The spleen microenvironment influences disease transformation in a mouse model of KITD816V-dependent myeloproliferative neoplasm

Activating mutations leading to ligand-independent signaling of the stem cell factor receptor KIT are associated with several hematopoietic malignancies. One of the most common alterations is the D816V mutation. In this study, we characterized mice, which conditionally express the humanized KITD816V receptor in the adult hematopoietic system to determine the pathological consequences of unrestrained KIT signaling during blood cell development. We found that KITD816V mutant animals acquired a myeloproliferative neoplasm similar to polycythemia vera, marked by a massive increase in red blood cells and severe splenomegaly caused by excessive extramedullary erythropoiesis. Moreover, we found mobilization of stem cells from bone marrow to the spleen. Splenectomy prior to KITD816V induction prevented expansion of red blood cells, but rapidly lead to a state of aplastic anemia and bone marrow fibrosis, reminiscent of post polycythemic myeloid metaplasia, the spent phase of polycythemia vera. Our results show that the extramedullary hematopoietic niche microenvironment significantly influences disease outcome in KITD816V mutant mice, turning this model a valuable tool for studying the interplay between functionally abnormal hematopoietic cells and their microenvironment during development of polycythemia vera-like disease and myelofibrosis.

Several KIT mutations have been described that cause constitutive receptor activation without ligand binding. The D816V substitution is one of the most commonly described mutations associated with hematopoietic neoplasia 8,9,18 . We previously described the generation of a humanized transgenic mouse model for conditional KIT D816V expression and analyzed effects of KIT D816V signaling on fetal liver erythropoiesis 19 . Here, we used R26-LSL-KIT D816V mice to investigate sustained KIT D816V signaling in the adult hematopoietic system and found development of a myeloproliferative neoplasm (MPN) reminiscent of polycythemia vera (PV), which was transplantable and characterized by massively increased red cell mass and splenomegaly. Furthermore, stem cells were mobilized from BM to the spleen. Splenectomy of KIT D816V mutants prevented the increase in red cell mass but promoted BM failure and myelofibrosis, clinical features observed upon transformation of PV to post polycythemic myeloid metaplasia. The fact that course of disease in KIT D816V mutants is influenced by splenectomy demonstrates the relevance of the niche and provides a unique model to study the interdependency of hematopoietic cells and the microenvironment.

Results
KIT D816V induces a polycythemia vera-like disease. We previously described the generation of the R26-LSL-KIT D816V mouse line, harboring a conditional knock in of a humanized KIT D816V receptor linked to a green fluorescent protein (GFP) in the ROSA26 genomic locus. The D816V mutation has been implicated in the pathology of acute myeloid leukemia, mastocytosis and other oncogenic malignancies [7][8][9]18,20 . To extend the knowledge on how KIT regulates hematopoiesis and contributes to myeloproliferative disorders, we studied the effects of ectopic KIT D816V expression in the adult hematopoietic system. We mated R26-LSL-KIT D816V with HSC-SCL-Cre-ER T mice, which express a tamoxifen-inducible Cre recombinase under control of the stem cell enhancer of the Scl gene locus 21 . HSC-SCL-Cre-ER T -mediated recombination has been demonstrated in HSCs/HPCs and endothelial cells. Double transgenic HSC-SCL-Cre-ER T :R26-LSL-KIT D816V animals (hereafter called HSC-SCL:KIT D816V ) were viable and developed normally. For induction of KIT D816V expression (Fig. 1a), we treated adult HSC-SCL:KIT D816V mice with a daily dose of 1.5 mg tamoxifen (TX) for 5 consecutive days. TX-treated wildtype and single transgenic littermates were used as controls. Quantitative real-time PCR validated KIT D816V expression in hematopoietic compartments of HSC-SCL:KIT D816V animals after induction, with transcript levels comparable to endogenous Kit expression in controls. To validate GFP co-expression, GFP-positive and -negative fractions were also analyzed for KIT D816V expression ( Supplementary Fig. S1).
