No tumor is an island. In the past two decades, it’s become increasingly appreciated that cancers are dependent on neighboring cells for their growth and survival. These tumor-friendly communities—or tumor microenvironments—often have stroma or connective tissues that support tumor growth and keep antagonists like chemotherapies or immune cells out of the neighborhood.
The cells primarily responsible for this cancer-supporting extracellular matrix (ECM) are cancer-associated fibroblasts (CAFs), which secrete collagens and fibers, cytokines to dampen the immune response and promote blood vessel formation, and growth factors that help the tumor proliferate.
Conventional drug screening often uses monocultures of tumor cells grown flat on a petri dish. This allows researchers to look for drugs that kill fast-growing tumor cells but lacks the complexity of the environment which allows the tumor to survive and expand within the body.
This is an issue for Taran Gujral, whose lab in the Human Biology Division seeks to understand cell-cell communication and how to pharmacologically target signaling networks to improve cancer outcomes. Previous work from his group has developed ways to recapitulate the tumor microenvironment by slicing fresh tumor samples into organotypic cultures—meaning cultures that mimic the cell type heterogeneity and three-dimensional structure of the original tumor. They dub these cultures 3D “microtumors.”
These 3D microtumors have “complex, tissue-like structures that preserve the tumor microenvironment,” says Dr. Gujral. “Unlike traditional 2D cell culture systems, this approach captures the intricate interactions between cancer cells and CAFs, which play a crucial role in tumor growth, drug response, and resistance.”
In a recent study published in Cell Reports Medicine, members of the Gujral Lab performed a drug screen to find drugs that might target the tumor microenvironment. They did this by comparing drug killing in 2D tumor cells lines to their performance on 3D microtumors.
The authors used a panel of 32 kinase inhibitors to test on breast and pancreatic cancer-derived microtumors, which they hoped would reveal shared microenvironment vulnerabilities. They then integrated the results of this screen and other published models to into their machine learning platform built to deconvolve kinase signaling networks. Using this program, they can computationally predict tumor responses to more than 400 kinase inhibitors.
They were surprised to find that two to three times as many drugs were predicted to be effective on 3D microtumors than in 2D culture. “The findings have strong implications for precision oncology, emphasizing that functional drug testing in microtumor models can reveal clinically meaningful vulnerabilities that are missed in conventional assays,” says Dr. Gujral.
One drug that stood out was doramapimod, an inhibitor of mitogen activated protein kinase (MAPK). Many cancers have mutations and alterations in the MAP family of kinases, which form a complex signaling cascade that helps the cell respond to various stimuli. However, doramapimod has never been pursued as cancer therapy (although it has undergone safety testing and human clinical trials for other medical conditions). The authors decided to focus on doramapimod since its role in cancer was not well established.
The authors added doramapimod to microtumor cultures of different cancers including triple-negative breast cancers, pancreatic tumors, hepatocellular carcinomas, and meningiomas. They observed that growth of breast and pancreatic microtumors—which have a fibrous and dense stroma—was inhibited to a greater extent than growth of the other cancers with less typically dense stroma. These results were recapitulated in mouse models of tumor growth: doramapimod inhibited pancreatic or breast cancer growth and weight by about 50% after 2 weeks of treatment.
The authors conducted RNA sequencing and proteomics of doramapimod-treated microtumor slices and found consistent downregulation of ECM proteins like metalloproteases and collagens. Because these are produced by cancer-associated fibroblasts, they hypothesized that doramapimod has an anti-tumor impact by disrupting the ability of CAFs to form a tumor-promoting microenvironment.
They turned to a co-culture assay to test this hypothesis: mixing pure tumor cell lines (not microtumor slices) with CAFs in serum-free medium pushes CAFs to secrete factors that promote tumor cell growth. However, in the presence of doramapimod, CAFs failed to support cancer proliferation. Importantly, doramapimod does not impact viability of the CAFs themselves but impairs their ability to remodel the ECM.
To better understand the mechanism behind this, they performed biochemical assays to find doramapimod’s targets out of a panel of 370 kinases. They found some anticipated members of the MAPK family such as MAPK12, but they also found that doramapimod inhibits other kinases that have not yet been reported including discoidin domain receptors 1 & 2 (DDR1/DDR2).
Through knockdown of doramapimod’s targets, the authors discovered that the MAPK12 and DDR1/2 signaling pathways converge on a transcription factor called glioma-associated oncogene homolog 1 (GLI). As its name implies, it was discovered by its pro-cancer impacts on gliomas, and it was later shown to be important for cell proliferation and tumor progression through signaling via the hedgehog signaling network.
They blocked GLI’s transcriptional activity with a GLI-specific inhibitor, RNA knockdown of MAPK12 and DDR1/2, and doramapimod treatment. All of these methods reduced tumor growth in the CAF co-culture assays. Together, these data indicate that the MAPK12-DDR1/2-GLI axis is important for CAFs to form the extracellular matrix important for tumor growth.
But the ECM is more than a support network; it is also a barrier to anti-tumor therapies. In the final experiment of the paper, the team showed that reduction of the ECM with doramapimod treatment enhanced killing activity of a chemotherapy agent and an immune checkpoint inhibitor. The anti-tumor effect of the drug combinations significantly outperformed any of the three agents alone.
“This work demonstrates the power of 3D microtumor screening using complex, tissue-like structures that preserve the tumor microenvironment,” concludes Dr. Gujral. However, there’s plenty of work still to be done.
“One key question is whether these mechanisms are tumor-type specific or represent a shared vulnerability across multiple solid tumors “Another emerging question is how stromal composition and spatial organization within the tumor microenvironment shape drug sensitivity in ways that can be predicted or targeted therapeutically,” Dr. Gujral says.
Going forward, the lab is excited to “focus on dissecting the molecular and functional diversity of CAFs using 3D microtumor and tissue slice models derived from patient tumors,” Dr. Gujral adds. “Ultimately, this direction will help translate 3D screening approaches into precision oncology frameworks, enabling personalized therapy selection that accounts for both tumor-intrinsic and microenvironmental factors.”
This publication was also featured by the Fred Hutch News Service. Click here to check out their story.
Fred Hutch/University of Washington/Seattle Children’s Cancer Consortium Members Drs. Nancy Davidson, Cyrus Ghajar, Venu Pillarisetty, and Taran Gujral contributed to this research.
The spotlighted research was funded by the National Cancer Institute, Cancer Center Support Grant, Breast Cancer Research Foundation, and Washington Research Foundation.
Nishida-Aoki N, Zhu S, Chan M, Kang Y, Fujita M, Jiang X, McCabe M, Vaz JM, Davidson NE, Ghajar CM, Hansen K, Welm AL, Pillarisetty, VG, & Gujral TS. 2025. Drug screening in 3D microtumors reveals DDR1/2-MAPK12-GLI1 as a vulnerability in cancer-associated fibroblasts. Cell Reports Medicine. https://doi.org/10.1016/j.xcrm.2025.102357
Hannah Lewis is a postdoctoral research fellow with Jim Boonyaratanakornkit’s group in the Vaccine and Infectious Disease Division (VIDD). She is developing screens to find rare B cells that produce protective antibodies against human herpesviruses. She obtained her PhD in molecular and cellular biology from the University of Washington.
For the Media