American Association for Cancer Research® (AACR)

Protein-protein interaction is one of the many mechanisms where individual cells communicate with nearby cells or extracellular matrix to modulate the tissue environment. Visualizing these interactions using a spatial platform can validate known interactions implicated in disease pathology within relevant spatial domains, especially in cancer. We developed the ProximityScope™ assay that can be used with the RNAscope™ Multiomic LS assay to detect protein interactions and their impact on molecular pathways by simultaneously visualizing proteins and RNA on a single tissue section. This workflow is fully automated on Leica Biosystems’ BOND RX staining platform. The assay detects one protein-protein interaction and can be combined with five RNA/protein targets simultaneously on the same section. Here, we demonstrate a complete end-to-end workflow by staining FFPE human tonsil tissue samples to detect PD1-PDL1 interactions along with immune cell markers such as PDCD1 RNA, CD68 RNA, CD4 protein, CD8 protein, and PanCK protein. Images were acquired using the Aperio FL 120 by Leica Biosystems.[AD1] Using the HALO® image analysis software from Indica Labs, we performed RNA and protein quantification as well as cell phenotyping. PD-1/PD-L1 interactions were successfully visualized between immune cells and tumor cells. T cells were identified by either CD8 or CD4 protein detection, macrophages were identified with CD68 RNA, while tumor cells were identified by PanCK protein staining. Interactions appear as punctate dots or dot clusters between two cells or on the cell surface. HALO software identified the PD-1/PD-L1 interaction between adjacent cells. This study demonstrates the importance of having an end-to-end spatial solution for staining, imaging, and quantification of target protein interactions to evaluate protein function. The ProximityScope assay has the potential to study a broad range of protein interactions to evaluate signaling pathways and gain biological insights, assess therapeutic success, or detect PD-1/PD-L1 and related interactions that can serve as biomarkers for patient stratification.For Research Use Only. Not for use in diagnostic procedures.

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related death worldwide, with limited treatment options and poor outcomes. Multikinase inhibitors remain the first-line therapy for advanced HCC. However, therapeutic resistance remains a significant challenge in HCC, and the molecular mechanisms underlying metabolic adaptation to drug resistance remain poorly understood. Here, we identify Y-box binding protein 1 (YBX1) as a key regulator of cholesterol metabolism that promotes tumor growth and drug resistance in HCC. Multi-omics and mechanistic analyses reveal that YBX1 transcriptionally activates Sterol Regulatory Element-Binding Protein 2 (SREBP2), a critical master regulator of cholesterol metabolism, and suppresses the cholesterol efflux transporter ABCA1, resulting in increased expression of cholesterol biosynthetic enzymes and intracellular cholesterol accumulation. Additionally, Cholesterol modulates TGFβ signaling and is implicated in drug resistance. This metabolic rewiring stabilizes membrane receptor tyrosine kinases (RTKs) and sustains downstream PI3K/Akt/mTORC1 and EMT signaling pathways, thereby fostering the development of drug resistance. Genetic silence or pharmacological inhibition of YBX1 or SREBP2 with SU056/Betulin restores sorafenib sensitivity and reduces tumor growth. Clinically, higher levels of YBX1 and SREBP2 expression are associated with poor therapeutic response and decreased overall survival in patients with HCC. These findings uncover a YBX1/SREBP2/Cholesterol metabolic axis that mediates adaptive resistance, offering a new therapeutic target to overcome drug resistance in HCC.
Keywords: Hepatocellular carcinoma, Drug resistance, YBX1, SREBP2, Cholesterol metabolism, Akt/mTOR signaling.

Desmoplastic small round cell tumor (DSRCT) is a rare and aggressive sarcoma driven by the EWSR1::WT1. Histologically, DSRCT is characterized by distinct spatial heterogeneity of tumor nests surrounded by a desmoplastic stroma. Our previous study showed that DSRCT exhibits heterogeneous expression of androgen receptor (AR)-associated and neuron-specific enolase (NSE)-associated markers, highlighting these as important lineage features of the disease. However, how these AR (Epithelial-like)- and NSE (Neural-like)- associated phenotypes are organized at the single-cell level and contribute to tumor biology remains poorly understood. This study aims to define these phenotypes and delineate their spatial and neighborhood patterns within the DSRCT microenvironment. We used the Lunaphore COMET system to evaluate 12 DSRCT patient specimens, comprising a 20-marker panel on nine slides and a separate 24-marker panel on three slides, which collectively covered epithelial, neural, fibroblast, endothelial, and immune cell types. Using Visiopharm, we applied deep-learning algorithms to identify individual cells and quantify protein expression data. Four custom RNAscope probes targeting EWSR1::WT1-associated neogenes were used to detect tumor cells and identify EWSR1::WT1 activity. Tumor cells were distributed along an AR-NSE expression spectrum, including AR-high, AR-low, double-negative (AR-NSE-), NSE-positive, top 1% NSE-strong, and hybrid phenotype (AR+NSE+). Among AR-associated phenotypes,gradient analysis revealed that AR-high tumor cells were enriched at the tumor nest center and gradually decreased in abundance toward the tumor-stroma interface, where AR-low tumor cells were more prevalent and in closer proximity to fibroblast-rich stromal regions. Among NSE-associated phenotypes, NSE-positive tumor cells were positioned closer to fibroblast-rich stromal regions. In contrast, both the top 1% NSE-strong and AR+NSE+ hybrid phenotypes localized deeper within the tumor region and were farther from fibroblasts. Our work identified novel conserved neighborhoods across samples: tumor-centered, transitional, and fibroblast-enriched neighborhoods. AR-high phenotypes predominantly mapped to tumor-centered neighborhoods, whereas AR-low phenotypes were enriched in transitional and stromal-interacting neighborhoods. These spatial distributions suggest distinct microenvironmental contexts at the center and periphery of the tumor nests, where stromal interactions are more pronounced. Future work will elucidate how stromal cues mediate phenotypic changes in DSRCT.

Glioblastoma (GBM) is the most common malignant brain tumor in adults. Despite extensive research, there haven’t been remarkable gains in resolving the seeds of GBM recurrence, and the outcomes for many patients suffering from this devastating disease remain poor. Our knowledge on GBM heterogeneity is mostly restricted to the surgically resectable tumor core, while functional characterization of tumor cells at the infiltrating edge remains largely elusive due to the presence of normal functional brain tissue in the peritumoral lesion. Edge-derived cells exhibit larger capacity for infiltrative expansion and are the main drivers of treatment failure and tumor recurrence, making them action targets for novel treatment approaches.In this study, we present a first-of-its-kind integrative spatial investigation of GBM, combining two complementary spatial omics modalities high-definition spatial transcriptomics (ST – Visium HD) and spatial proteomics (SP – COMET) to achieve a comprehensive morphological, transcriptomic, and proteomic characterization of invasive tumor edge in situ. This multimodal spatial framework enabled to resolve the complex molecular landscape of the GBM periphery and to identify druggable biomarkers specific to edge-derived malignant cell populations.By integrating pathologically annotated H&E images with high-resolution spatial gene expression, we delineated patterns of tissue architecture and captured continuous gradients of gene expression across tumor-brain interface. Using NicheCompass, we dissected tissue hierarchies and spatially localized cellular processes within the infiltrative compartment of GBM. Through unsupervised phenotyping, we identified top spatially variable genes and active gene programs, revealing modules indicative of tumor cell hijacking of neuronal pathways. These findings align with previously described mechanisms of glioma-neuron synaptic coupling and formation of neurite-like tumor microtubes, consistent with enrichment of OPC- and NPC-like cellular states at tumor margin.Complementing the transcriptomic layer, SP was used to guide single-cell segmentation for ST analysis, infer cell types based on canonical phenotypic markers, and characterize cell state and function through protein-level profiling. This dual-modality approach allowed for precise spatial mapping of highly invasive edge cell populations and their functional states.By integrating multiple spatial omic layers, our study provides an unprecedented multimodal view of GBM invasion and identifies novel, spatially defined biomarkers that distinguish malignant edge-derived cells. These findings hold potential translational value as diagnostic and prognostic tools, enabling early assessment of treatment response and facilitating personalized therapeutic strategies aimed at mitigating GBM progression and recurrence.

