Design and Synthesis of Inhibitors of the E3 Ligase SMAD Specific E3 Ubiquitin Protein Ligase 1 as a Treatment for Lung Remodeling in Pulmonary Arterial Hypertension

Pulmonary arterial hypertension (PAH) is a devastating rare disease, which despite currently available treatments, still represents a high unmet medical need. Specific E3 ubiquitin protein ligase 1 (SMURF1) is a HECT E3 ligase that ubiquitinates key signaling molecules from the TGFβ/BMP pathways, which are of great relevance in the pathophysiology of PAH. Herein, the design and synthesis of novel potent small-molecule SMURF1 ligase inhibitors are described. Lead molecule 38 has demonstrated good oral pharmacokinetics in rats and significant efficacy in a rodent model of pulmonary hypertension.

JDQ443, a Structurally Novel, Pyrazole-Based, Covalent Inhibitor of KRASG12C for the Treatment of Solid Tumors

Rapid emergence of tumor resistance via RAS pathway reactivation has been reported from clinical studies of covalent KRASG12C inhibitors. Thus, inhibitors with broad potential for combination treatment and distinct binding modes to overcome resistance mutations may prove beneficial. JDQ443 is an investigational covalent KRASG12C inhibitor derived from structure-based drug design followed by extensive optimization of two dissimilar prototypes. JDQ443 is a stable atropisomer containing a unique 5-methylpyrazole core and a spiro-azetidine linker designed to position the electrophilic acrylamide for optimal engagement with KRASG12C C12. A substituted indazole at pyrazole position 3 results in novel interactions with the binding pocket that do not involve residue H95. JDQ443 showed PK/PD activity in vivo and dose-dependent antitumor activity in mouse xenograft models. JDQ443 is now in clinical development, with encouraging early phase data reported from an ongoing Phase Ib/II clinical trial (NCT04699188).

Discovery, Preclinical Characterization, and Early Clinical Activity of JDQ443, a Structurally Novel, Potent, and Selective Covalent Oral Inhibitor of KRASG12C

Covalent inhibitors of KRASG12C have shown antitumor activity against advanced/metastatic KRASG12C-mutated cancers, though resistance emerges and additional strategies are needed to improve outcomes. JDQ443 is a structurally unique covalent inhibitor of GDP-bound KRASG12C that forms novel interactions with the switch II pocket. JDQ443 potently inhibits KRASG12C-driven cellular signaling and demonstrates selective antiproliferative activity in KRASG12C-mutated cell lines, including those with G12C/H95 double mutations. In vivo, JDQ443 induces AUC exposure-driven antitumor efficacy in KRASG12C-mutated cell-derived (CDX) and patient-derived (PDX) tumor xenografts. In PDX models, single-agent JDQ443 activity is enhanced by combination with inhibitors of SHP2, MEK, or CDK4/6. Notably, the benefit of JDQ443 plus the SHP2 inhibitor TNO155 is maintained at reduced doses of either agent in CDX models, consistent with mechanistic synergy. JDQ443 is in clinical development as monotherapy and in combination with TNO155, with both strategies showing antitumor activity in patients with KRASG12C-mutated tumors.

SIGNIFICANCE

JDQ443 is a structurally novel covalent KRASG12C inhibitor with a unique binding mode that demonstrates potent and selective antitumor activity in cell lines and in vivo models. In preclinical models and patients with KRASG12C-mutated malignancies, JDQ443 shows potent antitumor activity as monotherapy and in combination with the SHP2 inhibitor TNO155. This article is highlighted in the In This Issue feature, p. 1397.

Abstract P124: JDQ443, a covalent irreversible inhibitor of KRAS G12C, exhibits a novel binding mode and demonstrates potent anti-tumor activity and favorable pharmacokinetic properties in preclinical models

