Crystal Structure of Antagonist Bound Human Lysophosphatidic Acid Receptor 1.

Lipid biology continues to emerge as an area of significant therapeutic interest, particularly as the result of an enhanced understanding of the wealth of signaling molecules with diverse physiological properties. This growth in knowledge is epitomized by lysophosphatidic acid (LPA), which functions through interactions with at least six cognate G protein-coupled receptors. Herein, we present three crystal structures of LPA1 in complex with antagonist tool compounds selected and designed through structural and stability analyses. Structural analysis combined with molecular dynamics identified a basis for ligand access to the LPA1 binding pocket from the extracellular space contrasting with the proposed access for the sphingosine 1-phosphate receptor. Characteristics of the LPA1 binding pocket raise the possibility of promiscuous ligand recognition of phosphorylated endocannabinoids. Cell-based assays confirmed this hypothesis, linking the distinct receptor systems through metabolically related ligands with potential functional and therapeutic implications for treatment of disease.

A rational approach to assess off-target reactivity of a dual-signal integrator for T cell therapy.

Cell therapy is an emerging therapeutic modality with the power to exploit new cancer targets and potentially achieve positive outcomes for patients with few other options. Like all synthetic treatments, cell therapy has the risk of toxicity via unpredicted off-target behavior. We describe an empirical method to model off-tumor, off-target reactivity of receptors used for investigational T cell therapies. This approach utilizes an optimal panel of diverse human cell-lines to capture the large majority of protein-coding gene expression in adult human tissues. We apply this cell-line set to test Jurkat and primary T cells engineered with a dual-signal integrator, called TmodTM, that contains an activating receptor (activator) and a separate inhibitory receptor (blocker). In proof-of-concept experiments, we use CD19 as the activating antigen and HLA-A*02 as the blocker antigen. This specific Tmod system, which employs a blocker targeting a ubiquitously expressed HLA class I antigen to inhibit CAR activation, has an inherent mechanism for selectivity/safety, designed to activate only when a specific HLA class I antigen is lost. Nonetheless, it is important to test off-target reactivity in functional assays, especially given the disconnect between ligand-binding and function among T cell receptors (TCRs) and chimeric antigen receptors (CARs). We show these cell-based assays yield consistent results with high sensitivity and specificity. The general strategy is likely applicable to more traditional single-receptor CAR- and TCR-T therapeutics.

Antibody structure determination using a combination of homology modeling, energy-based refinement, and loop prediction.

We present the blinded prediction results in the Second Antibody Modeling Assessment (AMA-II) using a fully automatic antibody structure prediction method implemented in the programs BioLuminate and Prime. We have developed a novel knowledge based approach to model the CDR loops, using a combination of sequence similarity, geometry matching, and the clustering of database structures. The homology models are further optimized with a physics-based energy function (VSGB2.0), which improves the model quality significantly. H3 loop modeling remains the most challenging task. Our ab initio loop prediction performs well for the H3 loop in the crystal structure context, and allows improved results when refining the H3 loops in the context of homology models. For the 10 human and mouse derived antibodies in this assessment, the average RMSDs for the homology model Fv and framework regions are 1.19 Å and 0.74 Å, respectively. The average RMSDs for five non-H3 CDR loops range from 0.61 Å to 1.05 Å, and the H3 loop average RMSD is 2.91 Å using our knowledge-based loop prediction approach. The ab initio H3 loop predictions yield an average RMSD of 1.28 Å when performed in the context of the crystal structure and 2.67 Å in the context of the homology modeled structure. Notably, our method for predicting the H3 loop in the crystal structure environment ranked first among the seven participating groups in AMA-II, and our method made the best prediction among all participants for seven of the ten targets.

Structure-based virtual screening approach for discovery of covalently bound ligands.

