Sipa1 deficiency–induced bone marrow niche alterations lead to the initiation of myeloproliferative neoplasm

Normal hematopoiesis is well maintained by a rare population of hematopoietic stem cells (HSCs) residing in a specific bone marrow (BM) microenvironment, called niche.¹  The HSC fates are determined by both the extrinsic cues emanating from their niche and the intrinsic signals triggered by interactions with the niche cells via direct cell adhesion and secreted factors.²  The BM HSC niche is composed of various types of stromal cells, including osteoblasts, adipocytes, macrophages, megakaryocytes (MKs), perivascular cells, endothelial cells, and mesenchymal stem cells (MSCs).^2,3  MSCs are considered the precursor of osteoblasts, adipocytes, and chondrocytes.⁴  They can be functionally estimated by their ability to generate colony-forming unit-fibroblast (CFU-F) in vitro and are proposed to give rise to mesenchymal progenitors (MPCs) with single- or bi-lineage potential, but with no/little CFU-F activity.^3,5 

There is increasing evidence that BM niche alterations lead to the development of myeloid malignancies.^6,7  Mice deficient for retinoic acid receptor γ developed myeloproliferative neoplasm (MPN)-like disease, which was induced solely by the gene loss in the microenvironment.⁸  Deletion of Dicer1 from mouse BM osteoblast progenitors caused myelodysplasia (MDS) that could evolve to acute myeloid leukemia (AML).⁹  In addition, loss of Notch signaling in the BM niche led to lethal MPN-like disease.¹⁰  A recent study revealed the critical contribution of Ptpn11 mutations in BM MPCs to leukemogenesis.¹¹ 

Signal-induced proliferation-associated gene 1 (Sipa1), a principal RAP1 GTPase-activating protein, regulates signaling of integrins, growth factors, and cytokines by inactivating RAP1.^12-14 Sipa1 is expressed in mouse hematopoietic stem and progenitor cells (HSPCs) and human lymphocytes.^15,16  Loss of Sipa1 leads to constitutive hyperactivation of RAP1, cell proliferation, and development of malignancy.^14,17,18  Mutations or abnormal expression of SIPA1 have been reported in hematopoietic malignancies and solid cancer in humans.^19-21 SIPA1 gene point mutations were identified in patient mononuclear cells with juvenile myelomonocytic leukemia,²²  a childhood MDS/MPN,²³  and AML.²⁴ Sipa1^−/− mice show normal hematopoiesis at a younger age (before 5 months), but develop MPN, resembling human chronic myeloid leukemia (CML) or MDS, from 1 year of age.^15,25  However, it is unclear whether the MPN is caused by Sipa1 loss in hematopoietic cells or in BM stromal cells.

We here report that Sipa1 is expressed in BM stromal cells and downregulated in these cells from patients with MPN or MDS/MPN. Sipa1 deficiency in mice induces significant alterations in the BM niche prior to the initiation of MDS/MPN. Importantly, the altered Sipa1^−/− BM microenvironment is absolutely required for the MDS/MPN development in the Sipa1^−/− mice. Sipa1 loss confers greater capacity on BM MSCs and MPCs to promote myelopoiesis. The dysregulated cytokine signaling in the Sipa1^−/− BM niche may be involved in the pathogenesis of the disease.

Sipa1^−/− C57BL/6 mice¹⁵  were backcrossed to C57BL/6 for >15 generations. The Sipa1^−/− mice at 8 to 12 weeks and 16 to 17 months were used for experiments. Age- and sex-matched Sipa1^+/+ C57BL/6 mice were used as controls. CD45.1 B6.SJL-Ptprca Pepcb/BoyJ (Jackson Laboratory, Bar Harbor, ME) mice were used as recipients or donors for transplantations. All the mice were maintained in a specific pathogen-free condition in the animal facility at Karolinska Institute. Animal procedures were performed with approval from the local ethics committee (ethical number S40-14) at Karolinska Institute (Stockholm, Sweden).

