanted populations of
cells before it is transplanted (e.g., Filipovich et al., 1985).
In part because of the great promise of monoclonal antibodies and
their conjugates in many therapeutic and diagnostic areas, several
studies have been undertaken during the last few years in order to
identify those factors which influence the localization of antibodies
and their conjugates in whole animals. The important parameters which
have been identified by these studies include the kinetics of conju-
gate access to the target cells and the kinetics of conjugate clear-
ance. [These factors and many other issues related to the in vivo
performance of antibody conjugates have been critically reviewed
(Poznansky and Juliano, 1984).] From the available data, it appears
that the accessibility of tumor cells to conjugates is most strongly
correlated with the molecular weight of the conjugate. Smaller frag-
nents of antibodies have more rapid access to the cells which lie in
extravascular tissue spaces (Herlyn et al., 1983; Houston et al.,
1980). The localization of (Fab')2 fragments to target cells is not
only more rapid but more extensive than that of whole antibody. How-
ever, the utility of very small antibody fragments, such as Fab frag-
ments, may be limited by their loss from the circulation due to glo-
merular filtration (Arend and Silverblatt, 1975; Covell et al., 1986).
An even more important consideration is that the conjugate must not be
quickly swept from the bloodstream. Three groups of workers have re-
cently reported their studies on the mechanism of rapid clearance of
ricin and its conjugates from the circulation (Bourrie et al., 1986;
Skilleter and Foxwell, 1986; Worrell et al., 1986a; Worrell et al.,
1986b). They have found that these proteins were cleared from the
bloodstream in minutes by hepatic nonparenchymal cells which recognize
the mannose-containing oligosaccharides borne by ricin. This has
prompted efforts to modify ricin either enzymatically (Foxwell et al.,
1985) or chemically (Skilleter et al., 1985; Thorpe et al., 1985b) in
order to abrogate this unwanted problem. Pokeweed antiviral protein is
not a glycoprotein (Stirpe and Barbieri, 1986) and so would not be
expected to be cleared from the circulation by the same sort of mech-
anism as ricin. In fact, Ramakrishnan and Houston (1985) have reported
circulatory half-lives for pokeweed antiviral protein-antibody conju-
gates of approximately 24 hours.
Conjugates of Antibodies with Small Molecular Weight Cytotoxic Agents
The other major approach to the production of novel cancer thera-
peutic agents by means of modification of antibodies has utilized
SMWCAs. In most instances, investigators have employed chemotherapeu-
tic drugs which are already being applied to the treatment of cancer.
Thus, the antibody moiety is intended to provide additional specifi-
city of action to a drug which already has some specific antitumor
activity. Drugs which have been commonly employed include methotrexate
(Garnett et al., 1983; Kanellos et al., 1985; Kulkarni et al., 1981),
daunorubicin (Arnon and Sela, 1982; Shen and Ryser, 1981), vindesine
(Ford et al., 1983), radioisotopes (Badger et al., 1985; Meares et
al., 1984), and various alkylating agents (de Weger et al., 1982).
These conjugates have typically had much lower potency in vitro than
those prepared with the protein toxins. Despite this, experiments with
animals and small clinical trials with cancer patients have, in some
cases, been quite encouraging.
Some of the same factors outlined above (e.g., size of the conju-
gate) which modulate the efficacy of the protein toxin conjugates in
vivo, may also influence the efficacy of the SMWCA conju ates. In
addition, there may be other mechanisms by which SMWCA conjugates are
cleared rapidly from the circulation by the reticuloendothelial sys-
tem. This will be discussed in greater detail below.
Immun~otoxins Versus Conjugates Made with Small Molecular Weight
As of yet, a clear choice between ITs and SMWCA conjugates as the
superior alternative cannot be made. Each type of conjugate has its
own distinct advantages and disadvantages. Because of the extremely
high potency of the enzymatic toxins (Yamaizumi et al., 1978), ITs
need only contain a single molecule of toxin per antibody molecule.
Consequently, any deleterious effect on the antigen-binding activity
of the antibody molecule may be small. On the other hand, the low
molecular potency of the SMWCAs makes it desirable to prepare conju-
gates with as many SMWCA molecules linked to each antibody molecule as
can be tolerated. Unfortunately, difficulties with low solubility of
the conjugates and loss of antigen-binding activity often intervene at
relatively low levels of conjugation. For example, antibody conjugates
prepared with methotrexate have been reported to lose solubility and
antigen-binding activity with greater than ten (Kulkarni et al., 1981)
or twelve (Kanellos et al., 1985) molecules of methotrexate per anti-
body molecule. Ford et al. (1983) indicated that antibody conjugates
with more than eleven molecules of vindesine became poorly soluble.
Some workers have sought to circumvent this problem by linking the
SMWCA first to a highly soluble carrier and then coupling the loaded
carrier to an antibody. For example, Garnett et al. (1983) conjugated
an average of 32 molecules of methotrexate to each molecule of human
serum albumin. This drug-albumin conjugate was then linked to a mono-
clonal antibody. Another group of workers has used poly-L-glutamic
acid as an intermediate carrier for daunorubicin (Kato et al., 1984;
Tsukada et al., 1984). An advantage of the SMWCA conjugates is the
relative insensitivity of the SMWCAs to lysosomal hydrolases, while
the protein toxins are apparently very susceptible to proteolytic
digestion in 1ysosomes. Ammonium chloride has been found to greatly
increase the potency of ITs, presumably by suppressing the function of
lysosomes in the target cells (e.g., Jansen et al., 1982). Also in
their favor, the conjugated SMWCAs need not greatly increase the mo-
lecular weight of the antibody, while the protein toxins generally
have significant molecular weights (greater than 30,000 daltons). As
already discussed above, the higher molecular weight may have a sig-
nificant negative impact upon the rate at which the conjugate can
reach the tumor from the intravascular space.
Conjugates of Amatoxins with Proteins
Conjugation of amatoxins to proteins, which is the subject of
this work, has been studied in several laboratories (reviewed in
Faulstich and Finme, 1985). Amatoxins are the constituents of certain
mushrooms which cause the vast majority of fatal mushroom poisonings.
Work by Wieland and his colleagues has revealed the amatoxins to be
cyclic octapeptides which have several unusual amino acid residues
(Fi. I1).These peptides exert their toxic action by inhibiting the
C t O CH2 H CO
/ 'CtH ,`C/C"C"H NH
HO-CH r; C'I R ICH
,N O= S '-', R
CH2 CO CH2 HH C
Fig. I-1. Structures of
amanull in OH
some pertinent amanitins.
DNA-dependent RNA polymerases which are responsible for transcription
of nuclear genes. The inhibition of the class II enzyme, which gener-
ates messenger RNA, is particularly potent with a K of approximately
10-' M (reviewed in Wieland, 1983).
The first reports concerning amatoxin-protein conjugates arose
from efforts to prepare amatoxin-specific antisera. An amatoxin with a
free carboxyl group, R-amanitin, was linked to bovine serum albumin
(BSA) with the aid of a water-soluble carbodiimide. When administered
to rabbits, these conjugates proved to be much more toxic than the
free toxin (Bonetti et al., 1976; Cessi and Fiume, 1969). Subsequent
pathological studies showed that this increased toxicity was due to
uptake of the conjugate by Kupffer cells (Derenzini et al., 1973).
Thus, this represented accidental targeting of the toxin into the
reticuloendothelial system by virtue of its conjugation to BSA. An
increase in toxicity of this conjugate relative to free toxin was also
demonstrated for macrophages in culture (Barbanti-Brodano and Fiume,
1973; Fiume and Barbanti-Brodano, 1974).
A number of conjugates have also been prepared with azo
derivatives of a-amanitin (Faulstich and Trischmann, 1973; Preston et
al., 1981) which have been referred to as ABH and ABGG (short for
a-amanitiny1-7'-azobenzoy1-1,6-diaminohene and a-amanitiny1-7'-azo-
benzoylglycylglyci ne; previously called ADH and ADGG). Both of these
amatoxin derivatives have been coupled to proteins with the aid of a
water-soluble carbodiimide. When the toxicity of conjugates made from
BSA and ABH was tested on several cell lines, a positive correlation
between toxicity and phagocytic rate was found (Hencin and Preston,
1979). Conjugates prepared with the lectin concanavalin A proved to be
potent cytotoxins, the toxicity of which could be greatly diminished
by the addition of certain sugars which are bound by the lectin and
thus inhibit binding to the cell surface (Hencin, 1979). Linkage of
ABGG to a monoclonal antibody which recognizes Thy 1.2, an antigenic
marker of mouse T cells, yielded a conjugate with potent, specific
toxicity in vitro toward cells which bear the Thy 1.2 antigen (Davis
and Preston, 1981).
Because of certain characteristics of amatoxins, the amatoxin-
antibody conjugates fall within a class which is distinct from either
ITs or SMWCA-antibody conjugates. Amatoxins are like the ribosome-
inactivating enzymes used in ITs in that they are nonspecific cyto-
toxins, but they are like the SMWCAs in that they have a relatively
small molecular weight and resistance to hydrolases. Someday the
amatoxin-antibody conjugates may prove to be useful as therapeutic
agents and, as members of a distinct class, they may be uniquely
suited to some applications.
The amatoxins share with the SMWCAs a relatively low molecular
potency compared to the ribosome-inactivating toxins. Thus, one can
anticipate the need for relatively high levels of conjugation to anti-
body, especially in systems where there are few target antigenic sites
per cell. However, for amatoxin-antibody conjugates the importance of
retaining antigen-binding activity and favorable pharmacokinetic be-
havior (i.e., long circulatory half-life and rapid penetration to tar-
get cells) is more critical than for the conjugates made with the
chemotherapeutic drugs. This is true because the drugs possess consid-
erable selective toxicity toward many types of neoplastic cells with-
out being conjugated to an antibody, while amatoxins do not. Amatoxin-
protein conjugates can potentially kill any cell which takes them up.
These considerations have prompted a search for a mode of conjugation
of SMWCAs (as exemplified by amanitin) to antibodies which would re-
present a theoretical optimum.
Toward Optimal Conjugation of Small Molecular
Weight Cytotoxic Agents to Proteins
Because of the wide utility of chemically modified antibodies, a
number of investigators have compared the effect of different kinds of
modifications on the ability of the antibodies to retain their anti-
gen-binding activity. Most of the currently employed conjugation meth-
ods rely upon reaction with the amino groups of the antibody. This is
because many of the amino groups are disposed toward the surface of
the protein, they are reactive as nucleophiles at neutral to mildly
alkaline pH's, and they can be modified with high selectivity with
several classes of reagent.
A reasonable premise might be that for the conjugation of any
given chemical entity (e.g., SMWCA) to an antibody, a method which
preserves the native charges of the antibody should produce the least
possible alteration in its structure. In fact, modifications which
preserve the positive charge of the antibody amino groups, including
reductive alkylation and amidination, generally have a lesser effect
on antigen-binding activity than those modifications which abolish the
charge of the antibody amino groups (see below). Both amidination and
reductive alkylation are reactions which are highly specific for amino
groups of proteins. The utility of amidination is somewhat limited by
the harsh conditions required for generating the imidate esters which
are the amidinating reagents. On the other hand, reductive alkylation
with sodium cyanoborohydride (Borch et al., 1971) and an aldehyde has
received increasing attention as an excellent method for specifically
and gently modifying proteins (Hutchins and Natale, 1979).
