Biochimtca et Biophysica Acta, 1! 19 (1992) 27-34 © 1992 Elsevier Science Publishers B,V. All rights reserved 0167-4838/92/$05.00
27
BBAPRO 34088
Relative binding properties of fluorescein and 9-hydroxyphenylfluoron (HPF) with murine monoclonal anti-fluorescein antibodies * William D. Bedzyk, Cathy A. Swindlehurst and Edward W. Voss, Jr. Department of Microbiolo~" Unicersity of Illinois at Urbana-Champaign, Urbana, IL (U. S.A. ) (Received 5 June 1991) (Revised manuscript received 3 September 1991 )
Key words: Antibody active site: Conformational substatc: Fluorescence lifetime
Contribution of the fluorescein (FI) carboxyl group to hapten binding by idiotypically related murine monoclonal anti-Fl antibodies 4-4-20, 9-40 and 12-40 was studied by comparing relative liganded active site properties with bound FI or 9-hydroxyphenylfluoron (HPF). Kinetic studies revealed similar association rate constants between Fi and HPF to 4-4-20 ( = 1.1 • 107 M - t s - t ) ; however, the 4-4-20 dissociation rate for F! was = 200 times slower, relative to HPF, which resulted in relative intrinsic affinity values of 1.2- 10 TM and 6.5 • 107 M - t, respectively. Mabs 9-40 and 12-40 also displayed a reduced affinity for HPF and affinity constants of 5.5- l0 s M -t and 6.7. l0 s M were obtained from a competitive ELISA. Additionally, previous studies revealed that upon binding FI, Mabs 4.4-20 (92.1%), 9-40 (44.7%) and 12-40 (73.4%) quenched F! fluorescence. Similar analyses with HPF resulted in 64.4% and 2.0% fluorescence quenching by 4-4-20 and 12-40, respectively; however, 9-40 increased HPF fluorescence by = 24%. Steady-state fluorescence polarization experiments revealed that in solution, F! ( P = 0.019) and HPF ( P = 0.048) were polarized to different degrees. When bound, however, F! and HPF expressed similar polarization values ( P --- 0.455), except 9-40 bound HPF which was significantly depolarized ( P = 0.428). Fluorescence lifetime experiments revealed F! to possess two discrete lifetimes: a 3.96 ns component (free FI) and either a 0.52 ns (4-4-20), 2.23 ns (9-40) or 0.96 ns (12-40) short component that corresponded to bound FI. HPF, however, when bound by 4-4-20 or 9-40, was best fit by three discrete exponentials: a relatively long 4.0 ns component, a 1.11 ns lifetime (free HPF) and either a 0.52 ns (4-4-20) or 2.23 ns (9-40) component. Finally, HPF bound by Mab 12-40 exhibited a single lorenzian distributed lifetime of 1.36 ns ( _ 0.43 ns). Results are discussed in terms of Mab active site structure and conformational state dynamics.
Introduction Although genetic elements and mechanisms responsible for antibody active site diversity have been de-
* This work was supported in part by grant No. 2-3-013 from the Biotechnology Research and Development Corporation (Peoria, ILk Abbreviations: CDR, complementarity-determining region: Fl, fluorescein: H. immunoglobulin heavy chain: HPF, 9-hydroxyphenyifluoton: L, immunoglobulin light chain: Mab, monoclonal antibody: TMB, 3,3',5,5'-tetramethylbenzidine: V, immunoglobulin variable region. Correspondence: E.W. Voss, Department of Microbiology, University of Illinois at Urbana-Champaign, 407 S. Goodwin Avenue, Urbanao IL 61801, U.S.A.
scribed in detail [1,2], at the protein level, mechanisms involved in antibody-antigen complementarity remain relatively unresolved. Comparison of known variable region primary structures [3] with accumulated results from X-ray crystallographic studies (reviewed in Refs. 4 and 5) have suggested certain active site residues as being generally important in antibody recognition of a variety of antigens. These apparently critical side chains include Arg (which has a planar nature and delocalized charge) and Tyr and Trp (which have the capacity to hydrogen-bond and form hydrophobic and electrostatic interactions with antigen). Of particular interest has been the interaction between residue L32Tyr with both protein and haptenic antigens. For example, while L32Tyr participated in either a hydrophobic or hydrogen-bond interaction with hen egg lysozyme (Mab DI.3)
28 [6] and 2-phenyloxazolone (Mab NQ10/12.5) [7], respectively, L32Tyr interacted with the C helix of myohemerythrin (Mab B1312) [8] and fluorescein (Mab 4-4-20) via both mechanisms [9]. To specifically investigate the hydrogen-bond contributions of residue L32Tyr to overall ligand (Fl) binding, therefore, a structural analog of Fl has been employed. Historically, studies of interactions between anti-F! antibodies and haptenic structural analogs have proven difficult because most Fl derivatives (such as eosin, erythrosin and members of the rhodamine series) have shown minimal crossreactivity [10,11]. The three-dimensional co-crystal structure of Mab 4-4-20 and Fl revealed that the xanthenone moiety fit tightly into a pocket-like binding site [9] consistent with the lack of accommodation for these larger F! analogs. Additionally, although Mab 4-4-20 Fl contact residues were determined to be L27dHis, L32Tyr, L34Arg, LglSer, L96Trp and H33Trp [9], individual contributions of each residue to Fl-binding remained uncharacterized. Initially, in vitro heavy and light chain reassociations and site-directed mutagenesis experiments with the 4-4-20 single-chain derivative demonstrated L34Arg was responsible for increased 4-4-20 affinity and the L32Tyr hydrogen-bond with F! contributed -- 100-fold to the overall binding constant [12,13]. Based on these observations, a smaller Fi analog, 9-hydroxyphenylfluoron (HPF), was chosen as the cross-reactive ligand in these studies to test certain hypotheses. HPF differed from F! by absence of the carbox3'i group (Fig. 1) and thus, provided direct information on the L32Tyr hydrogen-bond contribution to Fl-binding by anti-F! Mabs. HPF was first synthesized by Martin and Lindqvist [14] to study pH dependency of Fl fluorescence and was later studied bound by rabbit polyclonal anti-Fl antibodies [15]. This study utilized HPF to examine the nature and extent of interactions present in a homogeneous binding site population. Finally, the use of HPF facilitated binding studies that probed the capacity and rate at which an antibody active site specific for the relatively rigid FI ligand stabilized the HPF phenyl ring (which possessed significant rotational freedom in relation to the xanthenone group). In this study, F! and HPF were both the ligand and an extrinsic fluorescent probe. Thus, a comparison of F! and HPF binding by a homogeneous antibody active site population permitted assessment of ring stabilization. Furthermore, since the fluorescence properties of each ligand correlated with the active site environment and xanthenone and phenyl ring orientation, fluorescence quenching, polarization, and timeresolved lifetime measurements were available. This study, therefore, also permitted preliminary analyses of antibody active site conformational state dynamics by determining relative Fl and HPF motion within each active site.
Materials and Methods
Reagents Fiuorescein disodium salt (Fi) and fluorescein-5amine (FI-NH 2) were purchased from Research Organics (Cleveland, OH). FI analog 9-hydroxyphenyifluoron (HPF) was kindly provided by Dr. W. Ware (The University of Western Ontario). Structural comparisons of F! and HPF are shown in Fig. 1, and HPF spectral properties have been described [14,15]. Tris (Ultra Pure) was obtained from Sigma (St. Louis, MO). Immulon II polystyrene wells, horseradish peroxidaselabeled Protein G and TMB (3,3',5,5'-tetramethylbenzidine) were procured from Dynatech Laboratories (Chantilly, VA), Zymed Laboratories (South San Francisco, CA) and Kirkegaard and Perry Laboratories (Gaithe~sburg, MD), respectively. Hybridoma cell lines and monoclonal antibodies Hybridoma cell lines were obtained by chemical fusion [16] of non-secreting murine myeloma cell line Sp2/0-Agl4 with hyperimmunized splenocytes from BALB/cV mice. Mabs 4-4-20 (IgG2a, r; K a = 1.7- 10~° M-l), 9-40 (IgGl, K; Ka=3.3" 10 7 M - I ) and 12-40 (IgGl, K; K~ = 1.0- 108 M - l ) F!-binding and idiotypic characteristics [11,17,18] and primary structures [19] were previously reported. Hybridoma proteins 4-4-20, 9-40 and 12-40 were produced as ascitic fluids in BALB/cV mice. Immunoglobulin purification was accomplished by dextran sulfate precipitation of lipoproteins followed by ammonium sulfate precipitation and gamma globulin fraction enrichment [20]. Subsequently, Mabs were affinity purified with FI-Sepharose and eluted with 0.23 M glycine pH 2.6 as described [211. Competitive ELISA A competitive ELISA was utilized to approximate HPF affinity constants for Mabs 4-4-20, 9-40 and 12-40. Microtiter wells were initially coated with 50 /zl (100 ng/ml) Fll2.sBSA in 50 mM Tris, 150 mM NaCl (pH
HO0" ~ 0
HO"~ °
d A
B
Fig. I. F! and HPF structural comparison. Fl (panel A) and HPF (panel B) contain identical immunodominant xanthenone moieties but differ by the presence (Fi) or absence (HPF) of a phenyl ring carboxyi group. At neutral pH or greater, both dianionic Fi and anionic HPF possess similar absorption maxima and molar extinction coefficients but different fluorescence quantum yields ( ~ = 0.92 for FI and 0.21 for HPF) [14].
