|Home | About | Journals | Submit | Contact Us | Français|
The X-linked Gata1low mutation in mice induces strain-restricted myeloproliferative disorders characterized by extramedullary hematopoiesis in spleen (CD1 and DBA/2) and liver (CD1 only). To assess the role of the microenvironment in establishing this myeloproliferative trait, progenitor cell compartments of spleen and marrow from wild-type and Gata1low mice were compared. Phenotype and clonal assay of non-fractionated cells indicated that Gata1low mice contain progenitor cell numbers 4-fold lower and 10-fold higher than normal in marrow and spleen, respectively. However, progenitor cells prospectively isolated from spleen, but not from marrow, of Gata1low mice expressed colony-forming function in vitro. Therefore, calculation of cloning activity of purified cells demonstrated that the total number of Gata1low progenitor cells was 10–100-fold lower than normal in marrow and >1,000-times higher than normal in spleen. This observation indicates that Gata1low hematopoiesis is favored by the spleen and is in agreement with our previous report that removal of this organs induces wild-type hematopoiesis in heterozygous Gata1low/+ females (Migliaccio et al, Blood 114:2107,2009). To clarify if rescue of wild-type hematopoiesis by splenectomy prevented extramedullary hematopoiesis in liver, marrow cytokine expression profile and liver histopathology of splenectomised Gata1low/+ females were investigated. After splenectomy, the marrow expression levels of TGF-β, VEGF, osteocalcin, PDGF-α and SDF-1 remained abnormally high while Gata1low hematopoiesis was detectable in liver of both CD1 and DBA/2 mutants. Therefore, in the absence of the spleen, Gata1low hematopoiesis is supported by the liver suggesting that treatment of myelofibrosis in these animals requires the rescue of both stem cell and microenvironmental functions.
Primary myelofibrosis (PMF) is a myeloproliferative neoplasm (Hoffman, 2000; Jacobson et al., 1978; Tefferi, 2000; Tefferi et al., 2007) characterized by distinct abnormalities in megakaryocyte (MK) development which include reduced levels of Gata1 expression (Vannucchi et al., 2005), increased proliferation with retarded maturation (Ciurea et al., 2007), abnormally high p-Selectin localization on the demarcation membrane system and increased pathological neutrophil emperipolesis (Schmitt et al., 2000). It is assumed that these MK abnormalities, by altering the growth factor milieu of the microenvironment, lead to fibrosis, neo-vascularization of the marrow and bone formation (Hoffman and Xu, 2006). The alterations of the microenvironment have been suggested to dislodge the stem/progenitor cells from their niches in the marrow, resulting in increased stem/progenitor cell trafficking and hematopoiesis in extramedullary sites, including the spleen (Migliaccio et al., 2008). Splenomegaly, one of the consequences of extramedullary hematopoiesis in the spleen, is associated with numerous clinical complications of PMF (Cervantes et al., 2007). Therefore, splenectomy is recommended as a palliative treatment strategy for symptomatic splenomegaly refractory to other treatments (Cervantes et al., 2007). Whether the involvement of the spleen in the pathogenesis of PMF is secondary to hematopoietic failure in the marrow or whether the spleen, by providing a specific microenvironment, plays an active role in the progression of the disease, is debatable. Since splenectomy involves a substantial risk in patients with PMF, the role of the spleen in the development of the disease has not been systematically addressed in humans.
