Kinetics and mechanism of deuterium oxide-induced fluorescence enhancement of fluorescyl ligand bound to specific heterogeneous and homogeneous antibodies.

Comparative kinetics studies of ligand dissociation and D2O enhancement were performed with both heterogeneous and homogeneous anti-fluorescyl immunoglobulin G antibodies. Heterogeneous rabbit and homogeneous mouse (monoclonal) antibody preparations were purified by immunoadsorption and found to be pure IgG by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoelectrophoresis. Relatively high affinities of all liganded antibody preparations were determined by dissociation rate studies, demonstrating comparatively long lifetimes for the dissociation of bound fluorescein. In addition, rabbit anti-fluorescyl preparations were found to display marked heterogeneity of off-rates while mouse monoclonal anti-fluorescyl preparations exhibited a single off-rate indicating homogeneity. D2O fluorescence enhancement studies showed that heterogeneous kinetics was observed with both heterogeneous and homogeneous antibody active sites. Temperature studies of ligand D2O enhancement and dissociation rates using homogeneous anti-fluorescyl antibodies revealed similar, yet different activation energies (22.7 +/- 0.8 cal and 20.2 +/- 0.3 cal, respectively) for both phenomena. The studies demonstrated that the anti-fluorescein antibody active site consists of both solvent accessible and relatively inaccessible components, and that the binding of ligand involves both exchangeable hydrogen atoms and other as yet unresolved interactions. The mechanism of D2O fluorescence enhancement is discussed in terms of its complexity involving heterogeneous rate mechanisms.

Comparative kinetics studies of ligand dissociation and DzO enhancement were performed with both heterogeneous and homogeneous anti-fluorescyl immunoglobulin G antibodies. Heterogeneous rabbit and homogeneous mouse (monoclonal) antibody preparations were purified by immunoadsorption and found to be pure IgG by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoelectrophoresis. Relatively high affinities of all liganded antibody preparations were determined by dissociation rate studies, demonstrating comparatively long lifetimes for the dissociation of bound fluorescein. In addition, rabbit anti-fluorescyl preparations were found to display marked heterogeneity of off-rates while mouse monoclonal anti-fluorescyl preparations exhibited a single off-rate indicating homogeneity. DzO fluorescence enhancement studies showed that heterogeneous kinetics was observed with both heterogeneous and homogeneous antibody active sites. Temperature studies of ligand DzO enhancement and dissociation rates using homogeneous anti-fluorescyl antibodies revealed similar, yet different activation energies (22.7 2 0.8 cal and 20.2 & 0.3 cal, respectively) for both phenomena. The studies demonstrated that the anti-fluorescein antibody active site consists of both solvent accessible and relatively inaccessible components, and that the binding of ligand involves both exchangeable hydrogen atoms and other as yet unresolved interactions. The mechanism of DzO fluorescence enhancement is discussed in terms of its complexity involving heterogeneous rate mechanisms.
The fluorescence quantum yield of fluorescyl ligand (@ = 0.92) is reduced by 90% or more when bound to the antifluorescyl antibody (IgG') active site at neutral pH (1). The mechanism of fluorescence quenching has been investigated and found to be complex (2).
Recent solvent-perturbation studies demonstrated that when liganded' anti-fluorescyl IgG antibody populations were equilibrated in high concentrations of deuterium oxide (D20), partial restoration (enhancement) of the quenched ligand's fluorescence resulted (1,3). Fluorescence enhancement of antibody-bound ligand in DzO was temperature-and pHdependent (3). Low temperature (-4 "C) studies revealed * Presented in part at the 71st Annual Meeting of the American Society of Biological Chemists (1980). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  complex, heterogeneous fluorescence enhancement kinetics with purified heterogeneous high affinity rabbit anti-fluorescyl IgG antibody populations. Original observations raised the question as to whether such kinetics was due to heterogeneity of antibody active sites within a population or heterogeneous mechanisms within the antibody active site.
