Human 3D Tumor Assembloids Reveal How Targeted Interferon-Gamma Therapies Could Outsmart the Tumor Microenvironment

Colorectal cancer remains one of the most stubborn targets in oncology. While immune checkpoint inhibitors have transformed treatment for patients with microsatellite instability-high tumors, the majority of colorectal cancer patients with microsatellite stable disease respond poorly to current immunotherapies. This therapeutic gap has pushed researchers toward next-generation strategies, including engineered cytokines, bispecific antibodies, and adoptive cell therapies. Yet developing these sophisticated treatments demands preclinical models that faithfully reproduce the human tumor microenvironment, something that traditional two-dimensional cultures, animal models, and even patient-derived xenografts have consistently failed to deliver. Now, a team of researchers at Roche Pharma Research and Early Development has applied and validated a human 3D immune-containing assembloid platform that brings together the epithelial, stromal, and immune compartments of colorectal cancer in a single physiologically relevant construct, offering a powerful new tool for evaluating targeted cytokine therapeutics.

The assembloid model, built upon a foundational shaking organoid co-culture approach, integrates patient-derived colorectal cancer organoids, autologous cancer-associated fibroblasts, and allogeneic peripheral blood mononuclear cells within a collagen matrix. Using a hydrogel-in-oil emulsion system, the researchers generated compact spherical structures where all three cell types are spatially organized to mimic the cellular heterogeneity and architecture of the native tumor microenvironment. Whole-mount multiplex imaging and immunohistological analysis confirmed successful co-engraftment, revealing large E-cadherin-positive epithelial cancer organoid structures embedded within a dense network of Vimentin-positive fibroblasts. Importantly, while CD8-positive and CD4-positive T lymphocytes and CD68-positive macrophages successfully infiltrated tumor-dense areas, the majority of immune cells preferentially localized within stroma-rich regions, reflecting the structural complexity found in real tumors.

Before deploying the platform for drug evaluation, the team performed a careful scaling analysis to determine the optimal balance between tissue complexity and metabolic stability. By proportionally scaling both construct volume and absolute cell number, doubling the input for 10 microliter droplets and quadrupling it for 20 microliter droplets, they observed significant size-dependent effects on tissue health. While larger assembloids successfully increased total tissue mass, the expanded physical dimensions resulted in prominent necrotic cores and morphological shifts in the fibroblast compartment, likely due to diffusion-limited oxygen and nutrient gradients. The researchers therefore selected the baseline 5 microliter architecture, which is sufficient to capture translatable features of tumor hypoxia while preserving optimal cellular health and morphology for evaluating specific stromal-immune responses.

Single-cell RNA sequencing on the assembloids identified distinct cell clusters corresponding to all input populations. The researchers successfully detected colorectal cancer cells robustly expressing epithelial markers such as EPCAM, KRT8, and KRT19, fibroblasts defined by mesenchymal and cancer-associated fibroblast genes including COL1A1, COL3A1, and FAP, alongside a highly diverse immune compartment encompassing CD4-positive T cells, CD8-positive T cells, B cells, plasma cells, natural killer cells, macrophages, and conventional dendritic cells. Assessment across multiple untreated patient-derived lines demonstrated the model’s ability to capture inherent inter-patient heterogeneity within both the immune and stromal compartments, a critical feature for a platform intended to evaluate personalized therapeutic responses.

With the baseline multicellular complexity established, the team evaluated the assembloids’ capacity to respond to interferon-gamma, a master regulator of cell-mediated immunity known for its potential to orchestrate anti-tumor responses and remodel the tumor stroma. Histological assessment via hematoxylin and eosin staining revealed that interferon-gamma treatment resulted in significant growth suppression, evidenced by a marked reduction in overall tissue area. A significant reduction in proliferation was also observed, quantified by the reduced percentage of Ki67-positive tissue area. As expected, a dose-dependent increase in CXCL10 levels was detected in the conditioned media following treatment, with patient-specific differences in secretion highlighting variability in how different patient-derived assembloids respond to the cytokine.

To investigate the functional relevance of CXCL10 secretion, the researchers designed a transwell migration assay placing two assembloids generated from the same patient material in the lower chamber with or without treatment, and CMFDA-labeled naive CD8-positive T cells in the upper compartment. Analysis confirmed that conditioned media from interferon-gamma-treated assembloids significantly enhanced CD8-positive T-cell migration compared to untreated controls. Broader multiplex cytokine profiling confirmed the transition to a highly inflamed microenvironment, characterized by significant elevations in pro-inflammatory cytokines including IL-1α, TNFα, IL-12, and CCL2. Immunohistochemical analysis further confirmed that interferon-gamma treatment potently drove widespread upregulation of both MHC class I and the immune checkpoint molecule PD-L1, predominantly in stromal regions, capturing the cytokine’s paradoxical dual role as both a driver of anti-tumor immunity and a trigger of adaptive immune resistance.

