|Home | About | Journals | Submit | Contact Us | Français|
Stem cells (SC) exhibit a unique capacity for self-renewal in an undifferentiated state. It is unclear whether the self-renewal of pluripotent embryonic SC (ESC) and of tissue-specific adult SC such as hematopoietic SC (HSC) is controlled by common mechanisms. The deletion of transcription factor Zfx impaired the self-renewal but not the differentiation capacity of murine ESC; conversely, Zfx overexpression facilitated ESC self-renewal by opposing differentiation. Furthermore, Zfx deletion abolished the maintenance of adult HSC, but did not affect erythromyeloid progenitors or fetal HSC. Zfx-deficient ESC and HSC showed increased apoptosis and SC-specific upregulation of stress-inducible genes. Zfx directly activated common target genes in ESC and HSC, as well as ESC-specific target genes including ESC self-renewal regulators Tbx3 and Tcl1. These studies identify Zfx as a shared transcriptional regulator of ESC and HSC, suggesting a common genetic basis of self-renewal in embryonic and adult SC.
The development and function of multicellular organisms are based on the activity of stem cells (SC). During early embryonic development, pluripotent stem cells give rise to all cell types in the body, including the germ line (Boiani and Scholer, 2005). In the adult, tissue homeostasis is maintained by multipotent adult stem cells, which continuously generate cell lineages comprising a given tissue. In adult mammals, SC have been identified in many tissues and organs, including the hematopoietic system, brain, skin, and intestine (Weissman et al., 2001). Whereas the precise differentiation potential varies between different SC types, all SC share the fundamental property of self-renewal, i.e. the maintenance of their number and identity in an undifferentiated state.
Embryonic stem cells (ESC) can be derived from pluripotent cells of the blastocyst inner cell mass and propagated in vitro as continuous clonal lines (Chambers and Smith, 2004). These ESC lines are capable of unlimited growth in culture while retaining their undifferentiated state and full developmental potential. Murine ESC are typically cultured on fibroblast feeder cells with serum and leukemia inhibitory factor (LIF), a cytokine that activates the essential gp130/Stat3 signaling pathway. Combined with LIF, bone morphogenetic proteins (BMP) support unlimited ESC self-renewal in the absence of serum or feeders (Ying et al., 2003). The pluripotency of ESC is regulated by a unique network of ESC-specific transcription factors including Oct4, Nanog, Sox2, and their binding partners (Boyer et al., 2005; Chambers and Smith, 2004; Wang et al., 2006). Indeed, a defined combination of Oct4, Sox2 and other transcription factors is sufficient to derive ESC-like pluripotent cells from mouse fibroblasts (Takahashi and Yamanaka, 2006). In addition, a distinct genetic pathway regulating ESC self-renewal involves transcription factors Tbx3 and Esrrb and a signaling co-factor Tcl1 (Ivanova et al., 2006).
Among adult SC types, one of the best characterized is the blood-producing hematopoietic SC (HSC) of the adult bone marrow (BM) (Kondo et al., 2003). These cells sustain hematopoiesis throughout adult life, and are capable of clonal, long-term hematopoietic reconstitution of irradiated recipients after serial transplantation. HSC are localized in a distinct structural and functional compartment in the BM, the HSC niche, which provides critical signals supporting HSC self-renewal (Calvi et al., 2003; Zhang et al., 2003). Upon differentiation, the bona fide long term-repopulating HSC rapidly lose their self-renewal potential, producing a hierarchy of multipotent short-term HSC and/or multilineage progenitors, common lymphoid and myeloid progenitors, and lineage-committed precursors. In the adult mouse BM, all HSC activity is contained in the rare (<0.1%) lineage marker (Lin)− Sca-1+ c-Kit+ (LSK) cell subset, which can be further subdivided to define the homogeneous long-term HSC population. The self-renewal of adult HSC is regulated by several broadly expressed transcription factors including Bmi-1 (Lessard and Sauvageau, 2003; Park et al., 2003), Tel/Etv6 (Hock et al., 2004) and FoxO (Tothova et al., 2007).
