Evi1 is essential for hematopoietic stem cell self-renewal, and its expression marks hematopoietic cells with long-term multilineage repopulating activity

To elucidate Evi1 expression within the hematopoietic system, we have generated gene-targeted mice in which an internal ribosome entry site (IRES)-GFP cassette is knocked-in to the Evi1 locus by homologous recombination (Fig. 1 A). This knock-in allele functions in a bicistronic manner in that expression of both Evi1 and GFP is under the endogenous transcriptional regulatory elements of the Evi1 gene, thus enabling us to track Evi1 expression on an individual cell basis. Appropriately targeted TT2 embryonic stem (ES) cell clones were identified by Southern blotting (Fig. 1 B). Mice heterozygous for the Evi1-IRES-GFP allele (Evi1^+/GFP) were distinguished from WT mice by genotyping PCR (Fig. 1 C). Western blot analysis showed the presence of GFP protein and comparable expression of Evi1 protein in embryonic fibroblast cells from Evi1^+/GFP mice compared with WT mice (Fig. 1 D). Evi1^+/GFP mice were phenotypically indistinguishable in survival, hematopoietic cellularity, and lineage composition from WT controls (unpublished data). Initial flow cytometric analysis of adult Evi1^+/GFP mice revealed a small, but discrete, population of GFP⁺ cells (0.15 ± 0.6%; Fig. 2 A), confirming the expression of the Evi1-IRES-GFP allele. To examine whether GFP expression levels correlated with those of endogenous Evi1 mRNA expression, Evi1 expression of sorted GFP⁻ and GFP⁺ cells from BM of Evi1^+/GFP mice was analyzed by real-time quantitative PCR (RQ-PCR). Evi1 mRNA was exclusively expressed in the GFP⁺ cells, and almost no expression was found in the GFP⁻ cells (Fig. 2 B), indicating that GFP expression in this mouse model faithfully marks cells with active Evi1 expression.

Evi1 mRNA has been shown to be expressed at significantly higher levels in HSPCs (Lin⁻ Sca-1⁺ c-kit⁺ [LSK]) and common lymphoid progenitors (CLPs) than in other hematopoietic cells (Yuasa et al., 2005; Chen et al., 2008). To gain insight into the biological function of Evi1 through its cell type–specific expression pattern, the distribution of GFP⁺ cells was examined in adult BM from Evi1^+/GFP mice. Beyond expectation, GFP expression was highly restricted to the LSK fraction (Fig. 2 A). To confirm stem/progenitor-specific expression of Evi1, we analyzed the GFP fluorescence of various hematopoietic cell populations from BM and spleen of Evi1^+/GFP mice. We found a heterogeneous expression of GFP in the LSK fraction, in which about half of the cells were GFP⁺ (Fig. 2, C and D). Conversely, only 2.5% of common myeloid progenitors (CMPs) expressed GFP, and almost no expression was found in granulocyte/monocyte progenitors (GMPs) and megakaryocyte/erythrocyte progenitors (MEPs; Fig. 2 C). In contrast to the previous study (Chen et al., 2008), GFP was not expressed in CLPs (Fig. 2 C). In addition, no GFP expression was observed in mature hematopoietic lineages or nonhematopoietic cells in BM (Fig. 2 C). Together, these results suggest that Evi1 is uniquely expressed in HSPCs, but its expression is sharply down-regulated along with differentiation.

Because LSK cells, a population which contains multipotent progenitors (MPPs), short-term HSCs (ST-HSCs), and long-term HSCs (LT-HSCs), include both a GFP⁺ fraction and a GFP⁻ fraction, we next resolved GFP expression within the LSK compartment for other markers characteristic of LT-HSCs. When LSK cells were subdivided according to CD34 and Flk-2 expression (Orford and Scadden, 2008), the Flk-2⁻ CD34⁻ LSK fraction, which is considered to contain most LT-HSC activity, had the highest expression of GFP, and its expression decreased with differentiation to hematopoietic progenitors (Fig. 2 E). In addition, further enrichment for LT-HSCs within the LSK fraction using SLAM family receptors (CD48 and CD150; Kiel et al., 2005) revealed that GFP⁺ cells were found in greatest abundance within CD48⁻ CD150⁺ LSK cells, in which LT-HSCs are highly enriched. In contrast, GFP expression was substantially down-regulated in CD48⁺ LSK cells, irrespective of CD150 expression (Fig. 2 F). When we examined how GFP⁺ cells were distributed within the LSK fraction, GFP expression was highly enriched in the Flk-2⁻ CD34⁻ LSK or CD48⁻ CD150⁺ LSK fractions (Fig. 2, G and H). Therefore, these results indicate that Evi1 is dynamically regulated within HSPCs; its expression is predominantly enriched in LT-HSCs and rapidly extinguished during early stages of lineage commitment.

To reinforce Evi1-IRES-GFP knock-in mice as a faithful tool for investigating HSCs, we assessed the number and function of LT-HSCs in BM from Evi1^+/GFP mice. Flow cytometric analysis revealed that the frequencies of Flk-2⁻ CD34⁻ LSK or CD48⁻ CD150⁺ LSK cells were comparable between Evi1^+/+ and Evi1^+/GFP mice (Fig. 2 I). In addition, a competitive repopulation assay (CRA) showed that Evi1^+/GFP BM cells exhibited slightly less, but not significantly different, long-term reconstitution capacity (Fig. 2 J), indicating that the number and function of HSCs in Evi1^+/GFP mice are similar to WT controls.

