Topic Modeling of Genotyping of Transcriptomes Reveals Collaboration between BTKC481S-Mutant and Wild Type Cells in Btki-Resistant Chronic Lymphocytic Leukemia

The acquired BTKC481S mutation is the most common resistance mechanism to Bruton's tyrosine kinase (BTK) inhibition in chronic lymphocytic leukemia (CLL). Yet, its variant allele frequency (VAF) is often low (<10%), raising the question of how wild-type (WT) cells survive in the context of low abundance of mutant (MUT) cells. To investigate this, we modeled the clonal dynamics of BTK-MUT and WT cells longitudinally in a cohort of 42 patients resistant to BTK inhibitors (BTKi). In 21 patients, clonal dynamics deviated from a simple model where MUT cells increase at the expense of WT cells. Instead, both MUT and WT cell numbers increased at relapse, suggesting that BTK-MUT and WT cells cooperate to drive resistance to BTKi. To explore the phenotypic variability that may underlie clonal cooperation, we expanded our genotyping of transcriptomes (GoT) profiling of seven CLL samples taken at relapse after BTKi [60,580 cells with a median of 8,713 cells per patient (6,819-9,838)] and carrying low VAFs (3-43%). GoT genotypes a mutation of interest (i.e. BTKC481S) jointly with single-cell transcriptomes of primary samples. This integration allows to link genotype and phenotype, and by comparing BTK-MUT and BTK-WT cells within the same patient, this method eliminates inter-patient variability and confounding factors, thus focusing on changes in expression and intratumoral heterogeneity (ITH). Transcriptomic ITH in CLL is largely driven by the recirculation of CLL cells between peripheral blood and lymphoid tissues. CXCR4hiCD5low CLL cells (i.e. quiescent fraction) migrate to lymph node (LN) proliferation centers (PR) via a CXCL12 gradient. In PR, they receive microenvironmental signals, upregulate mir155 (encoded by MIR155HG), increase BCR signaling, proliferate, and exit as CXCR4lowCD5hi cells (i.e. proliferative fraction). BTKi impairs CXCR4 signaling and homing to LN. Here, we curated gene signatures for the quiescent (CXCR4, CD24), proliferative (CD27, CXCR3, CD52, ITGAX) and MIR155HGhi fractions (MIR155HG, MYC, CD83, CDK4) leveraging published scRNA-seq datasets. We classified cells into the 3 fractions and uncovered an increased quiescent fraction (p=0.058) and a decreased proliferative fraction (p<0.05) in BTK-WT cells compared to BTK-MUT cells across all patients, suggesting that BTK-WT have an impaired homing capacity to LN. We reasoned that cancer cell states are plastic and continuous, requiring tailored methods to account for these features. We employed topic modeling, an algorithm that decomposes each gene expression profile into a set number of topics (k). Topics represent groups of correlated genes, linked to specific biological functions. We could annotate 3/5 topics to the fractions described above in most patients, with an increase of the MIR155HGhi topic (MIR155HG and MYC targets) in BTK-MUT cells (p<0.001). We also identified patient-specific topics that could not be captured through conventional methods. In one patient, we identified a topic enriched in interferon-stimulated genes (IFIT1, IFIT2, IFIT3, IFI44L, STAT1) that was specific to BTK-WT cells (p<0.001) within the quiescent fraction. As interferons protect CLL cells from apoptosis and BTKi sensitive cells are still responsive to interferons [Xia, J Immunol 2020], our data suggests that BTK-WT cells may survive by responding to circulating interferons without homing into LN. In the same patient, we observed a cell state enriched in BTK-MUT cells expressing IL4R, which is associated with resistance to BTKi through the IL4R-IL4 axis. Intriguingly, a cell state specific to BTK-WT cells in the same patient overexpressed Chronic Lymphocytic Leukemia Up-Regulated 1 (CLLU1), a gene that distinguishes CLL from normal B cells and other lymphoid malignancies and is a marker of poor prognosis. IL4R expression was mutually exclusive with CLLU1, suggesting a putative paracrine activation of the IL4 pathway, as CLLU1 was previously proposed to have structural homology to IL4. Notably, we observe subclonal expression of CLLU1 in 6/7 patients. Taken together, we propose two models of clonal collaboration between BTK-WT and BTK-MUT cells, involving paracrine signaling between interferons and CLLU1/IL4. We are currently investigating the relevance of these signaling axes in larger cohorts of longitudinally collected patient samples, including through in vitro functional validation assays.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.