Checkpoint immunotherapy rejects primary tumors without cDC1 cells or lasting immune memory

A new study has revealed that immune checkpoint blockade can drive the first wave of tumor rejection even when a key population of dendritic cells is missing—but the same immune response may fail to leave behind lasting protection. The findings, reported by researchers at Hokkaido University and collaborating institutions, challenge a widely held assumption about how checkpoint immunotherapy works and suggest that the immune machinery needed to destroy an established tumor may differ from that required to remember it.

Immune checkpoint blockade, or ICB, has transformed cancer treatment by releasing molecular brakes that restrain T cells. Drugs that block proteins such as PD-1, PD-L1 or CTLA-4 can restore the ability of T cells to recognize and attack malignant cells, producing long-lasting responses in some people with cancer. Yet the treatment does not work uniformly. Its success depends on a chain of events in which tumor antigens are captured, processed and displayed to T cells by antigen-presenting cells, particularly dendritic cells. Among these, conventional type 1 dendritic cells, known as cDC1s, have attracted intense attention because they are highly effective at cross-presenting tumor-derived proteins on major histocompatibility complex class I molecules, a process that activates tumor-killing CD8-positive T cells.

The new work examined whether cDC1s are absolutely required for ICB to eliminate a primary tumor. To do so, the researchers used mice carrying an immunogenic clone of Lewis lung carcinoma, or LLC, a transplantable mouse tumor model. They compared normal animals with genetically modified Batf3-deficient mice. The Batf3 transcription factor is necessary for the development of cDC1s, so Batf3-knockout animals lack this dendritic-cell subset. In normal mice, most animals rejected the LLC tumors after receiving ICB. Surprisingly, the treatment also triggered tumor rejection in 35.7 percent of the Batf3-deficient mice. The result indicates that, at least for this immunogenic LLC tumor, alternative antigen-presenting cells can support an initial antitumor response when cDC1s are absent.

That alternative route appears to involve XCR1-negative antigen-presenting cells. XCR1 is a chemokine receptor associated with cDC1s, making it a useful marker for distinguishing these cells from other antigen-presenting populations. After ICB, the researchers observed increased expression of costimulatory molecules on XCR1-negative APCs in both tumors and draining lymph nodes of Batf3-deficient mice. Costimulatory molecules such as CD40 and CD80 provide essential secondary signals during T-cell activation. Antigen recognition alone is often insufficient; without costimulation, T cells may become inactive or tolerant. The observed increase in CD40 and CD80 suggests that non-cDC1 APCs were not merely present but were being functionally activated in response to checkpoint therapy.

The researchers then investigated why the LLC model could provoke this backup immune pathway while another tumor model, the B16F10 melanoma, remained resistant to ICB. They collected conditioned culture media—the fluid containing molecules secreted by tumor cells—from LLC and B16F10 cultures and exposed bone-marrow-derived dendritic cells to it. Media from LLC cells stimulated both cDC1s and cDC2s, another conventional dendritic-cell subset, causing increased expression of CD40 and CD80. Media from ICB-resistant B16F10 cells did not produce the same effect. This experiment points to soluble factors released by the tumor microenvironment as potential drivers of dendritic-cell activation. The molecules were not identified in the study, but the contrast suggests that tumor cells can differ substantially in their ability to alert and organize immune responses.

To explore that difference at the molecular level, the team performed RNA sequencing on LLC and B16F10 tumor cells. The analysis showed that genes linked to antitumor immunity were more strongly expressed in LLC cells than in B16F10 cells. Such genes could influence how tumor antigens are released, how inflammatory signals are generated or how immune cells are recruited and activated. A tumor that produces the right combination of danger signals may effectively condition multiple APC populations, creating redundancy in the pathway leading to T-cell activation. By contrast, an immune-cold tumor such as B16F10 may fail at several points: it may present fewer recognizable antigens, release weaker activating signals or actively suppress the cells that would otherwise initiate immunity.

The most consequential finding emerged when the researchers tested immune memory. Mice that had become tumor-free after ICB were later exposed again to LLC tumors, a standard rechallenge experiment designed to determine whether the initial response created protective immunological memory. In Batf3-deficient animals that had rejected their primary tumors without cDC1 supplementation, the rechallenged tumors were not spontaneously eliminated. In other words, the mice could mount a successful first attack but did not retain a sufficiently powerful memory response to repel the same tumor a second time. This distinction is biologically important. Primary rejection can rely on a temporary or locally organized immune response, whereas durable memory requires the generation, survival and later reactivation of specialized T-cell populations, processes that depend on precise antigen presentation and coordination among immune cells.

The findings support a two-stage model of checkpoint immunotherapy. During the first stage, activated XCR1-negative APCs, including cDC2-like populations, may present tumor antigens and provide costimulation strong enough to initiate T-cell-mediated destruction. These cells could acquire tumor material in the tumor bed or draining lymph nodes, process it and present it to T cells through major histocompatibility complex molecules. Checkpoint blockade would then remove inhibitory signaling, allowing the newly activated T cells to expand and attack the cancer. During the second stage, however, cDC1s may be indispensable for shaping the quality and persistence of the response. Their specialized cross-presentation capacity could help sustain repeated T-cell stimulation, support the development of memory precursor cells and establish long-term surveillance against residual or returning tumor cells.

The work does not mean that cDC1s are unimportant in all cancers, nor does it show that patients lacking a direct equivalent of the mouse pathway would respond in the same way. The experiments used genetically modified mice and transplantable tumor models, and the percentage of Batf3-deficient animals rejecting LLC was substantially lower than the response seen in wild-type mice. The results instead highlight the complexity of immune responses within tumors and the danger of reducing immunotherapy to a single cellular mechanism. Future treatments may need to activate both cDC1-dependent and cDC1-independent pathways: one to generate a forceful initial attack and another to ensure that the immune system remembers what it has defeated. Identifying the soluble tumor-derived signals that activate alternative APCs could help explain why some cancers respond to ICB while others resist it, while strategies that restore or enhance cDC1 function could improve the durability of responses. The study’s central message is therefore both encouraging and cautionary: the immune system may find more than one way to destroy a tumor, but the route to lasting protection is narrower than the route to an initial victory.