The importance of electrostatic interactions in the binding of paraquat to its elicited monoclonal antibody

OMI-5890190$3.00 + 0.00 Pergamon Press plc

Mokcuiar hminolog,~, Vol. 27, No. 9, pp. 847-852, 1990 Printed in Great Britain.

THE IMPORTANCE OF ELECTROSTATIC INTERACTIONS IN THE BINDING OF PARAQUAT TO ITS ELICITED MONOCLONAL ANTIBODY* MARK R.

BOWLES

and SUSANM. POND

University of Queensland, Department of Medicine, Princess Alexandra Hospital, Brisbane, Queensland 4 102, Australia (Firs1 received 7 December

1989; accepted in revised form I February 1990)

Abstract--In this study the pH-dependent interactions interactions between a paraquat-specific murine monoclonal antibody and two antigens: paraquat (l,l’-dimethy~-4,~-bipyridinium dichloride), and a ~-nitrophenol analogue ( I-(N-methyl-4,4’-bipyridinium)-I-(2-hydroxy-5-nitrophenyl)methane dichloride; PQNP) were determined by ELISA. In each case the &,-pH profile reflected the titration of a single amino acid residue. pK; values derived from these plots were 8.90 (paraquat) and 8.13 (PQNP). Increasing pH led to a significant increase in the association constant for each antibody complex. A spectrophotometric titration of PQNP in the presence and absence of excess antibody indicated the presence of another charged amino acid residue at the binding site, which could be assigned as a carboxylic acid. From these studies, a model for paraquat-antibody binding has been proposed.

INTRODUCTION

Determination of the effect of pH on protein-ligand binding has often suggested which chemical determinants are important to the interaction (Koshland, 1959; Stoops ef al., 1975). In addition, a “reporter*’ group (Burr and Koshland, 1964), such as p-nitrophenol attached to either ligand or protein has often facilitated a more detailed examination of the nature of binding. We are particularly interested in examining the

chemicai determinants of antibody-ligand binding because this knowledge would facilitate the production of high affinity antibodies by site-directed mutagenesis. Some of the determinants can be deduced from crystal structures (Chothia et al., 1986) or molecular modelling (Roberts ef al., 1987) of the antibody-antigen complex. Using a molecular model of Gloop2, a monoclonal antibody against hen eggwhite lysozyme, Roberts et al. (1987) predicted that a glutamic acid residue and a lysine residue on the antibody were important to antigen binding. In contrast to their prediction, site-directed mutagenesis of Gloop2 to change these residues to neutral species increased both the affinity and specificity of binding. This effect was mimicked in an experiment in which an increase in ionic strength increased the affinity of Gloop2 for its antigen dramatically. This example emphasizes the importance of conventional physico--*Supported by the Australian National Health and Medical Research Council. Address correspondence to: Dr M. R. Bowles, Department of Medicine, University of Queensland, Princess Alexandra Hospital. Wooltoongabba, Queensland 4102, Australia

chemical studies of antibody-antigen binding prior to more sophisticated experiments. The effect of pH on ligand binding by polyclonal antibodies has been determined. The antigens studied include: 2,4-dinitrophenyl (DNP) substituted naphthols (Metzger et al., 1963; Froese, 1968); l-pD-arabinofuranosylcytosine (ara-C) and ara-uracil (Okabayashi et al., 1977); and D-group and B-group erythrocytes (Hughes-Jones et al., 1964; Gerbal ef al., 19’75). These studies revealed a variety of pH effects which are difficult to interpret because of the heterogeneous nature of the antibodies. Studies of homogeneous immunoglobulin preparations, monoclonal antibodies, can be interpreted more easily. One series of investigations of the effect of pH on antibody-ligand binding began with the observation that the pK(, of the DNP substituted naphthois increased dramatically when bound to polyclonal antibodies (Metzger et al., 1963; Froese, 1968). This effect also was demonstrated in studies of the binding of a series of these compounds to the anti-DNP monoclonal antibody MOPC 315 (Haselkorn et al., 1974). The production of the Fv fragment (molecutar weight 25 kD) of MOPC 315 allowed the use of proton nuclear magnetic resonance (pmr) difference spectroscopy to titrate three histidine residues in the presence and absence of DNP. The pK: values of the observed histidine resonances were altered in the presence of excess DNP and also appeared in a spin label difference spectrum. This demonstrated the proximity of these residues to the binding site (Dwek et al., 1975). Such “direct” observation of titrating groups in antibody-antigen complexes by pmr demonstrates the potential of this technique. However, its use is 847

