Differential solvation of "core" trimannoside complexes of the Dioclea grandiflora lectin and concanavalin A detected by primary solvent isotope effects in isothermal titration microcalorimetry.

The thermodynamics of binding of the Man/Glc-specific seed lectin from Dioclea grandiflora (DGL) to deoxy analogs of the "core" trimannoside, 3, 6-di-O-(alpha-D-mannopyranosyl)-alpha-D-mannopyranoside was determined by isothermal titration microcalorimetry (ITC) in the first paper of this series (Dam, T. K., Oscarson, S., and Brewer, C. F. (1998) J. Biol. Chem. 273, 32812-32817). The data showed binding of specific hydroxyl groups on all three residues of the trimannoside, similar to that observed for ConA (Gupta, D., Dam, T. K., Oscarson, S., and Brewer, C. F. (1997) J. Biol. Chem. 272, 6388-6392). However, differences exist in the thermodynamics of binding of monodeoxy analogs of the alpha(1-6) Man residue of the trimannoside to the two lectins. The x-ray crystal structure of DGL complexed to the core trimannoside, presented in the second paper in this series (Rozwarski, D. A., Swami, B. M., Brewer, C. F., and Sacchettini, J. C. (1998) J. Biol. Chem. 273, 32818-32825), showed the overall structure of the complex to be similar to that of the ConA-trimannoside complex. Furthermore, the trimannoside is involved in nearly identical hydrogen bonding interactions in both complexes. However, differences were noted in the arrangement of ordered water molecules in the binding sites of the two lectins. The present study presents ITC measurements of DGL and ConA binding to the monodeoxy analogs of the trimannoside in hydrogen oxide (H2O) and deuterium oxide (D2O). The solvent isotope effects present in the thermodynamic binding data provide evidence for altered solvation of the parent trimannoside complexes at sites consistent with the x-ray crystal structures of both lectins. The results indicate that the differences in the thermodynamics of DGL and ConA binding to alpha(1-6) monodeoxy analogs of the trimannoside do not correlate with solvation differences of the parent trimannoside complexes.

In the first paper in this series (15), the thermodynamics of binding of the Man/Glc-specific seed lectin from Dioclea grandiflora (DGL) 1 to deoxy analogs of the "core" trimannoside, 3,6-di-O-(␣-D-mannopyranosyl)-␣-D-mannopyranoside, was determined by isothermal titration microcalorimetry (ITC). The results showed evidence for the binding of the 2-, 3-, 4-, and 6-hydroxyl groups of the ␣(1,6) Man residue, the 3-and 4-hydroxyl groups of the ␣(1,3) Man residue, and the 2-and 4-hydroxyl groups of the central Man residue of the trimannoside. The ITC results are similar to those observed for the jack bean lectin, concanavalin A (ConA) (1), which also binds the trimannoside with high affinity; however, differences exist in the thermodynamics of binding of monodeoxy analogs of the ␣(1-6) Man residue of the trimannoside to the two lectins. The loss in the enthalpy of binding (⌬⌬H) of these analogs to DGL, relative to the parent trimannoside, is nearly 3 kcal mol Ϫ1 greater than that for ConA (1). Furthermore, the loss in ⌬⌬H for DGL binding to the 2-deoxy ␣(1-6) analog is not observed for ConA binding to the analog (1).
In the second paper in this series, the x-ray crystal structure of DGL complexed to the parent trimannoside (16) shows that the overall structure of the complex is similar to that of the ConA-trimannoside complex (2). The average deviation in ␣-carbon position between the DGL tetramer in complex with the trimannoside and the ConA tetramer in complex with the trisaccharide is only 0.84 Å (16). Furthermore, the location and conformation of the bound trimannoside as well as its hydrogen bonding interactions are nearly identical in both lectin complexes. However, differences exist in the location of two loops outside of the respective binding sites containing residues 114 -125 and residues 222-227. The latter residues effect the location of a network of hydrogen-bonded water molecules that interact with the trisaccharide. Differences in the arrangement of ordered water molecules in the binding site of the two lectins may account for their differences in the thermodynamics of binding to deoxy ␣(1-6) analogs of the trimannoside (15).
In the present paper, ITC studies of the binding of DGL and ConA to mono-and disaccharides, and to trimannoside 1 and monodeoxy analogs 2-11 ( Fig. 1) in hydrogen oxide (H 2 O) and deuterium oxide (D 2 O) are reported. The results show primary solvent isotope effects in the ⌬H values of both lectins binding to the carbohydrates that correlate with the altered solvation observed in the x-ray structures of the two lectins complexed to the parent trimannoside. EXPERIMENTAL PROCEDURES DGL was isolated from Dioclea grandiflora seeds obtained from northeastern Brazil (Albano Ferreira Martin Ltd., Sao Paulo, Brazil) as described previously (3). The concentration of DGL was determined spectrophotometrically at 280 nm using A 1 cm 1% ϭ 12.0 at pH 7.2 and expressed in terms of monomer (M r ϭ 25,000) (3). ConA was prepared from jack bean (Canavalia ensiformis) seeds (Sigma) according to the method of Agrawal and Goldstein (4). The concentration of ConA was determined spectrophotometrically at 280 nm using A 1 cm 1% ϭ 13.7 at pH 7.2 (5) and expressed in terms of monomer (M r ϭ 25,600).