We present a fast and effective covalent docking approach suitable for large-scale virtual screening (VS). We applied this method to four targets (HCV NS3 protease, Cathepsin K, EGFR, and XPO1) with known crystal structures and known covalent inhibitors. We implemented a customized "VS mode" of the Schrödinger Covalent Docking algorithm (CovDock), which we refer to as CovDock-VS. Known actives and target-specific sets of decoys were docked to selected X-ray structures, and poses were filtered based on noncovalent protein-ligand interactions known to be important for activity. We were able to retrieve 71%, 72%, and 77% of the known actives for Cathepsin K, HCV NS3 protease, and EGFR within 5% of the decoy library, respectively. With the more challenging XPO1 target, where no specific interactions with the protein could be used for postprocessing of the docking results, we were able to retrieve 95% of the actives within 30% of the decoy library and achieved an early enrichment factor (EF1%) of 33. The poses of the known actives bound to existing crystal structures of 4 targets were predicted with an average RMSD of 1.9 Å. To the best of our knowledge, CovDock-VS is the first fully automated tool for efficient virtual screening of covalent inhibitors. Importantly, CovDock-VS can handle multiple chemical reactions within the same library, only requiring a generic SMARTS-based predefinition of the reaction. CovDock-VS provides a fast and accurate way of differentiating actives from decoys without significantly deteriorating the accuracy of the predicted poses for covalent protein-ligand complexes. Therefore, we propose CovDock-VS as an efficient structure-based virtual screening method for discovery of novel and diverse covalent ligands.

Effect of membrane tension on the physical properties of DOPC lipid bilayer membrane.

Molecular dynamics simulations of a dioleoylphosphocholine (DOPC) lipid bilayer were performed to explore its mechanosensitivity. Variations in the bilayer properties, such as area per lipid, volume, thickness, hydration depth (HD), hydration thickness (HT), lateral diffusion coefficient, and changes in lipid structural order were computed in the membrane tension range 0 to 15dyn/cm. We determined that an increase in membrane tension results in a decrease in the bilayer thickness and HD of ~5% and ~5.7% respectively, whereas area per lipid, volume, and HT/HD increased by 6.8%, 2.4%, and 5% respectively. The changes in lipid conformation and orientation were characterized using orientational (S(2)) and deuterium (S(CD)) order parameters. Upon increase of membrane tension both order parameters indicated an increase in lipid disorder by 10-20%, mostly in the tail end region of the hydrophobic chains. The effect of membrane tension on lipid lateral diffusion in the DOPC bilayer was analyzed on three different time scales corresponding to inertial motion, anomalous diffusion and normal diffusion. The results showed that lateral diffusion of lipid molecules is anomalous in nature due to the non-exponential distribution of waiting times. The anomalous and normal diffusion coefficients increased by 20% and 52% when the membrane tension changed from 0 to 15dyn/cm, respectively. In conclusion, our studies showed that membrane tension causes relatively significant changes in the area per lipid, volume, polarity, membrane thickness, and fluidity of the membrane suggesting multiple mechanisms by which mechanical perturbation of the membrane could trigger mechanosensitive response in cells.

Effect of membrane tension on the electric field and dipole potential of lipid bilayer membrane.

The dipole potential of lipid bilayer membrane controls the difference in permeability of the membrane to oppositely charged ions. We have combined molecular dynamics (MD) simulations and experimental studies to determine changes in electric field and electrostatic potential of 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) lipid bilayer in response to applied membrane tension. MD simulations based on CHARMM36 force field showed that electrostatic potential of DOPC bilayer decreases by ~45mV in the physiologically relevant range of membrane tension values (0 to 15dyn/cm). The electrostatic field exhibits a peak (~0.8×10(9)V/m) near the water/lipid interface which shifts by 0.9Å towards the bilayer center at 15dyn/cm. Maximum membrane tension of 15dyn/cm caused 6.4% increase in area per lipid, 4.7% decrease in bilayer thickness and 1.4% increase in the volume of the bilayer. Dipole-potential sensitive fluorescent probes were used to detect membrane tension induced changes in DOPC vesicles exposed to osmotic stress. Experiments confirmed that dipole potential of DOPC bilayer decreases at higher membrane tensions. These results are suggestive of a potentially new mechanosensing mechanism by which mechanically induced structural changes in the lipid bilayer membrane could modulate the function of membrane proteins by altering electrostatic interactions and energetics of protein conformational states.

Consensus Induced Fit Docking (cIFD): methodology, validation, and application to the discovery of novel Crm1 inhibitors.