MARK

848

R. BOWLES and SUSANM. POND

limited by the following requirements: production of an antibody fragment of a size suitable for the identification of individual amino acid residues but which still conserves the antigen-binding capacity; substantiaf (milligram) quantities of the fragment; resolution of the resonances of interest from the spectral background, which includes any contribution from the antigen; determination of the spectrum by a Fourier transform nuclear magnetic resonance spectrometer. We have developed a method which allows the rapid determination of the pH dependence of antibody-antigen reactions using a conventional enzyme linked immunosorbent assay @LISA). We have applied this method in combination with studies involving a “reporter” analogue of paraquat to make a detailed examination of the binding of paraquat to a specific murine monoclonal antibody.

MATERIALSAND

METHODS

Paraquat (Sigma Chemical Co., St Louis, MO), 2-hydroxy-5-nitrobenzyl bromide (Aldrich Chemical Co., Milwaukee, WI), 4,4’-bipyridyl (BDH, Poole, England), non-specific human immunoglobulin (Sandoz Ltd, Basle, Switzerland) and goat antimouse immunoglobulin, labelled with horseradish peroxidase (Cappel Worthington Biochemicals, Malvern, PA) were of the highest quality available and were used without further purification. The monoclonal antibody to paraquat, an IgGzb with a kappa light chain, was produced and purified according to Bowles et al. (1988). Synthesis qf l-(~-meth~~i-4,~-b~pyrid~nium)-i-(2-hydroxy -5nitrophenyl)

methane dichloride

1-Methyl-4-(4’-pyridyl)-pyridinium methosulphate (monoquat methosulphate) was prepared by the dropwise addition of dimethyl sulphate to excess 4,~-bipyridyl in dry chforoform. The precipitated product was washed with twenty volumes of chloroform and dried in z’acuo over phosphorus pentoxide. Monoquat methosulphate (4.25 mmol.) was dissolved in IOOml isopropanol and filtered (Whatman 541). To this solution was added 2-hydroxy-5-nitrobenzyl bromide (22.9 mmol.) following which reflux was commenced. After 4X hr the precipitate which formed on cooling still contained 10% monoquat as determined by HPLC (Gill et al., 1983). After an additional 24 hr reflux, the hot reaction mixture was filtered to obtain 1.I g (50% theoretical yield) of yellow-orange crystals. Shiny white crystals, the dichloride salt. were obtained by recrystallization from HCi/acetone. The HPLC assay showed the product to be >99% pure. Structural confirmation was made by proton NMR in DzO using 2,2dimethyl-2-silapentane-5-sulphonic acid sodium salt (DSS) as the internal standard. Peaks were obtained

at 4.12 (3H), 5.04 (2H), 5.97 (lH), 6.94 (IH). 7.2 (5H), 7.64 (2H) and 7.80 ppm (2H) with the appropriate splitting. No extraneous peaks were observed. De~erminarion of association antigen binding

constants for antibody-

The intrinsic association constant (KAS) for antibody-paraquat interactions at each pH were determined according to the theoretical analysis of Hogg et al. (1987) using the ELISA developed by Johnston et al. (1988). Briefly, results in a competitive inhibition ELISA with total antibody concentration, [Ab], were analysed by the expression

