A dietary commensal microbe enhances antitumor immunity by activating tumor macrophages to sequester iron.

Innate immune cells generate a multifaceted antitumor immune response, including the conservation of essential nutrients such as iron. These cells can be modulated by commensal bacteria; however, identifying and understanding how this occurs is a challenge. Here we show that the food commensal Lactiplantibacillus plantarum IMB19 augments antitumor immunity in syngeneic and xenograft mouse tumor models. Its capsular heteropolysaccharide is the major effector molecule, functioning as a ligand for TLR2. In a two-pronged manner, it skews tumor-associated macrophages to a classically active phenotype, leading to generation of a sustained CD8+ T cell response, and triggers macrophage 'nutritional immunity' to deploy the high-affinity iron transporter lipocalin-2 for capturing and sequestering iron in the tumor microenvironment. This process induces a cycle of tumor cell death, epitope expansion and subsequent tumor clearance. Together these data indicate that food commensals might be identified and developed into 'oncobiotics' for a multi-layered approach to cancer therapy.

Cancer Mutations Converge on a Collection of Protein Assemblies to Predict Resistance to Replication Stress.

Rapid proliferation is a hallmark of cancer associated with sensitivity to therapeutics that cause DNA replication stress (RS). Many tumors exhibit drug resistance, however, via molecular pathways that are incompletely understood. Here, we develop an ensemble of predictive models that elucidate how cancer mutations impact the response to common RS-inducing (RSi) agents. The models implement recent advances in deep learning to facilitate multidrug prediction and mechanistic interpretation. Initial studies in tumor cells identify 41 molecular assemblies that integrate alterations in hundreds of genes for accurate drug response prediction. These cover roles in transcription, repair, cell-cycle checkpoints, and growth signaling, of which 30 are shown by loss-of-function genetic screens to regulate drug sensitivity or replication restart. The model translates to cisplatin-treated cervical cancer patients, highlighting an RTK-JAK-STAT assembly governing resistance. This study defines a compendium of mechanisms by which mutations affect therapeutic responses, with implications for precision medicine. SIGNIFICANCE: Zhao and colleagues use recent advances in machine learning to study the effects of tumor mutations on the response to common therapeutics that cause RS. The resulting predictive models integrate numerous genetic alterations distributed across a constellation of molecular assemblies, facilitating a quantitative and interpretable assessment of drug response. This article is featured in Selected Articles from This Issue, p. 384.

Integrated germline and somatic features reveal divergent immune pathways driving ICB response.

Immune Checkpoint Blockade (ICB) has revolutionized cancer treatment, however mechanisms determining patient response remain poorly understood. Here we used machine learning to predict ICB response from germline and somatic biomarkers and interpreted the learned model to uncover putative mechanisms driving superior outcomes. Patients with higher T follicular helper infiltrates were robust to defects in the class-I Major Histocompatibility Complex (MHC-I). Further investigation uncovered different ICB responses in MHC-I versus MHC-II neoantigen reliant tumors across patients. Despite similar response rates, MHC-II reliant responses were associated with significantly longer durable clinical benefit (Discovery: Median OS=63.6 vs. 34.5 months P=0.0074; Validation: Median OS=37.5 vs. 33.1 months, P=0.040). Characteristics of the tumor immune microenvironment reflected MHC neoantigen reliance, and analysis of immune checkpoints revealed LAG3 as a potential target in MHC-II but not MHC-I reliant responses. This study highlights the value of interpretable machine learning models in elucidating the biological basis of therapy responses.

Cell-cell communication network-based interpretable machine learning predicts cancer patient response to immune checkpoint inhibitors.

Immune checkpoint inhibitors (ICIs) have revolutionized cancer treatment. However, only some patients respond to ICIs, and current biomarkers for ICI efficacy have limited performance. Here, we devised an interpretable machine learning (ML) model trained using patient-specific cell-cell communication networks (CCNs) decoded from the patient's bulk tumor transcriptome. The model could (i) predict ICI efficacy for patients across four cancer types (median AUROC: 0.79) and (ii) identify key communication pathways with crucial players responsible for patient response or resistance to ICIs by analyzing more than 700 ICI-treated patient samples from 11 cohorts. The model prioritized chemotaxis communication of immune-related cells and growth factor communication of structural cells as the key biological processes underlying response and resistance to ICIs, respectively. We confirmed the key communication pathways and players at the single-cell level in patients with melanoma. Our network-based ML approach can be used to expand ICIs' clinical benefits in cancer patients.

Information about immune cell proportions and tumor stage improves the prediction of recurrence in patients with colorectal cancer.

Predicting cancer recurrence is essential to improving the clinical outcomes of patients with colorectal cancer (CRC). Although tumor stage information has been used as a guideline to predict CRC recurrence, patients with the same stage show different clinical outcomes. Therefore, there is a need to develop a method to identify additional features for CRC recurrence prediction. Here, we developed a network-integrated multiomics (NIMO) approach to select appropriate transcriptome signatures for better CRC recurrence prediction by comparing the methylation signatures of immune cells. We validated the performance of the CRC recurrence prediction based on two independent retrospective cohorts of 114 and 110 patients. Moreover, to confirm that the prediction was improved, we used both NIMO-based immune cell proportions and TNM (tumor, node, metastasis) stage data. This work demonstrates the importance of (1) using both immune cell composition and TNM stage data and (2) identifying robust immune cell marker genes to improve CRC recurrence prediction.

