Are Deep Learning Structural Models Sufficiently Accurate for Virtual Screening? Application of Docking Algorithms to AlphaFold2 Predicted Structures.

Machine learning-based protein structure prediction algorithms, such as RosettaFold and AlphaFold2, have greatly impacted the structural biology field, arousing a fair amount of discussion around their potential role in drug discovery. While there are few preliminary studies addressing the usage of these models in virtual screening, none of them focus on the prospect of hit-finding in a real-world virtual screen with a model based on low prior structural information. In order to address this, we have developed an AlphaFold2 version where we exclude all structural templates with more than 30% sequence identity from the model-building process. In a previous study, we used those models in conjunction with state-of-the-art free energy perturbation methods and demonstrated that it is possible to obtain quantitatively accurate results. In this work, we focus on using these structures in rigid receptor-ligand docking studies. Our results indicate that using out-of-the-box Alphafold2 models is not an ideal scenario for virtual screening campaigns; in fact, we strongly recommend to include some post-processing modeling to drive the binding site into a more realistic holo model.

Are Deep Learning Structural Models Sufficiently Accurate for Free-Energy Calculations? Application of FEP+ to AlphaFold2-Predicted Structures.

The availability of AlphaFold2 has led to great excitement in the scientific community─particularly among drug hunters─due to the ability of the algorithm to predict protein structures with high accuracy. However, beyond globally accurate protein structure prediction, it remains to be determined whether ligand binding sites are predicted with sufficient accuracy in these structures to be useful in supporting computationally driven drug discovery programs. We explored this question by performing free-energy perturbation (FEP) calculations on a set of well-studied protein-ligand complexes, where AlphaFold2 predictions were performed by removing all templates with >30% identity to the target protein from the training set. We observed that in most cases, the ΔΔG values for ligand transformations calculated with FEP, using these prospective AlphaFold2 structures, were comparable in accuracy to the corresponding calculations previously carried out using crystal structures. We conclude that under the right circumstances, AlphaFold2-modeled structures are accurate enough to be used by physics-based methods such as FEP in typical lead optimization stages of a drug discovery program.

Targeted Degradation of the Oncogenic Phosphatase SHP2.

SHP2 is a protein tyrosine phosphatase that plays a critical role in the full activation of the Ras-MAPK pathway upon stimulation of receptor tyrosine kinases, which are frequently amplified or mutationally activated in human cancer. In addition, activating mutations in SHP2 result in developmental disorders and hematologic malignancies. Several allosteric inhibitors have been developed for SHP2 and are currently in clinical trials. Here, we report the development and evaluation of a SHP2 PROTAC created by conjugating RMC-4550 with pomalidomide using a PEG linker. This molecule is highly selective for SHP2, induces degradation of SHP2 in leukemic cells at submicromolar concentrations, inhibits MAPK signaling, and suppresses cancer cell growth. SHP2 PROTACs serve as an alternative strategy for targeting ERK-dependent cancers and are useful tools alongside allosteric inhibitors for dissecting the mechanisms by which SHP2 exerts its oncogenic activity.

Activation loop targeting strategy for design of receptor-interacting protein kinase 2 (RIPK2) inhibitors.

Development of selective kinase inhibitors remains a challenge due to considerable amino acid sequence similarity among family members particularly in the ATP binding site. Targeting the activation loop might offer improved inhibitor selectivity since this region of kinases is less conserved. However, the strategy presents difficulties due to activation loop flexibility. Herein, we report the design of receptor-interacting protein kinase 2 (RIPK2) inhibitors based on pan-kinase inhibitor regorafenib that aim to engage basic activation loop residues Lys169 or Arg171. We report development of CSR35 that displayed >10-fold selective inhibition of RIPK2 versus VEGFR2, the target of regorafenib. A co-crystal structure of CSR35 with RIPK2 revealed a resolved activation loop with an ionic interaction between the carboxylic acid installed in the inhibitor and the side-chain of Lys169. Our data provides principle feasibility of developing activation loop targeting type II inhibitors as a complementary strategy for achieving improved selectivity.

Discovery of LRRK2 inhibitors using sequential in silico joint pharmacophore space (JPS) and ensemble docking.