Cohen and Becker (1968) have shown that the precipitin activity
of several different anti-hapten antibodies was greatly inhibited by
low levels of carbamylation, while it was preserved at high levels of
amidination with ethyl acetimidate. The carbamy1 and amidino groups
are similar in size and shape; they differ principally in that the
amidino group bears a positive charge, while the carbamy1 group is
uncharged. Thus, this study demonstrated especially well the impor-
tance of preserving the native charges of the antibody. Studies of
this kind have prompted the introduction of amidinating reagents for
the haptenation of antibodies to be used in hapten-sandwich tech-
niques. Wofsy and his colleagues (1978) have synthesized azo haptens
which can be coupled to antibodies at levels greater than thirty hap-
ten moieties per antibody molecule with good preservation of the anti-
gen-binding activity. A similar reagent for introducing the dinitro-
phenyl group has recently been reported which can be used to prepare
conjugates with up to thirteen dinitrophenyl groups per antibody mole-
cule without loss of antigen-binding activity (Hewlins et al., 1984).
Radiolabeling of antibodies by reductive methylation with tri-
tium-labeled sodium borohydride and formaldehyde has gained in popu-
larity because of the high degree of preservation of antigen-binding
activity which is achieved with this method (Tack and Wilder, 1981).
Reductive alkylation has, however, not been frequently used to attach
larger chemical moieties to antibodies despite the good results which
have been obtained with reductive methylation.
In addition to deleterious effects on antigen-binding activity,
chemical modifications of antibodies which remove positive charges may
also depress their circulatory half-life. Several kinds of chemical
modification have been found to promote clearance of proteins from the
circulation of mammals. These modifications include acylation, reac-
tion with formaldehyde, and dinitrophenylation. Much of the work on
acylated proteins has focused on lipoproteins (Brown et al., 1980).
Cells of the reticuloendothelial system bear specific, high-affinity
receptors for certain acylated proteins (Mahley et al., 1980;
Nagelkerke et al., 1983; Pitas et al., 1985). Formaldehyde-modified
albumin is taken up by the same cells, but apparently by a separate
receptor (Blomhoff et al., 1984; Horiuchi et al., 1985; Horiuchi et
al., 1986). Dinitrophenylated albumin has also been found to be
cleared by the reticuloendothelial system, primarily in the liver
(Kitteringham et al., 1985; Rhodes and Aasted, 1973; Skogh, 1982;
Skogh et al., 1983). These three classes of chemical modification each
reduce the number of native positive charges on the proteins and thus
make them more anionic. In the case of the receptor for acylated lipo-
proteins it has been specifically shown that certain polyanions can
compete for binding to the receptor (Mahley et al., 1980). This sug-
gests that the anionic character of the modified protein contributes
to its affinity for this receptor. Two groups of workers have observed
a correlation between increasing localization to the liver of anti-
bodies coupled with diethylenetriaminepentaacetic acid, a metal che-
lator, and increasing levels of conjugation (Anderson and Strand,
1985; Sakahara et al., 1985). In these studies, the chelator was
linked to the antibodies by an acylation reaction. Winkelhake (1977)
has also noted substantial decreases in the circulatory half-life of
an acylated antibody, while reductively methylated antibody was
cleared much more slowly.
Unfortunately, there is relatively little information concerning
the fate of SMWCA-antibody conjugates in vivo. The information ob-
tained in other systems, as described above, suggests that the most
frequently used method of preparing SMWCA-antibody conjugates (i.e.,
acylation) may not be optimal. The available data also suggest that a
conjugation method which preserves the native charges of the antibody,
such as amidination or reductive alkylation, may provide better reten-
tion of antigen-binding activity and longer circulatory half-life.
Rationale for the Current Work
The studies described below have been directed toward the devel-
opment of a practical method by which a suitable amatoxin derivative
can be conjugated to proteins by means of reductive alkylation. Devel-
opment of a method of this kind would provide a future opportunity to
examine its usefulness relative to existent methods for conjugation of
amatoxins to antibodies, especially with regard to any effects on
antigen-binding activity and circulatory half-life in animals.
Preparation of an amatoxin useful for reductive alkylation
requires the introduction of an aldehyde group into the molecule.
Fig. I-2 shows the sites at which a-amanitin (AMA), the most readily
available amatoxin, can be easily and selectively modified. The
dihydroxyisoleucine residue (site A) is susceptible to oxidation by
periodate to yield an aldehyde (Wieland and Fahrmeir, 1970). The
7-position of the hydroxytryptophan residue (site 8) is subject to
Os S H OH
Fig. I-2. Sites for selective chemical modification of a-amanitin.
electrophilic substitution (Faulstich and Trischmann, 1973; Morris and
Venton, 1983). For example, relatively complex aromatic amines can be
diazotized and reacted with AMA to yield azo derivatives like ABGG.
Also, the hydroxyl group of the hydroxytryptophan moiety (site C) has
been alkylated with alky1 halides to prepare useful derivatives
(Faulstich et al., 1981).
Two of these three possible sites were chosen for evaluation.
Since periodate oxidation of the dihydroxyisoleucine residue yields an
aldehyde function in high yield, the feasibility of using this ap-
proach was examined. Also, since the azo derivative ABGG had already
been utilized to synthesize several potent conjugates, a derivative of
ABGG which contained an aldehyde function was prepared and studied.
STUDIES ON THE NATURE OF 6'-0-METHYLALDO-a-AMANITIN
The oxidation of 6'-0-methyl-a-amanitin (0MA) with sodium
periodate to yield 6'-0-methylaldo-a-amanitin (0MAA) was first
reported in 1970 (Wieland and Fahrmeir). Like amanullin (Fig. I-1),
OMAA was demonstrated to be nontoxic when injected into mice. However,
when the aldehyde function of OMcAA was reduced with sodium borohydride
to a hydroxyl group, the resultant derivative, 6'-0-methyldehydroxy-
methyl-a-amanitin (0MDA), proved to be toxic to mice. These and other
data prompted the hypothesis that the y-hydroxyl of the dihydroxy-
isoleucine (residue 3) sidechain was critical for the toxic action of
amatoxins (Wieland and Fahrmeir, 1970).
After it was learned that the amatoxins express their toxicity by
means of potent inhibition of RNA polymerase II (RNAP II), studies
showed that OMAA was a very poor inhibitor of RNAP II (Buku et al.,
1971), while amanullin was found to be a potent inhibitor (Cochet-
Meilhac and Chambon, 1974). The low toxicity of amanullin in mice has
been rationalized in terms of unusual pharmacokinetic properties. A
possible explanation for the low inhibitory activity of OMAA toward
RNAP II was found when its circular dichroism (CD) spectrum was
discovered to be greatly different from that of other amanitins. This
alteration in the CD spectrum and attendant loss of toxicity were
interpreted as being due to a change in the conformation of the
peptide backbone. It was proposed that the conformational change was
induced by the formation of an intramolecular hydrogen bond to the
oxygen of the aldehyde (Faulstich et al., 1973). Further support for
an altered conformation of OMAA has come from radioimmunoassay
studies. Faulstich and Wieland (1975) found that more than 120 times
as much OMAA as compared to AMA was required to displace half of the
radiolabeled amatoxin from amatoxin-specific antibodies.
Unpublished work by Dr. James F. Preston has suggested that there
are actually at least two products which result from the periodate
oxidation of OMA and which can be chromatographically separated on a
column of Sephadex LH-20 eluted with water. These oxidation products
demonstrated the same mobility on TLC and reacted with trans-
cinnamaldehyde and HC1 (cinn-HC1) to yield the rust color which is
distinctive of OMAA (Wieland and Fahrmeir, 1970). Each one also was
reduced by sodium borohydride to a compound with the characteristics
of OMDA, including violet color reaction with cinn-HC1 and potent
inhibition of calf thymus RNA polymerase II (CT RNAP II).
The nature of these two periodate oxidation products was reeval-
uated here in order to determine whether one or both components should
be used in reductive alkylation studies and to obtain insights into
their chemistry which might be pertinent to the intended conjugation
to protein amino groups.
Materials and Methods
Reagents. All chemicals were analytical reagent grade unless
otherwise specified. e-Tolylsulphonylmethylnitrosamide and
tetramethylsilane (TMS) were obtained from Aldrich Chemical; sodium
periodate, sodium borohydride, 0-glucose, and d6-dimethylsulfoxide
(dg-DMSO) from Sigma Chemical; and sodium chlorite and amidosulfonic
acid from Alfa Chemical. Solvents and trifluoroacetic acid (TFA) were
obtained from Fisher Scientific. Water used in the experiments
described in this and subsequent chapters was deionized and then
distilled from glass.
Toxin. The amatoxin AMA was purified from carpophores of Amanita
suballiacea (Murt.) Murr. collected in the Gainesville, Florida, area
by a modification of methods previously described (Little and Preston,
1984). Briefly, the toxin was obtained as follows: Coarsely chopped
carpophores (800 g) were extracted with 1.2 1 of methanol for
approximately 24 h. The filtered and concentrated crude toxin solution
was extracted three times with three volumes of diethyl ether. The
defatted aqueous phase was then mixed with nine volumes of methanol
and refrigerated overnight. This was then filtered to remove the
precipitated polar compounds (primarily carbohydrates and salts).
After concentration the toxin was partially purified by column chroma-
tography on Sephadex LH-20 eluting with 50% methanol, on Bio-gel P-2
eluting with water, and on Sephadex LH-20 eluting with water. The
remaining impurities were removed by high performance liquid chromato-
graphy (HPLC) on a Zorbax ODS column (0.94 X 25 cm, Du Pont) eluting
with 15% acetonitrile and/or by recrystallization from methanol.
Estimates of the concentration of aqueous amanitin solutions were
based upon an extinction coefficient at 304 nm of 15,400 M cm (cf.
Cochet-Meilhac and Chambon, 1974). Because of the pH-dependence of its
absorbance spectrum, measurements of absorbance at 304 nm of AMA were
made in 5 mM sodium phosphate buffer, pH 7.0. Yields of amanitin
derivatives were calculated from these estimates of concentration and
the volume of the solutions.
Spectroscopy. Absorbance measurements were made with the aid of a
Gilford 2400 spectrophotometer. Absorbance spectra in both the ultra-
violet (UV) and visible ranges were obtained on either a Beckman 25
spectrophotometer or a Hewlett-Packard 8451A diode array spectrophoto-
meter. Circular dichroism (CD) spectra were recorded on a Jasco J-500C
spectropolarimeter with the aid of Jasco IF-500 and Okidata IF-800
data processing equipment. Proton magnetic resonance (PMR) spectra
were obtained on a Nicolet NT-300 spectrometer operating at 300 MHz in
the Fourier transform mode.
Chromatography. Thin-layer chromatography (TLC) was performed on
0.25 mm layers of silica gel 60 F254 (Merck). TLC solvent system I
contained 1-butanol: acetic acid: water (4:1:1). Amatoxins were
detected on chromatograms by means of their quenching of the layer's
fluorescence and by their color reaction after being sprayed with 2%
methanolic trans-cinnamaldehyde and subsequent exposure to HC1 fumes
(cinn-HC1). HPLC was performed with a Waters model 6000A pump which
was equipped with a U6K injector. Elution of compounds was detected by
a Gilson Holochrome variable wavelength absorbance monitor. Samples
were chromatographed on a Zorbax ODS column (0.94 X 25 cm, Du Pont)
which was protected by a 0.5 pm prefilter (Rainin) and a C5 guard
cartridge (Bio-Rad). Samples for HPLC were filtered through 0.22 pm
Miller GV or SR filters (Millipore).
RNA polymerase assay. Calf thymus RNA polymerase II (CT RNAP II)
was prepared and stored as previously described (Preston et al.,
1975). Components of the reaction mixture were those of Cochet-Meilhac
and Chambon (1974) except that C"H]UTP was employed for labeling the
product. The assay was performed as follows: The reaction tube with
10 pl of amanitin solution was incubated at 37*C for 10 min. The
enzyme solution was added to this and allowed to incubate for 10 min
more at 37*C. The reaction was started by adding a solution containing
the nucleoside triphosphates to yield a final reaction volume of 100
l.After 10 min at 37"C the reaction was stopped and processed for
scintillation counting as previously described (Preston et al., 1975).
