RACER-m leverages structural features for sparse T cell specificity prediction.

Reliable prediction of T cell specificity against antigenic signatures is a formidable task, complicated by the immense diversity of T cell receptor and antigen sequence space and the resulting limited availability of training sets for inferential models. Recent modeling efforts have demonstrated the advantage of incorporating structural information to overcome the need for extensive training sequence data, yet disentangling the heterogeneous TCR-antigen interface to accurately predict MHC-allele-restricted TCR-peptide interactions has remained challenging. Here, we present RACER-m, a coarse-grained structural model leveraging key biophysical information from the diversity of publicly available TCR-antigen crystal structures. Explicit inclusion of structural content substantially reduces the required number of training examples and maintains reliable predictions of TCR-recognition specificity and sensitivity across diverse biological contexts. Our model capably identifies biophysically meaningful point-mutant peptides that affect binding affinity, distinguishing its ability in predicting TCR specificity of point-mutants from alternative sequence-based methods. Its application is broadly applicable to studies involving both closely related and structurally diverse TCR-peptide pairs.

Development of a Novel CLDN18.2-directed Monoclonal Antibody and Antibody-Drug Conjugate for Treatment of CLDN18.2-Positive Cancers.

Gastric and pancreatic cancers are malignancies of high unmet clinical need. Expression of CLDN18.2 in these cancers, coupled with it's absence from most normal tissues, provides a potential therapeutic window against this target. We present preclinical development and characterization of a novel therapeutic mAb and antibody-drug conjugate (ADC) targeting CLDN18.2. A humanized CLDN18.2 specific mAb, CLDN18.2-307-mAb, was generated through immunization in mice followed by full humanization of the mouse mAb sequences. Antibody clones were screened by flow cytometry for selective binding to membrane bound CLDN18.2. A CLDN18.2-directed ADC (CLDN18.2-307-ADC) was also generated by conjugating MMAE to CLDN18.2 mAb using a cleavable linker. Tissue expression of CLDN18.2 was determined by IHC assay using a CLDN18.2-specific mAb. CLDN18.2-307-mAb binds with high affinity to CLDN18.2-positive (CLDN18.2+) cells and induces antibody-dependent cell-mediated cytotoxicity (ADCC). Treatment with this CLDN18.2-mAb blocked the growth of CLDN18.2+ gastric and pancreas cancer cell line xenograft (CDX) models. Upon binding to the extracellular domain of this target, the CLDN18.2-ADC/CLDN18.2 protein was internalized and subsequently localized to the lysosomal compartment inducing complete and sustained tumor regressions in CLDN18.2+ CDXs and patient-derived pancreatic cancer xenografts (PDX). A screen of human cancer tissues, by IHC, found 58% of gastric, 60% of gastroesophageal junction, and 20% of pancreatic adenocarcinomas to be positive for membrane expression of CLDN18.2. These data support clinical development of the CLDN18.2-307-mAb and CLDN18.2-307-ADC for treatment of CLDN18.2+ cancers. Both are now being investigated in phase I clinical studies.

MEK Inhibition Sensitizes Pancreatic Cancer to STING Agonism by Tumor Cell-intrinsic Amplification of Type I IFN Signaling.

PURPOSE: Stimulator of interferon genes (STING) agonists are currently in development for treatment of solid tumors, including pancreatic ductal adenocarcinoma (PDAC). Response rates to STING agonists alone have been promising yet modest, and combination therapies will likely be required to elicit their full potency. We sought to identify combination therapies and mechanisms that augment the tumor cell-intrinsic effect of therapeutically relevant STING agonists apart from their known effects on tumor immunity. EXPERIMENTAL DESIGN: We screened 430 kinase inhibitors to identify synergistic effectors of tumor cell death with diABZI, an intravenously administered and systemically available STING agonist. We deciphered the mechanisms of synergy with STING agonism that cause tumor cell death in vitro and tumor regression in vivo. RESULTS: We found that MEK inhibitors caused the greatest synergy with diABZI and that this effect was most pronounced in cells with high STING expression. MEK inhibition enhanced the ability of STING agonism to induce type I IFN-dependent cell death in vitro and tumor regression in vivo. We parsed NFκB-dependent and NFκB-independent mechanisms that mediate STING-driven type I IFN production and show that MEK signaling inhibits this effect by suppressing NFκB activation. CONCLUSIONS: Our results highlight the cytotoxic effects of STING agonism on PDAC cells that are independent of tumor immunity and that these therapeutic benefits of STING agonism can be synergistically enhanced by MEK inhibition.