To better understand kinetics and mechanisms, deuterium oxide enhancement of fluorescence of bound fluorescein was studied in detail with both purified heterogeneous and homogeneous (monoclonal) anti-fluorescyl IgG antibody populations. The results are discussed in terms of the rates of solvent exchange and the influence of bound ligand upon observed kinetics. Antibody-focused solvent perturbation studies coupled with a sensitive fluorophore-ligand system, provides a suitable experimental model to study varied effects of antigen-antibody interactions.

Materials
Chemicals-The disodium salt of fluorescein was obtained from Eastman Kodak Co. Chemical purity was examined by thin layer chromatography as previously described (4). Deuterium oxide (99.8 atom 5%) was obtained from Aldrich Chemical Co. Fluorescein (isomer I)-keyhole limpet hemocyanin was synthesized by reacting fluorescein isothiocyanate (I) (Molecular Probes), with keyhole limpet hemocyanin according to the method of Voss et al. (1). Fluorescyl-conjugated proteins were purified and characterized as previously described (1). Routinely, 200 to 300 fluorescyl groups were substituted per keyhole limpet hemocyanin molecule (8.0 X IO5 daltons).
Preparation of Immunoadsorbent and Purification of Anti-fluorescyl IgG-Immunoadsorbents were synthesized as previously described (5, 6). Anti-fluorescyl IgG antibodies were purified from hyperimmune rabbit antisera or mouse ascites fluid (1). Purification involved sodium dextran sulfate precipitation of lipoproteins, 50%. saturated ammonium sulfate precipitation of y-globulins and subsequent immunoadsorption (fluorescein-Sepharose 4B). Anti-fluorescein-specific antibodies were eluted (-48 h) with 0.1 M fluorescein disodium salt and dialyzed against 50 mM phosphate buffer, pH 8.0. Preparations were subjected to anion exchange chromatography on a Dowex I-X8 (Bio-Rad) column to remove free fluorescein dianion. Eluted fractions were monitored for absorbance at 278 nm and 500 nm, and the per cent of sites containing bound fluorescein was calculated from the relative absorbances ( 1 ) .

Methods
Immunization offtabbits-Adult albino rabbits (-4 kg) were given primary and secondary immunizations (spaced 6 weeks apart) with 4 mg of fluorescein-keyhole limpet hemocyanin emulsified in complete Freund's adjuvant (Difco). Bleedings were begun 15 days post-secondary immunization. Rabbits were bled repeatedly and reimmunized over a 1 -year period.
Production of Murine Monoclonal Anti-fluorescyl Antibodies-Adult BALB/c mice were injected intraperitoneally with 0.2 ml of fluorescein-keyhole limpet hemocyanin (200 pg) emulsified in complete Freund's adjuvant. Anti-fluorescein-secreting hybridomas were 4433 DzO Fhorescence Enhancement produced by fusion of spleen cells harvested 4 days after secondary immunization with the 8-azaguanine-resistant Sp 2/0-Ag 14 myeloma cell line ( 7 ) . Approximately 1 0 ' washed spleen cells from an immunized donor were fused with lo7 myeloma cells using 50% polyethylene glycol 1500 as described by Galfre et al. (8). Cells were distributed into 48 culture wells (Costar 24 well cluster plates) and grown in hypoxanthine/aminopterin/thymidine selective medium. After 10 to 21 days, cells were grown in hypoxanthine/thymidine medium and culture supernatants were assayed 1 week later for anti-fluorescein activity.
Anti-fluorescein-secreting hybrids were detected by both a radioimmunoassay and a fluorescence-quenching assay using supernatants from each well. Four hundred pl of culture supernatant, 100 ~ of rabbit anti-mouse immunoglobulin, 50 pl of 1.0 M phosphate buffer, pH 8.0, and 10 p1 of 1251-fluorescein-bovine serum albumin were incubated at 37 "C for 1 h, 4 "C for 6 to 12 h, centrifuged, and the amount of '251-fluorescein-bovine serum albumin in the precipitate was determined in a y counter. The fluorescence-quenching assay made use of the fact that fluorescein is efficiently quenched when bound to anti-fluorescyl antibody (6). In this assay, 500 p1 of supernatant was added to 500 pl of 0.2 p~ fluorescein disodium salt in 100 mM phosphate buffer, pH 8.0. The reduction in fluorescence observed when anti-fluorescyl antibodies are present was used as a method of screening for anti-fluorescyl antibody positive wells.