As only a subset of LSK cells expressed GFP in Evi1^+/GFP mice, we hypothesized that Evi1 expression functionally divides the LSK population and marks a more undifferentiated and quiescent state with multipotent differentiation properties in this population. To test this idea, we separated the LSK population into LSK GFP⁻ and LSK GFP⁺ cells and compared their biological functions. Initially, we confirmed that LSK GFP⁺ cells had a much higher level of Evi1 transcripts than LSK GFP⁻ cells by RQ-PCR analysis (Fig. 3 A). Interestingly, despite the negative GFP expression, LSK GFP⁻ cells expressed Evi1 mRNA at a higher level compared with CMPs and GMPs (Fig. 3 A), which also suggests that Evi1 expression is inversely proportional to the differentiation status. To achieve an estimate of the differentiation stage of these two populations, LSK GFP⁻ and LSK GFP⁺ cells were cultured in serum-free medium containing stem cell factor (SCF) and thrombopoietin (TPO). After 3 d of culture, the proportion that remained in the LSK fraction was significantly higher in LSK GFP⁺ cells than in LSK GFP⁻ cells (Fig. 3 B), suggesting that LSK GFP⁺ cells are more primitive HSCs. Next, to evaluate the differentiation potential of LSK GFP⁻ and LSK GFP⁺ cells, we performed colony-forming assays in vitro. Although both populations generated an equivalent number of myeloid colonies CFU-granulocyte/macrophage [CFU-GM]), LSK GFP⁺ cells gave rise to greater numbers of erythroid (burst-forming unit-erythrocyte [BFU-E]) and multipotential (CFU-granulocyte/erythrocyte/macrophage/megakaryocyte [CFU-GEMM]) colonies than LSK GFP⁻ cells (Fig. 3 C). These data suggest that Evi1 expression correlates with multipotent differentiation capacity. In addition, to assess the colony-forming capacity at the clonal level, single LSK GFP⁻ and LSK GFP⁺ cells were cultured in serum-free medium. LSK GFP⁻ cells formed detectable colonies at a frequency comparable to LSK GFP⁺ cells, but generated smaller numbers of highly proliferative colonies (>300 cells; Fig. 3 D), indicating that the LSK GFP⁺ fraction comprises a higher proportion of HSPCs with enhanced proliferative capacity.

Our observations suggested that Evi1 reporter activity is down-regulated as HSCs differentiate. To examine this issue, we forced LSK GFP⁺ cells to differentiate in vitro in response to SCF, TPO, IL-3, and IL-6. These LSK GFP⁺ cells predominantly generated GFP⁻ cells (Fig. 3 E). After culture, the majority of cells that had become GFP⁻ lost the LSK phenotype, whereas most cells that remained in GFP⁺ continued to express the LSK phenotype (Fig. 3 F), indicating that loss of GFP correlates with phenotypic differentiation. To confirm the differential phenotype of the GFP⁻ and GFP⁺ cells after culture reflected their functional status, we compared their ability to form colonies in methylcellulose. GFP⁺ cells yielded significantly more colonies than GFP⁻ cells (Fig. 3 G), suggesting that functionally primitive HSCs predominantly reside in the GFP⁺ fraction. Collectively, these data indicate that in vitro culture of LSK GFP⁺ cells leads to generation of GFP⁻ cells that are more differentiated, and lend credence to the use of GFP as a fluorescent sensor for the differentiation state of hematopoietic cells.

To determine the cell-cycle distribution of LSK GFP⁻ and LSK GFP⁺ cells, we performed Hoechst 33342 and pyronin Y staining, which revealed that the majority of LSK GFP⁺ cells were in G₀ phase, whereas a significant proportion of LSK GFP⁻ cells were actively cycling (Fig. 3 H). These data indicate that, within HSPCs, Evi1 expression represents a functionally distinct population that remains in an undifferentiated and quiescent state.

Based on the aforementioned data, we hypothesized that Evi1 expression would have the potential to effectively mark long-term multilineage repopulating HSCs. To examine this issue, we performed a CRA, in which 500 purified LSK GFP⁻ or LSK GFP⁺ cells were transplanted with 2 × 10⁵ competitor BM cells into lethally irradiated recipients (Fig. S1 A). At 16 wk after transplantation, flow cytometric analysis of donor-derived cells revealed long-term reconstitution in all recipients transplanted with LSK GFP⁺ cells (Fig. 4 A). Moreover, LSK GFP⁺ cells displayed multilineage potential with robust contribution to myeloid, B, and T cells in peripheral blood (PB) as well as the LSK fraction in BM (Fig. 4 B). In contrast, LSK GFP⁻ cells yielded an almost total inability to generate long-term chimerism (Fig. 4, A and B), which suggests that this population is devoid of self-renewal activity. To confirm the in vivo repopulating capacity of LSK GFP⁺ cells, we performed secondary transplantation. Similarly, LSK GFP⁺ cells showed remarkable long-term reconstitution, whereas LSK GFP⁻ cells consistently failed to produce detectable donor-derived cells (Fig. 4 C), demonstrating that in vivo long-term multilineage repopulating cells are exclusively enriched in the LSK GFP⁺ fraction in adult BM.