We present the Consensus Induced Fit Docking (cIFD) approach for adapting a protein binding site to accommodate multiple diverse ligands for virtual screening. This novel approach results in a single binding site structure that can bind diverse chemotypes and is thus highly useful for efficient structure-based virtual screening. We first describe the cIFD method and its validation on three targets that were previously shown to be challenging for docking programs (COX-2, estrogen receptor, and HIV reverse transcriptase). We then demonstrate the application of cIFD to the challenging discovery of irreversible Crm1 inhibitors. We report the identification of 33 novel Crm1 inhibitors, which resulted from the testing of 402 purchased compounds selected from a screening set containing 261,680 compounds. This corresponds to a hit rate of 8.2 %. The novel Crm1 inhibitors reveal diverse chemical structures, validating the utility of the cIFD method in a real-world drug discovery project. This approach offers a pragmatic way to implicitly account for protein flexibility without the additional computational costs of ensemble docking or including full protein flexibility during virtual screening.

Chimeric Antigen Receptors Directed at Mutant KRAS Exhibit an Inverse Relationship Between Functional Potency and Neoantigen Selectivity.

Neoantigens are among the most intriguing potential immuno-oncology targets because, unlike many cancer targets that are expressed on normal tissues, they are by definition restricted to cancer cells. Medicines directed at common neoantigens such as mutant KRAS are especially interesting because they may offer the convenience and cost of an off-the-shelf therapy. However, all common KRAS mutations produce proteins that differ from the wild type at a single amino acid, creating challenges for molecular discrimination. We have undertaken an effort to optimize single-chain variable fragments (scFv) against peptide/major histocompatibility antigen complexes composed of HLA-A*11 and either G12V- or G12D-mutant KRAS peptides. These scFvs could in principle be used in chimeric antigen receptor (CAR) T-cell therapies for selected patients whose tumors bear either of these mutations. Here we show that optimization of such CARs involves a trade-off between potency and selectivity. We further show that targeting this family without high selectivity engenders risks of cross-reactivity against other members of the G-protein family to which KRAS belongs. Significance: We report an effort to generate high potency, selective CARs directed at mutant KRAS peptides. Although the heavily optimized CARs maintain high selectivity against wild-type KRAS, they lose selectivity against other KRAS-related peptides derived from human proteins. To our knowledge, this work is the first to examine the trade-off between potency and selectivity with regard to KRAS pMHC-directed CARs, illustrating the challenge to achieve both sufficient potency and high selectivity.

Mesothelin-specific CAR-T cell therapy that incorporates an HLA-gated safety mechanism selectively kills tumor cells.

BACKGROUND: Mesothelin (MSLN) is a classic tumor-associated antigen that is expressed in lung cancer and many other solid tumors. However, MSLN is also expressed in normal mesothelium which creates a significant risk of serious inflammation for MSLN-directed therapeutics. We have developed a dual-receptor (Tmod™) system that exploits the difference between tumor and normal tissue in a subset of patients with defined heterozygous gene loss (LOH) in their tumors. METHODS: T cells engineered with the MSLN CAR Tmod construct described here contain (1) a novel MSLN-activated CAR and (2) an HLA-A*02-gated inhibitory receptor (blocker). A*02 binding is intended to override T-cell cytotoxicity, even in the presence of MSLN. The Tmod system is designed to treat heterozygous HLA class I patients, selected for HLA LOH. When A*02 is absent from tumors selected for LOH, the MSLN Tmod cells are predicted to mediate potent killing of the MSLN(+)A*02(-) malignant cells. RESULTS: The sensitivity of the MSLN Tmod cells is comparable with a benchmark MSLN CAR-T that was active but toxic in the clinic. Unlike MSLN CAR-T cells, the Tmod system robustly protects surrogate "normal" cells even in mixed-cell populations in vitro and in a xenograft model. The MSLN CAR can also be paired with other HLA class I blockers, supporting extension of the approach to patients beyond A*02 heterozygotes. CONCLUSIONS: The Tmod mechanism exemplified by the MSLN CAR Tmod construct provides an alternative route to leverage solid-tumor antigens such as MSLN in safer, more effective ways than previously possible.

