Shayu Deshpande1, Benedikt Bosbach1, Yasemin Yozgat1, Malcolm A Moore2, Chris-topher Y Park, Peter Besmer1#
1Developmental Biology and 2Cell Biology Programs, Sloan-Kettering Institute, New York, NY 10065
- in preparation -
Corresponding author#: Peter Besmer
Developmental Biology Program Sloan-Kettering Institute
1275 York Avenue New York, NY 10065.
Phone: 212-639-8188;
Fax: 646-422-2355.
E-mail: p-besmer@ski.mskcc.org
3.1 Abstract
The signaling mediated by the KIT receptor tyrosine kinase has an important role in the hematopoietic system. A wealth of studies has shown that mice with Kit loss-of-function mutations have varying degrees of macrocytic anemia, at least in part stem-ming from defects in hematopoietic stem cell and erythroid progenitor compartments.
We have recently engineered a mouse model bearing – in addition to the oncogenic KitV558Δ (human KITV559Δ) mutation found in patients with familial gastrointestinal stromal tumor (GIST) – the second-site mutation KitT669I (human KITT670I) which is ob-served in cases of imatinib-resistant GIST. These KitV558Δ;T669I/+ mice developed pro-nounced erythrocytosis with an increase in erythroid progenitor numbers, a phenotype previously seen only in mouse models of polycythemia vera (PV) with alterations in genes other than Kit, e.g., Epo or Jak2. As in non-Kit-mutant models of PV, erythroid progenitor numbers in KitV558Δ;T669I/+ mice normalized upon splenectomy. In accordance with the increased erythroid progenitors, myeloerythroid progenitor numbers were also elevated in the KitV558Δ;T669I/+ mice. Interestingly, hematopoietic stem cell (HSC) num-bers in the bone marrow (BM) of KitV558Δ;T669I/+ mice were unchanged in comparison to wild-type mice. However, increased HSC numbers were observed in the KitV558Δ;T669I/+
spleen and fetal livers. Importantly, stem cells from KitV558Δ;T669I/+
BM had competitive advantage over wild-type stem cells and stem cells from the spleen were capable of self-renewal. Although HSCs did not cycle differently in KitV558Δ;T669I/+
and wild-type mice, in response to 5-flurouracil treatment elevated numbers of Lin-Sca+ cells were found cycling in the KitV558Δ;T669I/+ BM compared to wild-type. Our study demonstrates that signaling from the KitV558Δ;T669I/+ receptor has important consequences on the hemato-poietic lineage affecting stem cell function and resulting in increased steady-state as well as stress erythropoiesis. Our knock-in mice provide an excellent model to explore the biology and therapeutic interventions of hematologic malignancies with Kit gain-of-function mutations.
3.2 Introduction
Hematopoiesis is a stringently regulated process under the control of various cytokines and growth factors and their receptors at different stages of the hematopoietic hierarchy.
The receptor KIT and its cognate ligand KitL are classically known to have important roles in mast cell biology, lymphopoiesis, erythropoiesis, and finally, stem cell function.
Loss of function mutations in murine Kit (W) and KitL (Sl) genes show varying degrees of mast cell deficiency, macrocytic anemia and stem cell renewal defects, based on the severity of the mutations.
The stem cell compartment in the bone marrow (BM) maintains steady state hemato-poietic numbers by self-renewal and differentiation into progenitors. Stem cells, based on their self-renewing abilities have been distinguished into long-term hematopoietic stem cells (LT-HSC), short-term hematopoietic stem cells (ST-HSC) and finally multipotent progenitor cells (Morrison and Weissman 1994). KitL and KIT have long been known to be critical components of the bone marrow niche. LT-HSC and ST-HSC from Kit loss-of-function mutations such as Wv, W41, W42 in either homozygous or het-erozygous conditions are ineffective at reconstituting bone marrow in irradiated recipi-ents indicating an important role for KIT in stem cell self-renewal and survival (Sharma et al. 2007).
Apart from its role in stem cell function, KitL and KIT have long been known to have a role in regulating adult definitive erythropoiesis. The mutations in the W and Sl genes have provided important insights into the role of KIT in erythropoiesis. Mice with partial loss-of-function mutations in either gene, are viable but have macrocytic anemia, while more severe mutations lead to embryonic lethality due to lack of erythropoiesis (Russell 1979; Nocka et al. 1989; Besmer 1991). KIT signaling is especially important in the proliferation and survival of erythroid progenitors, specifically at the BFU-E stage (Nocka et al. 1990; Muta et al. 1995). At the CFU-E stage KitL augments the ef-fects of erythropoietin (EPO) on CFU-E growth and survival (Muta et al. 1995). Signal-ing downstream of KIT, especially Src family kinases have also been shown to selec-tively affect erythropoiesis. Interestingly, knock-in mice carrying the mutation of the Src binding site of KIT (KitY567F/Y567F
) did not show defects in steady state hematopoie-sis but the mutation had pronounced effects on stress erythropoiehematopoie-sis (Agosti et al. 2009).
