Polyfunctional Effector PD-1+ CD8+ T Cells and M1 Macrophages Promote Immune Checkpoint Blockade Response in Richter Syndrome Mouse Models

Richter syndrome (RS) displays dismal prognosis and remains largely refractory to most existing therapies. Immune checkpoint blockade (ICB) and bi-specific antibodies have shown promising activity, yet determinants of response and resistance to these immunotherapies remain incompletely understood. We leveraged our previously developed mouse models of RS, achieved via CRISPR-Cas9 gene editing of combinatorial loss-of-function drivers in immune competent mice, to interrogate microenvironmental determinants of disease transformation and response to therapy.

We profiled immune cells from the spleen of 18 primary murine tumors displaying either CLL (n=4), CLL/RS (n=3) or RS (n=11) histology by sc-RNAseq. Analysis of a total of 104,731 T/NK cells revealed pronounced changes in the CD4⁺ and CD8⁺ T cell compartment, with CD8⁺ effector/exhausted T cells and CD4⁺ cytotoxic cells as the predominant subsets in RS compared to CLL (pAdj<0.038). Flow cytometric analysis of an independent set of 17 cases (5 CLL, 3 CLL/RS, 9 RS) confirmed increased CD8/CD4 T cell ratio as typical of RS (p=0.008), together with abundant effector/memory CD8⁺ T cell subsets expressing high PD-1 and TOX levels (p=0.001). Enrichment in cytotoxic/exhausted CD8⁺ T cells in RS was validated in bulk external RNA-seq data from 34 human RS lymph node (LN) samples and 34 CLL-LNs using a gene score based on human orthologs of 13 genes derived from the mouse data (p=0.0023). Importantly, ex vivo stimulation of murine splenocytes (5 CLL, 6 RS) with PMA/ionomycin showed marked secretion of IFNg, TNFa and Granzyme B in PD-1⁺ CD8⁺ T cells from RS cases (p<0.02 for all combinations). In contrast, CLL CD8⁺ PD-1⁺ T cells only showed induction of TNFa. TCR-seq revealed pronounced clonal expansion in RS compared to CLL (Shannon index p=0.008), indicative of chronic antigenic stimulation in RS.

The myeloid/monocytic compartment (41,956 cells) displayed more stability at transformation, aside from new enrichment of CXCL9⁺/CXCL10⁺ macrophages in RS compared to CLL (pAdj=0.093). This population displayed features of inflammatory M1 cells, including high expression of H2-Ab1 (class II), Cd274 (PD-L1), Cd40, Stat1, and Cd86. Flow cytometric analysis confirmed increased abundance of CXCL9⁺ macrophages in splenic preparations from RS compared to CLL mice (p= 0.007), and signatures extracted from CXCL9⁺ macrophages were validated as enriched in bulk RNA-seq from human RS cohorts (p<0.0001). Importantly, proportional levels of CXCL9⁺ macrophages correlated with those of CD8⁺ exhausted T cells (Pearson r= 0.69, p<0.0001). Numerous related ligand-receptor pairs, including PD-1/PD-L1, were significantly enriched in RS compared to CLL per the Liana algorithm. Physical proximity of IFNG, GZMB, and CD3D expressing cells, representing cytotoxic T cells, to CXCL9 and CD68 expressing cells, representing CXCL9/10⁺ macrophages (Moran’s R > 0.1), was evident in one RS-LN analyzed by 10X Visium.

To interrogate determinants of response/resistance to ICB therapy, we treated 4 different RS tumor models (achieved by transplantation of RS primary tumors into syngeneic immune competent recipients) and one CLL for 3 weeks with an anti-PD-1 monoclonal antibody (or isotype control) and analyzed changes in immune microenvironmental features at the end of treatment by combined scRNA-seq and flow cytometry. While CLL transplants remained insensitive to therapy, we observed a range of responses in the RS models, from non-response (1 line) to partial response (1 line) to marked response to treatment (2 lines). Importantly, responding cases showed lower tumor burden in spleen at euthanasia, higher PD-1 or PD-L1 tumor expression before treatment (p<0.001), and increased baseline (or post-treatment) polyfunctionality of CD8⁺ PD-1⁺ T cells. Complete tumor regression in one of the RS lines was accompanied by loss of CXCL9⁺ macrophages (p=0.001) and CD8⁺ PD-1⁺ T cells (p=0.0002), suggesting that tumor-intrinsic features favor retention of these subpopulations prior to therapy.

Overall, these data suggest key changes in immune microenvironmental features of RS compared to CLL, which associate to response to ICB in mice. Analysis of spatial datasets from LN of RS patients treated with ICB is ongoing to evaluate whether these features associate to response in humans, together with further investigation of their relevance to outcome through in vivo depletion experiments.