Gata1 is essential for appropriate erythroid and megakaryocytic differentiation (Orkin and Zon, 2008; Pang et al., 2005). In mice, the expression of Gata1 in MK can be experimentally reduced by ablating sequences upstream to the gene that regulate its expression in MK (Guyot et al., 2006; McDevitt et al., 1997; Vyas et al., 1999). Mice lacking these sequences, i.e. carrying the hypomorphic Gata1low mutation, are born anemic and thrombocytopenic and die soon after birth (McDevitt et al., 1997; Vyas et al., 1999). The mutation, however, is not lethal in other genetic backgrounds (CD1 and DBA/2) that efficiently activate erythropoiesis in the spleen in response to erythroid stress16. In these strains, the mutant mice recover from their anemia at 1-month of age by developing splenomegaly (Martelli et al., 2005; Vannucchi et al., 2001). These mice, however, remain thrombocytopenic because of severe abnormalities in MK maturation (Centurione et al., 2004). These abnormalities are identical to those observed in the MK of PMF patients (Centurione et al., 2004; Schmitt et al., 2000). It is therefore not surprising that, with age, Gata1low mutants develop myelofibrosis (MF), a phenotype that closely resembles the human disease including, in the CD1 background, extramedullary hematopoiesis in liver (Vannucchi et al., 2002).
Recently, we have demonstrated that the spleen and marrow microenvironment selectively support growth of Gata1low and wild-type stem cells, respectively (Migliaccio et al., 2009). Splenectomy is lethal for hemyzygous Gata1low/0 mice [Gata1 is on the X chromosome (Zon et al., 1990)] while favoring hematopoiesis from stem cells expressing the wild-type allele in the marrow, eliminating many features of MF, in heterozygous Gata1low/+ females (increasing blood platelet counts and reducing fibrosis and bone formation) (Migliaccio et al., 2009). The aim of the present study was to quantify the size of Gata1low stem/progenitor compartments in the spleen, to identify the mechanisms that support proliferation of these cells in this organ and to clarify whether the prevalently wild-type hematopoiesis observed in the marrow of heterozygous Gata1+/− females, in the absence of the spleen, restores the cytokine expression profile in this organ preventing stem/progenitor cell trafficking and development of extramedullary hematopoiesis in liver. In Gata1low mice, the majority of stem/progenitor cells capable of forming colonies in vitro were found in the spleen. In addition, progenitor cells from the spleen expressed uniquely high Gata2 levels that were likely responsible to rescue the Gata1low defect (Huang et al., 2009). However, after splenectomy, the cytokine expression profile of the marrow of heterozygous Gata1low/+ females remained abnormally high and Gata1low hematopoiesis was observed in the liver even in strains (DBA/2) in which this trait had not been observed before (Martelli et al., 2005). Therefore, Gata1low hematopoiesis is not stem cell autonomous but requires the permissive microenvironment of spleen and liver, suggesting that treatment of this animal model of myelofibrosis requires rescue of both stem cell and microenvironmental functions.
Gata1low mice were obtained by breeding 5th generation Gata1low/0 males (Martelli et al., 2005) with CD1 females (Charles River, Calco, Italy). The mice used in the experiments were represented by 5th generation Gata1low/0 males and by the heterozygous Gata1low/+ female and wild-type male offspring of these crossings. Heterozygous Gata1low/+ DBA/2NCrBR females were used in selected experiments. Mice were housed for up to 2 years under good animal care practice conditions in the animal facilities of Istituto Superiore Sanità. All the experiments were performed with sex- and age-matched mice under protocols approved by the institutional animal care committee.
Gata1low/+ females (Martelli et al., 2005) were anaesthetized with xylazine (10 mg/kg, Bayer, Milan, Italy) and ketamine (200 mg/kg, Gellini Farmaceutics, Latina, Italy) i.p. one day following food withdrawal. The spleen was removed after double ligation of the splenic artery and vein. The muscle, peritoneum and skin were closed in separate layers using sterile 5-0 absorbable suture. Animals received the analgesic butorphanol s.c. (5 mg/kg/day, Intervet Italia Srl, Milan, Italy) for 4 days post-surgery. Groups of at least 3 mice were sacrified 3, 6–8 and 9–10 months post splenectomy for further analyses.