Thermodynamic Binding Studies-ITC was performed using an MCS microcalorimeter from Microcal, Inc. (Northampton, MA). In individual titrations, injections of 4 l of carbohydrate were added from the computer-controlled 100-l microsyringe at an interval of 4 min into the lectin solution (cell volume ϭ 1.35 ml) dissolved in the same buffer as the saccharide, while stirring at 350 rpm. Control experiments performed by making identical injections of saccharide into a cell containing buffer with no protein showed insignificant heats of dilution. The experimental data were fitted to a theoretical titration curve using software supplied by Microcal, with ⌬H (enthalpy change in kcal mol Ϫ1 ), K a (association constant in M Ϫ1 ), and n (number of binding sites per monomer), as adjustable parameters. The quantity c ϭ K a M t (0), where M t (0) is the initial macromolecule concentration, is of importance in titration microcalorimetry (8). All experiments were performed with c values 1 Ͻ c Ͻ 200. The instrument was calibrated using the calibration kit containing RNase A and 2Ј-CMP supplied by the manufacturer. Thermodynamic parameters were calculated from the equation, Eq. 1) where ⌬G, ⌬H, and ⌬S are the changes in free energy, enthalpy, and entropy of binding. T is the absolute temperature, and R ϭ 1.98 cal mol Ϫ1 K Ϫ1 . Experiments performed in deuterium oxide were found to have Ͼ99% deuterium by 1 (1,10,11). We have consistently obtained the values in Table I using two different ITC instruments, the Omega and MCS instruments from Microcal Inc.
Since the primary hydrogen bonding contacts for DGL and ConA with trimannoside 1 are essentially the same (16), differences in the ⌬⌬H (H 2 O Ϫ D 2 O) values for the two lectins suggest that other factors are involved. One of these could be differences in the ordered water molecules in the extended binding site regions of the ConA-and DGL-trimannoside complexes observed in their respective x-ray crystal structures (2) (16). In order to investigate this possibility, ITC primary solvent isotope studies in H 2 O and D 2 O were carried out with the two lectins and deoxy analogs 2-11 in Fig. 1. Primary solvent isotope effects have been used to characterize the kinetic and thermodynamic involvement of hydrogen bonding between water and donor/acceptor groups in small molecules and in protein complexes (cf. Refs. 12 and 13).   Tables I and  II, respectively, as well as  Tables I and II as well as in Fig. 3 (Tables I and II,  respectively).
Maltose and isomaltose, which are ␣(1-4) and ␣(1-6) disaccharides of Glc, respectively, were tested, and the results are shown in Tables I and II Fig. 4 shows schematic representations of the x-ray crystal structures of the binding site regions of ConA and DGL complexed to the core trimannoside. The structures show the presence of ordered water in the two complexes (16). Although the primary contacts between the two lectins and the trimannoside are essentially the same in the two complexes, the location of ordered water molecules in the binding site regions differs in the complexes. This is due to amino acid substitutions in the regions adjacent to the contact residues in both proteins.
The x-ray data reveal differences in the location of ordered water in the following three regions of the two complexes. The first region is associated with the 2-hydroxyl group on the ␣(1-3) Man of 1. In the ConA complex (Fig. 4), the 2-hydroxyl on the ␣(1-3) Man interacts with W69 and indirectly with W58 via W69. In the DGL complex (Fig. 4), the 2-hyroxyl interacts with W69 and W68. W69, in turn, interacts with W70 and W68.  The second region with altered order water in the two complexes is that associated with the 4-hydroxyl group of the central Man residue of 1. In the ConA complex, the 4-hydroxyl is bonded to the side chain hydroxyl of Tyr 12 . However, in DGL, the 4-hydroxyl of the central Man residue is hydrogen-bonded to W41 in addition to the aromatic hydroxyl of Tyr 12 . W41, in turn, is close to W58, which binds to the side chain carbonyl oxygen of Glu 205 . In ConA, His 205 is present at this site. Thus, the 4-hydroxyl group of the central Man residue of 1 is differentially solvated by ordered water in the two complexes.
