Scientists Engineer CAR-T Cells to Target Amyloid in Alzheimer's Disease

Scientists led by Jonathan Kipnis at Washington University in St. Louis reported that T cells can be engineered to express chimeric antigen receptors (CARs) that recognize aggregated Aβ. In mice, T cells stably expressing these CARs reduced amyloid in the dura, while a version with transient receptor expression cleared plaques in the brain. The findings were published in the February 9 PNAS.

The researchers created the CAR by fusing the coding sequences for the heavy and light chains of lecanemab's variable region into a single-chain amyloid-binding fragment, then linked it to CD28 co-stimulatory and CD3ζ-signaling domains. They packed this construct into a viral vector.

First author Pavle Boskovic and colleagues isolated CD4+ helper T cells from wild-type spleens and transduced them to stably express the construct. Unlike cytotoxic CD8+ cells used in cancer therapy to destroy targeted cells, CD4+ cells rally other immune cells, including microglia, when activated. The scientists injected two doses of CAR CD4+ T cells, spaced three weeks apart, into the veins behind the eyes of 6-month-old 5xFAD mice. These mice express human APP and PSEN1 transgenes carrying multiple AD-linked variants and develop abundant amyloid pathology by this age.

After six weeks, about 2 percent of the dura membrane near bridging veins contained amyloid, as compared to 7 percent in saline-injected controls. In 2024, Kipnis' group had reported that molecules enter and exit the brain parenchyma through "cuffs" surrounding these bridging veins. No changes in amyloid pathology occurred within the brain itself with this approach.

The authors think viral gene delivery for stable CAR expression poses the risk of excessive immune activation that, in rare cases, can lead to neurodegeneration. They tried an alternative strategy, transfecting CD4+ T cells with the lecanemab CAR mRNA so the cells would only express the receptors temporarily. They injected three doses of these transient CAR-T cells into the eye veins of 5xFAD mice, spaced 10 days apart. In contrast to the two doses of stably transfected CAR-Ts given three weeks apart, this more frequent regimen reduced amyloid plaques in the brain itself.

Kipnis thinks more doses over a shorter period likely played a role in the brain plaque reduction. Repeated administration may also be safer than using virally transduced cells. The authors do not yet know how these lecanemab CAR-T cells reduce amyloid from dura or parenchyma. Kipnis suspects the cells secrete factors that recruit the host's immune cells, such as microglia, to do the job, and said future work will clarify this mechanism.

CARs are lab-made, cell-membrane receptors that redirect T cells to specific targets. In the late 1980s, Zelig Eshhar and colleagues at the Weizmann Institute of Science in Rehovot, Israel, created the first versions by fusing a single-chain antibody fragment to an intracellular signaling domain that activates T cells once the CAR receptor binds its target. Though this was a conceptual breakthrough, these CARs weakly triggered T-cell activation. In the late 1990s and early 2000s, scientists began adding co-stimulatory domains to CARs to mimic a secondary activation signal. This dramatically improved the cells' activation and survival in animal models of cancer and, later, in people.

CAR-T therapy works best against blood cancers, directing T cells toward malignant B cells by targeting proteins on their surface, such as B-cell maturation antigen or CD19.

In parallel research addressing CAR-T safety concerns, a team of scientists from Ludwig Lausanne, led by Melita Irving and Greta Maria Paola Giordano Attianese, alongside collaborators at the École Polytechnique Fédérale de Lausanne, has engineered an innovative CAR-T cell platform capable of being remotely and reversibly switched off. Their findings, published in Nature Chemical Biology, address formidable obstacles such as the risk of collateral damage to healthy cells and potentially life-threatening immune hyperactivation.

The novel technology enhances control over CAR-T cells via a 'drug-regulated off-switch protein-protein interaction CAR' (DROP-CAR) that modulates receptor integrity at the cell surface. This system does not rely on degrading CAR components or inducing CAR-T cell death, as previous off-switch designs did. Instead, it harnesses a finely engineered protein interface consisting entirely of human-derived elements, thus minimizing immunogenicity. The extracellular domain includes a computationally designed human protein, dubbed dmLD3, which binds BCL-2 with exceptional affinity. The CAR's antigen-binding moiety is appended with a complementary BCL-2 fragment.

Venetoclax, an FDA-approved cancer drug known for its high-affinity binding to BCL-2, serves as the molecular remote control in this system. Administration of venetoclax competitively disrupts the dmLD3-BCL-2 interaction, causing the extracellular CAR architecture to dissociate and the receptor to disassemble, effectively silencing the CAR-T cell's tumor-targeting function. Crucially, the CAR receptors then swiftly reassemble upon venetoclax withdrawal, restoring cytotoxic activity. This reversible mechanism allows precise temporal modulation of CAR-T cell functionality without triggering apoptosis or cell removal, preserving the therapeutic cell population across treatment cycles.

The DROP-CAR design provides a potential solution to antigen-driven exhaustion, a phenomenon whereby persistent stimulation in the immunosuppressive tumor microenvironment leads to a dysfunctional T cell state, marked by epigenetic and transcriptomic remodeling that curb effector functions. The design enables treatment protocols in which CAR-T cells can be transiently 'paused,' allowing them to rest and recover function before reactivation. Such temporal control could extend the durability and efficacy of CAR-T therapies against solid tumors, which often exhibit highly suppressive milieus.