SETD2 deficiency accelerates MDS-associated leukemogenesis via S100a9 in NHD13 mice and predicts poor prognosis in MDS

Myelodysplastic syndrome (MDS) is one of the most common myeloid malignancies characterized by bone marrow (BM) dysplasia, inefficient hematopoiesis, cytopenia, and a risk of progression to acute myeloid leukemia (AML).^1,2  Several mouse models have been developed to mimic human MDS, and the NUP98-HOXD13 (NHD13) transgenic mouse is probably the most accurate, in which NHD13 is driven by the Vav1 promoter to express in hematopoietic cells.³  The NHD13 mice show anemia, neutropenia, and lymphopenia with hypercellular/normocellular BM, suggesting ineffective hematopoiesis. It is also an excellent model for the study of MDS progression to leukemia, as approximately one-third of NHD13 mice with MDS develop leukemia, most commonly AML, which mimics disease progression in MDS patients.^3,4  Mutations in epigenetic modifiers, such as DNMT3A, TET2, EZH2, ASXL1, and BCOR, are commonly found in, and contribute to, MDS.^5-9  Thus, MDS appears to be an epigenetic malignancy, consistent with its response to DNA demethylating drugs.^10-12  For example, decitabine, the hypomethylating agent, has been the standard of MDS treatment for more than a decade.¹³ 

We had identified the histone methyltransferase SETD2 as the only enzyme that catalyzes histone H3 lysine 36 trimethylation (H3K36me3) in mammals.^14,15  SETD2 binds to elongating RNA polymerase II and generates H3K36-trimethylated nucleosomes in the gene body regions of active genes, which provides docking sites for many important chromatin regulators, including DNMT3A/B.^14,16  We have studied the role of Setd2 in embryonic development and normal hematopoiesis using genetically modified mouse models. We reported that Setd2 constitutive knockout mice exhibit embryonic lethality related to defects in embryonic vasculargenesis,¹⁵  and we and others also established that hematopoiesis-specific Setd2 deficiency impairs self-renewal and differentiation of hematopoietic stem cells (HSCs).^17,18 

SETD2 has been found to be mutated in different types of acute leukemia, including ∼6% of AML cases and ∼10% of cases of acute lymphoid leukemia.^19-21  A particularly high frequency (22%) of SETD2 mutation is associated with MLL-rearranged leukemia.²⁰ SETD2 genetic polymorphisms are also associated with AML prognosis.²²  In the MLL-fusion–driven mouse BM transplantation models of AML, loss of a single allele of Setd2 or knockdown of Setd2 increases the growth of AML cells and shortens disease latency.^20,23,24  However, it has also been observed that a complete loss of Setd2 (eg, by deletion of both alleles of Setd2) blocks development of leukemia.^23-25  These studies indicate a pivotal but complicated role of Setd2 in the pathogenesis of acute leukemia.

Some studies have shown that NHD13-induced MDS/leukemia is controlled by epigenetic regulators (eg, MLL1, NSL, and p300), transcription factors (eg, p53 and Lmo2/Lyl1), HSC regulators (eg, MSI2 and Flt3), and apoptosis-related proteins (eg, Bcl2, PKR, and Puma). However, the role of SETD2 in the pathogenesis of NHD13 MDS/AML is unknown. Consistent with the above-described evidence implicating Setd2 as a tumor suppressor, we recently found that the Mx1-Cre–driven Setd2-knockout mice develop MDS when aged, indicating the causal function of SETD2 in myeloid malignancies.¹⁷  Both nonsense and frameshift-deletion mutations of SETD2 have been identified in MDS patients.²⁶  Our clinical analysis showed that low expression of SETD2 was associated with shorter survival in MDS patients, and thus we hypothesized that SETD2 plays a vital role in the progression of MDS to AML.

The microarray data, which contains data of 183 MDS patients and 17 control samples, are available at GEO (GSE19429).²⁷  The cDNA, BM paraffin-embedded sections, and BM cells of MDS patients and samples from healthy subjects were collected from Shanghai Children’s Medical Center by J.S.