Introduction: The spatial relationships between tumor cells, immune infiltrates, and the surrounding stroma—characterized by both protein expression and transcriptional activity—are critical determinants of therapeutic response and disease prognosis. Traditional spatial profiling approaches often require serial sections, compromising exact cellular co-localization, or utilize harsh processing steps that degrade target integrity. There is an urgent need for robust, high-throughput, single-slide methods to co-detect protein and RNA with high fidelity.
Methods: We developed and validated a novel, streamlined hybrid workflow combining Ultivue’s highly multiplexed InSituPlex technology (protein detection) with Advanced Cell Diagnostics’ (ACD) newly developed protease-free capability with RNAscopeTM in situ hybridization (ISH) workflow. This pairing is crucial: the protease-free ISH step prevents the degradation of critical cell surface epitopes required for robust downstream multiplex immunofluorescence (mIF) protein staining. A comprehensive panel was rapidly assembled, including key immuno-oncology protein markers for antibody drug conjugates (ex. HER2) via InSituPlex, co-detected with relevant transcriptional targets (e.g., cytokine mRNA, therapeutic target transcripts) via ISH. The assay was applied to FFPE tissue sections from Non-Small Cell Lung Cancer (NSCLC), Triple-Negative Breast Cancer (TNBC), and Gastric Cancer (GC) to demonstrate broad applicability.
Results: We successfully achieved simultaneous, high-resolution co-localization of up to 5 protein markers and 3 RNA transcripts within a single tissue section across all three heterogeneous cancer types. The protease-free pre-treatment maintained optimal tissue and epitope morphology, enabling exceptional signal-to-noise ratios for both protein and RNA channels. Critically, this combined multi-omic readout allowed for the precise spatial phenotyping of cell populations (e.g., CD8+ T cells co-expressing specific cytokine mRNA) and the assessment of spatial proximity between cells defined by combined protein/transcriptional signatures. The modularity of the Ultivue assay allowed for rapid assembly and substitution of antibody panels, enabling high-speed assay optimization.
Conclusion: This innovative, combined Ultivue InSituPlex and protease-free ACD ISH assay provides a powerful, single-slide platform for spatially resolved multi-omic analysis. It overcomes the technical limitations of traditional sequential staining, offering a high-throughput, high-fidelity tool that can be rapidly deployed for complex biomarker validation, detailed tumor microenvironment characterization, and improved patient stratification strategies in translational oncology and clinical trials.

ntroduction: Identifying unique cell phenotypes in cyclic immunofluorescence (cycIF) images is typically achieved through manual gating or unsupervised clustering of segmented cells based on marker expression profiles. While effective, segmentation-based methods are limited by predefined thresholds and cell type classification schemes. MORPHAEUS is a new Python-based software that infers cell types and multicellular structures directly from pixel-level imaging data using the variational autoencoder (VAE) deep learning architecture. This method enables unsupervised identification of cellular and morphological patterns without relying on image segmentation. Here, we compared segmentation-based phenotyping with MORPHAEUS-derived cell type classifications in a liver metastasis from a patient with PDAC imaged for 32 markers on the Lunaphore COMET platform.
Methods: Cell segmentation was performed on DAPI-counterstained nuclei using the U-net algorithm in Visiopharm. Mean per-cell marker intensities were quantified, and cell types were assigned based on a nested classification scheme using binary thresholding and prior biological knowledge. For MORPHAEUS analysis, 9×9µm image patches centered on nuclear centroids were extracted and stored in Zarr file format for VAE model training. Image patch encoding containing information on marker intensity, morphology, and local neighborhood contexture were clustered using Leiden community detection to identify cell types.
Results: MORPHAEUS identified several clusters consistent with those identified by manual gating, including a cluster with high CD3, CD8, CD69, and CD103 expression corresponding to tissue-resident memory CD8+ T cells. Clusters enriched for PanCK were consistent with tumor cells, with a subset co-expressing Ki67 indicative of proliferating tumor. MORPHAEUS also revealed novel clusters not captured by manual classification, including one with unexpected co-expression of CD4 and CD11C. Inspection of the primary image revealed that this cluster represented cell-cell interactions involving CD4+ T helper cells and CD11C+ dendritic cells, consistent with their known cooperative roles in antigen recognition.
Conclusions: Traditional classification provides a robust framework for quantifying predefined cell populations, but its reliance on manual gating and fixed marker definitions limits the discovery of novel or context-dependent phenotypes. MORPHAEUS offers a complementary, unsupervised approach capable of identifying rare and previously uncharacterized cell states as well as biologically meaningful spatial interactions that may be missed by conventional segmentation-driven analyses. These findings underscore the value of pixel-level deep learning as a powerful adjunct to traditional spatial phenotyping, enabling deeper insights into tissue organization and biomarker discovery.

ntroduction: Tertiary lymphoid structures (TLSs) are ectopic lymphoid formations. In cancer, they promote antitumor immunity by supporting B-cell maturation, antibody production, and sustained T-cell activation. TLSs can convert immunologically non-inflamed tumors into inflamed ones. However, the vascular and molecular cues that govern TLS induction in human tumors remain poorly defined.
Methods: Formalin-fixed paraffin-embedded tissue sections from twelve primary lung adenocarcinomas were analyzed. Tumor-associated vasculature was histologically annotated and stratified into three categories based on TLS status: (1) noTLS (absence of TLS), (2) iTLS (presence of immature TLS), and (3) mTLS (presence of mature TLS). GeoMx spatial transcriptomics was performed on endothelial regions. Protein-level validation was conducted using sequential immunofluorescence on the COMET platform.
Results: Endothelial cells in noTLS regions showed high expression of COL5A1, COL3A1, FN1, ERRFI1, COL1A1, IFI6, SPP1, MDK, and BGN, whereas CXCL13 and CCL19 were enriched in mTLS. Gene-set analyses revealed that noTLS endothelium exhibited a myofibroblast-like, extracellular matrix-producing, EndoMT-associated program consistent with a fibrotic, immunosuppressive microenvironment. In contrast, mTLS endothelium displayed immune-activated signatures aligned with lymphoid organogenesis and fibroblastic reticular cell-like function. Receptor-ligand analysis showed preferential Wnt signaling in noTLS, while chemokine signaling dominated in mTLS. COMET immunofluorescence confirmed that FN1 was highly expressed in noTLS regions, whereas CXCL13 was highly expressed in mTLS.
Conclusion: Tumor endothelial cells display striking context-dependent plasticity. noTLS vasculature undergoes EndoMT and adopts a matrix-producing, immunosuppressive phenotype that may actively suppress TLS formation. Conversely, mTLS endothelium acquires immune-organizing properties that favor TLS maturation and lymphoid compartmentalization. These findings identify endothelial reprogramming as a potential therapeutic strategy to induce TLSs and enhance response to immune checkpoint blockade.

Background: Glioblastoma malignancy is strongly driven by the formation of tumor microtubes (TMs), which promote intercellular connectivity and calcium (Ca²⁺) wave propagation. The mechanisms by which Tumor Treating Fields (TTFields) influence these TM-mediated processes remain incompletely understood.
This study aims to elucidate the effects of TTFields on the structural and functional organization of tumor cell networks and to assess potential frequency-dependent modulation of signaling pathways, including NF-κB.
Methods: A comprehensive set of biological model systems is being utilized, including 2D glioblastoma cell monolayers, 3D brain organoids, and in vivo, awake, head-fixed mouse models with chronic cranial windows for longitudinal imaging. Live-cell imaging with confocal and multiphoton microscopy enables real-time observation of morphological and functional tumor dynamics. Quantification of Ca²⁺ signaling is being performed using Cellpose-based segmentation and custom Python analysis pipelines. Immunohistochemistry and spatial transcriptomics (Visium HD) are currently employed to dissect molecular mechanisms; COMET-based immunofluorescence and RNAscope FISH are planned to enable spatially resolved multi-omics.
Results: TTFields induced a marked disruption of glioblastoma network architecture and function. Specifically, treatment resulted in >50% reduction in global GCaMP8s-mediated Ca²⁺ activity, a decrease in pacemaker-like cell populations, and significant reductions in synchronization and Ca²⁺ co-activity within S24 glioblastoma cells in both 2D and 3D models.
Conclusion & Outlook: TTFields disrupt glioblastoma network integrity and Ca²⁺ signaling, potentially reducing tumor aggressiveness. Preliminary data from 3D brain tumor organoids and in vivo models support our previous in vitro results regarding TTFields-induced activity changes. Parallel studies in patient-derived organoids explore frequency-dependent signaling effects – including NF-κB and MAPK pathways – via spatial transcriptomic profiling. Together, these efforts aim to further elucidate the mechanistic underpinnings of TTFields action and their impact on glioblastoma plasticity and network organization.