Although both ESC and adult SC are capable of unlimited self-renewal, their cellular and molecular features are strikingly different. First, ESC can self-renew in defined culture conditions in vitro, whereas the self-renewal of adult SC typically depends on their in vivo niche (Li and Xie, 2005). Furthermore, ESC are highly proliferative and exhibit unusual features of cell cycle control, such as the lack of a G1 checkpoint (Burdon et al., 2002). In contrast, a cardinal feature of adult SC is their low proliferation rate (Li and Xie, 2005). Indeed, genetic manipulations that increase HSC proliferation lead to HSC exhaustion and eventual failure (Cheng et al., 2000; Yilmaz et al., 2006; Zhang et al., 2006). In view of these fundamental differences between ESC and various adult SC, a major unresolved question is whether their self-renewal is regulated by common genes and/or pathways (Eckfeldt et al., 2005; Mikkers and Frisen, 2005). Such "stemness" genes have been postulated on the basis of global expression analysis of different SC populations (Ivanova et al., 2002; Ramalho-Santos et al., 2002); however, they remain to be reconciled between different studies and verified in functional assays (Sakaguchi et al., 2006). Furthermore, other studies emphasized the lack of a common gene expression program in ESC and adult SC (D'Amour and Gage, 2003; Fortunel et al., 2003).
Zfx is a zinc finger protein of the Zfy family, whose members are highly conserved in vertebrates. Mammalian Zfx is encoded on the X chromosome and contains an acidic transcriptional activation domain (AD), a nuclear localization sequence (NLS) and a DNA binding domain (DBD) consisting of 13 C2H2-type zinc fingers (Schneider-Gadicke et al., 1989). A targeted mutation of murine Zfx resulted in partially penetrant neonatal lethality and hypofertility in surviving females (Luoh et al., 1997). However, the possibility that the targeting created a hypomorphic Zfx allele could not be ruled out. Recently, two microarray-based expression studies suggested that Zfx might be expressed at elevated levels in several SC types (Fortunel et al., 2003; Ramalho-Santos et al., 2002), prompting us to investigate the role of Zfx in SC function.
Here, we use conditional gene targeting to examine the role of murine Zfx in two prototypical SC types, ESC and adult HSC. We report that Zfx is required for the self-renewal of both SC types, whereas it is dispensable for the growth and differentiation of their progeny. These results establish Zfx as a shared transcriptional regulator of embryonic and adult SC, suggesting a common molecular basis of SC self-renewal.
To confirm the elevated expression of Zfx in SC, we measured Zfx levels in SC by quantitative PCR (qPCR). As shown in Fig. 1A, the expression of Zfx was twice as high in HSC and in undifferentiated ESC compared to their respective differentiated progeny. To test the function of Zfx in SC, we generated a conditional LoxP-flanked ("floxed") allele of murine Zfx by homologous recombination in ESC (Fig 1B and Suppl. Fig. 1). Cre-mediated recombination of the only Zfxflox allele in male Zfxflox/y cells produced Zfxnull allele, which did not encode a detectable protein product (Fig. 1C). Embryos with germ line Zfx deletion developed normally until 9.5 d.p.c. and subsequently died due to extraembryonic tissue abnormalities (data not shown). The embryos with Zfx deletion in the embryo proper died neonatally but were grossly normal at 18.5 d.p.c., with Zfxnull cells contributing extensively to all tissues (Suppl. Fig. 2). Similarly, a nearly complete deletion of Zfx in the absence of obvious abnormalities was observed in embryos and newborns with early brain- or hematopoietic-specific Zfx deletion (Suppl. Fig. 2). These data suggest that Zfx is dispensable for cell growth and differentiation of most tissues in vivo.
To analyze the growth of Zfx-deficient ESC, we transiently induced Cre recombination in Zfxflox ESC and cultured the resulting mixture of recombined (Zfxnull) and un-recombined ESC in the conditions favoring self-renewal (serum/LIF/feeders). As shown in Fig. 2A, Zfxnull ESC were rapidly out-competed by Zfxflox cells and disappeared after two passages. In contrast, Zfxnull cells persisted when ESC were allowed to differentiate in the absence of LIF/feeders. Moreover, a similar competition assay revealed normal growth of Zfx-deficient primary mouse embryonic fibroblasts (MEF), confirming a specific role of Zfx in undifferentiated ESC.