In total, 44 HSC-SCL:KIT D816V and 45 control mice were monitored for 4-10 weeks after treatment. Only a limited group of 2 HSC-SCL:KIT D816V and 4 control animals was monitored up to 18 weeks, as we observed a high rate of spontaneous mortality for HSC-SCL:KIT D816V animals (29.55%; 13/44 mice) within the first 10 weeks after induction. Aside from moderate enlargement of the abdomen in some cases, HSC-SCL:KIT D816V mice showed no signs of morbidity before death. From the control group, all animals survived the observation period.
Cell counts in peripheral blood (pB) were analyzed 4, 8, 10 and 18 weeks after induction. The red blood cell count (RBC), hemoglobin (Hb), mean platelet volume (MPV) and white blood cell count (WBC) were significantly elevated in HSC-SCL:KIT D816V mice compared to controls (Fig. 1b, Supplementary Fig. S2). The hematocrit (Hct) was initially elevated to high levels and slightly decreased thereafter. The mean corpuscular volume (MCV) and the platelet count (PLT) were decreased in HSC-SCL:KIT D816V mice.
We further investigated alterations in pB cell lineages via flow cytometry. Analyses demonstrated elevated amounts of CD45-positive cells in pB of HSC-SCL:KIT D816V mice, caused by increased monocyte and B-cell populations and a mild increase in granulocytes. Staining for erythroid markers CD71 and Ter-119 22 revealed the presence of CD45-negative erythroblasts, which are normally not released into circulation (Fig. 1c, Supplementary Fig. S2). In addition, the reticulocyte frequency was increased ( Supplementary Fig. S2). Figure 1d shows pB smears of HSC-SCL:KIT D816V and control mice 8 weeks after induction. Accumulation of erythrocytes and mature myeloid cells is a hallmark of PV 23 . PV furthermore often includes thrombocytosis, but similar to murine PV models harboring the JAK2 V617F mutation [24][25][26] we found no elevation of platelets in HSC-SCL:KIT D816V mice. As PV is additionally marked by endogenous erythroid colony formation and decreased serum erythropoietin 23 , we further analyzed these parameters. Indeed, for HSC-SCL:KIT D816V mice markedly decreased serum erythropoietin levels and high numbers of splenic colony forming unit-erythroids (CFU-Es) in assays with low erythropoietin concentration were found, indicating erythropoietin hyper-responsiveness (Fig. 1d). Thrombopoietin serum levels were also reduced in HSC-SCL:KIT D816V mice ( Supplementary Fig. S2).

HSC-SCL:KIT D816V mice develop splenomegaly with massive extramedullary erythropoiesis.
BM and spleen were analyzed 10 weeks after KIT D816V induction or when enlargement of the abdomen was observed. Hematoxylin and eosin (HE) staining revealed no apparent differences in BM histology of HSC-SCL:KIT D816V and control animals. However, siderophages (mononuclear phagocytes containing hemosiderin, a product of hemoglobin catabolism) were scattered throughout control BM, whereas HSC-SCL:KIT D816V mutants showed almost no siderophages, indicating reduced iron storage (Fig. 2a). Quantification further demonstrated a slight reduction in BM megakaryocytes. Immunohistochemical staining for active Caspase3 revealed a slight elevation of apoptosis in HSC-SCL:KIT D816V BM, while Ki67 staining showed no differences in proliferation ( Supplementary Fig. S3). Compared to controls, total BM cellularity of HSC-SCL:KIT D816V mice was moderately increased (Fig. 2b). GFP-fluorescence was examined to estimate efficiency of KIT D816V induction in HSC-SCL:KIT D816V animals, demonstrating transgene expression in 57.40 ± 6.41% of total BM cells (Fig. 2b, dashed bar). While the frequency of CD45-positive cells was decreased in HSC-SCL:KIT D816V BM, the overall frequency of CD45-negative erythroblasts was elevated in comparison to controls. Flow cytometric quantification of discrete developmental stages 19,27 demonstrated a trend towards increased early and late erythroblasts, whereas reticulocytes were reduced, indicating a mild shift to more immature cells (Fig. 2b, Supplementary Fig. S3). To also investigate early BM erythropoiesis, we performed assays for CFU-E progenitors, showing no differences between HSC-SCL:KIT D816V and control mice ( Supplementary Fig. S3).
