ZRSR2 loss causes aberrant splicing in JAK2V617F-driven myeloproliferative neoplasm but is not sufficient to drive disease progression

Myeloproliferative neoplasms (MPNs), including polycythemia vera (PV), essential thrombocythemia (ET), and primary myelofibrosis (PMF), are clonal disorders driven by mutations in hematopoietic stem cells (HSCs).^{1, 2} JAK2^V617F is the most recurrent driver mutation in MPN and results in the constitutive activation of the JAK-STAT pathway.^3-6 Mutations in ZRSR2 have been found in JAK2^V617F-driven MPNs and are associated with disease progression and poor prognosis. However, the functional consequences of mutations in ZRSR2 in terms of disease progression in JAK2^V617F-driven MPN have not been determined. In MPN, somatic mutations in ZRSR2 occur across the entire coding transcript, are predicted to confer a loss-of-function, and are primarily observed in male patients (Supporting Information: Figure S1).^7-9 Herein, using CRISPR-Cas9, we generated concomitant ZRSR2 loss in JAK2-mutated human megakaryoblastic cell lines (SET2, CHRF-288-11) and a mouse model of MPN to investigate its role in promoting disease progression. We confirmed ZRSR2 loss in both human cell lines and this mouse model by sequencing and immunoblotting (Supporting Information: Figures S2 and S3).

ZRSR2 is mainly involved in minor spliceosome assembly, with ZRSR2 loss previously shown to result in the mis-splicing of U12-type introns in myelodysplastic syndromes (MDSs).^{10, 11} We therefore performed replicate multivariate analysis of transcript splicing (rMATS)¹² to detect differential intron retention between ZRSR2 knockout (KO) megakaryoblastic cells and controls, and identified significantly increased retained intron (RI) events false discovery rate [FDR] < .05) in SET-2 ZRSR2 KO cell lines, but not in CHRF-288-11. However, in both megakaryoblastic cell lines, RI events tended to be more prominent in genes containing U12-type introns in the context of ZRSR2 loss (Figure 1A). Differential gene expression analysis was performed to assess how ZRSR2 loss impacts transcription. Gene set enrichment analysis (GSEA) showed enrichment for genes containing U12-type introns in both ZRSR2 KO megakaryoblastic cell lines versus control cells (Supporting Information: Figure S4A). We observed significant enrichment for megakaryocyte progenitors (MkPs) and E2F target gene sets, which are involved in MkPs and myeloid differentiation, in CHRF-288-11 ZRSR2 KO cells, but not in SET-2 ZRSR2 KO cells (Supporting Information: Figure S4B). Taken together, we conclude that ZRSR2 loss causes U12-type intron retention and transcriptional changes in genes containing U12-type introns in human JAK2-mutant megakaryoblastic cell lines.

Zrsr2 loss in Jak2^V617F-driven murine MPN was achieved by isolating bone marrow (BM) Lineage⁻Sca-1⁺c-Kit⁺ (LSK) cells from male donor mice (Jak2^+/fl-V617F, Rosa26^{creER/lsl-Cas9-T2A-GFP}), transducing them with a lentiviral vector expressing sgRNA targeting Zrsr2 (sgZrsr2), and transplanting into irradiated wild-type recipient mice. After 4 weeks of engraftment, mice were fed Tamoxifen-chow for 2 weeks to induce CreER and Zrsr2 editing (Figure 1B). Zrsr2-edited recipients and unedited controls all developed a lethal MPN resembling human PV with elevated hematocrit, splenomegaly, and expansion of erythroid precursors (Figure 1C–E). The disease phenotype observed was not associated with the sex of Zrsr2-edited recipients (Supporting Information: Figure S5). No significant differences in peripheral MPN disease parameters, main hematopoietic lineages, or survival were observed with Zrsr2 loss, except for an increase in peripheral progenitor cells (cKit⁺) (Figure 1C–F, Supporting Information: Figure S6A–D). ZRSR2 mutations have been described to occur frequently in PMF, a disease state characterized by abnormal megakaryocyte development.¹³ However, except for a slight decrease in granulocyte/macrophage progenitors (GMPs) in Jak2^V617F-sgZrsr2 mice, we did not observe any difference in the proportion of megakaryocyte/erythroid progenitors (MEPs), common myeloid progenitors (CMP), MkPs, or long-term HSCs (LT-HSCs) in the transduced donor cell population across conditions (Figure 1G–I). Furthermore, no difference was detected in the frequency of transduced donor cells within these populations (Supporting Information: Figure S6E), demonstrating that Zrsr2 disruption does not confer a competitive advantage in combination with Jak2^V617F within the HSPC compartment. Histopathological analysis across both conditions was consistent with a PV phenotype, showing effacement of splenic architecture with erythroid and megakaryocytic hyperplasia and the megakaryocytes in BM displayed atypical nuclear features and clustering, with no evidence of disease progression observed with Zrsr2 loss (Supporting Information: Figure S6F).