The NHD13 transgenic mice were purchased from Jackson Laboratory, and the Setd2^flox/flox (Setd2^f/f) mice were bred as described previously.¹⁷  The Mx1-Cre mice and CD45.1 B6SJL mice were purchased from Shanghai Research Center for Model Organisms. The NHD13 mice were crossed with Mx1-Cre/Setd2^f/f or Setd2^f/f mice, and Setd2 was deleted by injection of poly(I:C) into Mx1-Cre/NHD13/Setd2^f/f mice, relative to control NHD13/Setd2^f/f mice. All mice used in the experiments were on a pure C57BL/6 genetic background. The mice were used according to animal care standards, and animal studies were approved by the Committee of Animal Use at Shanghai Institute of Nutrition and Health.

To investigate the role of SETD2 in MDS, we first used microarray-based gene expression profiling data of primary BM CD34⁺ cells from 183 MDS patients²⁷  to learn whether the expression of SETD2 is associated with prognosis. The results showed that, in refractory anemia with excess blasts, a high-risk subtype of MDS, patients with low SETD2 expression had a significantly worse overall survival than the patients with normal SETD2 expression (Figure 1A), whereas the high-SETD2 group showed no significant difference (supplemental Figure 1A; available on the Blood Web site). We then performed quantitative polymerase chain reaction (qPCR) analysis of our MDS patient samples, and the results showed that the expression of SETD2 was significantly lower in the MDS patient samples than in those from healthy subjects (Figure 1B). Given the association of high CD34 expression with short survival in MDS patients (Figure 1C), we further analyzed the microarray data of the MDS patient samples, and we found that the expression of SETD2 correlated negatively with CD34 expression (Figure 1D). SETD2 is the only enzyme responsible for H3K36me3, and we therefore performed immunohistochemical analysis of MDS patient BM samples, using anti-H3K36me3 or CD34 antibodies. The results showed that H3K36me3 correlated negatively with CD34 expression (supplemental Figure 1B-C), consistent with the microarray analysis. In addition, the MDS patients with a low platelet/megakaryocyte ratio showed less SETD2 expression than those with a higher platelet/megakaryocyte ratio (Figure 1E). We did not observe a correlation of low expression of SETD2 with known MDS-associated gene mutations such as NF1, PTPN11, KRAS, NRAS, RUNX1, or SH2B3, in the analyzed MDS patient samples (supplemental Figure 1D). Low expression of SETD2 is associated with short survival in MDS, and we hypothesized that SETD2 expression could be altered by therapeutic drug treatments. To test this hypothesis, we used decitabine, a demethylating agent commonly used in MDS therapy, to treat BM CD34⁺ cells from MDS patients. qPCR analysis showed that SETD2 expression was upregulated significantly by decitabine, albeit with variable fold change, in the MDS samples, but not in the samples from healthy subjects (Figure 1F). The variable fold change was not related to the basic level of SETD2 expression (supplemental Figure 1E). As expected, decitabine significantly reduced the number of colonies produced by the primary MDS cells in the colony-forming unit (CFU) assay (supplemental Figure 1F-G). Thus, the increased expression of SETD2 may indicate a good response to decitabine treatment in MDS patients. Altogether, our clinical sample analysis suggests that low expression of SETD2 indicates a poor prognosis in MDS patients.