To further refine our analysis designating Evi1 expression as a robust and reliable HSC marker, we compared the repopulating capacity of GFP⁻ and GFP⁺ cells within the CD48⁻ CD150⁺ LSK fraction, which is enriched for LT-HSCs (Fig. S1 B). Intriguingly, CD48⁻ CD150⁺ LSK GFP⁺ cells exhibited long-term multilineage reconstitution, whereas no engraftment was observed in recipients of CD48⁻ CD150⁺ LSK GFP⁻ cells (Fig. 4, D and E), suggesting that long-term repopulating HSCs predominantly reside in the GFP⁺ fraction even within the highly subfractionated LT-HSC fraction. We then examined whether Evi1 expression is associated with repopulating capacity in the CD48⁺ CD150⁻ LSK fraction, which is enriched for ST-HSCs/MPPs with limited self-renewal activity (Fig. S1 B). Although CD48⁺ CD150⁻ LSK GFP⁻ cells provided only a transient reconstitution, CD48⁺ CD150⁻ LSK GFP⁺ cells showed declining, but sustained engraftment 16 wk after transplantation (Fig. 4 F). In contrast to CD48⁻ CD150⁺ LSK GFP⁺ cells, CD48⁺ CD150⁻ LSK GFP⁺ cells mediated faint myeloid but superior lymphoid reconstitution (Fig. 4, F and G). Although it is controversial whether CD48⁺ CD150⁻ LSK cells are transiently reconstituting MPPs/ST-HSCs or lymphoid-biased LT-HSCs with limited long-term engraftment and strong predominance of lymphoid reconstitution (Kiel et al., 2005; Weksberg et al., 2008; Grassinger et al., 2010), Evi1-expressing cells possess higher repopulating capacity within this fraction. When we subfractionated the LSK fraction according to CD34 and Flk-2 expression, and compared the repopulating capacity of GFP⁻ and GFP⁺ cells within these subsets (Fig. S1 C), we obtained similar results to the aforementioned findings using SLAM markers (Fig. 4, H–K). These data reveal that, irrespective of the combination of HSC surface markers used, GFP⁺ cells are the exclusive reservoir of HSC activity, with no reconstitution ability being observed in GFP⁻ cells within the LT-HSC compartment. Altogether, our results demonstrate that Evi1 expression can further augment the conventional HSC purification strategy, and suggest that Evi1-IRES-GFP knock-in mice allow us to functionally identify HSCs on the ground of self-renewal capacity.

The formation of blood cells begins in the yolk sac of the embryo, and then shifts to the aorta-gonad-mesonephros (AGM) region, and then sequentially to the placenta, fetal liver (FL), and adult BM. There are several major phenotypic and functional differences between fetal and adult HSCs in surface marker profile, cell cycle status, self-renewal potential, gene expression profile, and regulatory mechanism (Mikkola and Orkin, 2006; Orkin and Zon, 2008). Fetal HSCs, in particular, divide rapidly and undergo massive expansion, whereas adult HSCs are mostly quiescent (Bowie et al., 2006). It is known that Evi1 is highly expressed in the yolk sac, paraaortic splanchnopleura, and HSPCs (CD45⁺ CD34⁺ c-kit⁺) in early embryo (Yuasa et al., 2005). Therefore, we sought to determine whether Evi1 expression can mark fetal HSCs despite their distinct features from adult HSCs, and thus analyzed the expression pattern of GFP in Evi1^+/GFP embryos. As expected, GFP expression was highly restricted to HSPCs in the embryonic tissues; CD45⁺ CD34⁺ c-kit⁺ cells in embryonic day 10.5 (E10.5) AGM, CD34⁺ c-kit⁺ CD48⁻ cells in E12.5 placenta, and Mac-1⁺ Sca-1⁺ Lin⁻ (MSL) CD48⁻ cells in E14.5 FL (Fig. 5, A-C; Takakura et al., 2000; Kim et al., 2006; McKinney-Freeman et al., 2009). When the distribution of GFP⁺ cells in the fetal hematopoietic system was analyzed, most GFP⁺ cells exhibited the HSPC-specific marker profile in all embryonic tissues examined (Fig. 5, D–F), indicating the predominant expression of Evi1 in HSPCs during fetal hematopoiesis.

To determine whether Evi1 expression is associated with hematopoietic activity in the embryonic tissues, we performed colony-forming assays in vitro using CD45⁺ GFP⁻ and CD45⁺ GFP⁺ cells from E10.5 AGM, and found that CD45⁺ GFP⁺ cells contained almost all colony-forming cells, with few detectable hematopoietic colonies in CD45⁺ GFP⁻ cells (Fig. 5 G). In the same manner, within the CD34⁺ c-kit⁺ CD48⁻ fraction from E12.5 placenta of Evi1^+/GFP embryos, colony-forming activity was exclusively present in GFP⁺ cells, regardless of colony type (Fig. 5 H). These data suggest that clonogenic hematopoietic progenitors predominantly reside in the Evi1-expressing fraction in fetal hematopoiesis.

To examine whether Evi1 expression would have the potential to effectively mark long-term repopulating HSCs in embryo, we performed a CRA using sorted CD34⁺ c-kit⁺ CD48⁻ GFP⁻ and CD34⁺ c-kit⁺ CD48⁻ GFP⁺ cells from E12.5 placenta of Evi1^+/GFP embryos (Fig. S2 A). It was obvious that CD34⁺ c-kit⁺ CD48⁻ GFP⁺ cells contributed to the long-term reconstitution of irradiated recipients, whereas donor chimerism was almost undetectable in mice transplanted with CD34⁺ c-kit⁺ CD48⁻ GFP⁻ cells (Fig. 5 I), which is in agreement with the results obtained with their adult counterparts. To further assess whether Evi1 expression can enrich long-term repopulating HSCs in embryo, we performed a CRA using purified MSL GFP⁻ and MSL GFP⁺ cells from E14.5 FL (Fig. S2 B). Along with cells in E12.5 placenta, MSL GFP⁺ cells gave rise to long-term multilineage reconstitution, whereas no engraftment was observed in recipients of MSL GFP⁻ cells (Fig. 5 J). These results indicate that fetal HSCs with active Evi1 transcription exclusively harbor stem cell activity. Collectively, despite the functional differences between fetal and adult HSCs, Evi1 expression marks long-term multilineage repopulating HSCs throughout ontogeny, suggesting a specific relationship between Evi1 expression and HSC self-renewal capacity.