HLA-A∗02-gated safety switch for cancer therapy has exquisite specificity for its allelic target antigen.

Innovative cell-based therapies are important new weapons in the fight against difficult-to-treat cancers. One promising strategy involves cell therapies equipped with multiple receptors to integrate signals from more than one antigen. We developed a specific embodiment of this approach called Tmod, a two-receptor system that combines activating and inhibitory inputs to distinguish between tumor and normal cells. The selectivity of Tmod is enforced by the inhibitory receptor (blocker) that recognizes an antigen, such as an HLA allele, whose expression is absent from tumors because of loss of heterozygosity. Although unwanted cross-reactivity of the blocker likely reduces efficacy rather than safety, it is important to verify the blocker's specificity. We have tested an A∗02-directed blocker derived from the PA2.1 mouse antibody as a safety mechanism paired with a mesothelin-specific activating CAR in our Tmod construct. We solved the crystal structure of humanized PA2.1 Fab in complex with HLA-A∗02 to determine its binding epitope, which was used to bioinformatically select specific class I HLA alleles to test the blocker's functional specificity in vitro. We found that this A∗02-directed blocker is highly specific for its cognate antigen, with only one cross-reactive allele (A∗69) capable of triggering comparable function.

Extensive functional comparisons between chimeric antigen receptors and T cell receptors highlight fundamental similarities.

Though TCRs have been subject to limited engineering in the context of therapeutic design and optimization, they are used largely as found in nature. On the other hand, CARs are artificial, composed of different segments of proteins that function in the immune system. This characteristic raises the possibility of altered response to immune regulatory stimuli. Here we describe a large-scale, systematic comparison of CARs and TCRs across 5 different pMHC targets, with a total of 19 constructs examined in vitro. These functional measurements include CAR- and TCR-mediated activation, proliferation, and cytotoxicity in both acute and chronic settings. Surprisingly, we find no consistent difference between CARs and TCRs as receptor classes with respect to their relative sensitivity to major regulators of T cell activation: PD-L1, CD80/86 and IL-2. Though TCRs often emerge from human blood directly as potent, selective receptors, CARs must be heavily optimized to attain these properties for pMHC targets. Nonetheless, when iteratively improved and compared head to head in functional tests, CARs appear remarkably similar to TCRs with respect to immune modulation.

Re-examination of MAGE-A3 as a T-cell Therapeutic Target.

In 2013, an innovative MAGE-A3-directed cancer therapeutic of great potential value was terminated in the clinic because of neurotoxicity. The safety problems were hypothesized to originate from off-target T-cell receptor activity against a closely related MAGE-A12 peptide. A combination of published and new data led us to test this hypothesis with current technology. Our results call into question MAGE-A12 as the source of the neurotoxicity. Rather, the data imply that an alternative related peptide from EPS8L2 may be responsible. Given the qualities of MAGE-A3 as an onco-testis antigen widely expressed in tumors and largely absent from normal adult tissues, these findings suggest that MAGE-A3 may deserve further consideration as a cancer target. As a step in this direction, the authors isolated 2 MAGE-A3 peptide-major histocompatibility complex-directed chimeric antigen receptors, 1 targeting the same peptide as the clinical T-cell receptor. Both chimeric antigen receptors have improved selectivity over the EPS8L2 peptide that represents a significant risk for MAGE-A3-targeted therapeutics, showing that there may be other options for MAGE-A3 cell therapy.

Potent, Selective CARs as Potential T-Cell Therapeutics for HPV-positive Cancers.

Next-generation T-cell therapies will likely continue to utilize T-cell receptors (TCRs) and chimeric antigen receptors (CARs) because each receptor type has advantages. TCRs often possess exceptional properties even when tested unmodified from patients' T cells. CARs are generally less sensitive, possibly because their ligand-binding domains are grafted from antibodies selected for binding affinity or avidity and not broadly optimized for a functional response. Because of the disconnect between binding and function among these receptor types, the ultimate potential of CARs optimized for sensitivity and selectivity is not clear. Here, we focus on a thoroughly studied immuno-oncology target, the HLA-A*02/HPV-E629-38 complex, and show that CARs can be optimized by a combination of high-throughput binding screens and low-throughput functional assays to have comparable activity to clinical TCRs in acute assays in vitro. These results provide a case study for the challenges and opportunities of optimizing high-performing CARs, especially in the context of targets utilized naturally by TCRs.

