Chemical-induced respiratory irritation is a critical human health concern because of its potential to cause both acute and chronic injuries. Therefore, assessing the respiratory irritation potential of chemicals is crucial for protecting human health. Due to limited human data, human risk assessment often relies on extrapolating animal data, which can be imprecise and resource intensive. New approach methodologies are being developed to reduce the reliance on animal testing and streamline chemical risk assessment methods. In the present study, we developed a computational workflow to integrate explainable artificial intelligence (XAI) approaches into the quantitative structure-activity relationship (QSAR) modeling pipeline to predict chemical-induced respiratory irritation. We developed and assessed multiple QSAR models by combining different machine learning algorithms and diverse molecular representations. The models achieved an average cross-validated area under the receiver operating characteristic curve of 0.88, accuracy of 0.80, and Matthews correlation coefficient of 0.61, with external test set evaluation indicating good generalizability. We applied different methods, including Shapley additive explanations (SHAP), to explain the predictions of our QSAR models. By providing both global and local explanations of model predictions, the SHAP analyses highlighted key molecular descriptors and fingerprints driving respiratory irritation predictions, revealing physicochemical properties that may provide insights into irritation potential. The explainable models developed in this work have the potential to be an alternative screening tool for traditional animal models to obtain faster and more cost-effective human risk assessment of respiratory irritants.

Abstract Background: Chromosomal instability (CIN) frequently exerts innate immune response, being expected to sensitize immunotherapy. cGAS-STING pathway is a hub to regulate CIN-mediated innate immune activation. We previously demonstrated that small-molecule inhibitors targeting Centromere-associated Protein-E (CENP-E) profoundly accumulated CIN in cancer cells, leading to activation of cGAS-STING and its related innate immune pathways. In this study, aiming to deeply understand molecular mechanisms of CENP-E inhibitor (CENP-Ei)-induced CIN and innate immune activation, we conducted large-scale chemical screening with ~1,400 kinase inhibitors in CENP-Ei-treated reporter cells for NF-κB and IRF. We also experimentally validated the effects of hit compounds on CENP-Ei-induced innate immune activation. Material and Methods: GSK923295 and Cmpd-A were used as CENP-E inhibitors. HeLa, A549, A549 dual reporter cells were treated with the CENP-Ei at the indicated concentrations, subjected to immunofluorescent, transcriptome, FACS, reporter activity, or gene expression analyses. High-throughput screening (HTS) with ~1,400 kinase library was conducted in A549 dual reporter cells in combinational treatment with GSK923295. Combination effects of the representative hit compounds were confirmed in matrix-combination studies with the CENP-Ei. Results: Treatment with the CENP-Ei, Cmpd-A and GSK923295, induced CIN (e.g., multinucleation) in both A549 and A549 dual reporter cells, which elevated NFkB and IRF reporter activities as well as gene expressions of inflammatory cytokines and chemokines in innate immune pathways. Transcriptome analyses also confirmed that gene ontologies (GO) for innate immune response, such as cytosolic DNA sensing pathway and cytokine-cytokine receptor interaction pathway, were significantly enriched in CENP-Ei-treated CIN cells. Next, we conducted HTS of IRF-reporter activity with ~1,400 kinase library in combination with GSK923295 in A549 dual reporter cells. The HTS revealed that a series of inhibitors targeting PI3K-AKT-mTOR signaling pathways intensively suppressed CENP-Ei-induced IRF activation, which occupied ~40% of hit compounds. The matrix-combination studies confirmed that sapanisertib (TORC1/2 inhibitor) significantly attenuated CENP-Ei-induced IRF activity. Conclusions: CENP-Ei-induced CIN potently activates innate immune response pathways in cancer cells, in which PI3K-AKT-mTOR signaling pathway appears to be involved. We are initiating AI-based drug discovery to develop novel CENP-E inhibitors. Citation Format: Ryo Kamata, Hitoshi Saito, Yumi Hakozaki, Yukie Kashima, Gaku Yamamoto, Tomoko Yamamori Morita, Pinyi Lu, Akihiro Ohashi. PI3K-AKT-mTOR signaling pathways play important roles in chromosomal instability-induced innate immune response in cancer cells [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 411.

