Plasticity in the primary binding site of galactose/N-acetylgalactosamine-specific lectins. Implication of the C-H...O hydrogen bond at the specificity-determining C-4 locus of the saccharide in 4-methoxygalactose recognition by jacalin and winged bean (basic) agglutinin I.

It is currently believed that an unsubstituted axial hydroxyl at the specificity-determining C-4 locus of galactose is indispensable for recognition by galactose/N-acetylgalactosamine-specific lectins. Titration calorimetry demonstrates that 4-methoxygalactose retains binding allegiance to the Moraceae lectin jacalin and the Leguminosae lectin, winged bean (basic) agglutinin (WBA I). The binding reactions were driven by dominant favorable enthalpic contributions and exhibited significant enthalpy-entropy compensation. Proton NMR titration of 4-methoxygalactose with jacalin and WBA I resulted in broadening of the sugar resonances without any change in chemical shift. The alpha- and beta-anomers of 4-methoxygalactose were found to be in slow exchange with free and lectin-bound states. Both the anomers experience magnetically equivalent environments at the respective binding sites. The binding constants derived from the dependence of NMR line widths on 4-methoxygalactose concentration agreed well with those obtained from titration calorimetry. The results unequivocally demonstrate that the loci corresponding to the axially oriented C-4 hydroxyl group of galactose within the primary binding site of these lectins exhibit plasticity. These analyses suggest, for the first time, the existence of C-H.O-type hydrogen-bond(s) in protein-carbohydrate interactions in general and between the C-4 locus of galactose derivative and the lectins jacalin and WBA I in particular.

The recognition of sugar molecules by carbohydrate-specific proteins, lectins, involves the establishment of an organized set of interactions within the binding site. Hydrogen-bonding interactions are one of the most important factors of molecular recognition in lectinϪsugar interactions, along with van der Waals forces, which, although rather weak (often contributing only a fraction of 1 kcal mol Ϫ1 for each pair of atoms), are frequently numerous and together make a significant contribution to binding (1,2). Despite the differences in the lectin folds and their modes of sugar binding, the specificity for the recognition of galactose is determined by interactions involving the C-4 locus of the saccharide (1)(2)(3)(4)(5). Stereochemical evidence is emerging for a distinct sugar binding specificity-dependent distribution of hydrogen bond donors vis à vis the acceptors in the combining site of lectins (4). The C-4 locus of the monosaccharide within the primary binding site of galactose/N-acetylgalactosamine-specific lectins has hitherto been considered to be absolutely invariant.
Though belonging to different families, jacalin, a Moraceae member, and WBA 1 I, a Leguminosae member, both are galactose/N-acetylgalactosamine-specific lectins (6,7). Whereas jacalin displays a ␤-prism tertiary structural fold wherein a unique post-translationally generated N-terminal glycine residue serves as a critical determinant of galactose specificity (8), WBA I contains a legume lectin fold (9). During the course of mapping and establishing the hydrogen bond donor-acceptor relationship of the primary combining site of galactose-specific lectins (10), unexpectedly, we have discovered that the 4-methoxy derivative of D-galactopyranoside (4-methoxygalactose) binds, with affinities comparable with that of methyl-␣-galactose, to the Moraceae lectin jacalin and the Leguminosae lectin WBA I but not to the related WBA II. 2

EXPERIMENTAL PROCEDURES
Materials-All reagents were of analytical or ultrapure grade. Methyl-␣-galactose was purchased from Sigma. 4-methoxygalactose was synthesized as described, and its purity was checked by melting point, thin-layer chromatography, and high resolution 1 H-NMR at 250 MHz on a Bruker spectrometer (10). Deionized Milli-Q water was used for all studies.
Preparation and Analysis of Solutions-All protein as well as sugar solutions were prepared in 20 mM phosphate buffer (pH 7.2) containing 150 mM sodium chloride (PBS). Jacalin (6), WBA I (7, 10), and WBA II (12) were prepared by affinity chromatography as described. Their concentrations were measured spectrophotometrically using ⑀ 280 nm 1%,1 cm ϭ 15.8, 9.37, and 7.7, respectively, determined by weight method as well as from amino acid sequence data. Jacalin is a tetramer, whereas WBA I and WBA II are dimers.
