Spatial omics has advanced our understanding of tissue-level biology, yet tools to systematically link gene functional perturbations to spatial phenotypes and signaling pathways remain limited. To address this, we developed spatial CRISPR screen sequencing (SPAC-seq), a high-throughput spatial CRISPR screen platform, and TARDIS (target prioritization toolkit for perturbation data in spatial omics), a statistical spatial perturbation analysis toolkit. Using SPAC-seq and TARDIS, we linked gene perturbations to spatial phenotypes and pathways, uncovering how Icam1 loss in tumor cells promotes metastasis via immune suppression and macrophage polarization. In CD8+ T cells, we revealed Cd44's role in regulating spatial phenotypes by interacting with Spp1 on macrophages. We also demonstrated the model of the transcription factor-chemokine receptor axis coupling cell states with chemotaxis. SPAC-seq and TARDIS provide an effective framework to study spatially resolved functional genomics and pathways across diverse biological and disease contexts.

Although HLA-II molecules are classically associated with professional antigen-presenting cells, their expression by cancer cells has been recognized for several decades. It has been linked to immune infiltration, responses to immune checkpoint blockade, and clinical outcomes. However, the regulatory mechanisms governing tumor-associated HLA-II expression remain incompletely understood. Genome-wide CRISPR-Cas9 screening was employed to identify candidate regulators of HLA-II expression in human melanoma cells. Key candidates were functionally validated through genetic and pharmacological perturbation approaches. Integrated transcriptomic and epigenomic analyses were conducted to characterize regulatory mechanisms. Retrospective clinical analyses were performed using publicly available The Cancer Genome Atlas (TCGA) datasets to assess associations with immune infiltration, immunotherapy response, and survival. We identified the aryl hydrocarbon receptor (AHR) and its dimerization partner ARNT as critical, FICZ-responsive, positive regulators of HLA-II expression. AHR-ARNT promoted transcription of CIITA through direct binding to its promoter II (pII), in the absence of IFN-γ signaling. Clinically, an AHR-ARNT loss-of-function signature correlated with reduced immune infiltration, poorer response to immunotherapy, and inferior survival across cancer types. These findings reveal a previously unrecognized regulatory axis controlling HLA-II expression on cancer cells, suggesting that targeting the AHR-ARNT pathway may enhance tumor immunogenicity and improve immunotherapy efficacy.

Microsatellite-stable/proficient mismatch repair (MSS/pMMR) colorectal cancer (CRC) is characterized by a cold tumor microenvironment, with limited CD8⁺ T cell infiltration and poor responsiveness to immune checkpoint inhibitors (ICIs). Here, using an in vivo CRISPR/Cas9 screen in a CMT93 cell-derived murine tumor model, we identify Arid3b as a key negative regulator of CD8⁺ T cell infiltration and antitumor activity. Genetic ablation of Arid3b in CD8⁺ T cells significantly enhances their intratumoral accumulation and promotes robust tumor control. Mechanistically, Arid3b deficiency upregulates Runx3, driving a tissue-resident memory-like phenotype and effector function. Notably, the benefits conferred by Arid3b deficiency are abrogated upon Runx3 deletion, indicating a RUNX3-dependent mechanism. Together, targeting ARID3B could offer a promising strategy to reshape the tumor microenvironment and sensitize MSS CRC to immunotherapy.