Joint pharmacophore space (JPS), ensemble docking and sequential JPS-ensemble docking were used to select three panels of compounds (10 per panel) for evaluation as LRRK2 inhibitors. These computational methods identified four LRRK2 inhibitors with IC50 values <12µM. The sequential JPS-ensemble docking predicted the majority of active hits. One of the inhibitors (Z-8205) identified using this method was also found to inhibit the G2019S mutant of LRRK2 25-fold better than wild-type enzyme. This bias for the G2019S mutant is proposed to arise from an interaction with S2019 in this form of the enzyme. In addition, Z-8205 was found to only inhibit one other kinase when profiled against a panel of 97 kinases at 10µM.

Type II kinase inhibitors show an unexpected inhibition mode against Parkinson's disease-linked LRRK2 mutant G2019S.

A number of well-known type II inhibitors (ATP-noncompetitive) that bind kinases in their DFG-out conformation were tested against wild-type LRRK2 and the most common Parkinson's disease-linked mutation, G2019S. We found that traditional type II inhibitors exhibit surprising variability in their inhibition mechanism between the wild type (WT) and the G2019S mutant of LRRK2. The type II kinase inhibitors were found to work in an ATP-competitive fashion against the G2019S mutant, whereas they appear to follow the expected noncompetitive mechanism against WT. Because the G2019S mutation lies in the DXG motif (DYG in LRRK2 but DFG in most other kinases) of the activation loop, we explored the structural consequence of the mutation on loop dynamics using an enhanced sampling method called metadynamics. The simulations suggest that the G2019S mutation stabilizes the DYG-in state of LRRK2 through a series of hydrogen bonds, leading to an increase in the conformational barrier between the active and inactive forms of the enzyme and a relative stabilization of the active form. The conformational bias toward the active form of LRRK2 mutants has two primary consequences. (1) The mutant enzyme becomes hyperactive, a known contributor to the Parkinsonian phenotype, as a consequence of being "locked" into the activated state, and (2) the mutation creates an unusual allosteric pocket that can bind type II inhibitors but in an ATP-competitive fashion. Our results suggest that developing type II inhibitors, which are generally considered superior to type I inhibitors because of desirable selectivity profiles, might be especially challenging for the G2019S LRRK2 mutant.

Discovery and Optimization of a STING Agonist Platform for Application in Antibody Drug Conjugates.

While STING agonists have proven to be effective preclinically as anti-tumor agents, these promising results have yet to be translated in the clinic. A STING agonist antibody-drug conjugate (ADC) could overcome current limitations by improving tumor accessibility, allowing for systemic administration as well as tumor-localized activation of STING for greater anti-tumor activity and better tolerability. In line with this effort, a STING agonist ADC platform was identified through systematic optimization of the payload, linker, and scaffold based on multiple factors including potency and specificity in both in vitro and in vivo evaluations. The platform employs a potent non-cyclic dinucleotide STING agonist, a cleavable ester-based linker, and a hydrophilic PEG8-bisglucamine scaffold. A tumor-targeted ADC built with the resulting STING agonist platform induced robust and durable anti-tumor activity and demonstrated high stability and favorable pharmacokinetics in nonclinical species.

Fluorescence polarization assay for inhibitors of the kinase domain of receptor interacting protein 1.

Necrotic cell death is prevalent in many different pathological disease states and in traumatic injury. Necroptosis is a form of necrosis that stems from specific signaling pathways, with the key regulator being receptor interacting protein 1 (RIP1), a serine/threonine kinase. Specific inhibitors of RIP1, termed necrostatins, are potent inhibitors of necroptosis. Necrostatins are structurally distinct from one another yet still possess the ability to inhibit RIP1 kinase activity. To further understand the differences in the binding of the various necrostatins to RIP1 and to develop a robust high-throughput screening (HTS) assay, which can be used to identify new classes of RIP1 inhibitors, we synthesized fluorescein derivatives of Necrostatin-1 (Nec-1) and Nec-3. These compounds were used to establish a fluorescence polarization (FP) assay to directly measure the binding of necrostatins to RIP1 kinase. The fluorescein-labeled compounds are well suited for HTS because the assays have a dimethyl sulfoxide (DMSO) tolerance up to 5% and Z' scores of 0.62 (fluorescein-Nec-1) and 0.57 (fluorescein-Nec-3). In addition, results obtained from the FP assays and ligand docking studies provide insights into the putative binding sites of Nec-1, Nec-3, and Nec-4.