Gross examination of the mice revealed massive splenomegaly in HSC-SCL:KIT D816V animals, another diagnostic criterion for PV 23 . KIT D816V mutants showed a 19.81 ± 10.09-fold increase in spleen weight and altered spleen histology (Fig. 2c). Lymphoid nodules, normally marked by clusters of CD3-positive cells, were virtually absent in the spleen of mutant mice ( Supplementary Fig. S3).

Mono Gran B-cells T-cells CD45+
x 10 6 cells per ml   Flow cytometric examination demonstrated CD45 expression in almost 80% of control spleen cells. In HSC-SCL:KIT D816V mice, this frequency was reduced to less than 20% (Fig. 2d). This reduction was due to a massive increase in splenic erythropoiesis, as CFU-Es and erythroblast frequency were significantly elevated in HSC-SCL:KIT D816V mice (Figs 1c and 2d, Supplementary Fig. S3). The high ratio of GFP-positive erythroblasts substantiates that this expansion was KIT D816V -dependent.
In summary, these data demonstrate that chronic KIT D816V signaling causes a PV-like disease based on moderately increased BM erythropoiesis and a massive induction of splenic erythropoiesis. As PV is often associated with high incidence of thrombosis due to elevated Hct 23 , we assume that abrupt death of HSC-SCL:KIT D816V animals is a consequence of thrombotic events.
KIT D816V signaling causes stem cell mobilization from BM and a shift of hematopoiesis to extramedullary sites. We next examined effects of KIT D816V signaling on stem cells by analyzing the frequencies of LK (KIT pos Sca-1 neg Lin neg ), LSK (KIT pos Sca-1 pos Lin neg ) and HSC (LSK-CD48 neg CD150 pos ) populations among CD45-positive cells. BM HSC and LSK populations were increased in HSC-SCL:KIT D816V animals compared to controls (Fig. 3a, Supplementary Fig. S4). To investigate the reasons for this expansion, we sorted Lin neg KIT pos BM cells and analyzed expression of important transcriptional regulators. Increased Gata2 transcript was found in the stem cell-enriched compartment of KIT D816V mutants (Fig. 3b, Supplementary Fig. S4). Consistently, Gata2 has been shown to mediate proliferation and survival in hematopoietic stem cell compartments and in vitro differentiation of Gata2-deficient embryonic stem cells demonstrated impaired SCF-responsiveness 28,29 .
To examine if KIT D816V signaling confers a proliferative advantage to stem cells, we analyzed Ki67 expression in BM stem cell populations (Fig. 3c, Supplementary Fig. S4). Interestingly, we found that the effects varied between the different populations. While actively cycling cells were increased in the HSC population of HSC-SCL:KIT D816V Reticulocytes. Data are presented as mean ± standard deviation. P-values were determined using two-tailed, unpaired Student's t-test.
Scientific RepoRts | 7:41427 | DOI: 10.1038/srep41427 animals, they were decreased in the LK population, suggesting that KIT stimulates proliferation in HSCs but supports cell cycle exit in the LK population. To see if we could find alterations in downstream signaling of KIT D816V expressing stem cell populations, we stimulated BM cells with SCF or thrombopoietin and used phospho-flow cytometry to analyze phosphorylation of Erk1/2 and Akt in LK and LSK populations. However, for the analyzed pathways we could find no differences on the level of the examined cell populations ( Supplementary Fig. S4). So, studies of additional pathways and more precisely defined cell populations will be necessary to find out more about signaling changes in KIT D816V stem cells.
In the spleen, HSCs showed a mild but not significant elevation, while the LSK and LK populations bearing lower stem cell potential were significantly increased (Fig. 3a, Supplementary Fig. S4).