Background: Triple-negative breast cancer (TNBC) is an aggressive subtype with high recurrence and resistance to standard therapies. Midkine (MDK), a heparin-binding growth factor that functions as a cytokine, is overexpressed in TNBC and promotes tumor progression, immune evasion, and chemoresistance. We recently developed HBS-101, as a first-in-class small-molecule MDK inhibitor and demonstrated its therapeutic efficacy in preclinical TNBC models. This study evaluates whether MDK inhibition enhances chemo and immunotherapy efficacy as well as the mechanistic understanding of combination therapy effects.
Methods: TNBC cell lines were treated with HBS-101 alone or in combination with chemotherapy (doxorubicin and paclitaxel), and assessed for viability, apoptosis, and synergy. 3D organoids derived from patient-derived xenografts (PDX) were employed to examine the ex vivo effects of combination therapy. The efficacy of HBS-101 with chemotherapy and immunotherapy (PD-L1 inhibitor) was evaluated using human and murine cell line-derived xenograft (CDX) models of TNBC respectively. In vivo, orthotopic TNBC tumors were established in immunocompetent mice and treated with combination treatments. Tumor growth, and immune cell infiltration were analyzed using xenograft models. RNA-seq was performed on HBS-101 treated cells to identify changes in immune and apoptotic pathways. Molecular and immunological effects were examined using RT-qPCR, Western blotting, flow cytometry, and immunohistochemistry.
Results: Combination treatment with HBS-101 and chemotherapy produced synergistic anti-tumor activity in TNBC models, significantly reducing 2D and 3D cell viability, stemness, and tumor growth relative to monotherapies. Similarly, combining HBS-101 with immunotherapy also yielded strong synergistic effects in syngeneic TNBC models, outperforming individual treatments. Mechanistic studies revealed that HBS-101 disrupted MDK-mediated signaling pathways, including STAT3, thereby sensitizing tumor cells to treatment and enhancing immune activation. Combination therapy also increased infiltration of CD8 positive T cells and macrophages, accompanied by elevated levels of IFN-gamma and granzyme B. RNA-seq analysis showed upregulation of immune and apoptotic pathways and suppression of immunosuppressive signals. These molecular changes correlated with reduced tumor burden, and increased apoptosis, supporting the mechanistic synergy of MDK inhibition in combination therapy.
Conclusion: MDK inhibition with HBS-101 enhances the efficacy of chemo-immunotherapy in TNBC by disrupting oncogenic signaling and promoting immune activation. These findings support the therapeutic potential of HBS-101 as part of a combination strategy to overcome resistance and improve outcomes in TNBC.

Resistance and metastasis continue to make breast cancer a clinical challenge. VC2-GM-CSF drives antitumor immunity; paclitaxel remodels tumor environment to boost immune infiltration. Thus, combined use may yield superior therapeutic benefit over the use of either agent alone. This study investigates the potential synergistic interactions of VC2-GM-CSF with paclitaxel using the murine 4T1 breast cancer model in vitro and in vivo. In vitro, 4T1 cells were treated with paclitaxel alone or in combination with VC2-GM-CSF, and cell viability was assessed by MTT assay. Viral entry was quantified using a flow-cytometric assay. We further explored the in vivo antitumor effect of combined VC2-GM-CSF and paclitaxel in an orthotopic 4T1 stage IV metastatic model of breast cancer. Established 4T1 tumors in BALB/c mice were treated intratumorally with VC2-GM-CSF combined with paclitaxel. On day 31, tumors and lungs were collected for histopathological evaluation. Flow cytometric quantification of lymphocytes infiltrating the tumors was performed for CD45⁺, CD3⁺, CD4⁺, and CD8⁺ cells. The combined treatment was substantially more effective in inducing dose-dependent decreases in the viability of 4T1 cells and cell death than either single therapy in vitro. Flow cytometric entry assays showed that paclitaxel does not impair the entry of VC2-GM-CSF into cancer cells. In vivo, the combination significantly reduced primary tumor growth compared with the control treatment. Flow cytometry and immunohistochemistry analyses indicated increased T-cell infiltration in tumors after combination therapy. Together, these results demonstrate an enhanced antitumor effect resulting from the complementary cytotoxic and immunomodulatory mechanisms of paclitaxel and oncolytic virotherapy.

Background
The tumor microenvironment (TME) plays a pivotal role in cancer progression, immune evasion, and therapeutic response. Understanding the spatial organization and functional states of cells within the TME is essential for advancing immuno-oncology and precision medicine [PMID: 40102282]. However, simultaneous visualization of secreted molecules and cellular phenotypes in situ remains a major challenge in the spatial biology field [PMID: 39930476].  

Here, we employed an automated hyperplex multiomics assay to simultaneously detect RNA and protein expressions to spatially map cell phenotypes and their functional states in the TME of multiple cancer types. 

Methods
We examined a formalin-fixed paraffin-embedded tissue microarray (TMA) comprising various human cancer types: prostate, lung, breast, colorectal, melanoma, and lymphoma. A TMA section was stained and imaged on the COMET™ platform, integrating RNAscope™ HiPlex Pro for transcript detection and sequential immunofluorescence (seqIF™) for proteomic analysis [PMID: 22166544; 37813886; 41065276]. Image was analyzed using HORIZON™ software to extract single-cell and spatial features. 

Results
The automated multiomics approach enabled the concomitant in situ detection of over 100 biomolecular targets, including 12 transcripts and more than 90 proteins on the same section. High-resolution spatial profiling of the TME allowed accurate mapping of cancer, stromal, vascular and diverse immune cell subsets. It further revealed key molecular features associated with tumor-suppressor or proto-oncogene activity, including markers of proliferation, apoptosis, and immune checkpoint regulation. Concurrent detection of cytokine and chemokine transcripts highlighted localized immune signaling and cell-cell communication within tumor and stromal compartments.  

Conclusions
This multiomics workflow offers a powerful tool for in-depth characterization of the TME across multiple cancer types, while significantly reducing sample consumption. Spatial profiling provides new opportunities to dissect the tissue architecture and immune dynamics to identify functional cell states and interactions to be exploited in immunotherapy and personalized medicine

Background Genetic changes in mammalian genes including exon skipping and point mutations play a crucial role in oncogenesis and predicting patient survival and disease prognosis. Studies have shown that cancer cell lines can be distinguished from non-cancer cell line based on gene transcript isoform. As example, MET (receptor tyrosine kinase) exon 14 skipping mutation and EGFR exon 19, 21 deletion or exon 20 insertion mutation have emerged as a biomarker in various cancer types. Additionally, alternative-splicing plays a crucial role in disease progression and therapeutic response. Spatial visualization of gene isoform holds the potential of providing better understanding of tumor microenvironment.
Methods We developed a next-generation, manual and fully automated fluorescent assays on the Leica platform that enables specific and simultaneous detection of highly similar RNA targets – isoform-specific exon-exon junctions. The assay utilizes specificity and sensitivity of RNAscope^TM technology to visualize single RNA molecule on formalin fixed paraffin embedded (FFPE) tissue and cell pellets. To validate the new workflow, endogenous control genes were detected using 1zz probes on HeLa cell pellets and Mouse multi tissue array. Signal and background were quantitatively compared to original BaseScope 1zz probes. To visualize MET exon 14 skipping (METΔ14), MET WT exon and exon 14-15 junction specific probes were detected on the WT and variant positive cell pellets. Further, exon 13-15 or exon 14-15 junction probe and exon 15 probes were duplexed and co-detected in MET WT and Δ14 cell pellets, respectively using BaseScope duplex assay that utilizes streamlined RNAscope chemistry to potentially co-detect mRNAs, proteins and protein-protein interactions.
Results Signal (as determined by dot counts/cell) from control gene using new BaseScope technology was comparable to signal generated using original BaseScope assay. Next, we visualized specific signal from METΔ14 exon junction probe on MET mutant pellet and WT MET probe on WT cell pellet, using dual fluorophores to identify both isoforms on the same slide. No signal was observed from mutant probe in WT pellet indicating specificity of the assay. Similarly, MET exon 14-15 junction probe and exon 15 probes were duplexed in the assay and visualized using two distinct fluorophores on wild-type cell pellet. Exon 13-15 junction probe and exon 15 probe were duplexed and co-detected using two fluorophores on mutant cell pellet. These results demonstrate the assay’s efficacy in same-slide detection of cells expressing two short targets.
Conclusions This BaseScope Duplex assay provides spatial resolution of isoform and point mutation-specific gene expression, offering a powerful tool for exploring cell-specific transcript variants within the tumor microenvironment and potentially advancing precision oncology research.