To examine whether Setd2 is crucial for MDS progression to AML, we crossed NHD13 mice with Setd2 conditional knockout mice (supplemental Figure 2A-B). The NHD13/Setd2^Δ/Δ mice developed AML with a shorter median survival (267 days) than that of the NHD13 mice (333 days; P < .0001; Figure 2A; supplemental Table 1). The expression of Setd2 decreased significantly in the NHD13/Setd2^Δ/Δ BM cells (Figure 2B), and H3K36me3 was abolished by Setd2 deletion (Figure 2C), whereas the H3K36me2 level was not changed (supplemental Figure 2C). The spleens of the NHD13/Setd2^Δ/Δ mice were enlarged compared with those of the NHD13 and wild-type (WT) mice (Figure 2D-E). Phenotypically, all the NHD13/Setd2^Δ/Δ mice had an increased percentage of c-Kit⁺ cells in the BM and spleen compared with the NHD13 mice (Figure 2F-G). At the leukemia stage, the percentage of Mac-1⁻c-Kit⁺ cells in the NHD13/Setd2^Δ/Δ mice was higher than in the NHD13 mice (Figure 2H), which was consistent with our morphology analysis results showing that the NHD13/Setd2^Δ/Δ AML cells had less myeloid differentiation (supplemental Figure 2D). The leukemic NHD13/Setd2^Δ/Δ mice had enlarged spleens and livers (supplemental Figure 2E) and extensive extramedullary hematopoiesis and infiltration of myeloblasts in multiple organs (supplemental Figure 2F), as well as increased white blood cell (WBC) counts and decreased hemoglobin (HGB) levels compared with age-matched NHD13 mice (Figure 2I). Notably, analyses of the mice at relatively early, nonleukemic stages showed that the deletion of Setd2 could modify the NHD13-driven phenotypes in a short period (supplemental Figure 2G-H), indicating a direct rather than secondary effect. In addition, single-allele deletion of Setd2 could not accelerate leukemia progression in the NHD13 mice (supplemental Figure 3A), although it caused spleen and liver enlargement (supplemental Figure 3B) and a relatively high number of myeloblasts (supplemental Figure 3C-D). These observations imply that Setd2 plays a unique role in NHD13–associated, relative to MLL-fusion–associated, leukemia models.

To further study the effect of Setd2 on MDS-associated leukemia, NHD13 and NHD13/Setd2^Δ/Δ BM cells were transplanted into lethally irradiated recipient mice, and Setd2 was deleted in the recipient mice at 4 weeks after transplantation by poly(I:C) injection (supplemental Figure 4A-B). Consistent with the observations in the primary mice, Setd2 deficiency accelerated the development of AML and shortened survival in the transplant-recipient mice, as well (Figure 2J). Whereas the NHD13 mice died of MDS/leukemia, with a median survival of 284 days, all the NHD13/Setd2^Δ/Δ recipient mice died of AML, with a median survival of 172 days (Figure 2J). The leukemic mice receiving transplants of NHD13/Setd2^Δ/Δ BM cells had significantly enlarged spleens (supplemental Figure 4C) and had more leukemia blast cells in spleen and liver in the histology analysis (supplemental Figure 4D) than did the NHD13-recipient mice.

To further understand the rapid development of leukemia observed in the NHD13/Setd2^Δ/Δ mice, we analyzed the BM HSPC compartment in the mice receiving NHD13 or NHD13/Setd2^Δ/Δ BM cell transplants. We found that loss of Setd2 did not severely impair BM cellularity (Figure 3A; supplemental Figure 5A), but rapidly altered the BM compartment of the recipient mice with a marked expansion of the Lin⁻, Lin⁻c-Kit⁺Sca-1⁻ (LK), and Lin⁻c-Kit⁺Sca-1⁺ (LSK) cells (Figure 3A-C).

To avoid the potential effect of BM ablation caused by irradiation, we analyzed the HSPCs of the primary mice at pre-MDS stage (2 months of age). The results showed that the Lin⁻ and LK cell populations were significantly expanded in the NHD13/Setd2^Δ/Δ mice compared with those in the WT and NHD13 mice (Figure 3A; supplemental Figure 5B-C), whereas the LSK cells seemed not to be dramatically affected (supplemental Figure 5B-C). However, a more detailed analysis by separating the LSK cells into multipotential progenitors (MPPs), long-term HSCs (LT-HSCs), and short-term HSCs (ST-HSCs), based on their Flt3 and CD34 expression levels, showed that the deletion of Setd2 increased MPPs, but not LT-HSCs or ST-HSCs, in the NHD13 mice (Figure 3D-E). This expansion of HSPCs by Setd2 deficiency in the NHD13 mice was in sharp contrast to the previous observations that Setd2 knockout in normal mice decreased HSPCs, including MPP and LSK cells,^17,18  thus suggesting differential regulation of MDS/AML and normal hematopoiesis by Setd2.