To further characterize the phenotype of Zfx-deficient ESC, we derived clonal Zfxnull ESC lines and then stably re-expressed Zfx in these cells (Fig. 2B). The resulting Zfxnull ESC remained undifferentiated, as judged by the expression of ESC marker alkaline phosphatase (Fig. 2C) and key ESC regulators Nanog and Oct4 (Suppl. Fig. 3). Nevertheless, all Zfxnull clones (n=10) exhibited abnormal colony morphology, reduced clonogenic capacity, slower growth and spontaneous apoptosis in serum/LIF (Fig. 2C,D,E). The impaired self-renewal of Zfxnull ESC was particularly evident in serum-free culture with LIF and BMP4, in which Zfxnull clones failed to expand and collapsed after 2–3 passages (Fig. 2F). During the initial growth in serum-free culture, Zfxnull cells underwent massive apoptosis as revealed by increased proportion of Annexin V-positive and hypodiploid cells (Fig. 2G and Suppl. Fig. 3). Notably, nearly all (9/10) Zfxnull ESC clones reconstituted with Zfx (Zfxnull+Zfx) regained normal morphology and vigorous growth (Fig. 2C,D,F,G). These data demonstrate a specific and reversible defect in the growth and survival of undifferentiated Zfx-deficient ESC.
In contrast to their impaired self-renewal, the differentiation capacity of Zfx-deficient ESC was intact. When injected into mice, Zfxnull ESC efficiently produced teratomas containing differentiated tissues from the three germinal layers (Fig. 3A). Despite the low efficiency of chimera formation after blastocyst injection, Zfxnull ESC contributed to the majority of adult tissues in chimeric mice, including the germ line (Fig. 3B). The lack of contribution to the thymus and BM is consistent with the essential role of Zfx in adult hematopoiesis (see below). Furthermore, Zfxnull ESC clones grew normally in short-term differentiation cultures without LIF (Fig. 3C,D) and expressed appropriate differentiation markers (Suppl. Fig. 3). Finally, Zfxnull ESC efficiently formed embryoid bodies (EB) that expressed the expected markers of hematopoietic differentiation (Keller, 2005) such as β-globin and Flk-1 (Fig. 3D,E and data not shown). These results suggest that the loss of Zfx impairs the self-renewal of undifferentiated ESC, but does not affect their developmental potential or the growth of their differentiated progeny.
In the course of these experiments, we noticed that ESC clones expressing higher levels of Zfx (Zfxnull+Zfx*, Fig. 2B) were less prone to differentiation. In short-term differentiation culture without LIF, Zfx-overexpressing ESC grew better than Zfxnull or Zfxflox ESC (Fig. 3C) and retained undifferentiated phenotype (Fig. 3D), followed by massive cell death after re-plating. Furthermore, Zfxnull+Zfx* ESC generated small EB without cavities, which lost the pluripotency marker Nanog but failed to express hematopoietic differentiation marker β-globin (Fig. 3D,E). Finally, Zfx-overexpressing ESC failed to efficiently generate teratomas or mouse chimeras (data not shown). Thus, overexpression of Zfx appears to facilitate ESC self-renewal in sub-optimal conditions and to oppose ESC differentiation, the phenotype complementary to the loss of Zfx. Altogether, these data establish Zfx as a critical and specific regulator of ESC self-renewal.
To examine the role of Zfx in adult hematopoiesis, we generated Mx1-Cre+ Zfxflox/y conditional knockout (CKO) mice, in which Cre-mediated Zfx deletion is induced by interferonogen pI·C (Kuhn et al., 1995). Two weeks after induction, total numbers of BM cells and of c-Kit+ Sca-1− myeloid progenitors were unaffected in CKO mice, whereas the HSC-containing LSK population was significantly (P<0.0001) decreased (Fig. 4A). Time course analysis revealed a progressive loss of the HSC at 2–5 wk after induction, resulting in the decreased ratio of HSC to c-Kit+ Sca-1− myeloid progenitors (Fig. 4B). The fractions of erythroid and myeloid differentiated cells were normal, while lymphocytes were initially depleted by pI·C treatment and failed to regenerate in the CKO BM (Suppl. Fig. 4). A similar loss of the LSK population was observed after Zfx deletion induced by tamoxifen-activated Cre construct (R26-CreER, Suppl. Fig. 5), ruling out potential side effects of pI·C injection.