Livers were fixed in 10% (v/v) phosphate-buffered formalin (Sigma, St. Louis, MO, USA). Paraffin embedded, 2.5–3 µM sections were prepared and stained with hematoxylin-eosin or incubated with an anti-CD45 monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and then stained with avidin-biotin immunoperoxidase (Vectastain Elite ABC Kit; Vector laboratories, Burlingame, CA, USA), as described by the manufacturer and then counterstained with hematoxylin-eosin. Samples not incubated with the primary antibody, or incubated with non-immune IgGs, served as negative controls. Histopathological evaluations were performed with a DM RB microscope (Leica LTD, Heidelberg, Germany) set in a transillumination mode and equipped with a Coolsnap videocamera in order to capture computerized images (RS Photometrics, Tucson, AZ, USA).
The cells analyzed in these experiments were mononuclear cells (MNC) of the liver and marrow and light density cells (LDC) of the spleen. LDC were isolated by centrifugation over a mouse-specific Ficoll cushion (ρ<1.083, Histopaque-1083, Sigma-Aldrich, St Luois, MO, USA). Cells were suspended in Ca++ Mg++-free phosphate buffered saline supplemented with 1% (v/v) bovine serum albumin, 2 mM EDTA, 0.1 % NaN3 and incubated for 30 min on ice with 1 µg/106 cells of phycoerythrin (PE)-conjugated CD117 (anti-cKit), CD71 and CD61 and fluorescein isothiocyanate (FITC)-conjugated anti-Sca1, CD34, TER119 and CD41 (all from PharMingen, San Diego, CA) (Migliaccio et al., 2009). Dead cells were excluded by propidium iodide staining (5 µg/mL, Sigma) and non-specific signals were excluded with appropriate isotype controls (PharMingen). Cell fluorescence was analyzed with the ARIA cell sorter (Becton Dickinson, Franklin Lakes, NJ). In selected experiments, cells in the prospective stem cell (CD117pos/Sca1pos) (Spangrude et al., 1988), progenitor cell (CD117pos/CD34high and CD117posCDlow) (Akashi et al., 2000; Na Nakorn et al., 2002), erythroblast (CD71pos/TERpos) (Socolovsky et al., 2001) and megakaryocytic (CD41pos/CD61pos) (Migliaccio et al., 2009) gates were isolated by sorting and shown to be >90% pure by reanalysis.
The progenitor cells in hematopoietic tissues from representative normal and Gata1low/+ littermates were assayed by plating cells of the marrow (2×104 MNC/mL), blood (2 µL/mL), spleen (2×104 LDC/mL) and liver (105 MNC/mL) in standard methylcellulose cultures (0.9% w/v) containing fetal bovine serum (30% v/v, Sigma), rat stem cell factor (SCF, 100 ng/mL), mouse interleukin-3 (IL-3, 10 ng/mL) (from Sigma) and human erythropoietin (EPO, 2 U/mL; Boehringer Mannheim, Mannheim, Germany). Prospectively isolated progenitor cells were cultured under conditions of limiting dilution (100 cells/mL). The cultures were incubated at 37 °C in a humidified incubator containing 5% CO2 in air and hematopoietic colonies scored after either 8 or 15 days. BFU-E-, CFU-GM- and CFU-GEMM-derived colonies were scored individually but results expressed as total colony numbers for simplicity. The numbers of colonies (Table I and andII)II) were then used to calculate the size of the progenitor cell compartments in the marrow and spleen according to the following algorithms:
These calculations considered the number of cells present in a femur to correspond to 3.3% of total marrow cells of a mouse (Migliaccio et al., 2003).