Preclinical Efficacy of the Antibody-Drug Conjugate CLDN6-23-ADC for the Treatment of CLDN6-Positive Solid Tumors.

PURPOSE: Claudin-6 (CLDN6) is expressed at elevated levels in multiple human cancers including ovarian and endometrial malignancies, with little or no detectable expression in normal adult tissue. This expression profile makes CLDN6 an ideal target for development of a potential therapeutic antibody-drug conjugate (ADC). This study describes the generation and preclinical characterization of CLDN6-23-ADC, an ADC consisting of a humanized anti-CLDN6 monoclonal antibody coupled to monomethyl auristatin E (MMAE) via a cleavable linker. EXPERIMENTAL DESIGN: A fully humanized anti-CLDN6 antibody was conjugated to MMAE resulting in the potential therapeutic ADC, CLDN6-23-ADC. The antitumor efficacy of CLDN6-23-ADC was assessed for antitumor efficacy in CLDN6-positive (CLDN6+) and -negative (CLDN6-) xenografts and patient-derived xenograft (PDX) models of human cancers. RESULTS: CLDN6-23-ADC selectively binds to CLDN6, versus other CLDN family members, inhibits the proliferation of CLDN6+ cancer cells in vitro, and is rapidly internalized in CLDN6+ cells. Robust tumor regressions were observed in multiple CLDN6+ xenograft models and tumor inhibition led to markedly enhanced survival of CLDN6+ PDX tumors following treatment with CLDN6-23-ADC. IHC assessment of cancer tissue microarrays demonstrate elevated levels of CLDN6 in 29% of ovarian epithelial carcinomas. Approximately 45% of high-grade serous ovarian carcinomas and 11% of endometrial carcinomas are positive for the target. CONCLUSIONS: We report the development of a novel ADC, CLDN6-23-ADC, that selectively targets CLDN6, a potential onco-fetal-antigen which is highly expressed in ovarian and endometrial cancers. CLDN6-23-ADC exhibits robust tumor regressions in mouse models of human ovarian and endometrial cancers and is currently undergoing phase I study.

Rapid Assessment of T-Cell Receptor Specificity of the Immune Repertoire.

Accurate assessment of TCR-antigen specificity at the whole immune repertoire level lies at the heart of improved cancer immunotherapy, but predictive models capable of high-throughput assessment of TCR-peptide pairs are lacking. Recent advances in deep sequencing and crystallography have enriched the data available for studying TCR-p-MHC systems. Here, we introduce a pairwise energy model, RACER, for rapid assessment of TCR-peptide affinity at the immune repertoire level. RACER applies supervised machine learning to efficiently and accurately resolve strong TCR-peptide binding pairs from weak ones. The trained parameters further enable a physical interpretation of interacting patterns encoded in each specific TCR-p-MHC system. When applied to simulate thymic selection of an MHC-restricted T-cell repertoire, RACER accurately estimates recognition rates for tumor-associated neoantigens and foreign peptides, thus demonstrating its utility in helping address the large computational challenge of reliably identifying the properties of tumor antigen-specific T-cells at the level of an individual patient's immune repertoire.

Targeting activated PI3K/mTOR signaling overcomes acquired resistance to CDK4/6-based therapies in preclinical models of hormone receptor-positive breast cancer.