While a wealth of studies is available on the hematopoietic phenotypes of Kit-deficient mutant mice, few studies are available with Kit gain-of-function mutations. In our la-boratory, we have generated a mouse model with an activating mutation in the juxtamembrane domain of the KIT receptor (KitV558Δ/+ (Sommer et al. 2003)). Mice with heterozygous KitV558Δ/+ mutation develop gastrointestinal stromal tumors mirroring fa-milial GIST seen in patients. The drug imatinib is the frontline therapy used to treat ad-vanced GIST. However, many patients develop second-site mutations and thereby re-sistance to imatinib after prolonged imatinib treatment (Antonescu et al. 2005). These second-site mutations are found to occur in conjunction with the 558Δ mutation and are commonly found to occur in exon 17, which encodes part of the catalytic domain II of KIT, as well as in exon 13 (V654A) and exon 14 (T670I) which encode the N-terminal kinase domain (Antonescu et al. 2005). Recently, we engineered a mouse model bearing in addition to the KitV558Δ mutation the “gatekeeper” KitT669I mutation (Bosbach et al.
2012). These KitV558Δ;T669I/+ mice developed small cecal GISTs with pronounced inter-stitial cell of Cajal hyperplasia in the stomach and colon. Interestingly, these mice also showed increased hematocrit and BFU-E numbers and were distinguished by the pres-ence of red paws reminiscent of mice with polycythemia vera and other myeloproliferative neoplasms (Bosbach et al. 2012).
In the present study we have investigated the molecular basis of the increased hema-tocrit observed in the KitV558Δ;T669I/+ mice. We have assessed erythroid progenitor num-bers by flow cytometry and then analyzed whether splenectomy can normalize hemato-crit numbers in KitV558Δ;T669I/+ mice. We have then investigated whether the increased erythrocytosis is a result of expansion of myeloerythroid progenitors and stem cells. We also investigated if stem cell function is altered in the KitV558Δ;T669I/+ mice by transplan-tation assays and the analysis of stem cell cycling. Our data show that the KitV558Δ;T669I
mutation affects progenitor numbers and in particular affects stem cell renewal.
3.3 Results
Increased Erythroid Generation in KitV558Δ;T669I/+ Mice and Effect of Splenectomy.
We have recently generated the KitV558Δ;T669I/+ mouse model bearing the “gatekeeper”-mutation KitT669I in addition to the oncogenic KitV558Δ mutation. This mouse recapitu-lates imatinib-resistant GIST and, interestingly, displays red paws and increased
hema-tocrit (Bosbach et al. 2012). The KitV558Δ;T669I/+
mice consistently show elevated hema-tocrit from 5 wk of age, with a concomitant increase in the number of BFU-E progeni-tors in the BM and spleen. Interestingly, our results revealed that while BFU-E growth in KitV558Δ;T669I/+ mice remained EPO-dependent their growth was more sensitive to KitL than wild-type progenitors (Bosbach et al. 2012).