We further examined distribution of long-term (LT)-HSCs, short-term (ST)-HSCs and multipotent progenitors (MPPs) between BM, pB and spleen of HSC-SCL:KIT D816V and control mice. We found BM LT-HSCs slightly increased in HSC-SCL:KIT D816V mice, while the other populations in BM were not significantly altered. In contrast, ST-HSC and MPP frequencies were considerably elevated in pB and spleen (although changes were not significant for pB), indicating increased HSC activation and mobilization from BM to circulation. Consistently, progenitor cell populations were slightly reduced in BM but increased in spleen (Fig. 3d, Supplementary Fig. S4). HSC analysis includes Kit as a cell surface marker. It is thus important to note that the KIT D816V mutant receptor localizes to intracellular compartments 30 . In line with this, we found the fraction of cells positive for Kit surface expression similar in GFP-positive and GFP-negative BM populations in HSC-SCL:KIT D816V mice, suggesting that only the endogenous receptor reaches the cell surface and ectopic KIT D816V expression does not affect HSC analysis (Supplementary Fig. S5).
Unexpectedly, the frequency of GFP-positive cells was quite low in KIT D816V HSCs, while it increased with progressive differentiation. To examine potential loss of KIT D816V -positive HSCs or high Cre-recombination frequencies in differentiated cells, we analyzed kinetics of reporter gene expression in different hematopoietic populations 2, 4 and 6 weeks after induction. The GFP-frequency was stable in the LT-HSC compartment and gradually increased in ST-HSC and MPP populations, indicating no negative influence of KIT D816V signaling on stem cell survival (Supplementary Fig. S5). Initially, the GFP-frequency in most progenitor and mature populations was lower than in stem cells, demonstrating Cre-recombination primarily in HSCs, albeit at low frequency. However, analyses suggested also low recombination in granulocytes and B-cells. As the GFP-positive fraction was especially high in the erythroid compartment, we analyzed erythroblasts directly after TX-induction, further revealing recombination in proerythroblasts to a comparatively high degree (Supplementary Fig. S5).
As we found alterations in stem cell distribution, we also analyzed distribution of mature blood cells. In BM, we found no alterations in granulocyte, B-cell and monocyte populations, while in the spleen we found a reduction in lymphocyte frequency for HSC-SCL:KIT D816V animals (Fig. 4a, Supplementary Fig. S6). Analysis of dendritic cells (DCs) revealed a decrease in BM and an increase in the spleen of HSC-SCL:KIT D816V mice. As KIT D816V is strongly associated with mastocytosis 8,9 , we checked for mast cell infiltrations. Surprisingly, flow cytometry revealed a reduced number of peritoneal and skin mast cells in mutants, although changes were not significant (Fig. 4b, Supplementary Fig. S6).
Taking together, our analyses demonstrate that stem cell and differentiated blood cell populations tend to be reduced in BM but increased in spleen (with the exception of lymphocytes), indicating a shift in hematopoiesis to extramedullary sites (Fig. 4c). The KIT D816V -mediated PV-like phenotype is transplantable. In addition to the hematopoietic compartment, HSC-SCL-Cre-ER T -mediated recombination also occurs in endothelial cells 21 . Hence, KIT D816V signaling might also be activated in cells of the perivascular niche. Thus, we investigated whether the PV-like phenotype was a secondary event due to changes in the hematopoietic microenvironment. We treated HSC-SCL:KIT D816V and control animals with TX and performed transplantation of unfractionated BM 6 weeks later. Recipients were analyzed 3 and 6 weeks after transplantation. In HSC-SCL:KIT D816V transplanted mice we observed a gradual increase in RBC, Hb, Hct and WBC (Fig. 5a, Supplementary Fig. S7). Moreover, HSC-SCL:KIT D816V recipients developed splenomegaly and progressively elevated erythroblast numbers (Fig. 5b, Supplementary Fig. S7). While splenic stem cell frequencies were generally elevated post-transplantation, absolute numbers were markedly higher for HSC-SCL:KIT D816V recipients compared to controls (Fig. 5c, Supplementary Fig. S7). In summary, these data demonstrate that early PV-like disease develops autonomously from the medullary hematopoietic compartment.
