Srsf2 P95H delays Jak2V617F-induced myelofibrosis.
Bone marrow (BM) cells collected from Srsf2P95H, Jak2V617F, Jak2V617F / Srsf2P95H and Scl-Cre mice were engrafted into lethally irradiated congenic recipient animals and tamoxifen was given five weeks after transplantation (Supplemental Fig. 1A and Fig. 1A). Sequential peripheral blood analyses showed an increase in hemoglobin level (Supplemental Fig. 1B), white blood cell count (Supplemental Fig. 1C) and platelet count (Fig. 1B) as soon as 2 weeks after tamoxifen gavage in Jak2V617F expressing animals. The increase in hemoglobin level and white blood cell count was initially slowed down by Srsf2 P95H co-mutation (Supplemental Fig. 1B and 1C), with Srsf2P95H mostly decreasing B220+ B-cell number among white blood cells (Supplemental Fig. 1D), as previously described in Srsf2P95H/+ animals.19 Within the first 8 weeks after tamoxifen gavage, Jak2V617F -induced thrombocytosis was not affected by Srsf2P95H (Fig. 1B). At later time points, the platelet count decreased in Jak2V617F while being still rising in animals co-expressing Srsf2P95H (Fig. 1C).
BM histopathology detected an increased reticulin fibrosis in Jak2V617F expressing animals as soon as 8 weeks after initiation of tamoxifen gavage (supplemental Fig. 1E), which further increased with time (Fig. 1D). Fibrosis was significantly delayed by Srsf2P95H co-expression (Fig. 1D and 1E). At 8 and 24 weeks after tamoxifen gavage, Srsf2P95H also reduced Jak2V617F-induced splenomegaly and spleen fibrosis (Supplemental Fig. 1F) as well as the two-fold increase in the serum level of TGFβ1 induced by Jak2V617F (Fig. 1G). Together, these experiments provided the unexpected observation that co-expression of Srsf2P95H in mouse hematopoietic cells delayed rather than promoted Jak2V617F induced myelofibrosis.
Srsf2 P95H delays exhaustion of Jak2V617F stem cells in stressful conditions.
We analyzed HSPCs in the BM and spleen of transplanted mice between 16 and 24 weeks after tamoxifen gavage. We observed an increase in the fraction of LSK (Lin− Sca-1+ Kit+), MPP (LSK CD48+ CD150−) and SLAM (LSK CD48− CD150+) cells in animals engrafted with either Jak2V617F or Jak2V617F/Srsf2P95H cells when compared to Scl-Cre and Srsf2P95H, without significant difference between Jak2V617F or Jak2V617F/Srsf2P95H engrafted animals (Fig. 2A).
In competitive experiments, CD45.1 recipient mice were first engrafted with an equal fraction of CD45.2 Jak2V617F cells expressing the green fluorescent protein (GFP) and CD45.2 Jak2V617F/Srsf2P95H GFP-negative cells (Supplemental Fig. 2A). Double mutant cells demonstrated a decreased competitiveness attested by the reduced fraction of GFP-negative SLAM in the BM (Supplemental Fig. 2B) and the time-dependent decrease in GFP-negative peripheral blood cells (Supplemental Fig. 2C).
We also transplanted CD45.1 recipient animals with 30% genetically modified cells mixed with 70% wild-type cells expressing the GFP (Fig. 2B). In accordance with previous report,19 Srsf2P95H alone decreased cell competitiveness in transplanted animals, as indicated by the rapid loss of GFP-negative Ter119+ cells in the peripheral blood. When combined with Jak2V617F, Srsf2P95H slightly delayed the rapid increase in Ter119+ cells induced by Jak2V617F alone (Supplemental Fig. 2D). Analysis of GFP-negative SLAM cells in the BM collected 24 weeks after tamoxifen gavage demonstrated a decrease in the fraction of Srsf2P95H cells while Jak2V617F cells almost completely invaded the SLAM compartment. Srsf2P95H co-mutation significantly reduced Jak2V617F SLAM cell expansion (Fig. 2C).
BM cells isolated from these animals were used to perform serial transplantations in recipient CD45.1 animals (Fig. 2D). The third transplantation was associated with a significant decrease in the survival of animals engrafted with Jak2V617F compared to Jak2V617F/Srsf2P95H containing BM cells (Fig. 2E). After this third transplantation, Srsf2P95H or Jak2V617F SLAM cells had almost disappeared from the BM, contrasting with the expansion of Jak2V617F/Srsf2P95H double mutant cells (Fig. 2F), amplifying an effect already observed after the second transplantation (Supplemental Fig. 2E). Finally, monitoring of peripheral blood cells after the third transplantation showed significantly higher white blood cell and platelet counts in mice engrafted with BM cells containing Jak2V617F/Srsf2P95H compared to Jak2V617F alone cells (Fig. 2G). Together, Srsf2P95H co-mutation preserved Jak2V617F SLAMs from exhaustion in stressful conditions provoked by serial transplantations.