ITC Measurements and Analyses-The titration calorimetric measurements and analyses were performed with a Microcal Omega titration calorimeter as described (10,13,14). Aliquots of the sugar solution at 10 -100 times the binding site concentration were added via a 250-l rotating stirrer syringe to the lectin solution in the calorimetric cell under isothermal conditions. The dimensionless quantity c, a product of the binding constant K b and the total concentration of macromolecule [M t ] in the cell, was Ͼ2.5 but Ͻ50, corresponding to a binding regime best suited for the most precise measurements of the binding stoichi-* This work was funded by grants from the Departments of Science and Technology and Biotechnology, 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 "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Recipient of a research associateship from the Council of Scientific and Industrial Research, Government of India.
NMR Measurements and Analyses-NMR samples were prepared in PBS in D 2 O, pD 7.2. The resultant protein concentration in the samples was 0.2 mM for jacalin, WBA I, and WBA II. The concentration of 4-methoxygalactose was varied from 0.4 to 3 mM. The 400 MHz 1 H-NMR spectra of 4-methoxygalactose were recorded on a Bruker AMX-400 NMR spectrometer. The chemical shifts were referenced to the methyl signal of tetramethylsilane as internal standard. Water suppression was performed by the pre-saturation method. The assignment of 4-methoxyl signal in the NMR spectra was performed as described (15). The NMR spectra of 4-methoxygalactose are in agreement with the 4 C 1 pyranoid conformation (15,16). Line broadening of resonances was measured at half-height of the resonance under observation after correction for magnetic field inhomogeneity. The line broadening of a small molecule such as 4-methoxygalactose, because of its binding to a macromolecule such as the lectin jacalin or WBA I, was treated according to the method of Swift and Connick (17): T 2p is the reciprocal of net change in respective line width at half-height, m is the residence time of the anomers in the protein binding site, T 2m is the spin-spin relaxation time in the bound environment, [P] t is the total protein concentration, and R s ([␤]/[␣]) is the ratio of the anomers in equilibrium. A plot of reciprocal line broadening, T 2p , as a function of 4-methoxygalactose concentration displayed linear dependence for both WBA I and jacalin. The negative intercept on the x axis gives the dissociation constant (1/K b ) for the anomer, and the y intercept yields the residence time m . In the slow exchange limit ( m Ͼ ϾT 2m ), as is the case here, the line broadening is governed by the exchange rate 1/ m , which is equal to the dissociation rate constant k Ϫ1 of the anomer-lectin complex (17,18).

RESULTS AND DISCUSSION
The ITC experiments directly detected the evolution of heats of binding when fixed aliquots of 4-methoxygalactose solution were added into either jacalin solution (Fig. 1A) or WBA I solution (Fig. 1C). These data, analyzed by iterated non-linear least squares fitting procedures, were found to best fit the simplest identical site model indicating that both the lectins bind to 4-methoxygalactose ( Fig. 1, B and D). The reactions were driven by dominant favorable enthalpic contributions (Table I). These results, contrary to the usual expectation of the indispensability of the unsubstituted C-4 hydroxyl of the saccharide, clearly demonstrate the capability of 4-methoxygalactose to bind to both WBA I as well as jacalin. This reflects the existence of 4-methoxygalactose binding ability in galactose/Nacetylgalactosamine-specific lectin members from at least two unrelated families. To test whether 4-methoxygalactose displays a promiscuous binding to other lectins with similar monosaccharide specificity, the binding of 4-methoxygalactose to WBA II, a dimeric acidic lectin sequentially and structurally similar to WBA I, was tested. The results of such an ITC experiment indicate no binding of 4-methoxygalactose to WBA II (Fig. 1, E and F). That the non-binding of 4-methoxygalactose to WBA II was not due to loss of activity of WBA II was confirmed by testing the binding of 2Јfucosyllactose to WBA II; the thermodynamic parameters obtained, K b ϭ 1.0 ϫ 10 5 and ⌬H b 0 ϭ 43.2 kJ mol Ϫ1 at 298.2 K, agreed very well with previously published values (12). In contrast, none of these galactose/N-acetylgalactosamine-specific lectins, WBA I, WBA II, or jacalin, bound 4-methoxyglucose. The inherent inability of WBA II to recognize 4-methoxygalactose is perhaps due to dominant steric factors around its specificity-determining C-4 locus. The ability of 4-methoxygalactose to bind to WBA I as well as jacalin, together with its inability to bind to WBA II, is consistent with the formation of favorable interactions from the vicinity of the C-4 locus of 4-methoxygalactose with the corresponding binding site residues in WBA I and jacalin. It appears that the disposition of amino acid residues in the vicinity of the C-4 locus of 4-methoxygalactose within the binding sites of WBA I and jacalin permits the efficient binding of 4-methox-ygalactose by WBA I and jacalin. That 3-methoxygalactose and 6-methoxygalactose do not bind to either WBA I (10) or jacalin 3 indicates that this unusual binding of 4-methoxygalactose by WBA I and jacalin is due to specific interactions. In addition, WBA IϪ4-methoxygalactose as well as jacalinϪ4-methoxygalactose complexes saturated with 4-methoxygalactose did not bind to a molar excess of exogenously added methyl-␣galactose, suggesting that 4-methoxygalactose was not bound to a site other than the primary combining site of either jacalin or WBA I.