The consequences of constitutive Zfx deletion in hematopoietic cells were analyzed using the Tie2-Cre strain, in which pan-hematopoietic Cre recombination is initiated in the early embryo (Koni et al., 2001). Zfx deletion was inefficient in most adult CKO mice, suggesting a strong selection against Zfxnull cells in adult hematopoiesis. Indeed, the rare adult CKO mice with a complete Zfx deletion (n=5) showed the same specific loss of the HSC population (Fig. 4C). In contrast, the CKO fetal livers were uniformly Zfxnull (Suppl. Fig. 2), exhibited normal cellularity and contained only a slightly reduced fraction of HSC (Fig. 4C). Furthermore, primitive hematopoiesis in the yolk sac of Zfxnull embryos was normal (Suppl. Fig. 6). Thus, the absence of Zfx abolishes the maintenance of adult BM HSC under steady-state conditions, but largely spares embryonic hematopoiesis.
Further analysis of the HSC/progenitor compartment in induced Mx1-Cre CKO mice (Fig. 4D) and in Tie2-Cre CKO mice (Suppl. Fig. 7) confirmed the loss of Flt3−LSK subset that contains the self-renewing long-term HSC. In addition, the Flt3+ LSK population containing short-term HSC and c-Kitlow Flt3+ lymphoid progenitors including IL-7Rα+ common lymphoid progenitors (CLP) were severely reduced (Fig. 4D and Suppl. Fig. 4,5,7). The HSC in induced Mx1-Cre+ CKO mice proliferated normally, did not show increased egress from the BM, and displayed normal homing to and lodging in the BM after adoptive transfer (Suppl. Fig. 8). On the other hand, Zfx-deficient HSC showed a significant (P=0.0002) increase in Annexin V-positive apoptotic cells (Fig. 4E). On the other hand, the Lin− Sca-1− c-Kit+ erythromyeloid progenitors were present in normal numbers, contained common and lineage-restricted subsets (Fig. 4D), did not show increased apoptosis (Fig. 4E) and maintained their characteristically high proliferation rate (Suppl. Fig. 8). Indeed, control and CKO BM generated similar numbers of erythroid and myeloid colonies in short-term colony-forming assays (Fig. 4F). These observations suggest that Zfx is required for the long-term survival of HSC but is dispensable in erythromyeloid progenitors.
Consistent with the defective maintenance of Zfx-deficient HSC, the BM from induced Mx1-Cre CKO mice failed to contribute to hematopoiesis in competitively reconstituted BM recipients (Fig. 5A,B). To measure the maintenance of HSC independent of their homing or engraftment, we induced the loss of Zfx in pre-established BM chimeric mice (Mikkola et al., 2003). Irradiated recipient mice (CD45.1+ CD45.2+) were transplanted with a mixture of donor BM from un-induced Mx1-Cre+ CKO or control mice (CD45.2+) and competitor wild-type BM (CD45.1+). Four weeks later, chimeric mice were injected with pI·C to induce Zfx loss in the engrafted CKO BM, and the fraction of donor-derived peripheral blood leukocytes (PBL) was subsequently monitored (Fig. 5C). In contrast to all control recipient groups, the pI·C-injected recipients of CKO BM manifested a sharp decline of donor-derived PBL (Fig. 5D,E). The endpoint analysis 6 months after induction confirmed the virtual absence of Zfx-deficient donor cells in all hematopoietic cell types, including the HSC-containing LSK subset (Fig. 5F). These results demonstrate at the functional level that Zfx is absolutely required for the long-term HSC maintenance.
To identify Zfx target genes in SC, we compared global expression profiles of control and Zfx-deficient ESC and HSC. Several genes that were affected by Zfx loss in both ESC and HSC were identified, including 5 down-regulated genes (6 probes, P =0.0075) and 20 up-regulated genes (22 probes, P=7 x 10−8) (Fig. 6A and Suppl. Table I). In the latter group, a more permissive analysis suggested several additional candidates. By qPCR (Fig. 6B and Suppl. Table II), four down-regulated genes (6720467C03Rik, AV340375, Phca and Retsat) and six up-regulated genes (Procr, Pkia, Ier3, Cryab, Gem and Ptgs2) were validated as differentially expressed, revealing a common expression signature of Zfx-deficient ESC and HSC.