The ⌬⌬H (H 2 O Ϫ D 2 O) data for the two deoxy analogs of the central Man residue of 1 show that the 4-deoxy analog (11) exhibits a significant difference in this parameter for DGL relative to that for ConA (Tables I and II ; Fig. 3). The ⌬⌬H value for DGL of 2.3 kcal mol Ϫ1 is much greater than the value for ConA of 0.3 kcal mol Ϫ1 . In contrast, the ⌬⌬H (H 2 O Ϫ D 2 O) values for DGL and ConA binding to 2-deoxy analog (10) of 1.9 kcal mol Ϫ1 and 1.5 kcal mol Ϫ1 , respectively, show little difference.
The third area of the x-ray crystal structures of the DGL and ConA complexes that differs in ordered water structures is near the 2-hydroxyl on the ␣(1-6) arm of 1 (Fig. 4). W66 and W87 are both present in the two complexes, with both binding to the 2-hydroxyl on the ␣(1-6) arm. W87, however, is adjacent to a water molecule bonded to Ser 168 in ConA, while in DGL W87 is adjacent to the side chain carbonyl oxygen of Asn 168 . In addition, W67, which is adjacent to W66 in both complexes, is bonded to the carbonyl oxygen of Thr 226 in ConA but to the carbonyl oxygen of Gly 226 in DGL. Furthermore, W66 in ConA is directly bonded to the side chain hydroxyl of Thr 226 , but in DGL W66 is bonded to W89. Thus, the secondary ordered water layer near the 2-hydroxyl on the ␣(1-6) arm of 1 is different in the two complexes.
The  Tables I and II for deoxy analogs 2, 6, and 11 also show a correlation with the numbers and strength of water molecules interacting with corresponding hydroxyl groups of the parent trimannoside in the respective complexes. For example, the largest ⌬⌬H (H 2 O Ϫ D 2 O) value for DGL is for binding 2 (3.3 kcal mol Ϫ1 ). This contrasts with the corresponding ConA value of 1.2 kcal mol Ϫ1 . Fig. 4 shows that in DGL the 2-hydroxyl of the ␣(1-3) Man of 1 is directly hydrogen-bonded to two water molecules (W69 and W68), while in ConA only one water molecule (W69) is observed to bind to this hydroxyl group. Furthermore, W69 is closer to the hydroxyl group in DGL (3.1 Å) than in ConA (3.4 Å). Thus, the number and apparent strength of hydrogen-bonding water molecules to the 2-hydroxyl of the ␣(1-3) Man of 1 appears greater in DGL than in ConA.
The next largest value of ⌬⌬H (H 2 O Ϫ D 2 O) for DGL is 2.3 kcal mol Ϫ1 for 11. The corresponding value for ConA is 0.3 kcal mol Ϫ1 . Fig. 4 shows that the 4-hydroxyl of the core Man of 1 is directly bonded to W41 in DGL. However, there is an absence of electron density for a solvent molecule at this position in the ConA complex.
The ⌬⌬H (H 2 O Ϫ D 2 O) value for DGL binding to 6 of 1.6 kcal mol Ϫ1 is lower than those for 2 and 11. The corresponding value for ConA is 0.7 kcal mol Ϫ1 . In this case, the x-ray data (Fig. 4) shows altered solvent structure in the second hydration layer near the 2-hydroxyl of the ␣(1-6) Man of 1 in both complexes.
Thus, differences in numbers and strength of the water molecules in DGL and ConA interacting with specific hydroxyl groups of trimannoside 1 appear to be reflected in the magni-  Tables I and  II also show a correlation with the altered ordered water structures observed in the binding site regions of the DGL and ConA complexed with the trimannoside (Fig. 4). Since Me␣Man occupies the same site as the ␣(1-6) Man residue of 1 and makes similar contacts with ConA (2), it is reasonable to assume that the altered ordered water near the 2-hydroxyl of the ␣(1-6) Man residue of 1 in the DGL and ConA complexes is present in their respective complexes with the monosaccharide. The ⌬⌬H (H 2 O Ϫ D 2 O) value for DGL binding to Me␣Man (1.7 kcal mol Ϫ1 ) is considerably greater than that for ConA (0.5 kcal  Tables I and II. This suggests that solvation of these longer oligosaccharides in the respective DGL and ConA complexes moderates any differential solvation effects observed for the monosaccharides, trimannoside 1 and its deoxy analogs.  Table I  to the differences in the ⌬⌬H values of the deoxy analogs of the ␣(1-6) Man of 1 reported in the first paper in this series (15).