The temperature dependence of the enthalpy (⌬H 0 b ) (Fig.  1G) and entropy (⌬S 0 b ) (Fig. 1H) was linear for the binding of 4-methoxygalactose to both jacalin and WBA I. There is almost no temperature dependence of the binding free energy (i.e. ⌬⌬G 0 b Х 0) (Fig. 1I) within the temperature range examined because of significant enthalpy-entropy compensation (EEC) (Fig. 1J). EEC apparently masks the differences in binding 3 C. P. Swaminathan and A. Surolia, unpublished results. thermodynamics (14,19). A close examination of the EEC plot ( Fig. 1J) reveals that the effect of EEC, in terms of the slope of the linear plots, is nearly the same for 4-methoxygalactose binding both to jacalin and WBA I. The y intercept (i.e. a position in the EEC plot where ϪT⌬S 0 b is zero) of the EEC plot provides a measure of the condition at which all contributions from the enthalpic components proceed entirely to the free energy of the system, without any net change in entropic losses or gains. Thus, for enthalpically driven systems, such as the binding of 4-methoxygalactose to jacalin and WBA I, the differences in y intercepts of the respective EEC plots (25.2 Ϯ 0.4 kJ mol Ϫ1 for jacalin and 22.4 Ϯ 0.2 kJ mol Ϫ1 for WBA I) suggest different net contributions to the binding free energy emerging only in the form of different enthalpic components.
In 400 MHz 1 H-NMR experiments, the addition of WBA I or jacalin, but not WBA II, to a solution of 4-methoxygalactose led to broadening of the sugar resonances (Fig. 2, A-C). The broadening was due to the slow exchange ( m Ͼ ϾT 2m ) of the sugar between free and lectin-bound states as attested to by a pronounced increase in line width with increase in temperature. Hence the line broadening effects are governed by the residence time of the anomers at the binding site. Analyses of the dependence of the net change in reciprocal line broadening as a function of 4-methoxygalactose concentration yielded K b values of 7.4 ϫ 10 3 and 7.3 ϫ 10 3 for the binding of ␣and ␤-anomers, respectively, of 4-methoxygalactose to WBA I, whereas the K b for the jacalinϪ4-methoxygalactose complex was 1.5 ϫ 10 4 (Fig. 2D). These values are in the range of titration calorimetric data ( Table I). The ITC and NMR results together demonstrate unequivocally that the C-4 locus of galactose can tolerate substitution and yet not disrupt the specific carbohydrate binding ability of jacalin and WBA I, thus throwing light on the plasticity of their primary combining sites. Hence the binding site should be sufficiently flexible to accommodate this plasticity, which permits a promiscuous recognition of both galactose and 4-methoxygalactose.