Cells from anti-fluorescein positive welis were cloned on soft agar (0.2% SeaKem) using 3T6 mouse fibroblasts as a feeder layer (9). After 7 to 10 days, clones were transferred to new culture plates, grown to confluence and analyzed for anti-fluorescein activity. Clones exhibiting activity were subcultured and approximately 5 X IO6 antifluorescin-secreting hybridoma cells were injected intraperitoneally into pristane primed adult BALB/c mice to induce ascites fluid. Clones were designated by a three-number system, indicating fusion, well, and specific clone, respectively (i.e. 4-4-20 refers to the 20th clone isolated from well number 4 of the 4th fusion).
Monoclonal anti-fluorescyl antibodies were purified from ascites fluid by immunoadsorption, as described above. Purity and homogeneity of the anti-fluorescyl IgG antibody preparations were established by immunoelectrophoresis, sodium dodecyl sulfate polyacrylamide gel electrophoresis, isoelectric focusing, and fluorescyl ligandbinding studies. Data characterizing several anti-fluorescyl hybridomas will be presented elsewhere.
Purity and Characterization of Antibody Preparations"Immunoelectrophoresis was performed using either rabbit anti-mouse yglobulin or mouse anti-rabbit y-globulin as the developing agent in a system described by Watt and Voss (10). Sodium dodecyl sulfatepolyacrylamide gel electrophoresis was performed using 12% gels in the discontinuous-SDS buffer system of Laemmli (11) as described by Watt and Voss (10). Analytical isoelectric focusing was performed using a modified procedure of Briles and Davie (12). Assuming that k l = k2 = k (i.e. the anti-fluorescyl antibody active site has equivalent solvent exchange properties for D20 and HzO), the equation can be expressed in the following useful form: Finally, converting to fluorescence units (FU): where (FU)o is the initial fluorescence before D20 perturbation, (FU), is fluorescence observed at time (t), and (FU), is the maximum fluorescence observed when DzO exchange is complete. AIl data obtained were analyzed in terms of equation 4, with initial fluorescence values obtained from antibody solutions prepared in H20, and maximum fluorescence values extrapolated from double reciprocal plots of (FU)" versus (time)". Data were computed and plotted using a 9821A Hewlett-Packard calculator equipped with a 9862A plotter.
Arrhenius activation energies were determined by plotting In k versus 1/T (IO. The rate constant k was computed from the least squares slope of linear data obtained from equation 4. Dissociation Rate Studies-The dissociation rates of ligand bound to anti-fluorescyl antibody preparations were determined using the following protocol. Fifty pl of antibody (as described for D20 enhancement) was added to a cuvette at the appropriate temperature. One ml of 5-aminofluorescein (1 PM) in 50 mM phosphate buffer, pH 8.0, was added and fluorescence was monitored as described above. Since 5-aminofluorescein has a comparatively low quantum yield (0.003), yet binds anti-fluorescyl antibodies with the same affinity as fluorescein (13), one can observe an increase in fluorescence due to the dissociation of previously quenched, bound fluorescein. Fluorescein dissociation data were analyzed and plotted by first order kinetics (equation 4). Arrhenius activation energies were determined as described above.
Statistical Methods-Linear data (dissociation plots, linear portion of D20 enhancement plots, and Arrhenius plots) were regressed by least squares to obtain values for slope, x intercept, and y intercept. Standard error of slope and intercept values were determined as described by Snedecor and Cochran (14). Significance of the observed differences in Arrhenius activation energies, calculated from fluorescein dissociation and DzO enhancement experiments, was quantitated by t statistics. T values were calculated as described by Snedecor and Cochran (14) and significance levels computed with a HP-67 programmable calculator. Similarly, higher affinity antibody preparations have generally shown a greater D20 enhancement effect, It is possible that this phenomenon is due to the greater fraction of bound ligand which would be present at equilibrium with such antibodies when compared to lower affinity preparations.

