A variety of hematopoietic growth factors and cytokines have been shown to induce phosphorylation of STAT5, although functional data on the effects of STAT5A activation in human hematopoietic cells is still limited. Here, we show that a persistent activation of STAT5A in CB CD34+ cells results in enhanced human HSC self-renewal and erythropoiesis.
One critical component in the phenotypes observed by overexpression of the constitutively active mutant STAT5A(1*6) in CB CD34+
cells is the interaction of HSCs with the surrounding stroma. Within a few days after plating, CB CD34+
cells transduced with STAT5A(1*6) form CAFCs on various stromal cell lines, including MS5, and these CAFCs can be cultured for at least up to 18 wk. Typically, the production of CAFCs at weeks 5 or 6 in ex vivo expansion cultures has been shown to be indicative of the presence and quantity of in vivo long-term repopulating HSCs, as measured using the quantitative NOD-SCID transplantation model, whereas CAFCs arising earlier have been associated with short-term engrafting cells (33
). In the case of overexpressed STAT5A(1*6), the CAFCs that arise within 5–10 d can be grown for at least 18 wk by serial passage on new stromal cells every 2–3 wk, giving rise to nonadherent progeny that is mostly erythroid, but also contains a small subset of CD11b+
myeloid cells, suggesting that these CAFCs represent self-renewing cells with the capacity to give rise to predominantly erythroid and, to a lesser extent, myeloid progeny. Interestingly, in liquid cultures, STAT5A(1*6)-transduced cells did not have a proliferative advantage over control MiGR1 cells, as observed in stromal cocultures. Also, no pronounced erythroid differentiation was observed in STAT5A(1*6) cells in liquid cultures. These data indicate that the interaction with stroma is critical for the self-renewal, proliferative advantage, and erythropoiesis of STAT5A(1*6)-expressing cells. Perhaps, VEGF or LIF, which are both up-regulated in CB CD34+
cells transduced with STAT5A(1*6), are critical in ensuring an appropriate microenvironment for HSC self-renewal in stromal niches, a possibility that will certainly be a focus of future research. Indeed, it has been demonstrated that LIF can up-regulate the production of stem cell expansion–promoting factors in stromal cells, enabling the maintenance of highly enriched competitive repopulating stem cells (37
Although in control cells ~1 in 100 cells will give rise to a CAFC at week 5 on MS5, in STAT5A(1*6)-expressing cells this has increased to ~12 in 100 cells within 10 d. Upon secondary passage, one STAT5A(1*6) CAFC gave rise to four to five secondary CAFCs. The self-renewing HSCs reside in the immature CD34+
population. In experiments in which GFP+
populations were sorted directly after transduction and analyzed for CAFC, secondary CAFC, and CFC/CAFC ratios, we found that the CD38high
fraction contains some CAFCs, but no secondary CAFCs and few CFCs/CAFC, whereas the majority of CAFCs, secondary CAFCs, and CFC/CAFC activities was present in the CD38low
population. The CAFCs generated by this CD34+
fraction contained mostly BFU-E progenitors and some CFU-GM and CFU-mix progenitors ( F). The secondary CAFCs generated predominantly erythroid and some myeloid cells (not depicted). Although it has been described that the CD34+
population contains both long-term and short-term repopulating stem cells, our data indicate that a persistent activation of STAT5A imposes self-renewal characteristics on this population. Microarray and RT-PCR analyses revealed that STAT5A(1*6) does not enhance the expression of genes that have previously been associated with HSC self-renewal, such as HOXB4, NOTCH1, BMI11, or β-catenin (4
), although these genes are being expressed, as determined by our microarray analysis (not depicted). The proto-oncogenes FOSB and PIM1 were up-regulated in CB CD34+
cells, and PIM1 has previously been identified as a STAT5 target gene (40
). In particular, PIM1 has been associated with enhanced cell proliferation, cytokine independent growth, and cellular transformation processes (28
), and it will be of interest to determine its role in HSC self-renewal.
The early CAFCs can engraft sublethally irradiated NOD-SCID mice with frequencies of 0.1–1%. As we injected 2 × 105
CAFCs into NOD-SCID mice, of which ~5% were CD34+
, these engraftment frequencies are comparable to those that have been reported for fresh CB (33
). It has been suggested that upon plating on stromal cells, HSCs are initially in a quiescent state and reside in a nonproliferative niche, comparable to the situation in the bone marrow under nonmyelosuppressive conditions (44
). Bone marrow ablation induces the release of soluble KL, enabling bone marrow–repopulating cells to translocate to a permissive vascular niche, favoring differentiation and reconstitution of the stem/progenitor cell pool (44
). On stroma, stem cells start to proliferate within 5 wk, giving rise to CAFCs with in vivo multilineage reconstitution potential. As overexpression of STAT5A(1*6) results in a shift toward early CAFCs, we hypothesize that these HSCs are less quiescent, but rather “activated,” resulting in the generation of CAFCs with long-term self-renewal potential ex vivo. Recently, Bradley et al. (45
) reported that under 5-FU–induced myelosuppressive conditions, stat5a−/− b−/−
repopulating stem cells are indeed more quiescent and much less responsive to early-acting cytokines that may play a role in repopulation. These observations and our data suggest that STAT5 fulfills an important function in the hematopoietic reconstitution and self-renewal properties of HSCs.
