Dissociation kinetics of antigen-antibody interactions: studies on a panel of anti-albumin monoclonal antibodies

Molecular Immunology, Vol. 26, No. 2, pp. 129-136, Printedin Great Britain.

1989

0161-5890/89 $3.00 + 0.00 PergamonPress plc

DISSOCIATION KINETICS OF ANTIGEN-ANTIBODY INTERACTIONS: STUDIES ON A PANEL OF ANTI-ALBUMIN MONOCLONAL ANTIBODIES WILLIAM

C.

OLSON,*

THOMAS M.

SPITZNACEL

and

MARTIN L. YARMUSH~~

*Department of Chemical Engineering, Massachusetts Institute of Technology, U.S.A. and TRutgers University, Department of Chemical and Biochemical NJ 08554, U.S.A. (Received 26 April 1988; accepted in revisedform

Cambridge, Engineering,

MA 02139, Piscataway,

28 July 1988)

Abstract-Kinetic parameters and equilibrium association constants (K) are reported for a panel of antibovine serum albumin (BSA) monoclonal antibodies (MAb) immobilized onto agarose particles. For 12 covalently immobilized MAb of moderate affinity (K = 0.25 x IO*-1.2 x 10’ M-‘) measured dissociation time constants varied two orders of magnitude, from 2.1 to 410 min. Directly measured association rate parameters agree with values calculated from measured equilibrium and dissociation rate parameters. Dissociation time constants and equilibrium association constants were also determined for eight MAb immobilized biospecifically (via their F, regions). A significantly lower K was observed with those MAb which were covalently immobilized as opposed to biospecifically immobilized. These decreases in K appear to reflect decreased association rates rather than increased dissociation rates. The data suggest that, for the MAb described herein, dissociation rates do not correlate with equilibrium association constants.

INTRODUCTION

Monoclonal use

as

antibodies

primary

tools

(MAb) for

are finding

identification,

increasing assay

and

(Ag). In all of these applications, the MAbAg dissociation kinetics may play a crucial role in determining the operating parameters for a given test. For example, in vitro diagnostic assays such as ELISA typically employ a wash step.during which assay sensitivity may be lost if MAbAg dissociation times are comparable to wash times. In affinity chromatography, nonchemical elution techniques (i.e. competitive elution, electrophoretic elution) are ultimately limited by the MAbAg dissociation rate. In addition, dissociation kinetics will likely influence the valency of MAb binding (Mason and Williams, 1980) and extent of MAb penetration in in vivo immunoconjugate therapies (Weinstein et ul., 1986). Only a few studies have examined the kinetics of antibody-protein Ag interactions. Most investigations have employed either singular MAb systems (Roe et purification

of a variety

of antigens

fAuthor to whom reprint requests should be addressed. Abbreviations: c, free Ag concn; cO, initial Ag concn; h fraction of Ag irreversibly bound; K, equilibrium association constant; Kb, equilibrium association constant of biospecifically immobilized MAb; K,, equilibrium association constant of covalently immobilized MAb; k, , second-order association rate constant; n, number of active binding sites per mole of MAb; 4, bound Ag concn; qO, initial concn of immobilized MAb; f, time; 0, fraction undissociated; r, dissociation time constant (inverse of the first-order dissociation rate constant); tb, dissociation time constant of biospecifically immobilized MAb; T,, dissociation time constant of covalently immobilized MAb.

al., 1985; Mason and Williams, 1980; Brimijoin et al., 1985) or polyclonal antibody preparations (Levison et al., 1970; Noble et al., 1969, 1972; Sachs et al., 1972). Although kinetic parameters have been measured for a panel of anti-hapten MAb (Kranz and Voss, 1981) no similar study has been undertaken for MAb-protein Ag systems. Because hapten systems are inadequate models of protein Ag systems, important questions remain concerning the rates of MAb-protein Ag interactions. For example, for many anti-hapten antibodies, the association rate is relatively constant and near the diffusion limit (Pecht, 1982) and the equilibrium association constant is determined primarily by the dissociation rate. It is not clear whether similar correlations exist for protein Ag systems. Even fewer studies have examined the kinetics of antibodies which have been immobilized onto solid supports. These investigations have focused on association rates and have reported either semiquantitative parameters (Sportsman et al., 1983) or effective rate constants that may reflect mass transfer limitations (Chase, 1984). Thus, only imprecise estimates are available for the dissociation rates of immobilized MAb and the effects of different immobilization methods on MAb reaction rates have not been investigated. This report describes measurements of the intrinsic dissociation kinetics of bovine serum albumin (BSA) from a panel of anti-BSA MAb. Dissociation rates are found to vary by two orders of magnitude for MAb displaying similar equilibrium association constants. Association rate parameters directly measured for two MAb agree with values predicted from dissociation and equilibrium measurements. Dissociation rate 129