29 8.0) (TBS) and incubated for 2 h at 37 ° C. Wells were washed 2 ×with TEBT (TBS plus 1% BSA, 1 mM E D T A and 0.05% Tween-20) and excess protein binding sites masked with TEBT for 30 min at 37 ° C. 50 #! of 4-4-20 (100 ng/ml), or 9-40 or 12-40 (1 /.tg/mi), pre-incubated for 30 min at 4 ° C with various FI or H P F concentrations, was added to the wells and incubated for 4 h at 4 ° C . To detect solid-phase bound antibody molecules, horseradish peroxidase-labeled protein G was added and incubated for 2 h at 37 o C. Finally, TMB substrate was added (50 ~tl) and enzyme reactions ultimately terminated with 1 M H 2SO4 . Optical density measurements employed a Dynatech MR600 automatic 96-wel plate reader equipped with a 450 nm cut-off filter. Fluorescence instrumentation Association rate constants and fluorescence polarization analyses were performed with a photon counting spectrophotometer (Gregg-P.C., ISS Instruments, Champaign, IL) and utilized a xenon arc lamp (excitation at 493 nm and emission at 530 nm). Dissociation rate constants and amax values were determined as described [22,23] and utilized an Aminco-Bowman spectrophotofluorometer equipped with a xenon arc lamp (excitation at 493 nm and emission at 530 nm). Fluorescence lifetime measurements were performed on a multifrequency phase and modulation fluorometer [24]. Samples were excited by the 488 nm line of an argon ion laser passed through an acousto-optic modulator [25] and a polarization filter set at 35 °. A cut-on filter selected emission wavelengths greater than 520 nm and all measurements were taken using glycogen as a reference solution. Phase and modulation data were collected at 8 to i2 frequencies ranging from 10 to 210 MHz. The sample chamber was thermostat-controlled with a circulating water bath at 20 ° C. Association rate analysis Association rate constants for FI and H P F binding by Mab 4-4-20 were determined by monitoring decreases in fluorescence intensity as a function of time [22]. Equal volumes of 4-4-20 and ligand were mixed with a manual stopped-flow apparatus (Hi-Tech Scientific model SFA-11 Rapid Kinetics Assembly) thermostated at 2°C. A 10-fold molar excess of 4-4-20 (27.0 nM), compared to ligand (2.0 nM), was used to ensure first-order kinetics. Fluorescence intensity was measured as a function of time using ISS kinetics software. Technical rate constants (Ltech) were determined from the slope of the fraction free ligand ( L / L o) plotted as a function of time (t). The fraction free ligand was calculated from the equation: F(t) - F(eq) L / L o = In
F(0) - F ( e q )
(1)
where F(t) was the fluorescence intensity at time t, F(eq) the equilibrium fluorescence intensity, and F(0) the initial fluorescence intensity. A bimolecular association constant (k t) was calculated from the equation: Ft.,(eq) • Ktc,:h kl = [Sites]
(2)
where Fb(e q) was the fraction of bound ligand at equilibrium, [Sites] the active site concentration and Ktech the slope of L / L o vs. t. Experiments were repeated at least in triplicate. Dissociation rate analysis F! and HPF dissociation rates (k 2) were determined by initially monitoring increased fluorescence as a function of time after addition of a 10-fold molar excess of FI-NH 2 [22]. Fluorescence intensities were measured and analyzed using OffRate acquisition and analysis software (Interactive Software, Urbana, IL). An affinity constant (K a) was subsequently calculated from the formula: K.=kl/k
z.
(3)
Fluorescence lifetime measurements Stock FI and HPF solutions, 2.0 and 4.0.10 -8 M, respectively, were prepared in 50 mM Tris (pH 8.0) at 20 o C. Following fluorescence lifetime determination for free Fi or HPF in solution, affinity purified antibody (1 nM active sites) was added and phase and modulation data acquired. Analogous titrations were repeated until a 100:1 molar ratio (active sites:ligand) was achieved. For each frequency, fluorescence lifetime measurements were taken until phase angle error was less than or equal to 0.200 and modulation ratio error was less than or equal to 0.004. Multifrequency phase and modulation data analysis was performed on a personal computer and used Globals Unlimited software [26]. Data were analyzed using several different fitting routines and the best fit was obtained for either discrete exponentials or lorenzian distributed fluorescence lifetimes. Results
Association rate determination Kinetic studies were performed to characterize the interaction of 4-4-20 with Fl and HPF. Ligand association rates with 4-4-20 were determined to be 1.01 + 0.16- 107 M - l S-I and 1.21 _+ 0.55" 10 / M - I s - i f and HPF, respectively (data not shown). The association rate constant between 4-4-20 and Fl was approximately twice that previously reported [10]. This difference was probably due to differences in both experi-
30 mental conditions and instrumentation. For example, association rate constant determination utilized a thermostated stopped-flow apparatus with computerized data acquisition from a photon counting spectrophotometer. Previously reported values, however, were determined by manual ligand injection into a thermostated cuvette in a spectrophotofluorometer connected to a strip chart recorder. Nevertheless, association rate constants for FI and H P F with 4-4-20 were identical within error. Ligand affinity constants
Relative dissociation rate constants for 4-4-20 with the two ligands were measured to determine affinity constants. The dissociation rate was 4.5 + 0.8-10 -4 s - t for F1 and 7.8 + 3.0' 10 -2 s - i for HPF. Affinity constants, therefore, were determined to be 1.2-10 ~° M - ! and 6.5.10 7 M - l for Fi and H P F (Table I), respectively. Similarly, Mab 9-40 possessed an affinity for FI of 3.3- 10 r M - ! while the affinity of 12-40 was 1.0- 10 s M-~. Due to assay limitations, an off-rate for 9-40 and 12-40 bound H P F was too rapid for detection. This indicated that the affinities of 9-40 and 12-40 for H P F were < 5 . 0 . 1 0 6 M - z To determine an approximate 9-40 and 12-40 H P F affinity, a competitive ELISA, where free FI and H P F were soluble inhibitors of 9-40 or 12-40 binding to solid-phase FI~2.sBSA , was employed (Fig. 2). A 62.5fold concentration difference was observed between 50% inhibition by FI and H P F and indicated 9-40 possessed an affinity of - 5 . 5 . 1 0 5 M -~ for HPF. Similarly, 12-40 possessed an affinity for H P F of = 6.7 • 105 M - ~. Validity of thes~ estimations was verified by inhibition of 4-4-20 binding to solid-phase FI under identical conditions. A 200-fold concentration difference required for 50% inhibition by FI and H P F (Fig. 2) was consistent with the 200-fold difference in affinity as determined by off-rate analysis. Finally, a relatively smaller FI analog.~ resorufin, was an ineffective inhibitor of 4-4-20, 9-40 and 12-40 binding to solid-phase FI (data not shown').
A
"
v
C.
~p-
:~ o~ 80
60 40
20
| 10
9
8
7
6
5
4
10
9
8
7
.
,
,
6
5
4
1 0 9 8 7 6 5 4
-log [inhibitor] (M)
Fig. 2. Competitive ELISA inhibition curves. The ability of various soluble F! (closed symbols) and HPF (open symbols)concentrations to inhibit 4-4-20 (panel A), 9-40 (panel B) and 12-40 (panel C) binding to solid phase F!n2.5BSA. Following incubations and washes, solid-phase bound antibody molecules were detected with horseradish peroxidase-labeled Protein G and TMB substrate. Differences between Fi and HPF concentrations required for 50% inhibition were used to estimate HPF affinity constants for Mabs 9-40 and 12-40.
Fluorescence measurements
Initially, Mab bound Fl and H P F Qma~ values were obtained to determine if the L32Tyr hydrogen-bond with Fl contributed to this observed phenomenon. As shown in Table I, Fl Qmax values for 4-4-20, 12-40 and 9-40 were identical to previously reported values [12,18]. H P F fluorescence, however, was quenched to a lesser extent by 4-4-20 (64.4%) and 12-40 (2.0%) (Table I). Under identical conditions, surprisingly, 9-40 bound H P F exhibited an increase of ~ 24% in total fluorescence units (Table I). To resolve this apparent 9-40 bound H P F anomaly and to investigate potential explanations for reduced 4-4-20 and 12-40 H P F Qmax values, free and Mab bound Fl and H P F fluorescence lifetimes were determined (Table II). Following phase and modulation data acquisition for free ligand in solution, an aliquot of affinity purified Mab was added and phase and modulation data obtained. Analogous titrations were performed until the molar active site concentration was 100-fold greater than the molar ligand concentration. When free in solution, Fl and H P F possessed single discrete 3.96 ns and 1.11 ns lifetimes, respectively.
TABLE I Ligand-binding characteristics
Mab
FI Ka
AG
4-4-20 12-40 9-40
(M-t) 1.2.10 t° 1.0.10s 3.3.107
(kcal/mol) "~ - 12.6 - 10.0 -9.3
HPF o
Qmaxb
Ka
AG
92.1 +_0.1% 73.4+_ 1.3% 44.7+_3.8%
(M- x) 6.5-107 6.7.105 c 5.5" 105 c
(kcal/mol) ~ - - 9.7 -7.3 c -7.2 c
o
Qmax
64.4+ 0.4% 2.0+_0.5% -23.9+_0.3% d
" Binding free energy derived from AG o = _ R T in K a. b Fluorescence of antibody bound relative to free iigand at protein saturation. Measurements were performed at 20 °C and each value represents the mean of triplicate trials + standard deviation. c Approximated from compet;,~ave ELISA. d A negative value indicates fluorescence enhancement.
