Mechanisms of heparanase inhibition by the heparan sulfate mimetic PG545 and three structural analogues.

The tetrasaccharide heparan sulfate (HS) mimetic PG545, a clinical anti-cancer candidate, is an inhibitor of the HS-degrading enzyme heparanase. The kinetics of heparanase inhibition by PG545 and three structural analogues were investigated to understand their modes of inhibition. The cholestanol aglycon of PG545 significantly increased affinity for heparanase and also modified the inhibition mode. For the tetrasaccharides, competitive inhibition was modified to parabolic competition by the addition of the cholestanol aglycon. For the trisaccharides, partial competitive inhibition was modified to parabolic competition. A schematic model to explain these findings is presented.

Discovery of PG545: a highly potent and simultaneous inhibitor of angiogenesis, tumor growth, and metastasis.

Increasing the aglycone lipophilicity of a series of polysulfated oligosaccharide glycoside heparan sulfate (HS) mimetics via attachment of a steroid or long chain alkyl group resulted in compounds with significantly improved in vitro and ex vivo antiangiogenic activity. The compounds potently inhibited heparanase and HS-binding angiogenic growth factors and displayed improved antitumor and antimetastatic activity in vivo compared with the earlier series. Preliminary pharmacokinetic analyses also revealed significant increases in half-life following iv dosing, ultimately supporting less frequent dosing regimens in preclinical tumor models compared with other HS mimetics. The compounds also displayed only mild anticoagulant activity, a common side effect usually associated with HS mimetics. These efforts led to the identification of 3ß-cholestanyl 2,3,4,6-tetra-O-sulfo-α-d-glucopyranosyl-(1→4)-2,3,6-tri-O-sulfo-α-d-glucopyranosyl-(1→4)-2,3,6-tri-O-sulfo-α-d-glucopyranosyl-(1→4)-2,3,6-tri-O-sulfo-ß-d-glucopyranoside, tridecasodium salt (PG545, 18) as a clinical candidate. Compound 18 was recently evaluated in a phase I clinical trial in cancer patients.

Polyamine-based small molecule epigenetic modulators.

Chromatin remodelling enzymes such as the histone deacetylases (HDACs) and histone demethylases such as lysine-specific demethylase 1 (LSD1) have been validated as targets for cancer drug discovery. Although a number of HDAC inhibitors have been marketed or are in human clinical trials, the search for isoform-specific HDAC inhibitors is an ongoing effort. In addition, the discovery and development of compounds targeting histone demethylases are in their early stages. Epigenetic modulators used in combination with traditional antitumor agents such as 5-azacytidine represent an exciting new approach to cancer chemotherapy. We have developed multiple series of HDAC inhibitors and LSD1 inhibitors that promote the re-expression of aberrantly silenced genes that are important in human cancer. The design, synthesis and biological activity of these analogues is described herein.

Synthesis and biological evaluation of polysulfated oligosaccharide glycosides as inhibitors of angiogenesis and tumor growth.

A series of polysulfated penta- and tetrasaccharide glycosides containing alpha(1-->3)/alpha(1-->2)-linked mannose residues were synthesized as heparan sulfate (HS) mimetics and evaluated for their ability to inhibit angiogenesis. The compounds bound tightly to angiogenic growth factors (FGF-1, FGF-2, and VEGF) and strongly inhibited heparanase activity. In addition, the compounds exhibited potent activity in cell-based and ex vivo assays indicative of angiogenesis, with tetrasaccharides exhibiting activity comparable to that of pentasaccharides. Selected compounds also showed good antitumor activity in vivo in a mouse melanoma (solid tumor) model resistant to the phase III HS mimetic 1 (muparfostat, formerly known as PI-88). The lipophilic modifications also resulted in reduced anticoagulant activity, a common side effect of HS mimetics, and conferred a reasonable pharmacokinetic profile in the rat, as exemplified by the sulfated octyl tetrasaccharide 5. The data support the further investigation of this class of compounds as potential antiangiogenic, anticancer therapeutics.

