Tumor Protein CEMIP Hijacks Immune Cells to Fuel Colorectal Cancer Drug Resistance

Colorectal cancer remains one of the most common and lethal malignancies worldwide, and while immune checkpoint blockade has transformed the treatment landscape for many cancers, a large fraction of colorectal tumors stubbornly refuse to respond. Now, a team of researchers in China has uncovered a previously underappreciated molecular conspiracy between tumor cells and the immune cells they surround, offering a fresh explanation for why immunotherapy so often fails in this disease. The study, published in the Journal of Translational Medicine, identifies a protein called CEMIP as a central puppeteer that reprograms tumor-associated macrophages and helps tumors cloak themselves in stolen molecular disguises.

CEMIP, short for cell migration inducing hyaluronan binding protein, has long been associated with aggressive tumor behavior, but its precise role in the immune dynamics of colorectal cancer has remained murky. The new research, led by Qingling Hua, Yuanhang Yu, Dejun Zhang and Jinge Zheng under the senior authorship of Tao Zhang and Lei Zhao at Union Hospital, Tongji Medical College of Huazhong University of Science and Technology, reveals that CEMIP operates on two coordinated fronts. Inside tumor cells, it ramps up the production of extracellular vesicles, tiny membrane-bound packages that cells use to ship molecular cargo to their neighbors. Outside the tumor cell, those same vesicles deliver CEMIP into macrophages, the abundant immune cells that populate the tumor microenvironment, where it sabotages a critical signaling pathway that would otherwise keep the macrophages in an anti-tumor mode.

The technical backbone of the study is impressive in its breadth. The researchers combined syngeneic subcutaneous and orthotopic colorectal cancer mouse models with human tumor specimens, primary human macrophages, cell-line-derived macrophages, and extracellular vesicle preparations purified by differential ultracentrifugation. They layered on RNA sequencing to capture global gene expression changes, flow cytometry to quantify immune cell phenotypes, immunoblotting and co-immunoprecipitation to probe protein interactions, and nanoparticle tracking analysis to count and size the vesicles being released. An exploratory cohort of 71 colorectal cancer patients who had received immunotherapy was then analyzed using prespecified CEMIP immunohistochemistry scoring and multivariable statistical models to ask whether the laboratory findings translated into clinical relevance.

The first major discovery concerns how CEMIP controls vesicle output. The team found that endogenous CEMIP within colorectal cancer cells was associated with elevated cytosolic calcium, a versatile intracellular messenger known to regulate membrane trafficking and vesicle release. Higher calcium levels, in turn, corresponded to increased production of extracellular vesicles. When the researchers manipulated CEMIP expression, they observed corresponding shifts in both calcium signaling and the quantity of vesicles shed by the tumor cells. This establishes CEMIP as an intrinsic regulator of the tumor’s secretory machinery, effectively turning up the volume on the intercellular messaging system that tumors use to influence their surroundings.

The second and arguably more striking discovery involves what happens when those vesicles reach macrophages. Using donor-protein tracing techniques, the researchers showed that vesicle-associated CEMIP is acquired by recipient macrophages, where it competitively interferes with the complex formed between heat shock protein 90, commonly known as HSP90, and Janus kinase 2, or JAK2. This chaperone complex is essential for activating the JAK2/STAT1 signaling cascade, a pathway that macrophages rely on to respond to interferon-gamma and mount inflammatory, tumor-killing behavior. By wedging itself into the HSP90-JAK2 interaction, vesicular CEMIP dampened the phosphorylation of both JAK2 and STAT1, pushing macrophages away from their inflammatory state and toward a CD206-high, CD86-low profile, a phenotype classically associated with immunosuppressive, tumor-promoting activity.

Crucially, the researchers demonstrated that this reprogramming is reversible at the molecular level. When they forced macrophages to overexpress HSP90, the chaperone outcompeted the incoming CEMIP, restoring JAK2/STAT1 pathway activation, reverting the macrophage phenotype, and rescuing downstream effector readouts from CD8-positive T cells, the cytotoxic lymphocytes that carry out much of the immune system’s tumor-killing work. This rescue experiment is more than a mechanistic nicety; it pinpoints the HSP90-JAK2 axis as the functional chokepoint through which CEMIP exerts its influence, and it suggests that bolstering this pathway could counteract the immunosuppressive programming that tumors impose on their macrophage entourage.