RAS is the most frequently mutated oncogene in cancer. KRAS G12C mutations are most prevalent in lung adenocarcinoma (~13%) and colorectal adenocarcinoma (~4%), and occur less commonly in other solid tumor malignancies. First generation KRASG12C inhibitors show anti-tumor activity in early phase clinical trials. However, the emergence of resistance, mediated at least in part by RAS gene mutations that disrupt inhibitor binding and reactivation of downstream pathways, limit the duration of response. Here we report the identification of JDQ443 (NVP-JDQ443), a novel KRASG12C inhibitor which binds under the switch II loop with a novel binding mode, exploiting unique interactions with the KRASG12C protein compared to sotorasib and adagrasib. JDQ443 potently inhibits KRASG12C cellular signaling and proliferation in a mutant selective manner by irreversibly trapping the GDP-bound state of KRASG12C through formation of a covalent bond with cysteine at position 12. Consistent with its mechanism as an irreversible inhibitor, JDQ443 shows sustained target occupancy (TO) in vivo (KRASG12C TO t1/2 ~ 66 h in the MiaPaCa2 model) despite a blood half-life of ~ 2 hours, and exhibits a linear PK/PD relationship. JDQ443 has dose-dependent anti-tumor activity in mice bearing KRAS G12C mutated tumor xenografts comparable to sotorasib and adagrasib. In mouse, rat, and dog, JDQ443 is orally bioavailable, achieves exposures in a range predicted to confer anti-tumor activity, and is well-tolerated. Continuous delivery of JDQ443 using mini-pump administration demonstrates that area under the curve (AUC), rather than maximal concentration (Cmax), is the driver of efficacy. Combination of JDQ443 with the SHP2 inhibitor TNO155 further increases KRAS G12C target occupancy in vivo, enhanced pre-clinical anti-tumor activity, and delayed the emergence of resistance in xenografts. A genome-wide CRISPR screen in 5 KRAS G12C mutated lung cancer cell lines identifies novel mechanisms of resistance to the KRAS/SHP2 drug combination. Furthermore, the characterization of JDQ443 alone and in combination with TNO155 in BaF/3 pools addicted to KRAS alleles that have previously been shown to mediate resistance to adagrasib in clinical samples will be discussed. Collectively, these data show that JDQ443 is a potent, mutant-selective, covalent irreversible KRASG12C inhibitor with favorable pharmaceutical properties. A phase Ib/II clinical trial of JDQ443 alone and in combination with TNO155 in patients with advanced solid tumors harboring the KRAS G12C mutation is ongoing (NCT04699188).

Citation Format: Saskia M. Brachmann, Andreas Weiss, Daniel A. Guthy, Kim Beyer, Johannes Voshol, Michel Maira, Anirudh Prahallad, Diana Graus Porta, Christian Schnell, Nils Ostermann, Andrea Vaupel, Marc Gerspacher, Catherine Leblanc, Dirk Erdmann, Dario Sterker, Grainne Kerr, Giovannoni Jerome, Victoria Head, Rowan Stringer, Ruben De Kanter, Kearns Jeff, Danielle Roman, Toni Widmer, Peter Wessels, Eloisa Jimenez Nunez, Richard Sedrani, Frederic Zecri, Francesco Hofmann, Jeff Engleman, Edwige Lorthiois, Simona Cotesta. JDQ443, a covalent irreversible inhibitor of KRAS G12C, exhibits a novel binding mode and demonstrates potent anti-tumor activity and favorable pharmacokinetic properties in preclinical models [abstract]. In: Proceedings of the AACR-NCI-EORTC Virtual International Conference on Molecular Targets and Cancer Therapeutics; 2021 Oct 7-10. Philadelphia (PA): AACR; Mol Cancer Ther 2021;20(12 Suppl):Abstract nr P124.

Evolution of Novartis' Small Molecule Screening Deck Design

This article summarizes the evolution of the screening deck at the Novartis Institutes for BioMedical Research (NIBR). Historically, the screening deck was an assembly of all available compounds. In 2015, we designed a first deck to facilitate access to diverse subsets with optimized properties. We allocated the compounds as plated subsets on a 2D grid with property based ranking in one dimension and increasing structural redundancy in the other. The learnings from the 2015 screening deck were applied to the design of a next generation in 2019. We found that using traditional leadlikeness criteria (mainly MW, clogP) reduces the hit rates of attractive chemical starting points in subset screening. Consequently, the 2019 deck relies on solubility and permeability to select preferred compounds. The 2019 design also uses NIBR's experimental assay data and inferred biological activity profiles in addition to structural diversity to define redundancy across the compound sets.