Splenectomy protects HSC-SCL:KIT D816V mice from increased red blood cell mass but promotes rapid hematopoietic failure. We assumed that the high mortality rate of HSC-SCL:KIT D816V animals was a consequence of thrombotic events caused by the elevated Hct due to massive splenic erythropoiesis. We therefore hypothesized that splenectomy (SplE) might have a protective effect and subjected HSC-SCL:KIT D816V and control animals to SplE prior to TX-treatment (afterwards termed SplE HSC-SCL:KIT D816V and SplE control mice). Blood parameters determined 4 and 10 weeks after KIT D816V induction demonstrated that SplE HSC-SCL:KIT D816V mice indeed were protected from excessive red blood cell production, as RBC, Hb, Hct, MCV and reticulocyte frequency were similar to SplE controls (Fig. 6a, Supplementary Fig. S8). Moreover, although pB monocytes were significantly elevated after 18 weeks, the WBC and the overall number of CD45-positive cells were unaffected in SplE HSC-SCL:KIT D816V animals ( Supplementary Fig. S8). Surprisingly, however, 18 weeks after induction SplE HSC-SCL:KIT D816V animals became anemic, with RBC, Hb, Hct and PLT values falling significantly below control values (Fig. 6a, Supplementary Fig. S8). Despite the decrease in erythrocytes, a considerable number of erythroblasts was found in pB of SplE HSC-SCL:KIT D816V mice ( Supplementary Fig. S8).
Histologically, BM of SplE HSC-SCL:KIT D816V mice appeared hypocellular compared to SplE controls. Consistently, the total cell number per femur and BM megakaryocyte counts were dramatically decreased (Fig. 6b, Supplementary Fig. S8). In SplE controls BM siderophages were abnormally abundant, indicating partial take-over of red blood cell destruction after spleen removal. In contrast, siderophages were absent in KIT D816V mutants after SplE, again demonstrating affected iron storage (Supplementary Fig. S8).
After 18 weeks we found the frequency of CD45-positive cells in SplE HSC-SCL:KIT D816V BM considerably reduced in favor of a significant increase in erythroblast frequency (Fig. 6b). However, an elevated ratio of early and late erythroblasts to reticulocytes indicated ineffective cell maturation. Moreover, given the overall cell loss in BM, total cell numbers were reduced for all erythroid populations (Supplementary Fig. S8). In line with the increase in erythroblast frequency, CFC-assays revealed significantly more CFU-E colonies for SplE HSC-SCL:KIT D816V animals compared to SplE controls. In contrast, the number of non-erythroid colonies was reduced for SplE HSC-SCL:KIT D816V mice (Fig. 6c). These data indicate, that KIT D816V preferentially promotes erythropoiesis.
As extramedullary hematopoiesis can also occur in liver, SplE animals were examined for liver erythropoiesis ( Supplementary Fig. S8). Indeed, a mild induction of erythropoiesis was found in livers of SplE HSC-SCL:KIT D816V mice, but the overall extent was low and no hepatomegaly was observed.

Splenectomy is accompanied by stem cell loss and BM fibrosis in KIT D816V mice.
Analyzing stem cell populations of SplE HSC-SCL:KIT D816V animals, we found a strong progressive reduction in BM HSCs/HPCs compared to SplE controls (Fig. 7a, Supplementary Fig. S9). This was in contrast to KIT D816V mutants not subjected to SplE, which showed no indication of stem cell depletion after 10 weeks (Fig. 3a). We also examined Erk1/2 and Akt phosphorylation in BM LK and LSK populations of SplE animals, but again found no constitutive activation or altered reaction to SCF or thrombopoietin stimulation in KIT D816V cells (Supplementary Fig. S9).
Our results demonstrate, that although SplE protects HSC-SCL:KIT D816V mice from excessive red blood cell accumulation, it promotes rapid BM failure. Clinically, PV can progress to its spent phase polycythemic myeloid metaplasia (PPMM) 31 . There, the initially hypercellular BM becomes hypocellular and secondary myelofibrosis develops. Red blood cell production becomes ineffective and pB cell counts decrease, resembling aplastic anemia. In this study, HSC-SCL:KIT D816V mice without SplE developed a malignancy reminiscent of early PV, whereas the phenotype observed in SplE HSC-SCL:KIT D816V mice resembled PPMM. To further validate this, we performed staining for reticulin fibers to check for BM fibrosis. Indeed, fibrotic changes were found in SplE HSC-SCL:KIT D816V BM 10 and 18 weeks after KIT D816V induction 32 . In contrast, HSC-SCL:KIT D816V mice not subjected to SplE neither displayed any signs of fibrosis in BM after 10 or 18 weeks nor increased reticulin deposition in spleen (Fig. 7b, Supplementary Fig. S9). In SplE HSC-SCL:KIT D816V BM we also observed considerably elevated apoptosis, analyzed by immunostaining for active Caspase3 (Fig. 7b). Ki67 staining showed no apparent differences (Supplementary Fig. S9).