31 TABLE II
Fluorescence polarization and lifetime data for free and Mab bound FlandHPF Sample
Temperature
Polarization a
Tt
~'2
~'3
~c2 h
FI 4-4-20/Fi 12-40/F! 9-40/F1 HPF 4-4-20/HPF 12-40/HPF 9-40/HPF
20 20 20 20 20 20 20 20
0.019 _+ 0.002 0.458 _-_+0.001 0.453 _+0.000 0.456 _+0.001 0.048_+ 0.001 0.455 -1-0.001 0.449 _+0.003 0.428 +_ 0.002
3.96 c 3.96 3.97 3.96 1.11 1.04 1.36( _+0.43) d 1.04
0.52 0.963 2.23 0.52 _ 2.23
3.90 4.14
0.849 1. i 56 I. 136 0.524 0.674 0.177 0.933 0.344
a b c d
oC oC oC oC °C oC °C oC
Steady-state polarization value. C h i - s q u a r e f o r d a t a fitting r o u t i n e . L i f e t i m e v a l u e s a r e in n a n o s e c o n d s . N u m b e r in p a r a n t h e s e s r e p r e s e n t s the w i d t h f o r a I o r e n z i a n d i s t r i b u t e d c o m p o n e n t .
Since titrations were performed, most Mab bound measurements contained a percentage of this discrete lifetime component which corresponded to free ligand. F1 titrated separately with 4-4-20 (0.52 ns), 12-40 (0.96 ns) and 9-40 (2.23 ns) resulted in one additional shorter discrete lifetime (Table ll). HPF titrated separately with 4-4-20 and 9-40 resulted in two discrete lifetime components (in addition to free HPF-I.ll ns). Besides a shorter exponential (0.52 ns for 4-4-20 and 2.23 ns for 9-40), both 4-4-20 and 9-40 bound HPF displayed a discrete 3.9 to 4.1 ns component (Table II). HPF bound by 12-40, however, exhibited a single lorenzian distributed fluorescence lifetime with a center at i.36 ns and a width of _+0.043 ns. Finally, X2 values for each phase and modulation data fitting routine are also presented in Table II and were less than 1.16. To examine the extent of FI and HPF flexibility when bound by Mabs 4-4-20, 9-40 and 12-40, steadystate fluorescence polarization values were obtained. Free FI and HPF were depolarized as expected (P = 0.019 and P -- 0.048, respectively) (Table II). Both 4-420 and 12-40 bound Fl and HPF were polarized to similar degrees (P --- 0.453). Fl bound by Mab 9-40 was also polarized to this extent (P = 0.456); however, HPF bound by 9-40 was significantly depolarized and had a value of 0.428 (Table II). Discussion
Since accumulated evidence (i.e. V region primary and quaternary structure comparisons) has suggested that antibody active site tryptophan and tyrosine residues are generally important in antigen complementarity and binding [5,7], the contribution of Fl contact residue L32Tyr [9] to ligand binding was initially investigated in three idiotypically related anti-F!
Mabs (4-4-20, 9-40 and 12-40). To accomplish this, comparative binding assays and fluorescence polarization and lifetime measurements with free and Mab bound Fl and HPF (Fig. 1)were performed. Although HPF was a more flexible ligand (see below), due in part to lack of a stabilizing phenyl ring carboxyl group, association rate constants between 4-4-20 and either Fl or HPF were identical within error (data not shown). Ligand flexibility and hydrogen-bond formation, therefore, apparently did not effect ligand association with the active site. Fl and HPF dissociation rates from 4-4-20 active sites yielded binding free energies of -12.6 kcal/mol (Fl) and - 9 . 7 keal/mol (HPF) (Table I). These values were similar to values obtained by 4-4-20 active site L32Tyr to L32Phe site-directed mutagenesis studies [13]. Mabs 12-40 and 9-40 HPF binding free energies were estimated by a competitive ELISA to be - 7 . 3 kcal/mol and - 7 . 2 kcal/mol, respectively. Differences in Fl and HPF binding free energies (AAG °) for 4-4-20, 12-40 and 9-40 were 2.9, 2.7 and 2.2 kcal/mol, respectively. In percentage terms of the total binding free energy, this hydrogen-bond was responsible for 27.0% (12-40), 23.0% (4-4-20) and 23.4% (9-40) of the kcal/moles in active site binding. Although the Fl carboxylate group contribution to overall 4-4-20, 9-40 and 12-40 Fl-binding (2.0-3.0 kcal/mol) was less than previously reported values (= 4.0 kcal/mol) for additional anti-F! Mabs [27], all reported values were consistent with contributional differences associated with a hydrogen-bond [9] compared to a potential salt bridge [27]. In either case, the L32Tyr hydrogen-bond with Fl contributed differently to the overall binding free energy of an individual Mab. This was consistent ~ith V region primary structure comparisons of idiotypically related anti-Fl Mabs [19] which suggested potentially subtle active site micro-environment differences (e.g.,
32 the degree of solvent exposure of the L32Tyr hydrogen-bond) due, in part, to H CDR3 residue differences. Time-resolved fluorescence lifetime measurements of free and Mab bound F! and HPF were obtained to investigate each anti-Fl active site micro-environment (Table II). Identical discrete fluorescence lifetime components (0.52 ns) were observed for both FI and HPF when bound by 4-4-20. This suggested that the two ligands adopted similar conformations in the active site. Similar results were observed for 9-40 bound FI and HPF (2.23 ns). Finally, 12-40 bound Fl and HPF probably expressed identical fluorescence lifetimes; however, 0.96 ns (bound) and 1.11 ns (free HPF) were undiscernible and resulted in a single lorenzian distributed lifetime. From these results, several mechanisms of Fl-binding by these anti-FI Mabs were elucidated. The L32Tyr hydrogen-bond with FI, although significant in binding affinity determination, was not required for the F! fluorescence quenching phenomenon. This was consistent with previous studies that investigated the pH dependency of F! fluorescence [14] and showed that the carboxyl group ionization state did not influence the quantum yield of Fl. The enolic oxygen ionization state and the extent and nature of aromatic stacking between the Fl xanthenone group and aromatic active site residues, therefore, determined maximum fluorescence quenching. Additionally, 9-40 bound HPF fluorescence enhancement (2.23 ns component) was due to relative HPF flexibility and identical active site interactions with both Fl and HPF. Under steady-state conditions at 20°C, fluorescence quantum yields for Fl (~--0.92 or ~-= 3.96 ns) and HPF (q~ = 0.21 or ~-- 1.11 ns) differed, in part, due to the phenyl ring orientation relative to the xanthenone group. The Fl carboxyl group fixed the phenyl ring at an approx. 62 ° angle relative to the xanthenone; however, since HPF lacked this carboxyl group, the phenyl ring possessed greater flexibility. The molecular configuration of other bi-aromatic chromophores has been previously shown to dictate the absorption and fluorescence profiles of the fluorophore [28,29]. This phenomenon was explained by the 1;-electron orientations (e.g., parallel aromatic groups in the same fluorophore possessed aligned w-electrons and higher fluorescence quantum yields and perpendicular aromatic groups possessed unaligned w-electrons and lower quantum yields). With HPF, therefore, a steady-state quantum yield ( ~ = 0.21 or ~-= 1.11 ns) represented an average of rapidly interconverting higher and lower quantum yield species dependent on xanthenone ring and phenyl ring orientation. This interconversion rate, however, must be accomplished rapidly since the association rate constants between Mab 4-4-20 and either F! and HPF were identical. The HPF molecular configuration when being bound by 9-40 (or 4-4-20 or 12-40), therefore,
was similar to Fi and possessed a lifetime near the solution phase lifetime for Fi (~-3.96 ns). In fact, a 3.96 ns fluorescence lifetime was observed (see below). In addition, a shorter HPF lifetime, comparable to quenched Fi (i.e. 2.23 ns), was observed. On a steadystate basis, therefore, the 9-40 bound HPF 2.23 ns fluorescence lifetime was greater than free HPF (1.11 ns) and fluorescence enhancement was observed. Besides a fluorescence lifetime that corresponded to free ligand, 4-4-20 and 9-40 bound HPF displayed a relatively long lifetime species (1-= 4.0 ns). A similar discrete component was observed when HPF was bound by rabbit polyclonal anti-Fl antibodies [15]. Similar to these polyclonal antibody studies, the 4.0 ns component was not due to contaminating Fl in either the Mab or HPF preparations. This was verified both spectrophotometricaUy (at the observed percentages, adsorbance at 500 nm would be expected) and experimentally (no 4.0 ns component was present when non-fluorescent FI-NH 2 was titrated with each Mab preparation). Although Templeton and Ware [15] offered no explanation, this observation may have been due to two possibilities. Since fluorescence lifetime determinations were performed at 20 °C and bound and free ligand were in equilibrium, the 4.0 ns component may have been higher quantum yield HPF molecules (as described above) entering or exiting the active sites. Alternatively, active site bound HPF molecules may have existed in two different