Counting was performed on Beckman LS-133 and LS-8000 liquid scintil-
lation counters. Inhibition data were analyzed according to Dixon
(1953). The best fit of lines to data points was estimated by the
method of least squares using the statistical functions of a Texas
Instruments TI-55-II calculator.
Synthesis of Amanitin Derivatives
Synthesis of 6'-0-methy1-a-amanitin. To 30 plmol of AMA in 10.5 ml
of ice-cold methanol were added 4.5 ml of an ice-cold ethereal-
ethanolic solution (de Boer and Backer, 1954) of diazomethane which
was generated by the action potassium hydroxide on e-tolylsulphony1-
methylnitrosamide and adjusted to 0.10 M with ice-cold ether. The
diazomethane solution was titrated according to Arndt (1943) prior to
use. (Note: Diazomethane is a toxic and potentially explosive gas. It
should only be prepared and used in a well-ventilated hood, preferably
behind a safety shield, in glassware which lacks ground or abraded
surfaces.) The reaction mixture was permitted to rise to room tempera-
ture in a water bath in the dark. After 30 min the remaining diazo-
methane and the bulk of the ether were evaporated under a stream of
nitrogen. The resultant solution was concentrated to dryness in a
rotary evaporator at 30*C, redissolved in water, filtered, adjusted to
22% acetonitrile, and subjected to HPLC with the same solvent mixture.
The desired product eluted between 24 and 32 mi. The yield was 14.6
plmol (49%); 14.4 pmol (48%) of the AMA was recovered unchanged. TLC:
RF, I-0.47; color-reaction, violet.
Synthesis of 6'-0-methylaldo-a-amanitin under neutral conditions.
To 20.3 pmol of OMA in 1.0 ml of water was added 0.5 ml of a solution
containing 22.0 pmol of sodium periodate. After 5 min of reaction in
the dark at room temperature residual sodium periodate was quenched by
the addition of 10 pll of ethylene glycol. The solution was filtered,
adjusted to 22% with respect to acetonitrile, and subjected to HPLC
eluting with 22% acetonitrile. The desired products eluted between 28
and 45 ml. The yield was 18.8 pmol (92%). TLC: RF, I-0.43;
color -react ion ru st -red .
Synthesis of 6'-0-methyl dehydroxymethy1 -a- amani tin. 0MA (2.0
pmol) was oxidized with periodate as described above. The components
of the reaction mixture were separated by HPLC eluting with 18%
acetonitrile in order to achieve complete separation of the products.
The resultant fractions were placed on ice as they were collected.
Under these chromatographic conditions the products OMAA-IA and
DMAA-IB eluted between 57 and 67 ml and between 85 and 95 ml, respec-
tively. To 0.20 pmol each of OMAA-IA and OMAA-IB was added 1.0 pmol of
sodium borohydride in 80 pl of water. After gentle mixing, this was
permitted to react for an hour in the dark at 230C. Residual sodium
borohydride was quenched by reaction with 10 pl of 1.0 M D-glucose for
a further hour under the same conditions. The reaction mixture was
rinsed from the vial with 1.0 ml of 22% acetonitrile and then chroma-
tographed by HPLC using the same solvent. A single product eluted
between 29 and 34 ml in 75-90% yield. TLC: RF, I-0.51; color-reaction,
Synthesis of 6'-0-methylaldo-a-amanitin under acidic conditions.
To OMA (8 pimol) in 0.75 ml of 50 mM sodium phosphate, pH 3.0, was
added sodium periodate (8.8 pmol) in 30 pl of water. After reaction
for 5 min at 230C in the dark, the solution was adjusted to 22% with
respect to acetonitrile and applied to HPLC eluting at 1 ml/min with
acetonitrile: 0.05% TFA (22:78). The desired product (7.0 Clmol, 88%)
eluted between 30 and 35 ml. TLC: RF, I-0.61; color-reaction, violet.
One-half of this material was prepared for PMR studies. After the
acetonitrile was removed at 30*C in vacuo, the sample was lyophilized
and redissolved in 1.0 ml of d6-DMSO.
Synthesis of 6'-0-methy1-a-amanitin
The synthesis of OMA has been patterned after that of Wieland and
Fahrmeir (1970). The major disadvantage of this synthetic route is the
use of the dangerous diazomethane; alternatively, AMA can be methyl-
ated with methyl iodide (Faulstich et al., 1981). Studies on the
synthesis of OMA using diazomethane (Little, 1984) have demonstrated
that the choice of reaction conditions is quite critical. Both reac-
tant stoichiometries and concentrations are important. At stoichio-
metries and/or concentrations less than the optimum the yield of OMA
falls rapidly, but virtually quantitative recovery of input toxin is
achieved as the sum of AMA and OMA. At stoichiometries and/or concen-
trations greater than the optimum the yield of OMA falls because of
the production of byproducts, one of which has been identified as
1',6'-N,0-dimethyl-a-amanitin (Little, 1984). The conditions described
here are similar to those of Little (1984) and provide for recovery of
virtually all of the input toxin as the sum of OMA and recovered AMA.
OMA shares many characteristics in common with AMA, including
potent inhibition of CT RNAP II, UV absorbance spectrum, and violet
color reaction with trans-cinnamaldehyde. However, it can be readily
distinguished from AMA on the basis of a distinct CD spectrum and the
resistance of its UV absorbance spectrum to bathochromic shift upon
alkalinization (Faulstich et al., 1973; Faulstich et al., 1981).
Products of Periodate Oxidation of 6'-0-methyl-a-amanitin under
conditions of Neutral pH
HPLC purification of the periodate oxidation products. Separation
of the products of periodate oxidation of OMA by reverse-phase HPLC
yielded two components, just as was seen previously when the sepa-
ration was performed by conventional chromatography. These two
entities will be referred to as OMAA-IA for the material eluting first
and OMAA-IB for the material which eluted later (Fig. II-1). When the
HPLC fractions were held at room temperature, reanalysis at intervals
of the purified components by HPLC demonstrated a slow intercon-
version of the two forms (Fig. II-2). Within 2 h OMAA-IA elutingg at
IO 20 30 40 50
Fig. II-1. Preparative HPLC purification of various forms of OMAA
on a reverse-phase C18 column. The upper panel shows the purification
of OMAA-II with acetonitrile: 0.05% TFA (22:78) as mobile phase. The
lower panel shows the purification of OMAA-I with 22% acetonitrile as
the mobile phase. Fractions contained 1.0 ml.
0 6 12 18 24 O 6 12 18 24
RETENTION T ME, minutes
Fig. II-2. HPLC analysis of components of OMAA-I. Samples (30pl)
of OMAA-IA (upper tracings) and OMAA-IB (lower tracings) were
rechromatographed on a reverse-phase C18 column with 22% acetonitrile
at 2 ml/min at 2 h (panel A) and 26 h (panel B) after their initial
purification by preparative HPLC.
about 14 min) was contaminated with significant amounts of OMAA-IB
which eluted at about 19 min (upper portion of panel A) and after 26 h
further conversion was apparent (upper portion of panel B). An even
more dramatic conversion of OMAA-IB to OMAA-IA was observed (lower
portion of panels A and B).
Spectral and TLC properties of the periodate oxidation products.
The two OMAA-I components have UV absorbance spectra which are
indistinguishable from each other and very similar to that of OMA.
When held on ice, the interchange between the two products described
above was retarded. Indeed, when HPLC fractions which derived from
separation of the periodate oxidation products of OMA were placed on
ice immediately after they were collected, no contamination of one
form with the other could be detected by HPLC analyses repeated over
48 hours. This aided in the acquisition of CD spectra of the two
products in their pure forms. Thus, HPLC fractions were placed on ice
as they were collected and warmed to room temperature just before the
CD spectrum was obtained. HPLC analysis also showed that fractions did
not undergo significant transformation to the other form during the
time at room temperature which was needed for acquisition of the CD
spectrum. The CD spectra of the periodate oxidation products were
indistinguishable from each other and from that previously published
for OMAA (Fig. II-3).
A PMR spectrum of OMAA-I dissolved in d6-DMS0 showed only a very
small peak in the region expected to contain the signal of an
aldehydic proton (a=9.5-10). In addition, the spectrum lacked the
sharp definition of peaks which has been characteristic of the PMR
spectra of other amanitin derivatives (data not shown).
WAVEL ENGT H ( nm )
Fig. II-3. CD spectra of OMA (dashed line) and OMAA5I (solid
line) in water at concentrations of approximately 4 X 10" M.
Chemical reactivity of the periodate oxidation products.
Reduction of OMAA-IA and OMAA-IB with sodium borohydride produced in
75-90% yield a toxin derivative with the properties expected of OMDA.
The reduction products from these forms of OMAA were indistinguishable
in terms of UV absorbance and CD spectra (which were identical to
those of OMA), TLC and HPLC mobility, violet color reaction with
cinn-HC1, and inhibition of CT RNAP II with K. of 3.4 + 0.2 X 10-9 M.
In contrast to their ready reduction with sodium borohydride, neither
OMAA-IA nor OMAA-IB was susceptible to the action of the oxidant
sodium chlorite at pH 2.0 or 3.0 (Launer and Tomimatsu, 1954; Lindgren
and Nilsson, 1973).
Product of Periodate Oxidation of 6'-0-methyl-a-amanitin at pH 3.0
HPLC purification of the periodate oxidation product. In contrast
to the reaction at neutral pH, periodate oxidation of OMA at mildly
acidic pH's yielded only a single product which was distinct from the
periodate oxidation products described above and which could be puri-
fied by reverse-phase HPLC using an acidic mobile phase (Fig. II-1).
This form of OMAA will be referred to as OMA~A-II.
Spectral and TLC properties of the periodate oxidation product.
The material prepared and purified under acidic conditions had many
properties which were strikingly different from those of OMAA-IA and
OMAA-IB. TLC analysis of HPLC fractions (or of the reaction mixture)
showed predominantly a single component with higher mobility which
stained violet with cinn-HC1; there was trace contamination with the
rust-staining OMAA-I forms. TLC analysis of these same HPLC fractions
in solvent system I after they were neutralized with sodium phosphate
buffer and immediately spotted onto a TLC plate showed primarily the
rust-staining OMAA-I and a trace of the violet-staining OMAA-II. The
UV absorbance and CD spectra of OMAA-II were indistinguishable from
those of OMA. However, within thirty minutes of its neutralization
with sodium phosphate buffer the CD spectrum of OMAA-II changed to
that typical of OMAA-IA and OMAA-IB. A PMR spectrum of OMAA-II in
d6-DMS0 showed the general features expected of a methylamanitin
(Wieland et al., 1983) and also a peak of appropriate size at a=9.55
as expected for an a~dehyde (Fig. II-4).
Chemical reactivity of the periodate oxidation product. When both
sodium chlorite and sodium periodate were added to a solution of OMIA
which was buffered at pH 2.0 or 3.0, a novel product was generated.
This product has been identified as the carboxylic acid (0MA-C00H)
which results from the oxidation of the aldehyde function of OMAA (see
The chemical nature of OMAA is of interest for at least two
reasons. First, since OMAA-I has an altered conformation, detailed
information about its nature can provide some insight into the forces
which maintain amatoxins in their toxic conformation. Second, know-
ledge of its chemistry provides a sound base upon which the planning
of semisynthetic modifications to this part of the toxin can rest. It
is perhaps because the chemical nature of OMAA has been misunderstood
that very little in the way of synthetic modification of the aldehyde
group has been reported since the preparation of OMAA was first de-
scribed by Wieland and Fahrmeir in 1970. The only report of a modifi-
cation of this kind has been the synthesis of a dinitrophenylhydrazone
in modest yield by reaction of a periodate oxidation product of AMA
with dinitrophenylhydrazine (Morris and McSwine, 1983).
Until recently, OMAA has been thought to be a true aldehyde.
