Analysis of Alternative Splicing in Waldenstrom’s Macroglobulinemia Reveals Novel Targets, Insights into IgM Production and Improved Gene Level Expression Estimates

We recently described two distinct subtypes of MYD88 mutated WM: B-cell like (BCL) and Plasma cell like (PCL), both of which evolve from an unsubtyped clone that is prevalent in early/smoldering WM. We further described a gene signature known as the WM Evolutionary Score (EScore) that steadily increases from smoldering WM to symptomatic disease, regardless of subtype. EScore values significantly correlated with time to first therapy, mutation burden, bone marrow (BM) infiltration, and symptomatic disease (Hunter et al, ASH 2023). To better understand the role of alternative splicing in the biology of WM as well as the EScore, WM Subtypes, and the role of CXCR4 mutations, data from our multiomic data set of 249 MYD88 mutated untreated patients with Waldenstrom’s Macroglobulinemia (WM) was used for transcript quantification using a Gencode GRCh38 gene model augmented with novel isoforms identified from the PacBio IsoSeq analysis of 11 WM patients, 2 CD19+CD27+ healthy donor memory B-cell (HDMB) and 2 CD138+ healthy donor plasma cell (HDPC) samples (Richardson et al ASH 2023). Isoform transcripts per million (TpM) were then transformed into percent of total gene expression and filtered based on mean TpM values per analysis cohort and prepared for downstream analysis in limma using Bioconductor in R.

This approach identified 363 significant transcripts impacting 320 genes between the BCL and PCL WM subtypes as well as over 1,944 differentially spliced genes associated with the WM EScore. Comparisons with HDMB revealed 640 and 658 genes with differential transcript utilization (DTU) unique to PCL and BCL subtyped WM, respectively and 9,696 genes with DTUs in common. Notable DTU genes included CCND3, CARD9, TNFSF13, UNG, S100A6, DUSP22, FCRLB, HDAC4, and PIK3R2. Some significantly impacted genes driven by novel isoforms had corresponding changes in their gene effective length estimates relative to the plain Gencode model. This resulted in resulting in improved gene level TpM estimates and more than doubling the median gene level TpM values in some instances. Two examples of this were DUSP22 and PIK3R2 both of which had dominant undocumented short isoforms in either PCL subtyped WM (DUSP22) or HDMB (PIK3R2). The case of PIK3R2 is also illustrative of the importance of DTU level analysis. PIK3R2 is part of the PI3K/AKT signaling cascade and known as p85Beta at the protein level. The gene level TpM differences between WM and HDMB is unremarkable (p=0.2) but the DTU analysis reveals that a median of 75% (range 50-87%) of PIK3R2 in HDMB and 28% (0-87%; p<0.0001) in WM consists of a novel short isoform that is predicted to be non-coding. The dominate isoform in WM is a full-length coding variant making up a median of 41% (range 0-93%) of expression compared and 0% (range 0-24%; p<0.0001) in HDMB. Another example is the IgM heavy chain gene IGHM. While expression is over 9 times higher in WM (p<0.0001), the DTU analysis reveals that there is also a switch between the membrane bound and secreted isoforms of the gene with a secreted to bound ratio in HDMB of 0.34 (range 0.15-0.98%) compared with a median of 44.7 (range 0.2- 18,178.5; p<0.0001) in WM. We previously documented the higher levels of serum IgM associated with CXCR4 mutations in WM with a median of 4,188 mg/dl (range 241 – 10,000 mg/dl) and 2,597 mg/dl (range 104 - 10,321 mg/dl; p<0.0001) in CXCR4 mutated and wild-type, respectively. Likewise, the secreted IGHM isoform was significantly more expressed in CXCR4 mutated WM (p=0.0006). This was particularly true for the BCL subtype with a median CXCR4 mutant expression of 44,702 TpM (range 2,723 - 190,798) compared with wild-type 8,217 TpM (range 1,244 – 49,900 TpM; p < 0.0001).

Analysis of DTUs is essential to understanding the underlying pathology of WM and has reveals important differences in targetable pathways such as with PIK3R2 and insights into regulation of serum IgM production in WM.