We further investigated whether Zrsr2 disruption affected intron retention in BM LSK cells and identified a higher number of RIs in Jak2^V617F-sgZrsr2 mice. However, these RIs were less restricted by type, with fewer occurring in genes containing the U12 splice site in mouse LSK cells compared to human ZRSR2 KO megakaryoblastic cells (Figure 1A,J). Furthermore, we identified only four genes with RIs regulated by ZRSR2 in both species (Supporting Information: Figure S7A), indicating that Zrsr2 loss results in different levels of mis-splicing of genes containing U12-type introns between cell types and/or between humans and mice, with notably less predilection in murine cells. Consistent with this, the bulk gene expression profiles of murine genes containing U12-type introns, in contrast to human cells, did not demonstrate significant enrichment of transcripts containing U12-type introns in either direction (Figure 1K). However, consistent with genetic profiles of chronic-stage MPN patients that progress to MF who show megakaryocytic hyperplasia and aberrant mobilization of HSCs, we observed significant enrichment for MkP-associated gene sets and transcriptional changes associated with genes upregulated in HSCs in Jak2^V617F-sgZrsr2 compared to Jak2^V617F-EV cells (Supporting Information: Figure S7B). Taken together, Zrsr2 loss causes modest intron retention and minor transcriptional changes in murine LSK cells. However, the transcriptional changes that occur as a consequence of Zrsr2 loss are not sufficient to induce disease progression of Jak2^V617F-driven murine MPN in vivo.

Our results are similar to the report of Madan et al., in that Zrsr2 deficiency also led to fewer U12-type mis-splicing events observed in murine hematopoietic cells compared to human MDS or leukemia cell lines.¹⁴ In this study, they raised the issue that Zrsr1, a homolog of Zrsr2 expressed in mice but not in humans, may compensate for the impact of Zrsr2 loss in regulating RNA splicing in mice. Zrsr1 expression is not affected by Zrsr2 KO in our murine model (Supporting Information: Figure S7C). The compensatory role of Zrsr1 for Zrsr2 in hematopoietic development in mice may explain the attenuated intron retention and milder disease phenotype observed in our murine model compared to human cells. However, the frequent co-occurrence of mutations in epigenetic regulators and splicing regulators in MPN samples harboring a ZRSR2 mutation may also suggest that a cooperative role for these molecular lesions with ZRSR2 is required for the progression of MPN.^15-18