Faulstich et al. (1973) proposed that the altered conformation of OMAA
(as evidenced by its unique CO spectrum) was due to an intramolecular
hydrogen bond between the carbonyl function of the aldehyde and the
peptide backbone. Garrity and Brown (1978) reported an infrared spec-
trum of OMAA with bands at 2800 cm-1 and 1350 cm-1 which were thought
to arise from the carbonyl of the aldehyde group. This view has been
challenged by Morris and McSwine (1983) who were unable to detect a
signal characteristic of an aldehydic hydrogen in the PMR spectrum of
a periodate oxidation product of AMA. They have proposed that the
aldehyde of aldoamanitins exists primarily as an intramolecular adduct
with a nucleophile such as an amide nitrogen (e.g., Fig. 11-5). The
data presented in this chapter support this hypothesis, but do not
provide direct proof of it.
An important indication that OMAA-I does not exist primarily as a
free aldehyde is its stability. OMAA-I has been manipulated and stored
for years in the unbient atmosphere in our laboratory without loss of
the material. This is atypical of most true aldehydes, which usually
must be stored under an inert atmosphere to prevent oxidation.
The available data suggest that the two forms of OMAA-I, although
chromatographically distinct, are very similar in other respects. They
both resemble OMAA, as it has been previously described by others, in
terms of CD spectrum, color reaction with cinn-HC1, and reduction to
form 0MDA (Wieland and Fahrmeir, 1970; Faulstich et al., 1973). The
Hz, C CH
- C -N-CH -C- N -CH -
Fig. II-5. Scheme of periodate oxidation of the dihydroxyiso-
leucine sidechain and hypothetical intramolecular reaction of the
aldehyde group. The asterisk marks the new chiral center formed by the
studies with the HPLC-purified forms of OMAA-I have shown that they
slowly interconvert in solution at room temperature. n equilibrium
between the two components is reestablished from one of the purified
forms over the course of several hours. This slow rate of intercon-
version is in sharp contrast with the short period of time (no more
than a few minutes) which is required to establish the equilibrium
following the periodate oxidation of OMA in neutral solution. A simple
explanation for this discrepancy might be that neither form of OMAA-I
is the entity first generated by the periodate oxidation (i.e., the
free aldehyde), but rather they derive from it. Thus, the free alde-
hyde, when formed under conditions of near-neutral pH, rapidly and
almost completely converts to the two forms of OMAA-I. The components
of OMAA-I might be interconverting by a slow back-reaction to the free
aldehyde followed by formation of the other component.
The PMR spectrum of OMAA-I in d6-DMS0 showed only a very small
signal with a chemical shift which would be expected for an aldehyde.
Thus, our PMR spectral data confirm those of Morris and McSwine
(1983). The inertness of 0MAA-I to reaction with sodium chlorite has
provided a further piece of evidence that supports the hypothesis that
it is not primarily a free aldehyde.
The discovery that periodate oxidation of OMA under mildly acidic
conditions generates a distinct entity has added greatly to our under-
standing of the nature of OMAA. This novel product, OMAA-II, appears
to be a free aldehyde by spectroscopic (characteristic signal by PMR)
and chemical (oxidation to carboxylic acid by sodium chlorite) crite-
ria. The demonstration by both TLC and CD spectroscopy that OMAA-II
changes to OMAA-I within minutes of neutralization of the solution
lends credence to the hypothesis outlined above that the periodate
oxidation of OMA yields a free aldehyde (0MAA-II) which converts
rapidly to the two forms of OMAA-I in solutions of near-neutral pH.
The finding that the CD spectrum of OMAA-II is indistinguishable
from that of OMA is very interesting. This suggests that the presence
of the aldehyde group per se does not cause the alteration in the
toxin's conformation. This indicates that the hydrogen bond hypothesis
for the altered conformation (Faulstich et al., 1973) is probably
incorrect. The altered conformation of OMAA-I is associated with sev-
eral characteristics, including a PMR spectrum which is not character-
istic of an aldehyde, inertness to oxidation by sodium chlorite, and
the presence of two similar, yet chromatographically distinguishable,
forms. The probable explanation for this behavior of OMAA-I is that it
exists in solution as a mixture of two diastereomers which are gener-
ated by the reaction of the aldehyde moiety with an intramolecular
nucleophile. The a-amino group of the hydroxytryptophan residue is a
reasonable candidate as this nucleophile on several grounds. As shown
in Fig. II-5, reaction of the aldehyde group with this amide nitrogen
to yield an alkanolamide would form a five-membered ring (thermody-
namically favored?) with a new chiral center arising from the r-carbon
of residue 3 (marked with an asterisk). The generation of this addi-
tional chiral center provides for the possibility of two chromat'ograph-
ically separable diastereomeric forms. There have been reports of the
ready formation of cyclic alkanolamides from the periodate oxidation
of an appropriate dio1 (e.g., van Tamalen et al., 1960). This proposal
for the structure of OMAA-I also provides a ready rationale for both
the inertness of OMAA-I to sodium chlorite and the stability of
DMAA-II at mildly acidic pH's. The interconversion between alkanol-
amides and their constituent aldehyde and amide are known to proceed
very slowly at mildly acidic pH's, but at much higher rates with
increasing pH (Hubert et al., 1975; Jencks, 1969, p. 495). Thus, at
the mildly acidic pH's at which the oxidation with sodium chlorite was
attempted OMAA-I was locked in the alkanolamide form and could not be
oxidized. On the other hand, the reduction of OMAA-I with sodium boro-
hydride was conducted at near-neutral pH where the conversion to free
aldehyde would be permitted. When OMA is oxidized in an acidic medium,
the formation of the alkanolamide is inhibited and OMAA-II, the free
aldehyde, can be isolated or oxidized by sodium chlorite. However,
when a solution of OMAA-II is neutralized the thermodynamically more
stable OMAA-I rapidly forms.
Although the proposed structure of OMAA-I is strictly hypothet-
ical, it is worthwhile to consider how the toxin's conformation might
be altered by this sort of intramolecular reaction. First, crystallo-
graphic (Kostansek et al., 1978; Wieland et al., 1983) and NMR spec-
troscopic (Wieland et al., 1983) studies indicate that the al-amino of
the hydroxytryptophan residue is hydrogen bonded to the terminal
carbonyl of the asparagine residue. Formation of the alkanolamide
would necessarily disrupt this hydrogen bond and perhaps destabilize
the native conformation. Also, formation of the alkanolamide would
generate what has been referred to as a "bridged lactam". Methods for
intentionally introducing bridged lactams into peptides have been
recently developed and applied to the synthesis of peptides with
conformationally constrained backbones (Freidinger et al., 1980;
Freidinger et al., 1982).
As will be seen in the next chapter, the chemical nature of
OMAA-I seems to have a very large impact upon the rate and course of
its reductive amination. Additional observations will be presented and
discussed which indirectly support the proposed structure of OMAA-I.
TRANSFORMATIONS OF 6'-0-METHYLALDO-a-AMANITIN
VIA REDUCTIVE AMINATION AND OXIDATION
Much of our knowledge about the relationship between structure
and activity of amatoxins which differ at residue 3 (residue 3
corresponds to the dihydroxyisoleucine residue of AMA) has derived
from the existence of three naturally-occurring variants in this
sidechain (Fig. I-1, R2). These variants correspond to three levels of
hydroxylation: (1) no hydroxylation, (2) a hydroxyl group on the
r-carbon, or (3) hydroxyl groups on both the y- and a-carbon atoms
(Wieland, 1983). Increasing extent of hydroxylation has been found to
correlate with a modest increase in inhibitory activity toward CT RNAP
II (Cochet-Meilhac and Chambon, 1974). In addition, the semisynthetic
derivative OMDA has been prepared which contains a r-hydroxyvaline
residue. This hydroxyvaline residue is a close homolog of the isoleu-
cine residue of amanullin. Thus, it is not surprising that OMDA and
6'-0-methyl amanullin (0ML) have similar inhibitory activity toward CT
RNAP II (Cochet-Meilhac and Chambon, 1974). A totally synthetic ama-
toxin analog which has norvaline (an amino acid without a B3-branch) at
this position is a very poor inhibitor of RNAP (Wieland et al., 1981).
However, the significance of this latter finding is unclear, since the
effect of this structural change on the overall conformation of the
peptide has not been examined. One can see that these data relate to
variations in the structure of this sidechain which are both very
conservative and which result in small changes in inhibitory activity.
In particular, no charged groups had been introduced into this residue
at the time that this work was begun.
Although the exact nature of the site on the RNAP molecule to
which amatoxins bind has not been elucidated, some information about
the manner in which amatoxins inhibit RNAP has been gathered. Although
their mode of inhibition is apparently noncompetitive (Cochet-Meilhac
and Chambon, 1974), the amatoxins may inhibit RNAP by binding to a
catalytic subsite. Both genetic (Greenleaf, 1983) and biochemical
(Brodner and Wieland, 1976) studies have implicated one of the large
subunits, which presumably contributes part of the active site, as
being involved in amanitin binding. Also, some recent work has led to
the proposal that amatoxins may inhibit RNAP by binding to a portion
of the catalytic site which is involved in translocation of the
nascent RNA molecule (Vaisius and Wieland, 1982).
Since the amatoxins bear no close resemblance to RNAP's sub-
strates or products, it has been difficult to conceptualize their
mechanism of inhibition and to generate a hypothetical framework upon
which studies of structure-activity relationships can be planned.
Thus, although reductive coupling of OMAA to protein amino groups was
envisioned as a possibly useful route for the conjugation of amatoxins
to proteins, there did not exist a large body of knowledge concerning
the likely effect of this kind of modification on the activity of the
toxin nor was there a theoretical basis upon which to predict such
effects. In addition, the results presented in the previous chapter
suggested that the forms of OMAA which could be useful for conjugation
to protein amino groups, e.g., OMAA-IA and OMAA-IB, do not have a free
aldehyde function and also that the form of OMAA which has a free
aldehyde, OMAA-II, is present in solutions of neutral pH in very small
quantities. It was reasonable to anticipate that this diminished
availability of the aldehyde group would have a negative impact upon
the rate of reductive coupling to amines. However, it was difficult to
predict a prior the magnitude of this effect.
In order to assess the importance of these factors, the reductive
coupling of OMAA-I to some simple amines was examined (Fig. III-1). In
addition, two oxidative transformations of the aldehyde group (to
carboxyl and nitrile functions) have been accomplished. These two
additional alterations of this sidechain provide conservative changes
in the structure and additional important information about the
structural requirements for potent inhibition of RNAP II.
Material and Methods
Reagents. Ammonium acetate, glycine, and L-proline were obtained
from Sigma Chemical. Sodium cyanoborohydride (Sigma Chemical) was
recrystallized (Jentoft and Dearborn, 1979) and kept desiccated over
P205. Hydroxyl amine-0-sulfonic acid was purchased from Alfa Chemical.
Triethylamine (TEA) was obtained from Pierce Chemical. The source of
other reagents has been described in the previous chapter.
Toxin. The syntheses of OMA and OMAA have been described in the
previous chapter. These two compounds served as the starting material
for all of the semisynthetic derivatives described here.
- N-CH- COOH
Fig. III-1. Scheme showing the reductive amination of OMAA.
Spectroscopy and chromatography. Absorbance and CD spectroscopy
and chromatographic procedures were performed as described in chapter
II. TLC sovent system II contained 2-butanol: methanol: 0.5 M sodium
chloride (4:2:1). Fast-atom bombardment mass spectroscopy (FAB-MS) was
performed by Triangle Laboratories (Durham, NC) using a VG 7070E mass
spectrometer. In some cases HPLC solvents contained a buffer instead
of water; this buffer contained 20 mM TFA which was adjusted to pH 3.0
with TEA and will be referred to as TFA-TEA buffer (Cohen et al.,
RNA polymerase assay. The analysis of inhibition of CT RNAP II
was conducted as described in the previous chapter.