Introduction
Head and neck squamous cell carcinoma (HNSCC) incidence has increased by at least 23 % globally over the last ten years and is predicted to continue to rise by 30 % annually. Treatment for HNSCC often includes a multidisciplinary approach (i.e., chemotherapy + surgery) but success rates are still limited with less than half surviving more than two years post-treatment (Nieszporek et al., 2025). Significant improvements in the number of biomarkers that can be screened at once, while preserving tissue integrity, have been made over the last decade. These staining and imaging improvements have contributed to the identification of novel therapeutic targets while limiting the egregious use of precious tissues. This study aimed to interrogate the tumor microenvironment of head and neck squamous cell carcinoma using the same tissue slice using a fully automated cyclic IF system.
Methods
Two panels of 20 biomarkers each were designed using immune markers, tissue architectural markers, and specific targets of interest in head and neck squamous cell carcinoma. Panel 1 was comprised of I/O markers (CD3, CD4, CD8, FoxP3, CD56, CD20, CD68, CD11c, aSMA, PD-L1, PD-1, CD45, CD27, CTLA-4, CD19, PCNA, CD14, CD16, SOX10, CD79a) while Panel 2 was largely comprised of discovery markers (ALDH2, IL-8, CK17, MMP9, MAGE-A4, EpCAM, EGFR, CK14, CK19, CK5/6, p53, CD44, ZEB1, ZEB2, beta-Catenin, E-Cadherin, Vimentin, COL1A1, COL4A1, FAP). Optimization was achieved using FFPE tissue microarrays containing normal and cancerous skin tissues. Pre-processing steps were done in the Epredia© PT Module using Tris-EDTA ph9 solution, at 100°C, for 1 hour. Panel 1 was stained first followed by Panel 2 without removal of the tissue from the instrument. Automated immunofluorescent staining and imaging of the samples was performed on the Lunaphore COMET™ system.
Results
The sequential staining of two 20-plex protein panels on the same tissue demonstrates the COMET’s capability of maximizing the number of biomarkers that can be evaluated. This protocol significantly reduces the use of tissue necessary, helps maintain tissue integrity for downstream processing (i.e., H&E staining), and minimizes costs by reducing chip and reagent use.
Conclusions
This comprehensive interrogation of the HNSCC tumor landscape with the use of high-plex, fully automated sequential immunofluorescent staining highlights many of the capabilities of the latest spatial technologies. In using this approach, deeper dives of tumor microenvironments can be achieved with minimal tissue use while keeping reagent and material costs low.

Prostate cancer exhibits diverse cellular composition, and the interaction between the tumor epithelial and tumor microenvironment (TME) cell types plays a crucial role in disease progression. However, due to technical limitations, systematic characterization of microenvironmental influences on prostate cancer progression remains insufficient. This study combined an assembled single-cell transcriptomic data (163 prostate samples; 756,000 high-quality cells) with the spatial data we obtained from 20 primary prostate tumors using the Visium HD platform to create high-resolution spatial profiles of the TME in prostate cancer. Our approach enabled detailed investigation of the dynamic remodeling of TME during prostate cancer progression from benign to low- and high-grade lesions. Our spatial transcriptomics data cohort includes 13 high-grade (grade group>=2) and 7 low-grade cases (grade group 1). We used the single cell clustering results as a reference and applied robust cell type decomposition (RCTD) to annotate the Visium HD data. In total, we identified 15 major cell types within the microenvironment and observed significant differences in the distribution of TME cell types and their frequency across benign, low-grade, and high-grade prostate cancers areas. Inflammatory fibroblasts and their gene signature were more prevalent in benign and low-grade microenvironments, whereas they decreased in high-grade TME. Conversely, myofibroblasts and their gene signature were enriched in high-grade TMEs. Within the smooth muscle populations, the more proliferative and migratory synthetic type (THY1-high) is increased in high-grade TMEs, while the contractile type (RERGL-high) was enriched in benign and low-grade microenvironments. Among endothelial cells, the abundant pro-angiogenesis subpopulation (EDNRB-high) was noted in high-grade TMEs, in contrast to the preferential enrichment of anti-angiogenesis subpopulation (RGS16-high), in localized to benign and low-grade microenvironments. Our findings uncover variations in the TME across tumor stages and implicate several important cell types that may influence prostate cancer progression, offering clues for future therapeutic exploration.

Introduction: Glioblastoma is a lethal brain tumor with poor response to current therapies, which include surgery, chemotherapy, and conventional radiation therapy (CONV-RT). Although, CONV-RT (0.01Gy/second) to brain tumors stimulate tumor antigen release, it also recruits immunosuppressive myeloid-derived suppressor cells and is associated with neurotoxicity. Ultrahigh-dose-rate or FLASH-RT, which delivers CONV-RT doses over a significantly shorter period (more than 40Gy/second) maintains tumor control, reduces normal tissue injury and is less immunosuppressive compared to CONV-RT across multiple cancer types. In our study, we compared the effects of FLASH-RT to CONV RT in syngeneic mouse GL261 glioblastoma-bearing mice. We hypothesized that FLASH-RT would be equally or more effective than CONV-RT for tumor control and result in less immunosuppression within the tumor and systemically.
Methods: We stereotactically implanted 2 × 10⁵ mouse GL261 cells into the right forebrain of C57BL/6 mice. Five days after tumor initiation, tumor bearing mice were treated with Sham-RT (control), CONV-RT (mean dose rate > 0.373 Gy/s), or FLASH-RT (mean dose rate >3.6×10⁶ Gy/s). Brain tumor tissue and peripheral blood were collected on days 5 and 12 after treatment. To assess for changes in the tumor microenvironment after CONV-RT or FLASH-RT, we performed 10x Xenium spatial transcriptomics analysis (stRNA-seq; brain tumor, n = 6 mice per group), Lunaphore COMET multiplexed immunofluorescence assay (brain tumor, n = 6 mice per group), and flow cytometry (peripheral mononuclear cells, n = 6 mice per group). We also evaluated survival outcomes following treatment (n = 8 mice per group).
Results: FLASH-RT significantly improved overall survival rate of GL261 bearing mice compared to CONV-RT (P <0.05) and Sham-RT (P < 0.001). FLASH-RT treatment markedly increased intratumoral CD8+ T-cell infiltration compared with CONV-RT (P <0.01) and Sham-RT (P < 0.01). We also found that compared to FLASH-RT, CONV-RT caused a significant decrease in circulating PD-1⁺CD8⁺ T-cells (P < 0.01), a potent antigen reactive cytotoxic T-cell population previously identified in human patients with glioblastoma.
Conclusion: FLASH-RT is associated with better tumor control in mouse GL261 glioblastoma, increased intratumoral CD8+ T-cell infiltration, and preserves circulating antigen reactive PD1⁺CD8⁺ T-cells. These results indicate that FLASH-RT may synergize better with immune checkpoint inhibitors to re-invigorate anti-tumor T-cell responses against glioblastoma.

Abstract

The cervical microbiome plays a pivotal role in shaping vaginal health and influencing tumor prognosis. Tumor associated microbes secrete bioactive metabolites that modulate both tumor behavior and its surrounding microenvironment. Our lab previously identified a species of L-Lactate producing Lactobacillus (LAB), Lactobacillus iners (L. iners), to be associated with poor chemoradiation response. In vitro experiments revealed that conditioned media from L. iners enhances radiation resistance in cervical cancer cells—a phenomenon replicated by direct lactate supplementation. This study aims to characterize the impact of microbiome-derived lactate on the tumor microenvironment.
Hypothesis: We propose that L-lactate produced by cancer-derived LAB supports tumor progression by fostering an immunosuppressive niche.
Methods: LAB strains were isolated via targeted culture from cervical, vaginal, vulvar, and oral cavity cancers. These isolates were used in co-culture systems and to generate cell-free bacterial conditioned media (CFS) to assess their capacity to interact with cancer cells and promote tumor-supportive conditions. For initial studies, we focused on cervical cancer isolates. To evaluate LAB-derived lactate’s role in radiation resistance, we performed cell viability assays (CyQUANT) on HeLa, SiHa, and CaSki cervical cancer lines exposed to CFS from patient-derived LAB and radiation. Lactate production was quantified using the Diazyme D-/L-Lactate rapid test kit, enabling precise source attribution between bacterial and tumor origins. We developed a syngeneic lactate exposure model using mEER (mouse oral line) xenografts in C57BL/6 mice, administering weekly intratumoral lactate injections. The tumors were radiated after the tumors reached ~75mm³. To evaluate the immune landscape, multiplex immunofluorescence using the Lunaphore COMET microfluidic platform was performed on mEER tumors.
Results: Cervical cancer cells exhibited increased radiation resistance when co-cultured with L-lactate-producing LAB, the isoform preferentially metabolized by human cells. LAB emerged as the dominant source of lactate in co-culture, confirming its role in driving lactate accumulation within the tumor microenvironment. In vivo, lactate-treated tumors grew significantly larger post-radiation (mean volume 48 mm³ vs. 74 mm³, p < 0.0001). Immune profiling revealed elevated T-cell infiltration, with a skew toward regulatory T cells and reduced CD8⁺ populations. The lactate-rich microenvironment showed marked upregulation of immune checkpoint proteins CTLA4 and PD-L1, suggesting functional exhaustion of infiltrating immune cells.