conformations that resulted in both a 4.0 ns component and a shorter lifetime component. The 4.0 ns component, therefore, can be described as specifically bound HPF which existed in an active site 'encounter complex' [30,31]. If this phenomena was observed, then the bound 4.0 ns HPF molecules eventually needed to convert to either a 0.52 ns (4-4-20) or 2.23 ns (9-40) component. This conversion would require either HPF realignment in an active site, active site residues rearrangement around HPF, or a combination of both. Steady-state fluorescence polarization measurements were performed to distinguish between each possibility. Initially, free Fl (P = 0.019) and HPF (P = 0.048) exhibited slightly different polarization values (Table il) due to their different fluorescence lifetimes (as described by Perrin, 1929). When bound by 4-4-20 and 12-40, however, both Fi and HPF were identically polarized (P = 0.450). F! bound by 9-40 was also polarized to this extent (P = 0.456), but HPF bound by 9-40 was significantly depolarized (P - 0.428). As with fluorescence lifetime measurements, polarization values were determined by Mab titrations until active site excess. Although a polarization value is directly proportional to the rotational correlation coefficient (p) and inversely proportional to the fluorescence lifetime (~') [32], depolarized HPF when bound by 9-40 was not due to these factors. For example, while 4-4-20 (IgG2a)
33 and 12-40 (IgG1) possessed different H chain isotypes (and potentially different p values) each polarized HPF to an identical extent. In this regard, Yguerabide and colleagues have determined rotational correlation coefficients to be 168 ns and 33 ns for lgG molecules and Fab fragments, respectively [33,34]. Additionally, assuming all bound HPF molecules possessed a 4.0 ns lifetime and Po and p values remained constant, the theoretical (0.449) and observed (0.428) P values were still significantly different. Finally, depolarized 9-40 bound HPF was not due to a low binding constant since 12-40 expressed a similar affinity for HPF as 9-40 (Table I). It appeared, therefore, that 9-40 bound HPF was depolarized due to greater 9-40 active site motion a n d / o r greater HPF flexibility within the 9-40 active site. In either case, both phenomena resulted in an apparent decrease in p since P0 was constant and the bound F! and HPF lifetimes were 2.23 ns. Utilization of fluorescent probes, which possess fluorescence lifetimes that are both sensitive to their environment and occur in a similar timescale (ps to ns) as molecular events, represent experimental approaches to measure protein structural dynamic~ t3538]. Smaller proteins such as myoglobin have been shown, using in part fluorescence techniques, to exist in multiple conformational substates [39]. Only recently, however, has the potential existence of multiple antibody active site conformational substates been addressed [40-43]. To investigate this, Fl/anti-Fi interactions have been a valuable model system since FI was both the immunogen and the active site-specific extrinsic fluorescent probe. Because FI was an inherently rigid molecule, however, HPF was employed in this study to circumvent potential active site stabilizations imposed by FI [40,42]. Use of the more flexible HPF molecule, therefore, facilitated experiments designed to observe active site dynamics. All results presented in this preliminary study were consistent with the idea that structurally similar antibody active sites may exhibit quite different dynamic properties. In the case of 4-4-20 and 9-40, although still unproven, structural differences in H CDR3 could result in a more dynamic 9-40 active site which, in turn, could affect the binding constant. This affect would be in addition to L34 Arg in the 4-4-20 active site and explained why the 9-40 H chain reassociated with the 4-4-20 L chain did not express an increased affinity for FI [12]. In the synthesis of designer antibodies, therefore, ligand and/or active site flexibility may be important [44,45]. For example, intermediate affinity antibody-antigen interactions may require either the active site or ligand to be relatively rigid; whereas high affinity antibody-antigen interactions may require both to be relatively rigid. Current experiments (e.g. low temperature and active site site-directed mutagenesis studies) are being per-
formed to further characterize the role that active site dynamics contribute to antigen recognition.
Acknowledgments Fluorescence measurements and analyses were performed at the Laboratory for Fluorescence Dynamics (LFD) at the University of Illinois Urbana-Champaign (UIUC). The LFD is supported jointly by the Division of Research Resources of the NIH (RR03155-01) and the UIUC. We thank the LFD staff (in particular Drs. Catherine Royer, Joseph Beechem, James Matthies and William Mantulin) for their help in this study. We also thank Dr. William Ware for the generous gift of HPF and Sandra Henson for expert secretarial assistance.
References 1 Tonegawa, S. (1983) Nature 302, 575-581. 2 Yancopoulos, G.D. and A!L F.W. (1986) Ann. Rev. Immunol. 4. 3~o-2~. 3 Kabat, E.A., Wu, T.T., Reid-Miller, M., Perry, H.M. and Gottesman, K.S. (1987) Sequences of Proteins of Immunological Interest, U.S. Department of Health and Human Servies, Public Health Service. National Institutes of Health, Bethesda, MD. 4 Davies, D.R.. Padlan, E.A. and Scheriff, S. (1990) Annu. Rev. Biochem. 59. 439-474. 5 Mian, I.S., Bradwell, A.R and Olson, A.J. (1991) J. Mol. Biol. 217, 133-151. e ,"nit, A.G., Mariuzza, R.A., Phillips, S.E.U. and Poljak, R.J. ~1986) S,:ience 233, 747-753. 7 Alzari, P.M., Spinelli, S., Mariuzza. R.A., Boulot, G., Poljak, R.J., Jarvis, J.M. and Milstein, C. (1990) EMBO J. 9, 3807-3814. 8 Stanfield, R.L., Fieser, T.M., Lerner, R.A. and Wilson, I.A. (1990) Science 248, 712-719. 9 Herron, H.N., He, X.-M., Mason, M.L, Voss. E.W., Jr. and Edmundson, A.B. (1989) Proteins 5, 271-280. 10 Kranz, D.M., Herron, J.N. and Voss, E.W., Jr. (1982) J. Biol. Chem. 257, 6987-6995. 11 Bates, R.M., Ballard, D.W. and Voss, E.W., Jr. (1985) Moi. lmmunol. 22, 871-877. 12 Bedzyk, W.D. and Voss, E.W., Jr. (1991) Moi. lmmunol. 28, 27-34. 13 Denzin, LK., Whitlow, M. and Voss, E.W., Jr. (1991) J. Biol. Chem. 266, 14095-14103. 14 Martin, M. and Lindqvist, L. (1975) J. Luminescence 10, 381-390. 15 Templeton, E.F.G. and Ware, W.R. (1985) Mol. lmmunol. 22, 45-55. 16 Galfre, G., Howe. S.C., Milstein, C., Butcher, G.W. and Howard, J.C. (1977) Nature 266, 550-552. 17 Kranz, D.M. and Voss, E.W., Jr. (1981) Mol. Immunol. 18, 889-898. 18 Bedzyk, W.D., Reinitz, D.M. and Voss, E.W., Jr. (1986) Mol. Immunol 23, 1319-1328. 19 Bedzyk, W.D., Herron, J.N., Edmundson, A.B. and Voss, E.W., Jr. (1990) J. Biol. Chem. 265, 133-138. 20 Kranz, D.M. and Voss, E.W., Jr. (1984) in Fluorescein Hapten: An Immunological Probe. (Voss, E.W., Jr., ed.), pp. 15-20: CRC Press, Boca Raton. 21 Reinitz, D.M. and Voss, E.W., Jr., (1985) J. lmmunol. 135, 3365-3371.
34 22 Herron, J.N. 1984. In Fluorescein Hapten: An Immunological Probe. Voss, E.W., Jr., ed.), pp. 49-76, CRC Press, Boca Raton. 23 Watt, R.M. and Voss, E.W., Jr. (1978) Immunochemistry 15, 875 -882. 24 Gratton, E. and Limkeman, M. (1983) Biophys. J. 44, 315-324. 25 Piston, D.W., Marriott, G., Radivozevieh, T., Clegg, R.M., Jovin, T.M. and Gratton, E. (1991) Rev. Sci. Inst., in press. 26 Beechem, J.M. and Gratton, E. (1988) SPIE 909, 70-81. 27 Janjic, N., Schloeder, D. and Tramontano, A. (1989) J. Am. Chem. Soc. 111, 6374-6377. 28 Holloway, H.E., Nauman, R.V. and Wharton, J.H. (1968) J. Phys. Chem. 72, 4474-4482. 29 Berlman, I.B. (1970) J. Phys. Chem. 74, 3085-3093. 30 Lancet, D. and Pecht, I. (1976) Proc. Natl. Acad. Sci. USA 73, 3548-3553. 31 Pecht, I. (1982) The Antigens 6, 1-69. 32 Perrin, F. (!929) Ann. Phys. 12, 169-175. 33 Yguerabide, J. (1972) Methods Enzymol. 26C, 498-541. 34 Yguerabide, J., Epstein, H.F. and Stryer, L. (1976) J. Mol. Biol. 51,573-590.
35 Beechem, J. and Brand, L. (1985) Biochemistry 54, 43-71. 36 Alcala, R.J., Gratton, E. and Prendergast, F.G. (1987) P,iophys. J. 51,587-596. 37 Aicala, R.J., Gratton, E. and Prendergast, F.G. (1987) Biophys. J. 5 I, 597-604. 38 Alcala, R.J., Gratton, E. and Prendergast, F.G. (1987) Biophys. J. 51,925-936. 39 Frauenfeider, H., Parak, F. and Young, R.D. (1988) Annu. Rev. Biophys. Chem. 17, 541-479. 40 Voss, E.W., Jr., Dombrink-Kurtzman, M.A. and Miklasz, S.D. (1988) Immunol. Invest. 17, 25-39. 41 Stevens, F.J., Chang, C.-H. and Schiffer, M. (1988) Proc. Natl. Acad. Sci USA 85, 6895-6898. 42 Weidner, K.M. and Voss, E.W., Jr. (1991) J. Biol. Chem. 266, 2513-2519. 43 Swindlehurst, C.A. and Voss, E.W., Jr. (1991) Biophys. J. 59, 619-628. 44 Novotny, J., Bruccoleri, R.E. and Saul, F. (1989) Biochemistry 28, 4735 -4749. 45 Novotny, J. (1991) Mol. Immunol. 28, 201-207.