The aforementioned observations led us to predict that Evi1 plays a functional role specifically in LT-HSCs. To clarify this issue, we analyzed heterozygous Evi1 KO mice (Evi1^+/−). We previously showed that heterozygosity of Evi1 leads to decreased numbers of LSK and CD34⁻ LSK cells, as well as impaired long-term repopulating activity (Goyama et al., 2008). In the current study, although Flk-2⁺ CD34⁺ and Flk-2⁻ CD34⁺ LSK cells were moderately decreased, Flk-2⁻ CD34⁻ LSK cells from Evi1^+/− mice exhibited a marked reduction in frequency compared with WT controls (Fig. 6, A and C). Likewise, when LSK cells were subdivided according to SLAM markers, we observed substantial decreases in CD48⁻ CD150⁻ and CD48⁻ CD150⁺ LSK subsets (Fig. 6, A and D). Therefore, the number of each subpopulation within the LSK fraction in Evi1^+/− mice was declined in proportion to their expression level of Evi1, indicating that Evi1 has a dominating effect on the maintenance of LT-HSCs. In contrast, there were no significant differences in BM cellularity and the frequencies of lymphoid and myeloid progenitors, and mature blood cells between Evi1^+/+ and Evi1^+/− mice (Fig. 6 B and not depicted), indicating that the differentiation potential to all mature lineages and committed progenitors is normal in Evi1^+/− mice. Collectively, these observations suggest that Evi1 serves as a specific regulator in the earliest stage of adult hematopoietic development.

To exclude the possibility that a defect of BM microenvironment could be responsible for the observed hematopoietic abnormalities in Evi1^+/− mice, we performed reciprocal transplantation experiments, in which WT BM cells were transplanted into lethally irradiated Evi1^+/+ or Evi1^+/− mice. At 16 wk after transplantation, flow cytometric analysis showed no differences in the percentages of Flk-2⁻ CD34⁻ LSK or CD48⁻ CD150⁺ LSK cells in both groups of recipient mice (Fig. 6 E), demonstrating that the profound loss of LT-HSCs in Evi1^+/− mice is attributed to cell-intrinsic mechanisms.

To further characterize which subpopulation in HSPCs is most dependent on Evi1, we purified CD34⁺ and Flk-2⁻ CD34⁻ LSK cells from Evi1^+/+ and Evi1^+/− mice and compared their differentiation and self-renewal capacity in vitro and in vivo. First, to assess the effect of Evi1 heterozygosity on the biological functions of ST-HSCs/MPPs, we performed colony-forming assays in vitro using Evi1^+/+ and Evi1^+/− CD34⁺ LSK cells, which demonstrated no significant differences in the number and type of colonies (Fig. 7 A). Similarly, we found the capacity of Evi1^+/− CD34⁺ LSK cells to form colonies in the spleen 11 d after transplantation (CFU-spleen [CFU-S]) was also equivalent to that of WT littermates (Fig. 7 B), indicating Evi1 is dispensable for the regulation of the differentiation and proliferation capacity in ST-HSCs/MPPs. Moreover, to investigate the self-renewal ability of ST-HSCs/MPPs in vivo, we evaluated the short-term repopulating capacity of purified CD34⁺ LSK cells using a CRA. At 2 wk after transplantation, we detected comparable frequencies of Evi1^+/+ and Evi1^+/− CD34⁺ LSK cell–derived myeloid and B cells (Fig. 7 C), suggesting that heterozygosity of Evi1 does not affect the engraftment and differentiation potential of ST-HSCs/MPPs in vivo. However, at later time points in the experiment, we found a moderate but significant decline in the percentage of donor-derived cells from Evi1^+/− mice (Fig. 7 D). These data indicate that heterozygosity of Evi1 attenuates the self-renewal capacity of ST-HSCs/MPPs, but is not accompanied by any specific differentiation defects in them.

To assess whether Evi1 is required for the functions of LT-HSCs, we compared the self-renewal and proliferation capacity of Evi1^+/+ and Evi1^+/− Flk-2⁻ CD34⁻ LSK cells when cultured in serum-free medium. Evi1^+/− Flk-2⁻ CD34⁻ LSK cells showed comparable proliferation with WT cells for the first week of culture, but thereafter they exhibited pronouncedly impaired growth (Fig. 7 E). After incubation, a significantly lower proportion of cultured Evi1^+/− Flk-2⁻ CD34⁻ LSK cells remained in the LSK fraction than those from Evi1^+/+ mice (Fig. 7 F). In addition, we observed a prominent reduction of hematopoietic colonies contained in cultured Evi1^+/− Flk-2⁻ CD34⁻ LSK cells (Fig. 7 G). Besides, most of the colonies generated from cultured Evi1^+/− Flk-2⁻ CD34⁻ LSK cells consisted of only CFU-GM. These data indicate that heterozygosity of Evi1 results in accelerated loss of HSPCs, leading to the inefficient expansion of their progeny. To evaluate the colony-forming capacity at the single cell level, Evi1^+/+ and Evi1^+/− Flk-2⁻ CD34⁻ LSK cells were clonally sorted and cultured in serum-free medium. Evi1 heterozygosity diminished the colony-forming efficiency of clone-sorted Flk-2⁻ CD34⁻ LSK cells, and single Evi1^+/− Flk-2⁻ CD34⁻ LSK cells generated smaller colonies compared with control cells (Fig. 7 H), which indicates that the disruption of Evi1 gene not only decreases the number of clonogenic HSCs but also impairs the functional output per cell.

