SRSF2 mutation cooperates with ASXL1 truncated alteration to accelerate leukemogenesis

To the Editor:

Additional sex combs-like 1 (ASXL1) gene is highly mutated in a spectrum of myeloid malignancies, including ~49% of chronic myelomonocytic leukemia (CMML) [1], ~10% of acute myeloid leukemia (AML) [2], ~21% of myelodysplastic syndromes (MDS) [3], ~10% of myeloproliferative neoplasms (MPN) [4], and ~8% of juvenile myelomonocytic leukemia (JMML) [5]. The majority of ASXL1-mutated patients had other concurrent gene mutations, and splicing factors (SRSF2, U2AF1, ZRZR2, SF3B1) were most frequently mutated in myeloid malignancies [6, 7]. Of note, patients with a cooccurring mutation of ASXL1 and splicing factor mutations have a worse prognosis than patients with either mutation alone or without both mutations [8], suggesting a possible synergistic effect of the two mutations in myeloid malignancy progression.

To assess the impact of concomitant alterations of ASXL1 and splicing factors in accelerating the progression and aggressiveness of myeloid malignant, we performed mutual exclusivity analysis using 10 377 myeloid malignancies (https://www.cbioportal.org/) for ASXL1 and splicing factors mutations. We found significant mutation co-occurrence between ASXL1 and SRSF2, U2AF1, or ZRSR2 (log2 Odds Ratio: 1.974, 1.755, 1.177, respectively) (Fig. 1A, Supplementary Table S1). SRSF2 is most frequently co-mutated with ASXL1 (SRSF2 mutation in 28.07% ASXL1-mutated patients). In addition, patients with both ASXL1 and SRSF2 mutations had unique genetic characteristics and worse survival than patients with ASXL1 mutation only, SRSF2 mutation only, and neither (3 323 treatment-naive MDS samples [9], Supplementary Table S2, Fig. 1B, Supplementary Fig. 1A).

Fig. 1: Srsf2^P95H/+ mutation exacerbates Asxl1^Y588XTg-induced leukemogenesis.

A Mutual exclusivity analysis of ASXL1 mutation and splicing factors mutation (SRSF2, SF3B1, U2AF1 and ZRSR2). All 10 377 samples with myeloid malignancies collected in cBioPortal were used. B Overall survival analysis for 3 323 treatment-naive MDS samples, which were divided into four genotypes (Kaplan–Meier curves with log-rank test). C Survival analysis for the mice with different genotypes (Kaplan–Meier curves with log-rank test). The follow-up time is 800 days from the final pIpC injection. D For each genotype, the distribution of disease types (leukemia or MPN, MDS/MPN) in all diseased mice. E Timeline of disease progression in diseased Asxl1^Y588XTg mice and Asxl1^Y588XTg;Srsf2^P95H/+ mice. The red and blue triangle indicates that the onset type is AML and MPN, MDS/MPN, respectively. F PB counts showing the numbers of WBCs, neutrophils, lymphocytes, red blood cells, and platelets in WT, Asxl1^Y588XTg, Srsf2^P95H/+ and Asxl1^Y588XTg;Srsf2^P95H/+ mice. G Representative H&E stained femur sections are shown. Scale bar, 1 mm (top); 100 μm (bottom). *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

ASXL1 is mainly mutated in the last exon in the form of nonsense or frameshift, resulting in C-terminally truncated mutant proteins, and its mutations are always associated with aggressive disease and poor prognosis [1, 3]. To further decipher the impact of SRSF2 mutation on disease progression in ASXL1-mutated malignancies, we next crossed the Asxl1^Y588XTg [10] with Mx1Cre⁺;Srsf2^P95H/+ mice [11] to generate Asxl1^Y588XTg;Mx1Cre⁺;Srsf2^P95H/+ mice. The mutation of Srsf2 (Srsf2^P95H/+) was induced by polyinosine-polycytidine (pIpC) injection (Supplementary Fig. 1B-D) [10, 11]. Asxl1^Y588XTg;Srsf2^P95H/+ mice had a significantly shorter survival rate and a higher rate of myeloid leukemogenesis (72.22%) compared to Asxl1^Y588XTg, Srsf2^P95H/+ and WT mice (Fig. 1C–E). The AML onset time of Asxl1^Y588XTg;Srsf2^P95H/+ mice was 21.3 (13.1–27.2) months with a blast percentage of 43.65% (23.50–58.82%) in bone marrow (BM) (Supplementary Fig. 1E). While the peripheral blood (PB) counts revealed a comparable overall number of white blood cells among the four genotypes of mice, Asxl1^Y588XTg;Srsf2^P95H/+ mice had higher neutrophil and platelet counts and lower lymphocyte and red blood cell counts compared to other genotypes of mice (Fig. 1F, Supplementary Fig. 1F). Histologic analysis of the femur, spleen, and liver sections of Asxl1^Y588XTg;Srsf2^P95H/+ mice demonstrated pronounced blast cells and myeloid cell infiltration (Fig. 1G, Supplementary Fig. 1G). Analysis of BM cytospin preparations also revealed increased blast cells in Asxl1^Y588XTg;Srsf2^P95H/+ mice compared to other genotypes of mice (Supplementary Fig. 2A). Together, these data demonstrated that Srsf2^P95H/+ mutation exacerbates Asxl1^Y588XTg-induced leukemogenesis.