Background: Pancreatic ductal adenocarcinoma (PDAC) is characterized by extensive desmoplasia, with cancer-associated fibroblasts (CAFs) dominating the tumor microenvironment (TME) and driving collagen-rich extracellular matrix (ECM) deposition. While resident pancreatic fibroblasts exert natural tumor-restrictive functions, CAF/ECM units adopt both tumor-promoting and tumor-suppressive phenotypes, critically shaping tumor progression, therapeutic response, and patient outcomes. Defining inter-patient heterogeneity in CAF/ECM phenotypic traits and understanding their functional shifts in response to therapy is essential for predicting clinical outcomes, predicting therapy responses, and identifying new TME-targeting strategies.
Methods: We developed a human PDAC TME-within-a-chip integrating fresh tissue-derived 3D CAF/ECM units cultured in customized microfluidic chambers on glass slides. High-plex sequential immunofluorescence (seqIF™) on the COMET™ platform standardized ~20 mesenchymal biomarkers to distinguish tumor-supportive from tumor-restrictive CAF phenotypes. Automated image analysis (HORIZON™ software) quantified single-cell, subcellular, and extracellular marker expression patterns.
Results: CAF activation-state interventions revealed marked alterations in canonical TGFβ signaling, reduced Ki67-positive cell proliferation, and changes in ECM architecture and composition, reflecting shifts in stromal functional states. This integrated platform effectively resolved CAF/ECM unit heterogeneity and tracked TME adaptations in response to therapeutic perturbation.
Conclusions: This technological advancement provides a robust, medium-throughput framework for dissecting CAF heterogeneity and exploring stromal biology in PDAC (and other cancers). The platform enables comprehensive profiling from minimal starting material while substantially reducing antibody consumption and processing time compared to conventional methods, establishing a scalable foundation for functional screening and evaluating therapeutic interventions in complex TMEs.

Breast cancer is one of the most common cancers worldwide, but it is more fatal and rarer among men. There are no male patient derived breast cancer cell lines commercially available, limiting research on breast cancers in the male patients. Consequently, most research and therapeutic approaches are largely based on female research subjects because of limited information on male mammary carcinoma. Thus, development of male breast cancer cell lines would provide an excellent biological and preclinical model to study male specific mammary carcinoma. This study aims to establish novel male breast cancer cell lines using tumors derived from male mammary carcinoma patients. Mammary tumors were obtained from three consented male breast cancer patients undergoing mastectomy at the Simmons Cancer Institute. The breast tumor tissues were minced, enzymatically digested overnight at 370c, strained using sterile strainer, and centrifuged to pellet the cells on the following day. The pellets were suspended in HUMEC Ready Mix media and cultured for several weeks until adherent epithelial cells started to grow. The epithelial cell populations were purified by positive flow sorting using EpCAM antibody (epithelial marker) and considered immortal upon reaching passage 20. Out of three tumor samples, two unique male breast cancer cell lines from two different tumor samples have been successfully cultured and named as male breast cancer cell-1(MBCC-1) and male breast cancer cell-2 (MBCC-2). During the flow sorting, MBCC-1, which is currently in passage-50, was noted to be 70% epithelial (EpCAM positive) while MBCC-2 cells, which are currently at passage-19, were 20% EpCAM positive. The average circularity and diameter of the MBCC-1 ranges between 0.6-0.8uM and 12-19 microns while the average population doubling time is forty-two hours. Further, MBCC-1 is capable of anchorage independent growth by soft agar colony formation assay and shows positive expression for Cytokeratin, CD24, CD44, ALDH1, cKIT, ki-67 but is negative for Collagen-IV and fibroblast activation protein (FAP) protein. Studies are ongoing to further characterize both the male breast cancer cell lines, including biomarkers, gene expression profiling, karyotyping, cell proliferation rate, soft agar colony formation assay, nutrient dependency tests, and tumorigenicity assays. In summary, two novel male breast cancer cell lines have been established for studying the biology of male breast cancers.

Adenoid cystic carcinoma (ACC) is a rare malignancy characterized by indolent growth but a high rate of metastasis, with approximately 40-60% of patients developing distant disease. Systemic therapies, including immune checkpoint inhibitors (ICIs), have shown minimal efficacy, and multiple clinical trials have failed to identify effective treatments for recurrent or metastatic ACC. In previous work, we profiled the ACC immune microenvironment using multiplex immunofluorescence (mIF) and found that ACCs are “cold” tumors, with scarce tumor-infiltrating lymphocytes (TILs) and uniformly low expression of B2M and HLA class I across nearly all analyzed samples in our cohort of 24 ACCs. Only metastatic lesions displayed focal HLA class I expression. Spatial transcriptomic analysis revealed that these focally positive regions were associated with an interferon-γ-driven transcriptional program. Short-term ex vivo treatment of ACC tissues with interferon-γ or a STING agonist strongly upregulated HLA class I, B2M, and PD-L1, suggesting that immune visibility of ACC can be pharmacologically restored. However, using surrogate cell lines (given the lack of established ACC cell lines), we observed that STING activation induced a negative feedback loop leading to suppression of STING signaling pathway with reduced STING, IRF3 and TBK1 expression after 24 hours. Co-treatment with a range of agents – including proteasome and lysosome inhibitors, kinase pathway inhibitors, epigenetic modifiers, and modulators of NF-κB or PI3K/AKT signaling – did not prevent this feedback inhibition. Clinically, one ACC patient treated with a combination of a STING agonist dazostinag and pembrolizumab over nine months showed a ~70% reduction in tumor burden, while other patients experienced stable or progressive disease. One hypothesis is that differential activation of the STING feedback mechanism may underlie these varied clinical responses. These findings indicate that while ACC cells retain the machinery to upregulate antigen-presentation pathways, intrinsic mechanisms of STING pathway repression may limit the durability of immunostimulatory responses, underscoring the need for strategies to sustain interferon signaling in ACC. Ongoing work focuses on identifying strategies to overcome this feedback inhibition and on testing alternative agents capable of enhancing antigen presentation without triggering the same suppressive cascade, with the goal of developing more durable immunostimulatory therapies for ACC.

Background
Understanding the complexity of the tumor microenvironment (TME) requires simultaneous insight into multiple biological domains. Studying the molecular interactions that govern intercellular and intracellular signaling in combination with the analysis of cell phenotypes and transcriptional states, can guarantee an improved comprehension of key pathways driving tumor growth or the response of anti-cancer therapies. Protein–protein interactions (PPIs), such as PD-1/PD-L1, are central to immune evasion and targets of important immunotherapies. Despite the success of PD-1/PD-L1 and other checkpoint inhibitors, patient stratification for these therapies has been challenging, and marker expression alone has shown to not fully capture functional engagement or predict an efficient drug response. Here, we present a fully automated workflow that combines three omics layers on the same tissue section: protein-protein proximity, RNA, and protein expression, enabling a more comprehensive view of cellular interplay in cancer and modeling of treatment outcomes.

Methods
The multiomics assay runs on the COMET™ platform. It allows the co-detection of: (i) RNA profiling via RNAscope™ HiPlex Pro, (ii) Protein expression through sequential immunofluorescence (seqIF™, PMID: 37813886), (iii) Protein–protein proximity detection using oligonucleotide-conjugated secondary antibody pairs and RNAscope™ amplification chemistry. 

Proximity signals are interpreted as probabilistic indicators of molecular interactions and supported by multiple controls to ensure their specificity: from the colocalization of seqIF™ signals to negative controls run on the same section. 

Conclusions
In this study, we demonstrated that it is possible to combine the detection of proximity signals for the analysis of intercellular interactions controlling anti-tumoral immune responses, alongside protein markers for cell phenotyping and RNA targets for functional markers and cell activation status. In detail, across multiple human FFPE tumor samples, PD-1/PD-L1 interaction was detected in combination with multiple proteins, for immune and stromal phenotyping, and RNA transcripts responsible for the expression of key secreted molecules like cytokines and chemokines. Furthermore, we showed that multiple iterative cycles allow the detection of more than one PPI on the same FFPE section. 