Total RNA was prepared by lysing bone marrow or purified cells directly into a commercial guanidine thiocyanate/phenol solution (Trizol, Gibco BRL, Paisley, UK). RNA (1 µg) was reverse transcribed with 2.5 µM random hexamers using the superscript kit (Invitrogen, Milan, Italy). Osteocalcin, TGF-β1, PDGFα and VEGF cDNAs were quantified by semiquantitative RT-PCR based on increasing amplification cycles (20, 25, 30 and 35 cycles, indicated by triangles on the top of the figure), as described (Vannucchi et al., 2002). In this analysis, gene expression was quantified spectrophotometrically on the basis of the optical density of the amplified test bands compared to the amount of concurrently amplified Actin product, and expressed as ratio. SDF-1 and Gata1 expressions were quantified by Real Time RT-PCR using pre-developed assays (Applied Biosystem, Foster City, CA, USA) (Migliaccio et al., 2008). GAPDH cDNA was concurrently amplified as housekeeping control. Reactions were performed in a ABI PRISM 7700 Sequence Detection System (Applied Biosystems) using the following parameters: 40 cycles of a two-step PCR program at 95°C for 15 sec. and 60°C for 60 sec., after an initial denaturation/activation step at 95°C for 10 min. The SDS software was used to analyze and calculate the cycle threshold (CT). Quantitative normalization of cDNA was obtained by the ΔCt method (ΔCt = target gene Ct - GAPDH Ct) and the levels of mRNA expressed as 2−ΔCt. Samples were assayed in triplicate and appropriate negative and positive controls were included in each assay.
Results are presented as mean (±SD) of those observed in at least 3 independent experiments with 3 mice per experimental group. Statistical analysis was performed by analysis of variance (Anova test) using Origin 3.5 software for Windows (Microcal Software Inc., Northampton, MA). Differences between experimental groups were considered statistically significant with p<0.05-0.01.
The effect of the hypomorphic Gata1low mutation on the size of progenitor cell compartments in the marrow was assessed by comparing total progenitor cell numbers calculated on the basis of frequencies of hematopoietic colonies generated in cultures by MNC and by prospectively isolated progenitor cells (Table I,,IIII and Figure 1). Progenitor cells were phenotypically defined as cells that bind CD117 (the antibody that recognizes cKit, the receptor for stem cell factor) (Besmer, 1991), in combination with the stem cell marker Sca1 (Spangrude et al., 1988), or CD34, an antigen that distinguishes megakaryocytic/erythroid restricted (MEP, CD117pos/CD34low) and common myeloid (CMP) and granulo/monocytic restricted (GMP) (both CD117pos/CD34high) (Akashi et al., 2000; Na Nakorn et al., 2002) progenitor cells.
As described (Vannucchi et al., 2002), due to fibrosis, the femur from Gata1low mice contained lower numbers of MNC than the femur from wild-type animals (5.2 vs 18 ×106 cells, respectively) (Table I). MNC from Gata1low marrow generated numbers of hematopoietic colonies slightly lower (and statistically different only at day 15) than those observed in the corresponding wild-type cultures (Table I). In addition, in Gata1low mice, the frequencies of phenotypically defined progenitor cells in the marrow were also normal (Figure 1A and Table II). Therefore, by colony assay on MNC and by phenotype analyses, the difference in progenitor cell compartments in the marrow of Gata1low and wild-type littermates were modest and reflected mainly decreased cellularity due to fibrosis.
Progenitor cells prospectively isolated from the marrow of wild-type mice generated numerous hematopoietic colonies in culture (average cloning efficiency 0.85–32% at day 8 and 14–77% at day 15) (Table II and Figure 1) confirming that these phenotypes predict hematopoietic activity in vitro (Akashi et al., 2000; Na Nakorn et al., 2002; Spangrude et al., 1988; Vannucchi et al., 2002). By contrast, progenitor cells prospectively isolated from the marrow of Gata1low mice had poor cloning efficiency (0.1–14% at day 8 and 0.1–15% at day 15) (Table II). Therefore, in the mutant mice, progenitor cell phenotype/function correlations were lost in the marrow (Figure 1A). Calculations based on these results indicated that the size of Gata1low progenitor cell compartments in the marrow is 10–100-times lower than normal (Figure 1A).
The effect of the hypomorphic Gata1low mutation on the size of progenitor cell compartments in the spleen was also assessed (Figure 1 and Table I and andII).II). Progenitor cells were phenotypically defined according to the same criteria used to characterize cells from marrow (Akashi et al., 2000; Na Nakorn et al., 2002; Spangrude et al., 1988).
