Analysis of Dynamics and Mechanism of Ligand Binding to Artocarpus integrifolia Agglutinin A 13C AND 19F NMR STUDY*

Binding of 13C-labeled N-acetylgalactosamine (13C- GalNAc) and N-trifluoroacetylgalactosamine (lSF-GalNAc) to Artocarpus integrifolia agglutinin has been studied using 13C and "F nuclear magnetic resonance spectroscopy, respectively. Binding of these sac- charides resulted in broadening of the resonances, and no change in chemical shift was observed, suggesting that the a- and b-anomers of 13C-GalNAc and "F-GalNAc experience a magnetically equivalent environment in the lectin combining site. The a- and b-anomers of I3C-GalNAc and "F-GalNAc were found to be in slow exchange between free and protein bound states. Binding of 13C-GalNAc was studied as a function of temperature. From the temperature dependence of the line broadening, the thermodynamic and kinetic parameters were evaluated. The association rate constants obtained for the a-anomers of 13C-GalNAc and "F-GalNAc (k+l = 1.01 X 10' M-~*s" and 0.698 X lo5 M"*s", respectively) are in close agreement with those obtained for the corresponding 8-anomers (k+l = 0.95 X 10' M"*s" and 0.65 X 10' "'Os-', respectively), suggesting that the preventing bacterial growth and 25% D,O for field frequency locking of the spectrometer. I3C NMR spectra were recorded in tubes of 10-mm diameter on a Bruker WH-270 spectrometer at 67.8 MHz with quadrature detection. "F spectra were obtained using 5-mm tubes on a Varian FT 80A spectrometer. Equi-molar concentrations of p-dioxane and trifluoroethanol were used as internal reference for obtaining I3C and "F NMR spectra, respec-tively. In the case of 13C NMR spectra, inverse gated decoupling was done with 3.0-s delay time in order to avoid undue heating of the sample. 13C spectra with good signal to noise ratio were obtained within 6-8 h of accumulation. Experiments were carried out at different temperatures using a Bruker B-ST 100/700 temperature control unit. Line broadening of resonances was measured at half-height of the resonance under observation. Observed resonances were corrected for magnetic field inhomogeneity by using the width at half-height of the resonances of p-dioxane and trifluoroethanol in I3C and "F NMR experiments, respectively.

Lectins, due to their ability to bind to cell-surface carbohydrates, have become widely used tools for exploring the structure and dynamics of cell surfaces (1,2). An elucidation of the specificity and mechanism of saccharide binding to the lectins is necessary for understanding the mechanism of their interaction with cell-surface receptors. Lectin-sugar interac-* This work was supported by a grant from Department of Science and Technology, Government of India (to A. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Senior research fellow of University Grants Commission, India. Research Associate in a project funded by Department of Science and Technology, India (to A. S.). tions have been studied using uv absorption, fluorescence, stop-flow, and relaxation techniques. In such studies a chromophore or a fluorophore is usually attached to the ligand which may perturb the specificities of interactions. On the other hand nuclear magnetic resonance (NMR) spectroscopy is a convenient tool to study the dynamics of ligand binding to proteins (3). It has been particularly useful to study the nature of lectin-sugar interactions (4-8), using naturally occurring carbohydrates, by monitoring the changes in NMR parameters such as chemical shifts, change in line widths of the observed resonances, and spin-lattice relaxation times (Tl). In NMR spectroscopy the use of 13C and "F nuclei in the studies of protein-ligand interactions is more useful than that of the 'H nucleus because the former nuclei have larger chemical shifts and, therefore, yield simple spectra. However, the use of 13C is severely hampered because of its low natural abundance and poor sensitivity. Therefore, specific enrichment a t a required position is necessary for effective use of this isotope.
Artocarpus integrifolia lectin was shown by us to be specific for l-O-a-methylgalactose and T-antigenic disaccharide by fluorescence spectroscopy (9, 10). We have used N-acetylgalactosamine (GalNAc)' specifically 13C-labeled at carbonyl carbon and N-trifluoroacetylgalactosamine to study their binding to A . integrifolia lectin using 13C and "F NMR spectroscopy. The change in line widths of the resonances of aand p-anomers of GalNAc upon binding to the lectin has been used to obtain the activation parameters for the binding of these anomers to A. integrifolia lectin. The results suggest that the binding of these anomers is characterized by large activation entropy and a two-step binding process.