Development of bioinformatics and multi-omics analyses in organoids.

Pre-clinical models are critical in gaining mechanistic and biological insights into disease progression. Recently, patient-derived organoid models have been developed to facilitate our understanding of disease development and to improve the discovery of therapeutic options by faithfully recapitulating in vivo tissues or organs. As technological developments of organoid models are rapidly growing, computational methods are gaining attention in organoid researchers to improve the ability to systematically analyze experimental results. In this review, we summarize the recent advances in organoid models to recapitulate human diseases and computational advancements to analyze experimental results from organoids. [BMB Reports 2023; 56(1): 43-48].

Deconvolution of bulk tumors into distinct immune cell states predicts colorectal cancer recurrence.

Predicting colorectal cancer recurrence after tumor resection is crucial because it promotes the administration of proper subsequent treatment or management to improve the clinical outcomes of patients. Several clinical or molecular factors, including tumor stage, metastasis, and microsatellite instability status, have been used to assess the risk of recurrence, although their predictive ability is limited. Here, we predicted colorectal cancer recurrence based on cellular deconvolution of bulk tumors into two distinct immune cell states: cancer-associated (tumor-infiltrating immune cell-like) and noncancer-associated (peripheral blood mononuclear cell-like). Prediction model performed significantly better when immune cells were deconvoluted into two states rather than a single state, suggesting that the difference in cancer recurrence was better explained by distinct states of immune cells. It indicates the importance of distinguishing immune cell states using cellular deconvolution to improve the prediction of colorectal cancer recurrence.

Network-based machine learning approach to predict immunotherapy response in cancer patients.

Immune checkpoint inhibitors (ICIs) have substantially improved the survival of cancer patients over the past several years. However, only a minority of patients respond to ICI treatment (~30% in solid tumors), and current ICI-response-associated biomarkers often fail to predict the ICI treatment response. Here, we present a machine learning (ML) framework that leverages network-based analyses to identify ICI treatment biomarkers (NetBio) that can make robust predictions. We curate more than 700 ICI-treated patient samples with clinical outcomes and transcriptomic data, and observe that NetBio-based predictions accurately predict ICI treatment responses in three different cancer types-melanoma, gastric cancer, and bladder cancer. Moreover, the NetBio-based prediction is superior to predictions based on other conventional ICI treatment biomarkers, such as ICI targets or tumor microenvironment-associated markers. This work presents a network-based method to effectively select immunotherapy-response-associated biomarkers that can make robust ML-based predictions for precision oncology.

Evolutionary rewiring of regulatory networks contributes to phenotypic differences between human and mouse orthologous genes.

Mouse models have been engineered to reveal the biological mechanisms of human diseases based on an assumption. The assumption is that orthologous genes underlie conserved phenotypes across species. However, genetically modified mouse orthologs of human genes do not often recapitulate human disease phenotypes which might be due to the molecular evolution of phenotypic differences across species from the time of the last common ancestor. Here, we systematically investigated the evolutionary divergence of regulatory relationships between transcription factors (TFs) and target genes in functional modules, and found that the rewiring of gene regulatory networks (GRNs) contributes to the phenotypic discrepancies that occur between humans and mice. We confirmed that the rewired regulatory networks of orthologous genes contain a higher proportion of species-specific regulatory elements. Additionally, we verified that the divergence of target gene expression levels, which was triggered by network rewiring, could lead to phenotypic differences. Taken together, a careful consideration of evolutionary divergence in regulatory networks could be a novel strategy to understand the failure or success of mouse models to mimic human diseases. To help interpret mouse phenotypes in human disease studies, we provide quantitative comparisons of gene expression profiles on our website (http://sbi.postech.ac.kr/w/RN).

Creation of bladder assembloids mimicking tissue regeneration and cancer.

Current organoid models are limited by their inability to mimic mature organ architecture and associated tissue microenvironments1,2. Here we create multilayer bladder 'assembloids' by reconstituting tissue stem cells with stromal components to represent an organized architecture with an epithelium surrounding stroma and an outer muscle layer. These assembloids exhibit characteristics of mature adult bladders in cell composition and gene expression at the single-cell transcriptome level, and recapitulate in vivo tissue dynamics of regenerative responses to injury. We also develop malignant counterpart tumour assembloids to recapitulate the in vivo pathophysiological features of urothelial carcinoma. Using the genetically manipulated tumour-assembloid platform, we identify tumoural FOXA1, induced by stromal bone morphogenetic protein (BMP), as a master pioneer factor that drives enhancer reprogramming for the determination of tumour phenotype, suggesting the importance of the FOXA1-BMP-hedgehog signalling feedback axis between tumour and stroma in the control of tumour plasticity.

Network-based machine learning in colorectal and bladder organoid models predicts anti-cancer drug efficacy in patients.