Abstract Purpose: Centromere-associated protein E (CENP-E), a mitotic motor protein, is crucial for cell division. Inhibition of CENP-E can lead to chromosome misalignment and aneuploidy. Our recent studies revealed that the CENP-E inhibitor (CENP-Ei), Compound A, induces aneuploid-mediated cGAS-STING pathway activation in cancer cell lines, which is expected to trigger innate immune response and induce immunological conversion of the tumor microenvironment from cold to hot. However, known CENP-Eis exhibited limited efficacy, and none have gone beyond phase I trials since 2012. The aim of this study is to develop a workflow integrating structure and artificial intelligence-based modeling approaches to accelerate the discovery of novel CENP-Eis with improved efficacy and overcome related challenges. Method: The inhibitor-bound CENP-E structure was modeled using chimeric homology modeling followed by induced fit docking with known CENP-Eis. Structure-based virtual screening (VS) was performed on large compound libraries in Enamine REAL, Synthetically Accessible Virtual Inventory, and MolPort databases. Consensus scoring methods were applied by averaging ranks of individual molecules obtained from VS against different CENP-E conformations. An active learning-enhanced VS pipeline was developed using ATOM Modeling PipeLine (AMPL) to find the best-ranked ligands more efficiently. CENP-E ligand candidates were further filtered based on their safety properties predicted by machine learning models trained with curated databases of adverse drug reactions using AMPL. The success of this study was measured by its ability to reveal novel chemotypes that could modulate CENP-E in kinesin ATPase assays. Result: Conformations of CENP-E’s ligand binding site were predicted for VS using the homology model. Structure-based VS was performed to select high-ranked compounds based on consensus scoring. The active learning-enhanced VS pipeline achieved 75% accuracy in ranking the top 20% of hit compounds. The graph convolutional network model with a ROCAUC, 0.76, was created using AMPL to filter ligand candidates based on their predicted safety properties. Seventy-three potential CENP-E ligands computationally prioritized from the Molport database were experimentally tested using kinesin ATPase assays from which 4 potential inhibitors and 8 potential activators were identified indicating overlapping binding sites of inhibitors and activators. Conclusion: Our integrated workflow is effective in discovering novel CENP-E modulators. The discovery of CENP-E activators provides an opportunity to identify novel therapeutics in hepatocellular carcinoma malignancies known to experience decreased CENP-E activity leading to aneuploidy. The predictions generated by computational models combined with experimental validations could further accelerate anticancer drug discovery. Citation Format: Pinyi Lu, Ryo Kamata, Michael R. Weil, Naomi Ohashi, Akihiro Ohashi, Eric A. Stahlberg. Computational modeling-based discovery of novel anticancer drug candidates targeting centromere-associated protein E [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4945.

Background: cGAS-STING, which monitors cytosolic DNA and triggers innate immune response, has recently attracted attention as a molecular target to sensitize cancer immune therapies. Since DNAs enclosed in micronucleus frequently leak into cytoplasm, micronucleus generated by centromere-associated protein-E (CENP-E) inhibition is expected to activate cGAS-STING pathways. We validated that small-molecule CENP-E inhibitors profoundly generate micronucleus through misaligned chromosomes. The aim of this study is to elucidate the possibility that micronucleus formation by CENP-E inhibitors activates the cGAS-STING pathway in cancer cells and induces immunological conversion of the tumor microenvironment from cold to hot. Material and Methods: GSK923295 and Cmpd-A were used as CENP-E inhibitors. HeLa and A549 cells were treated with the CENP-E inhibitors at the indicated concentrations. RNA sequencing data were subjected to gene set enrichment analysis (GSEA) to determine the signaling pathways modulated by Cmpd-A treatment compared to DMSO (control). Reporter assays for IRF and NF-kB were performed in A549 dual-reporter cells. High-throughput screening on IRF3 reporter activity was performed in A549 dual reporter cells in combination with -1,400 kinase inhibitor library and GSK923295. Results: The CENP-E inhibitors, GSK-923295 and Cmpd-A, induced chromosome misalignment, resulting in micronucleus formation after mitotic slippage. Phosphorylations of TBK1 and IRF3, cGAS-STING pathway markers, were remarkably upregulated in Cmpd-A-treated cells concordantly with micronucleus formation. The reporter assays also demonstrated that both IRF and NF-kB pathways were significantly activated in Cmpd-A-treated cells in a time-dependent manner. Transcriptome analysis of GSEA demonstrated that the gene ontologies (GO) for innate immune pathways, such as cytosolic DNA sensing pathway and cytokine-cytokine receptor interaction pathway, were significantly enriched in Cmpd-A-treated cells. The kinase library screening revealed that a series of inhibitors targeting PI3K-AKT-mTOR signaling pathways intensively suppressed CENP-E inhibitor-induced IRF activation, which occupied ~40% of hit compounds. Conclusions: The CENP-E inhibitors generated micronucleus activating the cGAS-STING pathway in multiple cancer cell lines. PI3K-AKT-mTOR signaling pathway may play important roles in the CENP-E inhibitor-induced cGAS-STING pathway activation. We are also initiating AI-based drug discovery, aiming to generate novel CENP-E inhibitors by performing virtual screening and generative molecular design. Citation Format: Ryo Kamata, Hitoshi Saito, Yumi Hakozaki, Yukie Kashima, Tomoko Yamamori. Morita, Pinyi Lu, Akihiro Ohashi. CENP-E inhibitors activate cGAS-STING pathway by inducing chromosome misalignment and micronucleation after mitotic slippage [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 301.