Purity and Characterization
Kinetics of Fluorescence Enhancement in Deuterium Oxide-The relative fluorescence intensities of fluorescyl ligand bound to purified heterogeneous rabbit anti-fluorescyl IgG antibodies was monitored over time in DzO relative to H20 (Fig. I). D2O fluorescence enhancement appeared to consist of heterogeneous (Le. nonlinear) rate kinetics (Fig. 1). Since the kinetics of D20 enhancement involved more than a single rate, the question remained whether the complex kinetics was due to heterogeneity of the active sites or mechanisms within a single site. To resolve the problem, D20 fluorescence en-hancement studies were performed with homogeneous antibodies (z.e. populations in which all sites are identical). D20 enhancement plots for monoclonal antibody preparation 4-4-20 (Fig. 2) exhibited complex, heterogeneous kinetics, with two distinct components observed. The initial component was heterogeneous and accounted for 34.5 ? 3.2% of the D20 enhancement effect. The second component appeared fairly homogeneous, and was regressed by least squares, yielding a correlation coefficient of 298%. Rate constants, determined from least squares slope, exhibited a pronounced temperature effect varying from 2.4 X s-', at 4 "C to 0.35 s-' at 60 "C. For monoclonal preparation 4-4-20, average initial component kinetics was faster than ligand dissociation kinetics (see below), while the observed second component rate was t20% of the ligand dissociation rate.
Heterogeneous anti-fluorescyl antibody exhibited heterogeneous D20 enhancement kinetics, at all times and temperatures assayed. For a given preparation, D2O enhancement kinetics was always faster than ligand dissociation kinetics (see below). Dissociation Rates of Heterogeneous a n d Homogeneous Anti-fluorescyl Antibodies-To substantiate that fluorescence enhancement in D20 was not simply a dissociation of bound ligand in D2O ( l ) , and to further characterize the heterogeneity and/or homogeneity of these antibody preparations, dissociation rate studies were performed. Dissociation rates were determined by monitoring the increase in fluorescence of dissociated fluorescein (Q = 0.92) in the presence of a 10-fold excess of 5-aminofluorescein (@ = 0.003). Rate plots of ligand dissociation for heterogeneous and homogeneous IgG preparations are also shown in Figs. 1 and 2, respectively. Heterogeneous antibodies (Fig. 1) demonstrated a nonlinear curve, indicative of multiple populations of fluorescein-specific antibodies with different off-rates. However, the obvious differences between curves of D20 enhancement and ligand dissociation indicates that the D20 phenomena is not due to DnO-induced ligand dissociation. Homogeneous antibodies (4-4-20, Fig. 2) exhibited a linear curve, consistent with a homogeneous dissociation rate. Thus, it was further demonstrated that the D20 enhancement effect (heterogeneous) involves mechanisms distinct from ligand dissociation (homogeneous).
Fluorescence enhancement due to DZO perturbation or ligand dissociation was quantitatively compared for heterogeneous and homogeneous preparations (Table I). Both antibody preparations (heterogeneous and homogeneous) exhibited significantly greater per cent enhancements upon ligand dissociation than D20 perturbation. Such results might be anticipated if one considers that the dissociation of bound fluorescein results in the alleviation of all quenching mechanisms, while D20 enhancement is due to certain specific interactions (Le. hydrogen exchange) resulting in only partial relief of quenching mechanisms.
Temperature-dependent Kinetics of D20 Enhancement and Ligand Dissociation-It had been reported previously that D,O enhancement was a temperature-dependent phenomenon (I). The rate of D20 enhancement and ligand dissociation was monitored at different temperatures for both heterogeneous (Fig. 1) and homogeneous (Fig. 2 ) IgG antibody preparations. Both D,O enhancement and ligand dissociation were found to be temperature-dependent phenomena. In order to quantitate temperature effects, monoclonal antibody prep-  aration 4-4-20 was examined a t several temperatures (Table  11) and Arrhenius plots of both D20 enhancement and ligand dissociation were constructed (Fig. 3). Activation energies of DzO enhancement and ligand dissociation were calculated to be 22.70 +-0.86 and 20.24 k 0.33 cal/mol, respectively. The 2.46 cal/moI difference in these values was found significant at a 98.5% level by t statistics.

