Discovery of Undocumented, Cancer-Specific and Pro-Inflammatory Isoforms of WNK2 That Are Highly Expressed in MYD88 Mutated Waldenström’s Macroglobulinemia and Represent Novel Therapeutic Targets.

The genomic landscape of WM is characterized by mutations in MYD88 (MYD88^MUT) and CXCR4 in 95-97% and 30-40% of patients, respectively. Transgenic mouse studies suggest that MYD88^MUT alone is insufficient for lymphomagenesis and that additional “hits” are required. The serine/threonine protein kinase WNK2 emerged as a putative second “hit” candidate in our previous studies (Guerrera ML et al, ASH 2023). WNK2 negatively regulates ERK1/2 activation in a kinase-dependent manner and is a known tumor suppressor. In WM, ERK1/2 is a critical pro-survival pathway that triggers the release of inflammatory cytokines and promotes ibrutinib resistance. We therefore sought to clarify the expression and function of WNK2 in MYD88^MUT WM. Among 254 untreated WM patients, WNK2 was significantly upregulated in early-stage WM (n=74, median 8.6 TpM) and either further upregulated or silenced in cases evolving into plasma cell-like (PCL, n=77, median 25.1 TpM) or B-cell like (BCL, n=103, median 0.3 TpM) WM, respectively. Normal B cells lacked WNK2 expression. For further transcriptional studies, we performed PacBio Isoseq analysis and discovered novel isoforms that shared the skipping of the same two distal exons and grouped into two distinct families. The first (spliced kinase domain family, S-K) included aberrantly spliced variants of the otherwise canonical full-length isoforms containing the kinase domain (canonical kinase domain family, C-K). The second group (spliced non-kinase family, S-NK) included alternatively spliced variants of the ENST00000453718 isoform that lacks the kinase domain. We sought to investigate the cancer specificity of the novel spliced isoforms in normal tissues with endogenous WNK2 expression, such as the brain, and tumors aberrantly upregulating WNK2 vs the respective non-tumor tissues, such as WM and cholangiocarcinoma. To this end, we analyzed publicly available RNA-Seq datasets of normal brain (n=50, SRA# SRP319543) and paired intrahepatic cholangiocarcinoma/adjacent non-tumor tissues (n=15, SRA# SRP159264). In brain samples, total WNK2 expression reached a median of 46 TpM; C-K was significantly more expressed than both S-K and S-NK (p<0.0001), each accounting for a median of 14%, 0%, and 10.2% of total gene expression. Conversely, the 151/254 (60%) WM expressing WNK2 showed a significant increase in the relative usage of both S-K and S-NK over C-K (median 18.7%, 30.2%, 4.4%, respectively, p<0.0001). Patients’ CXCR4 mutational status did not affect the differential isoform usage. Likewise, WNK2 expression was significantly higher in cholangiocarcinoma vs non-tumor samples (median 16.9 vs 0.4 TpM, p=0.002). In tumors with meaningful WNK2 expression (n=10), S-K, S-NK, and C-K accounted for 9.1%, 17.3%, and 17.1% of total gene expression, respectively. Compared to normal brain tissues, S-K expression was tumor-specific in both WM (p<0.0001) and cholangiocarcinoma (p=0.0025); we also observed a significantly increased usage of S-NK in both WM (p<0.0001) and cholangiocarcinoma tissues (p=0.001), and decreased usage of C-K in WM (p<0.0001). Within the S-NK, the individual isoform usage significantly differed in brain vs WM and cholangiocarcinoma samples. To investigate the functional role of this family, we overexpressed the dominant S-NK isoform in WM patients via lentiviral transduction in WM cell lines. Stable lines showed increased phosphorylation of ERK1/2 and SRC members, including HCK. At the RNA-Seq level, they showed enrichment of the neutrophil chemotaxis and RNA splicing pathways vs vector controls. We validated the expression of 12 among the top differentially expressed genes, including the S100A9 and S100A8 alarmins and proinflammatory chemokines, such as CXCL13, involved in regulating inflammation, myeloid chemotaxis, and activating the ERK1/2 and NF-kB pathways.

Treatment of WM patients’ mononuclear cells with WNK463, a commercially available potent and selective oral kinase inhibitor of the four WNK kinase family members (WNK1-WNK4), induced a dose-dependent B-cell death. Taken together, our findings demonstrated novel and cancer-specific isoforms of WNK2, highly expressed in WM and aggressive cholangiocarcinoma tumors, of which the dominant S-NK isoform in WM promotes inflammation and myeloid chemotaxis. Our findings provide a critical framework for the selective therapeutic targeting of cancer-specific WNK2 isoforms.