The spleen from Gata1low mice was bigger than normal and contained 7-fold more LDC that the spleen from wild-type littermates (4.6 vs 31×106 LDC, respectively) (Table I). LDC from the spleen of Gata1low mice generated 3–4-times more hematopoietic colonies that the corresponding wild-type cells. Therefore, by calculation, the spleen from Gata1low mice contained 10-times more progenitor cells than the spleen of wild-type mice (1.7–2.4×105 vs 0.6–1.1×104 progenitors cells, respectively) (Figure 1B). Similarly, the frequencies of phenotypically defined progenitor cells in the spleen of Gata1low mice were 10-times greater than normal (Figure 1B and Table II). According to these data, the difference in progenitor cell content of the spleen from Gata1low and wild-type littermates was modest.
Progenitor cells prospectively isolated from the spleen of wild-type mice generated low numbers of colonies in vitro (average cloning efficiency <1%, Table II). Therefore, the spleen of wild-type mice contained barely detectable numbers of progenitor cells capable to form colonies under conditions of limiting dilution (<100 progenitor cells/spleen) (Figure 1B) confirming that in wild-type mice the vast majority of progenitor cell activity is prospectively isolated from marrow. By contrast, in the case of Gata1low mice, progenitor cells prospectively isolated from the spleen generated great numbers of colonies (cloning efficiency 15–68% at day 8 and 50–100% at day 15). Therefore, the size of progenitor cell compartments in the spleen calculated on the basis of colony numbers generated by purified cells and LDC were similar, indicating that in these mutant mice progenitor cell phenotype/function correlations were conserved in the spleen (Figure 1B).
In conclusion, in Gata1low mice, the vast majority (2-log more than in the marrow) of progenitor cell activity was prospectively isolated from spleen.
The function of hematopoietic progenitors is determined by appropriate levels of transcription factor expression. Hematopoietic progenitor cell maturation is sequentially controlled by Gata2 and Gata1(Orkin and Zon, 2008). The proteins encoded by these two genes are very similar and may exert redundant functions (Ferreira et al., 2007). Since the stage specificity of the function of these two proteins is ensured by their expression levels (Arinobu et al., 2007), retention of Gata2 expression with maturation has been proposed as the mechanism that rescues Gata1low progenitor cells (Huang et al., 2009). To confirm this hypothesis, the levels of Gata1 and Gata2 expressed by progenitor cells (CD117pos/CD34high, CMP/GMP, and CD117pos/CD34low, MEP) prospectively isolated from the spleen and marrow of wild-type and Gata1low mice were compared by quantitative RT-PCR (Figure 2). The levels of expression were standardized to those expressed by the corresponding cell populations isolated from the marrow of wild-type animals. In the case of wild-type mice, both MEP and CMP/GMP purified from the spleen expressed low levels of Gata1 but only MEP expressed high levels of Gata2 (>18-fold higher than those expressed by MEP purified from the marrow). Therefore, the poor cloning efficiency of wild-type cells prospectively isolated from the spleen was associated with low levels of Gata1 expression. In the case of Gata1low mice, MEP and CMP/GMP purified both from the marrow and from the spleen expressed reduced levels of Gata1 (Figure 2). However, progenitor cells purified from the spleen expressed high levels of Gata2 (>14-times the levels of expression in cells from wild-type marrow).
These results indicate that low levels of Gata1 expression is a property of progenitor cells purified from the spleen (wild-type and Gata1low alike) and that high levels of Gata2 expression may represent at least one of the mechanisms that rescue the function of Gata1low progenitor cells in this organ.