Immune correlates of protection remain elusive for most vaccines. An identified immune correlate would accelerate the down-selection of vaccine formulations by reducing the need for human pathogen challenge studies that are currently required to determine vaccine efficacy. Immunization via mosquito-delivered, radiation-attenuated P. falciparum sporozoites (IMRAS) is a well-established model for efficacious malaria vaccines, inducing greater than 90% sterile immunity. The current immunoprofiling study utilized samples from a clinical trial in which vaccine dosing was adjusted to achieve only 50% protection, thus enabling a comparison between protective and non-protective immune signatures. In-depth immunoprofiling was conducted by assessing a wide range of antigen-specific serological and cellular parameters and applying our newly developed computational tools, including machine learning. The computational component of the study pinpointed previously un-identified cellular T cell subsets (namely, TNFα-secreting CD8+CXCR3−CCR6− T cells, IFNγ-secreting CD8+CCR6+ T cells and TNFα/FNγ-secreting CD4+CXCR3−CCR6− T cells) and B cell subsets (i.e., CD19+CD24hiCD38hiCD69+ transitional B cells) as important factors predictive of protection (92% accuracy). Our study emphasizes the need for in-depth immunoprofiling and subsequent data integration with computational tools to identify immune correlates of protection. The described process of computational data analysis is applicable to other disease and vaccine models.

Background: cGAS-STING, which monitors cytosolic DNA and triggers innate immune response, has recently attracted attention as a molecular target to sensitize cancer to immune therapies. Micronuclei generated by aberrant mitosis are known to profoundly activate the cGAS-STING pathway in cancer cells when they leak into the cytoplasm. We have validated that small-molecule inhibitors targeting centromere-associated protein-E (CENP-E) generate micronuclei through chromosome misalignment and mitotic slippage. This suggests that CENP-E inhibitors (CIs) have potential to activate the cGAS-STING pathway in cancer cells at concentrations well below those used for a directly cytotoxic chemotherapeutic agent. The aims of this study is (1) to elucidate the potential that micronucleus formation by CIs activates the cGAS-STING pathway in cancer cells that may be rendered more susceptible to immunotherapeutic treatment and (2) to develop a novel workflow integrating structure and AI-based modeling approaches to accelerate the discovery of novel CIs with desirable ADMET properties. Material and Methods: (1) Misaligned chromosome and (2) micronucleus formation caused by CI treatment in cancer cells were determined using a microscope. (3) Downstream factors indicative of cGAS-STING pathway activation were measured by immunoblotting. (4) Reporter assays for IRF and NF-κB were performed. (5) Novel CI candidates will be identified from ultra-large chemical databases using the novel AI-based prediction and optimization approaches followed by the iterative generative molecular design. Results: The CI treatment caused chromosome misdistribution in cancer cells. In addition, micronuclei were formed in the cells. Immunoblotting revealed that CI markedly increased phosphorylation of TBK1 and IRF3, consistent with micronucleus formation. Furthermore, IRF-Lucia and NF-kB-SEAP reporter activity assays revealed that both IRF and NF-kB were significantly activated in a time-dependent manner in CI-treated cells but not by the control. Conclusions: The CI generates micronuclei and activates the cGAS-STING pathway in cancer cells. In our previous studies, transcriptome analysis has also revealed increased gene clusters involved in (1) the cytoplasmic DNA-sensing pathway, (2) the cytokine-cytokine receptor interaction pathway, and (3) the Toll-like receptor signaling pathway. Together, these findings suggest that the novel CI being researched have potential to induce immunological activation in cancer cells, which may convert tumors from cold to hot, making them more susceptible to treatment in the tumor microenvironment, while minimizing the side effects. The process of discovering immunomodulatory CIs will be accelerated using the novel AI-based workflow. That also benefits by reducing time and costs and identifying potential risks early on. Conflict of interest: Corporate-sponsored Research: AO was an employee of Takeda Pharmaceutical Company, Ltd. AO reported paid consulting or advisory roles for Ono Pharmaceutical Company Ltd., Craif Inc., and GEXVal Inc. out of this study. AO receives research fundings from Astellas Pharma Inc., Astellas Pharma Global Development Inc., and Daiichi Sankyo Company, Ltd. out of this study. Other Substantive Relationships: NCI Contract No. 75N91019D00024