The integration of pharmacophore-based 3D QSAR modeling and virtual screening in safety profiling: A case study to identify antagonistic activities against adenosine receptor, A2A, using 1,897 known drugs.

Safety pharmacology screening against a wide range of unintended vital targets using in vitro assays is crucial to understand off-target interactions with drug candidates. With the increasing demand for in vitro assays, ligand- and structure-based virtual screening approaches have been evaluated for potential utilization in safety profiling. Although ligand based approaches have been actively applied in retrospective analysis or prospectively within well-defined chemical space during the early discovery stage (i.e., HTS screening and lead optimization), virtual screening is rarely implemented in later stage of drug discovery (i.e., safety). Here we present a case study to evaluate ligand-based 3D QSAR models built based on in vitro antagonistic activity data against adenosine receptor 2A (A2A). The resulting models, obtained from 268 chemically diverse compounds, were used to test a set of 1,897 chemically distinct drugs, simulating the real-world challenge of safety screening when presented with novel chemistry and a limited training set. Due to the unique requirements of safety screening versus discovery screening, the limitations of 3D QSAR methods (i.e., chemotypes, dependence on large training set, and prone to false positives) are less critical than early discovery screen. We demonstrated that 3D QSAR modeling can be effectively applied in safety assessment prior to in vitro assays, even with chemotypes that are drastically different from training compounds. It is also worth noting that our model is able to adequately make the mechanistic distinction between agonists and antagonists, which is important to inform subsequent in vivo studies. Overall, we present an in-depth analysis of the appropriate utilization and interpretation of pharmacophore-based 3D QSAR models for safety screening.

Single variable domains from the T cell receptor ß chain function as mono- and bifunctional CARs and TCRs.

Cell therapy using T cell receptors (TCRs) and chimeric antigen receptors (CARs) represents a new wave of immunotherapies garnering considerable attention and investment. Further progress in this area of medicine depends in part on improving the functional capabilities of the engineered components, while maintaining the overall size of recombinant constructs to ensure their compatibility with existing gene delivery vehicles. We describe a single-variable-domain TCR (svd TCR) that utilizes only the variable domain of the ß chain (Vß). This Vß module not only works in TCR and CAR formats, but also can be used to create single-chain bispecific CARs and TCRs. Comparison of individual ligand-binding Vß domains in different formats suggests that the lone Vß sequence controls the sensitivity and a major part of the specificity of the CAR or TCR construct, regardless of signaling format, in Jurkat and primary T cells.

PREDICT modeling and in-silico screening for G-protein coupled receptors.

G-protein coupled receptors (GPCRs) are a major group of drug targets for which only one x-ray structure is known (the nondrugable rhodopsin), limiting the application of structure-based drug discovery to GPCRs. In this paper we present the details of PREDICT, a new algorithmic approach for modeling the 3D structure of GPCRs without relying on homology to rhodopsin. PREDICT, which focuses on the transmembrane domain of GPCRs, starts from the primary sequence of the receptor, simultaneously optimizing multiple 'decoy' conformations of the protein in order to find its most stable structure, culminating in a virtual receptor-ligand complex. In this paper we present a comprehensive analysis of three PREDICT models for the dopamine D2, neurokinin NK1, and neuropeptide Y Y1 receptors. A shorter discussion of the CCR3 receptor model is also included. All models were found to be in good agreement with a large body of experimental data. The quality of the PREDICT models, at least for drug discovery purposes, was evaluated by their successful utilization in in-silico screening. Virtual screening using all three PREDICT models yielded enrichment factors 9-fold to 44-fold better than random screening. Namely, the PREDICT models can be used to identify active small-molecule ligands embedded in large compound libraries with an efficiency comparable to that obtained using crystal structures for non-GPCR targets.

Structure and dynamics of an imidazoline nitroxide side chain with strongly hindered internal motion in proteins.

