Tracing the evolution of clonal T cell responses in LGL leukemia with single-cell whole genome sequencing Free

T cell large granular lymphocytic leukemia (T-LGLL) is a rare hematological disease of clonally expanded mature cytotoxic T cells. T-LGLL is thought to originate as a response to chronic (auto)antigen stimulus, as it often presents with autoimmune-mediated cytopenias and rheumatoid arthritis (RA). Approximately 40-50% of CD8+ T-LGLL patients have somatic STAT3 mutations in CD8+ T cells. However, previous analyses done in bulk have not been able to detangle whether STAT3 mutations initiate aberrant T cell expansion or accrue as subclonal events that give T cells a proliferative advantage independent of antigen drive.

Here, we used single-cell whole genome sequencing (scWGS) to profile the life histories of CD8+ T cells in five T-LGLL patients. Understanding the underlying clonal structure of STAT3 mutations across T cell clones could pave the way for improved diagnostic, monitoring, and treatment options for T-LGLL patients.

Methods We sequenced the whole genomes of 120 T cells flow-sorted from peripheral blood samples of 5 T-LGLL patients, leveraging primary-template directed amplification (ResolveDNA, Bioskryb Genomics). We genotyped DNA amplified from individually sorted T cells for STAT3 and selected 5-26 STAT3^mt and 8-15 STAT3^wt single T cells per patient for WGS with a target coverage of 15×.

T-LGLL patients were aged 42-78, sampled either at diagnosis (n=3) or follow-up (n=2). All patients harbored at least one STAT3 variant: Y640F (n=4), N647I (n=1), D661Y (n=2) or D661V (n=1). RA preceded T-LGLL diagnosis in 3 patients. 4 patients received treatment for T-LGLL (mean 2.6 lines of treatment).

ResultsSTAT3^mt LGL CD8+ T cells harbored an elevated single-nucleotide variant (SNV) burden compared to STAT3^wt CD8+ T cells in 2/5 patients and an elevated insertion-deletion burden in 2/5 patients (p<0.05, Wilcoxon). Overall mutation load increased with increasing patient age. Mutational signature analysis revealed that the clock-like mutational processes SBS1, SBS5, and SBSblood accounted for most SNVs in both STAT3^mt and STAT3^wt T cells.

We used shared and unique somatic variants across sequenced CD8+ T cells to reconstruct molecular phylogenies for each patient. We complemented our phylogenetic trees by extracting the T cell receptor (TCR) of each sequenced cell. Each patient had a unique phylogenetic tree structure, but several general patterns emerged. First, we observed few shared mutations between T cells that belonged to different TCR clones, consistent with T cells developing from a polyclonal population of hematopoietic stem and progenitor cells. Second, the CD8+ T cell repertoires of LGLL patients were strikingly clonal. Despite our small sampling of T cells from each patient, we captured at least two expanded TCR clones from each patient. Third, compared with TCR Vβ antibody staining of CD8+ T cells from the same samples, STAT3 variants did not necessarily arise in the dominant Vβ clone.

In patients with one STAT3 variant in one expanded TCR clone (n=3), the acquisition of STAT3 could be unambiguously timed after TCR rearrangement, implying that STAT3 variants arise as subclonal events during or after clonal expansion. The STAT3 mutation was timed to occur years or even decades before T-LGLL diagnosis and predated RA in the patients in whom the comorbidity was diagnosed.

Additionally, a patient with four different STAT3 variants displayed recurrent hits of the gene in numerous different TCR clones, demonstrating remarkably strong selection for STAT3 mutations. Finally, a patient with an unusual double-positive CD4+CD8+ T-LGLL phenotype had multiple different clones with the same STAT3 variant. In this patient, the acquisition of the STAT3 variant was timed before TCR rearrangement, early in T cell development.

Discussion We capitalized on the record resolution of scWGS to elucidate the clonal evolution of T-LGLL. Our study highlights the contribution of somatic mutations to functional shifts in healthy T cells, even when not instigating malignant transformation into an aggressive hematological cancer.

Our results uncover extensive heterogeneity in the phylogenetic histories of CD8+ T cells in different T-LGLL patients. This suggests that T-LGLL may be a more complex disease than previously thought, shaped by patient-specific selection pressures. Our findings hint at different models of STAT3 mutation acquisition in T-LGLL, which may reflect distinct clinical entities to be explored in larger cohorts.