Srsf2 P95H interferes with Jak2V617F-induced megakaryocyte phenotype
Given their acknowledged role in TGFβ synthesis and MF development, we explored the phenotype of megakaryocytes in transgenic animals. Initial analyses were performed 8 weeks post tamoxifen. We quantified megakaryocytes by von Willebrand factor (vWF) staining of BM sections. Jak2V617F induced an increase in the number of BM megakaryocytes as compared to Scl-Cre and Srsf2P95H-transplanted animals, without significant impact of concomitant Srsf2P95H mutation (Fig. 3A). Similarly, electron microscopy did not detect ultrastructural modifications of megakaryocytes according to their mutational status (Supplemental Fig. 3A). Srsf2P95H alone or co-expressed with Jak2V617F induced a slight decrease in the size of megakaryocytes (Fig. 3B). In agreement with decreased megakaryocyte size, Srsf2P95H co-expression with Jak2V617F was associated also with a slight decrease in cell ploidy compared to Jak2V617F megakaryocytes (Fig. 3C and supplemental Fig. 3B). The decreased expression of the thrombopoietin receptor MPL typically observed at the surface of Jak2V617F cells23 was also prevented by Srsf2P95H co-mutation (Fig. 3D).
At later time points, 16–24 weeks post tamoxifen, the number of megakaryocytes measured in vWF-stained BM remained high in both Jak2V617F and Jak2V617F / Srsf2P95H animals, yet this number had decreased in Jak2V617F mice as compared to earlier time points while it was still increasing in double-mutant mice (Fig. 3E and supplemental Fig. 3C). This evolution correlated with platelet count (Fig. 1C), with megakaryocyte erythroid progenitor (MEP) and megakaryocyte progenitor (MKP) fractions in the BM (supplemental Fig. 3D), and with megakaryocyte ploidy (Fig. 3F). Together, Srsf2P95H by itself only slightly decreased megakaryocyte size while it negatively interfered with Jak2V617F-induced longitudinal changes in BM megakaryocyte number and maturation.
Srsf2 P95H down-regulates cell signaling in megakaryocytes
We sequenced RNA of megakaryocytes enriched from Scl-Cre, Srsf2P95H, Jak2V617F and Jak2V617F/Srsf2P95H mouse BM samples. Dimensionnality reduction through principal component analysis (PCA) separated megakaryocytes from Jak2WT and Jak2V617F animals, whatever Srsf2 status (supplemental Fig. 4A). Pairwise differential analyses consistently showed a high impact of Jak2V617F and a more limited impact of Srsf2P95H on gene expression (as defined by a Padj < 0.05) in megakaryocytes. For example, compared to Scl-Cre megakaryocytes, Srsf2P95H mutation changed the expression of 75 genes while Jak2V617F mutation, with or without Srsf2P95H, altered the expression of about 3,000 genes (supplemental Fig. 4B). The upset plot showed that, among the six pairwise analyses, the largest intersection between differentially expressed genes (DEG) was across samples with Jak2V617F mutation (supplemental Fig. 4C).
Megakaryocytes collected from double-mutant Jak2V617F / Srsf2P95H mice clustered separately from single mutant Jak2V617F cells (Fig. 4A), with 71 DEG between these two conditions (Fig. 4B). Gene set enrichment analysis (GSEA) indicated the down regulation of multiple signaling pathways, including JAK-STAT, TNF and TGFβ signaling pathways, in Jak2V617F/Srsf2P95H compared to Jak2V617F cells (Fig. 4C). Comparison of enrichment scores further confirmed the upregulation of signaling pathways in Jak2V617F megakaryocytes and their down-regulation in cells expressing Srsf2P95H, either alone or with Jak2V617F (Fig. 4D). Together, these results indicate that Srsf2P95H reduces the enrichment in signaling pathways observed in Jak2V617F megakaryocytes.
Srsf2 P95H -induced Jak2 exon 14 skipping.