(1) in which R denotes the ratio of ELISA readings (&/A,,) in the absence and presence of a concentration [Ag] of antigen, andfis the antibody valence. The intrinsic binding constant for the interaction of the antibody with immobilised paraquat-ovalbumin conjugate (I(,,) was obtained from a series of ELfSAs. These assays measured the concentration dependence of antibody binding to the solid phase in the absence of additional paraquat. The value of fKAXr determined from a plot of A,/[Ab] US A, (Hogg et al., 1987), was then substituted into [I] to yield KAS. To determine the time required for the antibodyantigen reactions to reach equiIibrium, the antibody (20 nM) was incubated at room temp in 0.05 M phosphate (pH 6.31) buffer and 0.05 M carbonate buffer (pH 9.84) in the absence and presence of paraquat (100 nM). After 0, 15, 30, 60, and 120 min, ahquots from each pre-equilibration mixture were assayed for antibody activity by ELISA. Because of the pH range required in these experiments, unreacted sites on the ELISA plate were blocked with 2% (w/v) bovine serum albumin (Sigma Chemical Co.) in phosphate buffered saline (PBS). Buffers used in the ELISAs were 0.05 M sodium phosphate buffer (pH 6.0-7.0) 0.05 M Tris-HCl buffer (pH 7.4-8.4) and 0.05 M sodium carbonate buffer (pH 8.5-9.84). All were prepared at constant ionic strength (p = 0.16) by the addition of potassium chloride. All analyses were performed in triplicate. For each determination, antibody was pre-equilibrated in the presence or absence of paraquat in the appropriate buffer for 2 hr. Before the addition of these samples, the wells were washed with the preequilibration buffer (0.250 ml). Reaction mixtures (0.100 ml) containing antibody (40.7 nM) and paraquat (O-47.9 nM) were added to the coated, washed wells. To the remaining wells, which were used for analysis of the interaction between antibody and immobilized paraquat, was added 0.100 ml of the antibody (O-42.4 nM). After removal of the preequilibration mixtures from the polystyrene plate, the pH of each solution was measured and found not to have changed substantially (~0.01 pH units).

Antibody-paraquat

Spectral properties of PQNP presence of excess antibody

849

binding

in the absence and

All UV-visible spectra were measured in a Kontron Uvikon 8 10 spectrophotometer at 25.0 f 0.1 “C. Buffers adjusted to the same ionic strength (p = 0.16) with potassium chloride were used throughout these studies. The absorbance at 394 nm was determined as a function of PQNP concentration (O-0.126 mM) in 0.05 M phosphate buffer (pH 6.40). The spectrum (450 to 240 nm) of PQNP was recorded in 0.1 M HCI, 0.05 M phosphate buffer (pH 6.00) and 0.05 M carbonate buffer (pH 10.0). Spectra (450 to 350 nm) were also obtained in 0.05 M acetate buffer (pH 4.07 and pH 4.95) and 0.05 M phosphate buffer (pH 5.60 and pH 6.40). Difference spectra (450 to 350 nm) of PQNP (8.8830 PM) in the presence of either excess paraquat antibody (32 PM) or excess non-specific antibody (32 PM) were recorded in 0.05 M phosphate buffer at pH 5.96, 6.42, 6.88, 7.42 and 7.96 and in 0.05 M carbonate buffer (pH 10.00). The spectra in the presence of paraquat antibody were recorded under conditions in which the concentration of unbound PQNP (calculated from the KAs) was less than 1% of the total PQNP concentration. Such conditions precluded the need for any spectral correction. Isoelectric ,focusing of the paraquat antibody Polyacrylamide isoelectric focusing (IEF) gels, pH 3.5-9.5, purchased precast on plastic sheeting (LKB-Produkter AB, Bromma, Sweden), were run and stained according to the manufacturer’s instructions on a LKB Model 2117 Multiphor apparatus. Samples containing 30 pg of antibody were applied to the gel by paper wicks. After focusing (30 W, constant power) for 1.5 hr at lO“C, the pH gradient was measured using a surface pH electrode (LKB). Gels were fixed, then stained with Coomassie Brilliant Blue R-250. RESULTS

Determination of association constants for antibodyantigen binding Equilibrium of antibody-paraquat binding was established in less than one hour in all cases. Figure 1 shows the effect of pH on KAs for each of the two antigens. The solid line in each case is the theoretical titration curve based on equation 2 (Fersht, 1985): K,s = (K,[H+] + K2K:)/([H+]

+ K;)

(2)

in which K: is the dissociation constant for an acid group on the antibody, and K, and Kz are association constants for antibody-antigen binding at the low and high pH extremes of the range, respectively. For paraquat, values of K,, K2 and pKi were 4.1 x IO’M -I, 21.5 x IO’M-’ and 8.90, respectively. For PQNP, these values were 3.0 x lo4 M-‘, 1.10 x lO”M-‘, and 8.13, respectively.

PH Fig. 1. K,sspH profiles for the interaction of paraquat (0) and PQNP (0) with the paraquat-specific monoclonal antibody. The left axis shows the KAs range for paraquat and the right for PQNP. For both antigens, the solid line is a theoretical titration based on eqn 2 (see text). The error bars represent the standard error calculated by analysis of covariance.