The predicted regulatory regions of commonly downregulated genes contain multiple potential binding sites for Zfx (Suppl. Fig. 9). To analyze the binding of Zfx to these genomic regions in live SC, we performed chromatin immunoprecipitation (ChIP) of epitope-tagged Zfx expressed in Zfxnull ESC (Zfxnull+ZfxTAG, Suppl. Fig. 10). We found that promoter regions of 6720467C03Rik, AV340375 and Retsat were significantly enriched in ChIP with anti-tag antibodies; in contrast, no enrichment was seen in parental Zfxnull ESC that do not express tagged Zfx (Fig. 6C). Finally, the expression of all three genes was increased in Zfx-overexpressing ESC clones (Suppl. Table II). These results suggest that genes commonly downregulated in Zfxnull ESC and HSC are direct targets of Zfx.
Among the six genes coordinately up-regulated in Zfx-deficient ESC and HSC, four (Cryab, Ier3, Gem and Ptgs2) are known to be induced by stress, inflammatory stimuli and/or growth factors, with the latter three representing immediate-early genes. Therefore, their induction could be indirect and reflect the impaired function of Zfx-deficient SC. Indeed, four genes (Pkia, Cryab, Ier3 and Ptgs2) were up-regulated in Zfxnull ESC and HSC, but not in their respective differentiated progeny (Fig. 6D). Notably, three genes (Gem, Procr and Pkia) display highly HSC-specific expression pattern ((Ivanova et al., 2002) and this study). The specific up-regulation of Procr in Zfx-deficient HSC was confirmed by cell surface staining for its product CD201 (Suppl. Fig. 11). Thus, the impaired self-renewal of Zfx-deficient ESC and HSC correlates with SC-specific up-regulation of stress-inducible and/or immediate early genes.
To test whether Zfx controls genes and pathways specific only for a particular SC type, we analyzed known ESC-specific regulators of self-renewal. By protein staining (Suppl. Fig. 3), microarray analysis (Suppl. Table III) and qPCR (Fig. 7A), the components of the Nanog/Oct4/Sox2 transcriptional network were only minimally affected in Zfxnull ESC. On the other hand, we observed the downregulation of several ESC-specific genes including Dub1/2, Ceacam1 and particularly Tbx3, which encodes a transcription factor required for ESC self-renewal (Ivanova et al., 2006). Further qPCR analysis showed that another ESC-specific regulator, Tcl1, was similarly reduced in Zfxnull ESC. Short-term batch culture of ESC after Cre-mediated Zfx deletion showed a prominent reduction of Tbx3 and a slight decrease of Tcl1, but not of Nanog (Fig. 7B). Finally, Zfx-overexpressing ESC expressed higher levels of Tbx3 and Tcl1, whereas Nanog and Oct4 were unchanged (Fig. 7C).
The predicted promoter regions of Tbx3 and Tcl1 contain multiple consensus binding sites for Zfx (Suppl. Fig. 12). In the case of Tbx3, ChIP analysis localized the region of prominent Zfx binding to ~0.5 Kb upstream of the transcription start site (Fig. 7D). Indeed, the binding of Zfx to Tbx3 and Tcl1 promoters was detected only in Zfx-expressing Zfxnull+ZfxTAG ESC, but not in the parental Zfxnull ESC (Fig. 7E). Finally, Zfx binding to Tbx3 promoter was observed in undifferentiated ESC but not in ESC differentiated without LIF, consistent with the rapid loss of Tbx3 expression upon differentiation (Fig. 7F). These data suggest that Zfx directly controls the expression of self-renewal regulators Tbx3 and Tcl1 in undifferentiated ESC.
In this paper, we examined the biological role of Zfx, one of the first identified zinc finger-encoding genes (Schneider-Gadicke et al., 1989) whose properties and function remained unclear. We found that Zfx deletion impaired the growth of ESC in vitro and increased their apoptosis rate, particularly in defined serum-free conditions. Despite the growth and survival defect, Zfx-deficient ESC remained undifferentiated and gave rise to all lineages, including the germ line. Conversely, stable Zfx overexpression favored ESC self-renewal at the expense of differentiation. The requirement for Zfx in ESC self-renewal in vitro is not contradicted by the relatively normal early development of Zfxnull embryos, which survive until mid-gestation. Indeed, early embryonic development is not affected by the loss of several ESC self-renewal regulators such as Eras (Takahashi et al., 2003), Tcl1, Essrb or Tbx3 (Ivanova et al., 2006). Thus, our data suggest a critical role of Zfx in promoting ESC self-renewal, while it appears dispensable for ESC pluripotency.