Synthesis of Amatoxin Derivatives
Desalting of toxin solutions. Certain toxin solutions were
desalted (Bibhlen et al., 1980) with the aid of Sep Pak C18 cartridges
(Waters Associates). The C18 cartridges were prefinsed with 20 ml of
50% acetonitrile followed by 10 ml of water. Aqueous solutions of
toxin were loaded onto the cartridge at a rate no greater than
1 ml/min. After the cartridge was washed with 15 ml of water, the
toxin was finally eluted with 10 ml of 50% acetonitrile. When OMA-NH2
was desalted, this protocol was modified in that 20 mM ammonium
bicarbonate was substituted for water at each step.
Synthesis of OMA-NH To 10 ml of molar methanolic ammonium
acetate was added 6.3 mg (0.1 mmol) of sodium cyanoborohydride. A
portion of this solution (1.0 ml) was added to 2.0 plmol of lyophilized
OMAA. The reaction was allowed to proceed in the dark at 230C for 18
days. The solution was concentrated to dryness at 300C in vacuo,
redissolved in acetonitrile: TFA-TEA buffer (22:78), and chromato-
graphed by HPLC using the same solvent at 0.5 ml/min. The desired
product eluted as the major product between 21 and 26 ml. This mate-
rial was reconcentrated to dryness at 30*C in vacuo, dissolved in
acetonitrile: TFA-TEA buffer (15:85), and rechromatographed by HPLC in
this solvent at 0.5 ml/min. The toxin derivative eluted from the
column between 45 and 57 ml. This was concentrated at 300C in vacuo to
remove the acetonitrile and then desalted using the appropriate Sep
Pak C18 protocol (see above). The eluate from the Sep Pak C18 car-
tridge was concentrated at 300C in vacuo to remove the acetonitrile,
supplemented with 10 ml of water, and lyophilized. The final yield was
0.55 pmol (28%). TLC: RF, I-0.60, II-0.83; color-reaction, violet.
FAB-MS: [M+H] expected, 902.37; found, 902.33; [M-H] expected,
900.37; found, 900.14.
Synthesis of OMA-gly. Glycine (0.751 g, 10 mmol) and sodium
cyanoborohydride (31.5 mg, 0.5 mmol) were dissolved in sufficient 10%
methanol to make 5.0 ml of solution. Lyophilized OMAA (2.0 pmol) was
dissolved in 0.10 ml of this solution. After 96 h in the dark at 230C
the reaction mixture was taken up in 22% acetonitrile and chromato-
graphed by HPLC using the same solvent at 0.5 ml/min. The desired
product eluted between 16 and 25 ml yielding 1.68 pmol (84%). TLC: RF'
1-0.46, II-0.46; color-reaction, violet.
Synthesis of OMA-pro. A methanolic solution (0.10 ml) containing
0.5 M L-proline and 0.05 M sodium cyanoborohydride was added to 2.0
pmol of OMAA with dissolution of the toxin. After 170 h in the dark at
23*C the reaction mixture was taken up in 22% acetonitrile and applied
to HPLC using the same solvent at 0.5 ml/min. The desired product,
which eluted between 26 and 29 ml, was concentrated to dryness in
vacuo at 300C, redissolved in 15% acetonitrile, and rechromatographed
by HPLC using this solvent at 0.5 ml/min. The product (1.45 plmol, 73%)
eluted between 69 and 77 ml. TLC: RF, I-0.37, II-0.36; color-reaction,
Synthesis of OMA-C00H. A solution (0.4 ml) containing 0.015 M
sodium chlorite, 0.030 M amidosulfonic acid, and 0.10 M sodium phos-
phate, pH 2.0, was added to 2.0 pmol of OMA. To this was immediately
added 10 pl of a solution containing 0.24 M sodium periodate and 0.1 M
sodium phosphate, pH 2.0. After 10 min in the dark at 230C the toxin
solution was desalted using the Sep Pak C18 technique, concentrated to
dryness in vacuo at 40*C, redissolved in acetonitrile: TFA-TEA buffer
(18:82), and chromatographed by HPLC using the same solvent at 1.0
ml/min. The desired product eluted between 47 and 51 ml. After concen-
tration in vacuo at 3000 to remove acetonitrile the toxin solution was
desalted again to yield 0.85 pmol (47%). TLC: RF, I-0.49, II-:0.29;
color-reaction, violet. FAB-MS: [M+H] expected, 917.34; found,
917.27; [M-H] expected, 915.33; found, 915.49.
Synthesis of OMA-CN. Hydroxylamine-0-sulfonic acid (22.6 mg, 0.2
mmiol) was dissolved in sufficient 1.0 M sodium borate, pH 9.0, to make
1.0 ml of solution. A portion of this (0.20 ml) was used to dissolve
2.0 pmol of lyophilized OMAA. After 72 h in the dark at 230C the
reaction mixture was taken up in 22% acetonitrile and chromatographed
by HPLC using the same solvent at 1.0 ml/min. The desired product
(1.16 Ipmol, 58%) eluted between 36 and 45 ml. TLC: RF, I-0.63,
II-0.85; color-reaction, violet.
Results and Discussion
Reductive Amination of OMAA-I
Reaction of OMAA-I with ammonium acetate in the presence of
sodium cyanoborohydride was the first reductive amination reaction
attempted. This reaction would be expected to yield the primary amine
OMA-NH2 (Fig. III-2, structure VI with R=R'=H), the simplest amine
accessible via this route. Several unexpected difficulties were
encountered in the synthesis of OMA-NH2. Even when ammonium acetate
was included at a stoichiometry of 100 to 1 relative to OMAA, the
major products appeared to be dimeric and/or trimeric products (cf.
Borch et al., 1971). That is, they were bound by a cation-exchange
resin (Sephadex SP-50), yet were unreactive with fluorescamine. Their
appearance was suppressed by increasing the molar ratio of ammonium
acetate relative to OMAA from 100 to 500. However, instead of a single
fluorescamine-reactive entity being produced under these new condi-
tions, there were several. The concern was that OMAA and/or the
desired product were undergoing a secondary degradative reaction.
Since it has been well documented that many peptides which contain
residues of 2,4-diaminobutanoic acid are unstable at neutral pH
(Davies et al., 1969; Katrukha et al., 1968; Poduska et al., 1965), it
seemed plausible that OMA-NH2, which contains the related 2,4-diamino-
3-methylbutanoic acid residue, might display comparable liability.
Another area of concern was the tryptathionine residue which is
believed to be susceptible to base-catalyzed destruction (Wieland,
1983). However, when some of these byproducts were isolated by chroma-
tography on Sephadex LH-20, many of them were found to be unstable and
IV + CN
HC C -
c C'CN CH -
- C',N- CH -
-H 2 0
H,C CH H
HC C C H
Fig. III-2. Scheme showing the probable reactions and side-
reactions occurring in the reductive amination of OMAA.
to decompose back to the starting material, OMAA (data not shown).
These quasi-stable amines may be a-aminonitriles (Fig. III-2,
structure VII) (Bejaud et al., 1976) which have already been impli-
cated as undesirable, quasi-stable byproducts in the reductive methyl-
ation of proteins (Gidley and Sanders, 1982). One or more of these
amines might also arise from reaction between the intermediate immon-
ium (Fig. III-2, structure V) and an intramolecular nucleophile (such
as an amide nitrogen) to yield an N-Mannich compound (Tramontini,
1973). This intramolecular reaction to form an N-Mannich (Fig. 1II-2,
structure VIII) is comparable to .that proposed in chapter II for the
formation of an intramolecular alkanolamide (Fig. III-2, structure I)
in OMAA. Studies on the stability of N-Mannich compounds of this kind
have been recently reported (Bundgaard, 1985; Loudon et al., 1981).
Their stability increased with decreasing bulk of the substituents on
the amino group (i.e., R and R' in Fig. III-2, structure VIII). The
N-Mannich which might form during the synthesis of OMA-NH2 would bear
the smallest substituents possible (hydrogen) and thus impede the
progress of the reaction on account of its high stability. When the
reaction was permitted to progress for several weeks, most of these
fluorescamine-reactive compounds slowly disappeared and one increased
in amount. As noted above, chromatographic separation on Sephadex
LH-20 of the components of a reaction which still contained multiple
fluorescamine-reactive amatoxin derivatives showed that many of these
entities were unstable and decomposed back to the starting material,
OMAA. This supports the notion that labile compounds like structures
VII and VIII (Fig. III-2) are formed. Probably as a consequence of
these or similar side-reactions, the purified yield of OMA-NH2 has
never exceeded 30% despite numerous efforts to optimize the reaction.
Reductive coupling of ORAA to the amino acids glycine and
L-proline proceeded at a much higher rate than the reaction with
ammonium acetate. This higher rate may derive, in part, from the
larger substituents on amino group and a consequent lowering of the
stability of the N-Mannich byproduct. When these reactions were
fractionated by HPLC before the reaction reached completion, novel
compounds which decomposed to OMAA were again noted and in this case
isolated from void-volume fractions. 0MA-gly and OMA-pro can now be
obtained in approximately 75% yield.
Oxidative Transformations of OMAA
Reaction of OMAA-I with hydroxylamine-0-sulfonic acid produced
OMA-CN in 60% yield after a period of a few days. This reaction relies
upon the formation of an 0-sulfonated oxime by reaction of the reagent
with the aldehyde. This then udergoes an elimination reaction to yield
the nitrile and a sulfate ion (Fizet and Streith, 1974).
0MAA-II was generated by the reaction of OMA with sodium perio-
date at pH 2.0 (see previous chapter). By inclusion of sodium chlorite
(Launer and Tomimatsu, 1956; Lindgren and Nilsson, 1973) in this
system, OMAA-II was oxidized quickly to the carboxyl derivative,
Properties of the Derivatives
These derivatives had chromatographic properties expected for
their structures. The charged derivatives generally migrated more
slowly on TLC and had shorter retention times on HPLC than OMAA, while
the relatively nonpolar OMA-CN showed the opposite behavior. In con-
trast to OMAA-I, the CO spectra of all of the new derivatives were
indistinguishable from that of OMA (Fig. II-3). Also, their color-
reaction with cinn-HC1 was violet compared with the rust color ob-
tained with OMAA-I. As expected, only OMAA-NH2 reacted with fluores-
camine to yield a fluorescent product. FAB-MS of OMA-C00H and OMA-NH2
revealed mass ions of the appropriate size for the proposed structures
(Figs. III-3 and III-4).
Instability of the reductive amination products has been detected
after prolonged storage of frozen solutions. This destruction was
associated with loss of the distinctive UV absorbances, e~g., at 304
nm (data not shown). Loss of the compounds was retarded when solutions
were mildly acidic.
None of the new derivatives inhibited CT RNAP II as strongly as
AMA. OMA-CN was the strongest inhibitor with a Ki of 3 X 10-9 M.
OMA-gly, OMA-pro, OMA-C00H, and OMA-NH2 were much weaker inhibitors
with K 's of 2.5 X 10-7 M, 7 X 10-6 M, 1 X 10-7 M, and 1.7 X 10-7 M,
respectively. The inhibition by OMA-gly, OMA-pro, and OMA-NH2 was
studied more thoroughly and showed apparently noncompetitive behavior
(e.g., Fig. III-5). Since the amanitin amines were considerably less
potent inhibitors than other amanitins which might be generated in
small amounts as byproducts of the reaction (e.g., OMDA), steps were
taken to help ensure that a K obtained for the material in a peak was
characteristic of the bulk of the material in the peak and not grossly
biased by the presence of traces of more potent inhibitors. In order
to accomplish this, the Ki was determined for the contents of at least
two, and usually three, fractions across a peak. The material in the
SI l l 1 j i l l
O O O O O
_ mo co N
O OO O O O
O m e q ( ( c
L O w
-0.!8 -0.1 2 -0.0 6 O 0.0 6 0. 1 2 0.1 8
Fig. III-5. Dixon plot showing the apparently noncompetitive
inhibition of CT RNAP II by OMA-NH .The concentrations of UTP used
were 0.75 pM (circles), 1.10 pM (si uares), and 1.50 pM (triangles).
peak was subjected to additional chromatographic steps until it
appeared homogeneous with respect to inhibition of CT RNAP II. The Ki
of the pooled fractions from each final peak was consistent with that
of the constituent fractions (data not shown).
