1014 - Clonal architecture and single-cell transcriptome landscape in Richter’s syndrome

Richter’s syndrome (RS) is an uncommon clinicopathological condition in which chronic lymphocytic leukemia (CLL) evolves into aggressive lymphoma, the mechanism of which still remains difficult to elucidate.
To dissect the precise molecular mechanism of RS, paired samples of peripheral blood mononuclear cells and the lymphonodus tissue of the patient were analyzed by single-cell RNA-seq, and lymphoma-related gene mutations were examined by targeted next-generation sequencing.
An inferred branched revolutionary architecture from CLL clones to RS clones was shown in Fig. 1A. Overall, 27,058 single-cell transcriptomes were integrated into a two-dimensional map using the Uniform Manifold Approximation and Projection (UMAP) method. Four major cell types were identified, namely, B cells, mononuclear phagocytes, T cells and neutrophils. Compared to CLL, the cell composition and expression landscape of RS were significantly different. Nine clusters of B cells were identified using UMAP analysis. CLL cells were mostly mapped to Clusters 1, 5, 6, 7 and 8; RS B cells were mostly mapped to Clusters 2, 3 and 4; and Cluster 9 was mutually harbored by CLL and RS samples. Gene sets of downstream targets of MYC were significantly enriched in Cluster 4 and Cluster 9, indicating the potential therapeutic targets. PySCENIC analysis was used to identify master regulators of B cell clusters. The transcription factors BACH2 and JUNB were most associated with RS B cells. The upregulated RGS2 was the downstream target gene of BACH2 and JUNB. IRF9 was most associated with RS B cells, which could modulate the expression of PRDX1. GO enrichment analysis showed that differentially overexpressed genes of RS B cells were enriched in the proteasomal ubiquitin-independent protein catabolic process and regulation of cellular amino acid metabolic process, while scGSVA analysis also demonstrated that the proteasome pathway and amino acid metabolism pathway were enriched in the RS sample. CytoTRACE was used to predict the differentiation status of B-cell clusters. Cluster 4 was predicted to have lower capacity of differentiation. Subsequently, Monocle2 was used to perform pseudotime analysis on malignant B cells. Notably, proliferating cells were enriched at the end of Cluster 4 pseudotime trajectories.
The immune cell composition was remarkably changed between the CLL and RS samples. T cells were reclustered into six distinct populations by UMAP analysis: CD4+ naïve T cells, CD4+ regulatory T cells (Treg), CD8+ effector T cells (Teff), CD8+ exhausted T cells (Tex), CD8+ effector memory T cells (Tem), and natural killer cells (NK). In the RS sample, CD4+ naïve T cells, CD8+ Tex and Treg cells were significantly increased, while NK and CD8+ Tem cells were decreased. A volcano plot was created to show differentially expressed genes of T cell subtypes between CLL and RS samples. The upregulated genes of CD8+ Tex in the RS sample included ALOX5AP, which contributes to inflammation, and ABCG1, which is involved in T cell activation and apoptosis. CytoTRACE revealed the differentiation potential, and the exhausted T cells were predicted to have the least differentiation ability. In the RS sample, T cells deviated from the normal path. Meanwhile, more exhausted T cell subsets were observed in RS sample, which were also enriched at the end of the pseudotime trajectory. Genes correlated with exhaustion, toxicity or inflammation were mapped to T cell populations, which were measured by their uniform score (U-score). The exhaustion U-score was significantly higher and the toxicity U-score was significantly decreased in the T cell population of the RS sample, implying the possible contribution of T cell defects to the mechanism of RS. Meanwhile, we used CellPhoneDB to predict ligand-receptor interactions between malignant cells and immune cells. The cell-to-cell interactions were predicted to be impaired between Cluster 4 and T cell subtypes, suggesting that tumor progression may be attributed to immune escape. Furthermore, we observed that PDCD1/FAM3C, PTPRC/CD22, and CCL4/CNR2 were commonly enriched between B cells of the RS sample and CD8+ Tex, as well as CD8+ Teff cells. These interactions suggest that blocking these axes may affect the interaction of malignant B cells with surrounding immune cells and could be an effective strategy to treat RS.
In conclusion, we have demonstrated the clonal architecture and single-cell transcriptome landscape of RS using single-cell transcriptomics and bulk somatic mutation analyses, which is helpful to understand the potential molecular mechanisms of RS.