To investigate whether the expansion of the NHD13/Setd2^Δ/Δ HSPCs was related to enhanced self-renewal, we used the sorted LSK cells to perform the paired daughter cell assay, which could quantify different types of symmetric and asymmetric cell divisions of individual cells by determining combinations of differentiation potentials of their daughter cells.²⁸  The results showed that the symmetric self-renewal division frequency of the NHD13/Setd2^Δ/Δ LSK cells (38%) was much higher than that of the NHD13 (10%) and WT (10%) LSK cells, whereas the symmetric commitment divisions were dominant in the NHD13 LSK cells (50%) compared with the NHD13/Setd2^Δ/Δ (12%) LSK cells (Figure 3F). Thus, our results suggest an important role of Setd2 in regulating the balance between the stem cell–depleting divisions and the symmetric stem cell self-renewing divisions in the NHD13 mouse model.

We further analyzed the multipotent hematopoietic progenitors, including megakaryocyte erythroid progenitors (MEPs), granulocyte and monocyte progenitors (GMPs), and common myeloid progenitors (CMPs), in the BM of the primary mice and those receiving WT, NHD13, and NHD13/Setd2^Δ/Δ BM cell transplants. We found that the MEPs were most dramatically increased in both the primary mice and NHD13/Setd2^Δ/Δ BM-recipient mice compared with the counterpart NHD13-recipient mice (Figure 4A-B; supplemental Figure 5B-C). The effects of Setd2 deletion on the frequencies and number of GMPs and CMPs were relatively marginal or variable (Figure 4A-B; supplemental Figure 5B-C). Thus, these results suggest that the loss of Setd2 affects erythroid and megakaryocyte differentiation to a greater degree than it affects granulocyte and monocyte differentiation.

To investigate the role of Setd2 in terminal differentiation of the NHD13-expressing cells, we examined erythroid differentiation in the peripheral blood (PB), BM, and spleen of the primary WT, NHD13, and NHD13/Setd2^Δ/Δ mice. Flow cytometry analysis of cells double stained with Ter119 and CD71 was used to distinguish different erythroid developmental stages including proerythroblasts, basophilic erythroblasts, chromatophilic erythroblasts, and orthochromatophilic erythroblasts.^29,30  We observed altered erythroid differentiation stages with increased frequencies of the immature erythroid cells in the PB, BM, and spleen of the NHD13/Setd2^Δ/Δ mice, compared with the WT and NHD13 mice (Figure 4C-D). These results suggest that the deletion of Setd2 inhibits erythroid differentiation of the NHD13-expressing cells, which could accelerate the progression of MDS to leukemia.

To further understand the basis of the expansion and differentiation impairment of the NHD13/Setd2^Δ/Δ HSPCs, we performed a CFU assay, using HSPCs isolated from the transplant-recipient NHD13 and NHD13/Setd2^Δ/Δ mice. We found that, whereas the NHD13 HSPCs formed very few colonies, the NHD13/Setd2^Δ/Δ HSPCs showed enhanced self-renewal and gave rise to an increased number of erythroid colonies (BFU-E; Figure 4E-F). These results suggest that the improved HSPC expansion and the anemia in the NHD13/Setd2^Δ/Δ mice were caused by elevated self-renewal and defects in producing erythroid progenitors. Next, we analyzed the cell cycle status of the NHD13 and NHD13/Setd2^Δ/Δ HSPCs and found that loss of Setd2 increased the percentage of cells in the G2/M phase and reduced cell death (Figure 4G-H; supplemental Figure 5D-E). Altogether, these results suggest that loss of Setd2 accumulates more activated HSPCs and maintains a more aggressive myeloid malignancy.

To examine the effect of Setd2 deficiency on the maintenance of NHD13-driven AML, we performed secondary BM transplantation by transplanting the NHD13-expressing WT or Setd2^Δ/Δ BM cells collected from the primary leukemic mice into sublethally irradiated recipient mice. We found that, although these leukemias were all transplantable, the secondarily transplanted NHD13/Setd2^Δ/Δ group had a significantly shorter survival, with higher WBC and lower red blood cell counts, compared with the NHD13 group (Figure 5A-B). The spleens and livers of the NHD13/Setd2^Δ/Δ leukemic mice were significantly enlarged and massively infiltrated with myeloblasts (Figure 5C-D; supplemental Figure 6A). Flow cytometry analysis showed more c-Kit⁺Gr1⁺ and c-Kit⁺Mac1⁺ cells in the BM and spleen compared with counts in the NHD13 or WT mice (Figure 5E; supplemental Figure 6B-D). We also observed more leukemic blasts in the BM and spleen of the NHD13/setd2^Δ/Δ mice (Figure 5C-F). Thus, loss of Setd2 significantly promoted aggressiveness of NHD13-driven AML, which suggests that Setd2 deficiency accelerates not only the transformation of MDS to AML, but also the progression of leukemia.