Synthesis of a heparan sulfate mimetic disaccharide with a conformationally locked residue from a common intermediate.

A simple mimetic of a heparan sulfate disaccharide sequence that binds to the growth factors FGF-1 and FGF-2 was synthesized by coupling a 2-azido-2-deoxy-D-glucopyranosyl trichloroacetimidate donor with a 1,6-anhydro-2-azido-2-deoxy-beta-D-glucopyranose acceptor. Both the donor and acceptor were obtained from a common intermediate readily obtained from D-glucal. Molecular docking calculations showed that the predicted locations of the disaccharide sulfo groups in the binding site of FGF-1 and FGF-2 are similar to the positions observed for co-crystallized heparin-derived oligosaccharides obtained from published crystal structures.

The calculation of polar surface area from first principles: an application of quantum chemical topology to drug design.

The calculation of polar surface areas (PSA) from the electron density using quantum chemical topology (QCT) and a newly developed algorithm to determine isodensity surface areas is described. PSA values were calculated from the atomic partitioning of B3LYP/6-311G* wavefunctions and the results described herein represent the first application of this new algorithm. PSA values were calculated for forty drugs and compared to the topological polar surface area (TPSA) and those calculated by the QikProp program. Oral bioavailabilities predicted from the QCT PSA values for a subset of twenty drugs (the Palm set) were similar to those predicted by the dynamic polar surface area (DPSA) and in general, are in agreement with the observed values. Overall, PSA values obtained from QCT were generally similar to the DPSA, TPSA, and QikProp values, though differences in fragment contributions were found, with nitrogen-bearing functional groups showing the largest variation between methods. Differences between methods showed how the calculation of the PSA is dependent on the method used and, therefore, judicious application of the upper limits used in the prediction of oral bioavailability is warranted. These results also indicate that, because of the differences in the way PSA values are calculated, values from the different methods should not be used interchangeably.

Design, synthesis, FGF-1 binding, and molecular modeling studies of conformationally flexible heparin mimetic disaccharides.

Disaccharide mimetics of a heparin sequence that binds to fibroblast growth factors were prepared by coupling a D-galactose donor with a methyl beta-D-gluco- or xylopyranoside acceptor. When fully sulfated, the glucose or xylose moieties exist in solution in equilibrium between the (4)C1 and (1)C4 conformers, as confirmed by 1H NMR spectroscopy, thus mimicking the conformationally flexible L-iduronic acid found in heparin. Docking calculations showed that the predicted locations of disaccharide sulfo groups in the binding site of FGF-1 are consistent with the positions observed for co-crystallized heparin-derived oligosaccharides. Predicted binding affinities are in accord with experimental Kd values obtained from binding assays and are similar to the predicted values for a model heparin disaccharide.

PI-88 and novel heparan sulfate mimetics inhibit angiogenesis.

The heparan sulfate (HS) mimetic PI-88 is a promising inhibitor of tumor growth and metastasis expected to commence phase III clinical evaluation in 2007 as an adjuvant therapy for postresection hepatocellular carcinoma. Its anticancer properties are attributed to inhibition of angiogenesis via antagonism of the interactions of angiogenic growth factors and their receptors with HS. It is also a potent inhibitor of heparanase, an enzyme that plays a key role in both metastasis and angiogenesis. A series of PI-88 analogs have been prepared with enhanced chemical and biological properties. The new compounds consist of single, defined oligosaccharides with specific modifications designed to improve their pharmacokinetic properties. These analogs all inhibit heparanase and bind to the angiogenic fibroblast growth factor 1 (FGF-1), FGF-2, and vascular endothelial growth factor with similar affinity to PI-88. However, compared with PI-88, some of the newly designed compounds are more potent inhibitors of growth factor-induced endothelial cell proliferation and of endothelial tube formation on Matrigel. Representative compounds were also tested for antiangiogenic activity in vivo and were found to reduce significantly blood vessel formation. Moreover, the pharmacokinetic profile of several analogs was also improved, as evidenced primarily by lower clearance in comparison with PI-88. The current data support the development of HS mimetics as potent antiangiogenic anticancer agents.