Abstract High-grade serous ovarian carcinoma (HGSOC) is the most lethal gynecologic malignancy. HGSOC spreads within peritoneal cavity via transcoelomic dissemination, and the greater omentum is a frequent site of metastatic tumor deposition. Embedded beneath the omental mesothelium, the fat-associated lymphoid clusters (FALCs, known as milky spots) serve as portals for early peritoneal metastasis. Yet how metastatic ecosystems differ from the primary tumor microenvironment, and are spatially organized around FALCs remains unclear. To address this, we profiled matched primary and metastatic tumor samples from 17 HGSOC patients using single-cell RNA sequencing with paired TCR/BCR profiling and spatial multi-omics including Visium HD spatial transcriptomics, multiplex CODEX imaging and spatial metabolomics. In FALC-rich regions within metastatic lesions, we identify a stereotyped architecture in which small aggregates of B cells and plasma cells are interlaced with CD4+ and CD8+ T cells and dendritic cells. Stromal regions between these FALCs and adjacent tumor border are highly infiltrated by SPP1+C1Q+ macrophages. We show that the B cells in metastasis mainly consist of IgG+ and IgA+ populations, and are class-switched and clonally expanded. Integrating spatial transcriptomics with histology, we found the abundance of the niche enriched for IgA+ B cells associates with worse overall survival using a deep learning framework. Spatial metabolomics mapped spatial metabolic niches and showed that metastases shift toward glycolytic, lipid and phospholipid metabolism, whereas primary tumors retain more oxidative programs. Collectively, our findings provide a spatially resolved atlas of HGSOC metastasis and a foundation for spatially informed biomarker development. Citation Format: Yufeng He, Ce Luo, Yuanguang Meng, Zhe Zhang, Zexian Zeng. Spatially resolved multi-omics reveal metastatic ecosystem remodeling and immunomodulatory niches in HGSOC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 3975.

Advances in imaging- and sequencing-based spatial transcriptomics have increased molecular throughput and resolution, enabling the measurement and analysis of spatial transcriptomes at single-cell resolution. However, accurate cell segmentation remains challenging because cell morphology, tissue processing and staining methods vary across samples and platforms, limiting the accuracy and generalizability of existing algorithms. Here we show that DISSECT, a cell segmentation model integrating cytological images with spatial transcriptomic profiles, improves spatial single-cell transcriptome reconstruction. DISSECT uses a pretrained deep generative model to denoise multiscale image features, predicts cell instances with an instance-aware detection module and applies image- and transcriptome-derived gradient fields to refine segmentation masks. Benchmarking across multiple datasets showed that DISSECT achieved higher mean average precision than several existing segmentation tools. We further applied DISSECT to three pairs of gastric adenocarcinoma samples collected before and after anti-PD-1 treatment and profiled by Stereo-seq, illustrating its utility for downstream spatial biological interpretation.

Angiogenic and MHC-II TAM signatures correlate with different patients’ survivals, Related to Figure 1.

The mutually exclusive pattern of angiogenic and MHC-II also existed in identified TAM subsets, Related to Figure1

Gene expression signatures induced by different combinations of lactic acid, PGE2 and GM-CSF in macrophages, Related to Figure 3.

scRNA-seq datasets for correlation analysis of macrophages' functional programs with effector signature in T cells, Related to Supplementary Figure S12 and Methods

Customized CROP-seq sgRNA library of genes identified from ex vivo CRISPR screen, Related to Figure 5 and Methods

The angiogenic and MHC-II gene expression programs are mutually exclusive in human TAMs across a wide range of cancer types, Related to Figure 1.

MHC-II molecules are traditionally restricted to professional antigen-presenting cells (pAPCs), but increasing evidence highlights their expression in cancer cells, where they are associated with enhanced immune infiltration and improved clinical outcomes. However, the mechanisms governing cancer cell-intrinsic MHC-II expression remain poorly understood. Here, through genome-wide CRISPR-Cas9 screening in human melanoma cells, we identify the aryl hydrocarbon receptor (AHR) and its dimerization partner (ARNT) as critical, ligand-responsive regulators of MHC-II expression. Our analyses reveal that AHR–ARNT promotes transcription of the MHC-II transactivator CIITA through direct binding to its promoter II (pII), independently of IFN-γ signaling. Clinically, an AHR–ARNT loss-of-function signature correlates with reduced immune infiltration, poor response to immunotherapy, and inferior survival across cancer types. Together, our findings uncover a previously unrecognized, tumor-intrinsic regulatory axis of MHC-II expression and suggest that targeting the AHR–ARNT pathway may enhance tumor immunogenicity and improve responses to immunotherapy.