The study also tackled a second, parallel mechanism of immune evasion: the horizontal transfer of PD-L1, the molecular brake that tumors use to disable T cells. Using donor cells engineered to express tagged Flag-PD-L1, the researchers detected the tagged protein in the membrane fractions of recipient macrophages, indicating that macrophages can acquire tumor-derived PD-L1 through extracellular vesicles. When the team knocked down PD-L1 in the donor tumor cells, the surface PD-L1 on recipient macrophages diminished accordingly, supporting the acquisition model. The authors are careful to note that while these experiments support vesicle-mediated PD-L1 transfer, the precise intracellular trafficking route the protein follows after delivery has not yet been fully defined. Even so, the implication is significant: macrophages in the tumor microenvironment may be wearing molecular uniforms manufactured by the tumor itself, allowing them to suppress T cells through the very checkpoint pathway that immunotherapy drugs are designed to block.

The clinical correlative data lend weight to the laboratory findings. In the retrospective cohort of 71 immunotherapy-treated colorectal cancer patients, high CEMIP expression in tumor tissue was associated with poorer disease control and shorter progression-free survival. Because the cohort was analyzed with prespecified scoring criteria and multivariable models, the association carries more methodological rigor than many retrospective biomarker studies, though the authors appropriately frame it as exploratory. Larger, prospective validation will be needed before CEMIP can be considered a bona fide predictive biomarker, but the signal is consistent with the mechanistic story: tumors that produce more CEMIP should, in theory, be better at disarming macrophages and acquiring vesicular PD-L1, and therefore harder for checkpoint inhibitors to defeat.

What makes this work particularly compelling is the spatial coordination of the two mechanisms. CEMIP does not merely act within the tumor cell that produces it, nor does it act on macrophages through some diffuse, untraceable route. Instead, the same protein that boosts vesicle production inside tumor cells ends up packaged inside those very vesicles and delivered to macrophages, where it disrupts HSP90-dependent signaling. The tumor cell is thus both the factory and the distributor of its own immunosuppressive weapon, and the vesicle is the delivery vehicle that links the intrinsic and extrinsic arms of the axis. This elegant economy of mechanism, in which a single molecule orchestrates both the supply and the deployment of an immune-evasion payload, may explain why CEMIP-high tumors are so refractory to treatment.

The therapeutic implications are tantalizing, though the road from mechanism to medicine is long. If EV-associated CEMIP is the linchpin connecting tumor secretion to macrophage reprogramming and PD-L1 acquisition, then interventions that block vesicle release, neutralize CEMIP, or stabilize the HSP90-JAK2 complex in macrophages could theoretically restore anti-tumor immunity and sensitize tumors to checkpoint blockade. The HSP90 rescue experiments provide a proof of principle that the pathway can be restored even after CEMIP has done its damage. For now, the study stands as a meticulous dissection of a signaling axis that had been hiding in plain sight, and as a reminder that the battle between tumors and immunotherapy is fought not only at the checkpoint molecules on cell surfaces, but in the microscopic cargo ships that ferry instructions between the cells of the tumor microenvironment.

Subject of Research: CEMIP-driven extracellular vesicle signaling and macrophage reprogramming in colorectal cancer immunotherapy resistance

Article Title: Tumor-derived CEMIP drives macrophage reprogramming and EV-associated PD-L1 acquisition in colorectal cancer

Article References: Hua, Q., Yu, Y., Zhang, D., Zheng, J., Li, L., Zhang, P., Lin, Z., Yu, D., Zhang, J., Zhou, P., Zhang, P., Zhang, T., & Zhao, L. (2026). Tumor-derived CEMIP drives macrophage reprogramming and EV-associated PD-L1 acquisition in colorectal cancer. Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-09047-4

Image Credits: AI Generated

DOI: 10.1186/s12967-026-09047-4

Keywords: colorectal cancer, CEMIP, extracellular vesicles, tumor-associated macrophages, immunotherapy resistance, PD-L1, JAK2/STAT1 signaling, HSP90, tumor microenvironment, immune checkpoint blockade, Tumor-derived, drives