Purification of A . integrifolia Lectin-A. integrifolia
lectin was purified by affinity chromatography on cross-linked guar gum as reported (9, 11). Protein concentration was estimated according to the method of Lowry et al. (12).
Chemical Modification of Tyrosyl Side Chains of A . integrifolia Lectin by N-Acetylimidazole-Tyrosyl side chains of A . integrifolia lectin (5-6 mg/ml) were modified with 300-fold molar excess of Nacetylimidazole in 200 mM sodium phosphate buffer, pH 7.3, at 5 "C for 1 h. The protein was passed through a Sephadex G-25 column and dialyzed and concentrated by lyophilization.
Treatment for Demetalization of A . integrifolia Lectin-All glassware were washed with deionized water. Phosphate-buffered saline, p H 7.3, was prepared and passed through Chelex 100 (15)  varied from 4 to 12 mM and 7 to 26 mM, respectively. All samples contained 0.03% azide for preventing bacterial growth and 25% D,O for field frequency locking of the spectrometer. I3C NMR spectra were recorded in tubes of 10-mm diameter on a Bruker WH-270 spectrometer a t 67.8 MHz with quadrature detection. "F spectra were obtained using 5-mm tubes on a Varian FT 80A spectrometer. Equimolar concentrations of p-dioxane and trifluoroethanol were used as internal reference for obtaining I3C and "F NMR spectra, respectively. In the case of 13C NMR spectra, inverse gated decoupling was done with 3.0-s delay time in order to avoid undue heating of the sample. 13C spectra with good signal to noise ratio were obtained within 6-8 h of accumulation. Experiments were carried out at different temperatures using a Bruker B-ST 100/700 temperature control unit. Line broadening of resonances was measured a t halfheight of the resonance under observation. Observed resonances were corrected for magnetic field inhomogeneity by using the width at half-height of the resonances of p-dioxane and trifluoroethanol in I3C and "F NMR experiments, respectively.

RESULTS
The 13C and "F NMR spectra of 13C-GalNAc and NTFAGalN in the absence and in the presence of A. integrifolia lectin are shown in Figs. 1 and 2, respectively. The

FIG. 1. Proton-decoupled 13C NMR spectra of '%-labeled N-acetylgalactosamine in the absence ( a and b ) and presence of A . integrifolia lectin ( c ) .
A 6.0 mM concentration of GalNAc (90% "C-enriched a t carbonyl carbon) was used, and the protein concentration was 626 VM in tetramer. Line widths at half-height in expanded spectra correspond to 7.2 Hz for free sugar ( b ) and 11.2 and 10.8 Hz for a-and 0-anomers, respectively. Digital line broadening of 2 Hz was applied in each case. The resonance at high field in Q is due to the internal standard, p-dioxane.