High-Resolution Genetics Identifies the Lipid Transfer Protein Sec14p as Target for Antifungal Ergolines

Invasive infections by fungal pathogens cause more deaths than malaria worldwide. We found the ergoline compound NGx04 in an antifungal screen, with selectivity over mammalian cells. High-resolution chemogenomics identified the lipid transfer protein Sec14p as the target of NGx04 and compound-resistant mutations in Sec14p define compound-target interactions in the substrate binding pocket of the protein. Beyond its essential lipid transfer function in a variety of pathogenic fungi, Sec14p is also involved in secretion of virulence determinants essential for the pathogenicity of fungi such as Cryptococcus neoformans, making Sec14p an attractive antifungal target. Consistent with this dual function, we demonstrate that NGx04 inhibits the growth of two clinical isolates of C. neoformans and that NGx04-related compounds have equal and even higher potency against C. neoformans. Furthermore NGx04 analogues showed fungicidal activity against a fluconazole resistant C. neoformans strain. In summary, we present genetic evidence that NGx04 inhibits fungal Sec14p and initial data supporting NGx04 as a novel antifungal starting point.

Emerging resistance to antibiotics led to an inglorious revival of infectious diseases. Furthermore, in the past 30 years, only one novel anti-fungal target has been discovered which was used to develop therapies against. Therefore pathogen-selective targets and knowledge about possible resistance determinants are of utmost importance to successfully develop new medicines. Here we describe the identification of anti-fungal ergolines, targeting the lipid transfer protein Sec14p, and inhibiting the growth of two clinical isolates of the pathogenic fungus Cryptococcus neoformans. Both, compound and target represent attractive points for further investigations: Sec14p as it differs significantly from the human homolog and as it has been implicated in fungal viability and pathogenicity, and, ergolines as they are used in the clinic against a variety of diseases demonstrating both efficacy and safety.

A General Strategy for Targeting Drugs to Bone

Targeting drugs to their desired site of action can increase their safety and efficacy. Bisphosphonates are prototypical examples of drugs targeted to bone. However, bisphosphonate bone affinity is often considered too strong and cannot be significantly modulated without losing activity on the enzymatic target, farnesyl pyrophosphate synthase (FPPS). Furthermore, bisphosphonate bone affinity comes at the expense of very low and variable oral bioavailability. FPPS inhibitors were developed with a monophosphonate as a bone-affinity tag that confers moderate affinity to bone, which can furthermore be tuned to the desired level, and the relationship between structure and bone affinity was evaluated by using an NMR-based bone-binding assay. The concept of targeting drugs to bone with moderate affinity, while retaining oral bioavailability, has broad application to a variety of other bone-targeted drugs.

Advantages and Challenges of Phenotypic Screens

Phenotypic screens are effective starting points to identify compounds with desirable activities. To find novel antifungals, we conducted a phenotypic screen in Saccharomyces cerevisiae and identified two discrete scaffolds with good growth inhibitory characteristics. Lack of broad-spectrum activity against pathogenic fungi called for directed chemical compound optimization requiring knowledge of the molecular target. Chemogenomic profiling identified effects on geranylgeranyltransferase I (GGTase I), an essential enzyme that prenylates proteins involved in cell signaling, such as Cdc42p and Rho1p. Selection of resistant mutants against both compounds confirmed the target hypothesis and enabled mapping of the compound binding site to the substrate binding pocket. Differential resistance-conferring mutations and selective substrate competition demonstrate distinct binding modes for the two chemotypes. Exchange of the S. cerevisiae GGTase I subunits with those of Candida albicans resulted in an absence of growth inhibition for both compounds, thus confirming the identified target as well as the narrow antifungal spectrum of activity. This prenylation pathway is reported to be nonessential in pathogenic species and challenges the therapeutic value of these leads while demonstrating the importance of an integrated target identification platform following a phenotypic screen.

Discovery of Novel Allosteric Non-Bisphosphonate Inhibitors of Farnesyl Pyrophosphate Synthase by Integrated Lead Finding

Farnesyl pyrophosphate synthase (FPPS) is an established target for the treatment of bone diseases, but also shows promise as an anticancer and anti-infective drug target. Currently available anti-FPPS drugs are active-site-directed bisphosphonate inhibitors, the peculiar pharmacological profile of which is inadequate for therapeutic indications beyond bone diseases. The recent discovery of an allosteric binding site has paved the way toward the development of novel non-bisphosphonate FPPS inhibitors with broader therapeutic potential, notably as immunomodulators in oncology. Herein we report the discovery, by an integrated lead finding approach, of two new chemical classes of allosteric FPPS inhibitors that belong to the salicylic acid and quinoline chemotypes. We present their synthesis, biochemical and cellular activities, structure-activity relationships, and provide X-ray structures of several representative FPPS complexes. These novel allosteric FPPS inhibitors are devoid of any affinity for bone mineral and could serve as leads to evaluate their potential in none-bone diseases.

