Bladder cancer is often resistant to anti-PD-1 immunotherapy, however, responses do occur in a small fraction of patients. In order to better understand the specific T cells that mediate tumor rejection in those infrequent cases, Oh and Kwek et al. used single-cell RNAseq and TCRseq to profile T cells from muscle-invasive bladder tumors and adjacent uninvolved bladder tissue from 7 patients, including 4 who underwent treatment with anti-PD-L1. Analyzing CD8⁺ T cells, the researchers found that the phenotypic states and TCR repertoires remained largely the same between tumors and adjacent nonmalignant tissue. In contrast, analysis of CD4⁺ T cells showed evidence of cytotoxic and regulatory tumor-specific states as well as heterogeneity within known CD4⁺ T cell states. The results were recently published in Cell.

To investigate the contribution of CD4⁺ T cells to antitumor responses, Oh and Kwek et al. sequenced and analyzed 16,995 tumor-infiltrating CD4⁺ T cells and 2,847 CD4⁺ T cells from adjacent tissue. Using this whole-transcriptome data, the researchers identified 11 clusters, each of which represented cells from all patients. These clusters defined several known CD4⁺ T cell states, including CCR7-expressing central memory cells and exhausted cells expressing high CXCL13 and IFNG, whose presence has been associated with improved outcomes in other cancers. Tregs (expressing Foxp3 and high levels of immune checkpoint molecules) were also identified within CD4⁺ T cells and were found to be abundant in tumors. The Treg population was not homogeneous; regulatory CD4⁺ T cells were distinguishable by either high or low expression of IL2RA and were found to consist of distinct proliferating (Ki67⁺) and non-proliferating states.

In addition to regulatory T cells, Oh and Kwek et al. identified cytotoxic CD4⁺ T cells in two distinct states: one marked by high expression of GZMB, and one marked by high expression of GZMK. In both states, these cytotoxic CD4⁺ T cells expressed IFNG. Some GZMB-expressing cells co-expressed TNF, CXCR6, PRF1, and high levels of NKG7. They did not express TOX or regulatory molecules. Using flow cytometry, the researchers confirmed that the majority of CD4⁺CCR7^- cells were polyfunctional and could produce both IFNγ and TNFα, resembling cytotoxic CD8⁺ T cells. In another analysis using bulk expression data from carefully sorted T cell populations, CD4⁺ TILs expressing GZMB were found to be most similar to tumor-specific CD8⁺ T cells expressing ENTPD1 (CD39), whereas CD4⁺ TILs expressing high GZMK were most similar to central memory or naive CD8⁺ T cells. Like Tregs, CD4⁺ T cells in both cytotoxic states were also found to be significantly enriched in tumors (although to a lesser extent than Tregs) and appeared to consist of distinct proliferating and non-proliferating states.

Further analyzing CD4⁺ T cells, Oh and Kwek et al. performed paired alpha and beta chain TCR sequencing on the same cells. Overall, the TCR repertoire was more restricted in tumors than in adjacent non-malignant tissue. When TCR sequences were assigned to clusters, the researchers observed evidence of notable clonal expansion within Tregs and CXCL13-expressing CD4⁺ T cells from tumors, but little sharing of expanded clonotypes between Tregs and cytotoxic cells. Tumor-specific clonal expansion of cells suggests that cytotoxic CD4⁺ T cells might expand in response to tumor antigens presented on MHC-II. Of note, most bladder tumor cells expressed MHC-II, which would allow antigen recognition by CD4⁺ T cells.

Investigating whether cytotoxic CD4⁺ T cells might contribute to antitumor immunity, Oh and Kwek et al. isolated CD4⁺ TILs from a patient and cultured them ex vivo with IL-2. When these expanded CD4⁺ T cells were cocultured with autologous tumor cells, they produced granzymes and perforin and induced tumor cell apoptosis. This effect was further enhanced when Tregs were removed from the culture of CD4⁺ T cells, suggesting that Tregs have a suppressive effect on autologous cytotoxic CD4⁺ T cells. When MHC-II was blocked in these cocultures, tumor cell apoptosis was reduced, suggesting that the response to tumor cells is dependent on the recognition of antigens presented on MHC-II.

To determine whether cytotoxic CD4⁺ T cells contribute to antitumor responses in patients with bladder cancer, Oh and Kwek et al. identified genes that were differentially expressed between different branches of CD4⁺ T cells observed in pseudotime, which grouped cells as proliferating cytotoxic CD4⁺ T cells, non-proliferating cytotoxic CD4⁺ T cells, or regulatory T cells. From this, the researchers developed a gene signature for cytotoxic CD4⁺ T cells and tested its predictive value to checkpoint therapy responses. Using bulk RNAseq data on tumor samples from 244 patients enrolled in a phase II clinical trial of atezolizumab for metastatic bladder cancer, the researchers found that the gene signature for cytotoxic CD4⁺ T cells correlated with response to anti-PD-L1 therapy in inflamed tumor samples, but not immune-excluded or immune desert samples.

The data collected by Oh and Kwek et al. suggest that tumor-infiltrating cytotoxic CD4⁺ T cells may expand in response to recognition of antigens presented on MHC-II and that they play an important role in bladder cancer and in the response to anti-PD-L1 immunotherapy. The work also shows that these cytotoxic CD4⁺ T cells are suppressed by intratumoral Tregs, which may limit antitumor responses. This information could be useful in evaluating patients likely to respond to immunotherapy, as well as designing future immunotherapy and combination therapies in bladder cancer, and possibly in other solid tumors with MHC-II expression and limited responses to immunotherapy.

by Lauren Hitchings