In order to gain insight into the various stages of erythroid maturation flow cytome-tric analysis of erythroid progenitors was performed according to the protocol described previously (Liu et al. 2006b), which distinguishes different stages of erythroid matura-tion marker-based as R1: Ter119lowCD71high, A: Ter119highCD71highFSChigh, B: Ter-119high, CD71medFSClow and C:Ter119highCD71lowFSClow (Fig. 12A). Our results show that R1 progenitor frequency was higher in the mutant BM and significantly increased in the mutant spleen compared to wild-type (Fig. 12B left). Total R1 progenitor num-bers were also increased in the KitV558Δ;T669I/+ BM and spleen as were the type C erythro-id cells, which represent mature erythrocytes (Liu et al. 2006b) (Fig. 12B right). In ac-cordance with the increased R1 progenitors, spleen cytospins showed increased number of erythrocytes in the KitV558Δ;T669I/+ mice (Supplementary Fig. S20). In order to estab-lish if the increased R1 progenitor numbers were due to increased expansion of these progenitors, we analyzed cycling of these progenitors by Ki67 and DAPI staining fol-lowed by flow cytometry. In accordance with the increase in erythroid numbers in the KitV558Δ;T669I/+ spleen, we found significantly increased frequency of R1 progenitors in S/G2 phase of the cell cycle in the KitV558Δ;T669I/+
spleen (Fig. 12C). Cycling in the mu-tant BM however was not significantly different compared to wild-type (Fig. 12C). In order to ascertain the role of the spleen in the development of increased hematocrit in the KitV558Δ;T669I/+ mice, splenectomy was performed on 10-wk-old wild-type and mutant mice. Mice were bled two weeks prior to surgery to obtain pre-surgical hematocrit val-ues and then subjected to splenectomy. Mice were allowed to recover for one week after surgery. Subsequent bleeds post-surgery revealed a drop in hematocrit values in both the wild-type (55.92 ± 2.1 to 43.24 ± 2.38) and KitV558Δ;T669I/+ mice (84.48 ± 2.32 to 52.88 ± 3.28), although the reduction was much greater in the KitV558Δ;T669I/+ mice. Inter-estingly, these post-surgery hematocrit values in the KitV558Δ;T669I/+ mice were similar to pre-surgery wild-type levels and were maintained until the time point of last measure-ment, namely 8 weeks (Fig. 12D). This is in contrast to the JAK2V617F-GFP BM
trans-plantation model of PV, where splenectomy caused a gradual decrease in hematocrit levels and by 7 weeks normal wild-type hematocrit values were obtained (Mo et al.
2009). Erythroid progenitors assessed 8 wk post-splenectomy showed similar numbers
Fig. 12: Expansion of erythroid progenitors in the KitV558Δ;T669I/+ BM and spleen.
KitV558Δ;T669I/+ mice develop erythrocytosis which is remedied only partially by splenectomy. (A) Total bone marrow (BM) and spleen (SPL) cells were prepared for staining with Ter119 and CD71 antibodies for flow cytometry. Cells were gated as previously described. A representative flow cytometry graph shows the R1, Ter119high, A, B and C erythroid populations. (B) Graphical representation of flow cytometry results show increased frequency and total Ter119+CD71+ cells, especially R1 erythroid pro-genitor population in the BM and spleen of KitV558Δ;T669I/+ (GTK) mice in comparison to wild-type (WT) mice. (C) BM and spleen cells were stained with cell surface markers Ter119 and CD71 and processed for Ki67-DAPI staining. Erythroid R1 progenitor cycling measured by Ki67 flow cytometry revealed in-creased cycling of progenitors in the spleen but no change in the BM in KitV558Δ;T669I/+ mice. (D) In order to assess the contribution of the spleen toward increased erythropoiesis, spleens from wild-type and KitV558Δ;T669I/+ mice were surgically removed. Hematocrit levels 2 weeks post-surgery were reduced in splenectomized mice compared to pre-surgical values. However hematocrit levels of KitV558Δ;T669I/+ mice never reached wild-type levels when analyzed up to 8 weeks after splenectomy. (E) Flow cytometric analysis of wild-type and KitV558Δ;T669I/+ BM cells by Ter119 and CD71 staining at 8 weeks post-splenectomy showed increased erythroid R1 and A progenitors in the BM of KitV558Δ;T669I/+ mice.
of R1 progenitors in pre- and post-splenectomized wild-type and KitV558Δ;T669I/+
mice (compare Fig. 12B right and E). The numbers of differentiated erythroid cells, namely the population C was reduced in KitV558Δ;T669I/+ post-splenectomized mice which may corroborate with the reduced hematocrit levels. Collectively, these results indicate sig-nificantly elevated erythropoiesis in the KitV558Δ;T669I/+ mice and splenectomy normalizes hematocrit levels in the KitV558Δ;T669I/+ mice.
Myeloid Progenitor Expansion in KitV558Δ;T669I/+ Mice.
The increased number of erythroid progenitors in the KitV558Δ;T669I/+ mice led us to in-vestigate if myeloid subsets were also increased in these mice. Flow cytometric analysis of granulocyte and monocyte-macrophage populations in the BM and spleen (Fig. 13A) revealed increased numbers of both these cell types in the KitV558Δ;T669I/+ mice (Fig.