FIG. 2. ''F spectra of N-trifluoroacetylgalactosamine in the absence ( a and b) and presence of A. integrifolia lectin (c). A
9.3 mM concentration of "F-GalNAc was used, and the protein concentration was 523 WM in tetramer. Line widths a t half-height correspond to 1.3 Hz for free sugar ( b ) and 4.2 and 3.8 Hz for a-and 6-anomers, respectively, in the bound form. Triplet a t high field in a is due to trifluoroethanol internal standard.
Chemical shifts were listed assigning the middle resonance as 0.0. resonances to the upfield were assigned as Pand those of the downfield as a (16). The resonances of both the anomers broadened only in the presence of the lectin. We argue that this line broadening emanates from the specific binding of the saccharide(s) to the lectin. This was established by the following criteria. 1) The line broadening varies as a function of total concentration of the labeled saccharide. 2) In the presence of an excess unlabeled competing saccharide (in this case 1-0-a-methylgalactose), the line broadening was substantially diminished (=80-85%).
3) The line width of the resonances of 13C-GalNAc as well as NTFAGalN remains unchanged when a chemically modified protein was used. 4) The line width of resonances of a-and 0-anomers of 13C-GlcNAc as well as NTFAGlcN remains unchanged in the presence of 1.4 mM A. integrifolia lectin. These four control experiments not only clearly demonstrate the fact that the line broadening occurs only as a consequence of specific binding of these anomers ( a and P) to the lectin but also emphasizes the fact that the phenomenon of line broadening is not due to a change in viscosity of the sample due to the addition of large macromolecules or due to the presence of metal ions sticking to the surface of the protein. Identical line width for the a-and P-anomers of NTFAGalN was observed when protein subjected to demetalization was compared with untreated protein of identical concentration. Moreover, the line width of the a-and P-anomers remained unchanged when the spectrum was recorded in presence of 3 mM EDTA. These observations taken together with our failure to detect transi-tion metals in the protein suggest that metal ions have no role in the line broadening of the anomers as observed in the present study. In the present study both the anomers, viz. a and / 3 of I3C-GalNAc, as well as NTFAGalN, are in equilibrium with the protein. Thus it is possible to evaluate the kinetic parameters for each of the anomers as outlined below.
As the observed line widths are characteristic of the anomer present, it is possible to estimate the ratio of a-and p-anomers by measuring the area under each resonance. The possible equilibria within the sample tube are From the above equations, the total protein concentration where [PI, is total protein concentration, [P]f is free protein (or unbound), [Pa] and [PPI represent protein bound to the a-and @-anomers, respectively, of the ligand L. Similarly for B The line broadening of a small molecule due to its binding to a macromolecule can be treated according to the method of Swift and Connick (17). For a molecule undergoing chemical exchange between two sites, i.e. between free and bound to a macromolecule, the spin-spin relaxation rate is given by where f is the fraction of small molecule bound to the macromolecule (in this case it is a-and p-anomers bound to the A. integrifolia lectin), 7, is the residence time of the respective anomer in the protein binding site, and Tzm is the spin-spin relaxation time in the bound environment. In the fast exchange limit (TZm > > 7,,,) one observes the average width of the free and bound saccharide, while in slow exchange limit (7, >> TZm) the line broadening is governed by the exchange rate 1/~,,,, which is equal to the dissociation rate constant k-l of the anomer-protein complex (3). In this study, it has been observed that the anomers are in slow exchange with the protein as the line width at half-height of the a-and @anomers increases with increase in temperature (Fig. 3). The net change in line width at half-height (l/Tzp) of the resonance is given by where l/Tz is the observed line width in the presence of the protein, and l/Tzc is the line width measured for the resonance in the absence of the protein. Substituting for f with respective fraction of bound with f a or f B in Equation 9 This equation is essentially the same as the one derived by Representative plots are shown in Figs. 4 and 5 for the aanomer of 13C-GalNAc and NTFAGalN, respectively. The association rate constant was evaluated by the relationship k+l = K , X k-l. The values obtained for the binding of a-and @-anomers of 13C-GalNAc as well as NTFAGalN to A. integrifolia lectin are listed in Table I. Experiments were performed at temperatures ranging from 10 to 20°C. The association constants and dissociation rate constants were evaluated as mentioned above. From the temperature-dependent association and dissociation rates (Figs. 6 and 7), the thermodynamic parameters for the binding of a-and P-anomers of 13C-GalNAc to A. integrifolia lectin were    Table  11).

TABLE I1 Thermodynamic parameters for the binding of a-and 8-anomers of I3C-GalNAc to A. integrifolia lectin
Units: for Ai3 and AH, kJ . mol" and for AS, J . mol". K" (data calculated for 20 "C).