130

WILLIAM

c.

parameters and equilibrium association constants were also determined for MAb immobilized by biospecific as well as covalent methods. The effects of antibody immobilization methods on thermodynamic and kinetic parameters are described. MATERIALS

OLSON

CI d

Fisher Scientific (Boston, MA). Regenerated cellulose membranes (47 mm diameter, I .O lirn pore size) were purchased from Schleicher and Schuell (Keene, NH). Recombinant staphylococcal protein A was a gift of Repligen Inc. (Cambridge, MA).

AND METHODS

~~ier~it~~~ Superose 6 (11-15 pm porous agarose particles) was purchased from Pharmacia (Piscataway, NJ). Amine-terminated BioMag (1 pm nonporous paramagnetic particles) was obtained from Advanced Magnetics (Cambridge, MA). Pentex sul~ydryl modified bovine serum albumin was obtained from Miles Laboratory (Naperville, IL), purified by sizeexclusion chromatography (SEC) on a Sephadex G-l 50 column, and radiolabeled with “‘1 using either iodogen (Pierce, Rockford, IL) or the ICI method of McFarlane (1958). For experiments using Superose particles, only ICI -labeled BSA was employed. For experiments using BioMag particles, duplicate measurements were performed using BSA iodinated via both procedures. Equivalent kinetic and thermodynamic results were obtained with the two “‘I-BSA preparations; the results reported herein were obtained using iodogen-labeled BSA. The “‘I--BSA was 99 + % precipitated by 10% trichloroa~etic acid and determined to be free of oligomers by SEC over a GF-450 HPLC column (DuPont, Wilmington, DE). The sp. act. was determined by measuring radioactivity on a gamma counter (Packard Instruments. Downers Grove, IL) and the optical absorbance at 280 nm using an extinction coefficient of 0.60 mljmgi cm. One group of anti-BSA MAb was developed in this laboratory (Morel rr [II., 1988); others were the kind gift of David Benjamin (Benjamin cf al., 1985). The anti-BSA antibodies were isolated using affinity chromatography~ using a BSA-Sepharose 4B column and eluted using 0. I M gfycine-HCl at pW 2.5. They were then further puritied by SEC using a Sephadex G-200 column and concentrated over a YM-30 ultrafiltration membrane (Amicon. Danvers, MA). Each MAb preparation was determined to be free of oligomers by HPLC analysis as described above. A fresh BSA column and uit~filtration membrane was used for each MAb. Several MAb were also purified by conventional ion exchange chromatography methods using DEAE-cellulose (Whatman. Hillsboro. OR). MAb thermodynamic properties were observed to be independent of purification method (ion exchange vs affinity chromatography). Affinity purified goat antibodies to the F, fragment of mouse immunoglobuI~n were supplied by Jackson Immunoresearch (Avondale, PA). The buffer used for kinetic and thermodynamic experiments was 0.01 M potassium phosphate buffer with 0.02% sodium azide (PBA) at pH 7.4. Mono- and dibasic potassium phosphate and all other inorganic chemicals were supplied by Mallinckrodt (Paris, KY). Syringe filter tips (3 mm diameter, 5 jlrn pore size) were supplied by

T~le~~v~~~arni~ ~ea~s~re~~nt.~. Two sets of immobilized MAb were prepared: (I) a set which contained MAb directly coupled to the support matrix and (2) a set which contained MAb biospecifically bound to the support via an affinity ligand. In the first set, MAb were covalently immobilized onto Superosc 6 using the soluble CNBr technique of March et al. f 1974). Typically, 1-I .5 mg MAb were applied per ml swollen gel, with 80-95% coupling efficiencies being obtained. Superose 6 was selected for its small particle size. I3 pm. For BSA in Superose 6, intraparticle diffusion times are estimated to be less than 5 set and are thus small compared with the observed MAb dissociation times. The coupling buffer was 0.1 M potassium phosphate, pH 7.6. Active sites remaining after coupling were blocked using 1 M ethanolamine in 1 M sodium bicarbonate, pH 8. For the second set of immobilized MAb, adsorbents were prepared containing either (1) covalently immobilized antibodies specific for the F, region of murine antibodies or (2) immobilized protein A. The former were used for experiments with mouse IgG, MAb, while protein A was employed for IgGz MAb. The binding properties of the biospecifitally bound antibodies were determined as follows. Equal aliquots (10 p 1packed volume) of anti-F, or protein A beads were dispersed into 250 ~1 polyethylene centrifuge tubes. To these were added subsaturating amounts of anti-BSA MAb, varying amounts of 12’1-BSA and sufficient PBA to adjust the volume to 200 ~1. The bead suspension was rotated for 2 hr and then precipitated via low speed centrifugation. One hundred jr1 of the supernatant were transferred to a fresh tube and both supernatant and precipitate tubes were counted for ‘?‘I activity. Nonspecific adsorption of “‘I-BSA onto the adsorbents was negligible for these conditions, as determined by performing control assays with immobilized mouse immunoglobulin (Sigma, St Louis, MO). MAb equilibrium association constants and number of active binding sites were obtained by fitting the data to a Langmuir isotherm ‘I = il&!:c!(l + Kc)