BACKGROUND: Combined targeting of CDK4/6 and ER is now the standard of care for patients with advanced ER+/HER2- breast cancer. However, acquired resistance to these therapies frequently leads to disease progression. As such, it is critical to identify the mechanisms by which resistance to CDK4/6-based therapies is acquired and also identify therapeutic strategies to overcome resistance. METHODS: In this study, we developed and characterized multiple in vitro and in vivo models of acquired resistance to CDK4/6-based therapies. Resistant models were screened by reverse phase protein array (RPPA) for cell signaling changes that are activated in resistance. RESULTS: We show that either a direct loss of Rb or loss of dependence on Rb signaling confers cross-resistance to inhibitors of CDK4/6, while PI3K/mTOR signaling remains activated. Treatment with the p110α-selective PI3K inhibitor, alpelisib (BYL719), completely blocked the progression of acquired CDK4/6 inhibitor-resistant xenografts in the absence of continued CDK4/6 inhibitor treatment in models of both PIK3CA mutant and wild-type ER+/HER2- breast cancer. Triple combination therapy against PI3K:CDK4/6:ER prevented and/or delayed the onset of resistance in treatment-naive ER+/HER2- breast cancer models. CONCLUSIONS: These data support the clinical investigation of p110α-selective inhibitors of PI3K, such as alpelisib, in patients with ER+/HER2- breast cancer who have progressed on CDK4/6:ER-based therapies. Our data also support the investigation of PI3K:CDK4/6:ER triple combination therapy to prevent the onset of resistance to the combination of endocrine therapy plus CDK4/6 inhibition.

Mice Expressing Myc in Neural Precursors Develop Choroid Plexus and Ciliary Body Tumors.

Choroid plexus tumors and ciliary body medulloepithelioma are predominantly pediatric neoplasms. Progress in understanding the pathogenesis of these tumors has been hindered by their rarity and lack of models that faithfully recapitulate the disease. Here, we find that endogenous Myc proto-oncogene protein is down-regulated in the forebrain neuroepithelium, whose neural plate border domains give rise to the anterior choroid plexus and ciliary body. To uncover the consequences of persistent Myc expression, MYC expression was forced in multipotent neural precursors (nestin-Cre:Myc), which produced fully penetrant models of choroid plexus carcinoma and ciliary body medulloepithelioma. Nestin-mediated MYC expression in the epithelial cells of choroid plexus leads to the regionalized formation of choroid plexus carcinoma in the posterior domain of the lateral ventricle choroid plexus and the fourth ventricle choroid plexus that is accompanied by loss of multiple cilia, up-regulation of protein biosynthetic machinery, and hydrocephalus. Parallel MYC expression in the ciliary body leads also to up-regulation of protein biosynthetic machinery. Additionally, Myc expression in human choroid plexus tumors increases with aggressiveness of disease. Collectively, our findings expose a select vulnerability of the neuroepithelial lineage to postnatal tumorigenesis and provide a new mouse model for investigating the pathogenesis of these rare pediatric neoplasms.

Preclinical Activity of Abemaciclib Alone or in Combination with Antimitotic and Targeted Therapies in Breast Cancer.

The cyclinD:CDK4/6:Rb axis is dysregulated in a variety of human cancers. Targeting this pathway has proven to be a successful therapeutic approach in ER+ breast cancer. In this study, in vitro and in vivo preclinical breast cancer models were used to investigate the expanded use of the CDK4/6 inhibitor, abemaciclib. Using a panel of 44 breast cancer cell lines, differential sensitivity to abemaciclib was observed and was seen predominately in the luminal ER+/HER2- and ER+/HER2+ subtypes. However, a subset of triple-negative breast cancer (TNBC) cell lines with intact Rb signaling were also found to be responsive. Equivalent levels of tumor growth inhibition were observed in ER+/HER2-, ER+/HER2+ as well as biomarker selected TNBC xenografts in response to abemaciclib. In addition, abemaciclib combined with hormonal blockade and/or HER2-targeted therapy induced significantly improved antitumor activity. CDK4/6 inhibition with abemaciclib combined with antimitotic agents, both in vitro and in vivo, did not antagonize the effect of either agent. Finally, we identified a set of Rb/E2F-regulated genes that consistently track with growth inhibitory response and constitute potential pharmacodynamic biomarkers of response to abemaciclib. Taken together, these data represent a comprehensive analysis of the preclinical activity of abemaciclib, used alone or in combination, in human breast cancer models. The subtypes most likely to respond to abemaciclib-based therapies can be identified by measurement of a specific set of biomarkers associated with increased dependency on cyclinD:CDK4/6:Rb signaling. These data support the clinical development of abemaciclib as monotherapy or as a combination partner in selected ER+/HER2-, HER2+/ER+, and TNBCs. Mol Cancer Ther; 17(5); 897-907. ©2018 AACR.