Spectral properties of PQNP presence of excess antibody

in the absence and

A linear dependence of A,,, on concn (o-0.126 mM) of PQNP was observed, indicating that the Beer-Lambert law applied under these conditions. Spectra of PQNP in the absence of antibody are presented in Fig. 2. The conjugate acid of PQNP (in 0.1 M HCl) had a peak at 260 nm (c = 23,200 M-‘cm-‘) and a shoulder at 312 nm (6 = 10,400 M-‘cm-‘). The conjugate base of PQNP (0.05 M carbonate buffer, pH 10.00) had peaks at 257 nm (c = 21,600 M-’ cm-‘), due to the paraquat moiety, and at 394 nm (c = 16,400 M-’ cm-‘) due to the p-nitrophenol moiety. The similarity of these values to those for paraquat (A,,,,, = 256 nm, t = 21,100 M-’ cm-‘) and p-nitrophenolate ion (E.,,, = 400nm, L = 18,320 M-’ cm-‘; Kezdy and Bender, 1962) demonstrates that there is little conjugation between the paraquat and p-nitrophenol moieties of PQNP. In the presence of excess specific antibody, the spectrum of PQNP (Fig. 3) was dramatically altered

1

I

..\ I..

2.0

‘.

--r i.’ ;I

1.5 .-

1.0 .0.5 .-

\

‘..\‘\,,

_...,

: \ \ I '1' 1 \ 1, ' :\ .. .

\

'.\

.' , -\

\ \'

,:\

\'.

'.\ '._ 240

290

.. \

.,' _ 2:'

390

340

440

4910

X bd Fig. 2. Spectra of PQNP in the absence of antibody recorded in 0.1 M HCI (-.,-), 0.05 M phosphate buffer, pH 5.60 (---) and 0.05 M carbonate buffer, pH 10.0 (--.),

850

MARK R. BOWLESand 14.0--

SUSANM. POND

., ......-.... - - <.. .$

12.0-.;;;

10.0..

.>

LO--

i .i .

8.0-i .._._,_.4.0.-

.- .-.-.-.

-.

2.0.0.0 4

370

3!50 300

360

300

400

...... . ..;c - -...

14.0

.;i 12.0

i

t

. :,

t

.

I

430

440

430

430

440

4

‘;L.

.f

_. /.-.-.

i

.

.' / .t /.i / .

6.0

420

k

.;i;

10.0 8.0

410

i.

\ \

'

\

4.0 2.0 \

0.04

350

360

370

380

14.0..

i 6

12.0--

i

lO.O--

3 ” 7 2

LO-- :# 6.0-i

400

......‘.....

l&O-_

300

.5.2.<.

.‘/

.:;. 3.

0

410

420

..

.

pH 10

. ‘$.

:.:“ :

. .. v.

.&.. . b...

ai.: \I... . F

4.0-2.0-0.0 4

350

360

370

380

390

400

410

420

430

440

’

Fig. 3. Spectra of PQNP recorded in the absence of antibody (-,-.-), and in the presence of excess non-specific antibody (, ,) or paraquat-specific antibody (---). at pH 6.00 but still reflected a normal titration of PQNP. A linear transformation of the spectrophotometric titration of PQNP is shown in Fig. 4. The pK; (the intercept) of PQNP was increased from 5.41 to 6.19 when it was bound to the specific antibody. Isoelectric focusing C$ the paraquat antibody Isoelectric focusing of the paraquat antibody yielded a typical “family of bands” (Awdeh et al., 1970) within a span of 0.2 pH units (7.12-7.32). DISCUSSION

The intrinsic of the antibody

association constant for the reaction with both antigens increased with pH.

This relationship, which follows the theoretical titration for a single acid group (eqn 2) must be due to an amino acid residue on the antibody binding site. It does not reflect the net charge on the entire immunoglobulin molecule, as would be suggested by the theory of Endo et al. (1987). This theory relates the pH optimum of an antibody-antigen reaction to the molecular weights and pI values of clonotypic rabbit antibodies and their corresponding antigens. The empirically determined mid-points from KAs t’s pH curves for paraquat (8.90) and PQNP (8.13) do not mirror the isoelectric point of the antibody which is about 7.25. The apparent disparity between this result and the relationship proposed by Endo et al. (1987) is probably due to the following. The binding of a low molecular weight antigen such as paraquat

Antibody-paraquat binding

851

6.56.0.. 7.5.. 7.0.. 6.5.. pH

6.0.. 5.5.5.0.. 4.5.. 4.0-m 3‘S.l -2.0

-1.5

-1.0

-0.5

‘W f ( *A-

0.0 -

capp

0.5

1/

1.0

1.5

aopp 1

Fig. 4. The spectrophotometric titration of PQNP in the absence (0) and presence of excess antibody (0) according to an adaptation of the Hendersen-Hasselbalch equation. pH = pK; - log@,_ - tgpp)/(ta&, in which tA_ is the molar absorption coefficient of PQNP anion and cappis the apparent molar absorption coefficient at any given pH.