Tumor progression, immune evasion, and therapy resistance are not driven by single molecules but by complex interaction networks among proteins, signaling pathways, and cellular types. This automated spatial multiomics workflow incorporating protein–protein proximity as a third dimension, can help reveal how these processes are regulated. By combining RNA, protein, and proximity data, the assay offers a powerful approach to support biomarker discovery and improved patient stratification. 

Background
The use of multiplex immunofluorescence (mIF) to study the tumor microenvironment (TME) has significantly advanced our understanding of spatial dynamics within tumors. This technique has emerged as a valuable tool for identifying biomarkers and therapeutic targets. Despite its growing adoption, mIF protocols remain complex and technically demanding. Their manual execution and reliance on dedicated reagents make them time-consuming and expensive. Additionally, concerns persist regarding their reproducibility and transferability across different tissue types. Ready-to-use, validated antibody panels, such as the SPYRE™ Core Panels, help address these challenges. 

Methods
In this study, we demonstrate the development and validation of two new antibody panels covering relevant stromal and vessel biomarkers to enable spatial analysis of the TME on the COMET™ platform across various tissue types as well as two additional antibodies against SOX10 and S100B optimized for melanoma studies. The stroma panel enables simultaneous detection of Vimentin, E-Cadherin, Collagen I, and FAP, while the vessel panel contains CD31, CD34, Podoplanin, and LYVE-1. Formalin-fixed paraffin-embedded human tissue sections from a 24-cores multi-organ tissue microarray and whole-section melanoma samples were stained on COMET™ by fully automated sequential immunofluorescence (seqIF™, PMID: 37813886). Staining and detection are done via indirect immunofluorescence using unlabeled primary antibodies and fluorophore conjugated secondary antibodies. Both panels were developed and validated on several tumoral and non-tumoral tissues at the same time. The sections retrieved from COMET™ after seqIF™, were stained by a histology facility with standard immunohistochemistry (IHC) established for pathological diagnosis to compare seqIF™ and IHC staining patterns and verify antibody specificity. All markers demonstrate accurate detection with specific seqIF™ staining, comparable to gold-standard IHC counterparts, as well as robust performance across multiple tissues. Protocols were optimized to achieve high staining quality for all ten markers in terms of sensitivity and signal-to-background ratio. The repeatability and reproducibility of the automated stainings on the COMET™ platform was verified by day-to-day tests on one instrument and tests among multiple ones.  

Results
Our validated SPYRE™ Stroma and Vessel Focus panels, along with melanoma-specific antibodies, deliver highly specific and reproducible results across diverse tissues. Ready-to-use on the COMET™ platform and designed as modular extensions of the SPYRE™ Core Panels, they enable quantitative marker detection, combination with custom antibodies, and empower researchers with robust and scalable workflows for advanced spatial biology studies. 

Advances in precision medicine utilizing antisense oligonucleotides (ASOs) and small interfering RNAs (siRNAs) as therapeutic platforms have brought promising solutions for various neurodegenerative/neuromuscular disorders and rare diseases. Currently, 20 oligonucleotide drug products have been commercially approved by the FDA and EMA, and many more are in clinical phase I-III trials. The FDA has issued recommendations to generate nonclinical biodistribution (BD) data for gene therapy products to evaluate and interpret nonclinical pharmacology and toxicology findings before initiating human clinical trials. In situ hybridization (ISH) and immunohistochemistry (IHC) are increasingly used to spatially visualize the delivered therapeutic, target gene, transgene, and/or cell markers and complement information gathered from molecular technologies such as quantitative polymerase chain reaction (qPCR) and digital PCR. Direct visualization of oligonucleotides can also monitor the risk of off-target events by studying BD of potential therapies in various organs. A new RNAscope™ ISH assay enabling the detection of ASOs, siRNAs, microRNAs (miRNAs), and other small RNAs was developed on the Roche DISCOVERY™ ULTRA System. This fully automated assay allows the co-detection of small RNAs, RNAs and proteins within the same formalin-fixed paraffin-embedded sections of tissues. This workflow includes protease-free RNA detection to prevent disruption of protease-sensitive epitopes. Multiple RNA species and protein targets can be visualized either chromogenically or fluorescently at the single-cell level, leveraging translucent chromogens or TSA-fluorophores. We investigated the expression profile of miR-21, a microRNA implicated in cancer proliferation and progression, in the human cancer TMA tissue. MicroRNA detection was combined with PTEN RNA to assess the impact of miR21 on the expression of PTEN, a tumor suppressor that regulates cell growth, proliferation, and survival. Ki67, and CD31 proteins were co-detected in the same sample to correlate with tumor cell proliferation, and neo-angiogenesis, respectively. This new automated ISH assay showed high sensitivity and specificity for the detection of small RNAs with different expression profiles, and subcellular resolution in intact tissue context. This novel assay will be particularly valuable for the study of ASO/siRNAs delivery, biodistribution, stability, and expression profile of their associated RNA targets for the development of new oligonucleotide therapeutics.

Immunohistochemistry (IHC) is a widely utilized tool for visualizing biomarker expression in tissues of interest. However, not all biomarkers are amenable to investigation by IHC, including secreted factors, such as cytokines. In these cases, in situ hybridization (ISH) may be a suitable alternative. The combination of IHC and ISH approaches allows for investigation of protein and RNA targets in the same multiomic assay, offering greater utility and flexibility than traditional assays. Furthermore, image and spatial analyses can be applied to elucidate the colocalization, proximity, and spatial relationship of RNA and protein biomarkers in the same sample. Flagship Biosciences, in collaboration with Advanced Cell Diagnostics (ACD), is developing a multiplex IHC/ISH fluorescent, which aims to identify and characterize subsets of T lymphocytes in the context of cancer. Using the RNAscope™ Multiomic LS assay we detected CD8, CD3, and Ki67 proteins alongside IL4, IFNG, and GZMB RNA expression to assess the activation and functional state of different T cell subsets based on biomarker expression profiles. Here, we demonstrate the capacity of this analysis on a set of FFPE lung samples from healthy and non-small cell lung cancer subjects. Our approach and analysis highlights the potential of spatial multiomics assays to bridge immunobiology with cancer research. Data generated could prove useful for phenotyping tumors, monitoring therapy response, or better understanding immune cell dynamics within the tumor microenvironment.

Spatial proteomics (highly multiplexed tissue imaging) provides unprecedented insight into the types, states, and spatial organization of cells within preserved tissue environments. To enable single-cell analysis, high-plex images are typically segmented using algorithms that assign marker signals to individual cells. However, conventional segmentation is often imprecise and susceptible to signal spillover between adjacent cells, interfering with accurate cell type identification. Segmentation-based methods also fail to capture the morphological detail that histopathologists rely on for disease diagnosis and staging. Here, we present a method that combines unsupervised, pixel-level machine learning using autoencoders with traditional segmentation to generate single-cell data that captures information on protein abundance, morphology, and local neighborhood in a manner analogous to human experts while overcoming the problem of signal spillover. The result is a more accurate and nuanced characterization of cell types and states than segmentation-based analysis alone. We demonstrate the generality of this technique by applying it to a range of whole-slide, highly multiplexed human tissues acquired using platforms such as cyclic immunofluorescence (CyCIF), Lunaphore COMET, and Akoya PhenoCycler, and show that it can learn histological features across multiple spatial scales.