Although expression of STAT5A(1*6) in CB CD34+ cells results in the generation of CAFCs that can be cultured for up to 18 wk, give rise to new CAFCs with every passage, and engraft sublethally irradiated NOD-SCID mice, we cannot rule out the possibility that STAT5A activity promotes the expansion and self-renewal of early erythroid progenitors with reconstitution potential rather than true HSC expansion and self-renewal. Clearly, most of our data, including ex vivo long-term cultures on MS5 and in vivo hematopoietic reconstitution studies in NOD-SCID mice, indicate that cells predominantly differentiate along the erythroid lineage, which would support such a model. Nevertheless, after 18 wk on MS5, we still find ~10% CD11b+ cells. In secondary CFCs assays, we observed that only STAT5A(1*6) cells give rise to secondary CFCs, which are mostly CFU-GMs, suggesting that STAT5A(1*6) promotes self-renewal of an early cell that certainly can give rise to myeloid progeny. The CAFCs generated by STAT5A(1*6) still contained significant amounts of CFU-GM and CFU-mixed progenitors ( F), and upon long-term culture on MS5, we observed a gradual loss of myeloid progenitors. Further studies in which STAT5A(1*6) is expressed in various sorted subsets, including early HSCs and more differentiated erythroid and myeloid progenitors, are currently ongoing and will help to clarify some of these issues.
Our data are in agreement with previously published observations in mice that have suggested a role for STAT5 in HSC self-renewal (45
). In competitive repopulation assays, bone marrow and fetal liver cells of stat5a−/− b−/−
mice displayed a decreased repopulating activity in granulocyte, macrophage, erythroid, and B lymphocyte populations, with no detectable engraftment of T lymphocytes (46
). These results indicated that a significant proportion of the growth factor signals required for multilineage reconstitution potential of HSCs is dependent on STAT5. In a similar study, Snow et al. (47
) also demonstrated that STAT5-null HSCs have a profound impairment in repopulating potential, and they suggested that STAT5 is required to sustain a robust hematopoietic reserve that contributes to host viability through crucial survival effects on early progenitor cells. These data indicate that although STAT5-deficient mice are viable and steady-state hematopoiesis is fairly normal, STAT5A/B is an important positive factor for HSC fitness and multilineage hematopoiesis. Our data indicate that constitutive activation of STAT5 is in fact sufficient for long-term human HSC self-renewal. Currently, studies are ongoing to further evaluate the effects of STAT5A(1*6) on multilineage reconstitution potential of human HSCs in vivo in NOD-SCID mice, and we have observed similar phenotypes in murine embryonic stem cell–derived hematopoietic cells in which overexpression of STAT5A(1*6) facilitates hematopoietic differentiation and results in long-term HSC self-renewal in vitro and in vivo (48
Although adult mice lacking STAT5A/B have normal hematocrit and hemoglobin concentrations, stat5a−/− b−/−
embryos are severely anemic due to massive apoptosis and a reduced response to EPO (17
). These findings and others have implicated STAT5 in EPO downstream signaling and have suggested an antiapoptotic role for STAT5A in erythroid progenitors by up-regulating genes such as bcl-xL
. Besides effects on antiapoptosis, our studies indicate that a persistent activation of STAT5A directly drives erythropoiesis by up-regulating genes involved in erythroid differentiation. It will be of interest to determine whether the promoters of the STAT5A(1*6)–up-regulated erythroid genes contain bona fide STAT5 binding elements or whether these genes are up-regulated via an indirect mechanism. Furthermore, BCL2 and BCL-XL
were not up-regulated by STAT5A(1*6) within 48 h of expression in CB CD34+
cells, as determined by microarray analysis as well as by RT-PCR, whereas an increase of these transcripts was observed in week 2 nonadherent cells from MS5 cocultures. These data suggest that initially STAT5A(1*6) drives erythroid differentiation by the up-regulation of critical erythroid genes (and/or the concomitant down-regulation of other lineage-specific genes), whereas in later phases during differentiation STAT5 will also play a role in the survival of erythroid progenitors by preventing apoptosis.
Finally, our data provide some important insights into the possible roles that STAT5 proteins might play in the development of leukemic malignancies. A constitutive activation of STAT5 has been observed in acute myeloid leukemia (21
), CML (25
), and idiopathic myelofibrosis (49
), either as a result of chromosomal translocations such as Bcr/Abl in the case of CML, or possibly due to a disturbed cytokine production by leukemic blasts themselves or the surrounding stroma (50
). Our data indicate that a persistent activation of STAT5A in HSCs results in enhanced self-renewal, whereas differentiation along the myeloid lineage toward macrophages and granulocytes is severely impaired. Evidence is now accumulating that suggests that acute leukemias are clonal disorders in which blastic cells have stem cell self-renewal characteristics with a concomitant block in differentiation toward mature blood cells (52
). It is now conceivable that a persistent activation of STAT5A might play a causal role in the development of such disorders in humans and it would therefore be of great interest to develop therapeutically agents that specifically inhibit STAT5. Little is known about the activation patterns of STAT5 in erythroleukemias in humans, but this is certainly a focus of future research, as our data directly indicate that a constitutive activation of STAT5 is sufficient to drive erythropoiesis at the expense of myeloid differentiation.