Extramedullary hematopoiesis in MF is thought to be, at least partially, mediated by increased stem/progenitor cell trafficking (Hoffman and Xu, 2006) induced by the high levels of growth factors (TGF-β, osteocalcin, VEGF, PDGFα and SDF-1) released in the marrow microenvironment by stromal cells activated by immature Gata1low MK (Lataillade et al., 2008; Migliaccio et al., 2008). The observation that removal of the spleen favours wild-type hematopoiesis including the formation of Gata1pos MK, in the marrow of heterozygous Gata1low/+ females (Migliaccio et al., 2009), suggested studies to compare the levels of TGF-β, osteocalcin, VEGF, PDGFα and SDF-1 expressed in the marrow of untreated and splenectomised Gata1low/+ females (Figure 3). As expected, the marrow from untreated Gata1low/+ females expressed all the cytokines analysed at levels significantly higher than those expressed by wild-type marrow. Removal of the spleen did not reduce the levels of cytokine expression by the marrow and significantly increased the expression of osteocalcin (at 3 and 6 month) and VEGF (at 9-months) (Figure 3).
These results indicate that the activated stromal cells of the marrow from Gata1low/+ females remain activated even when Gata1low MKs are no longer formed in this organ.
The high levels of cytokines produced in the marrow after splenectomy suggested that removal of the spleen may not halt stem/progenitor cell trafficking and may even favour development of extramedullary hematopoiesis in liver. To test this hypothesis, the presence of progenitor cells in the blood and of extramedullary hematopoiesis in the liver of heterozygous Gata1low/+ females after splenectomy were investigated (Figure 4 and and5).5). These experiments included mice carrying the mutation in backgrounds permissive (CD1) and not (DBA/2) for development of this trait (Martelli et al., 2005), as well as wild-type mice as control.
By colony assay, hematopoietic progenitor cells were detectable in blood (5 progenitor cells/µL) and liver (10–20 progenitor cells/2×104 cells) of CD1 and DBA/2-Gata1low/+ females after splenectomy but not in blood from splenectomised wild-type mice. Therefore, in Gata1low mutants progenitor cell trafficking persisted after splenectomy.
As previously published (Martelli et al., 2005), at 15–18-months of age, the liver parenchyma of untreated CD1-Gata1low mice, but not that of DBA/2-Gata1low mice, contained numerous hemopoietic cell clusters recognizable by hematoxylin-eosin staining (Figure 4). By the more sensitive CD45 immunostaining method, numerous hematopoietic cells (CD45pos) were detected in the parenchyma of untreated CD1-Gata1low mutants while few were present in liver of untreated DBA/2-Gata1low mice (Figure 4). Removal of the spleen did not induce hematopoiesis in the liver parenchyma of wild-type mice. By contrast, massive numbers of hematopoietic foci were observed in the liver of both CD1- and DBA/2-Gata1low/+ mice after splenectomy (Figure 4).
The extent of extramedullary hematopoiesis in the liver of DBA/2 mutants was quantified by FACS analyses (Figure 5A). At 15–19 months of age, the liver of untreated DBA/2 Gata1low/+ mutants contained negligible numbers of cells expressing the phenotype of stem/progenitor cells (CD117pos, 0.7%), MK (CD61pos/CD41pos, 1.3%) and erythroblasts (CD71pos/TER119pos, 1.9%) (Figure 5A). By contrast, in the corresponding splenectomized littermates, the frequency of stem/progenitor cells, MK and erythroblasts rose to 6.1, 6.7 and 10.2%, respectively.