Distinct effects of IPSU and suvorexant on mouse sleep architecture

Dual orexin receptor (OXR) antagonists (DORAs) such as almorexant, SB-649868, suvorexant (MK-4305), and filorexant (MK-6096), have shown promise for the treatment of insomnias and sleep disorders. Whether antagonism of both OX₁R and OX₂R is necessary for sleep induction has been a matter of some debate. Experiments using knockout mice suggest that it may be sufficient to antagonize only OX₂R. The recent identification of an orally bioavailable, brain penetrant OX₂R preferring antagonist 2-((1H-Indol-3-yl)methyl)-9-(4-methoxypyrimidin-2-yl)-2,9-diazaspiro[5.5]undecan-1-one (IPSU) has allowed us to test whether selective antagonism of OX₂R may also be a viable strategy for induction of sleep. We previously demonstrated that IPSU and suvorexant increase sleep when dosed during the mouse active phase (lights off); IPSU inducing sleep primarily by increasing NREM sleep, suvorexant primarily by increasing REM sleep. Here, our goal was to determine whether suvorexant and IPSU affect sleep architecture independently of overall sleep induction. We therefore tested suvorexant (25 mg/kg) and IPSU (50 mg/kg) in mice during the inactive phase (lights on) when sleep is naturally more prevalent and when orexin levels are normally low. Whereas IPSU was devoid of effects on the time spent in NREM or REM, suvorexant substantially disturbed the sleep architecture by selectively increasing REM during the first 4 h after dosing. At the doses tested, suvorexant significantly decreased wake only during the first hour and IPSU did not affect wake time. These data suggest that OX₂R preferring antagonists may have a reduced tendency for perturbing NREM/REM architecture in comparison with DORAs. Whether this effect will prove to be a general feature of OX₂R antagonists vs. DORAs remains to be seen.

Kinetic properties of "dual" orexin receptor antagonists at OX1R and OX2R orexin receptors

Orexin receptor antagonists represent attractive targets for the development of drugs for the treatment of insomnia. Both efficacy and safety are crucial in clinical settings and thorough investigations of pharmacokinetics and pharmacodynamics can predict contributing factors such as duration of action and undesirable effects. To this end, we studied the interactions between various “dual” orexin receptor antagonists and the orexin receptors, OX₁R and OX₂R, over time using saturation and competition radioligand binding with [³H]-BBAC ((S)-N-([1,1′-biphenyl]-2-yl)-1-(2-((1-methyl-1H-benzo[d]imidazol-2-yl)thio)acetyl)pyrrolidine-2-carboxamide). In addition, the kinetics of these compounds were investigated in cells expressing human, mouse and rat OX₁R and OX₂R using FLIPR® assays for calcium accumulation. We demonstrate that almorexant reaches equilibrium very slowly at OX₂R, whereas SB-649868, suvorexant, and filorexant may take hours to reach steady state at both orexin receptors. By contrast, compounds such as BBAC or the selective OX₂R antagonist IPSU ((2-((1H-Indol-3-yl)methyl)-9-(4-methoxypyrimidin-2-yl)-2,9-diazaspiro[5.5]undecan-1-one) bind rapidly and reach equilibrium very quickly in binding and/or functional assays. Overall, the “dual” antagonists tested here tend to be rather unselective under non-equilibrium conditions and reach equilibrium very slowly. Once equilibrium is reached, each ligand demonstrates a selectivity profile that is however, distinct from the non-equilibrium condition. The slow kinetics of the “dual” antagonists tested suggest that in vitro receptor occupancy may be longer lasting than would be predicted. This raises questions as to whether pharmacokinetic studies measuring plasma or brain levels of these antagonists are accurate reflections of receptor occupancy in vivo.

Identification of a novel series of orexin receptor antagonists with a distinct effect on sleep architecture for the treatment of insomnia

Dual orexin receptor (OXR) antagonists (DORAs) such as almorexant, 1 (SB-649868), or suvorexant have shown promise for the treatment of insomnias and sleep disorders in several recent clinical trials in volunteers and primary insomnia patients. The relative contribution of antagonism of OX1R and OX2R for sleep induction is still a matter of debate. We therefore initiated a drug discovery project with the aim of creating both OX2R selective antagonists and DORAs. Here we report that the OX2R selective antagonist 26 induced sleep in mice primarily by increasing NREM sleep, whereas the DORA suvorexant induced sleep largely by increasing REM sleep. Thus, OX2R selective antagonists may also be beneficial for the treatment of insomnia.