(11

using nonweighted nonlinear least-squares regression. In equation (l), q is the bound Ag concn, n is the number of active binding sites per mole of MAb, X-is the equilibrium association constant and c is the free Ag concn. The n-values of the MAb used in this study ranged from I .2 to 1.9. Binding characteristics of the anti-f;, and protein A beads were determined using similar methods and “‘1-MAb of the approprjate subclass. For assays of covalently immobilized MAb the above methods were employed following the

Antigen-antibody

addition of MAb immunoadsorbents and lz511BSA to the microfuge tubes. ~~ss~~iatio~ rate measurements. A panel of antiBSA MAb were (1) immobilized onto agarose particles, (2) incubated with “‘1-BSA, (3) rapidly washed and then (4) contacted with a vast excess of unlabeled BSA. Periodically, samples of the bead suspension were withdrawn and filtered to determine the fraction of bound 1251-BSA. The reaction vessel for dissociation rate measurements was a 50 ml stirred ultrafiltration cell (Amicon, Danvers, MA). For biospecific immobilization studies, anti-F, or protein A adsorbent, anti-BSA MAb and “‘1-BSA (an amount sufficient to saturate 80% of the active MAb binding sites) were incubated with rotation for 2 hr at room temp and then transferred to the ultrafiltration cell. The contents were pressure filtered over a 1.O pm regenerated cellulose membrane (pre-incubated with unlabeled BSA to minimize nonspecific binding) and washed twice with - 150 adsorbent volumes of a 1 mg/ml solution of BSA in PBA. The wash steps required - 15 set total time. The cell was then refilled with -40 ml of the unlabeled BSA solution. At fixed time intervals thereafter, an -0.5 ml sample was removed, quickly filtered and washed over a syringe filter tip. The filtrate plus wash and the filter tip were then counted for ‘*‘I. Bound and free “‘1-BSA concns prior to the start of the experiment were measured to ensure that equilibrium had been attained. Corrections were made for the small amount of nonspecific “51-BSA adsorption onto the filter tips and beads and to account for the dissociation of the MAb from the anti-f;, or protein A adsorbent. Prior incubation of the regenerated cellulose membranes with unlabeled BSA reduced nonspecific binding of ‘*‘I-BSA to negligible levels. Both thermodynamic and kinetic measurements were carried out at room temp, - 23°C. The dissociation data are poorly fit by a simple exponential function. For each MAb adsorbent, a significant fraction (3-20%) of the “‘1-BSA showed no dissociation over extended time periods. This biphasic character in dissociation data was modeled as follows: bound BSA was assumed to consist of two fractions, (I) one which dissociates according to a first-order process (exponentially) and (2) one which is irreversibly bound. Equation (2) describes the bimodal dissociation kinetics: 6 = (I -.f)exp(

- t/T) +J

(2)

In equation (2) 7 is the dissociation time constant (inverse of the first-order dissociation rate constant), 6 is the undissociated fraction, 1 is time and f is the irreversibly bound fraction. The dissociation data were fit to equation (2) using nonlinear least-squares regression. Runs were carried out in triplicate. To ensure consistency in curve-fitting, data were analyzed for t < 75. Association rate measurements. MAb were immobilized onto BioMag using the CNBr procedure described above. For association measurements, an

reaction

kinetics

131

1251-BSA solution was gently layered onto a settled suspension of BioMag in PBAT (PBA with 0.05% Tween-20, a nonionic detergent) and the reaction was initiated by vortex mixing. At fixed time intervals, samples of suspension were withdrawn and magnetically precipitated. Aliquots of clear solution were removed and supernatant and precipitate tubes were counted for y-activity. The use of PBAT reduced to insignificant levels the competing adsorption of ‘*‘I-BSA to the polypropylene test tube used as the reaction vessel. The suspension was agitated periodically by mild vortex mixing to maintain a uniform suspension. The association rate data were fitted to the integrated form of a simple second-order, reversible rate expression: (B* - Q*)sinh(k, Qt/2) ’ = Z(Q cosh(k, @/2) + B sinh(k, Qt/2))’