Structure-activity relationship study of beta-carboline derivatives as haspin kinase inhibitors.

Haspin is a serine/threonine kinase that phosphorylates Thr-3 of histone H3 in mitosis that has emerged as a possible cancer therapeutic target. High throughput screening of approximately 140,000 compounds identified the beta-carbolines harmine and harmol as moderately potent haspin kinase inhibitors. Based on information obtained from a structure-activity relationship study previously conducted for an acridine series of haspin inhibitors in conjunction with in silico docking using a recently disclosed crystal structure of the kinase, harmine analogs were designed that resulted in significantly increased haspin kinase inhibitory potency. The harmine derivatives also demonstrated less activity towards DYRK2 compared to the acridine series. In vitro mouse liver microsome stability and kinase profiling of a representative member of the harmine series (42, LDN-211898) are also presented.

Structure-activity relationship study of selective benzimidazole-based inhibitors of Cryptosporidium parvum IMPDH.

Cryptosporidium parasites are important waterborne pathogens of both humans and animals. The Cryptosporidium parvum and Cryptosporidium hominis genomes indicate that the only route to guanine nucleotides is via inosine 5'-monophosphate dehydrogenase (IMPDH). Thus the inhibition of the parasite IMPDH presents a potential strategy for treating Cryptosporidium infections. A selective benzimidazole-based inhibitor of C. parvum IMPDH (CpIMPDH) was previously identified in a high throughput screen. Here we report a structure-activity relationship study of benzimidazole-based compounds that resulted in potent and selective inhibitors of CpIMPDH. Several compounds display potent antiparasitic activity in vitro.

Engineered disulfide bonds restore chaperone-like function of DJ-1 mutants linked to familial Parkinson's disease.

Loss-of-function mutations such as L166P, A104T, and M26I in the DJ-1 gene (PARK7) have been linked to autosomal-recessive early onset Parkinson's disease (PD). Cellular and structural studies of the familial mutants suggest that these mutations may destabilize the dimeric structure. To look for common dynamical signatures among the DJ-1 mutants, short MD simulations of up to 1000 ps were conducted to identify the weakest region of the protein (residues 38-70). In an attempt to stabilize the protein, we mutated residue Val 51 to cysteine (V51C) to make a symmetry-related disulfide bridge with the preexisting Cys 53 on the opposite subunit. We found that the introduction of this disulfide linkage stabilized the mutants A104T and M26I against thermal denaturation, improved their ability to scavenge reactive oxygen species (ROS), and restored a chaperone-like function of blocking alpha-synuclein aggregation. The L166P mutant was far too unstable to be rescued by introduction of the V51C mutation. The results presented here point to the possible development of pharmacological chaperones, which may eventually lead to PD therapeutics.

Design, Synthesis, and Biological Evaluation of a Series of Oxazolone Carboxamides as a Novel Class of Acid Ceramidase Inhibitors.

Acid ceramidase (AC) is a cysteine hydrolase that plays a crucial role in the metabolism of lysosomal ceramides, important members of the sphingolipid family, a diversified class of bioactive molecules that mediate many biological processes ranging from cell structural integrity, signaling, and cell proliferation to cell death. In the effort to expand the structural diversity of the existing collection of AC inhibitors, a novel class of substituted oxazol-2-one-3-carboxamides were designed and synthesized. Herein, we present the chemical optimization of our initial hits, 2-oxo-4-phenyl-N-(4-phenylbutyl)oxazole-3-carboxamide 8a and 2-oxo-5-phenyl-N-(4-phenylbutyl)oxazole-3-carboxamide 12a, which resulted in the identification of 5-[4-fluoro-2-(1-methyl-4-piperidyl)phenyl]-2-oxo-N-pentyl-oxazole-3-carboxamide 32b as a potent AC inhibitor with optimal physicochemical and metabolic properties, showing target engagement in human neuroblastoma SH-SY5Y cells and a desirable pharmacokinetic profile in mice, following intravenous and oral administration. 32b enriches the arsenal of promising lead compounds that may therefore act as useful pharmacological tools for investigating the potential therapeutic effects of AC inhibition in relevant sphingolipid-mediated disorders.

Inosine-5'-monophosphate dehydrogenase (IMPDH) inhibitors: a patent and scientific literature review (2002-2016).