27
BBAPRO 34088
Relative binding properties of fluorescein and 9-hydroxyphenylfluoron (HPF) with murine monoclonal anti-fluorescein antibodies * William D. Bedzyk, Cathy A. Swindlehurst and Edward W. Voss, Jr. Department of Microbiolo~" Unicersity of Illinois at Urbana-Champaign, Urbana, IL (U. S.A. ) (Received 5 June 1991) (Revised manuscript received 3 September 1991 )
Key words: Antibody active site: Conformational substatc: Fluorescence lifetime
Contribution of the fluorescein (FI) carboxyl group to hapten binding by idiotypically related murine monoclonal anti-Fl antibodies 4-4-20, 9-40 and 12-40 was studied by comparing relative liganded active site properties with bound FI or 9-hydroxyphenylfluoron (HPF). Kinetic studies revealed similar association rate constants between Fi and HPF to 4-4-20 ( = 1.1 • 107 M - t s - t ) ; however, the 4-4-20 dissociation rate for F! was = 200 times slower, relative to HPF, which resulted in relative intrinsic affinity values of 1.2- 10 TM and 6.5 • 107 M - t, respectively. Mabs 9-40 and 12-40 also displayed a reduced affinity for HPF and affinity constants of 5.5- l0 s M -t and 6.7. l0 s M were obtained from a competitive ELISA. Additionally, previous studies revealed that upon binding FI, Mabs 4.4-20 (92.1%), 9-40 (44.7%) and 12-40 (73.4%) quenched F! fluorescence. Similar analyses with HPF resulted in 64.4% and 2.0% fluorescence quenching by 4-4-20 and 12-40, respectively; however, 9-40 increased HPF fluorescence by = 24%. Steady-state fluorescence polarization experiments revealed that in solution, F! ( P = 0.019) and HPF ( P = 0.048) were polarized to different degrees. When bound, however, F! and HPF expressed similar polarization values ( P --- 0.455), except 9-40 bound HPF which was significantly depolarized ( P = 0.428). Fluorescence lifetime experiments revealed F! to possess two discrete lifetimes: a 3.96 ns component (free FI) and either a 0.52 ns (4-4-20), 2.23 ns (9-40) or 0.96 ns (12-40) short component that corresponded to bound FI. HPF, however, when bound by 4-4-20 or 9-40, was best fit by three discrete exponentials: a relatively long 4.0 ns component, a 1.11 ns lifetime (free HPF) and either a 0.52 ns (4-4-20) or 2.23 ns (9-40) component. Finally, HPF bound by Mab 12-40 exhibited a single lorenzian distributed lifetime of 1.36 ns ( _ 0.43 ns). Results are discussed in terms of Mab active site structure and conformational state dynamics.
Introduction Although genetic elements and mechanisms responsible for antibody active site diversity have been de-
* This work was supported in part by grant No. 2-3-013 from the Biotechnology Research and Development Corporation (Peoria, ILk Abbreviations: CDR, complementarity-determining region: Fl, fluorescein: H. immunoglobulin heavy chain: HPF, 9-hydroxyphenyifluoton: L, immunoglobulin light chain: Mab, monoclonal antibody: TMB, 3,3',5,5'-tetramethylbenzidine: V, immunoglobulin variable region. Correspondence: E.W. Voss, Department of Microbiology, University of Illinois at Urbana-Champaign, 407 S. Goodwin Avenue, Urbanao IL 61801, U.S.A.
scribed in detail [1,2], at the protein level, mechanisms involved in antibody-antigen complementarity remain relatively unresolved. Comparison of known variable region primary structures [3] with accumulated results from X-ray crystallographic studies (reviewed in Refs. 4 and 5) have suggested certain active site residues as being generally important in antibody recognition of a variety of antigens. These apparently critical side chains include Arg (which has a planar nature and delocalized charge) and Tyr and Trp (which have the capacity to hydrogen-bond and form hydrophobic and electrostatic interactions with antigen). Of particular interest has been the interaction between residue L32Tyr with both protein and haptenic antigens. For example, while L32Tyr participated in either a hydrophobic or hydrogen-bond interaction with hen egg lysozyme (Mab DI.3)
28 [6] and 2-phenyloxazolone (Mab NQ10/12.5) [7], respectively, L32Tyr interacted with the C helix of myohemerythrin (Mab B1312) [8] and fluorescein (Mab 4-4-20) via both mechanisms [9]. To specifically investigate the hydrogen-bond contributions of residue L32Tyr to overall ligand (Fl) binding, therefore, a structural analog of Fl has been employed. Historically, studies of interactions between anti-F! antibodies and haptenic structural analogs have proven difficult because most Fl derivatives (such as eosin, erythrosin and members of the rhodamine series) have shown minimal crossreactivity [10,11]. The three-dimensional co-crystal structure of Mab 4-4-20 and Fl revealed that the xanthenone moiety fit tightly into a pocket-like binding site [9] consistent with the lack of accommodation for these larger F! analogs. Additionally, although Mab 4-4-20 Fl contact residues were determined to be L27dHis, L32Tyr, L34Arg, LglSer, L96Trp and H33Trp [9], individual contributions of each residue to Fl-binding remained uncharacterized. Initially, in vitro heavy and light chain reassociations and site-directed mutagenesis experiments with the 4-4-20 single-chain derivative demonstrated L34Arg was responsible for increased 4-4-20 affinity and the L32Tyr hydrogen-bond with F! contributed -- 100-fold to the overall binding constant [12,13]. Based on these observations, a smaller Fi analog, 9-hydroxyphenylfluoron (HPF), was chosen as the cross-reactive ligand in these studies to test certain hypotheses. HPF differed from F! by absence of the carbox3'i group (Fig. 1) and thus, provided direct information on the L32Tyr hydrogen-bond contribution to Fl-binding by anti-F! Mabs. HPF was first synthesized by Martin and Lindqvist [14] to study pH dependency of Fl fluorescence and was later studied bound by rabbit polyclonal anti-Fl antibodies [15]. This study utilized HPF to examine the nature and extent of interactions present in a homogeneous binding site population. Finally, the use of HPF facilitated binding studies that probed the capacity and rate at which an antibody active site specific for the relatively rigid FI ligand stabilized the HPF phenyl ring (which possessed significant rotational freedom in relation to the xanthenone group). In this study, F! and HPF were both the ligand and an extrinsic fluorescent probe. Thus, a comparison of F! and HPF binding by a homogeneous antibody active site population permitted assessment of ring stabilization. Furthermore, since the fluorescence properties of each ligand correlated with the active site environment and xanthenone and phenyl ring orientation, fluorescence quenching, polarization, and timeresolved lifetime measurements were available. This study, therefore, also permitted preliminary analyses of antibody active site conformational state dynamics by determining relative Fl and HPF motion within each active site.
Materials and Methods
Reagents Fiuorescein disodium salt (Fi) and fluorescein-5amine (FI-NH 2) were purchased from Research Organics (Cleveland, OH). FI analog 9-hydroxyphenyifluoron (HPF) was kindly provided by Dr. W. Ware (The University of Western Ontario). Structural comparisons of F! and HPF are shown in Fig. 1, and HPF spectral properties have been described [14,15]. Tris (Ultra Pure) was obtained from Sigma (St. Louis, MO). Immulon II polystyrene wells, horseradish peroxidaselabeled Protein G and TMB (3,3',5,5'-tetramethylbenzidine) were procured from Dynatech Laboratories (Chantilly, VA), Zymed Laboratories (South San Francisco, CA) and Kirkegaard and Perry Laboratories (Gaithe~sburg, MD), respectively. Hybridoma cell lines and monoclonal antibodies Hybridoma cell lines were obtained by chemical fusion [16] of non-secreting murine myeloma cell line Sp2/0-Agl4 with hyperimmunized splenocytes from BALB/cV mice. Mabs 4-4-20 (IgG2a, r; K a = 1.7- 10~° M-l), 9-40 (IgGl, K; Ka=3.3" 10 7 M - I ) and 12-40 (IgGl, K; K~ = 1.0- 108 M - l ) F!-binding and idiotypic characteristics [11,17,18] and primary structures [19] were previously reported. Hybridoma proteins 4-4-20, 9-40 and 12-40 were produced as ascitic fluids in BALB/cV mice. Immunoglobulin purification was accomplished by dextran sulfate precipitation of lipoproteins followed by ammonium sulfate precipitation and gamma globulin fraction enrichment [20]. Subsequently, Mabs were affinity purified with FI-Sepharose and eluted with 0.23 M glycine pH 2.6 as described [211. Competitive ELISA A competitive ELISA was utilized to approximate HPF affinity constants for Mabs 4-4-20, 9-40 and 12-40. Microtiter wells were initially coated with 50 /zl (100 ng/ml) Fll2.sBSA in 50 mM Tris, 150 mM NaCl (pH
HO0" ~ 0
HO"~ °
d A
B
Fig. I. F! and HPF structural comparison. Fl (panel A) and HPF (panel B) contain identical immunodominant xanthenone moieties but differ by the presence (Fi) or absence (HPF) of a phenyl ring carboxyi group. At neutral pH or greater, both dianionic Fi and anionic HPF possess similar absorption maxima and molar extinction coefficients but different fluorescence quantum yields ( ~ = 0.92 for FI and 0.21 for HPF) [14].
