The research, led by Yanjun Ge and Haiyan Zhu of Shanghai First Maternity and Infant Hospital, Tongji University School of Medicine, together with colleagues at Wenzhou Medical University and OriCell Therapeutics, set out to answer a deceptively simple question: what does the tumor cell surface actually look like after MUC1α has been shed? Most MUC1-targeted CAR-T programs have focused on the shed-prone alpha subunit, particularly tumor-associated glycoforms of the variable number tandem repeat region. But when MUC1α is cleaved, it leaves behind a membrane-retained truncated MUC1β stub, and that stub exposes neoepitopes that were previously hidden or inaccessible. These shedding-exposed epitopes have received comparatively little attention from the CAR-T field, and the team suspected they could be exploited.

To find an antibody against truncated MUC1β, the researchers turned to an unusual but increasingly popular source: alpaca immune antibody libraries. Camelid antibodies are prized in antibody engineering because their single-chain variable fragments are compact, stable, and often recognize clefts or unusual conformations that conventional antibodies miss. By screening immune libraries from alpacas, the team isolated a binder specific to the truncated MUC1β form. That single-chain variable fragment became one of the two recognition domains in what the researchers call a biparatopic CAR, a construct carrying two different antigen-binding modules on a single chimeric antigen receptor.

Before building the therapeutic cells, however, the team needed to know whether the target landscape in actual patient tumors justified the design. Using multiplex immunohistochemistry with two antibodies, one recognizing truncated MUC1β and another recognizing tumor-associated MUC1α glycoforms, they mapped the distribution of MUC1 forms in situ in ovarian cancer tissue. The finding was striking: MUC1 commonly exists in a partially shed state, with full-length MUC1 and truncated MUC1β co-occurring within the same tumor tissue. In other words, shedding is not an all-or-nothing event that uniformly strips MUC1α from every cell. Instead, it creates a mosaic in which some cells display intact MUC1, some display only the truncated beta stub, and many display both. Any therapy that recognizes only one of these states would leave gaps in coverage, and those gaps are precisely where antigen escape could begin.

The biparatopic CAR-T cells were designed to close those gaps. In laboratory tests against ovarian cancer cell lines, the dual-target cells efficiently recognized and killed MUC1-positive tumor cells regardless of how the antigen was presented, whether as full-length MUC1 alone, truncated MUC1β alone, or both forms together. Monospecific CAR-T cells, by contrast, killed only those target cells displaying their corresponding epitope, leaving the rest untouched. This broader target coverage is the core conceptual advance of the study: rather than betting everything on a single epitope that shedding can erase, the biparatopic design hedges across both the intact and the shed forms of the antigen, so that the very act of MUC1α shedding creates a new target rather than destroying an old one.

The data also revealed a quantitative benefit beyond simple coverage. Against tumor cells co-expressing full-length MUC1 and truncated MUC1β, the biparatopic CAR-T cells showed higher cytotoxicity and greater cytokine release than their monospecific counterparts. This pattern is consistent with the idea that engaging two epitopes on the same cell provides stronger and more sustained activation signals, a phenomenon that has been observed in other biparatopic and dual-target CAR systems. The engineering principle is straightforward: when a CAR-T cell can dock at two points on its target antigen, the probability of productive activation rises, and the cell is less likely to be fooled by partial loss of one epitope.

One of the most clinically consequential findings concerns soluble MUC1 in the tumor microenvironment. Shed MUC1α circulates as a soluble antigen, and one clinically familiar readout of this shedding is CA153, cancer antigen 153, a marker routinely measured in the serum of patients with MUC1-expressing carcinomas. Soluble antigen is a classic saboteur of CAR-T therapy: it can bind the CAR before the cell ever reaches the tumor, triggering exhaustion or simply soaking up the receptor. The researchers therefore tested whether soluble CA153 would impair the killing ability of their biparatopic CAR-T cells. It did not. Not only was killing preserved, but exposure to soluble CA153 actually induced CAR-T activation. The dual-target design appears to have converted the soluble antigen from a decoy into a stimulus, although the authors note that the precise consequences of chronic activation by circulating antigen in patients will require careful clinical evaluation.

The therapeutic promise held up in living systems. In ovarian tumor xenograft models, the MUC1 biparatopic CAR-T cells exhibited superior tumor control compared with monospecific controls, maintaining pressure on tumors even as antigen presentation varied. Perhaps more importantly for translational relevance, the team evaluated the cells against patient-derived tumor-like cell clusters, three-dimensional cultures established directly from ovarian cancer patients that preserve more of the heterogeneity and architecture of real tumors than standard cell lines. Against these patient-derived clusters, the biparatopic cells again showed enhanced activity, suggesting that the approach can perform against the messy, heterogeneous antigen landscape that actual clinical CAR-T products will face.

The study’s framing of partial MUC1α shedding as a prevalent and exploitable feature of ovarian cancer is what elevates it beyond a single construct report. If multiplex immunohistochemistry can confirm in a given patient’s tumor that both full-length MUC1 and truncated MUC1β are present, clinicians would have a rational basis for selecting a biparatopic product, and the shed form itself becomes a biomarker of target availability rather than a warning sign of escape. The work also highlights a broader lesson for the CAR-T field: targets that shed are not necessarily poor targets, provided the receptor is engineered to recognize what shedding leaves behind. Several other tumor-associated antigens, including epidermal growth factor receptor variants and other mucins, are known to undergo proteolytic cleavage, and the biparatopic logic demonstrated here could in principle be adapted to them.

Caveats remain, as they always do at this stage. The work rests on xenograft models and ex vivo patient-derived clusters rather than human trials, and questions of safety, persistence, and the behavior of the cells in the presence of the high circulating MUC1 levels seen in some patients will need to be answered in carefully designed clinical studies. The patent application filed by several of the authors signals commercial interest in moving the construct toward the clinic. Still, the study establishes a clear proof of principle: by targeting both the intact antigen and the truncated remnant that shedding exposes, CAR-T cells can be built that treat antigen loss not as an obstacle but as an invitation. For a disease like ovarian cancer, where relapse and immune escape are the norm, that inversion of the problem may prove to be the study’s most lasting contribution.

Subject of Research: Biparatopic CAR-T cell therapy targeting MUC1 shedding to overcome antigen escape in ovarian cancer

Article Title: Biparatopic CAR-T cells targeting full-length MUC1 and truncated MUC1β circumvent MUC1α shedding in ovarian cancer

Article References: Ge, Y., Zhang, Y., Li, X., Zhao, J., Wang, H., Yang, S., He, X., & Zhu, H. (2026). Biparatopic CAR-T cells targeting full-length MUC1 and truncated MUC1β circumvent MUC1α shedding in ovarian cancer. Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-09032-x

Image Credits: AI Generated

DOI: 10.1186/s12967-026-09032-x

Keywords: ovarian cancer, CAR-T cell therapy, MUC1, antigen shedding, biparatopic CAR, immunotherapy, MUC1β, CA153, tumor escape, alpaca antibodies, patient-derived tumor models, Journal of Translational Medicine