(257.2 D) by an antibody

should not be affected by the net charge on the entire antibody molecule but only by the adjacent charged groups in the antigen binding site. The interaction of thyroxine (777.2 D) with its antibody also provides an example of a small hgand which does not fit the pH dependency observed for protein antigens (Endo et al., 1987). When interpreting the pH dependence of all protein-hgand interactions, the following must be considered. A hydrophobic environment causes a very small decrease in pK: for a cationic acid and a much larger increase in the pKi of a neutral acid. This effect results from stabilisation of the protonated form of a neutral acid or the conjugate base of a cationic acid by media of low dielectric constant. Secondly, an adjacent positive charge causes the pK: of any acid (neutral or cationic) to decrease. Finaliy, an adjacent negative charge causes the pKi of any acid (neutral or cationic) to increase. From the foregoing arguments, the increase in pK: of the PQNP in the presence of antibody suggests the presence of a hydrophobic binding site or an adjacent negative charge. The polar nature of paraquat implies that its elicited antibody would possess a binding site of a complementary polar nature. Therefore, an anionic amino acid residues is most probably responsible for the increase in the pK: of PQNP. For two reasons, thisamino acid residue cannot be the residue whose titration results in the observed fC,s-pH profhe. Firstly, the IY,s-pH profile was not perturbed by the titration of PQNP (pKi = 6.19). Analogously, the PQNP-antibody complex displayed a spectrophotometric titration in which there was no evidence for an adjacent group with a pK; of 8.13. Because of this absence of effect in the pH range 610, the anionic amino acid residue adjacent to the p-nitrophenol moiety of PQNP must have a pKi more than 1 pH unit lower than pH 6. By inference, this group must be a carboxylic acid residue (glu or asp). The structure in Fig. 5, in which RH is an antibody residue titrating with a pK: of 8.13, is the simplest

Fig. 5. Proposed model for the PQNP-antibody

complex.

scheme which accounts for the requirements imposed by the binding and spectral evidence. Firstly, there is an amino acid residue (RH) which titrates with a pK: of 8.13. The titration of this group while dramatically altering the binding of hgand to antibody does not affect the titration of then-nitrophenol moiety and so must be spatially removed from it. Secondly, the pK: of the p-nitrophenol group is increased due to an adjacent anionic (carboxylic) acid residue. In support of such a model, the association constant for monoquat at pH 7.4 (a monocation) is 2.3 + 0.5 x lo6 M-’ (S. Johnston, unpublished results) compared with 1.85 + 0.3 x 106M-’ for PQNP at the same pH. The values are in contrast to the KAs of 4.8 & I .I x lO’ M-’ for paraquat. The reason for the similarity between monoquat and PQNP and the disparity with the paraquat association constant is due to the formation of an ionic bond between the phenolate anion and the adjacent pyridinium nitrogen which effectively masks the positive charge. The formation of this ion pair decreases the interaction between the carboxylic acid and the adjacent pyridinium cation to such an extent that the interaction with the antibody resembles that of monoquat which is a monocation. That such an ionic bond is formed in PQNP can be deduced from a comparison of its pK: (5.41) with that of p-nitrophenol (7.04; KCzdy and Bender, 1962). From the change in pKi of p-nitrophenol, a change in free energy of 2.23 kcal/mol. can be calculated from the relationship between free energy and the equilibrium constant, AG” = - RT ln K, in which R is the gas constant and T is the absolute temperature. By considering the spectral evidence that there is very little conjugation between the two ring systems of PQNP, this change in free energy can only be ascribed to the formation of an ionic bond. The procedure for the determination of antibodyantigen pH dependence described here requires only the instrumentation involved in standard ELISAs and very small quantities of antibody and antigen. Together with the evidence obtained in the PQNP spectral studies, these pH dependences allow the construction of a model of paraquat-antibody binding. Combination of the knowledge of overall antibody-antigen structure obtained from the X-ray crystallography or molecular modelling, with the acid-base chemistry of the interaction, should allow strategic site-directed mutagenesis of the antibody

MARK R. BOWLES and

852 REFERENCES

Awdeh Z. L., Williamson A. R. and Askonas B. A. (1970) One cell-one immunoglobulin. Biochem. J. 116, 241-248. Bowles M. R., Johnston S. C., Schoof D. D., Pentel P. R. and Pond S. M. (1988) Large scale purification of paraquat and desipramine monoclonal antibodies and their Fab fragments. Int J. Immunopharmacol. 10, 537-545.