BM samples were collected from patients (49-78 years old) with CML, chronic neutrophilic leukemia (CNL), or chronic myelomonocytic leukemia (CMML), classified as MPN and MDS/MPN,^23,26  at diagnosis and from the age-matched (45-86 years old) healthy donors. The experiments were approved by the local Ethical Committee at Stockholm (2012/4:10 and 2013/3:1) and informed consent was obtained from the patients and healthy donors.

Human and mouse MSCs were isolated as described.^27,28  See supplemental Data for details.

CFU-F assay of freshly sorted BM MSCs or unfractionated cells was performed as described.^27,28 

BM cells from 7- to 10-week-old Sipa1^+/+ (CD45.1⁺ or CD45.2⁺) or Sipa1^−/− (CD45.2⁺) mice were enriched by magnetic-activated cell sorting using CD45 microbeads (Miltenyi Biotec). Three million of the cells per mouse were injected via tail vein into sublethally (6 Gy) or lethally (9.5 Gy) irradiated young (8- to 10-week-old) CD45.2 Sipa1^−/− and Sipa1^+/+ recipient mice. When transplanting the Sipa1^−/− cells, 8- to 10-week-old CD45.1 Sipa1^+/+ mice (lethally irradiated) were used as recipients. At the endpoints (7-9 months) of the primary transplantations, 3 million spleen cells from the primary Sipa1^−/− recipients (CD45.2), which had developed MDS/MPN, were transplanted into sublethally irradiated Sipa1^+/+ recipients (CD45.2). The analysis of hematopoiesis after transplantation is described in supplemental Data.

Total RNA was isolated from sorted cells, and libraries were constructed using NuGEN’s Ovation Ultralow Library systems (NuGEN Technologies, San Carlos, CA) and were subsequently subjected to 50 cycles of HiSequation 2000 or 76 cycles of NextSeq500 sequencing (Illumina, San Diego, CA). See supplemental Data for more details and data analysis.

The unpaired Student t test or Mann-Whitney U test, Welch’s correction, and Kolmogorov-Smirnov test were used to compare the differences based on the data distribution. The Kaplan-Meier survival curve of the mice was generated by Prism 6.0. All reported P values were obtained using Prism 5.0 or 6.0, and P < .05 was considered statistically significant.

See additional methods in supplemental Data.

Previous studies have shown that Sipa1 was expressed in hematopoietic progenitors and lymphoid cells.^15,16 Sipa1 expression in BM nonhematopoietic cells is unclear. Analysis of the microarray data from our previous studies^28,29  revealed that SIPA1 was also expressed in human BM MSCs, and mouse BM MSCs expressing early B-cell factor 2 (Ebf2)²⁷  (Figure 1A-B), a recently identified MSC population²⁷  that is partly overlapping with the Nestin⁺ MSCs.³⁰  To further determine the Sipa1 gene expression in different mouse BM stromal cell fractions, we performed quantitative real-time polymerase chain reaction (qPCR) analysis on FACS-sorted BM endothelial cells (CD45⁻LIN⁻CD31⁺), MSCs (CD45⁻LIN⁻CD31⁻CD44⁻CD51⁺SCA1⁺),^28,29  and MPCs (CD45⁻LIN⁻CD31⁻CD44⁻CD51⁺SCA1⁻), which contain most of the CXCL12-abundant cells.^5,31  We detected Sipa1 gene expression in all the stromal cell subsets, with the highest Sipa1 expression in the endothelial cells (Figure 1C-D). Interestingly, SIPA1 expression was significantly reduced in BM endothelial cells (P = .0027) of patients with CML, CNL, or CMML compared with age-matched controls, and to a lesser extent reduced in the MSCs (Figure 1E).

To investigate the contribution of the BM niche to the development of MDS/MPN, we characterized the BM niche in the 16-month-old Sipa1^−/− mice. Consistent with the previous report,¹⁵  the majority of the Sipa1^−/− mice developed MDS-like MPN, which can be categorized as MDS/MPN.³²  The disease was manifested by anemia, thrombocytopenia, increased granulocytes, pronounced B-lymphopenia, and splenomegaly (occurring in ∼28.6% of the female mice) (Figure 2A-C; supplemental Figure 1), macrocytic erythrocytes, and increased neutrophils in the peripheral blood (PB) (supplemental Figure 1D-E). We also observed an increased leukocyte infiltration concomitant with MK hyperplasia in the Sipa1^−/− BM (Figure 2D-E). The hyperplastic MK formed many loose clusters and showed smaller hypolobated, pyknotic, or fragmented nuclei and abnormal cytoplasmic morphology. There is a gender bias of the disorders toward the female Sipa1^−/− mice (Figure 2B; supplemental Figure 1A-C).