To assess the repopulating capacity of Evi1^+/− LT-HSCs in vivo, we performed a CRA using purified Flk-2⁻ CD34⁻ LSK cells from Evi1^+/+ and Evi1^+/− mice. Notably, Evi1^+/− Flk-2⁻ CD34⁻ LSK cells were almost unable to efficiently repopulate all mature lineages as well as stem and progenitor cells 16 wk after transplantation (Fig. 7, I and J), suggesting that LT-HSC function is critically dependent on Evi1 gene dosage. In a noncompetitive setting, although recipients of Evi1^+/+ and Evi1^+/− Flk-2⁻ CD34⁻ LSK cells had similar survival after transplantation (not depicted), Evi1^+/− Flk-2⁻ CD34⁻ LSK cells showed impaired engraftment (Fig. 7 K), suggesting that Evi1^+/− HSCs were outcompeted by residual host HSCs. However, some of those recipients exhibited long-term multilineage reconstitution (Fig. 7 K and not depicted), confirming that the multipotent differentiation capacity is not abrogated in Evi1^+/− mice. To further explore the competitive disadvantage of Evi1^+/− HSCs, we transplanted WT BM cells into unirradiated Evi1^+/+ or Evi1^+/− mice. We found no engraftment in both mice (Fig. 7 L), which indicates that the resistance to the donor HSC engraftment during steady-state hematopoiesis is maintained in Evi1^+/− mice. Collectively, these data suggest that Evi1 is dispensable for the regulation of proliferative and differentiation capacity of ST-HSCs/MPPs, but is strictly required for the maintenance of LT-HSC activity.

To investigate the mechanism behind the impaired HSC activity, we performed cell-cycle and apoptosis assays, but found no differences in the cell-cycle profile or apoptotic rates of Flk-2⁻ CD34⁻ LSK cells between Evi1^+/+ and Evi1^+/− mice (unpublished data). Collectively, in consideration of the accelerated loss of LT-HSC activity in Evi1^+/− mice, it is supposed that Evi1 heterozygosity directs LT-HSCs from self-renewal toward differentiation to generate more committed progenitors, which is uncoupled from cell-cycle progression or apoptosis.

The findings noted above led us to hypothesize Evi1 has the potential to inhibit differentiation and enhance HSC self-renewal independent of cell-cycle progression. To clarify this, we adopted a gain-of-function approach, in which WT LSK cells were transduced with Evi1, and then incubated in serum-free medium. Although forced expression of Evi1 gave no apparent growth advantage for the first 10 d of culture, Evi1-transduced LSK cells subsequently manifested a mild but significant increase in proliferation rate (Fig. 8 A). Moreover, we found a substantial increase in the frequency of the remaining LSK fraction in cultured Evi1-transduced cells compared with control cells (Fig. 8 B). In parallel, the number of colonies derived from cultured Evi1-transduced LSK cell was drastically increased (Fig. 8 C). These results suggest that Evi1 activation restricts lineage differentiation and enhances self-renewal activity of HSPCs. Collectively, our data provide compelling evidence that Evi1 regulates the developmental transition from HSPCs to more committed progenitors, suggesting a crucial role of Evi1 in controlling the balance between self-renewal and differentiation.

A recent work suggests that the longer, PR domain-containing isoform Mds1-Evi1 (ME) deficiency alone causes a reduction in the number of HSCs with a loss of long-term repopulation capacity (Zhang et al., 2011). Because both ME and Evi1 are inactivated in our Evi1 KO model (Goyama et al., 2008), we attempted to genetically dissect the relative roles of ME and Evi1 in maintaining LT-HSCs. For this purpose, we transduced Evi1 or ME into Evi1^+/− Flk-2⁻ CD34⁻ LSK cells and examined whether they could maintain stem cell phenotype after in vitro culture. Reintroduction of Evi1 led to a significant increase in the proportion that remained in the LSK fraction, similar to observations made in Evi1^+/+ cells (Fig. 8 D). However, retroviral transfer of ME was unable to normalize the frequency of the remaining LSK fraction (Fig. 8 D), indicating that Evi1 preferentially rescues Evi1^+/− LT-HSC defects. Given that ME has broader effects on the hematopoietic system than Evi1 and acts in part by maintaining HSC quiescence through up-regulation of Cdkn1c transcription (Zhang et al., 2011), Evi1 and ME may exert their functions in regulating hematopoiesis at different stages and by different mechanisms.

In this study, we show that the amount of Evi1 transcript can be indicative of an undifferentiated state with multipotent differentiation capacity within HSPCs. In both the fetal and adult hematopoietic systems, Evi1 expression can mark long-term multilineage repopulating HSCs, and enhance HSC purification with a combination of other surface markers, suggesting a specific relationship between HSC activity and Evi1 expression throughout ontogeny. This stem cell-specific expression pattern of Evi1 allows us to functionally identify self-renewing HSCs by using Evi1-IRES-GFP knock-in mice, and suggests the relevance of Evi1 in fine-tuning of stem cell properties. Indeed, we provide the genetic evidence confirming that Evi1 has a predominant effect on LT-HSCs by specifically regulating their self-renewal capacity.