In considering that the decreased in H3K36me3 may contribute to leukemia maintenance, we tested a Jumonji histone demethylases inhibitor, JIB-04,³¹  to treat the SKM-1 cell line, which was derived from human MDS-associated leukemia.³²  Western blot analysis showed that the JIB-04 treatment indeed increased H3K36me3 in the SKM-1 cells (Figure 5G). We treated the SKM-1 cells with different doses of JIB-04, and the results showed that JIB-04 significantly inhibited the growth of the SKM-1 cells in a dose-dependent manner (Figure 5H). Flow cytometry analysis showed that the JIB-04 treatment resulted in an increase in apoptosis of the SKM-1 cells (Figure 5I-J). Furthermore, we also used JIB-04 to treat the leukemia cells from the NHD13 and NHD13/Setd2^Δ/Δ mice, and the results showed that NHD13/Setd2^Δ/Δ cells were less sensitive to JIB-04 treatment than were the NHD13 cells (supplemental Figure 6E-F), probably because of the inability of the NHD13/Setd2^Δ/Δ cells to increase H3K36me3. These results suggest that increasing H3K36me3 may be a strategy or an effective indicator in the treatment of MDS-associated leukemia.

To understand how Setd2 deficiency promotes the transformation from MDS to leukemia, we performed comparative RNA-seq analysis of HSPCs from the NHD13 and NHD13/setd2^Δ/Δ primary mice. Notably, a comparison with the previously reported data of Setd2 knockout in WT HSPCs¹⁷  showed that the majority (97.6%) of altered genes by Setd2 deletion in the NHD13 HSPCs were different from those in the WT, suggesting a context dependency of the Setd2 target genes (supplemental Figure 7A-B). Gene Set Enrichment Analysis revealed that at the G2/M checkpoint, NF-κB and hematopoiesis mature gene sets were depleted of NHD13/Setd2^Δ/Δ HSPCs, whereas the HSC gene sets were enriched (Figure 6A), consistent with the phenotypic differences between the AML and MDS stages. Some genes that were dysregulated in the NHD13/Setd2^Δ/Δ HSPCs (eg, S100a9, E2f2, Tcf4, Cd14, Chac1, Spi1, Gata2, Fli1, and Srf) have been shown to play important roles in regulating these pathways (Figure 6B). The altered expression of these genes was validated by qPCR (Figure 6C). Thus, these results further support and provide possible mechanistic explanations for the enhanced self-renewal, impaired differentiation, and cell cycle abnormalities of NHD13/Setd2^Δ/Δ HSPCs.

It has been reported that H3K36me3 can be recognized by Dnmt3a/b, suggesting that Setd2 could be functionally connected with Dnmt3a/b-mediated DNA methylation.^33-35  Thus, we examined the change in DNA methylation by Setd2 deletion in the NHD13 mice. We performed whole-genome bisulfate sequencing (WGBS) analysis of the NHD13 and NHD13/Setd2^Δ/Δ LSK cells, and we found that the NHD13/Setd2^Δ/Δ cells had more hypermethylated loci than the NHD13 cells (2.4-fold; Figure 6D) and that their differentially methylated regions (DMRs) were also more enriched in the hypermethylated loci (1.8-fold; Figure 6E; supplemental Figure 7C). The mean methylation level of the DMRs was also significantly higher in the NHD13/Setd2^Δ/Δ than the NHD13 cells (Figure 6F). Moreover, the DNA methylation profiling showed that the NHD13/Setd2^Δ/Δ genome had decreased CpG methylation levels in the gene body regions but increased levels in the intergenic regions (supplemental Figure 7D-F), which may explain its overall hypermethylation state. Pathway analysis of the MDR-associated genes showed considerable overlaps with the gene expression profiles (supplemental Figure 7G-H). These results demonstrate the altered DNA methylation pattern that may be important for Setd2-regulated gene expression and the NHD13-expressing HSPC functions.