Cancer patient classification using predictive biomarkers for anti-cancer drug responses is essential for improving therapeutic outcomes. However, current machine-learning-based predictions of drug response often fail to identify robust translational biomarkers from preclinical models. Here, we present a machine-learning framework to identify robust drug biomarkers by taking advantage of network-based analyses using pharmacogenomic data derived from three-dimensional organoid culture models. The biomarkers identified by our approach accurately predict the drug responses of 114 colorectal cancer patients treated with 5-fluorouracil and 77 bladder cancer patients treated with cisplatin. We further confirm our biomarkers using external transcriptomic datasets of drug-sensitive and -resistant isogenic cancer cell lines. Finally, concordance analysis between the transcriptomic biomarkers and independent somatic mutation-based biomarkers further validate our method. This work presents a method to predict cancer patient drug responses using pharmacogenomic data derived from organoid models by combining the application of gene modules and network-based approaches.

Domain-mediated interactions for protein subfamily identification.

Within a protein family, proteins with the same domain often exhibit different cellular functions, despite the shared evolutionary history and molecular function of the domain. We hypothesized that domain-mediated interactions (DMIs) may categorize a protein family into subfamilies because the diversified functions of a single domain often depend on interacting partners of domains. Here we systematically identified DMI subfamilies, in which proteins share domains with DMI partners, as well as with various functional and physical interaction networks in individual species. In humans, DMI subfamily members are associated with similar diseases, including cancers, and are frequently co-associated with the same diseases. DMI information relates to the functional and evolutionary subdivisions of human kinases. In yeast, DMI subfamilies contain proteins with similar phenotypic outcomes from specific chemical treatments. Therefore, the systematic investigation here provides insights into the diverse functions of subfamilies derived from a protein family with a link-centric approach and suggests a useful resource for annotating the functions and phenotypic outcomes of proteins.

Evolutionary coupling analysis identifies the impact of disease-associated variants at less-conserved sites.

Genome-wide association studies have discovered a large number of genetic variants in human patients with the disease. Thus, predicting the impact of these variants is important for sorting disease-associated variants (DVs) from neutral variants. Current methods to predict the mutational impacts depend on evolutionary conservation at the mutation site, which is determined using homologous sequences and based on the assumption that variants at well-conserved sites have high impacts. However, many DVs at less-conserved but functionally important sites cannot be predicted by the current methods. Here, we present a method to find DVs at less-conserved sites by predicting the mutational impacts using evolutionary coupling analysis. Functionally important and evolutionarily coupled sites often have compensatory variants on cooperative sites to avoid loss of function. We found that our method identified known intolerant variants in a diverse group of proteins. Furthermore, at less-conserved sites, we identified DVs that were not identified using conservation-based methods. These newly identified DVs were frequently found at protein interaction interfaces, where species-specific mutations often alter interaction specificity. This work presents a means to identify less-conserved DVs and provides insight into the relationship between evolutionarily coupled sites and human DVs.

Epigenetic regulation of mammalian Hedgehog signaling to the stroma determines the molecular subtype of bladder cancer.

In bladder, loss of mammalian Sonic Hedgehog (Shh) accompanies progression to invasive urothelial carcinoma, but the molecular mechanisms underlying this cancer-initiating event are poorly defined. Here, we show that loss of Shh results from hypermethylation of the CpG shore of the Shh gene, and that inhibition of DNA methylation increases Shh expression to halt the initiation of murine urothelial carcinoma at the early stage of progression. In full-fledged tumors, pharmacologic augmentation of Hedgehog (Hh) pathway activity impedes tumor growth, and this cancer-restraining effect of Hh signaling is mediated by the stromal response to Shh signals, which stimulates subtype conversion of basal to luminal-like urothelial carcinoma. Our findings thus provide a basis to develop subtype-specific strategies for the management of human bladder cancer.

Exomic and transcriptomic alterations of hereditary gingival fibromatosis.

OBJECTIVE: Hereditary gingival fibromatosis (HGF) is a rare oral disease characterized by either localized or generalized gradual, benign, non-hemorrhagic enlargement of gingivae. Although several genetic causes of HGF are known, the genetic etiology of HGF as a non-syndromic and idiopathic entity remains uncertain. SUBJECTS AND METHODS: We performed exome and RNA-seq of idiopathic HGF patients and controls, and then devised a computational framework that specifies exomic/transcriptomic alterations interconnected by a regulatory network to unravel genetic etiology of HGF. Moreover, given the lack of animal model or large-scale cohort data of HGF, we developed a strategy to cross-check their clinical relevance through in silico gene-phenotype mapping with biomedical literature mining and semantic analysis of disease phenotype similarities. RESULTS: Exomic variants and differentially expressed genes of HGF were connected by members of TGF-ß/SMAD signaling pathway and craniofacial development processes, accounting for the molecular mechanism of fibroblast overgrowth mimicking HGF. Our cross-check supports that genes derived from the regulatory network analysis have pathogenic roles in fibromatosis-related diseases. CONCLUSIONS: The computational approach of connecting exomic and transcriptomic alterations through regulatory networks is applicable in the clinical interpretation of genetic variants in HGF patients.