To explore any potential BM stromal cell alterations in these mice, we dissected the BM stromal cell compartment by FACS. The frequencies of the MSCs and the mature stromal cells were significantly reduced, whereas the MPCs were increased in the aged Sipa1^−/− mice (Figure 2F-G). The reduction of the MSCs was confirmed by the decreased number of CFU-Fs, reflecting the functionally defined MSCs, in Sipa1^−/− mouse BM (Figure 2H left). Consistent with the observation of a female bias of the MDS/MPN, the reduced CFU-F activities were mainly detected in the MSCs from the Sipa1^−/− female mice (Figure 2H right). The frequencies of the endothelial cells remained unchanged (data not shown). Moreover, the Sipa1^−/− MSCs showed increased adipogenic and chondrogenic, but impaired osteogenic differentiation potential (Figure 2I). Micro-computed tomography (μCT) showed slightly reduced femoral bone volume in the aged Sipa1^−/− mice (Figure 2J-K). These data suggest phenotypic and functional alterations of BM cellular niches in the aged Sipa1^−/− mice with MDS/MPN.

The discovery of Sipa1 expression in the BM stromal cells and the BM niche alterations in the aged Sipa1^−/− mice raised a question of whether the MDS/MPN in the aged Sipa1^−/− mice was attributable to the Sipa1 deletion from the BM stromal cells. To answer this question, we transplanted BM hematopoietic cells from young adult Sipa1^−/− mice into lethally irradiated wild-type mice (Figure 3A). Surprisingly, we could not detect any signs of hematological disorders (Figure 3). These data clearly suggest that the loss of Sipa1 in the hematopoietic cells is not sufficient to initiate the MPN.

The absence of MPN after transplantation of Sipa1^−/− hematopoietic cells into a wild-type environment pointed to the potential critical contribution of the Sipa1^−/− BM niche to the MPN. To test this, we analyzed the BM cellular niche components in young (2-3 months old) Sipa1^−/− mice where no abnormal hematopoiesis was observed (supplemental Figure 2). FACS analysis indicated that both MSCs and MPCs, but not CD31⁺ cells, were significantly expanded in the young Sipa1^−/− BM (Figure 4A-C). Correspondingly, the CFU-F frequencies in the unfractionated BM cells and the sorted MSCs from the Sipa1^−/− mice were increased (Figure 4D). The Sipa1^−/− MSCs displayed increased adipogenic, but reduced osteogenic differentiation potential (Figure 4E). μCT analysis revealed a reduced femoral bone volume (Figure 4F-G left) in the young Sipa1^−/− mice compared with that in the age-matched Sipa1^+/+ mice, supporting the finding of impaired osteoblast differentiation capacity of the Sipa1^−/− MSCs.

To evaluate the hematopoiesis-supportive function of different Sipa1^−/− MSCs, MPCs, and endothelial cells, we cocultured these sorted cells with normal Lin⁻SCA1^highKIT^high (LSK) HSPCs (Figure 4H-J). Sipa1^−/− MSCs and MPCs displayed significantly stronger capacity in promoting myeloid cell differentiation, indicated by more granulocyte-macrophage (GM) colonies, but less multilineage (GM and erythrocyte, GME) colonies generated in the cultures with the Sipa1^−/− stromal cells compared with that with Sipa1^+/+ stromal cell counterparts (Figure 4I-J).

Taken together, these data demonstrated the dramatic alterations in composition and function of the BM niche in young Sipa1^−/− mice prior to the onset of the MDS/MPN.