Are intramolecular dynamic electron correlation effects detectable in X-ray diffraction experiments on molecular crystals?

In order to assess whether the effects of intramolecular dynamic electron correlation on the electron density would be experimentally detectable, X-ray structure factors which include thermal averaging effects have been calculated from the electron densities of a range of small-molecule molecular crystals [C(2)H(6), C(2)H(4), C(2)H(2), BH(3)NH(3), NH(3), NH(2)CN, OCl(2), CO(NH(2))(2)] using the procrystal, Hartree-Fock, B3LYP and QCISD wavefunction models with the superposition-of-independent-molecules method to create the electron density in the crystal. A naive R-factor-like criterion of 1% has been used to assess detectability, as well as a more sophisticated method based on real X-ray data for estimating experimental errors. Correlation effects on the density are found to be only marginally above the 1% detectability threshold, and are about one to two orders of magnitude smaller than deviations from the procrystal model. Further, only 10% of the data up to 1.2 A(-1) are significant for detecting correlation effects; and of those 10%, many are at low intensity and therefore difficult to measure. Another method to estimate the experimental errors indicates that the intramolecular correlation effects would not be measurable. Although thermal averaging effects are important for the absolute value of the calculated structure factors, the use of different thermal averaging models does not change our overall conclusion of detectability. Likewise, calculations using the B3LYP method for some molecules do not show significant changes in the amount of, or distribution of, the changes that would be detectable by experiment.

An experimental and molecular-modeling study of the binding of linked sulfated tetracyclitols to FGF-1 and FGF-2.

The experimental binding affinities of a series of linked sulfated tetracyclitols [Cyc2N-R-NCyc2, where Cyc = C6H6(OSO3Na)3 and R = (CH2)n (n = 2-10), p-xylyl or (C2H4)2-Ncyc] for the fibroblast growth factors FGF-1 and FGF-2 have been measured by using a surface plasmon resonance assay. The KD values range from 7.0 nM to 1.1 microM for the alkyl-linked ligands. The binding affinity is independent of the flexibility of the linker, as replacement of the alkyl linker with a rigid p-xylyl group did not affect the KD. Calculations suggest that binding modes for the p-xylyl-linked ligand are similar to those calculated for the flexible alkyl-linked tetracyclitols. The possible formation of cross-linked FGF:cyclitol complexes was examined by determining KD values at increasing protein concentrations. No changes in KD were observed; this suggesting that only 1:1 complexes are formed under these assay conditions. Monte Carlo multiple-minima calculations of low-energy conformers of the FGF-bound ligands showed that all of the sulfated tetracyclitol ligands can bind effectively in the heparan sulfate-binding sites of FGF-1 and FGF-2. Binding affinities of these complexes were estimated by the Linear Interaction Energy (LIE) method to within a root-mean-square deviation of 1 kcal mol(-1) of the observed values. The effect of incorporating cations to balance the overall charge of the complexes during the LIE calculations was also explored.

Validation of molecular docking calculations involving FGF-1 and FGF-2.

A predictive relationship between calculated and observed binding affinities for the complexation of ligands to the fibroblast growth factors FGF-1 and FGF-2 based on molecular docking calculations is described. The majority of the ligands examined in this study have high conformational flexibility, and to account for this, multiple conformers were generated for each and subsequently used in flexible docking calculations. Two scoring functions, Gscore and Emodel, were used to quantify the protein:ligand interaction of which the Emodel score showed the best correlation with experimental binding energies. Both scoring functions, however, predicted similar locations for the ligand sulfate groups in the binding site. The van der Waals radii of nonpolar atoms of both the protein and ligand, which modify the effective sizes of both the protein binding site and the ligand, were also systematically altered by factors of 1.0, 0.9, and 0.8 in order to optimize the conditions for predictive docking. Least squares analyses of the Emodel scores against experimental binding energies yielded best r(2) values of 0.91 and 0.83 for FGF-1 and FGF-2, respectively, with slightly lower q(2) values. Optimized scale factor combinations in conjunction with the least squares lines of best fit based on the Emodel function were used to define a predictive model that was tested against ligands not included in the original set. Acceptable predictions of binding affinity were obtained for use in the initial screening of potential leadlike molecules for both FGF-1 and FGF-2.