29 8.0) (TBS) and incubated for 2 h at 37 ° C. Wells were washed 2 ×with TEBT (TBS plus 1% BSA, 1 mM E D T A and 0.05% Tween-20) and excess protein binding sites masked with TEBT for 30 min at 37 ° C. 50 #! of 4-4-20 (100 ng/ml), or 9-40 or 12-40 (1 /.tg/mi), pre-incubated for 30 min at 4 ° C with various FI or H P F concentrations, was added to the wells and incubated for 4 h at 4 ° C . To detect solid-phase bound antibody molecules, horseradish peroxidase-labeled protein G was added and incubated for 2 h at 37 o C. Finally, TMB substrate was added (50 ~tl) and enzyme reactions ultimately terminated with 1 M H 2SO4 . Optical density measurements employed a Dynatech MR600 automatic 96-wel plate reader equipped with a 450 nm cut-off filter. Fluorescence instrumentation Association rate constants and fluorescence polarization analyses were performed with a photon counting spectrophotometer (Gregg-P.C., ISS Instruments, Champaign, IL) and utilized a xenon arc lamp (excitation at 493 nm and emission at 530 nm). Dissociation rate constants and amax values were determined as described [22,23] and utilized an Aminco-Bowman spectrophotofluorometer equipped with a xenon arc lamp (excitation at 493 nm and emission at 530 nm). Fluorescence lifetime measurements were performed on a multifrequency phase and modulation fluorometer [24]. Samples were excited by the 488 nm line of an argon ion laser passed through an acousto-optic modulator [25] and a polarization filter set at 35 °. A cut-on filter selected emission wavelengths greater than 520 nm and all measurements were taken using glycogen as a reference solution. Phase and modulation data were collected at 8 to i2 frequencies ranging from 10 to 210 MHz. The sample chamber was thermostat-controlled with a circulating water bath at 20 ° C. Association rate analysis Association rate constants for FI and H P F binding by Mab 4-4-20 were determined by monitoring decreases in fluorescence intensity as a function of time [22]. Equal volumes of 4-4-20 and ligand were mixed with a manual stopped-flow apparatus (Hi-Tech Scientific model SFA-11 Rapid Kinetics Assembly) thermostated at 2°C. A 10-fold molar excess of 4-4-20 (27.0 nM), compared to ligand (2.0 nM), was used to ensure first-order kinetics. Fluorescence intensity was measured as a function of time using ISS kinetics software. Technical rate constants (Ltech) were determined from the slope of the fraction free ligand ( L / L o) plotted as a function of time (t). The fraction free ligand was calculated from the equation: F(t) - F(eq) L / L o = In
F(0) - F ( e q )
(1)
where F(t) was the fluorescence intensity at time t, F(eq) the equilibrium fluorescence intensity, and F(0) the initial fluorescence intensity. A bimolecular association constant (k t) was calculated from the equation: Ft.,(eq) • Ktc,:h kl = [Sites]
(2)
where Fb(e q) was the fraction of bound ligand at equilibrium, [Sites] the active site concentration and Ktech the slope of L / L o vs. t. Experiments were repeated at least in triplicate. Dissociation rate analysis F! and HPF dissociation rates (k 2) were determined by initially monitoring increased fluorescence as a function of time after addition of a 10-fold molar excess of FI-NH 2 [22]. Fluorescence intensities were measured and analyzed using OffRate acquisition and analysis software (Interactive Software, Urbana, IL). An affinity constant (K a) was subsequently calculated from the formula: K.=kl/k
z.
(3)
Fluorescence lifetime measurements Stock FI and HPF solutions, 2.0 and 4.0.10 -8 M, respectively, were prepared in 50 mM Tris (pH 8.0) at 20 o C. Following fluorescence lifetime determination for free Fi or HPF in solution, affinity purified antibody (1 nM active sites) was added and phase and modulation data acquired. Analogous titrations were repeated until a 100:1 molar ratio (active sites:ligand) was achieved. For each frequency, fluorescence lifetime measurements were taken until phase angle error was less than or equal to 0.200 and modulation ratio error was less than or equal to 0.004. Multifrequency phase and modulation data analysis was performed on a personal computer and used Globals Unlimited software [26]. Data were analyzed using several different fitting routines and the best fit was obtained for either discrete exponentials or lorenzian distributed fluorescence lifetimes. Results
Association rate determination Kinetic studies were performed to characterize the interaction of 4-4-20 with Fl and HPF. Ligand association rates with 4-4-20 were determined to be 1.01 + 0.16- 107 M - l S-I and 1.21 _+ 0.55" 10 / M - I s - i f and HPF, respectively (data not shown). The association rate constant between 4-4-20 and Fl was approximately twice that previously reported [10]. This difference was probably due to differences in both experi-
30 mental conditions and instrumentation. For example, association rate constant determination utilized a thermostated stopped-flow apparatus with computerized data acquisition from a photon counting spectrophotometer. Previously reported values, however, were determined by manual ligand injection into a thermostated cuvette in a spectrophotofluorometer connected to a strip chart recorder. Nevertheless, association rate constants for FI and H P F with 4-4-20 were identical within error. Ligand affinity constants
Relative dissociation rate constants for 4-4-20 with the two ligands were measured to determine affinity constants. The dissociation rate was 4.5 + 0.8-10 -4 s - t for F1 and 7.8 + 3.0' 10 -2 s - i for HPF. Affinity constants, therefore, were determined to be 1.2-10 ~° M - ! and 6.5.10 7 M - l for Fi and H P F (Table I), respectively. Similarly, Mab 9-40 possessed an affinity for FI of 3.3- 10 r M - ! while the affinity of 12-40 was 1.0- 10 s M-~. Due to assay limitations, an off-rate for 9-40 and 12-40 bound H P F was too rapid for detection. This indicated that the affinities of 9-40 and 12-40 for H P F were < 5 . 0 . 1 0 6 M - z To determine an approximate 9-40 and 12-40 H P F affinity, a competitive ELISA, where free FI and H P F were soluble inhibitors of 9-40 or 12-40 binding to solid-phase FI~2.sBSA , was employed (Fig. 2). A 62.5fold concentration difference was observed between 50% inhibition by FI and H P F and indicated 9-40 possessed an affinity of - 5 . 5 . 1 0 5 M -~ for HPF. Similarly, 12-40 possessed an affinity for H P F of = 6.7 • 105 M - ~. Validity of thes~ estimations was verified by inhibition of 4-4-20 binding to solid-phase FI under identical conditions. A 200-fold concentration difference required for 50% inhibition by FI and H P F (Fig. 2) was consistent with the 200-fold difference in affinity as determined by off-rate analysis. Finally, a relatively smaller FI analog.~ resorufin, was an ineffective inhibitor of 4-4-20, 9-40 and 12-40 binding to solid-phase FI (data not shown').
A
"
v
C.
~p-
:~ o~ 80
60 40
20
| 10
9
8
7
6
5
4
10
9
8
7
.
,
,
6
5
4
1 0 9 8 7 6 5 4
-log [inhibitor] (M)
Fig. 2. Competitive ELISA inhibition curves. The ability of various soluble F! (closed symbols) and HPF (open symbols)concentrations to inhibit 4-4-20 (panel A), 9-40 (panel B) and 12-40 (panel C) binding to solid phase F!n2.5BSA. Following incubations and washes, solid-phase bound antibody molecules were detected with horseradish peroxidase-labeled Protein G and TMB substrate. Differences between Fi and HPF concentrations required for 50% inhibition were used to estimate HPF affinity constants for Mabs 9-40 and 12-40.
Fluorescence measurements
Initially, Mab bound Fl and H P F Qma~ values were obtained to determine if the L32Tyr hydrogen-bond with Fl contributed to this observed phenomenon. As shown in Table I, Fl Qmax values for 4-4-20, 12-40 and 9-40 were identical to previously reported values [12,18]. H P F fluorescence, however, was quenched to a lesser extent by 4-4-20 (64.4%) and 12-40 (2.0%) (Table I). Under identical conditions, surprisingly, 9-40 bound H P F exhibited an increase of ~ 24% in total fluorescence units (Table I). To resolve this apparent 9-40 bound H P F anomaly and to investigate potential explanations for reduced 4-4-20 and 12-40 H P F Qmax values, free and Mab bound Fl and H P F fluorescence lifetimes were determined (Table II). Following phase and modulation data acquisition for free ligand in solution, an aliquot of affinity purified Mab was added and phase and modulation data obtained. Analogous titrations were performed until the molar active site concentration was 100-fold greater than the molar ligand concentration. When free in solution, Fl and H P F possessed single discrete 3.96 ns and 1.11 ns lifetimes, respectively.
TABLE I Ligand-binding characteristics
Mab
FI Ka
AG
4-4-20 12-40 9-40
(M-t) 1.2.10 t° 1.0.10s 3.3.107
(kcal/mol) "~ - 12.6 - 10.0 -9.3
HPF o
Qmaxb
Ka
AG
92.1 +_0.1% 73.4+_ 1.3% 44.7+_3.8%
(M- x) 6.5-107 6.7.105 c 5.5" 105 c
(kcal/mol) ~ - - 9.7 -7.3 c -7.2 c
o
Qmax
64.4+ 0.4% 2.0+_0.5% -23.9+_0.3% d
" Binding free energy derived from AG o = _ R T in K a. b Fluorescence of antibody bound relative to free iigand at protein saturation. Measurements were performed at 20 °C and each value represents the mean of triplicate trials + standard deviation. c Approximated from compet;,~ave ELISA. d A negative value indicates fluorescence enhancement.