To determine the role of the altered BM niche in the young Sipa1^−/− mice for the initiation of the MPN, we next transplanted the Sipa1^+/+ BM hematopoietic cells into sublethally (6 Gy) irradiated Sipa1^+/+ and Sipa1^−/− recipients (Figure 5A). To avoid introducing donor BM stromal cells, we purified the CD45.1⁺ hematopoietic cells for the transplantation. About 60% to 70% donor-cell engraftment in the PB was achieved in both groups after transplantation (Figure 5B). However, we detected lower WBC, RBC, and hemoglobin counts in the Sipa1^−/− recipient PB compared with that in the Sipa1^+/+ recipients (Figure 5C), indicating ineffective hematopoiesis of the donor cells in the Sipa1^−/− niche. The donor-derived CD11B⁺GR1⁺ granulocytes were increased at ∼7 months after transplantation, whereas the CD19⁺ B cells, but not T cells, were reduced in the Sipa1^−/− recipients compared with that in the Sipa1^+/+ recipients (Figure 5D-E). The natural killer cells were also significantly increased after transplantation in the Sipa1^−/− recipient PB (Figure 5E). Moreover, PB smears of the Sipa1^−/− recipients showed the presence of erythroblasts, reticular erythrocytes, spherocytes, macrocytes, and microcytes in the Sipa1^−/− recipients (Figure 5F-G), supporting the notion of the dyserythropoiesis. In addition, we observed splenomegaly in 10% of the Sipa1^−/− recipients 9 months after transplantation (Figure 5H). The increased myeloid cells, but decreased B cells, were also detected in the Sipa1^−/− recipient BM (supplemental Figure 3A-B). Importantly, we did not observe any significant difference in host-derived hematopoietic lineage distribution between the Sipa1^+/+ and Sipa1^−/− recipients (supplemental Figure 3B), emphasizing little contribution of the host Sipa1^−/− hematopoietic cells to the MDS/MPN.

Expansion of HSPCs is a common phenotype of MPN.^33,34  The frequencies of donor-derived LSK, short-term–HSC (LSKCD34⁺FLT3⁻), granulocyte-macrophage progenitor (GMP; LIN⁻SCA1⁻KIT⁺CD34^highFcγR^high), and MK-erythrocyte progenitor (MEP; LIN⁻SCA1^-KIT⁺CD34⁻FcγR⁻) were increased in the Sipa1^−/− recipient BM 9 months after transplantation (Figure 5I). Meanwhile, the accumulations of CMPs and GMPs were also detected in the Sipa1^−/− recipient spleen (supplemental Figure 3C). However, the frequency of host-derived HSPCs in the Sipa1^−/− recipient spleen remained similar to that in the Sipa1^+/+ recipients (supplemental Figure 3D).

Taken together, transplantation of normal hematopoietic cells into the Sipa1^−/− niche resulted in myeloproliferation along with dysplastic erythropoiesis and leukopenia, which resembles the clinic features of the overlapping syndromes of the unclassifiable MDS/MPN³⁵  and is consistent with MDS/MPN in mice according to the Bethesda criteria³²  and the World Health Organization classification.³⁶ 