The prospective isolation of HSCs is the most important step to dissect their function. The strategy commonly used for HSC isolation is purification based on the expression of a combination of cell surface markers. However, some of these parameters differ between strains of mice, change dramatically during development, and are expressed on many non-HSCs. Here, we reveal that Evi1 expression specifically correlates with functional HSCs, whereas lack of Evi1 expression exclusively identifies cells without functional HSC activity in both the fetal and adult hematopoietic systems. In particular, Evi1 expression can segregate long-term repopulating HSCs from cells without self-renewal potential even within the highly subfractionated Flk-2⁻ CD34⁻ LSK or CD48⁻ CD150⁺ LSK compartments. In addition, Evi1-IRES-GFP knock-in mice offer advantages over the conventional HSC surface markers, as the GFP⁻ and GFP⁺ subfractions of the Flk-2⁻ CD34⁻ LSK or CD48⁻ CD150⁺ LSK cells show a quite similar distribution of these markers (unpublished data). Moreover, our findings that Evi1 specifically regulates the self-renewal capacity of HSCs guarantee the potential utility of Evi1 expression as an indicator of HSC activity. Therefore, the Evi1-IRES-GFP knock-in mouse line provides a powerful approach for the functional identification of self-renewing HSCs in vivo, thus opening a new avenue for investigating HSC biology.

Although functional HSCs exclusively reside in the GFP⁺ population, a proportion of GFP⁺ cells lack HSC function. As GFP protein is quite stable and degraded more slowly than Evi1 protein (unpublished data), these observations may reflect a remnant of GFP expression from cells that have just differentiated from GFP⁺ HSCs. However, it is possible that Evi1 expression distinguishes self-renewing ST-HSCs from cells with no self-renewal activity in the ST-HSC/MPP fraction, as Evi1 heterozygosity affects the short-term repopulating capacity of CD34⁺ LSK cells.

The present findings in the hematopoietic system encourage us to examine the possibility that Evi1 expression serves as a selective marker for stem cells in other tissues or in cancer systems. We show that mesenchymal stem cells (MSCs), one of the few tissue stem cell types that have been established to self-renew in vivo (Morikawa et al., 2009), do not express Evi1. However, a mouse gene expression atlas and prior studies examining the expression pattern of Evi1 in various tissues have reported Evi1 expression in the kidney, ovary, uterus, intestine, stomach, lung, trachea, and nasal cavity in the adult mouse (Morishita et al., 1990; Perkins et al., 1991; Su et al., 2004). It will be interesting to determine, using Evi1-IRES-GFP knock-in mice, whether Evi1-expressing cells in these organs are enriched with tissue stem cells.

Our data suggest a unique association between Evi1 expression and HSC self-renewal activity throughout hematopoietic ontogeny. Along with this stem cell–specific expression pattern of Evi1, the fact that the disruption of a single allele of Evi1 leads to a near total loss of self-renewing HSCs implicates Evi1 as a central regulator in HSC self-renewal. In addition, a recent gene expression profile analysis showed that Evi1 binding sites are enriched in the upstream region of genes expressed selectively in LT-HSCs (Forsberg et al., 2010). In fact, several molecules involved in the regulation of HSC self-renewal have been identified as downstream targets or interacting proteins of Evi1, including Gata2 (Sato et al., 2008; Yuasa et al., 2005), Pbx1 (Shimabe et al., 2009), Runx1 (Senyuk et al., 2007), and TGF-β (Kurokawa et al., 1998). Together with these findings, our data strongly support a model in which Evi1 gene dosage is a critical determinant of HSC self-renewal potential.

Inappropriate expression of EVI1 confers poor prognosis in patients with AML (Lugthart et al., 2008; Gröschel et al., 2010), and therefore improvement of the therapeutic outcome of leukemia with high EVI1 expression is needed. In this study, we reveal that Evi1 overexpression blocks differentiation and induces HSPC expansion. Our data fit with other studies showing that retroviral integration at the Evi1 locus can be associated with long-term in vivo clonal dominance, occasionally leading to leukemic transformation (Stein et al., 2010). The genetic events underlying AML pathogenesis fall into two groups: (1) mutations that enhance proliferation and survival of hematopoietic progenitors, or (2) mutations that result in impaired differentiation or aberrant acquisition of self-renewal properties of HSPCs (Fröhling et al., 2005). Our data indicate that Evi1 activation can function as the latter mutation and confer enhanced self-renewal capacity in myeloid neoplasms. In addition, we demonstrate that retroviral transfer of Evi1, but not ME, can ameliorate the self-renewal defects in Evi1^+/− HSCs, highlighting a distinct role of Evi1 in HSC self-renewal. These findings may explain the underlying mechanisms of the clinical observations that, irrespective of ME expression, aberrant EVI1 expression carries an adverse prognostic value in AML (Lugthart et al., 2008, 2010). As it is becoming evident that leukemic stem cells share self-renewal machinery with normal HSCs, the elucidation of how Evi1 controls HSC self-renewal may provide biological insight into the pathogenesis of Evi1-related leukemia.