Gene summary file generated from MAGeCK analysis of the ex vivo CRISPR screen, Related to Figure 2

Effective immunotherapy relies on the presentation of tumor-derived neoantigens on the major histocompatibility complex class I (MHC-I) to activate CD8+ T cells. Deficiencies in this process are a key mechanism of immune evasion and resistance to checkpoint blockade. In this study, using an in vivo CRISPR-Cas9 screen, we unexpectedly found that inactivation of calreticulin (CALR), and other selected components of the peptide-loading complex (PLC), induced robust CD8+ T cell-mediated immune responses. We show that this effect is dependent on the expression of classical MHC-I on tumor cells. Mechanistically, loss of CALR reshaped the MHC-I peptide repertoire, favoring the presentation of low-affinity peptides in murine and human cell lines. Genetic or pharmacological inhibition of PDIA3, another PLC component, similarly induced antitumor effects. These findings reveal a previously unrecognized role of CALR and the PLC in regulating antitumor immunity and suggest that targeting this pathway could be a promising strategy to overcome immune resistance and improve the efficacy of cancer immunotherapies.

In vivo relevance of LGP factors in determining TAM phenotypes, Related to Figure 5.

TME metabolites are spatially correlated with niche-specific phenotypes of TAMs in human cancer, Related to Figure 6.

ISG + phenotype potentiate anti-tumor immunity via CXCL10-CXCR3 and CD40-CD40L interaction with CD8+ T cells, Related to Figure7.

Inactivation of Adar in myeloid cells rewire TAMs into ISG + phenotype, Related to Figure7.

Many cancer drugs that target cancer cell pathways also impair the immune system. We developed a computational target discovery platform to enable examination of both cancer and immune cells so as to identify pathways that restrain tumor progression and potentiate anti-tumor immunity. Immune-related CRISPR screen analyzer of functional targets (ICRAFT) integrates immune-related CRISPR screen datasets, single-cell RNA sequencing (scRNA-seq) data, and pre-treatment RNA-seq data from clinical trials, enabling a systems-level approach to therapeutic target discovery. Using ICRAFT, we identified numerous targets that enhance both cancer cell susceptibility to immune attack and T cell activation, including tumor necrosis factor (TNF) alpha-induced protein 3 (TNFAIP3), protein tyrosine phosphatase non-receptor type 2 (PTPN2), and suppressor of cytokine signaling 1 (SOCS1). In cancer cells, Tnfaip3 (A20) deletion activated the TNF-nuclear factor kappa-B (NF-κB) pathway, promoting chemokine expression and T cell recruitment to the tumor. T cell-mediated elimination of Tnaifp3-null cancer cells was primarily driven by TNF-induced apoptosis. Inactivation of Tnfaip3 in T cells enhanced anti-tumor efficacy. By integrating diverse functional genomics and clinical datasets, ICRAFT provides an interactive resource toward a deeper understanding of anti-tumor immunity and immuno-oncology drug development.

Tumor-associated macrophages (TAM) display remarkable functional heterogeneity, yet the molecular mechanisms driving their diverse phenotypes remain elusive. Using CRISPR screens in primary macrophages, we identified tumor-derived factors, including lactic acid, prostaglandin E2, and GM-CSF, as key modulators of TAM polarization. These factors interacted cooperatively and antagonistically to shape distinct TAM phenotypes that were highly conserved across human cancers. Mechanistically, lactic acid and PGE2 jointly induced angiogenic gene programs while suppressing GM-CSF-driven MHC-II expression at the chromatin level, creating mutually exclusive distributions of proangiogenic and MHC-II+ TAMs, which were differentially localized to specific spatial niches in the tumor microenvironment. Furthermore, we showed that shifting TAMs to an interferon-responsive phenotype, triggered by Adar inactivation, significantly promoted the infiltration of effector CD8+ T cells through specific receptor-ligand interactions. These findings uncover a conserved mechanism of TAM polarization and offer insights into therapeutic strategies for TAM reprogramming to potentiate cancer immunotherapy.