(3)

where; B equals (qO+ c,, + l/K), cO is the initial Ag concn, k, is the second-order association rate constant, qOis the initial MAb concn and Q2 equals B2 - 4q,,c,. The affinity and dissociation rates of BioMagimmobiIized MAb were determined using the methods described above. Kinetic and thermodynamic results were corrected for nonspecific adsorption as determined from control assays performed using immobilized mouse IgG. RESULTS

MAbAg dissociation kinetics were investigated using competitive binding methods. A panel of immobilized anti-BSA MAb were (1) incubated with 1251BSA, (2) rapidly washed and (3) contacted with a vast excess of unlabeled BSA. Periodically, samples of the bead suspension were withdrawn and filtered to determine the fraction of bound ‘251--BSA. Dissociation rate parameters and equilibrium association constants were compared for MAb which were both covalently and biospecifically immobilized. To test the validity of the observed dissociation rate parameters, association rate measurements were carried out for two MAb displaying similar values of K but widely differing dissociation rates. Kinetics

ofcooalently immobilized MAb

Table 1 lists measured dissociation time constants (inverse of the first-order dissociation rate constants), measured equilibrium association constants and estimated association rate constants for 12 anti-BSA MAb covalently immobilized to agarose particles. The dissociation time constants vary nearly X&fold, from 2.1 to 410 min. In contrast, these MAb display only a five-fold variation in K, from 2.5 x IO’ to 12 x lo7 M-‘. In addition, there is no apparent correlation between K and the dissociation rate A two order of magnitude variation exists in the estimated association rate constants (calculated from the measured affinity constants and dissociation time

132

WILLIAM Table

1.Kinetic

and thermodynamic

MAb

tsotvne

3.1 2.1 IN12 6.1 2.3 5.1 SC4 9.1 1N9’ lN2h ICI” 3C2h

Yl,K yl,hYI,K )‘l,h. >‘Za. Ic 1’l,h. 72a, Y :‘I.% ;‘I.% Yl,K i’3, K y2b, K

C.

parameters

OLSON

et al.

for covalently

Measured equilibrium constant K (x10-XM 9

I.2i_0.07 I .2 ir 0.04 0.72 + 0.04 0.57*0.17 0.54 i: 0.10 0.46kO.li 0.27 i 0.01 0.25 * 0.02 0.91 0.51 0.34 0.28

immobilized Measured dissociation ttme constant”

anti-BSA

MAb at 23’C Estimated association rate constant

(min) 8.2 It 0.7 9.5 * 0.3 410 _t 38 7.6 * 0.07 30.0 F I .3 5.5 i 0.5 6.9 + 0.6 3.2 ir0.i 27.0 3.3 130 2.1

2.4 2.0 0.029 I .2 0.3 I I .4 0.65 1.3 0.57 2.6 0.045 2.2

“The dissociation time constant is equal to the inverse of the first-order dissociation rate constant. “Kinetic and thermodynamic parameters are each the result of a single measurement. In all other cases. values represent mean values (i SD) of experiments performed in triplicate.

constants) which range from 2.9 x IO’M ‘/see to 2.6 x 10’ M-‘/sec. Association rate measurements were also performed for MAb lN12 and 3.1 immobilized onto nonporous particles. These MAb have similar values of K but widely differing dissociation rates. Figure 1 is a plot of the association rate data for these MAb. The data have been fit to the integrated form of a simple second-order, reversible rate expression [equation (3)]. Reasonable agreement is observed between the experimental data and the theoretical fit. The average association rate constants of MAb IN12 obtained at 50, 100 and 200 nM MAb are, respectively, 5600, 5700 and 6200 M-‘/sec. As predicted, faster association kinetics were observed for MAb 3.1. The average values of k, for MAb 3.1 obtained at 20 and 40 nM MAb are 2.5 x lOsand 2.4 x 10’ M-‘/SIX, respectively, further indicating that k, is independent of reactant concn. Table 2 compares the measured values of k, with the values calculated from equilibrium and dissociation rate measurements. For MAb lN12, excellent agreement is obtained between the average measured value (5800 Mm’/set) and the predicted value (6000 M-‘/see). The averaged measured k, for MAb 3.1 (2.5 x 105&-‘/set) is 40-fold higher than that for MAb IN12 and is in reasonable agreement with the predicted value (1.9 x 10’ M -‘/set). In all, the association rate measurements corroborate the variation and magnitudes of the predicted MAbAg association rates.