Lack of Correlation of Altered Water Structures in the DGL and ConA Complexes with the Core Trimannoside and ⌬⌬H
The ⌬⌬H values of DGL in H 2 O for the 2-, 3-, 4-, and 6-deoxy analogs of 1 (6 -9) in Fig. 1 are observed to be ϳ3 kcal mol Ϫ1 greater than the corresponding values in ConA (15). However, as shown in Fig. 4, and together with the ⌬⌬H (H 2 O Ϫ D 2 O) data in the present study, the altered water structure in that region of the two complexes appears to effect primarily the 2-hydroxyl of the ␣(1-6) Man of 1 and not the 3-, 4-, and 6-hyroxyl groups. The 3-hydroxyl of the ␣(1-6) Man of 1 is in contact with W60 in both complexes, while the 4-and 6-hydroxyl groups are not directly bonded to water molecules. In addition, the ⌬⌬H values of DGL and ConA in D 2 O for deoxy analogs 7-9 show a difference of ϳ2.0 kcal mol Ϫ1 as compared with the ϳ3 kcal mol Ϫ1 difference in H 2 O. On the other hand, the 2-deoxy analog (6) shows a reversal (0.8 kcal mol Ϫ1 ) of this difference between the two lectins in H 2 O and D 2 O (Tables I  and II). Thus, the altered water structure near the ␣(1-6) Man of 1 may account for the differences in the ⌬⌬H values of DGL and ConA for the 2-deoxy analog (2) binding to the two lectins. However, the altered water structure in this region of the two lectin complexes does not appear to account for the differences in the ⌬⌬H values in H 2 O and D 2 O for ConA and DGL binding to deoxy analogs 7-9.
This conclusion is also supported by the fact that the altered water structures near the 4-hydroxyl of the central Man residue of 1 and altered water structures near the 2-hydroxyl of the ␣(1-3) Man of 1 in the two lectin complexes do not appear to have an effect on the ⌬⌬H values for the two lectins binding the core 4-deoxy analog (11) and the ␣(1-6) 2-deoxy analog (2) in H 2 O (Tables I and II). The ⌬⌬H value for ConA binding 2 is 0.6 kcal mol Ϫ1 , while the ⌬⌬H value for DGL binding 2 is 1 kcal mol Ϫ1 . The ⌬⌬H value for ConA binding 11 is 2.6 kcal mol Ϫ1 , while the ⌬⌬H value for DGL binding 11 is 3.4 kcal mol Ϫ1 (Tables I and II). Thus, the results indicate that altered structural water in these two regions of the DGL and ConA complexes with 1 do not correlate with the ⌬⌬H values in H 2 O of both lectins for 2 and 11.
One of the remaining possibilities for explaining the differences in the ⌬⌬H values of DGL and ConA binding to the ␣(1-6) 3-, 4-, and 6-deoxymannose analogs of 1 is different energetic conformational transitions leading to the bound trimannoside-deoxy analog complexes. In order to examine this possibility, the unbound structures of both lectins are required. While the structure of ConA is known (14), the unbound structure of DGL is not. This question awaits determination of the latter structure.
Enthalpy-Entropy Compensation Plots of DGL and ConA Binding to 1-11-Plots of Ϫ⌬H versus ϪT⌬S for DGL binding to 1-11 in H 2 O and D 2 O are shown in Fig. 5A. The plot for DGL in H 2 O is a straight line with a slope of 1.60 and a correlation coefficient of 0.99. This result is similar to that reported in our first paper in this series for the deoxy trimannosides, which included a tetradeoxy analog (15). The plot for D 2 O is a straight line with a slope of 1.72 and a correlation coefficient of 0.99. Thus, the solvent isotope effect on DGL binding to 1-11 is evident in the enthalpy-entropy compensation plots of the data in Table I. A similar plot of the data for ConA (Fig. 5B) also shows straight lines in H 2 O and D 2 O. The slope of the data in H 2 O is 1.45 with a correlation coefficient of 0.98. These results are similar to those recently reported by us for ConA binding to the deoxy analogs in Fig. 1 and to di-and trideoxy analogs of 1 (1). The slope of the data in D 2 O is 1.59 with a correlation coefficient of 0.97. Thus, the solvent isotope effect on ConA binding to the carbohydrates in Fig. 1 is evident in the enthalpyentropy compensation plots of the data in Table II.
Summary-The present study demonstrates that ITC measurements of solvent isotope effects in the ⌬H values of DGL and ConA binding to certain monosaccharides and deoxy analogs of trimannoside 1 in H 2 O and D 2 O are sensitive to altered solvation of the core trimannoside at specific sites, consistent with the x-ray crystal structures of the two respective lectin complexes. Differences in the numbers and strength of the water molecules in DGL and ConA interacting with specific hydroxyl groups of trimannoside 1 appear to be reflected in the magnitude of the ⌬⌬H (H 2 O Ϫ D 2 O) values for 2, 6, and 11. Thus, evidence that solvation differences exist in solution for the DGL and ConA complexes with 1 has been obtained. The results allow us to conclude that differences in the ⌬⌬H values of ␣(1-6) deoxy analogs of 1 binding to DGL and ConA in H 2 O do not correlate with solvation differences of the parent trimannoside complexes. The origin of these differences is presently being investigated.