Selective and specific inhibition of the plasmodium falciparum lysyl-tRNA synthetase by the fungal secondary metabolite cladosporin

With renewed calls for malaria eradication, next-generation antimalarials need be active against drug-resistant parasites and efficacious against both liver- and blood-stage infections. We screened a natural product library to identify inhibitors of Plasmodium falciparum blood- and liver-stage proliferation. Cladosporin, a fungal secondary metabolite whose target and mechanism of action are not known for any species, was identified as having potent, nanomolar, antiparasitic activity against both blood and liver stages. Using postgenomic methods, including a yeast deletion strains collection, we show that cladosporin specifically inhibits protein synthesis by directly targeting P. falciparum cytosolic lysyl-tRNA synthetase. Further, cladosporin is >100-fold more potent against parasite lysyl-tRNA synthetase relative to the human enzyme, which is conferred by the identity of two amino acids within the enzyme active site. Our data indicate that lysyl-tRNA synthetase is an attractive, druggable, antimalarial target that can be selectively inhibited.

Different membrane modifications revealed by atomic force/lateral force microscopy after doping of human pancreatic cells with Cd, Zn, or Pb

The interaction of the cytotoxic metals cadmium, zinc, and lead with pancreatic cells was studied by atomic force/lateral Force microscopy (AFM/LFM), an approach that provides both topographic (with nanometer scale lateral resolution) and chemical information on the membrane. Different morphological modifications of the overall cell shape and roughness took place as consequence of 100 muM metal-dependent treatment. Furthermore, after exposure to Cd(Cl(2)) and Zn(Cl(2)), but not Pb(Cl(2)), the LFM images revealed several areas of the cell's surface showing lateral friction contrasts that have been interpreted as marker of different alterations of the cell physiology induced by the metal loading. Thus, the coupling of LFM detection to topographic AFM characterization allows to distinguish, through a nondestructive and surface characterising approach, between different metal-induced cytotoxic effects on cells. In this framework, the role of the LFM as an important tool to discriminate between different alteration of a biological system has to be highlighted.

Virtual screening to enrich a compound collection with CDK2 inhibitors using docking, scoring, and composite scoring models

Docking programs can generate subsets of a compound collection with an increased percentage of actives against a target (enrichment) by predicting their binding mode (pose) and affinity (score), and retrieving those with the highest scores. Using the QXP and GOLD programs, we compared the ability of six single scoring functions (PLP, Ligscore, Ludi, Jain, ChemScore, PMF) and four composite scoring models (Mean Rank: MR, Rank-by-Vote: Vt, Bayesian Statistics: BS and PLS Discriminant Analysis: DA) to separate compounds that are active against CDK2 from inactives. We determined the enrichment for the entire set of actives (IC50 < 10 microM) and for three activity subsets. In all cases, the enrichment for each subset was lower than for the entire set of actives. QXP outperformed GOLD at pose prediction, but yielded only moderately better enrichments. Five to six scoring functions yielded good enrichments with GOLD poses, while typically only two worked well with QXP poses. For each program, two scoring functions generally performed better than the others (Ligscore2 and Ludi for GOLD; QXP and Jain for QXP). Composite scoring functions yielded better results than single scoring functions. The consensus approaches MR and Vt worked best when separating micromolar inhibitors from inactives. The statistical approaches BS and DA, which require training data, performed best when distinguishing between low and high nanomolar inhibitors. The key observation that all hit rate profiles for all four activity intervals for all scoring schemes for both programs are significantly better than random, is evidence that docking can be successfully applied to enrich compound collections.

The environment of amide groups in protein–ligand complexes: H-bonds and beyond

A comprehensive structural analysis of interactions involving amide NH and C=O groups in protein-ligand complexes has been performed based on 3,275 published crystal structures (resolution < or =2.5 A). Most of the amide C=O and NH groups at the protein-ligand interface are highly buried within the binding site and involved in H-bonds with corresponding counter-groups. Small percentages of C=O and NH groups are solvated or embedded in hydrophobic environments. In particular, C=O groups show a higher propensity to be solvated or embedded in a hydrophobic environment than NH groups do. A small percentage of carbonyl groups is involved in weak hydrogen bonds with CH. Cases of dipolar interactions, involving carbonyl oxygen and electrophilic carbon atoms, such as amide, amidinium, guanidium groups, are also identified. A higher percentage of NH are in contact with aromatic carbons, interacting either through hydrogen bonds (preferably with the NH group pointing towards a ring carbon atom) or through stacking between amide plane and ring plane. Comprehensive studies such as the present one are thought to be important for future improvements in the molecular design area, in particular for the development of new scoring functions. [Figure: see text].