To further determine the contribution of the altered Sipa1^−/− niche to the initiation of MDS/MPN and the underlying molecular mechanisms, we examined the hematopoietic activity of normal BM CD45.1⁺ cells after transplantation into lethally irradiated (9.5 Gy) Sipa1^−/− CD45.2⁺ mice. About 98% donor-cell engraftment was obtained in the recipients after transplantation (supplemental Figure 4A). We observed even more severe phenotype of MDS/MPN in these Sipa1^−/− recipients compared with that in the sublethally irradiated recipients, which might be due to the stronger inflammatory response following lethal irradiation. The Sipa1^−/− recipients exhibited reduced survival (Figure 6A), myeloproliferation accompanied with B-cell reduction (supplemental Figure 4B-C), BM hypercellularity (Figure 6B-C), and MK hyperplasia (Figure 6D-E), a characteristic feature of malignant or clonal hematological disorders.³⁷  The frequency of the donor-derived MEPs was increased in the Sipa1^−/− recipient BM compared with that in the Sipa1^+/+ BM (Figure 6F). One of the Sipa1^−/− recipients displayed enlarged lymph nodes (supplemental Figure 4E). Similar to what we observed in the aged Sipa1^−/− mice (Figure 2C), ∼28.6% (2/7) of the Sipa1^−/− recipients developed splenomegaly 7 to 9 months after transplantation. H&E staining illustrated reduced lymphocytic component of white pulp in the Sipa1^−/− recipient spleen (Figure 6G). This was associated with the increased frequencies of CMP, GMP, and MEP in the Sipa1^−/− recipient spleen (Figure 6H-I). The malignant transformation of the normal donor cells in the Sipa1^−/− recipients was indicated by the development of MDS in ∼75% of the secondary recipients after receiving the spleen cells from the primary Sipa1^−/− recipients with MDS/MPN. The secondary recipients exhibited increased mortality, anemia, thrombocytopenia, and splenomegaly 6 months after the transplantation (Figure 6J-K). To explore the molecular mechanisms underlying the pathogenesis of the disease, we performed whole-exome sequencing of the donor cells sorted from Sipa1^−/− and Sipa1^+/+ primary recipients. However, we only detected a few genetic variants in the donor cells from Sipa1^−/− BM and spleen in 1 of the experiments (data are not shown).

To investigate the molecular mechanisms underlying the niche-induced MDS/MPN, we performed RNA sequencing on the native BM MSCs, MPCs, and endothelial cells from young (8-10 week old) Sipa1^−/− and Sipa1^+/+ mice. As expected, gene set enrichment analysis showed that G-protein signal pathways, including Ras and Rap1 signaling, were elevated in the Sipa1^−/− MSCs and endothelial cells (supplemental Figures 5A-B and 6). Proinflammatory cytokines, including transforming growth factor-β (TGF-β) and the IL6/JAK/STAT3 signaling pathways, were significantly increased in the Sipa1^−/− MSCs (Figure 7A). Similarly, Il4 and Mapk in the FcεRI-mediated signaling pathway were also elevated in the Sipa1^−/− MPCs (supplemental Figure 5C). Within the top 25 changed genes in the Sipa1^−/− MPCs, thrombopoietin (Thpo), a growth factor critical for HSC maintenance and the development of MPN, was increased in the Sipa1^−/− MPCs (Figure 7B). The TGF-β and tumor necrosis factor-α (TNF-α) signaling pathways were also enhanced in the Sipa1^−/− endothelial cells (Figure 7C). In addition, the extracellular matrix receptor interactions, wingless-type MMTV integration site family β (WNT-β) catenin, interleukin-2 (IL-2), β-catenin (Ctnnb1), and Notch signaling were enriched in the Sipa1^−/− endothelial cells (supplemental Figure 5D-E).

Consistent with the impaired osteogenic differentiation potential of the Sipa1^−/− MSCs, the expression of runt-related transcription factor 2 (Runx2), a master gene for osteoblast differentiation, was reduced in the Sipa1^−/− MSCs and MPCs (Figure 7D-E). Furthermore, expression of the lymphopoiesis-promoting cytokine Il7 and Dicer1, a microRNA-processing gene, was reduced in the Sipa1^−/− MPCs (Figure 7E).

Cxcl12, Kitl, and Angptl1 are known to be critical for maintaining HSC quiescence.^38-42  The significant downregulations of Cxcl12, Kitl, and Angptl1 were observed in Sipa1^−/− BM MSCs and MPCs, but not in endothelial cells (Figure 7D-F) compared with the Sipa1^+/+ cell counterparts. However, erythropoietin (Epo) expression in the Sipa1^−/− endothelial cells was increased (Figure 7G). Together, these data indicate that Sipa1 loss in BM stromal cells causes molecular alteration of the niche factors that may contribute to the pathogenesis of MPN.

Recent studies have reported that genetic alterations of BM stromal cells can be involved in the pathogenesis of MPN and AML.^2,11  However, the cellular and molecular mechanisms underlying the microenvironment-induced leukemogenesis remain poorly understood. We here demonstrate that Sipa1 deletion results in BM niche alterations leading to the development of MDS/MPN.^32,35  The initiation of the MDS/MPN in this model is dependent on the altered BM niche resulting from the Sipa1 deficiency in BM microenvironment, not in hematopoietic cells. The dysregulated inflammatory cytokines in Sipa1-deficient niche might contribute to the pathogenesis of MDS/MPN.