SRSF2 splicing factor recognizes both CCNG and GGNG sequences.25 Using RNA sequences collected from sorted megakaryocytes, we compared the relative enrichment of all four SSNG variants, where S represents C or G, in cassette exons that were differentially spliced upon expression of mutant Srsf2P95H, in the context of wild-type or mutated Jak2. Megakaryocytes expressing Srsf2P95H either alone or in combination with Jak2V617F exhibited an enrichment for CCNG while the other sequences, especially GGNG, were depleted in exons that were included versus excluded, respectively (Fig. 5A). Based on a False Discovery Rate (FDR) < 0.05, we detected a total number of 1,729 differential splicing events in mutated compared to Scl-Cre megakaryocytes. We detected 666 differential splicing events in Srsf2P95H cells. We also identified 587 differential splicing events in Jak2V617F cells, which may reflect the ability of mutant JAK2 to phosphorylate proteins involved in mRNA processing.26 The highest number (n = 1,241) was observed in double mutant Jak2V617F/Srsf2P95H megakaryocytes. One hundred thirty four events (7.7% of total) were common to the three genetically modified cell populations (Fig. 5B).
Given the importance of the JAK/STAT pathway activation in MFs, we focused on differential splicing events that affect genes of this pathway. Of the 147 genes listed in KEGG_JAK_STAT_SIGNALING_PATHWAY, 16 showed abnormal splicing events, with only one common to the three cell types (il15ra gene). One of the two abnormally spliced genes common to Srsf2P95H and Jak2V617F / Srsf2P95H cells, but not detected in Jak2V617F cells, was Jak2 (Fig. 5C, supplemental table 1).
In Srsf2P95H expressing cells, an abnormal skipping of Jak2 exon 14 (Jak2Δex14), which encodes the pseudokinase domain including Valine 617, results in a frameshift (Fig. 5D). Using an isoform-specific RT-qPCR, Jak2Δex14 was detected in LSK (Lin−,Sca+,Kit+) cells and megakaryocytes sorted from the BM of Srsf2P95H and Jak2V617F / Srsf2P95H animals. The highest expression of Jak2Δex14 was detected in double-mutated megakaryocytes (Fig. 5D, Supplemental Fig. 5A). This alternatively spliced Jak2 was no more detected in megakaryocytes when these cells were generated by ex vivo differentiation of LSK in the presence of TPO, suggesting a counter-selection of cells expressing Jak2Δex14 isoform in these culture conditions (Fig. 5D). Importantly, a JAK2Δex14 isoform could be detected also in peripheral blood granulocytes collected from two JAK2V617F MF patients with a co-existing SRSF2P95 mutation, while not being detected in the cells of two healthy donors and three JAK2V617F MF patients without SRSF2 mutation (Fig. 5E).
The presence of a Jak2Δex14 isoform correlated with a decrease in the level of phosphorylated STAT5 measured by flow cytometry in sorted LSK cells (Fig. 5F and supplemental 5B). When transfected in γ-2A cells, Jak2Δex14 isoform was translated into a shortened protein at the expected 66kDa molecular weight (Fig. 5G). Using a dual luciferase assay in γ-2A cells expressing the thrombopoietin receptor MPL, we observed that the shortened protein encoded by JAK2Δex14 (Fig. 5G) did not transduce a P-STAT5 signal in response to thrombopoietin (Fig. 5G and 5H). When JAK2V617F was co-expressed with JAK2Δex14 at various ratio in γ-2A cells, JAK2Δex14 isoform behave like the empty vector, in the absence or presence of thrombopoietin (Fig. 5I). These results suggested a loss of function of JAK2Δex14 that may explain the weaker phenotype observed when Srsf2P95H mutation is combined to Jak2V617F through a quantitative decrease in the functional JAK2V617F (Fig. 5D and supplemental 5A).
Srsf2 P95H down-regulates romiplostim-induced MF
Since Srsf2P95H induces the abnormal skipping of JAK2 exon 14, generating a truncated and inactive protein, we wanted to further validate Srsf2P95 mutation effect on JAK/STAT signaling. Therefore, we used a previously described mouse model of MF induced by subcutaneous administration of high dose romiplostim, a thrombopoietin receptor agonist that generates MPL-mediated hypersignalling. Animals were engrafted with either Scl-Cre or Srsf2P95H bone marrow cells. Three weeks after tamoxifen gavage, they received three weekly administration of high dose (1 mg/kg/day) romiplostim (Fig. 6A). This treatment induced a rapid increase in the peripheral blood platelet count, which was significantly less important in Srsf2P95H engrafted animals (Fig. 6B). All the Scl-Cre engrafted animals developed a MF, which was reduced in Srsf2P95H engrafted animals (Fig. 6C and 6D), an effect that was also observed in the spleen (Fig. 6E). Together, these results validated that Srsf2P95H genotype negatively interfered with MF induction upon JAK/STAT hypersignaling, either induced by Jak2 mutation or MPL stimulation.