The targeting construct was assembled in the plasmid vector pBluescript KS. The 5′ arm of the targeting vector consists of a 5.1-kb fragment of BAC clone RP24-481A14 and the 3′ arm consists of a 3.1-kb fragment. The 5′ arm contains Evi1 intron 8 and exon 9, and the 3′ arm contains intron 9, exon 10, and intron 10. Both arms were obtained by PCR using BAC clone RP24-481A14 as a template, all sequenced, and then inserted into pBluescript KS. Mouse Evi1 cDNA was isolated from murine embryo cDNA by PCR, with an EcoRI site at the 5′ end and a BamHI site at the 3′ end, which was cloned into pBluescript KS. A 1.3-kb IRES-GFP cassette derived from pGCDNsam-IRES-GFP retroviral vector was inserted downstream of the aforementioned Evi1 fragment. A polyadenylation (pA) cassette was then ligated 3′ to the IRES-GFP cassette. A neomycin-positive selection (Neo^r) cassette, expressed under control of the PGK promoter, was inserted downstream of the pA cassette. Both pA cassette and loxP-flanked neomycin-positive selection cassette were derived from DT-A/AFP(EGFP)/Neo vector (a gift from the Institute of Physical and Chemical Research Center for Developmental Biology, Kobe, Japan). The partial Evi1 cDNA-IRES-GFP-pA-loxP-neo was released intact by digestion with Sse8387I and cloned into a unique Sse8387I site in exon 9 of BAC arm. A diphtheria toxin–negative selection cassette was cloned into pBluescript KS, 3′ to the targeting construct. The targeting construct was linearized by SacII and transfected into TT2 ES cells by electroporation. Homologous recombinant clones were identified by Southern blot analysis of genomic DNA isolated from individual G418/FIAU-resistant ES cell colonies. The DNA was digested with XbaI, blotted to nylon membranes, and hybridized with a 3′ external Evi1 probe. Confirmatory Southern blotting could detect a 9.1-kb WT Evi1 allele band and a 4.1-kb correctly targeted Evi1-IRES-GFP allele band with this 3′ probe. In EcoRV-digested genomic DNA from positive ES cell clones, a 5′ external probe detected 10- and 11-kb bands from the WT and targeted alleles, respectively. Next, appropriately targeted ES clones were aggregated with 8-cell stage of ICR embryo, and resultant blastocysts were injected into pseudopregnant foster ICR mothers. Chimeric males, which gave germ-line transmission, were crossed with C57BL/6 females. Blastocyst injection and breeding of chimeras were performed in the Animal Center for Biomedical Research, University of Tokyo, Tokyo, Japan.

Evi1-IRES-GFP knock-in mice were backcrossed onto a C57BL/6 background (Ly5.1 or Ly5.2) for at least four generations (Sankyo-Laboratory Service). Heterozygous Evi1 KO mice (Evi1^+/− mice) were previously described (Goyama et al., 2008). C57BL/6-Ly5.1 mice were crossed with Ly5.2 mice to obtain Ly5.1/Ly5.2 mice. Littermates were used as controls in all experiments. All animal experiments were approved by the University of Tokyo Ethics Committee for Animal Experiments and strictly adhered to the guidelines for animal experiments of the University of Tokyo.

Evi1^+/− mice were genotyped by PCR as previously described (Goyama et al., 2008). Evi1^+/GFP mice were genotyped using a multiplex PCR to detect both WT and Evi1-IRES-GFP alleles. Genomic DNA was isolated from tail biopsies and subjected to PCR using Neo and Evi1 primers. PCR with Neo primers detects the knock-in allele, and that with Evi1 primers detects the WT allele. The PCR samples were denatured at 94°C for 2 min, subjected to 30 cycles of amplification (94°C for 30 s, 65°C for 1 min, and 72°C for 1 min), and followed by a final extension step at 72°C. PCR products were resolved by agarose gel electrophoresis. PCR primers are listed below: Neo primer forward, 5′-AGGGGATCCGCTGTAAGTCT-3′, reverse, 5′-GCACTGACTGCTCATCCAAA-3′; Evi1 primer forward, 5′-ATGTCAGCAATTGAGAACATGG-3′, reverse, 5′-ATCCAAAGGTCCTGAGTTCAAA-3′.

A list of antibodies is provided in Table S1. Stained cells were sorted with a FACSAriaII, and analysis was performed on LSRII (both from BD). A mixture of antibodies recognizing CD3, CD4, CD8, B220, TER-119, Mac-1, or Gr-1 was used to identify Lin⁺ cells. Anti-CD127 antibody was added to the lineage mixture, except for the analysis of CLPs. The data analyses were performed with FlowJo software (Tree Star). In experiments with the Evi1-IRES-GFP knock-in mouse, a “fluorescence minus one” littermate control was analyzed in parallel to set GFP gates.

For Hoechst 33342 and pyronin Y staining, cells were incubated with 5 ng/ml Hoechst 33342 (Invitrogen) and 25 µg/ml verapamil at 37°C for 45 min. Next, pyronin Y (Sigma-Aldrich) was added to 1 µg/ml, and the cells were incubated for an additional 15 min.

For in vitro serum-free culture, cells were cultured in StemSpan SFEM (StemCell Technologies) supplemented with 20 ng/ml mouse SCF and 20 ng/ml human TPO, and subsequently subjected to flow cytometry or colony-forming assay at the indicated day after incubation. For colony-forming assay, cells were seeded in duplicate and cultured in cytokine-supplemented methylcellulose medium (MethoCult GF M3434; Stem Cell Technologies). Subsequently, colonies were counted and identified based on morphological examination on day 12. For in vitro differentiation, LSK GFP⁺ cells were cultured in RPMI-1640 medium (Wako) containing 10% serum with 50 ng/ml mouse SCF, 50 ng/ml human TPO, 10 ng/ml mouse IL-3, and 10 ng/ml human IL-6, and subjected to flow cytometry or colony-forming assay after 5 d of incubation.

Cells were clone-sorted into 96-well plates using FACS-based automated cell deposition unit and cultured in StemSpan SFEM supplemented with 20 ng/ml mouse SCF and 20 ng/ml human TPO. After 14 d of culture, cell numbers in each colony were analyzed.