Single-cell transcriptomics then revealed the cell-type-specific mechanics of interferon-gamma signaling. Expression analysis confirmed broad expression of the interferon-gamma receptors IFNgR1 and IFNgR2 across the assembloid compartments, with the highest enrichment in macrophages, fibroblasts, dendritic cells, and colorectal cancer cells. Upon treatment, these clusters exhibited robust upregulation of key T-cell attracting chemokines, namely CXCL9, CXCL10, and CXCL11, with fibroblasts and macrophages showing the most prominent induction. In parallel, interferon-gamma stimulation strongly induced expression of critical immune checkpoint and immunomodulatory molecules, including CD274 (PD-L1) and IDO1, primarily within the macrophage, fibroblast, dendritic cell, and colorectal cancer cell populations, indicating a prominent role for the stroma in mediating immune suppression. Quantification of both pro- and anti-tumor transcriptomic changes identified fibroblasts as the most highly responsive cell population, followed closely by macrophages, dendritic cells, and colorectal cancer cells.

Perhaps the most striking finding emerged from analysis of the stromal compartment. Module scoring demonstrated that interferon-gamma exposure drives a distinct phenotypic transition within the fibroblast population, shifting away from a classical myofibroblast baseline state toward an antigen-presenting cancer-associated fibroblast state. This transition was characterized by robust upregulation of MHC class II molecules such as HLA-DRA, HLA-DRB1, and CD74, alongside key transcription factors including IRF1, STAT1, and CIITA, and immune-modulatory genes such as GBP1, IDO1, and ICAM1. Conversely, structural myofibroblast markers like COL1A1 and COL1A2 were downregulated. Multiplex immunofluorescence staining for Vimentin and HLA-DR validated a consistent increase in the proportion of Vimentin-positive HLA-DR-positive antigen-presenting fibroblasts within the total cancer-associated fibroblast population following treatment, confirming the transcriptomic shift at the protein level.

Given that the clinical application of recombinant interferon-gamma has historically been severely limited by peripheral dose-limiting toxicities and the induction of broad immunosuppressive feedback loops, the team designed a novel panel of attenuated, targeted bispecific molecules intended to localize interferon-gamma receptor activation specifically to the tumor stroma. They utilized fibroblast activation protein, a marker predominantly overexpressed on cancer-associated fibroblasts, as a tumor-enriching target. The researchers engineered constructs with varying FAP binders and VHH domains that bind IFNgR1 or IFNgR2, exploring distinct spatial mechanisms of receptor activation: cis activity, where the molecule binds FAP and activates the interferon-gamma receptor on the surface of the same cell, and trans activity, where the molecule anchors to FAP on a fibroblast but bridges across to activate the receptor on a neighboring cell. They also developed an all-in-one construct integrating multiple anti-FAP binders and anti-IFNgR VHHs into a single unified format capable of engaging both targets simultaneously.

Functional evaluation within the 3D assembloids demonstrated that localizing interferon-gamma activity via FAP successfully uncouples beneficial anti-tumor mechanisms from detrimental immunosuppressive feedback loops. While standard recombinant interferon-gamma provoked massive upregulation of both MHC class I and PD-L1, the cis, trans, and all-in-one targeted molecules elicited minimal PD-L1 upregulation. The targeted constructs still successfully induced CXCL10 secretion at attenuated levels, demonstrating therapeutic potential, and transwell migration assays showed a trend toward enhanced CD8-positive T-cell recruitment, though these changes did not achieve statistical significance over untreated controls at the tested dose. Flow cytometric analysis in two-dimensional co-cultures revealed that exposure to the interferon-gamma variants induced dose-dependent upregulation of CD69 on both CD4-positive and CD8-positive T cells, providing evidence of early immune activation alongside chemotactic signaling. Collectively, these findings confirm that patient-derived assembloids provide a robust, physiologically relevant model to assess complex cellular interactions and screen the localized efficacy of novel, next-generation immunotherapies, paving the way for a more refined, human-derived in vitro strategy to accelerate the discovery and optimization of targeted cancer immunotherapies for solid tumors.

Subject of Research: Cancer immunotherapy preclinical modeling using human 3D assembloids

Article Title: Evaluating cytokine-based therapeutics in human 3D in vitro immune-containing assembloids

Article References: Maenhoudt, N., Lütge, M., Obenloch, M., Caironi, A., Kiratzis, M., Jean-Mairet, P., Graziano, S., Fröbel, P., Hüsser, T., Yángüez, E., Antsiferova, M., Filip, A. M., Müller, C., Leclair, S., Gasser, S., Umaña, P., Colombetti, S., Cremasco, F., & Calabro, S. (2026). Evaluating cytokine-based therapeutics in human 3D in vitro immune-containing assembloids. Frontiers in Cell and Developmental Biology, 14, Article 1930651. https://doi.org/10.3389/fcell.2026.1930651

Image Credits: AI Generated

DOI: 10.3389/fcell.2026.1930651

Keywords: assembloids, cancer immunotherapy, colorectal cancer, interferon-gamma, cancer-associated fibroblasts, tumor microenvironment, organoids, targeted cytokines, Evaluating, cytokine-based, therapeutics, human