DISCUSSION
Previous experiments using deuterium exchange have been performed to examine the solvent accessibility of hydrogen atoms located on native versus denatured proteins (16). Results using native insulin (17) have indicated that approximately 70% of the exchangeable hydrogen atoms undergo instantaneous exchange, while 30% are relatively solvent-inaccessible and undergo a slow rate of exchange over a period of hours. Such experiments have aided in the evaluation of a protein's native structure and conformation. The work presented in this paper has extended the application of deuterium exchange to the study of the antibody active site. These studies have illustrated the utility of deuterium oxide in probing the accessibility of the anti-fluorescyl active site to bulk solvent. Deuterium oxide is superior in several respects to other small molecules, like iodide anion (4) and 0 2 (181, which have previously been employed in solvent accessibility studies. First, buffered deuterium oxide bulk solvent provides a more native environment than the 110 mM concentrations of iodide anion or O2 required for their respective perturbation effects. Second, fluorescence enhancement by deuterium oxide is a more sensitive technique than either iodide or oxygen quenching. Aside from the obvious advantage of fluorescence enhancement, rather than quenching of the bound (and quenched) fluorophore, deuterium oxide perturbs only antibody-bound fluorescein, while ioide and oxygen perturb both bound and free species. However, it is likely that the method of D20 enhancement of bound fluorescein is unique to fluo-rophores, like fluorescein, which exhibit quantum yields near unity in aqueous milieu. Finally, the use of iodide in probing the accessibility of antibody active sites to bulk solvent is of limited value, since the iodide anion, by virtue of its charge and molecular weight, may interact with antibody molecules in a manner different than bulk solvent.
Neither solvent peturbation nor isotope exchange has been extensively used in studying the mechanisms of antigen-antibody interactions. Application of these techniques in conjunction with the sensitive fluorescent antigenic probe, fluorescein, has permitted the experimental examination of reaction mechanisms. In this report, previous observations have been extended which showed that the low fluorescence quantum yield of antibody-bound fluorescyl ligand can be partially restored in high concentrations of deuterium oxide (1, 3).
In order to gain insight into the effect of D20, the kinetics of D 2 0 fluorescence enhancement was defined using heterogeneous and homogeneous anti-fluorescyl IgG antibody preparations. Fluorescyl ligand bound to the active sites of heterogeneous rabbit antibodies has been shown to be at least 90% quenched (relative to free fluorescein). It has now been observed that the fluorescence quenching associated with specific binding to the active sites of anti-fluorescyl antibodies can be partially relieved in the presence of DzO. This phenomenon has been demonstrated with over 20 heterogeneous and 7 homogeneous IgG antibody preparations, some of which are presented in this report. The extent of this enhancement, in addition to distinctly different rate plots, lent credence to previous observations (1) that the D20 effect is not due to DzO-induced dissociation of ligand. D 2 0 enhancement and ligand dissociation experiments with heterogeneous antibodies were found to follow dissimilar, complex kinetics. T h e heterogeneity of ligand dissociation has been shown to be due to the presence of multiple subpopulations of antibodies with varying off-rates (19). In order to approach the question of whether the D2O effect was due to such subpopulations, we examined monoclonal antibodies, which are homogeneous with respect to their binding sites.
Two monoclonal anti-fluorescein clones (4-4-20 and 4-6-10) exhibited distinctly different D2O fluorescence enhancement properties. Similarly, variations in the DzO enhancement effect have been observed with heterogeneous anti-fluorescyl preparations (Table I). Such differences appear to be indicative of the diversity of mechanisms involved in the binding of fluorescyl ligand by the antibody active site. The sensitivity of isotope (DzO) exchange for probing subtle differences between antibody molecules binding the same ligand offers a unique means by which to analyze these binding mechanisms. The studies reported here have compared the kinetics of the D20 phenomenon for heterogeneous and homogeneous Ig preparations. These results are currently being used as a basis for further comparisons of different monoclonal anti-fluorescyl antibody active sites. Information which is potentially available from this method includes the relative contribution of hydrogen bonding, solvent accessibility of bound ligand and its microenvironment.