13B). In order to assess whether homeostasis was also altered upstream in the myeloid hierarchy, a detailed flow cytometric analysis of myeloid progenitors namely common myeloid progenitors (CMP), granulocyte macrophage progenitors (GMP) and myeloerythroid progenitors (MEP) was carried out according to the protocol established previously (Akashi et al. 2000). In the BM, CMP and GMP populations were similar in frequency and total numbers in wild-type and KitV558Δ;T669I/+
mice. Interestingly, MEP population in the KitV558Δ;T669I/+
BM was significantly increased compared to wild-type in terms of both frequency and total numbers (Fig. 13C). The changes in myeloid pro-genitors were also studied in the spleen and in contrast to the BM, interestingly in the KitV558Δ;T669I/+ spleen the frequency of all three progenitor types was significantly in-creased compared to wild-type spleen. A significant elevation in total cells numbers of CMP and MEP was observed in the KitV558Δ;T669I/+ spleen (Fig. 13D). In order to physio-logically assess the expansion of some of these myeloid populations, in vitro colony as-says were performed. As observed in flow cytometric analysis of granulocyte and mon-ocyte populations, GM colonies were two-fold increased in the KitV558Δ;T669I/+ BM and several-fold increased in the KitV558Δ;T669I/+ spleen. In addition, GEMM colonies were 5-fold greater in the KitV558Δ;T669I/+ spleen compared to wild-type reiterating the elevated expansion of myeloerythroid progenitors in the spleens of these mice (Tab. 6). Recon-stitution assays such as CFU-S also provide a measure of in-vivo progenitor function.
While it has been described previously that day 8 CFU-S correspond to MEPs, day 9
CFU-S reflect CMP numbers and day 12 CFU-S are a mixture of MPP and CMP (Morrison and Weissman 1994; Na Nakorn et al. 2002; Sharma et al. 2007). In accord-ance with the unchanged CMP numbers in the BM, transplantation of BM cells from wild-type or KitV558Δ;T669I/+ mice into irradiated C57BL6 mice, showed no change in CFU-S colonies when observed at d10 after transplantation (Supplementary Fig. S21).
Fig. 13: Expansion of the myeloid lineage in the adult BM and spleen of KitV558Δ;T669I/+ mice.
Cells from the BM and spleen were stained with Gr1 and Mac1 antibodies and analyzed by flow cytome-try. (A) Representative flow cytometry graph shows Gr1highMac1+ granulocyte and Gr1lowMac1+ mono-cyte populations from BM. (B) Analysis of Gr1/Mac1 population in the BM and spleen showed signifi-cantly increased granulocyte and monocyte populations in the KitV558Δ;T669I/+ mice. (C-D) BM and spleen cells were stained against lineage markers and Lin-Kit+Sca- cells were distinguished into CMP, GMP and MEP subsets according to their CD34 and FcγR surface expression. Flow cytometric analysis showed increased frequency and total numbers of CMP, GMP and MEP subsets in KitV558Δ;T669I/+ BM and spleen.
Tab. 6: CFU-GM and CFU-GEMM assays KitV558Δ;T669I/+ mice.
Colony forming assays were performed with total BM or spleen cells from wild-type and KitV558Δ;T669I/+
mice in presence of KitL, IL3 and EPO. In accordance with the increased myeloerythroid populations in KitV558Δ;T669I/+, colony forming assays showed increased number of CFU-GM and CFU-GEMM popula-tions in the BM and spleen of KitV558Δ;T669I/+ mice. Values are mean ± SEM. n = 3 mice per group.
While investigation of the lymphoid lineages by flow cytometry revealed reduced fre-quency of B220+ cells (Fig. 14A) as well as CD4+ and CD8+ cells (Fig. 14B) both in the KitV558Δ;T669I/+ BM and spleen compared to wild-type, respective total numbers of B220+, CD4+ and CD8+ cells were not statistically different (Fig. 14B). In summary, our results show expansion of myeloerythroid progenitors in conjunction with normal lymphopoiesis and thus a reduction in the frequency of lymphocytes in total BM and spleen in KitV558Δ;T669I/+ mice.
Fig. 14: Reduced frequency of lymphopoiesis in KitV558Δ;T669I/+ mice.
(A) BM and spleen cells were stained with B220 antibody and analyzed by flow cytometry. The fre-quency of B220+ cells was reduced in the KitV558Δ;T669I/+ BM and spleen, however total cell numbers were not significantly reduced. (B) The frequency of CD4+ and CD8+ cells was reduced in the KitV558Δ;T669I/+
BM and spleen, however total cell numbers were not significantly reduced. Values are mean ± SEM. n indicates number of mice.
Kit+/+-BM KitV558Δ;T669I/+
-BM Kit+/+-Spleen KitV558Δ;T669I/+
-Spleen GM 3,7020 ± 6,730 52,611 ± 1,645* 3,317 ± 718 91,218 ± 9,645***
GEMM 2,757 ± 682 3,801 ± 1,053 1,658 ± 718 19,608 ± 8,439*
The KitV558Δ;T669I/+
Mutation Results in Increased Stem Cell Numbers at Extramedullary Sites and in the Fetal Liver.