The dysfunctional behavior of hematopoietic stem/progenitor cells (HSC/HPCs) stands as a principal factor in leukemogenesis. Flow cytometric analyses revealed increased frequencies of Lin⁻Sca1⁺cKit⁺ (LSK) cells and long-term (LT)-HSC in the BM of Asxl1^Y588XTg;Srsf2^P95H/+ mice compared to other groups of mice (Fig. 2A, B). Furthermore, the frequency of the myeloid population (Gr1⁺/Mac1⁺) was significantly increased in the BM of Asxl1^Y588XTg;Srsf2^P95H/+ compared with Asxl1^Y588XTg and WT mice (Fig. 2C). MPO staining of spleen sections confirmed myeloid cell enrichment in Asxl1^Y588XTg;Srsf2^P95H/+ mice (Fig. 2D). In contrast, significantly decreased frequencies of CD71⁺/Ter119⁺ erythroid cells in the BM, CD4⁺ cells, CD8⁺ cells, and B220⁺ cells in the spleen were found in Asxl1^Y588XTg;Srsf2^P95H/+ mice (Supplementary Fig. 3A–E). These results indicate that Srsf2 mutation in Asxl1^Y588XTg mice increases the HSC pool and promotes more severe biased myeloid commitment. To further identify the mechanisms of Asxl1^Y588XTg;Srsf2^P95H/+-induced leukemogenesis, we carried out RNA-sequencing on sorted LSK from WT, Asxl1^Y588XTg, Srsf2^P95H/+ and Asxl1^Y588XTg;Srsf2^P95H/+ BM cells (n = 4 for each genotype, five months after pIpC injection). A significant difference in the transcriptome profile was observed amongst Asxl1^Y588XTg;Srsf2^P95H/+ mice and Asxl1^Y588XTg, Srsf2^P95H/+ LSK cells (Supplementary Fig. 4A), although several AML-associated pathways, such as HOXA9/MEIS1 targets and MYC pathway, were significantly upregulated in all three genotypes compared with WT mice (Supplementary Fig. 4B). 339, 450, and 1 307 differentially expressed genes (DEGs) were identified in Asxl1^Y588XTg, Srsf2^P95H/+, and Asxl1^Y588XTg;Srsf2^P95H/+ mice, respectively (|fold change | > 2 & FDR < 0.05). Although most DEGs of Asxl1^Y588XTg and Srsf2^P95H/+ mice were found in Asxl1^Y588XTg;Srsf2^P95H/+ mice, 48.75% up-regulated genes and 76.26% down-regulated genes in Asxl1^Y588XTg;Srsf2^P95H/+ cells were specifically identified such as Meis2, Sox18, and Id3 (Fig. 2E, Supplementary Fig. 4C, D). Scoring the pathways among all samples revealed a specific upregulation of HSC, AML, and megakaryocyte-related pathways in Asxl1^Y588XTg;Srsf2^P95H/+ LSK cells (Fig. 2F). Regardless of SRSF2 being an important splicing factor, we did not identify significantly differential splicing abnormalities in Asxl1^Y588XTg;Srsf2^P95H/+ and Srsf2^P95H/+ cells (Supplementary Fig. 4E). These data suggested that the co-existence of SRSF2 ^P95H and ASXL1 ^aa1-587 induced a malignant signature, which leads to the dysregulation of HSC/HPCs.

Fig. 2: Co-existence of Srsf2^P95H/+ and Asxl1^Y588XTg mutation alters the function of HSC/HPCs.

A Flow cytometric analysis of HSC/HPCs in BM cells from representative mice of each genotype and quantification of the percentages of LSK and LKS^- cells. B Flow cytometric analysis of LSK cells in BM cells from representative mice of each genotype and quantification of the percentage of LT-HSC and ST-HSC. C Flow cytometric analysis of myeloid cells in BM cells from representative mice of each genotype and quantification of the percentage of Gr1⁺/Mac1⁺ cells. D Representative MPO staining of spleen sections is shown. Scale bar, 50 μm. E The overlap of DEGs of Asxl1^Y588XTg, Srsf2^P95H/+ and Asxl1^Y588XTg;Srsf2^P95H/+ mice. F GSVA score distribution of representative pathways among all four genotypes (scaled among all 16 samples). *P < 0.05; **P < 0.01; ***P < 0.001.

In summary, this study demonstrated that co-occurring mutations of Asxl1 and Srsf2 accelerate the development and enhance the severity of myeloid malignancies. Although the proportion of monocytes in the PB of Asxl1^Y588XTg;Srsf2^P95H/+ mice is not significantly distinct from Asxl1^Y588XTg and Srsf2^P95H/+, it is significantly higher than that of WT, which is consistent with the report of monocytic differentiation in ASXL1 and SRSF2 double-mutated AMLs by Johnson et al. [12]. Mechanistically, the Asxl1^Y588XTg;Srsf2^P95H/+ induces an increase in the HSC/HPC pool and a biased commitment to myeloid lineage, along with upregulated HSC and AML-associated malignant signature in double mutated mice. Future studies of the contribution of alternative splicing to leukemogenesis in aged Asxl1^Y588XTg;Srsf2^P95H/+ mice are warranted.