Our results indicate that removal of the splenic hematopoietic niche in KIT D816V mutants dramatically influences the clinical picture of the PV-like disease.

Discussion
In this study, we analyzed the consequences of oncogenic KIT D816V expression in the adult hematopoietic system and found MPN development reminiscent of early and advanced forms of PV, depending on pre-treatment of animals with SplE.  Upon induction of KIT D816V expression, we observed massively increased red blood cell production. Under normal conditions, the erythrocyte pool is maintained by BM steady-state erythropoiesis, whereas extramedullary stress erythropoiesis mediates its rapid expansion upon acute anemia 33 . Different studies have reported  a role of KIT in stress erythropoiesis 34,35 . Accordingly, we found the increase in red blood cells to depend on splenic erythropoiesis. Similar observations have been reported for KIT V558∆;T669I/+ mice 36 . While this clearly demonstrates involvement of KIT in stress erythropoiesis, effects on BM steady-state erythropoiesis are difficult to evaluate in our model, since analysis is impeded by splenic erythropoiesis or hematopoietic failure. However, the erythroid expansion in BM of SplE KIT D816V mutants indicates that KIT signaling regulates proliferative expansion during both, steady-state and stress erythropoiesis.
HSC-SCL:KIT D816V mice had significantly increased pB monocytes, which is likely based on increased proliferation rather than mobilization, as the monocyte frequency in HSC-SCL:KIT D816V BM or spleen 37 was not reduced and pB monocytes were also elevated after SplE. Detailed investigation will be necessary to elucidate effects of KIT D816V on the monocytic lineage. Untypical for PV, pB B-cells were also elevated in HSC-SCL:KIT D816V mice. We suppose that this was primarily caused by displacement from the spleen, which showed markedly reduced B-cell frequency.
In HSC-SCL:KIT D816V animals we observed mobilization of HSCs to the spleen. Several studies have indicated a role of KIT in mobilization of stem cells from the quiescent niche 38,39 . Mobilization of HSCs/HPCs from the BM niche was shown to depend on the release of soluble SCF mediated by matrix metalloproteinase-9 (MMP9) 38 . Moreover, KIT signaling is involved in cell mobilization triggered by functionally blocking cytoadhesion molecules VLA4/VCAM-1, suggesting an integrin/cytokine crosstalk 39 . We also observed an increased frequency of cycling HSCs and increased ST-HSCs in KIT D816V mutants. Differences in the KIT expression level within the LSK-CD150 pos CD48 neg HSC pool have been reported, with lower expression in quiescent cells 40 . These data indicate that oncogenic KIT D816V contributes to activation and mobilization of dormant HSCs. Interestingly, mobilization of HSCs/HPCs has also been associated with primary myelofibrosis and PPMM [41][42][43] .
HSC-SCL:KIT D816V animals pre-treated by SplE showed stem cell loss and myelofibrosis already 10 weeks after KIT D816V induction. Interestingly, no fibrosis was detected in HSC-SCL:KIT D816V mice without SplE, suggesting that disease pathogenesis depends not only on time but also on the interplay between KIT D816V -positive cells and the medullary and extramedullary hematopoietic niches. This is in line with the "bad seeds in bad soil" concept proposed for primary myelofibrosis, presuming that an abnormal hematopoietic cell clone alters its microenvironment, resulting in niche dysfunctions 42 . One may hypothesize, that abnormal KIT D816V -positive cells cycle between hematopoietic niches and upon splenectomy remain in or repopulate the BM and produce high levels of fibrogenic cytokines, stimulating stromal reticulin production. Consequently, displacement from the niche caused by KIT D816V -induced stem cell mobilization and myelofibrosis leads to hematopoietic failure. Studies which have shown that the spleen serves as a reservoir of aberrant stem cells in primary myelofibrosis patients support this assumption 44,45 . Further, a study by Migliaccio et al. demonstrated that the spleen microenvironment is capable of supporting maturation of Gata1 low mutant stem cells that fail to mature in the BM, suggesting the possibility that the BM niche may likewise not sustain maturation of KIT D816V mutant cells 46 .