A disulfide-linked imidazoline nitroxide side chain (V1) has a similar and highly constrained internal motion at diverse topological sites in a protein, unlike that for the disulfide-linked pyrroline nitroxide side chain (R1) widely used in site directed spin labeling EPR. Crystal structures of V1 at two positions in a helix of T4 Lysozyme and quantum mechanical calculations suggest the source of the constraints as intra-side chain interactions of the disulfide sulfur atoms with both the protein backbone and the 3-nitrogen in the imidazoline ring. These interactions apparently limit the conformation of the side chain to one of only three possible rotamers, two of which are observed in the crystal structure. An inter-spin distance measurement in frozen solution using double electron-electron resonance (DEER) gives a value essentially identical to that determined from the crystal structure of the protein containing two copies of V1, indicating that lattice forces do not dictate the rotamers observed. Collectively, the results suggest the possibility of predetermining a unique rotamer of V1 in helical structures. In general, the reduced rotameric space of V1 compared to R1 should simplify interpretation of inter-spin distance information in terms of protein structure, while the highly constrained internal motion is expected to extend the dynamic range for characterizing large amplitude nanosecond backbone fluctuations.

Conformational analysis of a nitroxide side chain in an α-helix with density functional theory.

In site directed spin labeling, a nitroxide side chain is introduced at a selected site in a protein; the most commonly used is a disulfide-linked side chain designated R1. The electron paramagnetic resonance (EPR) spectra of R1, and the interspin distance between pairs of R1 residues as determined by dipolar EPR spectroscopy, encode a wealth of information on the protein structure and dynamics. However, extracting this information requires structural and dynamical models of the R1 side chain, that is, the favored rotamers, the intraresidue interactions that stabilize them, and the internal modes of motion. X-ray crystal structures of R1 in proteins have revealed a set of preferred rotamers in the crystal lattice. To identify the intraresidue interactions that stabilize particular rotamers of R1 in the absence of interactions with nearby side chains in a helix, and to evaluate models for the internal motion of the side chain, quantum mechanical calculations were performed on a relevant fragment of R1 in a 10-residue α-helix. Relative rotamer energies were determined in the gas phase, and solvation energies were estimated from a continuum solvent model that includes both electrostatic and hydrophobic contributions. The results identified preferred rotamers that are in agreement with the X-ray crystallographic studies. The rotamers are apparently stabilized by intraresidue sulfur-backbone interactions, suggesting that the preferred rotamers may be the same at all solvent-exposed helix sites.

Mapping the cofilin binding site on yeast G-actin by chemical cross-linking.

Cofilin is a major cytoskeletal protein that binds to both monomeric actin (G-actin) and polymeric actin (F-actin) and is involved in microfilament dynamics. Although an atomic structure of the G-actin-cofilin complex does not exist, models of the complex have been built using molecular dynamics simulations, structural homology considerations, and synchrotron radiolytic footprinting data. The hydrophobic cleft between actin subdomains 1 and 3 and, alternatively, the cleft between actin subdomains 1 and 2 have been proposed as possible high-affinity cofilin binding sites. In this study, the proposed binding of cofilin to the subdomain 1/subdomain 3 region on G-actin has been probed using site-directed mutagenesis, fluorescence labeling, and chemical cross-linking, with yeast actin mutants containing single reactive cysteines in the actin hydrophobic cleft and with cofilin mutants carrying reactive cysteines in the regions predicted to bind to G-actin. Mass spectrometry analysis of the cross-linked complex revealed that cysteine 345 in subdomain 1 of mutant G-actin was cross-linked to native cysteine 62 on cofilin. A cofilin mutant that carried a cysteine substitution in the alpha 3-helix (residue 95) formed a cross-link with residue 144 in actin subdomain 3. Distance constraints imposed by these cross-links provide experimental evidence for cofilin binding between actin subdomains 1 and 3 and fit a corresponding docking-based structure of the complex. The cross-linking of the N-terminal region of recombinant yeast cofilin to actin residues 346 and 374 with dithio-bis-maleimidoethane (12.4 A) and via disulfide bond formation was also documented. This set of cross-linking data confirms the important role of the N-terminal segment of cofilin in interactions with G-actin.