bilized MAb (3.1 min vs 3.5 min) are not significantly different. This constancy of dissociation rates suggests that the decreased K of the covaiently immobilized MAb reflects a decreased association rate. Table 3 compares kinetic and the~odyn~mic parameters measured for eight covaiently and biospecifically immobilized anti-BSA MAb. A significant (P = 0.05) decrease in K is observed for each of the MAb; the largest decreases are the five-fold reductions observed for MAb 9.1 and 8C4. With the exception

n 50nM MAb

&flect of’ immobilization method Figures 2 (a) and (b) are plots of equilibrium adsorption isotherms and dissociation rate data, respectively, for biospecificaily and covaiently immobilized MAb 9. I. The average equilibrium association constants are 0.25 x 10’ M ~’ for covalent immobiiization and 1.2 x 10” M ml for biospecific immobilization. This nearly five-fold difference in K is significant at a 99% confidence level. On the other hand, the respective dissociation time constants for covaiently and biospecificaiiy immo-

Fig. 1. 23°C association kinetics of “‘I-BSA to MAb covalently immobilized onto BioMag particles. The curves are fits to the integrated form of a second-order reversible rate law [equation (3)]. (A) MAb lN12. The fitted values of the association rate constant k, are 6500 (O), 5200 (a) and 5800 M-‘/set (A). (B) MAb 3.1. The fitted values of k, are 2.4 x 10s (0) and 2.7 x lO”M-‘/set (w).

Antigen-antibody

reaction

kinetics

133

Table 2. Comparison of observed and predicted 23-C association rate constants of MAb covalently immobilized onto BioMag beads Measured equilibrium constant MAb IN12 3.1

(xl/L4

.-~ ‘)

I .3 + 0.02 0.87 t 0. I2

Measured dissociation time constant no (min) -. - _._,._ 3 360 t 30 4 7.5 + 1.4

n” 3 6

Association rate constant k, (X K-‘M-‘jsec) ~....““.__________ nn predicted measured 5.8 + 1.o 250 t 10

7 6

6.0 190

“n is the number of repetitions.

of this latter MAb, the observed dissociation rates are inde~ndent of the immobili~tion method, Thus when covalently immobilized, seven of the eight MAb examined exhibit a decreased K which appears attributable to a decreased association rate rather than to an increased dissociation rate. For each MAb, the dissociation data are poorly described by a simple exponential. More satisfactory fits were obtained by assuming that a fraction of the ‘25f-BSA becomes irreversibly bound to the beads [equation (2)]. The fittedf-values range from 0.03 to 0.20 and are MAb-specific; i.e. thef-values obtained for the various MAb are significantly different (P = 0.01). Moreover, the f-value appears independent of whether the MAb is biospecifically or covalently bound to the support: thef-values observed for the two immobilization methods are significantly different (P = 0.05) for MAb 3.1 but not for the remaining six MAb. These results suggest that the.f-value is a MAb-specific property that reflects heterogeneity in dissociation rates. Table 4 lists these irreversibly bound fractions or f-values observed for both biospecifically and covalently immobilized MAb. Figure 3 is a plot of the dissociation data for covalently immobilized MAb 9.1 and 8C4. This plot further serves to illustrate the MAb-specific nature of the -f-values. These data have already been corrected for nonspecific adsorption effects.

obviously undesirable in immunoadsorption processes employing com~titive or electrophoretic elution. This may also hold true for chemical elution methods as it is unclear whether chemical elution methods accelerate the initial dissociation of bound Ag. On the other hand, MAb with long dissociation times are advantageous for ELISA and similar assays in which the MAk-Ag equilibrium is disturbed by wash steps. With slowly dissociating MAb longer wash steps can be employed to reduce background noise without

DISCUSSION

Magnitude

of dissociation time constants

Dissociation time constants were measured for 12 immobilized anti-BSA MAb of moderate affinity (K = 0.25 x IO*-1.2 x IO*M-l). Marked variability was observed in dissociation times, which ranged from 2.1 to 410 min. The magnitudes of these values are in rough agreement with previous reports of - 10 min dissociation times for polyclonal (Levison, 1970) and monoclonal (Brimijoin et al., 1985; Roe et al., 1985) antibodies possessing similar equilibrium association constants. Even longer dissociation times (in excess of 700 min) have been reported for the F,, fragments of a high affinity (K - 10’“-lO’i M-‘) MAb to a cell surface Ag (Mason and Williams, 1980). The vacation in dissociation rates among MAb of similar affinity may have an impact on several MAb applications. MAb with long dissociation times are