INTRODUCTION: Inosine-5'-monophosphate dehydrogenase (IMPDH) is an enzyme involved in the de novo biosynthesis of guanine nucleotides. To date human IMPDH inhibitors have been approved for prevention of organ transplant rejection and as anti-viral agents. More recently, the use of IMPDH inhibitors for other indications including cancer and pathogenic microorganisms has been pursued. Areas covered: IMPDH inhibitors disclosed primarily in the patent and scientific literature from 2002 to the present are discussed. Several interesting chemotypes that have not been pursued by patent protection are also highlighted. Expert opinion: Progress has been made in the development of IMPDH inhibitors, particularly compounds that are structurally distinct from mycophenolic acid and nucleoside-based inhibitors. However, clinical progression has been hampered primarily by a limited understanding of the enzyme's role in disease pathophysiology. Finally, most of the IMPDH inhibitors developed over the past fourteen years fall within a relatively narrow set of chemotypes. This provides opportunities for expanding IMPDH inhibitor chemical space to further evaluate this class of molecular targets.

Cocrystal structures of diaminopimelate decarboxylase: mechanism, evolution, and inhibition of an antibiotic resistance accessory factor.

Cocrystal structures of Methanococcus jannaschii diaminopimelate decarboxylase (DAPDC) bound to a substrate analog, azelaic acid, and its L-lysine product have been determined at 2.6 A and 2.0 A, respectively. This PLP-dependent enzyme is responsible for the final step of L-lysine biosynthesis in bacteria and plays a role in beta-lactam antibiotic resistance in Staphylococcus aureus. Substrate specificity derives from recognition of the L-chiral center of diaminopimelate and a system of ionic "molecular rulers" that dictate substrate length. A coupled-enzyme assay system permitted measurement of kinetic parameters for recombinant DAPDCs and inhibition constants (K(i)) for azelaic acid (89 microM) and other substrate analogs. Implications for rational design of broad-spectrum antimicrobial agents targeted against DAPDCs of drug-resistant strains of bacterial pathogens, such as Staphylococcus aureus, are discussed.

Structure guided design of potent and selective ponatinib-based hybrid inhibitors for RIPK1.

RIPK1 and RIPK3, two closely related RIPK family members, have emerged as important regulators of pathologic cell death and inflammation. In the current work, we report that the Bcr-Abl inhibitor and anti-leukemia agent ponatinib is also a first-in-class dual inhibitor of RIPK1 and RIPK3. Ponatinib potently inhibited multiple paradigms of RIPK1- and RIPK3-dependent cell death and inflammatory tumor necrosis factor alpha (TNF-α) gene transcription. We further describe design strategies that utilize the ponatinib scaffold to develop two classes of inhibitors (CS and PN series), each with greatly improved selectivity for RIPK1. In particular, we detail the development of PN10, a highly potent and selective "hybrid" RIPK1 inhibitor, capturing the best properties of two different allosteric RIPK1 inhibitors, ponatinib and necrostatin-1. Finally, we show that RIPK1 inhibitors from both classes are powerful blockers of TNF-induced injury in vivo. Altogether, these findings outline promising candidate molecules and design approaches for targeting RIPK1- and RIPK3-driven inflammatory pathologies.

Enzymatic Characterization of ER Stress-Dependent Kinase, PERK, and Development of a High-Throughput Assay for Identification of PERK Inhibitors.

PERK is serine/threonine kinase localized to the endoplasmic reticulum (ER) membrane. PERK is activated and contributes to cell survival in response to a variety of physiological stresses that affect protein quality control in the ER, such as hypoxia, glucose depravation, increased lipid biosynthesis, and increased protein translation. Pro-survival functions of PERK are triggered by such stresses, suggesting that development of small-molecule inhibitors of PERK may be efficacious in a variety of disease scenarios. Hence, we have conducted a detailed enzymatic characterization of the PERK kinase to develop a high-throughput-screening assay (HTS) that will permit the identification of small-molecule PERK inhibitors. In addition to establishing the K(m) of PERK for both its primary substrate, eIF2α, and for adenosine triphosphate, further mechanistic studies revealed that PERK targets its substrate via either a random/steady-state ordered mechanism. For HTS, we developed a time-resolved fluorescence resonance energy transfer-based assay that yielded a robust Z' factor and percent coefficient of variation value, enabling the successful screening of 79,552 compounds. This approach yielded one compound that exhibited good in vitro and cellular activity. These results demonstrate the validity of this screen and represent starting points for drug discovery efforts.