By colony assay, both well hemoglobinized and poorly hemoglobinized bursts were observed in cultures of liver cells from the mutant mice after splenectomy (Figure 5B), suggesting that the liver of these animals contained both Gata1+ and Gata1low progenitor cells. Since genetic markers for clonal analyses of hematopoietic colonies in mice are not available, the progenitor cell type, Gata1+ and/or Gata1low that colonized the liver of these females and its contribution to hematopoiesis in this organ was inferred from the levels of Gata1 expressed by cells prospectively isolated from the liver of untreated and splenectomized CD1-Gata1low/+ mice (Figure 5C). The levels of Gata1 expressed by the corresponding populations purified from the marrow of untreated wild-type, Gata1low and Gata1low/+ mice served as comparison. In the case of the marrow, all the Gata1low cell populations expressed levels of Gata1 significantly lower than those expressed by the corresponding wild-type cells. Cells purified from Gata1low/+ females expressed instead Gata1 levels intermediate between the wild-type and Gata1low ones, indirectly confirming that in the presence of the spleen both wild-type and Gata1low progenitor cells contribute to hematopoiesis in the marrow of heterozygous females20. Also, cells purified from the liver of untreated CD1-Gata1low/+ females expressed levels of Gata1 intermediate between values expressed by wild-type and Gata1low cells. By contrast, CD117pos cells purified from the liver of splenectomised CD1-Gata1low/+ females expressed levels of Gata1 similar to those expressed by wild-type marrow cells, suggesting that the majority of them expressed the Gata1+ allele (Figure 5C). Erythroid cells from the liver of these females expressed levels of Gata1 similar to those expressed by cells purified from the marrow of untreated Gata1low/+ mice and therefore, were likely to be derived both from Gata1+ and Gata1low progenitor cells. MK purified from the liver of these splenectomized animals, however, expressed levels of Gata1 as low as those expressed by MK purified from the marrow of Gata1low mice (Figure 5A), indicating that megakaryocytopoiesis in the liver was mostly Gata1low.
These results indicate that, althought the majority of progenitor cells in the liver of splenectomized heterozygous Gata1low/+ females express the Gata1+ allele, both Gata1+ and Gata1low progenitor cells contribute to erythropoiesis and only Gata1low progenitor cells contribute to megakaryocytopoiesis in the liver after splenectomy.
The results presented in this study indicate that the spleen plays an important role in the Gata1low mouse model of MF by rescuing the hematopoietic functions of Gata1low stem/progenitor cells and by trapping these cells, possibly preventing colonization of additional extramedullary sites. Although calculations based on colony numbers in cultures from MNC indicated modest differences in progenitor cell compartment sizes between marrow and spleen of wild-type and Gata1low littermates (Figure 1), calculations based on colony numbers generated by prospectively isolated cell populations indicated that Gata1low progenitor cells capable to form colonies in vitro were localized mainly in spleen (Figure 1 and Table I and andII).II). Differences in estimates of progenitor cell populations determined by the two assays reflect the fact that progenitor cell phenotype/function correlations were lost in the marrow of Gata1low mice (Figure 1). This loss of correlation was associated with low Gata1 expression (Figure 2). Also Gata1low progenitor cells from the spleen expressed low levels of Gata1. These cells, however, unexpectedly expressed levels of Gata2 14-fold higher than normal. These results suggest that in the spleen Gata1low hematopoiesis is rescued, at least in part, by microenvironmental cues (the nature of which remains to be identified) which allow retention of Gata2 expression with maturation.
In contrast to the extensive knowledge obtained in recent years on microenviromental control of hematopoiesis in the marrow (Adams and Scadden, 2006; Moore and Lemischka, 2006; Papayannopoulou and Scadden, 2008; Yin and Li, 2006), much less is known about the microenviromental properties of the spleen. Spleen hematopoiesis, in fact, was actively considered in the past (Curry et al., 1967; Jandl et al., 1965; La Pushin and Trentin, 1977) but few studies (Slayton et al., 2002; Yanai et al., 1991) have analyzed the hematopoietic role of this organ in recent years. A recent publication has suggested BMP4 as an extramedullary-specific cue whose expression is specifically involved in activation of spleen erythropoiesis under conditions of stress (Perry et al., 2009). BMP4, however, has also been identified as a growth factor that regulates hematopoiesis in the marrow (Goldman et al., 2009). It is possible then that other pathways, such as the retinoblastoma pathway, may cooperate with BMP4 in determining the Gata1low specificity of the microenvironmental cues. In fact, the concept that microenvironmental cues could be permissive for the development of myeloproliferative disorders was developed on the basis of the phenotype expressed by mice in which either the retinoblastoma gene (Walkley et al., 2007a; Walkley and Orkin, 2006), or the gene encoding the γ subunit of the retinoic acid receptor (Walkley et al., 2007b), was inactivated. These mutants experience profound myeloproliferation with increased numbers of stem/progenitor cells in bone marrow, increased levels of stem/progenitor cell trafficking and extramedullary hematopoiesis in the spleen. Transplantation experiments, however, demonstrated that the myeloproliferation trait segregates with the genotype of the host microenvironment and not with that of the hematopoietic stem cells of the donor suggesting that these mutations may alter the ability of the extramedullary microenvironment to support stem cell proliferation. Further studies are necessary to clarify whether microenvironmental cues that rescue Gata1low hematopoiesis in the spleen are represented by BMP4 and/or the retinoblastoma pathway or other factors still to be identified.