Fig, 2, 23°C equilibrium adsorption isotherms and dissociation rate data observed for biospecifically and covalently immobilized MAb 9.1. (A) Adsorption isotherms. Open symbols: data for bios~cifically jmmobilized MAb (two assays); closed symbols: data for covalently immobilized MAb (three assays). Curves are nonlinear fits to normalized Langmuir isotherms, y = Kx/(l + KY). Equilibrium association constant for biospecifically immobilized MAb: Kb = 1.2 x 10’ M-‘: equilibrium association constant for covaiently immobilized MAb: K, = 0.25 x 10s M-‘. The difference between K,, and & is significant (P = 0.01). (B) Dissociation rate data. Open symbols: data for biospecifically immobilized MAb (three runs); closed symbols: data for covalently immobilized MAb (three runs). Dissociation time constant for biospecific immobilization: Q, = 3.5 min; Dissociation time constant for covalent immobilization: r, = 3.1 min. The difference between zb and 5, is not statistically significant.

WILLIAM C. OLSON et al.

134 Table

3. Comparison

of kinetic and thernmdynamx properties biospecifically immobilized MAb at 23°C

Equilibrium constant biospecific mmmbiltzatio”” MAb 3.1 2.1 5.1 8C4 9.1 6.1 IN12 2.3

(x 10%

‘)

2.4 i; 0.3 2.0 + 0.01 I.6 &0.2 1.4kO.3 1.2,O.Ol 1.2* 0.09 0.93 -i_0.02 0.90 + 0.1

for covalently

f$Kk

Dissociation time biospecitic immobilization” constant TTh (min)

0.50 0.59 0.29 0.20 0.22 0.50 0.77 0.60

9.3 +0.3 9.5 f 0.4 s-7 * 0.3 18.0 * 5 3.5 2 0.09 8.2 2 0.4 4252 I 29.0 + 5

and

T&,

--

0.89 1.0 0.97 0.3x 0.89 0.94 0.97

1.02

“Data shown represent mean values (+ SD) of experiments performed in triplicate. “The equilibrium association constants measured for covalently immobilized (K,) and biospecifically immobilized (K,,) MAb are significantly different (as determined by analysis of variance, P = 0.05) for all MAb. ‘The dissociation time constants measured for covalent immobilization (TV) and biospecific immobilization (r,,) are significantly different (as determined by analysis of variance, P = 0.05)only for MAb SC4

compromising assay sensitivity. Lastly, dissociation kinetics will likely influence MAb penetration and clearance rates in in viz;o immunoconjugate therapies (Weinstein ef al., 1986). The large variation in dissociation rates for MAb of similar affinity requires that there be a commensurate variation in association rates. This variation was confirmed by association rate measurements carried out for two MAb. Good agreement between measured and predicted k, values was obtained for MAb 1N 12, which possesses the lowest predicted value of k, . Substantially higher association kinetics were observed for MAb 3.1, as expected from equilibrium and dissociation rate measur~m~nts. These results also confirm that the equilibrium and dissociation rate measurements yield meaningful estimates of association rate parameters. The association rate constants estimated for the anti-ISA MAb vary from 0.029 to 2.4 x IO’ M ‘/sec. These values are well below the diffusional limit of 1Ox-IO9 M ‘jsec expected for macromolecular species in the absence of any reaction potential. Thus both the magnitude and large variation of association par-

ameters indicate that only a small fraction of MAbAg collisions successfully lead to complex formation. The estimated and measured association rate constants reported here are somewhat lower than the 105P106 M-‘/see values typically observed for soluble antibody (Noble et al., 1969, 1972; Levison, 1970: Mason and Williams, 1980; Icenogle et al., 1983; Roe et al., 1985; Brimijoin et al., 1985). Moreover, to our knowledge MAb lN12 has the slowest association rate observed to date. As discussed below, these low association rates can be partly attributed to covalent immobilization of the MAb. E#ect o~imm~bili~ation d~namie parameters

method on kinetic and therm~j -

Dissociation rates and equilibrium association constants were determined for MAb immobilized onto agarose particles via two methods: (1) CNBr chemistry, which leads to a random, multipoint, covalent attachment and (2) biospecific ligands which bind to the MAb F, region. CNBr chemistry was selected as a prototypical covalent immobilization method, while biospecific immobilization offers a site-directed method

Table 4. Percent of “‘I-BSA modeled as being irreversibly bound to covalently biosuecifically immobilized MAb Percent

MAb

rsotype

8C4 3.1 6.1 2.1 5.1 2.3 9.1 1N2” 3C2 I N9”