Improving binding specificity of pharmacological chaperones that target mutant superoxide dismutase-1 linked to familial amyotrophic lateral sclerosis using computational methods.

We recently described a set of drug-like molecules obtained from an in silico screen that stabilize mutant superoxide dismutase-1 (SOD-1) linked to familial amyotrophic lateral sclerosis (ALS) against unfolding and aggregation but exhibited poor binding specificity toward SOD-1 in presence of blood plasma. A reasonable but not a conclusive model for the binding of these molecules was proposed on the basis of restricted docking calculations and site-directed mutagenesis of key residues at the dimer interface. A set of hydrogen bonding constraints obtained from these experiments were used to guide docking calculations with compound library around the dimer interface. A series of chemically unrelated hits were predicted, which were experimentally tested for their ability to block aggregation. At least six of the new molecules exhibited high specificity of binding toward SOD-1 in the presence of blood plasma. These molecules represent a new class of molecules for further development into clinical candidates.

Destabilization of DJ-1 by familial substitution and oxidative modifications: implications for Parkinson's disease.

Parkinson's disease (PD) is a neurodegenerative disorder characterized by oxidative stress and protein aggregation. Both toxic phenomena are mitigated by DJ-1, a homodimeric protein with proposed antioxidant and chaperone activities. The neuroprotective function of DJ-1 is modulated by oxidation of cysteine 106, a residue that may act as an oxidative stress sensor. Loss-of-function mutations in the DJ-1 gene have been linked to early onset PD, and age-dependent over-oxidation of DJ-1 is thought to contribute to sporadic PD. The familial mutant L166P fails to dimerize and is rapidly degraded, suggesting that protein destabilization accounts for the dysfunction of this mutant. In this study, we investigated how the structure and stability of DJ-1 are impacted by two other pathogenic substitutions (M26I and E64D) and by over-oxidation with H2O2. Whereas the recombinant wild-type protein and E64D both adopted a stable dimeric structure, M26I showed an increased propensity to aggregate and decreased secondary structure. Similar to M26I, over-oxidized wild-type DJ-1 exhibited reduced secondary structure, and this property correlated with destabilization of the dimer. The engineered mutant C106A had a greater thermodynamic stability and was more resistant to oxidation-induced destabilization than the wild-type protein. These results suggest that (i) the M26I substitution and over-oxidation destabilize dimeric DJ-1, and (ii) the oxidation of cysteine 106 contributes to DJ-1 destabilization. Our findings provide a structural basis for DJ-1 dysfunction in familial and sporadic PD, and they suggest that dimer stabilization is a reasonable therapeutic strategy to treat both forms of this disorder.

Structural basis for conformational plasticity of the Parkinson's disease-associated ubiquitin hydrolase UCH-L1.

The ubiquitin C-terminal hydrolase UCH-L1 (PGP9.5) comprises >1% of total brain protein but is almost absent from other tissues [Wilkinson, K. D., et al. (1989) Science 246, 670-673]. Mutations in the UCH-L1 gene have been reported to be linked to susceptibility to and protection from Parkinson's disease [Leroy, E., et al. (1998) Nature 395, 451-452; Maraganore, D. M., et al. (1999) Neurology 53, 1858-1860]. Abnormal overexpression of UCH-L1 has been shown to correlate with several forms of cancer [Hibi, K., et al. (1998) Cancer Res. 58, 5690-5694]. Because the amino acid sequence of UCH-L1 is similar to that of other ubiquitin C-terminal hydrolases, including the ubiquitously expressed UCH-L3, which appear to be unconnected to neurodegenerative disease, the structure of UCH-L1 and the effects of disease associated mutations on the structure and function are of considerable importance. We have determined the three-dimensional structure of human UCH-L1 at 2.4-A resolution by x-ray crystallography. The overall fold resembles that of other ubiquitin hydrolases, including UCH-L3, but there are a number of significant differences. In particular, the geometry of the catalytic residues in the active site of UCH-L1 is distorted in such a way that the hydrolytic activity would appear to be impossible without substrate induced conformational rearrangements.