The finding that the inhibition of CT RNAP II by each of the
amanitin amines showed noncompetitive behavior with respect to UTP
provides preliminary evidence that these amanitin derivatives inhibit
this enzyme in a manner analogous to other amanitins, presumably
through localization at an amanitin binding site on the enzyme. This
evidence for inhibition by a mechanism common to other amanitins,
along with the resumption of a "native" conformation as suggested by
the CO spectra, provides a basis for interpreting their diminished
affinities in tems of interactions between the enzyme and the newly
introduced substituents. In this regard, OMA-NH2 is most readily
compared to the previously examined amanitins, OMDA and OML. These
three derivatives differ only with respect to a single group. They
each have a different group of roughly the same size and shape append-
ed to the r-carbon atom of residue 3 (see Figs. I-1 and III-1). While
stanitins OMDA and OML, which have a hydroxyl and a methyl group,
respectively, at this location, bind to CT RNAP II almost as tightly
as the most inhibitory of the naturally-occurring toxins, OMA-NH2
which bears an amino group is a much poorer inhibitor. At first
glance, the primary difference between OMA-NH2 and the other two
amanitins appears to be the positive charge expected for an aliphatic
amino group at near-neutral pH. However, because of its charge the
amino group may also be effectively somewhat larger than the analogous
hydroxyl and methyl groups due to a larger layer of bound water. Thus,
it is difficult to attribute the difference in affinity unequivocally
to the presence of a positive charge, but this seems likely to be the
most important factor. The Ki's of OMA-gly and OMA-pro are much more
difficult to rationalize because of the multiplicity of functional
groups which have been introduced and because of the high likelihood
of chelation of the Mn2+ in the RNAP assay system by the amino acid
Since OMA-NH2, which bears a positive charge, was a relatively
poor inhibitor of CT RNAP II, it was anticipated that OMA-C00H, which
should bear a negative charge, would be a quite potent inhibitor. It
was surprising to find that it inhibits the enzyme almost as poorly as
DMA-NH2. In contrast, OMA-CN, a semisynthetic derivative which is like
OMDA in that it lacks a group with a charge, is a rather potent inhi-
bitor of CT RNAP II. These data taken together suggest that the por-
tion of the amanitin binding site of CT RNAP II which interacts with
the sidechain of residue 3_ may contain closely spaced positively and
negatively charged moieties. Thus, derivatives with charged sidechains
are strongly repulsed, while derivatives with uncharged sidechains can
bind more strongly. Alternatively, this portion of the binding site
may be somewhat nonpolar so that the addition of highly polar
substituents to this sidechain inhibits binding.
It seems clear that OMAA is not well-suited for reductive
coupling to proteins. Even under forcing conditions with very high
concentrations of amines, the rate of reaction is very slow. However,
some the derivatives described here and future subderivatives may
prove useful in a number of different applications.
STUDIES ON THE CONJUGATION OF N-ACYLATED AMINO SUGARS
TO BOVINE SERUM ALBUMIN BY MEANS OF REDUCTIVE ALKYLATION
The results presented in the previous chapter indicated that
reductive coupling of OMAA to protein amino groups does not represent
a feasible route for the conjugation of amanitin to proteins. Conse-
quently, attention was turned to the preparation of an azo amanitin
derivative which could be employed in this way.
Synthesis of azo ananitins requires diazotization of an aromatic
amine. However, since aldehydes are reactive to diazotizing reagents,
the aldehyde group needed to be introduced in protected or precursor
form. This need for protection is complicated by the fact that amatox-
ins are unstable toward most of the conditions which are used to
deprotect aldehydes or to generate them from a precursor (Wieland,
Many aldoses contain aldehyde groups which are inherently protec-
ted by intramolecular reaction with a hydroxyl group to form a hemi-
acetal. Despite the fact that only a very small proportion of an aldo-
hexose (e.g., glucose) is present in solution in the free aldehyde
form, these sugars have been successfully coupled to proteins by
reductive alkylation with sodium cyanoborohydride (Gray, 1974). Under
the conditions which were initially described, this reaction was quite
slow. However, Roy et al. (1984b) have recently noted that borate
increases the amount of free aldehyde which is in equilibrium with the
cyclic forms of aldoses. This effect was most prominent for glucose
and lactose. They were further able to show that borate greatly in-
creased the rate of reductive coupling of lactose to BSA (Roy et al.,
These observations by Roy and his coworkers suggested a possible
route for preparing an azo amanitin with the desired properties. A
derivative of the already well-studied ABGG was envisioned which would
bear an amino sugar linked to the carboxyl group of the glycylglycine
linker (see Fig. V-I). This chapter describes experiments which were
undertaken to define the feasibility and optimal reaction conditions
for the reductive coupling of N-acylated amino sugars to protein.
Model compounds which contai n the NJ-4-ni troben zoyl glycyl glyci ne moi ety
linked to an amino sugar (0-galactosamine or D-glucosamine) were
prepared and utilized for these studies. The conjugation of these
compounds to protein amino groups is schematically depicted in Fig.
IV-1. The effects of temperature, pH, buffer, and various reactant
concentrations on the reaction rate have been studied.
Materials and Methods
N-4-nitrobenzoylglycylglycine (PNBGG) was obtained from United
States Biochemicals. Nickel (II) chloride, BSA, sodium cyanoboro-
hydride, 2-amino-2-deoxy-D-glucose (glucosamine) hydrochloride,
2-amino-2-deoxy-D-galactose (galactosamine) hydrochloride, 1-ethy1-
3-(3-dimethylamino)propylcarbodiimide (EDC), N-2-hydroxyethyl-
pi perazine-N -2-ethanesulfoni c acid (HEPES), and sodium 3-trimethyl-
PROT- N ,H
Fig. IV-1. Scheme illustrating the reductive coupling of
PNBGG-sugar derivatives (or ABGG-GLU) to proteins; "R" represents the
PNBGG (or ABGG) moiety.
sily1-d4-propanoate (TSP) were obtained from Sigma Chemical. Sodium
cyanoborohydride was recrystallized according to Jentoft and Dearborn
(1979). Trichloroacetic acid (TCA) was obtained from Fisher Scien-
tific. Other reagents were obtained as described in the previous
Chromatography. TLC and HPLC were performed as described in the
previous chapters. TLC solvent system III contained 1-butanol: metha-
nol: water (4:2:1); solvent system IV contained 2-butanol: ethyl ace-
tate: water (14:12:5). Column chromatography was also performed with
Sephadex LH-20 and Sephadex SP-50 (Pharmacia Fine Chemicals).
Bio-Beads SM-4 (Bio-Rad) were soxhlet-extracted with methanol for
seven days prior to use.
Spectroscopy. Spectroscopic studies were performed as outlined in
the previous chapters. In addition, optical rotation was determined
with a Jasco DIP-360 digital polarimeter equipped with a 10 cm cell.
Elemental analysis. Elemental analysis was performed by the
instrument facility in the Department of Chemistry, University of
Synthesis of PNBGG Derivatives
Synthesis of PNBGG-GLU. To PNBGG (0.56g, 2.0 mmol) in 4.0 ml of
water was added with stirring 0.18 ml of 10 M Na0H, EDC (0.575 g,
3.0 mmol), and glucosamine HC1 (0.86 g, 4.0 mmol). After 12 h of
reaction at 230C in the dark the resultant gel was dissolved in 100 ml
of water. The solution was added to 125 ml of Bio-Beads SM-4, stirred
for 1 h, and then filtered. The resin was washed for another hour with
100 ml of water and filtered again. Bound material was eluted from the
resin by extracting six times for 10 min with 100 ml of 75% methanol.
The combined methanolic extracts were concentrated in vacuo at 40*C to
a small volume. After filtration through Whatman # 40 paper the
solution was applied to a 2.5 X 15 cm column of Sephadex SP-50 (Na )
and eluted with 100 ml of water. The eluant was concentrated in vacuo
at 50*C to a small volume and applied to a 2.5 X 95 cm column of
Sephadex LH-20 which was eluted with water. Fractions of 7.5 ml were
collected at approximately 0.2 ml/min. The desired product (1.35 mmol,
67%) was located in fractions 52 to 60. TLC: RF, III-0.68, IV-0.34.
For C17H22N4010 calculated: C 46.16, H 5.01, N 12.66; found: C 45.22,
H 5.13, N 12.32. [a]25=+210. e268=1.2 X 104 M1 cm-1 PMR (Fig. IV-2):
3.46-3.99 (m, 6H, sugar); 4.04 (s, 2H, glycine); 4.21 (s, 2H,
glycine); 4.77 (d, 0.4H, J=8.2 Hz, anomeric); 5.20 (d, 0.6H, J=3.3 Hz,
anomeric); 8.04 (d, 2H, J=8.6 Hz, aromatic); 8.37 (d, 2H, J=8.6 Hz,
Synthesis of PNBGG-GAL. The synthesis of PNBGG-GAL was similar to
that of PNBGG-GLU. PNBGG-GAL eluted from the Sephadex LH-20 column in
fractions 59 to 65 (57% yield). TLC: RF, III-0.68, IV-0.34. For
C17H22N4010 calculated: C 46.16, H 5.01, N, 12.66; found: C 45.10,
H 5.11, N 12.23. [n]25=+32o. e268=1.2 X 104 M1 cm-1 PMR (Fig. IV-3):
3.6-4.3 (m, 10H, sugar and glycines); 4.70 (d, 0.5H, J=8.4 Hz,
anomeric); 5.24 (d, 0.5H, J=3.6 Hz, anomeric); 8.03 (d, 2H, J=8.6 Hz,
aromatic); 8.36 (d, 2H, J=8.9 Hz, aromatic).
Conjugation of PNBGG-GLU and PNBGG-GAL to BSA. BSA stocks were
dialyzed against water or the appropriate buffer before use. Molar
concentrations of BSA were estimated on the basis of E1% of 6.67 at
279 nm (Janatova et al., 1968) and a molecular weight of 66,300 (Reed
et al., 1980). Reaction components were sterilized by filtration
through 0.22 pm Millex-GV filters (Millipore) and added aseptically to
autoclaved vials. All reactions contained 1.0 mg/m1 (1.51 X 10-5 M)
BSA and 3.62 X 10-3 M of a carbonyl compound (PNBGG-GLU or PNBGG-GAL).
Other components were incorporated as indicated below. Reactions were
allowed to proceed in the dark at either 200C or 37*C.
Purification and analysis of conjugates. The BSA conjugates were
freed of unconjugated PNBGG derivatives by one of two methods. In
experiments where the buffer concentration was 0.2 M, samples of 200
pl were removed aseptically and placed on ice. The chilled solution
was adjusted to 10% TCA with 40 pl of ice-cold 60% TCA. After 15 min
the precipitate was pelleted by centrifugation for 5 min in a table-
top centrifuge (Fisher Scientific). The supernatant fluid was decanted
and the pellet dissolved in 0.1 M sodium phosphate, pH 7.0. This
precipitation with TCA was repeated and the final pellet was dissolved
in 0.5 ml of the sodium phosphate buffer.