These studies point to the existence of a CϪH⅐⅐⅐O hydrogen bond between the 4-methoxy group of 4-methoxygalactose and the corresponding loci in the binding site of jacalin and WBA I. The presence of oxygen atoms in a large majority of molecules raises the possibility that the CϪH⅐⅐⅐O hydrogen bond is widespread though not identified in many cases (20)(21)(22). However, x-ray and neutron diffraction studies have shown that crystals of various organic molecules and biomolecules exhibit close CϪH⅐⅐⅐X contacts (where X is an electronegative atom, in most cases oxygen) that show all the stereochemical hallmarks of hydrogen-bonds (21). Recently, CϪH⅐⅐⅐O interactions in collagen triple helix (23), DNAϪprotein complex (24), thrombin-inhibitor complex (25), trypanothione reductase-trypanothione disulfide complex (26), active sites of serine hydrolases (27), and helices involving proline residues (28) have also been identified. Cytosine-rich intercalated DNA quadruplexes not only contain intra-cytidine CϪH⅐⅐⅐O hydrogen bonds but also display a systematic intermolecular CϪH⅐⅐⅐O hydrogen bonding network between the deoxyribose sugar moieties of antiparallel backbones in the four-stranded molecule (29). More recently, CϪH⅐⅐⅐O hydrogen bonds have been found in the minor grooves of A-tracts in DNA double helices (30), and, as a caveat, in cubanecar-   (Fig. 1G). NB, non-binding. There is ample evidence that "hydrogen-bonds" are composite, multi-center interactions that span a wide range of geometry and energy, with large chemical variations among the donor (XϪH) and acceptor groups (21). The angular distributions of CϪH⅐⅐⅐O interactions for different types of CϪH groups show that the directionality decreases with decreasing CϪH polarization but is still clearly recognizable for methyl groups; on the other hand, for CϪH⅐⅐⅐HϪC van der Waals contacts, the isotropic angular characteristics are observed (32). There are many cases of multiple approaches by CϪH groups in some donor-rich systems, with as many as four CϪH groups forming contacts with a carbonyl oxygen atom (33). Based on studies on simple alkynes and alkenes, it was found that the more acidic a particular type of CϪH group, the shorter are the CϪH⅐⅐⅐O bonds it forms (21). It is interesting to note that the strength of a hydrogen bond in C'CϪH⅐⅐⅐O is close to 4 -5 kcal mol Ϫ1 , rivalling the OϪH⅐⅐⅐O hydrogen bond in water (21). The C-4 hydroxyl group serves as a donor of bidentate and tridentate hydrogen bonds in the complexes of WBA IϪmethyl-␣-galactose (Ref. 9 and Fig. 3A) and jacalinϪmethyl-␣-galactose (Ref. 8 and Fig. 3C), respectively. As neither 4-deoxygalactose nor 4-fluorodeoxygalactose binds to either WBA I (10) or to jacalin, 3 the C-4 hydroxyl group contributes significantly to the reaction, predominantly as a hydrogen bond donor. Apparently the methoxy group at the C-4 locus of 4-methoxygalactose is able to fulfill such a role. This is consistent with the relatively greater acidity of a methyl group covalently bound to an oxygen atom than of that bound to an aliphatic carbon atom. These results thus substantiate the existence of CϪH⅐⅐⅐O hydrogen bond(s) in the vicinity of the specificity-determining C-4 loci of the bound saccharide in the complexes of WBA IϪ4-methoxygalactose (Fig. 3B) and jacalinϪ4-methoxygalactose (Fig. 3D). These CϪH⅐⅐⅐O hydrogen bonding interactions involved in the recognition of 4-methoxygalactose by WBA I and jacalin appear to be stronger in magnitude (Table I) than those involved in the binding of galactose with WBA I (10) and jacalin. 3 The examples of 4-methoxygalactoseϪWBA I/jacalin interactions thus suggest that CϪH⅐⅐⅐O hydrogen bonds in lectinϪsugar interactions could contribute at least as significantly to the binding reaction as other hydrogen bonds.
Aside from the main chain carbonyl of Thr-210 in WBA I at a distance of 4.39 Å from the methyl group oxygen at the C-4 locus of the modeled 4-methoxy-␣-D-galactopyranoside, no atom other than the cognate binding site residues was present within 5.5 Å in the primary binding sites of WBA I and jacalin (11). The absence of nonpolar residues around the C-4 loci of methyl-␣-galactose-bound complexes of WBA I and jacalin provide grounds to believe that the interaction of the 4-methoxy group of 4-methoxygalactose with the binding sites of WBA I and jacalin is not of a nonpolar nature. This is also supported by the observation that these binding reactions are predominantly enthalpically driven (Table I). Moreover, the root mean square deviation of C-␣ atoms of residues around the binding sites of native WBA I compared with those of WBA I complexed with methyl-␣-galactose is less than 0.5 Å, suggesting the absence of significant conformational changes upon sugar binding.
In conclusion, we have presented here for the first time evidence for the existence in lectinϪcarbohydrate recognition of CϪH⅐⅐⅐O hydrogen bond(s) in the vicinity of the specificitydetermining C-4 locus of the saccharide 4-methoxygalactose and the lectins WBA I and jacalin.