To identify functionally important Setd2 target genes and to understand the regulatory mechanism, we performed ChIP-seq analysis of the c-Kit⁺ cells of the NHD13 and NHD13/Setd2^Δ/Δ mice with an anti-H3K36me3 antibody. A significant decrease in H3K36me3 on the gene body regions of active genes was detected in the NHD13/Setd2^Δ/Δ cells compared with levels in the NHD13 controls (Figure 7A; supplemental Figure 8A-C). Given the important role of the cross talk between H3K36me3 and DNA methylation in leukemogenesis, we also surveyed the dysregulated genes that associate with the DMRs. We found that S100a9, which encodes a calcium-binding protein, was significantly downregulated by Setd2 deletion in the NHD13 mice (Figure 6B-C; supplemental Figure 8D-E), accompanied by a decreased H3K36me3 level (Figure 7B). Meanwhile, S100a9 fell into the category of genes with DMR-associated expression (Figure 7C-D). qPCR analysis of human MDS BM samples also showed a relatively lower expression of S100A9 compared with samples from healthy subjects (supplemental Figure 8F). Overexpression of SETD2 in SKM-1 cells upregulated S100A9 and S100A8 (supplemental Figure 8G) and decreased colony formation by the cells (supplemental Figure 8H-I). Thus, we identified S100a9 as a target gene of Setd2 based on RNA-seq, ChIP-seq, and whole-genome bisulfate sequencing (WGBS) analyses.

S100a9 has been reported to trigger cell death of HSPCs, including MEPs, and to contribute to ineffective hematopoiesis in MDS.^36-38  S100A9 has also been identified as a differentiation inducer in AML.³⁹  Mechanistically, it has been shown that the activation of the JNK and NF-κB signaling pathways is essential in S100A9-induced AML differentiation through Toll-like receptor 4 in leukemia cells.³⁹  We indeed found that the expression of Tlr4 and the Jnk/NF-κB pathway–related genes, such as Jnk and Ikba, was decreased in the NHD13/Setd2^Δ/Δ HSPCs (Figure 7E), which is consistent with the Kyoto Encyclopedia of Genes and Genome analysis showing the MAPK signaling pathway to be significantly regulated by Setd2 in NHD13 HSPCs (supplemental Figure 7G). To understand the role of S100a9 in NHD13/Setd2^Δ/Δ HSPCs, we performed a CFU assay using sorted HSPCs that were treated with 4 μg/mL recombinant mouse S100a9 protein (rmS100a9). We found that the rmS100a9 treatment reduced the promoted expansion of the NHD13/Setd2^Δ/Δ HSPCs, including the capacity of generating various types of colonies (Figure 7F-H). Conversely, knockdown of S100a9 accelerated the progression of AML in the NHD13 mice (supplemental Figure 8G-I). Last, S100A9 could be upregulated in SKM-1 cells by treatment with JIB-04, especially when SETD2 was suppressed by shRNA-mediated knockdown (supplemental Figure 8M-N). Thus, the mimicking and rescue of the phenotypes of Setd2 deficiency in the MDS-associated HSPCs by manipulation of S100a9 suggests that S100a9 downregulation contributes to the increased self-renewal and defective myeloid differentiation in the NHD13/Setd2^Δ/Δ mice.

In this study, low expression of SETD2 predicted a poor outcome in MDS and identified a previously unrecognized function of Setd2 in an NHD13-driven MDS mouse model: controlling the progression of MDS to leukemia. Deletion of Setd2 significantly accelerated development of AML, indicating that Setd2 acts as an important tumor suppressor in the transformation of NHD13-driven MDS to AML. Mechanistically, loss of Setd2 enhanced self-renewal and impaired myeloid differentiation in the NHD13-expressing HSPCs. Early progression of MDS to AML in Setd2-deleted NHD13 mice was also associated with abnormal erythroid differentiation and reduced cell death. Symmetric self-renewal division was increased by Setd2 deletion in the NHD13-expressing HSCs, which may also account for the higher risk for the development of leukemia in the NHD13/Setd2^Δ/Δ mice.⁴⁰  The NHD13/Setd2^Δ/Δ mice developed less differentiated AML, which may reflect differentiation blocking in the early stage and enhanced self-renewal of leukemia-initiating cells, induced by Setd2 deletion.