Wave functions derived from experiment. V. Investigation of electron densities, electrostatic potentials, and electron localization functions for noncentrosymmetric crystals.

The constrained Hartree-Fock method using experimental X-ray diffraction data is extended and applied to the case of noncentrosymmetric molecular crystals. A new way to estimate the errors in derived properties as a derivative with respect to added Gaussian noise is also described. Three molecular crystals are examined: ammonia [NH(3)], urea [CO(NH(2))(2)], and alloxan [(CO)(4)(NH)(2)]. The energetic and electrical properties of these molecules in the crystalline state are presented. In all cases, an enhancement of the dipole moment is observed upon application of the experimental constraint. It is found that the phases of the structure factors are robustly determined by the constrained Hartree-Fock model, even in the presence of simulated noise. Plots of the electron density, electrostatic potential, and the electron localization function for the molecules in the crystal are displayed. In general, relative to the Hartree-Fock model, there is a depletion of charge around hydrogen atoms and lone pair regions, and a build-up of charge within the molecular framework near nuclei, directed along the bonds. The electron localization function plots reveal an increase in the pair density between vicinal hydrogen atoms.

Can a multipole analysis faithfully reproduce topological descriptors of a total charge density?

Total charge densities rho(r) of solid NH(3) have been derived using an ab initio crystalline molecular-orbital approach and also from multipole refinement of the structure factors obtained from the same charge density. Comparison of the topological features of these charge densities, as defined by the quantum theory of atoms in molecules, has been used to probe the ability of the multipole analysis to reproduce exactly known total charge-density distributions. For the most part, multipole refinement satisfactorily returns the features of the original density, although the fit to theoretical data is not as good as that to the experimental data. The one topological parameter that is poorly reproduced is the Laplacian nabla (2)rho(r(b)) at NH bond critical points.

Wavefunctions derived from experiment. III. Topological analysis of crystal fragments.

A constrained wavefunction model has been used to extract a Hartree-Fock wavefunction for C2H2O4.4H2O from both low-angle (sin theta /lambda < 0.71 A-1, 571 reflections) and full (sin theta / lambda < 1.00 A-1, 968 reflections) experimental X-ray diffraction data for crystalline alpha-oxalic acid dihydrate (alpha-C2H2O4.2H2O) using polarized double-zeta and triple-zeta Gaussian basis sets. Properties obtained from the zero-flux partitioning of the total charge-density distribution derived from these wavefunctions, as well as from multipole refinement of the experimental data, are calculated and compared. This work represents the first calculation of integrated atomic properties derived from the fitting of Gaussian density functions to experimental X-ray diffraction data. In particular, atomic kinetic energies derived from experimental data are presented for the first time. The results obtained from the constrained (experimental) charge density show qualitatively similar properties to those obtained from conventional ab initio gas-phase calculations, though the quantitative differences are often substantial. The accuracy of integrated properties calculated using this procedure was established from the analysis of a wavefunction derived from simulated random-error diffraction data; that is, data obtained by adding normally distributed errors to the experimental structure factors. Analysis of this random-error wavefunction indicated that most topological properties are accurate to within approximately 5%, although the error is much larger for those properties that have a steep gradient in the region being evaluated [e.g. the value of V(2)rho(rb) at bond critical points] or are very small (e.g. the atomic dipole moment). Calculations of the constrained wavefunction using both the larger basis set and the complete set of experimental data yield results that agree quantitatively with the smaller calculations.