In addition to significant D20 enhancement effects, ligand dissociation from monoclonal preparation 4-4-20 exhibited a relatively long lifetime (>6 min) at 4 "C, thereby allowing accurate measurements of temperature effects. Linear dissociation rate plots confirmed the homogeneity of this preparation and allowed us to approach the mechanism of D 2 0 enhancement. The kinetics of DzO fluorescence enhancement were decidedly heterogeneous, substantiating the observation that the D20 effect is not due to a simple dissociative event and indicating that the heterogeneous D20 kinetics is not due to different subpopulations of antibodies.
Heterogeneity of DzO enhancement plots, obtained from monoclonal preparation 4-4-20 (Fig. l), was more pronounced, initially, than at longer times. The initial heterogeneous enhancement exhibited faster average kinetics than observed for ligand dissociation and accounted for -34.5% of the total observed DzO enhancement effect. Subsequent deuterium enhancement exhibited relatively homogeneous kinetics with a rate significantly less (<20%) than observed for ligand dissociation, and accounted for the remaining 65.5% of the observed enhancement effect. Linear regression by least squares of the slow component showed nearly homogeneous kinetics (correlation coefficient 98%), but since the precision of the assay was -1-2%, the presence of very slow components could not have been detected under these conditions. Arrhenius activation energies were determined for both fluorescein dissociation and for the linear component of DzO enhancement (Fig. 3). Activation energies of 20.2 f 0.3 cal/ mol and 22.7 k 0.8 cal/mol were determined for fluorescein dissociation and D 2 0 enhancement, respectively. The 2.5 cal/ mol difference in these values was significant to >98.5% (by t statistics). However, since both activation energies were similar, the mechanism of D20 enhancement (of the slow component) probably involves a dissociative event. The rate of D 2 0 enhancement of the slow component is ~2 0 % of the dissociation rate, implying that ligand dissociation is necessary, but not sufficient for D 2 0 enhancement. The additional 2.5 cal/mol required for D 2 0 enhancement probably accounts for the remaining events (such as deuterium-proton exchange) necessary for enhancement. Ligand dissociation was not required for the initial heterogeneous D2O enhancement component, since the observed enhancement rate was much faster than the ligand dissociation rate.
In summary, D20 enhancement kinetics, for monoclonal preparation 4-4-20, contained two macroscopic components: a relatively heterogeneous rapid component that accounted for 34.5% of the observed enhancement effect and did not require ligand dissociation, and a more homogeneous slow component for which ligand dissociation was necessary but not sufficient.
It is presumed that deuterium exchange observed in these studies involves both solvent accessibility of exchangeable hydrogen atoms, as with the original work on insulin (17), and various binding mechanisms involving exchangeable protons within the active site. Therefore, the complexity of D 2 0 enhancement kinetics most likely results from the combined effects of solvent accessibility and fluorescyl ligand binding within the active site of the antibody molecule. The rapid component (34.5% enhancement) may involve the former effect while the slower component (65.5% enhancement) may involve both effects.
Finally, it has become apparent that D,O enhancement is a complex phenomenon, involving heterogeneous rate mechanisms. The effect is dependent on the internal diffusion of D20 and the solvent exchange with protons involved in the fluorescein-quenching mechanisms within the active site.
In addition, based on the total per cent fluorescence enhancement achievable in the dissociation experiment (-500%) and that observed with the same preparation with D 2 0 (130%), it becomes obvious that the quenching mechanisms within an active site are many and varied. It is likely that protonation of the fluorescyl ligand within the active site is only one means by which the fluorescence of bound fluorescein may be quenched. In addition, to a variety of quenching, and hence binding modes within a single active site, we have now observed numerous differences in the affinities and binding mechanisms of different monoclonal preparations." We are Unpublished data.

I)ZO Fluorescence Enhancement
continuing to use the method of deuterium exchange to further elucidate and compare the nature of the combining sites of these anti-fluorescyl antibody clones.