The impact of pre-existing immunity on the efficacy of artemisinin combination therapy is largely unknown. We performed in-depth profiling of serological responses in a therapeutic efficacy study [comparing artesunate-mefloquine (ASMQ) and artemether-lumefantrine (AL)] using a proteomic microarray. Responses to over 200 Plasmodium antigens were significantly associated with ASMQ treatment outcome but not AL. We used machine learning to develop predictive models of treatment outcome based on the immunoprofile data. The models predict treatment outcome for ASMQ with high (72-85%) accuracy, but could not predict treatment outcome for AL. This divergent treatment outcome suggests that humoral immunity may synergize with the longer mefloquine half-life to provide a prophylactic effect at 28-42 days post-treatment, which was further supported by simulated pharmacokinetic profiling. Our computational approach and modeling revealed the synergistic effect of pre-existing immunity in patients with drug combination that has an extended efficacy on providing long term treatment efficacy of ASMQ.

Human immunodeficiency virus type 1 (HIV-1) infection remains a major public health threat due to its incurable nature and the lack of a highly efficacious vaccine. The RV144 vaccine trial is the only clinical study to date that demonstrated significant but modest decrease in HIV infection risk. To improve HIV-1 vaccine immunogenicity and efficacy, we recently evaluated pox-protein vaccination using a next generation liposome-based adjuvant, Army Liposomal Formulation adsorbed to aluminum (ALFA), in rhesus monkeys and observed 90% efficacy against limiting dose mucosal SHIV challenge in male animals. Here, we analyzed binding antibody responses, as assessed by Fc array profiling using a broad range of HIV-1 envelope antigens and Fc features, to explore the mechanisms of ALFA-mediated protection by employing machine learning and Cox proportional hazards regression analyses. We found that Fcγ receptor 2a-related binding antibody responses were augmented by ALFA relative to aluminium hydroxide, and these responses were associated with reduced risk of infection in male animals. Our results highlight the application of systems serology to provide mechanistic insights to vaccine-elicited protection and support evidence that antibody effector responses protect against HIV-1 infection.

FLT3 is a frequently mutated gene that is highly associated with a poor prognosis in acute myeloid leukemia (AML). Despite initially responding to FLT3 inhibitors, most patients eventually relapse with drug resistance. The mechanism by which resistance arises and the initial response to drug treatment that promotes cell survival is unknown. Recent studies show that a transiently maintained subpopulation of drug-sensitive cells, so-called drug-tolerant "persisters" (DTPs), can survive cytotoxic drug exposure despite lacking resistance-conferring mutations. Using RNA sequencing and drug screening, we find that treatment of FLT3 internal tandem duplication AML cells with quizartinib, a selective FLT3 inhibitor, upregulates inflammatory genes in DTPs and thereby confers susceptibility to anti-inflammatory glucocorticoids (GCs). Mechanistically, the combination of FLT3 inhibitors and GCs enhances cell death of FLT3 mutant, but not wild-type, cells through GC-receptor-dependent upregulation of the proapoptotic protein BIM and proteasomal degradation of the antiapoptotic protein MCL-1. Moreover, the enhanced antileukemic activity by quizartinib and dexamethasone combination has been validated using primary AML patient samples and xenograft mouse models. Collectively, our study indicates that the combination of FLT3 inhibitors and GCs has the potential to eliminate DTPs and therefore prevent minimal residual disease, mutational drug resistance, and relapse in FLT3-mutant AML.