Background: Lung cancer accounts for 12% of all cancers and 20% of cancer-related deaths. Rapid advances in spatial biology are uncovering mechanisms of disease, patient-to-patient variability, and immune biomarkers predictive of response to immunotherapy. Using a combination of spatial multiomics approaches and data analysis in the same section, we characterized the spatial distribution, phenotypes, and functional states of tumor, immune, and stromal cells between stage IA (early) and IIIA (late) non-small cell lung cancer (NSCLC).
Method: 5 µm formalin-fixed paraffin-embedded (FFPE) tissue sections were prepared from stages IA and III A human NSCLC with confirmed histology of adenocarcinoma. Same sections were analyzed using two complementary platforms, (1) multiplex immunofluorescence (mIF) with a 30-antibody panel on Lunaphore COMET system and (2) spatial transcriptomics using 10x Genomics Visium CytAssist. Protein-level data were processed using HORIZON, and transcriptomic data were analyzed in Loupe Browser. Unbiased clustering (Leiden) and cellular phenotyping were performed to identify region-specific cell populations and marker co-expression patterns.
Results: In early NSCLC tumor, CD8+, CD4+ T cells were more concentrated within the tumor core. In addition, fewer Tregs, immunosuppressive macrophages, and cells expressing immune checkpoint were observed. There were less stromal remodeling, fibrosis and angiogenesis and more TLSs in stage IA tumor. On the other end, late NSCLC tumor showed an increased expression of markers associated with immunosuppression/evasion and stromal remodeling, including an increase in Treg cells, TAMs, CAFs and immune checkpoint expression. A difference in the gene expression signature between early and late-stage NSCLC support observations from protein expression. Early-stage tumor was enriched for genes associated with immune surveillance and active cytotoxic response, whereas late-stage tumor was enriched for immune suppression and stromal remodeling genes. Spatial analysis also highlighted changes in various chemokines and cytokines between the two tumor types.
Conclusions: The integration of multiomics technologies on the same section helped identify diverse processes and transcriptional changes associated with a varied tumor and immune landscape between early and late-stage NSCLC. Cellular phenotyping triaged cell types based on marker expression in pathologist annotated regions. These findings highlight intra and inter-tumor changes, the biology within and helped with the identification of potential biomarkers for immunotherapy response and intervention.

Background: Radiation resistance remains a critical barrier to enhancing cures after chemoradiation therapy (CRT) in esophageal adenocarcinoma (EAC). Ferroptosis, a lipid peroxidation-driven cell death pathway, is increasingly recognized as a regulator of tumor-immune interactions. We investigated how ferroptotic susceptibility of macrophages might differ in radiation responders (GR) and non-responders (NR) in EAC patients using both single cell sequencing and spatial proteomic analysis.
Methods: scRNA-seq and sequential multiplex-IF using Lunaphore COMET were performed on patient biopsies before, during and after CRT to characterize cellular subsets and ferroptosis markers at single-cell resolution. In vitro ferroptosis assays were performed in THP-1, and human PBMC-derived M1/M2 macrophages following 0-12 Gy radiation ± RSL3 or ferrostatin. Ferroptotic activity was assessed by CellTiter-Glo, BODIPY-C11 oxidation, and GPX4 staining (Confocal, flowcytometry).
Results: scRNA-seq confirmed robust myeloid cell expansion in both GR and NR and revealed that pro-ferroptosis gene programs (BH4 biosynthesis, iron utilization and glycolysis pathway) were upregulated in NR both at baseline and during CRT, particularly in the myeloid population, which greatly expands during CRT. Conversely, anti-ferroptosis pathways (GPX4) were enriched in GR and suppressed in NR. COMET profiling revealed that non-responders (NR) displayed higher baseline M2 macrophage density and significantly elevated ferroptosis marker 4-HNE in both M1 and M2 subsets. During CRT, NRs exhibited a further increase in M2 macrophage infiltration, reinforcing an immunosuppressive TME, whereas good responders (GR) maintained higher M1 representation. In vitro, radiation significantly potentiated RSL3-induced ferroptosis in both human and murine macrophage-like cells; ferrostatin rescued viability, confirming ferroptotic cell death. Human macrophages displayed polarization-dependent susceptibility: M2 macrophages were sensitive to radiation-induced ferroptosis, while M1 macrophages were resistant
Conclusion: CRT response appears to relate to macrophage polarization and ferroptosis susceptibility. Non-responders are characterized by (i) M2-dominant macrophage landscapes, (ii) elevated lipid peroxidation, (iii) expansion of ferroptosis-primed myeloid subsets, and (iv) persistent activation of pro-ferroptosis pathways. Collectively this data ascribes macrophage ferroptosis as a potential targetable axis for radio-sensitization in EAC.

High-resolution tissue profiling increasingly relies on integrated spatial multiomic approaches that unify spatial transcriptomics and antibody-based proteomics to reveal coordinated molecular patterns within complex tissues. This enables a detailed exploration of spatial niches, cell-cell interactions, and tissue microenvironments. However, these modalities are often performed on consecutive sections, limiting precise correlation between molecular and spatial features. Here, we present optimized workflows that combine high-plex imaging and sequencing-based spatial transcriptomic assays with antibody-based proteomics from the same tissue section in a coordinated and customizable manner.
We developed and evaluated experimental adaptations to ensure high data quality and optimal tissue handling across multiple platforms, including Xenium, Visium, and COMET. Quality control procedures were implemented to assess antigen retrieval compatibility, as these conditions can be antibody dependent. We examined the balance between epitope exposure, tissue integrity, and background signal, providing specific recommendations tailored to different research objectives.
We further compared the sensitivity of these technologies and offer guidance on selecting and combining commercially available transcriptomic and proteomic workflows in a controlled, flexible setup. As in all multiomic approaches, signal loss can occur, particularly in the second readout of consecutive analyses. For proteomics, photobleaching and antigen retrieval are key considerations, especially for low-abundance or difficult-to-detect targets. Transcriptomic data can be enhanced by using HiPlex RNAscope Pro on COMET to detect lowly expressed transcripts. For data integration, we employed a straightforward pipeline that includes cell segmentation based on protein data, image registration using nuclear staining and/or segmentation masks, and extraction of single-cell transcript counts and pixel intensity data for downstream analyses within the SpatialData framework.
We applied these workflows to tonsil, skin, and colon tissues using immuno-oncology-focused panels. The combined approach improved molecular resolution and reduced data sparsity, enabling more precise definition of cell states, spatial neighborhoods, and functional niches. These spatial multiomics workflows expand the analytical capabilities and facilitate deeper biological interpretation across diverse tissue contexts.

Hepatocellular carcinoma (HCC) is the primary form of liver cancer, because early symptoms may be absent, it is often detected late, resulting in a median survival of less than a year. Metabolically-dysfunction-associated steatotic liver disease (MASLD), previously known as non-alcoholic fatty liver disease (NAFLD), has been identified as the fastest-growing cause of HCC globally, including in the United States. Vitellius et al. reported that MASLD may account for 35% of HCC cases. Furthermore, Younossi et al. mentioned it is estimated that by 2040, MASLD will be present in 55% of the population. Therefore, understanding the molecular mechanisms by which MASLD progresses to HCC is critical.
In this study, we leveraged the existing whole-transcriptomic data from over 200 patients with MASLD or Metabolic Dysfunction-Associated Steatohepatitis (MASH) to identify all differentially expressed transcripts, compared to normal liver tissue. We examined the expression of unique noncoding transcripts that are associated with metabolism in HCC. To explore their potential functions, we used the Genomic Regions Enrichment of Annotations Tool (GREAT). Currently, we are employing various molecular and phenotypic based assays to identify their function in HCC. We will link their functions to HCC patient outcomes using RNA in situ hybridization in human tissues.
The analysis identified differentially expressed transcripts in MASLD and MASH. We observed overlapping gene expression patterns across both conditions, along with distinct transcriptomic profiles unique to each stage. In MASLD, 4,403 genes were significantly upregulated and 4,411 downregulated compared to controls. Similarly, in MASH, 4,702 genes were upregulated and 4,433 downregulated. Comparative analyses revealed substantial overlap between the two cohorts, while also highlighting transcripts unique to each condition. These findings suggest that as metabolic liver disease progresses from fatty liver to steatohepatitis, many transcriptional alterations are conserved and become increasingly dysregulated. GREAT analyses indicate that these may be linked to various cellular processes known to play critical role in HCC, such as long-chain fatty-acyl-CoA and fatty-acyl-CoA metabolism, cellular catabolism, apoptosis, and kidney development, including mesonephros. Notably, many of these transcripts exhibit distinct expression patterns in HCC, suggesting their potential as diagnostic and therapeutic markers. We have begun elucidating their mechanism of action using loss- and gain-of approaches in HCC and how their expression is regulated in patient tissue samples.