Transplantation assays were performed using the Ly5 congenic mouse system. In CRAs, lethally irradiated (9.5 Gy) mice were reconstituted with the indicated subsets from Evi1^+/+, Evi1^+/GFP, or Evi1^+/− mice, in competition with 2 × 10⁵ unfractionated BM cells from congenic mice. For second BM transplantation, BM cells (1/2 femur equivalent) were obtained from recipient mice 16 wk after transplantation, and transplanted into a second set of lethally irradiated (9.5 Gy) mice. For reciprocal transplantation assays, lethally irradiated (9.5 Gy) or unirradiated Evi1^+/+ and Evi1^+/− mice were transplanted with 2 × 10⁵ WT BM cells without competitor cells. In non-CRAs, lethally irradiated (9.5 Gy) mice were reconstituted with Evi1^+/+ and Evi1^+/− Flk-2⁻ CD34⁻ LSK cells without competitor cells. Reconstitution of donor-derived cells was monitored by staining PB cells with antibodies against Ly5.1, Ly5.2, CD3, CD4, CD8, B220, Mac-1, and Gr-1. When CD48⁺ CD150⁻ LSK cells were transplanted, nonblocking anti-CD48 antibody (MRC OX78 clone) was used (Grassinger et al., 2010).

For CFU-S assay, 100 CD34⁺ LSK cells were injected into lethally irradiated (9.5 Gy) mice. Spleens in transplanted mice were isolated 11 d later, and visually inspected for the presence of macroscopic colonies after fixation in Tellyesniczky’s solution.

Total RNA was prepared using RNeasy Mini kit (QIAGEN), then cDNA was synthesized with a QuantiTect Reverse Transcription kit (QIAGEN), and used for RQ-PCR with FastStart SYBR Green Master and LightCycler 480 System (Roche Applied Science) according to the manufacturer’s instructions. All assays were performed in triplicate and relative expression was normalized to the internal control GAPDH. PCR primers are listed below: GAPDH primer forward, 5′-CCATCACCATCTTCCAGGAG-3′, reverse, 5′-CCTGCTTCACCACCTTCTTG-3′; Evi1 primer forward, 5′-ATCGGAAGATCTTAGATGAGTTTTG-3′, reverse, 5′-CTTCCTACATCTGGTTGACTGG-3′

Western blotting was performed as previously described (Goyama et al., 2008). In brief, mouse embryo fibroblast cells were lysed in TNE buffer, subjected to 7% SDS-PAGE, and transferred to a PVDF membrane (Millipore). The blot was incubated with an Evi-1 (C50E12) rabbit monoclonal antibody (Cell Signaling Technology), and visualized by ECL Plus (GE Healthcare).

The day of vaginal plug observation was considered as day 0.5 postcoitum (E0.5). E10.5 AGM region or E12.5 placenta were carefully dissected from embryos, dissociated by incubation with 250 U/ml dispase (Godo Shusei) for 20 min and cell dissociation buffer (Invitrogen) for 20 min at 37°C, and followed by passages through 18–25 G needles. Single cell suspensions were filtered through 70-µm cell strainer and analyzed by flow cytometry.

After BM cells were flushed out, bones were crushed with a pestle and mortar. Then bone fragments were incubated with a Collagenase/Dispase (1 mg/ml; Roche Applied Science) in MEM α (Wako) with 20% serum and gently agitated for 90 min at 37°C. The dissociated cells were collected, and bone-associated mononuclear cells were isolated with the use of density centrifugation with Histopaque-1083 (Sigma-Aldrich). ECs were defined as CD31⁺ TER-119⁻ CD45⁻, OBs were defined as CD31⁻ TER-119⁻ CD45⁻ Sca-1⁻ ALCAM⁺ cells (Nakamura et al., 2010), and MSCs were defined as CD31⁻ TER-119⁻ CD45⁻ Sca-1⁺ PDGFRα⁺ cells (Morikawa et al., 2009).

The murine Evi1 or ME cDNA were inserted into a site upstream of an IRES-EGFP cassette in the retroviral vector pGCDNsam. To produce Evi1-GFP–expressing retrovirus, Plat-E packaging cells were transiently transfected with retroviral constructs by using FuGENE 6 transfection reagent (Roche). LSK cells were purified and incubated in StemSpan SFEM medium and cytokines (100 ng/ml mouse SCF and 100 ng/ml human TPO) for 24 h. Next, cultured LSK cells were infected with Evi1-GFP retrovirus in the presence of RetroNectin (Takara Bio Inc.). The infected LSK cells were harvested 36 h after retrovirus infection, and GFP⁺ cells were sorted and subjected to in vitro culture. For retroviral transduction of Evi1 or ME into Evi1^+/+ and Evi1^+/− Flk-2⁻ CD34⁻ LSK cells, those cells were sorted and immediately infected with Evi1- or ME-GFP retroviruses in the presence of RetroNectin. These cells were incubated in RPMI-1640 medium containing 10% serum and cytokines (50 ng/ml mouse SCF, 50 ng/ml human TPO, 10 ng/ml mouse IL-3, and 10 ng/ml human IL-6). After 5 d of culture, the percentage of the remaining LSK fraction in GFP⁺ cells was analyzed by flow cytometry.

Statistical significance of differences between parameters was assessed using a two-tailed unpaired Student’s t test or Wilcoxon rank sum test.

Fig. S1 shows FACS gating strategy used to identify GFP⁺ population from adult BM of Evi1^+/GFP mice. Fig. S2 shows FACS gating strategy used to identify GFP⁺ population from E12.5 placenta or E14.5 FL of Evi1^+/GFP embryos. Table S1 lists the antibodies used for flow cytometry.

We thank T. Kitamura for Plat-E packaging cells; H. Nakauchi and M. Onodera for pGCDNsam-IRES-EGFP retroviral vector; Y. Shimamura, Y. Sawamoto, R. Takizawa, and Y. Oikawa for expert technical assistance; and Institute of Physical and Chemical Research Center for Developmental Biology for DT-A/AFP(EGFP)/Neo plasmid.

This work was supported in part by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science and by Health and Labour Sciences Research grants from the Ministry of Health, Labour and Welfare. K. Kataoka is a Research Fellow of the Japan Society for the Promotion of Science.

The authors have no conflicting financial interests.