In experiments where buffer concentrations exceeded 0.2 M the
conjugate was purified with Centricon CM-30 ultrafiltration devices
(Amicon). The sample (0.2 ml) of the reaction mixture was added to the
device and diluted to 2.0 ml with 0.2 M sodium borate, pH 8.0, with
mixing. This was then centrifuged at 5,500 rpm for 2 h in either a
Beckman JA-40 or Sorvall type 30 rotor at SoC in order to concentrate
the conjugate solution to approximately 50 pl. The conjugates were
diluted and concentrated three more times in a similar fashion. The
second wash was with 0.1 M disodium ethylenediaminetetraacetic acid
(EDTA), while the third and fourth washes were with 0.1 M sodium
phosphate, pH 7.0. The final concentrate was diluted with 0.5 ml of
0.1 M sodium phosphate, pH 7.0
The extent of coupling was estimated by comparing the absorbances
at 279 nm and 300 nm. Molar extinction coefficients for BSA at 279 nm
and 300 nm were determined to be approximately 4.54 X 104 -1 cm- and
3 -1 -1
3.0 X 10 M cm respectively. The comparable values for the PNBGG
4 -1 -1 3 -1
derivatives were found to be 1.0 X 10 M cm and 3.95 X 10 M
cm .These extinction coefficients reflect a significant difference
in the absorbance spectra of BSA and the PNBGG derivatives. The
absorbance of the PNBGG derivatives falls slowly in the near UV (Fig.
IV-4), while that of BSA falls rapidly. It is this difference which
has been exploited to provide a method for estimating the extent of
conjugation. Simultaneous equations which incorporated these extinc-
tion coefficients were solved to yield an algorithm (cf. Mishell and
Shiigi, 1980, pp. 345-7) by which the absorbance at 300 nm could be
partitioned into the contributions from BSA and the conjugated PNBGG
derivative: A300 from PNBGG = 1.21(A300) 0.0832(A279). Based upon
the contribution to A300 from each component and their respective
extinction coefficients, the molar ratio of PNBGG derivative to BSA in
the conjugate was easily calculated. Absorbance measurements at 279 nm
and 300 nm were corrected for light scattering by extrapolation from
390 nm, a wavelength at which neither BSA nor the PNBGG derivatives
absorb significantly. Light scattering was assumed to obey Rayleigh's
WAVELENGTH (nm )
Fig. IV-4. UV absogbance spectrum of PNBGG-GLU in water at
approximately 2.5 X 10' M.
theorem ideally and thus to vary with the fourth power of wavelength
(cf. Leach and Scheraga, 1960).
Initial Studies ~on the Effects of pH, Temperature, Nature of Sugar
Residue, Borate, and Sodium Cyanoborohydride Cncefr+rntrationo the
Conjugation of PNBGG-Sugars to BSA
The results of initial studies to examine the influence of sever-
al reaction parameters that were thought likely to be important are
summarized in Fig. IV-5. Conjugation of PNBGG-GLU and PNBGG-GAL to BSA
-was found to be strongly dependent upon sodium cyanoborohydride. In
the absence of sodium cyanoborohydride there was a slow time-dependent
increase in the extent of conjugation, but the degree of conjugation
after 225 hours was less than three PNBGG-sugar molecules per BSA
molecule for all of the conditions examined. There was no significant
change in the A279 or A300 (i.e., the spectral parameters used to
follow the course of the reaction) of the BSA in sham reactions which
lacked PNBGG-GLU or PNBGG-GAL or in reactions where lactose was conju-
gated to the BSA (data not shown). The most important factors affect-
ing the rate of conjugation were the use of borate instead of other
buffer salts (phosphate or HEPES) and the temperature (reactions pro-
ceeded much more rapidly at 370C than at 200C). In accord with the
results of Roy et al. (1984b), the effect of borate on the reaction
rate was more marked for the glucosamine than for the galactosamine
derivative. In fact, in the presence of borate PNBGG-GLU coupled at a
significantly higher rate than PNBGG-GAL, while there was little
difference in their rates of reaction in the absence of borate. The
influence of pH (pH 8.0 versus pH 9.0) was found to be less signifi-
Fig. IV-5. Effect of pH, temperature, and nature of sugar residue
on the extent of conjugation of PNBGG-GAL (upper panel) and PNBGG-GLU
(lower panel) to BSA over time. All reaction mixtures contained 1.0
mg/m1 BSA, 3.62 mM PNBGG derivative, 7.24 mM sodium cyanoborohydride,
10 mM nickel (II) chloride, and 6 mM sodium citrate in a 0.2 M buffer.
Closed symbols represent borate buffer-containing reactions, while the
open upright triangle contained phosphate buffer. Upright triangle,
pH 9.0 at 370C; circle, pH 8.0 at 37*C; inverted triangle, pH 9.0 at
200C; square, pH 8.0 at 20*C.
cant, but reactions did proceed more rapidly at pH 9.0. The inclusion
of nickel (II) chloride as a cyanide-absorbing reaction component
(Jentoft and Dearborn, 1980) had no demonstrable negative effect upon
either the reaction rate or the properties of the conjugate (data not
Further Studies on the Effect of Sodium Cyanoborohydride
Concentration, Borate Concentration, and pH on the Conjugation of
The initial studies described above showed that the glucosamine
residue permitted much higher reaction rates than the galactosamine
residue. Thus, further work was restricted to PNBGG-GLU. The depen-
dence of rate on the sodium cyanoborohydride concentration is present-
ed in Fig. IV-6. Under the conditions utilized the rate increased ra-
pidly with increasing sodium cyanoborohydride concentration and level-
ed off at approximately 0.10 M. Furthermore, under conditions of opti-
mized sodium cyanoborohydride concentration there was no significant
effect of varying borate concentration between 0.2 M and 2.0 M on the
rate of coupling (Fig. IV-7). Variation of the pH between 6.0 and 9.0
had a profound effect on the rate of conjugation (Fig. IV-8). While
there was a large increase in rate between pH 7.0 and 9.0, at pH's
less than 7.0 the rate was very low, almost negligible.
The coupling of sugars to proteins by reductive alkylation with
sodium cyanoborohydride was first demonstrated by Gray (1974). Even
with concentrations of the reactants at near saturation the rate of
OQ 10 20
Fig. IV-6. Effect of sodium cyanoborohydride concentration on the
extent of conjugation of PNBGG-GLU to BSA after various times. All
reactions were conducted with 1 mg/m1 BSA, 10 mM nickel (II) chloride,
and 6 mM sodium citrate in 0.2 M sodium borate, pH 8.0, at 370C. The
triangles, squares, and circles represent samples at 25, 50, and 75
hours of reaction, respectively y.
O 50 10 0
Fig. IV-7. Effect of borate concentration on the extent of conju-
gation of PNBGG-GLU to BSA at various times. All reactions were con-
ducted with 1 mg/ml BSA, 10 mM nickel (II) chloride, 6 mM sodium
citrate, and 72.5 mM sodium cyanoborohydride in borate buffers of
varying concentration with pH 8.0 and temperature 370C. Inverted
triangle, 0.2 M; square, 0.5 M; upright triangle, 1.0 M; and circle,
2.0 M sodium borate.
.J 32 -
6.0 7.0 8.0 9.0
Fig. IV-8. Effect of pH on the extent of conjugation of PNBGG-GLU
to BSA at various times. All reaction mixtures contained 1 mg/m1 BSA,
0.1 M sodium cyanoborohydride, 10 mM nickel (II) chloride, and 6 mM
sodium citrate in a buffer consisting of 0.2 M sodium borate and 0.05
M PIPES. All reactions were kept at 37*C at the appropriate pH. The
triangles, squares, and circles represent samples analyzed after 25,
50, and 75 hours of reaction time, respectively.
conjugation was very slow. Despite this sluggishness, the linkage of
sugars to proteins by this method has served as an important route for
the synthesis of neoglycoproteins. The demonstration by Roy and his
coworkers (1984a, 1984b) that borate increases the proportion of
acyclic sugar in solution and thus greatly accelerates the rate of
reductive coupling of lactose to BSA has served to enhance the attrac-
tiveness of this approach for conjugation of sugars to proteins. If
this phenomenon of borate-enhanced reductive coupling might also apply
to N-acylated 2-amino-2-deoxy-0-aldobexoses, then its utility could
conceivably be extended to the conjugation of a wide variety of com-
pounds to proteins.
The PNBGG adducts of glucosamine and galactosamine were prepared
as synthetic intermediates of the corresponding ABGG derivatives.
These proved to have characteristics which made them useful for pre-
liminary studies on the coupling of this class of compounds to pro-
tein. In particular, they were quite soluble in water and their UV
absorbance spectra were sufficiently different from that of BSA to
permit estimation of the extent of conjugation based upon spectral
The reductive coupling of both PNBGG-GLU and PNBGG-GAL was
markedly stimulated by 0.2 M borate. As might be expected from the
studies of Roy et al. (1984b), PNBGG-GLU reacted substantially more
rapidly than PNBGG-GAL in the presence of borate, while their rate of
reaction was similar in solutions of other buffers. -Increasing the
temperature from 200C to 37*C was also found to greatly enhance the
rate of coupling, while increasing the pH from 8.0 to 9.0 had a much
smaller effect. The reductive coupling of the PNBGG derivatives to BSA
was strongly cyanoborohydride-dependent, but very slow, time-dependent
conjugation of the PNBGG derivatives to BSA was detected in the
absence of sodium cyanoborohydride. This reaction may correspond to
that previously seen in solutions of glucose with BSA (Baynes et al.,
Optimization of the conjugation reaction was examined further
with the more active PNBGG-GLU. The reaction rate was seen to increase
markedly as the sodium cyanoborohydride concentration was increased to
approximately 0.1 M, beyond which the rate plateaued. This finding is
in sharp contrast to reductive methylation with formaldehyde and
sodium cyanoborohydride where the rate varies little with the cyano-
borohydride concentration (Jentoft and Dearborn, 1979). However, there
have been indications of a similar strong dependence on cyanoborohy-
dride concentration for the rate of reductive coupling of raffinalde-
hyde to proteins (van Zile et al., 1979). Under conditions of opti-
mized sodium cyanoborohydride concentration at pH 8.0, increasing the
borate concentration from 0.2 to 2.0 M had no significant effect upon
A more extensive examination of the effect of pH was undertaken.
This showed a large increase in the rate of coupling between pH 7.0
and 8.0. Previous studies on the reductive coupling of sugars to
proteins in buffers other than borate have noted a significant in-
crease in the rate with increased pH (Marsh et al., 1977; Schwartz and
Gray, 1977). This presumably resulted from the fact that sugars have a
higher proportion of their acyclic form in solution at higher pH's
(Roy et al., 1984b). The pH-dependence of conjugation noted here may
derive in part from this effect, but it is likely to be much more
strongly related to the pH-dependence of the interaction between
borate and the sugar.
This study, in contrast with others, has focused upon the reduc-
tive coupling of complex compounds to proteins using relatively low
reactant concentrations. Most previous workers have utilized concen-
trations which could only be achieved with sugars and highly soluble
proteins. The current studies have defined a system in which compounds
can be coupled to proteins at a practical rate using reactant concen-
trations which may be readily achieved with drugs (or their deriva-
tives) and immunoglobulins. The optimized rate of conjugation attained
here with PNBGG-GLU and BSA appears to approach the rate obtained by
Lee and Lee (1980) for reductive coupling of some free aldehydes to
From the work of Hall (1956) on the pKa of various amines it can
be anticipated that the pKa of protein amino groups which have been
linked to one of the PNBGG derivatives will drop to about 9.0. This
should provide considerable preservation of charge at physiologic pH.
In the next chapter the application of this method to the conju-
gation of amanitin to BSA is described.
CONJUGATION OF A NOVEL AZO AMANITIN TO BOVINE SERUM ALBUMIN
VIA REDUCTIVE ALKYLATION WITH SODIUM CYAN080ROHYORIDE
The results obtained with the model compounds in the preceding
chapter suggested that preparation of a derivative of AMA containing
D-glucosamine would provide a reasonable route to an amatoxin which
could be conjugated to proteins by means of reductive alkylation.