When MDS progressed to AML, there was increased cell survival and decreased cell death. Loss of Setd2 triggered a unique gene signature in the NHD13-expressing HSPCs at both the MDS and AML stages, including upregulation of the HSC pathway and downregulation of the myeloid cell differentiation pathway. It appears that elevated MAPK signaling induced by Setd2 deletion in the NHD13 BM cells confers a growth advantage to the preleukemic blasts. We identified S100a9 as a target gene of Setd2 in the NHD13 leukemic cells. S100A9 and S100A8 have been implicated in the pathology of RPS14-associated MDS and regulation of myeloid and erythroid differentiation,^39,41  and, in particular, S100A9 can induce AML differentiation, whereas S100A8 inhibits differentiation induced by S100A9.³⁹  In the NHD13 MDS model, S100a9 acts to prevent the transformation to AML. Although S100a9 and S100a8 are highly homologous, these 2 proteins may have opposite functions in NHD13-mediated leukemogenesis.

We observed different effects of Setd2 deletion in the HSPCs of MDS/AML relative to that of normal hematopoiesis. Setd2 exerts cellular context–specific effects in HSPCs, as the decrease in LSK and MPP cells was found in the Setd2^Δ/Δ mice, but not in the NHD13/Setd2^Δ/Δ mice. The NHD13/Setd2^Δ/Δ mice had an increased WBC count and decreased HGB, but the Setd2^Δ/Δ mice had a decreased WBC count and normal HGB. Intriguingly, the Setd2-regulated genes including S100a9 are significantly different in the NHD13 and WT cells, which may cause the genetic background–specific effects of Setd2. Thus, our experiments showed that Setd2 deletion–induced transformation differs in WT and NHD13 HSPCs.

Altogether, our results suggest that deficiency in Setd2 function promotes the progression of MDS to AML. Moreover, the low expression of SETD2 correlates strongly with the survival of patients with MDS. Our findings indicate that promoting SETD2 function or activating downstream effects of SETD2 may be helpful in elimination of MDS-associated leukemia.

The authors thank Dangsheng Li for critical reading and helpful comments on the manuscript and Y. Zhai, X. Miao, S. Yan, Y. Chen, K. Wang, and Y. Wang (Shanghai Institute of Nutrition and Health core facilities) for technical support.

This work was supported by National Key Research and Development Plan of China grants 2018YFA0107200, 2018YFA0800203 (L.W.), and 2018YFA0107802 (X.-J.S.); National Natural Science Foundation of China (NSFC) General Program grants 81670122, 81970150 (L.W.), and 81670094 (X.-J.S.); NSFC Excellent Young Scholar Program grant 81622003 (L.W.); CAS Bureau of Major R&D Program grant XDA12020376 (L.W.); the CAS Bureau of Frontier Sciences and Education Program grant QYZDBSSWSMC027 (L.W.); Shanghai Municipal Education Commission-Gaofeng Clinical Medicine grant 20152506 (X.-J.S.); Shanghai Collaborative Innovation Program on Regenerative Medicine and Stem Cell Research 2019CXJQ01 (S.-J.C., X.-J.S.); Innovative Research Team of High-Level Local Universities in Shanghai (X.-J.S.); Shanghai Guangci Translational Medical Research Development Foundation; and the Samuel Waxman Cancer Research Foundation.

Contribution: L.W., X.-J.S., and B.-Y.C. designed the research, analyzed the data, and wrote the paper; B.-Y.C., J.S., C.-L.H., S.-B.C., Q.Z., J.-C.W., D.H., M.S., N.L., P.-C.Y., P.L., J.-Y.Z., and R.-F.D. performed the research; and S.-J.C., Z.C., S.D.N., Q.Z., C.-H.X. Y.-L.Z., S.-J.Z., F.L., L.-J.Z., and Q.-H.H. analyzed the data and contributed new tools.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Lan Wang, Shanghai Institute of Nutrition and Health, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200031, China; e-mail: lwang@sibs.ac.cn; and Xiao-Jian Sun, Shanghai Institute of Hematology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai 200025, China; e-mail: xjsun@sibs.ac.cn.