72a, K l’1.K yl,h. l’l,fi L.1.h. y2a, K yl,h. j’l,K ;12b, x i’ I . x

Covalent

of

‘%BSA

immobilization 19i_0.9 11 il 8.7 i_ 0.5 6.7 + I 6.4 * I 5.3 i_ I 3.3 i: 0.9 8.2 6.2 3.2

irreversibly

and

bound, .f x 100

Biospecific

immobilization

20 -i_2 7.7 + 0.7 7.1 i: I 6.3 i_ 0.2 xi_4 5.2 _t I 3.0 ?r 0.5 ND ND ND

ND = not determined “Results based on a single measurement. In all other cases, data shown represent mea” values (+ SD) of experiments performed in triplicate. The differences between percentages obtained for covafently and bios~ci~cally immobilized MAb are significant only for MAb 3.1 c&sdetermined by analysis of variance, P = 0.05). The differences among MAb are significant (P = 0.01) for both covalent and biospecific immobilization.

Antigen-antibody

0

20

40

60

lwIfvurEs Fig. 3. Dissociation rate data for covalently immobilized MAb 9.1 and SC4 at 23”C, demonstrating MAb-specific differences in the irreversibly bound Ag fractions. Data were fit to the expression y =(I -f)exp(-x/r)+5 r is the dissociation time constant and f is the irreversibly bound Ag fraction. Upper curve: MAb 8C4, r = 6.9 min, f = 0.20; lower curve: MAb 9.1, r = 3.2 min,f= 0.03.

that leaves the MAb binding sites comparatively unhindered. For seven of eight MAb observed dissociation rate parameters were independent of immobilization method. In addition, for the two MAb examined, dissociation rates were independent of the solid phase used (Superose or BioMag). These observations are consistent with the view that, regardless of the method or environment of immobilization, the dissociation event is governed primarily by the detailed interactions between the residues of the MAb binding site and Ag epitope. These observations might also indicate that the dissociation rates of immobilized MAb are similar to the dissociation rates of MAb in solution. For all eight MAb examined, the values of K measured for covalently immobilized MAb are significantly lower (by as much as five-fold) than those for biospecifically immobilized MAb. This decrease in K appears attributable to a decreased association rate, as dissociation rates are largely unaffected. Covalent immobilization methods could reduce MAb association rates and K via a variety of mechanisms, such as unfavorable orientation of the MAb binding site with respect to the solid support. The use of a single value for K for covalently immobilized MAb is clearly an idealization. Rather than reflecting values of K which have been discretely altered by immobilization, these parameters probably represent averages of a spectrum of binding properties. Evidence for this heterogeneity may be reflected in the association rate data of Fig. 3, which cannot be effectively modeled by a single association rate constant. For both biospecifically and covalently immobilized MAb, the dissociation of “‘1-BSA also showed a departure from first-order kinetics. This deviation takes the form of a slowly dissociating Ag fraction [or

reaction

kinetics

13.5

f-value, cf. equation (2)] that was modeled as being irreversibly bound. The nature of this slowly dissociating material is not known. We have, however, made the following observations: (1) the f-value appears MAb-specific and subclass-independent; (2) the amount of ‘251-BSA which slowly dissociates is substantially greater than that which binds to mouse IgG-Superose; (3) the f-value appears largely independent of the immobilization method; (4) the MAb and ‘251-BSA preparations were free of oligomeric species as determined by HPLC analysis; (5) similar behaviour has been observed in the electrophoretic elution of ‘251-BSA to those MAb immobilized onto Sepharose CL-2B (Olson and Yarmush, 1987); (6) the MAb used herein appear to bind monogamously to the BSA molecule (Morel et al., 1988) and thus the f-values are unlikely to reflect BSA, which is bound to two MAb binding sites (either on the same or distinct MAb molecules); (7) the slowly dissociating material does appear to be noncovalently bound protein: this material is completely eluted by contact with 10% sodium dodecyl sulfate, retained over a 12,00&14,000 mol. wt dialysis membrane and subsequently 99 + % precipitated by 10% trichloroacetic acid. Finally, biphasic kinetics have been previously observed for the dissociation of monoclonal ‘251-F,h fragments from cell surface Ag (Mason and Williams, 1980).