SIPA1 is a G-protein signaling component of many growth factors and cytokine, including IL-3, granulocyte-macrophage colony-stimulating factor, and CXCL12 in hematopoietic cells.^12-14 SIPA1 point mutations were detected in mononuclear cells of patients with juvenile myelomonocytic leukemia but lacking detectable mutations of other G-protein signaling molecules KRAS, NRAS, and PTPN11.²²  However, the cell origin of the mutation remains unclear. Previous studies showed that the majority of the Sipa1^−/− mice and 15% Sipa1^+/− mice developed CML-like MPN.^15,25  We here observed MDS-like MPN mainly in the aged Sipa1^−/− female mice. This discrepancy in the observed disease phenotype might be due to different genetic background of the mice. The female-biased development of the MDS/MPN could be attributed to the declined bone-forming ability in the aged female mice,^43,44  which could deteriorate the phenotype of the impaired osteoblast differentiation induced by Sipa1 deletion. Nevertheless, the impact of Sipa1 deletion in BM microenvironment on the MDS/MPN remains unexplored.

We here for the first time report SIPA1 expression in human and mouse native BM stromal cells. The importance of Sipa1 expression in BM niche is supported by the abnormal expansion and differentiation of MSCs and MPCs in young Sipa1^−/− mouse BM prior to the initiation of the MDS/MPN, which could contribute to the pathogenesis of the disease. Furthermore, the SIPA1 gene is downregulated in BM MSCs and endothelial cells from patients with CML, CMML, and CNL. This finding merits future study on the impact of SIPA1 gene alteration in BM niche on the treatment outcome in patients with MPN, particularly in the patients who undergo allogenic stem cell transplantation.

The absolute requirement of Sipa1^−/− BM niche for the generation of the MPN in the Sipa1^−/− aged mice is demonstrated by the reciprocal transplantation of normal Sipa1^+/+ hematopoietic cells into irradiated Sipa1^−/− and Sipa1^+/+ mice, or vice versa. The sustained normal hematopoiesis after transplantation of Sipa1^−/− hematopoietic cells in normal microenvironment suggests that the intrinsic loss of Sipa1 gene in hematopoietic cells is not sufficient to induce the MPN. In contrast, the Sipa1^−/− recipients developed lethal MDS/MPN after transplantation of Sipa1^+/+ hematopoietic cells. The features of the disease in the Sipa1^−/− recipients almost completely recapitulated the MDS/MPN in the aged Sipa1^−/− mice. Our data suggest that the MDS/MPN in the aged Sipa1^−/− mice is actually driven by the BM niche, not the Sipa1 deletion in hematopoietic cells. This finding is in contrast to the previous report that Sipa1^−/− hematopoietic cells from the mice with established MPN continued developing CML-like disease after transplantation.¹⁵  This discrepancy could be mainly due to the fact that the donor Sipa1^−/− cells in the previous study were already transformed to be leukemic, and thus, might have developed niche independence, as reported.⁴⁵  Therefore, it is critical that we transplanted Sipa1^−/− hematopoietic cells from young adult mice where hematopoiesis remained normal. The reoccurrence of the MDS-like phenotype in the secondary wild-type recipients supports the notion of the niche-induced malignant transformation of the donor cells. The disease phenotype is likely due to potential epigenetic alterations or selection of clonal hematopoiesis in the donor hematopoietic cells. However, the exact underlying mechanisms require further investigation in the future.