Wavefunctions derived from experiment. IV. Investigation of the crystal environment of ammonia.

Constrained Hartree-Fock calculations have been performed to obtain wavefunctions that reproduce experimental X-ray structure-factor magnitudes for crystalline NH3 to within the limits of experimental error. Different model densities using both a single molecule and clusters of NH3 in the calculation of X-ray structure-factor magnitudes have been examined. The effects of the crystalline lattice on the experimental wavefunction of the NH3 unit can be reproducibly recovered. The construction of structure-factor magnitudes based on normally distributed random perturbations of the experimental values has also been used to gauge the accuracy of integrated atomic properties obtained from the wavefunctions, the point at which the constraint procedure should be terminated, and the approximate error in the experimental sigma(k) values.

Bond Lengths and Bond Angles in Oxo, Hydroxo, and Alkoxo Molecules of Be, B, and C: A Close-Packed Nearly Ionic Model.

We have surveyed the experimental data for oxo, hydroxo, and alkoxo molecules of Be, B, and C and have shown that the intramolecular interligand distances for a given central atom are remarkably constant and independent of coordination number and of the presence of other ligands. Atomic charges obtained from the analysis of the calculated electron densities for a large selection of molecules of this type have shown that these molecules are predominately ionic. On the basis of these results we suggest that the bond lengths and geometries of these molecules can be best understood in terms of a model in which anion-like ligands are close-packed around a cation-like central atom. Values of the interligand radius of each ligand obtained from the intramolecular interligand contact distances are smaller than the crystal ionic radii and decrease as expected with decreasing ligand charge. This model provides a simple quantitative explanation of the decrease in the bond lengths in these molecules with decrease in the coordination number from four to three and of the changes in bond length caused by the presence of other ligands with different ligand radii. With decreasing bond length the electron density at the bond critical point increases correspondingly for Be-O, B-O, and C-O bonds. The nontetrahedral angles found in all A(OX)(4) molecule are explained on the basis of a noncylindrically symmetrical charge distribution around oxygen.

Geometry of the Fluorides, Oxofluorides, Hydrides, and Methanides of Vanadium(V), Chromium(VI), and Molybdenum(VI): Understanding the Geometry of Non-VSEPR Molecules in Terms of Core Distortion.

This paper describes a study of the topology of the electron density and its Laplacian for the molecules VF(5), VMe(5), VH(5), CrF(6), CrMe(6), CrOF(4), MoOF(4), CrO(2)F(2,) CrO(2)F(4)(2)(-) and CrOF(5)(-) all of which, except VF(5,) CrF(6), and CrOF(5)(-) have a non-VSEPR geometry. It is shown that in each case the interaction of the ligands with the metal atom core causes it to distort to a nonspherical shape. In particular, the Laplacian of the electron density reveals the formation of local concentrations of electron density in the outer shell of the core, which have a definite geometrical arrangement such as four in a tetrahedral arrangement or five in a square pyramidal or trigonal bipyramidal and six in an octahedral arrangement. Ligands that are predominately covalently bonded are found opposite regions of charge depletion between these core charge concentrations. In VH(5), VMe(5), CrOF(4), and MoOF(4), these core charge concentrations have a square pyramidal arrangement, and the regions of charge depletions have the corresponding inverse square pyramidal arrangement so that these molecules have a square pyramidal geometry rather than a trigonal prism geometry. In CrMe(6), there are five core charge concentrations with a trigonal bipyramidal arrangement so that the regions of charge depletion have a trigonal prismatic arrangement and the molecule has the corresponding trigonal prism geometry rather than an octahedral geometry. In contrast, molecules in which the only ligand is the more ionically bound fluorine are less affected by core distortion and have VSEPR-predicted structures. The unexpected bond angles in CrO(2)F(2) and the preference of CrO(2)F(4)(2)(-) for a cis structure are also discussed in terms of the pattern of core charge concentrations.