CONCLUSIONS

Intrinsic dissociation rates and equilibrium association constants were measured for 12 anti-BSA MAb immobilized onto Superose (agarose) particles. Dissociation rate constants varied nearly 200-fold, from 2.1 to 410 min, whereas equilibrium association constants varied only five-fold, from 0.25 x lo8 to 1.2 x lOa Mm’. Thus for this system, K is not a useful predictor of reaction kinetics and the variation in kinetics may have an important impact on several applications of immunotechnology. Predicted differences in MAb association rate constants were experimentally confirmed via association rate measurements carried out for two MAb with fast and slow association kinetics. Dissociation and affinity parameters were also determined for MAb biospecifically immobilized via their F, regions. For all eight MAb the equilibrium association constant for covalent immobilization was significantly lower than that for biospecific immobilization. The lower K of covalently immobilized MAb appears attributable to a decreased association rate rather than to an increased dissociation rate.

Acknowledgements-The authors kindly thank Barbara J. Hanson for excellent technical assistance. This work was supported in part by an NSF grant under the Engineering Research Center Initiative Agreement CDR-850003 and NIH grant CA-45272. M. L. Yarmush is a Lucille P. Markey Scholar in Biomedical Science.

WILLIAM C. (

136 REFERENCES

Benjamin D. C., Wilson L. D., Riley R. L., Holaday B. and DeCourcy K. (1985) Antigenic structure of albumin. In Immune Rec#g~~tio~ of Protein Antigens (Edited by Laver W. G. and Air G. M.), pp. 98-104. Cold Spring Harbor Laboratory Press, New York. Brimijoin S., Mintz K. P. and Prendergast F. G. (1985) An inhibitory monoclonal antibody to rabbit brain acetylcholinesterase. Molec. Pharmac. 28, 530-545. Chase H. A. (1984) Prediction of the performance of preparative affinity chromatography. 1. Chromnt. 297, 179.-202. lcenogle J., Shiwen H., Duke G., Gilbert S., Rueckert R. and Anderegg J. (1983) Neutralization of poliovirus by a monoclond antibody: kinetics and stoichiometry. Yirolog~ 127, 412425. Kranz D. M. and Voss E. W. (1981) Partial elucidation of an anti-hapten repertoire in BALB/c mice: comparative characterization of several monoclonal anti-fluorescylantibodies. Molec. inunlcn. 18, 889-898. Levison S. A., Kierszenbaum F. and Dandliker W. B. (1970) Salt effects on antigen--antibody kinetics. Biochemistry 9, 322-331. March S. C., Parikh 1. and Cuatrecasas P. (1974) A simplified method for cyanogen bromide activation of agarose for affinity chromatography. Analyt. Biochem 60, 149.. 152. Mason D. and Williams A. (1980) The kinetics of antibody binding to membrane antigens in solution and at the cell surface. Biochem. J. 187, l-20. McFarlane A. S. (1958) Efficient trace-labelling of proteins with iodine. Nature 182, 53. Morel G. A,, Yarmush D. M., Colton C. K.. Benjamin D. C. and Yarmush M. L. (1988) Monoclonal antibodies

to bovine serum albumin: affinity and specificity determinations. M&c. Immun. 22, 7-15. Noble R. W., Reichlin M. and Gibson Q. H. (1969) The reactions of antibodies with hemoprotein antigens. J. hioi Chem. 244, 2403.--24 I I_ Noble R. W., Reichlin M. and Schreiber R. D. (1972) Studies on antibodies directed toward single antigenic sites on globular proteins. Biochemistry 11, 3326-3332. Olson W. C. and Yarmush M. L. (1987) Electrophoretic elution from monoclonal antibody immunoadsorbents: a theoretical and experimental investigation of controlling parameters. Bio&m. Prog. 3, l77- 188. Pecht I. (1982) Dynamic aspects of antibody function. The Antigens 6, I--68. Roe R., Robbins R. A., Laxton R. R. and Baldwin R. W. (1985) Kinetics of divalent monoclonal antibody binding to tumour cell surface antigens using flow cytometry: standardization and mathematical analysis. Molec. Immun. 22, 11 21. Sachs D. H., Schechter A. N.. Eastlake A. and Anfinsen C. B. (1972) Inactivation of staphytococcai nuclease by the binding of antibodies to a distinct antigenic determinant. Biochemistry 11, 42684213. Sportsman J. R., Liddll J. D. and Wilson G. S. (19X3) Kinetic and equilibrium studies of insulin immunoatfinity chromatography. Ana/,vr. Chem. 55, 771. 775. Weinstein J. N., ChristoDher D. V., Black J. B., Rencc R. E., Parker R. J.. Holion 0. D., Mulshine J. L.. Keenan A. M., Larson S. M., Carrasquiiio J. A., Sieber S. M. and Cove11 D. G. (1986) Selected issues in the pharmacology of monoclonal antibodies. In Sitc~-.Sp~rific Drug De/it~ev>~ (Edited by Tomlinson E. and Davis S. S.), pp. XI 91 John Wiley & Sons, New York.