Protein dynamics, thermal stability, and free-energy landscapes: a molecular dynamics investigation

Proteins have a complex free-energy landscape because of their rich topology and the nature of their nonbonded interaction potential. This has important consequences because the roughness of the landscape affects the ease with which a chain folds and also determines the dynamic behavior of the folded structure, thus influencing its functional and stability properties. A detailed description of the free-energy landscape is therefore of paramount importance for a quantitative understanding of the relationships between structure, dynamics, stability, and functional behavior of proteins. The free-energy landscape of a protein is a high-dimensional hypersurface, difficult to rationalize. Therefore, achieving its detailed graphical representation in a way that goes beyond the familiar funnel-like free-energy model is still a big challenge. We describe here an approach based on global structural parameters that allows a two-dimensional representation of the free-energy landscape from simulated atomic trajectories. As shown in this and in the accompanying article, our representation of the landscape, combined with other conformational analyses, provides valuable information on its roughness and on how atomic trajectories evolve with time.

Dynamics of RNase-A and S-protein: a molecular dynamics simulation of the transition toward a folding intermediate

The description at atomic level of protein folding is an ambitious goal in biophysics, particularly because of the difficulty in obtaining structural information on unfolded states. Computer simulations can contribute in achieving this goal. Here we report the results of a 10-ns comparative simulation on bovine ribonuclease A and its S-protein, obtained by removal from the native molecule of the first 20 residues, the so-called S-peptide. The atomic trajectories have been analyzed by standard procedures and by applying concepts previously developed for disordered systems. Furthermore, we used a novel approach, described in the preceding paper, to represent graphically the energy landscape of the simulated systems. Relative to RNase-A, the S-protein, while largely maintaining its structural organization, displays an increased structural flexibility, it gains ergodicity and its core loses order, thus indicating that the removal of the S-peptide from ribonuclease A triggers the transition to a folding intermediate with reduced compactness. This finding also has biochemical relevance since the S-protein is recognized as not properly folded by the machinery responsible for the control of the folding quality in the endoplasmic reticulum.

Assessment of Docking Poses: Interactions-Based Accuracy Classification (IBAC) versus Crystal Structure Deviations

Six docking programs (FlexX, GOLD, ICM, LigandFit, the Northwestern University version of DOCK, and QXP) were evaluated in terms of their ability to reproduce experimentally observed binding modes (poses) of small-molecule ligands to macromolecular targets. The accuracy of a pose was assessed in two ways: First, the RMS deviation of the predicted pose from the crystal structure was calculated. Second, the predicted pose was compared to the experimentally observed one regarding the presence of key interactions with the protein. The latter assessment is referred to as interactions-based accuracy classification (IBAC). In a number of cases significant discrepancies were found between IBAC and RMSD-based classifications. Despite being more subjective, the IBAC proved to be a more meaningful measure of docking accuracy in all these cases.

Novel Scoring Functions Comprising QXP, SASA, and Protein Side-Chain Entropy Terms

Novel scoring functions that predict the affinity of a ligand for its receptor have been developed. They were built with several statistical tools (partial least squares, genetic algorithms, neural networks) and trained on a data set of 100 crystal structures of receptor-ligand complexes, with affinities spanning 10 log units. The new scoring functions contain both descriptors generated by the QXP docking program and new descriptors that were developed in-house. These new descriptors are based on solvent accessible surface areas and account for conformational entropy changes and desolvation effects of both ligand and receptor upon binding. The predictive r(2) values for a test set of 24 complexes are in the 0.712-0.741 range and RMS prediction errors in the 1.09-1.16 log K(d) range. Inclusion of the new descriptors led to significant improvements in affinity prediction, compared to scoring functions based on QXP descriptors alone. However, the QXP descriptors by themselves perform better in binding mode prediction. The performance of the linear models is comparable to that of the neural networks. The new functions perform very well, but they still need to be validated as universal tools for the prediction of binding affinity.