How to Define Potent Ligand Mimetics

Taking the meaning of the word "carbohydrate" (C (H2O)n) literally, the abundant display of hydroxyl groups with their sp3-hybridized oxygen atoms acting as acceptors with two lone electron pairs and the protons as donors nourishes the view that hydrogen bonds will dominate the spectrum of binding forces. When the spacing between two hydroxyl groups or the axial 4'-hydroxyl group and the ring oxygen atom matches that of an amino acid side chain (amide or carboxylate), two neighboring sites on the ligand can well be engaged in bidentate hydrogen bonds. The necessity for topological complementarity to yield the intricate network, schematically shown in Fig. 6, may not only be a source for enthalpy but also for selectivity, distinguishing anomers such D-Gal versus D-Man/D-Glc. It can thus be expected that the axial 4'-position for recognition of D-Gal and the equatorial 3',4'-positions for binding D-Man/D-Glc play decisive roles. This assumption is strikingly verified by x-ray crystallography and in solution by chemically engineered ligand derivatives (Rini, 1995; Solis et al., 1996; Weis and Drickamer, 1996; Gabius, 1997a, Solis and Diaz-Maurino, 1997; Gabius, 1998; Lis and Sharon, 1998; Loris et al., 1998; Rüdiger et al., 1999;). With this structural explanation it becomes obvious why the change of the position of one hydroxyl group to form an epimer discussed during the presentation of the individual members of the mon-osaccharide alphabet unmistakably has the effect of creating a new letter. By the way, the same principle holds true for the characteristic formation of two coordination

Fig. 6 The potential of D-galactose (see Fig. 1, bottom, and Fig. 3) for establishing interactions with constituents of the binding pocket of a sugar receptor. While the rather polar upper side can be engaged in frequent hydrogen bonds exploiting lone electron pairs of oxygen atoms as acceptors and the protons of appropriately positioned hydroxyl groups as donors (and also coordination bonds with a Ca2+-ion in the case of C-type lectins), C-H/n-electron interactions and entropically favorable stacking can be engendered by an intimate contact of an aromatic (here: indolyl) amino acid side chain and the sugar's less polar bottom section. (Kindly provided by Dr. C.-W. von der Lieth, Heidelberg.)

Fig. 6 The potential of D-galactose (see Fig. 1, bottom, and Fig. 3) for establishing interactions with constituents of the binding pocket of a sugar receptor. While the rather polar upper side can be engaged in frequent hydrogen bonds exploiting lone electron pairs of oxygen atoms as acceptors and the protons of appropriately positioned hydroxyl groups as donors (and also coordination bonds with a Ca2+-ion in the case of C-type lectins), C-H/n-electron interactions and entropically favorable stacking can be engendered by an intimate contact of an aromatic (here: indolyl) amino acid side chain and the sugar's less polar bottom section. (Kindly provided by Dr. C.-W. von der Lieth, Heidelberg.)

bonds with the central Ca2+-ion in the mentioned C-type lectins. Thereby, any wrong combination for the two adjacent hydroxyl groups involved in contacting the metal ion is excluded and sugar specificity is assured, unless the access-restricting impediment by a constraining loop close to the metal ion is lifted (Weis and Drickamer, 1996; Gabius, 1997a; Lis and Sharon, 1998; Loukas et al., 1999).

Inspecting Fig. 6 more closely, another important feature to drive ligand binding can be discovered. While the upper side of D-Gal is rather polar, the B-face exhibits a hydrophobic character. Stacking to the bulky aromatic amino acid side chain in the binding pocket removes both nonpolar surfaces from solvent accessibility, although the two rings may not be perfectly aligned in parallel. In fact, their positioning can tolerate distortions with angles between 17° and 52° in lectins (Weis and Drickamer, 1996). Nonetheless, this alignment will still contribute to complex stability and also to ligand selection despite its lower degree of directionality relative to hydrogen bonds (Quiocho, 1988; Vyas, 1991). The ensuing shielding of the indolyl side chain by the ligand is reflected for galectins in molecular dynamics calculations as well as differential UV, fluorescence, and laser photo chemically induced dynamic nuclear polarization (CIDNP) spectra (Levi and Teichberg, 1981; Siebert et al., 1997). Beyond this impact on solvent molecules by reducing the apolar surface area the n-electron cloud of the aromatic ring is likely to interact with the aliphatic D-Gal protons which harbor a net positive charge (Dougherty, 1996; Nishio et al., 1995; Weis and Drickamer, 1996). That the ensuing hydrophobic effect and van der Waals interactions do not deserve to be underestimated for impinging on the overall Gibb's free energy gain is underscored by the analysis of dominant forces in tight ligand binding for a variety of cases, where these factors can even surpass by far the contribution of hydrogen bonds (Davis and Teague, 1999; Kuntz et al., 1999).

This observation illustrates the complexity of the question how to account for the global enthalpic and entropic parameters on the level of molecules. For that galectin, whose data set from isothermal titration calorimetry is given in Fig. 5, it has recently been described by crystallographical work that six structural water molecules occupy the binding site in the ligand-free state stabilizing its topology and yielding a not yet precisely quantitated contribution to the Gibbs' free energy change upon displacement (Varela et al., 1999). In the case of a related galectin from the conger eel one additional water molecule even takes place of D-Gal's B-face substituting stacking by forming a n-electron hydrogen bond with a distance of 3.36 A and an angle of 6.5° between the vector of the weight center of the five-membered section of the indole ring to the water molecule and the vector perpendicular to the ring plane (Shirai et al., 1999). The total exchange of the water molecules with the ligand will not only directly affect these solvent molecules but may also have a bearing on the proteins' intramolecular motions in solution. Remarkably, also the impact of ligand binding on protein flexibility is to be reckoned with. An increase in its vibrational entropy (14.6 kJ/mol for binding of one water molecule to bovine pancreatic trypsin inhibitor as model (Fischer and Verma, 1999)) can offset a substantial portion of the entropic penalty of the immobilization. The extent of this factor will certainly depend on the inherent mobility dynamics of the carbohydrate ligand free in solution. This parameter has already been inferred above to be often restricted due to spatial interference of the rather bulky rings and substituents. Graphically drawing on E. Fischer's (1894) classical "lock and key" paradigm, the metaphor has tentatively been introduced for this ligand type to view certain oligosaccharides as "bunch of keys" moving in solution through a limited set of shapes (Hardy, 1997). Only one of them may be selected by a receptor.

