Mimicking tumor cell heterogeneity of colorectal cancer in a patient-derived organoid-fibroblast model

Velina S. Atanasova, Crhistian de Jesus Cardona, Václav Hejret, Andreas Tiefenbacher, Loan Tran, Carina Binder, Theresia Mair, Julijan Kabiljo, Janik Clement, Katharina Woeran, Barbara Neudert, Markus Hengstschläger, Markus Mitterhauser, Leonhard Müllauer, Boris Tichý, Michael Bergmann, Gabriele Schweikert, Markus Hartl, Helmut Dolznig, Gerda EggerbioRxiv 2022 · DOI

Abstract

Patient-derived organoids (PDOs) are powerful tools for modeling cancer, yet they often lack the complex tumor microenvironment (TME) that drives disease progression. This study presents a colorectal cancer (CRC) model that integrates epithelial tumor cells with patient-matched stromal fibroblasts. By characterizing cancer-associated fibroblasts (CAFs) and normal fibroblasts (NFs), the researchers demonstrate that these stromal cells support organoid growth without exogenous niche factors. Crucially, the co-culture model restores the cellular heterogeneity and morphological characteristics of the original tumor, providing a more physiologically relevant platform for studying CRC mechanisms and therapy response.

Introduction: The Missing Stroma

The shift from 2D cell lines to 3D patient-derived organoids (PDOs) has significantly advanced oncology by preserving the molecular characteristics of individual tumors. However, most standard PDOs are "naked," consisting almost entirely of epithelial tumor cells. In a clinical setting, colorectal tumors exist within a dense neighborhood of immune cells, blood vessels, and fibroblasts.

Fibroblasts, particularly Cancer-Associated Fibroblasts (CAFs), are known to drive tumor growth, promote metastasis, and shield cancer cells from chemotherapy. Because standard PDOs lack these neighbors, they often fail to accurately predict how a patient will respond to treatments like oxaliplatin. This research addresses this gap by creating a "patient-matched" model that includes both the cancer cells and the supporting stromal cells from the same individual.

Methodology and Multi-Omic Validation

The research team developed a 3D co-culture system using cells from patients with moderately differentiated colorectal tumors. Their approach involved several key innovations:

  • Patient-Matched Isolation: Epithelial tumor cells and fibroblasts (both CAFs from the tumor and Normal Fibroblasts from adjacent healthy tissue) were isolated from the same surgical specimens.
  • Niche-Independent Growth: Standard organoids require an expensive "cocktail" of growth factors (such as Wnt3a and EGF). The researchers tested a "minimal medium" (1% serum) to determine if fibroblasts could naturally provide these essential signals.
  • Bioinformatic Deconvolution: To prove the co-culture was more realistic, the team used computational techniques to "deconvolve" bulk RNA-seq data. By comparing their results to published single-cell RNA-seq (scRNA-seq) datasets from CRC tissues (including the SMC and Li et al. datasets), they estimated the exact proportions of cell subtypes, such as stem cells versus mature colonocytes.

Key Insights: Fibroblasts as Signal Factories

A striking finding was that fibroblasts alone could support organoid growth and survival in minimal media. This confirms that the stroma provides the necessary "niche factors" that researchers typically add manually. Proteomic analysis further revealed that CAFs maintain a "constantly activated" state in vitro, exhibiting higher motility and increased secretion of pro-metastatic factors compared to normal fibroblasts.

Restoring Cellular Heterogeneity

In standard expansion media, organoids tend to become uniform and stem-cell-like. Adding fibroblasts restored the natural cellular diversity of the tumor. Histological staining (H&E, PAS, and markers like CK20) showed that co-cultures closely mirrored the architecture and mucus production of the original patient biopsy. The presence of fibroblasts induced the development of specialized cells like goblet cells and enterocytes, which are often missing in mono-cultures.

Mutual Molecular Instruction

The crosstalk between tumor cells and fibroblasts is a two-way street. RNA-seq revealed 1,742 significantly deregulated genes in organoids when grown with fibroblasts, primarily in pathways related to collagen biosynthesis and the innate immune system. Simultaneously, tumor cells "instructed" fibroblasts to adopt pro-inflammatory and matrix-remodeling phenotypes, upregulating specific markers such as WNT2, WNT5a, and VCAM1.

Conclusion and Future Directions

By incorporating the stromal component, this patient-matched model bridges the gap between simplified organoids and complex human tumors. While the model currently lacks immune and endothelial components, it provides a blueprint for the next generation of physiological cancer models. Future work will focus on longer-term cultures and clinical validation to determine if this co-culture model can better predict a patient's actual response to chemotherapy than standard methods, moving us closer to truly personalized oncology.

Summary updated

Abstract

Abstract Patient-derived organoid (PDO) cancer models are generated from epithelial tumor cells. Although they reflect the molecular tumor characteristics, they lack the complexity of the tumor microenvironment, which is a key driver of tumorigenesis and therapy response. Here, we present a colorectal cancer (CRC) organoid model that incorporates epithelial cells and stromal fibroblasts from the same patient. Molecular characterization of primary cancer associated fibroblasts (CAFs) and matched normal fibroblasts (NF) revealed proteomic, secretome and gene expression differences in pathways associated with tumor related fibroblast function. Further, CAFs retained higher motility compared to NFs in vitro . Importantly, both CAFs and NFs supported cancer cell proliferation in 3D co-cultures, without the addition of classical niche factors. PDOs grown together with fibroblasts displayed a larger cellular heterogeneity of tumor cells compared to mono-cultures, and closely resembled the in vivo tumor morphology. This was also confirmed by the calculation of cellular proportions of epithelial cell subtypes in organoid mono-versus co-cultures, which were inferred through bioinformatics deconvolution of bulk RNA sequencing data using published single cell RNA sequencing datasets from CRC tissues. Additionally, we observed a mutual crosstalk between tumor cells and fibroblasts in the co-cultures. This was manifested by majorly deregulated pathways such as cell-cell communication and extracellular matrix remodeling in the organoids. For the fibroblasts, we observed enhanced expression of tumor induced marker genes and cytokines characteristic for myo- and immunogenic fibroblasts. This model will be vital as a physiological personalized tumor model to study disease mechanisms and therapy response in CRC. One Sentence Summary Patient matched fibroblasts support tumor organoid growth in 3D co-culture and maintain intratumoral cellular heterogeneity and histo-morphology.
BibTeX
@misc{atanasova2022mimicking,
  author = {Velina S. Atanasova and Crhistian de Jesus Cardona and Václav Hejret and Andreas Tiefenbacher and Loan Tran and Carina Binder and Theresia Mair and Julijan Kabiljo and Janik Clement and Katharina Woeran and Barbara Neudert and Markus Hengstschläger and Markus Mitterhauser and Leonhard Müllauer and Boris Tichý and Michael Bergmann and Gabriele Schweikert and Markus Hartl and Helmut Dolznig and Gerda Egger},
  title = {{Mimicking tumor cell heterogeneity of colorectal cancer in a patient-derived organoid-fibroblast model}},
  year = {2022},
  doi = {10.1101/2022.03.07.483214},
  howpublished = {bioRxiv},
}
Figure