Ewing sarcoma (ES) is a pediatric cancer of the bone and soft tissues with poor outcomes for patients with metastatic or relapsed disease. Ewing sarcoma cells are characterized by the presence of a driver fusion oncogene, most commonly EWSR1::FLI1. In the absence of a genetic animal model due to the severe toxicity of the oncofusion, the developmental aspects of ES initiation, including its cellular origin, have remained poorly understood. To address these questions, we developed a stable zebrafish transgenic model enabling tissue-specific expression of the human EWSR1::FLI1 oncofusion in neural crest cells, one of the proposed cell of origin for ES (Vasileva et al., Cell Reports 2025). Using this model, we demonstrated that expression of human EWSR1::FLI1 oncofusion in neural crest cells can lead to their transformation and the development of tumors in vivo. Single-cell analysis of tumor initiation shows that EWSR1::FLI1 reprograms neural crest-derived cells to a mesoderm-like state, strikingly resulting in ectopic fin formation throughout the body. Such hijacking of the limb development program led to abnormal activation of developmental signaling pathways in EWSR1::FLI1-induced outgrowths, resulting in dysregulation of the FGF signaling cascade and HOX gene expression. EWSR1::FLI1 reprograms neural crest cells by hijacking developmental enhancers and upregulating the expression of mesodermal regulators. One such regulator is tbxta (Brachyury or T), a key transcription factor controlling mesodermal specification. Notably, tbxta/TBXT expression was maintained in a subset of zebrafish and human tumors. Our model provides a mechanism by which a neural crest cell lineage can be transformed into Ewing sarcoma, a malignancy with predominant mesenchymal features. Taken together, these findings show how a single mutation can disrupt normal developmental trajectories, driving neural crest cells reprogramming and initiating malignant transformation.

Ovarian cancer can be subdivided into different histologic types. Among them, clear cell ovarian cancer (CCOC), which constitutes 8% of ovarian cancer, differs from the other types with respect to its clinical characteristics. Most of CCOC frequently presents at an early stage compared to high-grade serous ovarian cancer (HGSOC), with most cases diagnosed at Stage I or II, which offers a favorable prognosis. However, those diagnosed with advanced disease experience poorer clinical outcomes compared to those with HGSOC, since CCOC is usually more resistant to systemic chemotherapy than other types. Despite multiple studies showing promise of immune checkpoint inhibitors (ICIs) treatment in patients with CCOC, the molecular mechanisms by which CCOC confers improved response to ICIs, and biomarkers that can predict treatment response to these ICIs in CCOC have not been thoroughly explored. In addition, it remains unclear whether the immune microenvironment plays a role in the early presentation and in metastatic potential of CCOC.Recent spatial transcriptomics (ST) analyses demonstrated that increased AOC1 expression was found in the epithelial cell cluster of early stage CCOC than in HGSOC, which was subsequently validated by sequential immunofluorescence (seqIF) analysis on 17 CCOC and 34 HGSOC patient samples. Increased AOC1 expression is associated with improved overall survival in HGSOC patients. Functional studies showed that despite the lack of a direct effect on the growth of ovarian cancer (OC) cells, syngeneic mouse cells transfected with full-length AOC1 had significantly lower tumor burden than the control mice, suggesting that the tumor microenvironment (TME) mediates the effect of AOC1 on tumor growth. Integrating ST and mass spectrometry imaging (MSI) revealed significantly inverse correlation between AOC1, and histamine and cell membrane VISTA expression levels in cancer cells and/or macrophages in the TME of CCOC and HGSOC, which was confirmed by seqIF. These findings suggest that histamine and VISTA mediate the tumor suppressive effect of AOC1. Indeed, our in vitro studies demonstrated that AOC1 abrogates the growth promoting effect of histamine in OC cells expressing high levels of histamine receptor HRH1, and AOC1 attenuates histamine induced OC proliferation and enhance T cell-mediated anti-tumor immunity via the histamine/HRH1/VISTA axis. Further studies demonstrated that in addition to VISTA, histamine can upregulate PD-L1 in both macrophages and OC cells. AOC1 may increase immune surveillance through attenuating histamine-induced VISTA and PD-L1 expression in the OC TME. Studies to further delineate the immune modulation role of AOC1 and exploring whether enhancing circulating AOC1 levels or targeting histamine with repurposed drugs to improve the efficacy of ICIs as a new strategy in the treatment of OC patients are warranted.

Background: Mucosal melanoma (MuM) is a rare melanoma subtype, accounting for only 1-2% of all melanoma cases, yet is highly aggressive, demonstrating poorer responsiveness to immune checkpoint blockade than the more common cutaneous melanoma. However, the biological mechanisms driving therapeutic resistance in MuM remain poorly understood. This study aims to uncover spatial and molecular mechanisms underlying response and resistance of MuM to immune checkpoint blockade therapies.
Study Design and Methods: An integrative spatial multi-omics framework was applied, which combined single-cell spatial proteomics using COMET platform with high mass resolution spatial metabolomics using imaging mass spectrometry (MALDI-IMS). We comprehensively profiled cellular compositions and spatially defined cellular neighborhoods (CNs) across 97 FFPE tissue cores from 26 MuM patients treated with PD-1/PD-L1 or CTLA-4 inhibitors. Spatial organization, cell-cell interactions, proteomic profiles and metabolomic features of CNs were further compared across responders and non-responders.
Results: The comprehensive approach enabled spatially resolved profiling of 695,444 single cells, which were categorized into 25 cellular phenotypes spanning 9 major cell lineages. Spatial cellar neighborhood analysis revealed 15 biologically distinct CNs that differed in their composition and spatial distribution of tumor, immune, and stromal cell compartments. In patients who responded to immunotherapy, three tumor-associated CNs—the central tumor, invasive tumor, and tumor boundary CNs—were significantly enriched and collectively formed unique spatial organization patterns. Notably, the invasive tumor CN and tumor boundary CN were characterized by spatial proximity among Ki67⁺ tumor cells, CD163⁺ macrophages, and CD11c⁺ dendritic cells. These CD163⁺ macrophages exhibited reduced expression of IRF4 and Arg1, consistent with lower immunosuppressive activity. Conversely, non-responders exhibited a stromal-dominant CN composed primarily of SMA^– stromal cells and demonstrated reduced immune infiltration in both pre-treatment and post-treatment samples. Spatial metabolomic profiling further revealed a pronounced reduction of tryptophan-derived indole metabolites in responders, which significantly correlated with CD11c and CD163 expression, indicating coordinated immunometabolic remodeling within the tumor microenvironment.
Conclusions: These findings highlight that spatial tumor-immune architecture, stromal exclusion, and metabolic rewiring collectively shape immunotherapy response in MuM. The identified spatially resolved tryptophan-derived metabolite signatures offer promising biomarkers and potential therapeutic targets to improve treatment outcomes in this clinically challenging melanoma subtype.

Background: Immunosuppression is a key characteristic of pancreatic ductal adenocarcinoma (PDAC), contributing to metastasis and poor survival. Our studies have identified tumor cell intrinsic Rho-associated coiled-coil containing protein kinase-2 (ROCK2) as a key regulator of extracellular matrix remodeling. In this study, we investigated how ROCK2 regulates immunosuppression in PDAC by modulating Leukemia inhibitory factor (LIF) and its effects on STAT3.
Methods: TCGA PDAC patient dataset was used to compare the ROCK2 and LIF expression in normal and PDAC tissues. CIBERSORT analysis of the PDAC dataset estimated the proportion of tumor infiltrating immune cell subsets. Genomic editing using the CRISPR/Cas9-system in LSL-Kras^G12D/+; Trp53 ^R172H/+; Pdx1^Cre/+ (KPC) cells was performed to generate KPC Rock2 knockout (Rock2^KO) cells. Cytokine array was performed on Rock2^EV and Rock2^KO conditioned media and ELISA was used to validate the results. Flow cytometry was used to profile LIF receptor (LIFR) expression in different cell types from KPC orthotopic tumors. Bone marrow-derived macrophages from C57BL/6 mice were treated with recombinant LIF (rLIF) and LIFR inhibitor (EC359), then analyzed for polarization by flow cytometry. KPC orthotopic tumors were generated, and Immune cell profiling was performed to evaluate alterations in immune cell subsets following treatment with EC359. Findings from ROCK2 and ROCK2-regulated LIF-STAT3 targeting, both in vitro and in vivo were validated using Immunoblotting and immunohistochemistry.
Results: Analysis of the TCGA dataset revealed that Human PDAC tissues have increased expression and correlation of ROCK2 and LIF. Further analysis showed that macrophages constitute a substantial proportion of the immune cell infiltrate. Cytokine array and ELISA-based studies revealed decreased LIF secretion with ROCK2 knockout, providing evidence for ROCK2 dependent regulation of LIF. KPC orthotopic tumors demonstrated higher LIFR expression in tumor-associated macrophages (TAMs) and EC359 treatment reduced ARG1 and PD-L1 expression on these cells. Additionally, EC359 treatment led to a significant increase in the activated effector and effector memory T cell populations. Furthermore, rLIF treatment increased pSTAT3 levels in macrophages, while EC359 lowered its expression, highlighting the role of ROCK2-regulated LIF-LIFR in STAT3 activation.
Conclusion: These findings demonstrate that tumor cell-intrinsic ROCK2 regulates LIF-STAT3, which mediates immunosuppression and can serve as a potential therapeutic target for PDAC.