With a digalactoside (Gal01-2Gal) as model, the formation of two "keys" from the same sequence is displayed in Fig. 7. Based on the y, E-plot, shown in its left section, molecular dynamics calculations and nuclear Overhauser effect (NOE) NMR-spectroscopy (Siebert et al., 1996, 1999; von der Lieth et al., 1998), two distinct conformations are present in solution, each molecule rapidly fluctuating between these two topological constellations (Fig. 7, right side). Due to the inability to acquire spectroscopic snapshots with a resolution in the ps range, spectroscopical monitoring will be subject of time and ensemble averaging (Carver, 1991; Jardetzky, 1980). Since the term "key" implies its accurate fit into an appropriate lock, monitoring of transferred NOE signals, reflecting through space dipolar interactions between two protons in the bound ligand in double-resonance experiments, will resolve the gripping question as to which ligand topology will be accommodated in the binding pocket (Gabius, 1998; Jiménez-Barbero et al., 1999; Peters and Pinto, 1996; Poveda and Jiménez-Barbero, 1998; Rüdiger et al., 1999; von der Lieth et al., 1998).

Fig. 7 Illustration of the principle of differential conformer selection. Based on NMR-spectroscopi-cal analysis and molecular mechanics/dynamics calculations the disaccharide Gal|31-2Gal can adopt two distinct conformations in solution, which reside in energetically preferred regions of the y, E-plot, symbolized by circles (left). Keeping the topological positioning of the nonreducing Gal-unit constant, the two sets of y-values are readily visualized to translate into two significantly different conformers (right) which harbor disparate ligand properties. (Kindly provided by Priv.-Doz. Dr. H.-C. Siebert, Munich, and Dr. C.-W. von der Lieth, Heidelberg.)

Fig. 7 Illustration of the principle of differential conformer selection. Based on NMR-spectroscopi-cal analysis and molecular mechanics/dynamics calculations the disaccharide Gal|31-2Gal can adopt two distinct conformations in solution, which reside in energetically preferred regions of the y, E-plot, symbolized by circles (left). Keeping the topological positioning of the nonreducing Gal-unit constant, the two sets of y-values are readily visualized to translate into two significantly different conformers (right) which harbor disparate ligand properties. (Kindly provided by Priv.-Doz. Dr. H.-C. Siebert, Munich, and Dr. C.-W. von der Lieth, Heidelberg.)

These experiments provide two captivating answers for the studied case of lectins. Firstly, a lectin can actually select a distinct conformer, as seen for galac-toside-binding lectins and selectins (Asensio et al., 1999; Espinosa et al., 1996; Gilleron et al., 1998; Harris et al., 1999; Poppe et al., 1997; Siebert et al., 1996; von der Lieth et al., 1998). Despite the same sequence the shape of other con-formers renders them unsuitable for binding. Of course, a wrong key will not open a nonadaptable (rigid) lock designed for a different shape. Secondly, different receptors even with the same saccharide specificity harbor the capacity to bind different conformers. Thus, freezing a distinct conformation should have a dramatic impact on receptor binding as alluded to above. This principle is referred to as "differential conformer selection." It is visualized in Fig. 7 by noting that the conformer defined by the upper ^-combination is exclusively bound by a plant (mistletoe) agglutinin, while the tested galectin homes in on the second conformer (Gabius, 1998; Gilleron et al., 1998; Siebert et al., 1996; von der Lieth et al., 1998). Thus, not only the hydrogen-bonding patterns of these lectins toward D-Gal differ, as delineated by chemical mapping with deoxy and fluoro derivatives (Lee et al., 1992; Rüdiger et al., 1999; Solis et al., 1996), but also the pair of ^-torsion angles of ß-Gal-terminated disaccharides. Because the importance of the intramolecular flexibility of the free ligand and conformer selection is only gradually explored as factor to be rationally manipulated, this result together with insights into favorable energetic interactions between the binding partners including solvent molecules warrants consideration for the design of mimetics. Thereby, they can eventually meet the high expectations for potency expressed in Fig. 4. When the geometry of crucial groups is maintained or even improved, the obtained substances do not even need to belong to the class of carbohydrates. To grant adequate heed to mimetics is probably a means to open a wide field for rational drug design, currently, for example, exploited for the influenza A/B neuraminidase and selectins (Sears and Wong, 1999; Simanek et al., 1998; von Itzstein and Thomson, 1997). As caveats to caution against prematurely advocating clinical effectiveness of anti-adhesion therapy in inflammation or of sugar-based drugs in epidemic flu, detrimental long-term effects in an animal model mimicking both acute and chronic intestinal inflammation has been reported (McCafferty et al., 1999). Similarly, stress has been laid upon the necessity to prove clinical benefit for an elegantly invented but costly antiflu drug in terms of an obvious impact on mortality beyond that of common, less expensive medications including vaccination (Cox and Hughes, 1999; Institut für Arzneimittelinformation, 1999; Yamey, 1999).

Was this article helpful?

0 0

Post a comment