I n ( k / T )
where AH, AG, and A S are enthalpy, free energy, and entropy, respectively, k is the appropriate rate constant, k' is the Boltzman constant, and h is the Planck's constant.

DISCUSSION
The control experiments as described under "Results" show that the observed broadening is due to specific binding of cy- The rate parameters can be determined more accurately by stop-flow or temperature-jump studies; however, these methods necessitate the introduction of chromogenic reporter group(s) in the saccharide which may cause some perturbation and/or significantly alter the mechanism. On the other hand the 13C-and lgF-labeled ligands represent more closely the naturally occurring carbohydrates, and hence the information obtained is near to the physiological situation. Moreover, in this study it has been possible to elucidate the dynamics of binding of a-and P-anomers to the lectin, which was not feasible with the former techniques. On the other hand the elucidation of the dynamics of sugar binding to lectins by following the changes in the intrinsic fluorescence of the protein has not been possible due to a lack of correlation between the ligand binding and fluorescence change (20).
The temperature-dependent changes in the line widths of the resonances reveal that these anomers are in slow exchange with the protein (7, >> 2' 2,). If the anomers were in fast exchange with the protein one would expect a decrease in line width with increase in temperature (6, 21). Since our studies show a pronounced increase in line width with increase in temperature, the line broadening effects are governed by the residence time (7, >> 2'2,) of the anomers at the binding site, and hence the kinetic parameters for the interaction of the lectin with sugar could be evaluated as per the procedure of Swift and Connick (17). This condition enabled the determination of the residence times of the anomers in the proteinbinding site (7, = 1/kl). The association constants for both 13C-GalNAc and NTFAGalN are in good agreement with the earlier studies using fluorescence spectroscopy (9, 10). It can be seen from Table I that the association constants as well as the association and dissociation rate constants obtained for the a-and P-anomers of I3C-GalNAc are similar to those obtained for the a-and P-anomers of NTFAGalN. The thermodynamic parameters, AHo and ASo, obtained for the two anomers of 13C-GalNAc are, however, at some variance with those obtained earlier ( A H " = -55.73 kJ-mol-'; A S = -119 J . mol". K-') by the fluorescence method (see Table 11). This could arise because of the difference in the principles in the two techniques (in the NMR method determination of association constant is direct, whereas in the fluorescence method it is indirect). In order to verify the values using another direct method, we have determined these values for the interaction of GalNAc (mixture of anomers) by monitoring the ligand-induced changes in the protein fluorescence. The values obtained (AH' = -44.87 kJ. mol" and A S = -87.58 J . mol". K-') are in better agreement with the values obtained from the NMR experiments (for example A H o = -37.3 kJ. mol" and A S = -72.48 J. mol" for the a-anomer). A reasonable agreement between the thermodynamic data obtained for the binding of GalNAc by intrinsic fluorescence and those determined by NMR measurements suggests that NMR parameters reflect the events related with the association of the lectin with the anomers. Examination of Table I1 reveals that the association of these anomers with the lectin is entropic and the activation enthalpy needed is small (AH,, = 8.68 kJ. mol" and 8.18 kJ .mol" for the a-and p-anomers, respectively). The large enthalpy of dissociation ( A H o f f = 45.85 kJ. mol" and 38.3 kJ . mol" for the a-and p-anomers, respectively) could be due to an energetic requirement to break hydrogen bonds between the protein and the saccharide during the dissociation process. The situation found here is similar to that observed for the binding of 13C-labeled 1-O-amethylgalactose and its @-methyl derivative to peanut agglu-tinin (8). The large activation entropy values obtained for the a-and P-anomers of 13C-GalNAc (AS,, = -119.38 J.mol-'. K" and -121.0 J.mol".K", respectively) could be due to the requirement of specific configuration of the reactants during the association process.