Preparation of the anino analog of PNBGG-GLU (PABGG-GLU) has permitted
access to a simple synthetic route to an azo amanitin derivative,
ABGG-GLU, which bears the D-glucosamine residue. This new compound is
closely related chemically to the azo amanitin ABGG which has already
been successfully utilized to prepare several conjugates that have
demonstrated potent and specific cytotoxicity (see chapter I).
The conjugation of ABGG-GLU to BSA was studied in order to permit
direct comparisons with the data obtained with the model compounds.
The behavior of ABGG-GLU in this system appeared more complex than
that of PNBGG-GLU in some important respects. Therefore, further
studies were undertaken to identify and optimize parameters which were
felt likely to be particularly relevant to the conjugation of
Materials and Methods
N-4-aminobenzoylglycylglycine (PABGG) was obtained from Dr. James
F. Preston and was prepared by hydrogenation of PNBGG as previously
descri bed (Preston et al., 1981). Piperazine-N,N_-bis (2-ethanesulfonic
acid) (PIPES) was obtained from Sigma Chemical. Other reagents were
obtained or prepared as indicated in previous chapters.
Chromatography, spectroscopy, and elemental analysis. These
methods are fully described in previous chapters.
RNA polymerase assay and source of toxin. The protocols for
examining the inhibition of RNA polymerase activity and for purifi-
cation of AMA are outlined in chapter II.
Synthesis of ABGG-GLU
Synthesis of PABGG-GLU. Two methods for the synthesis of
PABGG-GLU have been developed (Fig. V-1):
In the first method PABGG (0.126 g, 0.50 mmol) and glucosamine
hydrochloride (10.8 g, 50 mmol) were added to 43.5 ml of water in a
flask. The PABGG was brought into solution by dropwise addition of
0.55 ml of 10 M Na0H with continuous rapid stirring. EDC (0.192 g, 1.0
mmol) was added to this solution four times at 30 min intervals;
during this time the reaction was allowed to proceed at 230C in the
dark. After a total reaction time of 2 h the solution was concentrated
in vacuo at 30*C to near dryness. Remaining water was removed as an
azeotrope with ethanol by twice suspending the residue in 25 ml of
H2N O CONHCH2CONHCH2C00H + H2 O
02N -~ -s~oOH H20H
+ Pd/C + FORMAT
CONHCOH CH20H C
H2N O2 2@
~CONHCH CO FCH2COm O
Fig. V-1. Scheme showing the two routes for synthesis of
PABGG-GLU and the synthesis of ABGG-GLU from AMA and PABGG-GLU.
absolute ethanol and evaporating in vacuo at 30*C. The dried residue
was suspended in 50 ml of methanol and filtered through Whatman # 40
paper. The filter cake was washed twice with 25 ml of methanol at room
temperature. The pooled filtrates were concentrated to dryness in
vacuo at 300C dissolved in 10 ml of water, and applied to a 4 X 20
cm column of Sephadex SP-50 (Na ) which was eluted with 500 ml of
water. This aqueous Sephadex SP-50 eluate was concentrated in vacuo at
40*C to a small volume, applied to a 2.5 X 95 cm column of Sephadex
LH-20, and eluted with water at 0.2 ml/min to collect 7.5 ml frac-
tions. The desired product (0.19 mmol, 38%) eluted between fractions
53 and 60.
A second method of preparing PABGG-GLU involved catalytic trans-
fer hydrogenation (cf. Anwer and Spatola, 1980) of PNBGG-GLU. To
PNBGG-GLU (0.50 mmol) in 45 ml of water was added 5 ml of 1.0 M sodium
format, pH 3.5. The solution was sroarged with nitrogen for 15 min and
then 0.375 g of 10% Pd on charcoal (Kodak) was added. This mixture was
stirred for 12 h in the dark at 23*C under a nitrogen atmosphere. The
cata-lyst was then removed by filtration through Whatman # 40 paper;
the catalyst was washed with 50%; acetonitrile until the A278 of the
filtrate was negligible. The combined filtrates were concentrated to a
small volume in vacuo at 500C and applied to a 2.5 X 95 cm column of
Sephadex LH-20. Fractions of 7.5 ml were collected at approximately
0.2 ml/min. The desired product (0.41 mmol, 82%) eluted in fractions
51 to 58. TLC: RF, I-0.31, III-0.56;. For C17H24N408 calculated:
C 49.51, H 5.87, N 13.58; found: C 47.23, H 6.28, N 13.18. [n]25=+16*.
'278=1.38 X 104 M1 cm-1 PMR: 3.4-4.0 (m, 6H, sugar); 4.015 (s, 2H,
glycine); 4.12 (s, 2H, glycine); 4.75 (d, 0.4H, J=8.0 Hz, anomeric);
5.20 (d, 0.6H, J=3.3 Hz; anomeric); 6.87 (d, 2H, J=8.4 Hz, aromatic);
7.70 (d, 2H, J=8.6 Hz, aromatic).
Synthesis of ABGG-GLU. PABGG-GLU (30 pimol) was dissolved in 0.75
ml of ice-cold 0.1 M HC1. The PABGG-GLU was diazotized by addition of
30 pl of 1 M sodium nitrite followed by incubation at 23*C in the dark
for 30 min. A portion of this solution (0.65 ml) was added to 25 pmol
of AMA which was dissolved in 1.20 ml of 0.5 M sodium phosphate, pH
8.0. There was immediate development of a deep purple color. After 5
min the crude product was desalted using the Sep Pak C18 protocol (see
chapter III). The desalted material was dried in vacuo at 50"C,
redissolved in acetonitrile: TFA-TEA buffer (18:82), and purified by
HPLC. The desired product (13.8 plmol, 55%) eluted between 32 and 40
ml. After evaporation of the acetonitrile in vacuo at 300C the final
product was desalted by the Sep Pak C18 procedure. TLC: I-0.17,
III-0.44. FAB-MS: [M+H] expected, 1342.50; found, 1342.54; [M-H] -
expected, 1340.485; found, 1340.50.
Conjugation of ABGG-GLU to BSA
Reaction mixtures were assembled as described in the previous
chapter except that ABGG-GLU was substituted for the PNBGG deriva-
tives. Excess ABGG-GLU was removed from the conjugates by the ultra-
filtration method which was detailed in chapter IV. The only modifi-
cation made in this protocol was that the conjugates were incubated in
the presence of EDTA for a full hour before the second centrifugation.
The extent of conjugation was estimated spectrophotometrically.
The molar extinction coefficient of ABGG-GLU at 395 nm was taken to be
4 -1 -1
1.4 X 10 M cm (Faulstich and Trischmann, 1973); the extinction at
279 nm in 0.1 M sodium phosphate, pH 7.0, was determined to be
4 -1 -1 1%
1.23 X 10 M cm .The E at 279 nm and molecular weight of BSA
(see chapter IV) were used to calculate a molar extinction at 279 nm
of 4.54 X 104 M1 cm-1; BSA does not absorb at 395 nm. Using these
parameters the molar ratio of ABGG-GLU to BSA in the conjugates was
estimated by the formula: 3.24A395/(A279-0.88A395 '
The fate of ABGG-GLU in the reaction mixtures was determined by
analytical HPLC. Duplicate samples (2-10 pl each) were diluted in
0.1 M sodium phosphate, pH 3.0, to yield a final concentration of
approximately 0.2 mM ABGG-GLU. The diluted samples were applied to
HPLC by an automated sample injector (Waters Model 7108 WISP) and
chromatographed at 1 ml/min on a C18 Z-module (Waters) using aceto-
nitrile: TFA-TEA buffer (18:82) as the eluant. The absorbance at 304
nm was monitored and peaks were integrated by a Waters Model 720 Data
Synthesis and Properties of ABGG-GLU
Preparation of PABGG-GLU. Two routes for the synthesis of
PABGG-GLU have been developed (Fig. V-1). When initial efforts to
reduce PNBGG-GLU failed, a first method was developed by which PABGG
was coupled to the anino group of D-glucosamine with the aid of EOC.
This method required a large excess of D-glucosamine in order to pre-
vent polymerization of the PABGG. In the second method PNBGG-GLU was
selectively reduced to PABGG-GLU by catalytic transfer hydrogenation
with palladium on charcoal as the catalyst and sodium format as the
reductant (cf. Brieger and Nestrick, 1974; Anwer and Spatola, 1980).
The products of the two syntheses were found to be identical by seve-
ral criteria, including PMR spectrum (Fig. V-2), UV absorbance spec-
trum (Fig. V-3), TLC mobility, elemental analysis, and optical rota-
Azo coupling of diazotized PABGG-GLU to o-amanitin. A synthesis
of ABGG-GLU was devised which provided significantly increased yield
over previously reported procedures (Falck-Pederson et al., 1983;
Preston et al., 1981). First, because of its low nucleophilicity, the
ary1 amine function of PABGG-GLU was diazotized in dilute HC1 (Fig.
V-1) in which the potent nitrosating agent nitrosy1 chloride is gener-
ated (Challis and Butler, 1968). Second, the azo coupling reaction was
conducted in a sodium phosphate buffer at pH 8.0. At this pH the
phenolic hydroxyl of AMA, which has a pKa of approximately 10 (Falck-
Pederson, 1981), can be expected to be slightly ionized and the ten-
dency of the diazotized PABGG-GLU to be converted to an unreactive
diazotate is slight (Zollinger, 1961). The yield of ABGG-GLU obtained
using these conditions is reliably 50-60% and most of the remainder of
the AMA is recovered in unreacted form.
Properties of ABGG-GLU. ABGG-GLU is in many ways very similar to
its predecessor ABGG. Their UV/visible absorbance spectra (see Fig.
V-4) are indistinguishable. They demonstrated very similar potency of
inhibition of CT RNAP II; ABGG-GLU showed a K. of 4 X 109 M, while
that of ABGG was 3 X 10-9 M. Examination of ABGG-GLU by reverse phase
HPLC under analytical conditions showed two poorly separated peaks.
These apparently represent the two diastereomers which exist in solu-
tion as a consequence of the anomeric nature of the sugar residue. A
similar phenomenon has been observed for PNBGG-GLU and PNBGG-GAL
200 250 300
Fig. V-3. UV absorbance spectrum of PABGG-GLU at 1.8 X 10-5 M in
5 mM sodium phosphate, pH 7.0, (light line) and 0.1 M HC1 (dark line).
L 0 *-I
to CV O
(data not shown). FAB-MS studies of ABGG-GLU supported the predicted
structure, showing mass ions of the expected sizes (Fig. V-5).
Effects of Alkaline pH and Nickel (II) lon on ABGG-GLU
Preliminary trials of conjugation of ABGG-GLU to proteins re-
vealed some unexpected problems. The most readily noticed of these was
that the nickel (II) ion, which was added to absorb the cyanide gener-
ated in the course of the reaction, reacted with ABGG-GLU to form a
complex as evidenced by an altered absorbance spectrum (Fig. V-6).
ABGG-GLU could be recovered intact from the complex by adding EDTA to
the solution, but at least 45 min was required for the action of the
EDTA to be completed.
Also, when conjugation was undertaken at pH 8.0, decolorization
of the solution was noted over a period of days. This was faster in
reactions where the nickel (II) chloride was omitted. The loss of
color appeared to correlate with the appearance of a new toxin-related
entity which had very low mobility on TLC. When ABGG-GLU was placed at
370C in 0.2 M sodium borate buffer, pH 8.0, a slow alteration of the
UV/visible absorbance spectrum was noted (Fig. V-7).
Conjugation of ABGG-GLU to BSA
Influence of pH and reactant concentrations on the rates of toxin
conjugation and alteration. Since difficulties with degradation of the
toxin were noted under the conditions which were developed for cou-
pling the PNBGG derivatives to BSA, means of circumventing this pro-
blem by altering the reaction conditions were sought. It had been
noted in work with PNBGG-GLU that significant coupling occurred at
+ E g t
a O -
O I O
Z O -f
+ O rC
I 'Ull s OO
O O O O O~ O O O O O-
9 0NO co co q