Understanding the molecular mechanisms mediating the niche-induced MPN is fundamental for identifying predicting factors for leukemogenesis. The increased expression of proinflammatory cytokines like TGF-β and TNF-α is a common feature of mouse models with MPN.²  This was also observed in patients with MPN.^46,47  IL6/JAK/STAT3 signaling has been implicated in MPN progression.^48-52  The elevated IL-6, TGF-β, and TNF-α signaling pathways in the Sipa1^−/− stromal cells may contribute to myeloproliferation possibly via altering HSC lineage fate decision in the Sipa1^−/− mice, as previously suggested.^50,53 

THPO-MPL signaling is critical for maintaining HSCs in adult BM⁵⁴  and is involved in the pathogenesis of MPN.^55,56  The upregulation of Thpo in the Sipa1^−/− BM stromal cells may be associated with the enhanced granulopoiesis and the overproduction of MEPs and MK hyperplasia in the aged Sipa1^−/− mice under steady state and after transplantation. The increased number of MKs in BM might in turn promote the progression of the MDS/MPN via platelet factor secretion, as described recently.⁵⁷  In addition, it has been shown that EPO can direct multipotent HSPCs toward committed erythrocyte progenitors.⁵⁸  The increased Epo gene expression in Sipa1^−/− endothelial cells might also contribute to the increase of MEPs.⁵⁷ 

Il7 is critical for maintaining normal lymphopoiesis in adult mice.⁵⁹  The reduced Il7 expression in the Sipa1^−/− MPCs could indirectly lead to myeloproliferation in the Sipa1^−/− recipients after transplantation. Furthermore, Cxcl12, Kitl, and Angptl1 are critical to maintain normal hematopoiesis through keeping HSC quiescence,^5,40,60,61  and the reduction of these genes in BM niche cells were detected in previous MPN mouse models.^29,62  The downregulation of these genes in the Sipa1^−/− MSCs and MPCs could affect HSC retention and quiescence and eventually promote the myelopoiesis. Deletion of Dicer1 from osteoprogenitors leads to the development of MDS and AML.⁹  Thus, the downregulation of Dicer1 in the Sipa1^−/− MPCs, enriched with osteoblast progenitors,³¹  might contribute to the niche alteration and initiation of the MPN in the Sipa1^−/− mice.

In conclusion, we have demonstrated that Sipa1 loss in the BM stromal cells results in BM niche alterations, which are absolutely required for the development of MDS/MPN. The enhanced inflammatory cytokine signaling and myelopoiesis-promoting activities in the Sipa1^−/− BM niche may have contributed to the disease pathogenesis. Our data provide novel evidence for niche-driven MPN and the underlying cellular and molecular mechanisms.

The authors are grateful to Minna Taipale and Andranik Durgaryan at Karolinska Institute and Liseotte Lenner at Linköping University for their technical assistance. All major computations were performed on resources provided by the Swedish National Infrastructure for Computing through Uppsala Multidisciplinary Center for Advanced Computational Science under Project b2014299. The authors acknowledge the MedH Core Flow Cytometry facility (Karolinska Institute) for providing cell-sorting/analysis services.

This study was supported by Swedish Research Council (K2013-99X-22241-01-5), Swedish Childhood Society (TJ2013-0048, PR2015-0142, and PROJ12/081), Swedish Cancer Society (CAN 2009/1589 and CAN 2012/891), Åke Olsson Foundation, Radiumhemmets Forskningsfonder, and Karolinska Institute Wallenberg Institute for Regenerative Medicine (H.Q.), and Knut and Alice Wallenbergs Foundation (M.S.).

Contribution: H.Q. and P.X. designed, performed, analyzed data and wrote the manuscript; M. Dolinska, L.S., M.K., A.-S.J., M. Dimitriou, and T.B. assisted part of the study and manuscript review; Y.Z. performed μCT analysis and wrote the corresponding method in the manuscript; X.L. analyzed RNA sequencing data and wrote the corresponding method in the manuscript; N.M. provided the Sipa1^−/− mice; G.Z.R. provided the morphologic assessment of histologic and cytologic preparations; D.T.S., E.H.-L., and J.W. provided scientific input on the study and manuscript review; M.S. performed the RNA and exome sequencing experiments and provided scientific input, and all authors have read and approved the manuscript.

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

Correspondence: Hong Qian, Center for Hematology and Regenerative Medicine, Department of Medicine, Karolinska Institute, Karolinska University Hospital, SE-141 86 Stockholm, Sweden; e-mail: hong.qian@ki.se.