31 TABLE II
Fluorescence polarization and lifetime data for free and Mab bound FlandHPF Sample
Temperature
Polarization a
Tt
~'2
~'3
~c2 h
FI 4-4-20/Fi 12-40/F! 9-40/F1 HPF 4-4-20/HPF 12-40/HPF 9-40/HPF
20 20 20 20 20 20 20 20
0.019 _+ 0.002 0.458 _-_+0.001 0.453 _+0.000 0.456 _+0.001 0.048_+ 0.001 0.455 -1-0.001 0.449 _+0.003 0.428 +_ 0.002
3.96 c 3.96 3.97 3.96 1.11 1.04 1.36( _+0.43) d 1.04
0.52 0.963 2.23 0.52 _ 2.23
3.90 4.14
0.849 1. i 56 I. 136 0.524 0.674 0.177 0.933 0.344
a b c d
oC oC oC oC °C oC °C oC
Steady-state polarization value. C h i - s q u a r e f o r d a t a fitting r o u t i n e . L i f e t i m e v a l u e s a r e in n a n o s e c o n d s . N u m b e r in p a r a n t h e s e s r e p r e s e n t s the w i d t h f o r a I o r e n z i a n d i s t r i b u t e d c o m p o n e n t .
Since titrations were performed, most Mab bound measurements contained a percentage of this discrete lifetime component which corresponded to free ligand. F1 titrated separately with 4-4-20 (0.52 ns), 12-40 (0.96 ns) and 9-40 (2.23 ns) resulted in one additional shorter discrete lifetime (Table ll). HPF titrated separately with 4-4-20 and 9-40 resulted in two discrete lifetime components (in addition to free HPF-I.ll ns). Besides a shorter exponential (0.52 ns for 4-4-20 and 2.23 ns for 9-40), both 4-4-20 and 9-40 bound HPF displayed a discrete 3.9 to 4.1 ns component (Table II). HPF bound by 12-40, however, exhibited a single lorenzian distributed fluorescence lifetime with a center at i.36 ns and a width of _+0.043 ns. Finally, X2 values for each phase and modulation data fitting routine are also presented in Table II and were less than 1.16. To examine the extent of FI and HPF flexibility when bound by Mabs 4-4-20, 9-40 and 12-40, steadystate fluorescence polarization values were obtained. Free FI and HPF were depolarized as expected (P = 0.019 and P -- 0.048, respectively) (Table II). Both 4-420 and 12-40 bound Fl and HPF were polarized to similar degrees (P --- 0.453). Fl bound by Mab 9-40 was also polarized to this extent (P = 0.456); however, HPF bound by 9-40 was significantly depolarized and had a value of 0.428 (Table II). Discussion
Since accumulated evidence (i.e. V region primary and quaternary structure comparisons) has suggested that antibody active site tryptophan and tyrosine residues are generally important in antigen complementarity and binding [5,7], the contribution of Fl contact residue L32Tyr [9] to ligand binding was initially investigated in three idiotypically related anti-F!
Mabs (4-4-20, 9-40 and 12-40). To accomplish this, comparative binding assays and fluorescence polarization and lifetime measurements with free and Mab bound Fl and HPF (Fig. 1)were performed. Although HPF was a more flexible ligand (see below), due in part to lack of a stabilizing phenyl ring carboxyl group, association rate constants between 4-4-20 and either Fl or HPF were identical within error (data not shown). Ligand flexibility and hydrogen-bond formation, therefore, apparently did not effect ligand association with the active site. Fl and HPF dissociation rates from 4-4-20 active sites yielded binding free energies of -12.6 kcal/mol (Fl) and - 9 . 7 keal/mol (HPF) (Table I). These values were similar to values obtained by 4-4-20 active site L32Tyr to L32Phe site-directed mutagenesis studies [13]. Mabs 12-40 and 9-40 HPF binding free energies were estimated by a competitive ELISA to be - 7 . 3 kcal/mol and - 7 . 2 kcal/mol, respectively. Differences in Fl and HPF binding free energies (AAG °) for 4-4-20, 12-40 and 9-40 were 2.9, 2.7 and 2.2 kcal/mol, respectively. In percentage terms of the total binding free energy, this hydrogen-bond was responsible for 27.0% (12-40), 23.0% (4-4-20) and 23.4% (9-40) of the kcal/moles in active site binding. Although the Fl carboxylate group contribution to overall 4-4-20, 9-40 and 12-40 Fl-binding (2.0-3.0 kcal/mol) was less than previously reported values (= 4.0 kcal/mol) for additional anti-F! Mabs [27], all reported values were consistent with contributional differences associated with a hydrogen-bond [9] compared to a potential salt bridge [27]. In either case, the L32Tyr hydrogen-bond with Fl contributed differently to the overall binding free energy of an individual Mab. This was consistent ~ith V region primary structure comparisons of idiotypically related anti-Fl Mabs [19] which suggested potentially subtle active site micro-environment differences (e.g.,
32 the degree of solvent exposure of the L32Tyr hydrogen-bond) due, in part, to H CDR3 residue differences. Time-resolved fluorescence lifetime measurements of free and Mab bound F! and HPF were obtained to investigate each anti-Fl active site micro-environment (Table II). Identical discrete fluorescence lifetime components (0.52 ns) were observed for both FI and HPF when bound by 4-4-20. This suggested that the two ligands adopted similar conformations in the active site. Similar results were observed for 9-40 bound FI and HPF (2.23 ns). Finally, 12-40 bound Fl and HPF probably expressed identical fluorescence lifetimes; however, 0.96 ns (bound) and 1.11 ns (free HPF) were undiscernible and resulted in a single lorenzian distributed lifetime. From these results, several mechanisms of Fl-binding by these anti-FI Mabs were elucidated. The L32Tyr hydrogen-bond with FI, although significant in binding affinity determination, was not required for the F! fluorescence quenching phenomenon. This was consistent with previous studies that investigated the pH dependency of F! fluorescence [14] and showed that the carboxyl group ionization state did not influence the quantum yield of Fl. The enolic oxygen ionization state and the extent and nature of aromatic stacking between the Fl xanthenone group and aromatic active site residues, therefore, determined maximum fluorescence quenching. Additionally, 9-40 bound HPF fluorescence enhancement (2.23 ns component) was due to relative HPF flexibility and identical active site interactions with both Fl and HPF. Under steady-state conditions at 20°C, fluorescence quantum yields for Fl (~--0.92 or ~-= 3.96 ns) and HPF (q~ = 0.21 or ~-- 1.11 ns) differed, in part, due to the phenyl ring orientation relative to the xanthenone group. The Fl carboxyl group fixed the phenyl ring at an approx. 62 ° angle relative to the xanthenone; however, since HPF lacked this carboxyl group, the phenyl ring possessed greater flexibility. The molecular configuration of other bi-aromatic chromophores has been previously shown to dictate the absorption and fluorescence profiles of the fluorophore [28,29]. This phenomenon was explained by the 1;-electron orientations (e.g., parallel aromatic groups in the same fluorophore possessed aligned w-electrons and higher fluorescence quantum yields and perpendicular aromatic groups possessed unaligned w-electrons and lower quantum yields). With HPF, therefore, a steady-state quantum yield ( ~ = 0.21 or ~-= 1.11 ns) represented an average of rapidly interconverting higher and lower quantum yield species dependent on xanthenone ring and phenyl ring orientation. This interconversion rate, however, must be accomplished rapidly since the association rate constants between Mab 4-4-20 and either F! and HPF were identical. The HPF molecular configuration when being bound by 9-40 (or 4-4-20 or 12-40), therefore,
was similar to Fi and possessed a lifetime near the solution phase lifetime for Fi (~-3.96 ns). In fact, a 3.96 ns fluorescence lifetime was observed (see below). In addition, a shorter HPF lifetime, comparable to quenched Fi (i.e. 2.23 ns), was observed. On a steadystate basis, therefore, the 9-40 bound HPF 2.23 ns fluorescence lifetime was greater than free HPF (1.11 ns) and fluorescence enhancement was observed. Besides a fluorescence lifetime that corresponded to free ligand, 4-4-20 and 9-40 bound HPF displayed a relatively long lifetime species (1-= 4.0 ns). A similar discrete component was observed when HPF was bound by rabbit polyclonal anti-Fl antibodies [15]. Similar to these polyclonal antibody studies, the 4.0 ns component was not due to contaminating Fl in either the Mab or HPF preparations. This was verified both spectrophotometricaUy (at the observed percentages, adsorbance at 500 nm would be expected) and experimentally (no 4.0 ns component was present when non-fluorescent FI-NH 2 was titrated with each Mab preparation). Although Templeton and Ware [15] offered no explanation, this observation may have been due to two possibilities. Since fluorescence lifetime determinations were performed at 20 °C and bound and free ligand were in equilibrium, the 4.0 ns component may have been higher quantum yield HPF molecules (as described above) entering or exiting the active sites. Alternatively, active site bound HPF molecules may have existed in two different