Transcriptomic Characterization of SF3B1 Mutation Reveals Its Pleiotropic Effects in Chronic Lymphocytic Leukemia.

Mutations in SF3B1, which encodes a spliceosome component, are associated with poor outcome in chronic lymphocytic leukemia (CLL), but how these contribute to CLL progression remains poorly understood. We undertook a transcriptomic characterization of primary human CLL cells to identify transcripts and pathways affected by SF3B1 mutation. Splicing alterations, identified in the analysis of bulk cells, were confirmed in single SF3B1-mutated CLL cells and also found in cell lines ectopically expressing mutant SF3B1. SF3B1 mutation was found to dysregulate multiple cellular functions including DNA damage response, telomere maintenance, and Notch signaling (mediated through KLF8 upregulation, increased TERC and TERT expression, or altered splicing of DVL2 transcript, respectively). SF3B1 mutation leads to diverse changes in CLL-related pathways.

Spatially heterogeneous choroid plexus transcriptomes encode positional identity and contribute to regional CSF production.

A sheet of choroid plexus epithelial cells extends into each cerebral ventricle and secretes signaling factors into the CSF. To evaluate whether differences in the CSF proteome across ventricles arise, in part, from regional differences in choroid plexus gene expression, we defined the transcriptome of lateral ventricle (telencephalic) versus fourth ventricle (hindbrain) choroid plexus. We find that positional identities of mouse, macaque, and human choroid plexi derive from gene expression domains that parallel their axial tissues of origin. We then show that molecular heterogeneity between telencephalic and hindbrain choroid plexi contributes to region-specific, age-dependent protein secretion in vitro. Transcriptome analysis of FACS-purified choroid plexus epithelial cells also predicts their cell-type-specific secretome. Spatial domains with distinct protein expression profiles were observed within each choroid plexus. We propose that regional differences between choroid plexi contribute to dynamic signaling gradients across the mammalian cerebroventricular system.

Efficient feeder-free episomal reprogramming with small molecules.

Genetic reprogramming of human somatic cells to induced pluripotent stem cells (iPSCs) could offer replenishable cell sources for transplantation therapies. To fulfill their promises, human iPSCs will ideally be free of exogenous DNA (footprint-free), and be derived and cultured in chemically defined media free of feeder cells. Currently, methods are available to enable efficient derivation of footprint-free human iPSCs. However, each of these methods has its limitations. We have previously derived footprint-free human iPSCs by employing episomal vectors for transgene delivery, but the process was inefficient and required feeder cells. Here, we have greatly improved the episomal reprogramming efficiency using a cocktail containing MEK inhibitor PD0325901, GSK3ß inhibitor CHIR99021, TGF-ß/Activin/Nodal receptor inhibitor A-83-01, ROCK inhibitor HA-100 and human leukemia inhibitory factor. Moreover, we have successfully established a feeder-free reprogramming condition using chemically defined medium with bFGF and N2B27 supplements and chemically defined human ESC medium mTeSR1 for the derivation of footprint-free human iPSCs. These improvements enabled the routine derivation of footprint-free human iPSCs from skin fibroblasts, adipose tissue-derived cells and cord blood cells. This technology will likely be valuable for the production of clinical-grade human iPSCs.