From a published MPN cohort containing 12 patients with concurrent ZRSR2 mutations in 1289 JAK2^V617F patients,⁷ we found that ZRSR2 mutations were significantly more frequent in JAK2^V617F cases with MF compared to those with PV and ET (P < 0.0001, Pearson's chi-squared test), with frequencies of 3.69%, 0.28%, and 0.41%, respectively. Among the patients with co-mutations in JAK2^V617F and ZRSR2, eight cases had one or more additional mutations, most frequently detected in the epigenetic regulators ASXL1 and TET2 (Figure 2A). It is noteworthy that these co-occurring mutations were more commonly associated with MF rather than ET or PV, including chromatin modifiers ASXL1 and EZH2, or DNA methylation regulator TET2 (Figure 2A). Moreover, the frequency of ASXL1, EZH2, and TET2 mutations was significantly higher in cases co-mutated with ZRSR2 compared to those without ZRSR2 mutations (ASXL1: 33.3% vs. 6.81%, P = 0.0074; EZH2: 16.67% vs. 1.72%, P = 0.019; TET2: 33.3% vs. 14.41%, P = 0.0839; Fisher's exact test). These data suggest that high-risk mutations in chromatin modifiers ASXL1 and EZH2 that predispose to progression of MPN to MF are more frequently present in JAK2^V617F MPN patients with ZRSR2 mutations. We validated these findings in an independent data set that included 18 ZRSR2-mutated MPN patients within a cohort of 990 MPN patients seen at DFCI (Supporting Information: Table S1). Of these JAK2 and ZRSR2 co-mutated patients, we observed similar findings, with eight patients having at least one additional mutation, occurring most frequently in ASXL1 (46.2%) and TET2 (30.8%), respectively (Figure 2B). Interestingly, two patients lost ZRSR2 mutations upon progression to SMF, and three patients acquired their ZRSR2 mutation in a late disease stage, after onset of MF (Figure 2C), suggesting that ZRSR2 mutations are not primary drivers of myelofibrosis progression. Using the DFCI cohort, we further explored the relationship between the ZRSR2 mutations and the MPN phenotype by analyzing the correlation matrix of ZRSR2 variant allele frequency (VAF) with blood parameters. Consistent with clinical observations, patients with MF or SMF showed a trend of higher ZRSR2 VAF and elevated platelet counts, as well as a trend of lower hemoglobin levels and white blood cell counts (Supporting Information: Figure S8). Together, using two independent retrospective data sets, we identified that ZRSR2 mutations are enriched in patients with MF and are typically found in genetically complex MPN. These findings suggest that ZRSR2 mutations alone are insufficient to drive fibrotic progression of MPN. Consequently, functional and mechanistic studies to determine whether ZRSR2 mutations predispose to genetic instability and how they cooperate with epigenetic regulators to promote disease progression in MPN are planned.

In summary, we determine that ZRSR2 loss using CRISPR-Cas9 gene editing causes minor changes in intron retention and transcription; however, this is not sufficient to induce disease progression in preclinical models of JAK2^V617F-driven MPN.

Ranran Zhang: Conceptualization; methodology; data curation; investigation; formal analysis; writing—original draft; visualization; writing—review and editing; validation. Jasmin Straube: Writing—review and editing; formal analysis; data curation; methodology; software; validation. Yashaswini Janardhanan: Investigation. Rohit Haldar: Investigation. Leanne Cooper: Investigation. Noah Hayes: Investigation. Charles Laurore: Investigation. Chulwoo J. Kim: Formal analysis. Marlise R. Luskin: Resources. Robert C. Lindsley: Resources. Maximilian Stahl: Resources. Ann Mullally: Conceptualization; writing—review and editing; funding acquisition; resources; project administration. Anna E. Marneth: Methodology; investigation; data curation; formal analysis; writing—review and editing; conceptualization; validation; project administration; resources. Megan Bywater: Conceptualization; methodology; supervision; writing—review and editing; project administration; data curation; investigation. Steven W. Lane: Conceptualization; supervision; funding acquisition; project administration; resources; writing—review and editing; methodology.

A.M. has received research funding from Morphic and Incyte and has consulted for Cellarity. M.J.B. has received research funding from BMS and Cylene Pharmacueticals for unrelated projects. S.W.L. has received research funding from BMS for unrelated projects and has consulted for Novartis, AbbVie, and GSK.

A.M. acknowledges funding from NIH NHLBI (R01HL131835) and the Department of Defense Congressionally Directed Medical Research Programs (W81XWH2110909). A.E.M. acknowledges funding from the US Department of Defense (Horizon Award W81XWH-20-1-0904). S.W.L. acknowledges funding from NHMRC investigator grant (1195987). Open access publishing facilitated by The University of Queensland, as part of the Wiley - The University of Queensland agreement via the Council of Australian University Librarians.