We first sought to determine whether in addition to the expansion of MEPs in the KitV558Δ;T669I/+ mice, there was any change in the HSC-enriched Lin-Sca+Kit+ (LSK) population. Interestingly, analysis of LSK by flow cytometry revealed no significant differences in LSK frequency or numbers in the wild-type and KitV558Δ;T669I/+ BM (Fig.
15A). However, in contrast to the BM, LSK frequency and total LSK numbers were significantly higher in the KitV558Δ;T669I/+ spleen compared to wild-type (Fig. 15A).
Based on these results, we further analyzed if there were any differences in the long term-HSC (LT-HSC) compartment defined phenotypically as Lin-Sca+Kit+CD34 -CD150+ (Czechowicz et al. 2007). We detected no significant differences in the LT-HSC numbers in the wild-type and KitV558Δ;T669I/+ BM. However as with the LSK subset, interestingly, LT-HSC frequency and total cell numbers were significantly higher in the KitV558Δ;T669I/+ spleen compared to wild-type (Fig. 15B). In addition a higher frequency of LT-HSC was also found in the KitV558Δ;T669I/+ peripheral blood (Fig. 15C).
Fig. 15: Analysis of the stem cell compartment in wild-type and KitV558Δ;T669I/+ mice.
(A) Total BM and spleen cells were stained against lineage cocktail and gated as Lin-Kit+Sca+. LSK fre-quency and total cell numbers were not significantly altered in the KitV558Δ;T669I/+ BM compared to wild-type but significantly elevated in the KitV558Δ;T669I/+ spleen compared to wild-type. (B) LT-HSC defined as Lin-Sca+Kit+CD150+CD34- were significantly higher in frequency and total cell numbers in KitV558Δ;T669I/+
spleen compared to wild-type spleen. LT-HSC numbers were not significantly different in the BM of the two genotypes. (C) The frequency of LT-HSC numbers in the peripheral blood was increased in KitV558Δ;T669I/+ compared to wild-type mice. Values are mean ± SEM. n indicates number of mice.
These results indicate elevated stem cell numbers in the KitV558Δ;T669I/+
mice with dis-semination of stem cells from their primary location, i.e., the BM to extramedullary sites. Previous studies have defined a clear role for KIT and KitL in the expansion of fetal liver HSC (FL-HSC) (Bowie et al. 2007a). Interestingly, it has been observed in vitro that FL-HSC are more sensitive to KitL than adult HSC, which has been attributed to possible differences in KIT-mediated downstream signaling during the developmen-tal progression of HSC (Bowie et al. 2007a). In order to delineate the effect of the KitV558Δ;T669I/+ mutation on FL-HSC, we analyzed fetal livers from timed-pregnant fe-males obtained by crossing C57BL/6 fefe-males with KitV558Δ;T669I/+ males. At E14.5, KitV558Δ;T669I/+ fetal livers had significantly greater cellularity compared to wild-type (Fig. 16A). At this fetal age, equivalent numbers of LSK progenitor cells were observed in KitV558Δ;T669I/+ fetal livers and wild-type (Fig. 16B).
Fig. 16: Analysis of fetal livers from wild-type and KitV558Δ;T669I/+ mice.
(A) The total number of nucleated cells is higher in KitV558Δ;T669I/+ fetal livers at E14.5. (B) LSK numbers and frequency were not significantly different in KitV558Δ;T669I/+ compared to wild-type mice. (C) However both LT-HSC number and frequency was significantly higher in KitV558Δ;T669I/+ fetal livers. (D) Analysis of erythroid progenitors by staining fetal livers against Ter119 and CD71 antibodies revealed significantly higher R1 progenitors in KitV558Δ;T669I/+ compared to wild-type fetal livers, while the erythroid subsets A, B and C did not vary significantly between the two groups. (E-F) Lineage analysis revealed no significant difference in B-lymphoid (B220+), or myeloid (Gr1+) cell numbers between KitV558Δ;T669I/+ and wild-type mice. Values are mean ± SEM. n = 7-8 fetal livers per group.
We next investigated LT-HSC in the fetal liver using the markers LSKCD150+CD48 -and omitting CD34 since it is expressed on HSC in the embryo (Ogawa et al. 2001;
McKinney-Freeman et al. 2009). Interestingly, similar to the adult KitV558Δ;T669I/+ spleen,
McKinney-Freeman et al. 2009). Interestingly, similar to the adult KitV558Δ;T669I/+ spleen,