Splenectomy favoured wild-type hematopoiesis in the marrow of heterozygous Gata1low/+ mice (Migliaccio et al., 2009) but did not halt stem/progenitor cell trafficking and promoted development of extramedullary hematopoiesis in the liver even in DBA/2 mutants (Figure 4 and and5)5) that would not normally develop this trait (Martelli et al., 2005). The observation that splenectomy restored maturation of Gata1pos MK in the marrow (Migliaccio et al., 2009), but did not reduce the high cytokine expression profile of this organ (Figure 3), suggests that marrow stromal cells, once activated by Gata1low MKs, remain activated even in the absence of Gata1low MKs, perhaps explaining why splenectomy did not halt stem/progenitor cell trafficking. In addition to cytokines, activated stromal cells of PMF patients (Hoffman and Xu, 2006) and Gata1low mice (Centurione et al., 2004) express high levels of the metalloprotease MMP9. Recently it has been suggested that MMP9 forms with other as-yet unidentified proteins a supramolecular structure defined as “the invadosome” that mediates extramedullary infiltration by leukemic cells (Stefanidakis et al., 2009). It is conceivable that MMP9 is organized into an invadosome-like complex also in PMF patients and in Gata1low mice contributing to stem/progenitor cell infiltration in extramedullary sites. The fact that stromal cells of Gata1low/+ females remained activated after splenectomy suggests that these cells, in addition to cytokines, are likely to produce great levels of “invadosome” explaining why removal of the spleen facilitated infiltration of other organs (the liver) by stem/progenitor cells (wild-type and Gata1low alike). The failure of wild-type stem/progenitor cells to generate erythroid and megakaryocytic cells in the liver of splenectomized Gata1low/+ mutants may then reflect again the fact that the microenvironmental cues of the extramedullary sites favors maturation of Gata1low progenitor cells. Although additional studies are required to assess whether an invadosome-like structure is involved in extramedullary hematopoiesis in the Gata1low model of PMF, our results are consistent with the recent hypothesis that the spleen in patients with PMF might also actively participate in the development of disease manifestations by selectively interacting with subpopulations of cells belonging to the malignant clone (Lane et al., 2009; Lataillade et al., 2008). In fact, splenectomy in patients with PMF has also been associated with a higher incidence of development of extramedullary hematopoiesis in liver and higher rate of blast transformation (Cervantes et al., 2007). Further evidence for such a role of the spleen in the biology of PMF has recently been provided by observations made during the initial clinical trials with JAK2 inhibitors in patients with PMF. The administration of these drugs led to a dramatic rapid reduction of spleen size without affecting other haematological parameters, clearly indicating the existence of a subpopulation of malignant cells that might preferentially reside in the spleen (Verstovsek et al., 2007a; Verstovsek et al., 2007b).
In conclusion, these results support the notion that the spleen plays an important role in the development of MF in Gata1low mice by providing a permissive microenvironment for the mutant stem/progenitor cells. However removal of this organ restored stem/progenitor cell function but not microenvironmental function in Gata1low/+ females resulting in development of hematopoiesis in extramedullary sites. These results suggest that treatment of MF in Gata1low mice requires restoration of both hematopoietic and microenvironmental functions of the marrow.
This study was supported by Ministero per la Ricerca Scientifica and Alleanza sul Cancro, Italy, the National Cancer Institute, grant no. P01-CA108671 and NY STAR program, USA.