The Gene Expression Landscape of BTK Inhibitor Treatment in Chronic Lymphocytic Leukemia Exposed By Shallow Depth RNA Sequencing

Efforts to understand chronic lymphocytic leukemia (CLL) subgroups have focused on multi-omics profiling of patient cohorts (in steady-state) and linking biomarkers to responses. Here, we aim to combine ex-vivo drug screening and gene expression profiling (GEP) in primary CLL samples to dissect disease biology based on the observed phenotype as a qualitative and quantitative function of steady-state features and drug perturbation effects. As most differential gene expression changes in CLL are retrieved from highly expressed genes, we hypothesized that a relevant set of genes could still be reliably captured by reduced sequencing coverage. We profiled five samples with traditional whole transcript TruSeq and 3'-end QuantSeq (Moll et al., Nat Methods, 2014) library preparation with deep and shallow depth sequencing, respectively. We obtained 43.48-72.68 million mappable reads with TruSeq and 2.24-3.78 million with QuantSeq. Although the read coverage was reduced, QuantSeq was able to recover 71.17 ± 0.87% of all genes that were detected by TruSeq with reproducible quantification ( R > 0.85, P < 0.001 for all samples). Shallow sequencing of CLL samples reliably profiled intermediate to highly abundant genes with reduced sensitivity only for lowly expressed genes. We then selected 105 representative CLL patient samples from a well-characterized cohort, treated them ex-vivo with 100 nM ibrutinib for 48h and performed GEP using shallow depth QuantSeq. Sample viability was measured by flow cytometry before and after perturbation. Library preparation and sequencing was successful for 91% of samples. With a median coverage of 2.19 million mappable reads per sample, the principal components (PC) of transcriptional variation in DMSO-treated samples included IGHV (12.0%), methylation subgroup and trisomy 12 status (7.1%). These data show that the screening method reliably captured the transcriptional features of CLL at steady-state and thus support the use of shallow depth RNA sequencing. We were interested in the landscape of transcriptional change after perturbation of the B-cell receptor (BCR) pathway. Samples from the same patients clustered together and spread along a gradient of sample viability in the first PC. After correlation of gene counts from DMSO-treated samples with sample viability and subsequent gene set enrichment analysis, we found activation of the canonical apoptosis pathway via NFκB-mediated TNF-α signaling and cell stress to be associated with