conformations that resulted in both a 4.0 ns component and a shorter lifetime component. The 4.0 ns component, therefore, can be described as specifically bound HPF which existed in an active site 'encounter complex' [30,31]. If this phenomena was observed, then the bound 4.0 ns HPF molecules eventually needed to convert to either a 0.52 ns (4-4-20) or 2.23 ns (9-40) component. This conversion would require either HPF realignment in an active site, active site residues rearrangement around HPF, or a combination of both. Steady-state fluorescence polarization measurements were performed to distinguish between each possibility. Initially, free Fl (P = 0.019) and HPF (P = 0.048) exhibited slightly different polarization values (Table il) due to their different fluorescence lifetimes (as described by Perrin, 1929). When bound by 4-4-20 and 12-40, however, both Fi and HPF were identically polarized (P = 0.450). F! bound by 9-40 was also polarized to this extent (P = 0.456), but HPF bound by 9-40 was significantly depolarized (P - 0.428). As with fluorescence lifetime measurements, polarization values were determined by Mab titrations until active site excess. Although a polarization value is directly proportional to the rotational correlation coefficient (p) and inversely proportional to the fluorescence lifetime (~') [32], depolarized HPF when bound by 9-40 was not due to these factors. For example, while 4-4-20 (IgG2a)
33 and 12-40 (IgG1) possessed different H chain isotypes (and potentially different p values) each polarized HPF to an identical extent. In this regard, Yguerabide and colleagues have determined rotational correlation coefficients to be 168 ns and 33 ns for lgG molecules and Fab fragments, respectively [33,34]. Additionally, assuming all bound HPF molecules possessed a 4.0 ns lifetime and Po and p values remained constant, the theoretical (0.449) and observed (0.428) P values were still significantly different. Finally, depolarized 9-40 bound HPF was not due to a low binding constant since 12-40 expressed a similar affinity for HPF as 9-40 (Table I). It appeared, therefore, that 9-40 bound HPF was depolarized due to greater 9-40 active site motion a n d / o r greater HPF flexibility within the 9-40 active site. In either case, both phenomena resulted in an apparent decrease in p since P0 was constant and the bound F! and HPF lifetimes were 2.23 ns. Utilization of fluorescent probes, which possess fluorescence lifetimes that are both sensitive to their environment and occur in a similar timescale (ps to ns) as molecular events, represent experimental approaches to measure protein structural dynamic~ t3538]. Smaller proteins such as myoglobin have been shown, using in part fluorescence techniques, to exist in multiple conformational substates [39]. Only recently, however, has the potential existence of multiple antibody active site conformational substates been addressed [40-43]. To investigate this, Fl/anti-Fi interactions have been a valuable model system since FI was both the immunogen and the active site-specific extrinsic fluorescent probe. Because FI was an inherently rigid molecule, however, HPF was employed in this study to circumvent potential active site stabilizations imposed by FI [40,42]. Use of the more flexible HPF molecule, therefore, facilitated experiments designed to observe active site dynamics. All results presented in this preliminary study were consistent with the idea that structurally similar antibody active sites may exhibit quite different dynamic properties. In the case of 4-4-20 and 9-40, although still unproven, structural differences in H CDR3 could result in a more dynamic 9-40 active site which, in turn, could affect the binding constant. This affect would be in addition to L34 Arg in the 4-4-20 active site and explained why the 9-40 H chain reassociated with the 4-4-20 L chain did not express an increased affinity for FI [12]. In the synthesis of designer antibodies, therefore, ligand and/or active site flexibility may be important [44,45]. For example, intermediate affinity antibody-antigen interactions may require either the active site or ligand to be relatively rigid; whereas high affinity antibody-antigen interactions may require both to be relatively rigid. Current experiments (e.g. low temperature and active site site-directed mutagenesis studies) are being per-
formed to further characterize the role that active site dynamics contribute to antigen recognition.
Acknowledgments Fluorescence measurements and analyses were performed at the Laboratory for Fluorescence Dynamics (LFD) at the University of Illinois Urbana-Champaign (UIUC). The LFD is supported jointly by the Division of Research Resources of the NIH (RR03155-01) and the UIUC. We thank the LFD staff (in particular Drs. Catherine Royer, Joseph Beechem, James Matthies and William Mantulin) for their help in this study. We also thank Dr. William Ware for the generous gift of HPF and Sandra Henson for expert secretarial assistance.
References 1 Tonegawa, S. (1983) Nature 302, 575-581. 2 Yancopoulos, G.D. and A!L F.W. (1986) Ann. Rev. Immunol. 4. 3~o-2~. 3 Kabat, E.A., Wu, T.T., Reid-Miller, M., Perry, H.M. and Gottesman, K.S. (1987) Sequences of Proteins of Immunological Interest, U.S. Department of Health and Human Servies, Public Health Service. National Institutes of Health, Bethesda, MD. 4 Davies, D.R.. Padlan, E.A. and Scheriff, S. (1990) Annu. Rev. Biochem. 59. 439-474. 5 Mian, I.S., Bradwell, A.R and Olson, A.J. (1991) J. Mol. Biol. 217, 133-151. e ,"nit, A.G., Mariuzza, R.A., Phillips, S.E.U. and Poljak, R.J. ~1986) S,:ience 233, 747-753. 7 Alzari, P.M., Spinelli, S., Mariuzza. R.A., Boulot, G., Poljak, R.J., Jarvis, J.M. and Milstein, C. (1990) EMBO J. 9, 3807-3814. 8 Stanfield, R.L., Fieser, T.M., Lerner, R.A. and Wilson, I.A. (1990) Science 248, 712-719. 9 Herron, H.N., He, X.-M., Mason, M.L, Voss. E.W., Jr. and Edmundson, A.B. (1989) Proteins 5, 271-280. 10 Kranz, D.M., Herron, J.N. and Voss, E.W., Jr. (1982) J. Biol. Chem. 257, 6987-6995. 11 Bates, R.M., Ballard, D.W. and Voss, E.W., Jr. (1985) Moi. lmmunol. 22, 871-877. 12 Bedzyk, W.D. and Voss, E.W., Jr. (1991) Moi. lmmunol. 28, 27-34. 13 Denzin, LK., Whitlow, M. and Voss, E.W., Jr. (1991) J. Biol. Chem. 266, 14095-14103. 14 Martin, M. and Lindqvist, L. (1975) J. Luminescence 10, 381-390. 15 Templeton, E.F.G. and Ware, W.R. (1985) Mol. lmmunol. 22, 45-55. 16 Galfre, G., Howe. S.C., Milstein, C., Butcher, G.W. and Howard, J.C. (1977) Nature 266, 550-552. 17 Kranz, D.M. and Voss, E.W., Jr. (1981) Mol. Immunol. 18, 889-898. 18 Bedzyk, W.D., Reinitz, D.M. and Voss, E.W., Jr. (1986) Mol. Immunol 23, 1319-1328. 19 Bedzyk, W.D., Herron, J.N., Edmundson, A.B. and Voss, E.W., Jr. (1990) J. Biol. Chem. 265, 133-138. 20 Kranz, D.M. and Voss, E.W., Jr. (1984) in Fluorescein Hapten: An Immunological Probe. (Voss, E.W., Jr., ed.), pp. 15-20: CRC Press, Boca Raton. 21 Reinitz, D.M. and Voss, E.W., Jr., (1985) J. lmmunol. 135, 3365-3371.
34 22 Herron, J.N. 1984. In Fluorescein Hapten: An Immunological Probe. Voss, E.W., Jr., ed.), pp. 49-76, CRC Press, Boca Raton. 23 Watt, R.M. and Voss, E.W., Jr. (1978) Immunochemistry 15, 875 -882. 24 Gratton, E. and Limkeman, M. (1983) Biophys. J. 44, 315-324. 25 Piston, D.W., Marriott, G., Radivozevieh, T., Clegg, R.M., Jovin, T.M. and Gratton, E. (1991) Rev. Sci. Inst., in press. 26 Beechem, J.M. and Gratton, E. (1988) SPIE 909, 70-81. 27 Janjic, N., Schloeder, D. and Tramontano, A. (1989) J. Am. Chem. Soc. 111, 6374-6377. 28 Holloway, H.E., Nauman, R.V. and Wharton, J.H. (1968) J. Phys. Chem. 72, 4474-4482. 29 Berlman, I.B. (1970) J. Phys. Chem. 74, 3085-3093. 30 Lancet, D. and Pecht, I. (1976) Proc. Natl. Acad. Sci. USA 73, 3548-3553. 31 Pecht, I. (1982) The Antigens 6, 1-69. 32 Perrin, F. (!929) Ann. Phys. 12, 169-175. 33 Yguerabide, J. (1972) Methods Enzymol. 26C, 498-541. 34 Yguerabide, J., Epstein, H.F. and Stryer, L. (1976) J. Mol. Biol. 51,573-590.
35 Beechem, J. and Brand, L. (1985) Biochemistry 54, 43-71. 36 Alcala, R.J., Gratton, E. and Prendergast, F.G. (1987) P,iophys. J. 51,587-596. 37 Aicala, R.J., Gratton, E. and Prendergast, F.G. (1987) Biophys. J. 5 I, 597-604. 38 Alcala, R.J., Gratton, E. and Prendergast, F.G. (1987) Biophys. J. 51,925-936. 39 Frauenfeider, H., Parak, F. and Young, R.D. (1988) Annu. Rev. Biophys. Chem. 17, 541-479. 40 Voss, E.W., Jr., Dombrink-Kurtzman, M.A. and Miklasz, S.D. (1988) Immunol. Invest. 17, 25-39. 41 Stevens, F.J., Chang, C.-H. and Schiffer, M. (1988) Proc. Natl. Acad. Sci USA 85, 6895-6898. 42 Weidner, K.M. and Voss, E.W., Jr. (1991) J. Biol. Chem. 266, 2513-2519. 43 Swindlehurst, C.A. and Voss, E.W., Jr. (1991) Biophys. J. 59, 619-628. 44 Novotny, J., Bruccoleri, R.E. and Saul, F. (1989) Biochemistry 28, 4735 -4749. 45 Novotny, J. (1991) Mol. Immunol. 28, 201-207.