Burr M. and Koshland D. E., Jr (1964) Use of “reporter groups” in structure-function studies of proteins. Eiochemistry 52, 1017~1024. Chothia C., Lesk A. M., Levitt M., Amit A. G., Mariuzza R. A., Phillips S. E. and Poljak R. J. (1986) The predicted structure of immunoglobulin Dl.3 and its comparison with the crystal structure. Science 233, 755-758. Dwek R. A., Jones R., Marsh D., McLaughlin A. C.. Press E. M., Price N. C. and White A. I. (1975) Antibody-hapten interactions in solution. Phil. Trans. R. Sot.

Land.

B. 272,

53.-74.

Endo Y., Miyai K., Hata N. and Iijima Y. (1987) Effect of pH on antigen binding by clonotypic antibodies with different isoelectric points. Biotechnol. appl. Biochem. 9, 7448

1,

Fersht A. (1985) Enzyme Structure and Function, 2nd edn, Chapter 5, pp. 1555174. W. H. Freeman and Co., New York. Froese A. (1968) Kinetic and equilibrium studies on 2,CDinitrophenyl hapten-antibody systems. Immunochemistry

5, 253-264.

Gerbal A., Maslet C. and Salmon C. (1975) Immunological aspects of the acquired B antigen. VOX Sang. 28, 3988403. Gill R., Qua S. C. and Moffat A. C. (1983) High-performance liquid chromatography of paraquat and diquat in urine with rapid sample preparation involving ion-pair extraction on disposable cartridges of octadecyl-silica. J. Chromar.

225,

483490.

SUSAN M. POND

Haselkorn D., Friedman S., Givol D. and Pecht I. (1974) Kinetic mapping of the antibody combining site by chemical relaxation spectrometry. Biochemistry 13, 2210-2222.

Hogg P. J., Johnston S. C., Bowles M. R., Pond S. M. and Winzor D. J. (1987) Evaluation of eauilibrium constants for antigen-antibody interactions by solid phase immunoassay: the binding of paraquat to its elicited mouse monoclonal antibody. Molec. Immun. 24, 797-801, Hughes-Jones N. C., Carter B. and Telford R. (1964) The effect of pH and ionic strength on the reaction between anti-D and erythrocytes. Immunology 7, 72-81. Johnston S. C., Bowles M. R., Winzor D. J. and Pond S. M. (1988) Comparison of paraquat-specific murine monoclonal antibodies produced by in ttitro and in oico immunization. Fund. appl. Toxicol. 11, 261-267. Kezdy F. J. and Bender M. L. (1962) The kinetics of the z-chymotrypsin-catalyzed hydrolysis of p-nitrophenyl acetate. Biochemistry 1, 1097-I 106. Koshland D. E.. Jr (1959) In The Enzymes, 2nd edn, Vol. I, (Edited by P. D. Boyer, H. Lardy and K. Myrback) pp. 3 17-3 18. Academic Press, New York. Metzger H., Wofsy L. and Singer S. J. (1963) A specific antibodyyhapten reaction with novel spectral properties. Archs Biochem. Biophys. 103, 206-2 15. Okabayashi T., Mihara S., Repke D. B. and Moffatt J. G. (1977) A radioimmunoassay for 1-beta-D-arabinofuranosyluracil with reference to cross-reactivity of I-beta-Darabinofuranosylcytosine with an antibody. Cancer Res. 37, 3132.-3135.

Roberts S., Cheetham J. C. and Rees A. R. (1987) Generation of an antibody with enhanced affinity and specificity for its antigen by protein engineering. Nature 328, 731-734. Stoops J. K., Hamilton S. E. and Zerner B. (1975) Carboxylesterases (EC 3.1.1). Comparison of some kinetic properties of horse, sheep, chicken, pig, and ox liver carboxylesterases. Can. J. Biochem. 53, 565-572.