The activation energies required for the association of aand @-anomers of 13C-GalNAc to the lectin are substantially smaller than observed for the association of N-dansylgalactosamine and A. integrifolia lectin ( A H o n = 56.4 k J . mol", AS,, = 59.94 J .mol". K-', see Ref. 9). This suggests that a large part of the activation energy is probably expended in fitting the bulky "dansyl" substituent of DnsGalN in the binding pocket of the protein.
The association rate constants (k,, = 1.0 X lo5 M" . s-, and 0.96 X lo5 M" . s" for the a-and 0-anomers of 13C-GalNAc and k+l = 0.68 X lo5 M".s" and 0.53 X lo5 M".s" for the a-and P-anomers of NTFAGalN) obtained here are in the range of those obtained for several other lectin-sugar interactions reported (22)(23)(24)(25). The replacement of CHs by CF3 did not change the binding process markedly as the association constants and association and dissociation rate constants are comparable ( Table I).
The association rate constants obtained in this study are several orders of magnitude slower than those observed for diffusion controlled reactions. Moreover, the activation energy for the binding reaction is very much higher than that expected for diffusion controlled processes. This activation barrier is essentially due to a high activation entropy term and the enthalpic barrier is not significant. The large negative activation entropy suggests that the reaction is limited by steric factors. In other words, the reaction requires specific configuration of reactants. The slow bimolecular association rate constants coupled with high activation energy term, mostly comprised of activation entropy, indicate that the binding of GalNAc to the lectin cannot be explained by a simple bimolecular reaction but involves a more complex mechanism. A two-step process as outlined in the expressions (26, 27) below offers a satisfactory explanation for these observations k+l k+z k-2 P + L ' k _ l P L ; -P L If the first step is substantially faster than the second step and if we assume that k+z >> k-2 then the residence time of the saccharide on the protein is given by In other words the second step is the rate-limiting step, An alternative explanation for the slow association reaction assumes the presence of two conformational states of the protein (P and P*), only one of which (P*) binds to the saccharide. The saccharide binding induces a shift in the equilibrium between the two conformers toward the one which binds the protein as shown below: P + P', P* + L + PL* However, such a mechanism is very unlikely here in view of the linearity of the ln(K/T) uersus (l/T) plots (Figs. 6 and  7). This mechanism is also ruled out by the fact that similar values of k+l and k-l were obtained when protein concentration was varied between 0.3 and 2 mM. If this mechanism were to be operative then one would have expected an increase in k+l with increase in the concentration of the lectin.
The data obtained here suggest a two-step mechanism based on the following observations. The first step ( i e . formation of PLi), which is probably a diffusion controlled reaction, would be a loose association of the saccharide with the protein.
The second step is a mutual fitting of the saccharide in the lectin-binding site. The step is associated with a large activation entropy, due to the requirement of a specific orientation of the reactants for the association process. Our failure to detect the intermediate complex [PL]i could be due to the fact that k+2 > k l , and, therefore, it is possible to observe only the overall reaction (28).
It is interesting to compare the binding of a-and B-anomers to the A. integrifolia lectin with that of the binding of NTFAGlcN to concanavalin A (6) and the binding of GlcNAc to lysozyme (29). In the case of GlcNAc binding to lysozyme, the acetamido groups of a-and p-anomers of GlcNAc experience a magnetically nonequivalent environment in the enzyme binding site resulting in a chemical shift of the acetamido resonances to varying degrees (29), whereas in the case of binding of a-and p-anomers of NTFAGlcN to ConA, no shift in the resonances of either of the anomers was observed.
In the present study, the a-and P-anomers and 13C-GalNAc as well as NTFAGalN do not show any chemical shift difference between free and bound states. This observation in conjunction with the similar association and dissociation rate constants for the binding of a-and p-anomers of 13C-GalNAc as well as NTFAGalN to A. integrifolia lectin suggests that the acetamido groups of a-and P-anomers experience a magnetically equivalent environment in contrast to the binding of a-and p-anomers of GalNAc to lysozyme.