Future experiments with SplE HSC-SCL:KIT D816V mice should investigate the disease-promoting cell population(s) and cytokine production to further substantiate this hypothesis.
As SplE can manipulate the course of disease, HSC-SCL:KIT D816V mice provide an excellent model to study the interplay between hematopoietic cells and microenvironment in PV-like disease.
Noteworthy, while Philadelphia-negative MPN are highly associated with the JAK2 V617F mutation found in hematopoietic and endothelial-like cells 47,48 , there is only one study reporting an association of KIT mutations with PV 49 . However, other sequencing studies did not confirm this association, raising the question for the relevance of activating KIT mutations in human MPN. It might be possible that alternative mechanisms lead to altered KIT signaling in human MPN. For instance, it was shown that cultured pB cells from primary myelofibrosis patients produce elevated levels of activated MMP9 50 . As mentioned before, MMP9 can mediate release of soluble KIT ligand 38 . Furthermore, phospho-proteomic analysis of erythroblasts from PV patients revealed reduced total KIT and phospho-KIT-Y719 protein content, indicating altered KIT signaling 51 . It remains unclear if this results from hyper-or hypoactivation of the pathway, as the relative KIT-Y719 phosphorylation in PV and control groups was not compared. However, the reduction in total KIT on protein but not transcript level points to increased internalization upon activation in PV cells.
KIT D816V represents a frequent mutation in mastocytosis in man 8,9 and development of cutaneous mastocytosis with variable speed and degree has been demonstrated in a BAC transgenic Kit D814V mouse model 52 . Thus, it was surprising that peritoneal and skin mast cells were reduced in HSC-SCL:KIT D816V mice and no mastocytosis was observed. This discrepancy might be due to differential expression levels, as the the Kit promoter exhibits strong physiological activity in the mast cell lineage 15,53 compared to moderate ROSA26-mediated expression. Moreover, mastocytosis in man is associated with additional mutations, such as Tet2, Srsf2, Asxl1, Cbl and Runx1, which are not present in HSC-SCL:KIT D816V mice 54,55 . A recent study by Jawhar et al. has shown that mutations in TET2, SRSF2 or ASXL1 precede the KIT D816V mutation in mastocytosis 56 . Thus, in the absence of such seed mutations the KIT D816V mediated disease phenotype might well be different from the phenotype in KIT D816V mastocytosis patients. In addition to mastocytosis, the KIT D816V mutation is also frequently identified in the core binding factor leukemias, involving the AML1-ETO or CBFB/MYH11 genes (t [8;21] or inv [16]/t [16;16], respectively) 7,57 . A study by Wang et al. found strong evidence that in t(8;21) AML, AML1-ETO is the first genetic hit which is responsible for disease initiation, while KIT D816V is a secondary mutation 58 .
Features reminiscent of early PV have also been observed in KIT V558∆;T669I/+ mice 36,59 . However, the effects were rather mild, probably due the endogenous Kit promoter driving the mutated KIT, mediating no expression at later developmental stages. We show that KIT D816V expression driven by the ROSA26 promoter causes a more severe and complex phenotype (PPMM/myelofibrosis). We speculate that our KIT D816V allele, which is not downregulated upon differentiation might better mirror pathology of some MPN entities, where deregulated gene expression patterns are found [60][61][62] . Taking together, R26-LSL-KIT D816V mice provide a valuable model to investigate interactions of aberrant KIT D816V cells with the hematopoietic microenvironment leading to PV-like disease and myelofibrosis.