Malaria-071, a controlled human malaria infection trial, demonstrated that administration of three doses of RTS,S/AS01 malaria vaccine given at one-month intervals was inferior to a delayed fractional dose (DFD) schedule (62.5% vs 86.7% protection, respectively). To investigate the underlying immunologic mechanism, we analyzed the B and T peripheral follicular helper cell (pTfh) responses. Here, we show that protection in both study arms was associated with early induction of functional IL-21-secreting circumsporozoite (CSP)-specific pTfh cells, together with induction of CSP-specific memory B cell responses after the second dose that persisted after the third dose. Data integration of key immunologic measures identified a subset of non-protected individuals in the standard (STD) vaccine arm who lost prior protective B cell responses after receiving the third vaccine dose. We conclude that the DFD regimen favors persistence of functional B cells after the third dose.

The role of humoral immunity on the efficacy of artemisinin combination therapy (ACT) has not been investigated, yet naturally acquired immunity is key determinant of antimalarial therapeutic response. We conducted a therapeutic efficacy study in high transmission settings of western Kenya, which showed artesunate-mefloquine (ASMQ) and dihydroartemisinin-piperaquine (DP) were more efficacious than artemether-lumefantrine (AL). To investigate the underlying prophylactic mechanism, we compared a broad range of humoral immune responses in cohort I study participants treated with ASMQ or AL, and applied machine-learning (ML) models using immunoprofile data to analyze individual participants’ treatment outcome. We showed ML models could predict treatment outcome for ASMQ but no AL with high (72-92%) accuracy. Simulated PK profiling provided evidence demonstrating specific humoral immunity confers protection in the presence of sub-therapeutic residual mefloquine concentration. We concluded patient humoral immunity and partner drug interact to provide long prophylactic effect of ASMQ.

BACKGROUND:Peroxisome proliferator-activated receptor gamma (PPARγ) is a member of the nuclear receptor superfamily that functions as a ligand-inducible transcription factor. It regulates glucose and lipid metabolism, immunity, and cellular growth and differentiation. Thiazolidinediones (TZDs) are potent insulin sensitizers that function by activating PPARs, with a high specificity for PPARγ. Due to their ability to preserve pancreatic beta cell function and reduce insulin resistance, TZDs have become one of the most prescribed classes of medications for type 2 diabetes (T2D) since their approval by the US Food and Drug Administration (FDA) and initial use in 1997.OBJECTIVE:However, adverse effects, including weight gain, bone loss, fluid retention, congestive heart failure, and risk to bladder cancer, have weakened the benefits of TZDs in T2D therapies. Therefore, there is an urgent need to have a deeper understanding of regulatory mechanisms of PPARγ expression and activity so that novel classes of PPARγ-modulating therapeutics with fewer or weaker side effects can be developed.CONCLUSION:This article systematically reviews PPARγ's mechanisms of action and multilayer regulations. In addition, novel classes of therapeutics modulating PPARγ and new direction of research on genetic variants that affect PPARγ function and antidiabetic drug response are highlighted, which sheds light on PPARγ as a promising target for developing safer and precision medicine based therapeutic strategies.

Imputation is a key step in Electronic Health Records-mining as it can significantly affect the conclusions derived from the downstream analysis. There are three main categories that explain the missingness in clinical settings–incompleteness, inconsistency, and inaccuracy–and these can capture a variety of situations: the patient did not seek treatment, the health care provider did not enter the information, etc. We used EHR data from patients diagnosed with Inflammatory Bowel Disease from Geisinger Health System to design a novel imputation that focuses on a complex phenotype. Our approach is based on latent-based analysis integrated with clustering to group patients based on their comorbidities before imputation. IBD is a chronic illness of unclear etiology and without a complete cure. We have taken advantage of the complexity of IBD to pre-process the EHR data of 10,498 IBD patients and show that imputation can be improved using shared latent comorbidities. The R code and sample simulated input data will be available at a future time.