In addition, we found that Zfx is not required for fetal hematopoiesis but essential in adult HSC, which failed to maintain their numbers or contribute to hematopoiesis after Zfx deletion. Zfx-deficient HSC interacted normally with their BM niche but showed an increase in apoptosis, suggesting a defect of long-term self-renewal due to impaired survival. In contrast, the survival and function of erythromyeloid progenitors were not affected, at least in the short term. Very similar phenotypes have been observed after the inducible deletion of Tel/Etv6 (Hock et al., 2004) or FoxO1/3/4 (Tothova et al., 2007), or after germ line deletion of Bmi-1 (Park et al., 2003) or Ikaros (Nichogiannopoulou et al., 1999) transcription factors. Similar to the latter three models, lymphopoiesis was severely impaired after the loss of Zfx. Thus, Zfx and possibly other HSC regulators appear to control both HSC self-renewal and lymphoid differentiation, but not erythromyeloid differentiation. Alternatively, the failure of lymphoid development might be secondary to defective HSC function, while erythromyeloid progenitors might be less dependent on HSC input (Nichogiannopoulou et al., 1999).
Although several transcriptional regulators were shown to control multiple SC types, the common regulation of ESC and adult SC has been difficult to demonstrate. For instance, Polycomb group (PcG) protein Bmi-1 promotes the self-renewal of HSC and of neural SC through the repression of Cdkn2a products p16ink4a and/or p19Arf (Lessard and Sauvageau, 2003; Molofsky et al., 2003; Park et al., 2003). On the other hand, ESC are resistant to p16ink4a and to other cyclin inhibitors (Burdon et al., 2002), and are not known to depend on Bmi-1. Conversely, PcG protein Eed contributes to the repression of differentiation-associated genes in ESC (Azuara et al., 2006; Boyer et al., 2006); however, Eed promotes proliferation and antagonizes Bmi-1 in hematopoietic cells (Lessard et al., 1999). In another case, transcription factor Sox2 controls the pluripotency of ESC and of neural stem/progenitor cells (Boyer et al., 2005; Graham et al., 2003; Ivanova et al., 2006). However, Sox2 is not expressed in other adult SC types, including HSC, and its specific role in adult neural SC remains to be established. Thus, Zfx provides a rare example of a transcription factor that is specifically required in ESC and in a canonical adult SC population such as the HSC.
Because the common transcriptional regulation of ESC and HSC has not been previously described, it is likely to be based on equally novel molecular mechanisms. Indeed, we have identified several direct target genes of Zfx in ESC and HSC, all of which are either predicted or poorly characterized. While the exact function of these direct Zfx targets is currently unknown, their coordinate regulation in ESC and HSC supports the common role of Zfx in these SC types. In addition, we identified several genes upregulated in Zfx-deficient ESC and HSC, likely as an indirect consequence of Zfx loss. Most of the coordinately induced genes are targets of stress, growth factor signaling and/or immediate-early response, suggesting that Zfx-deficient SC undergo excessive stress. This is consistent with increased apoptosis rate of Zfxnull ESC and HSC; furthermore, both immediate-early gene induction and decreased survival occur in undifferentiated ESC and HSC, but not in their differentiated progeny. Of note, HSC are uniquely sensitive to certain forms of stress such as physiological oxidative stress (Tothova et al., 2007). Altogether, our data reveal the apoptosis and concomitant upregulation of stress-inducible genes as common features of impaired self-renewal in ESC and HSC.
In addition to the common target genes, Zfx can directly activate cell type-specific targets, such as Tbx3 and Tcl1 in ESC. Both genes are rapidly downregulated upon ESC differentiation, and facilitate ESC self-renewal through a putative common pathway (Ivanova et al., 2006). Recently, Tcl1 was shown to control ESC growth but not their differentiation status (Matoba et al., 2006). Thus, Zfx may promote ESC self-renewal at least in part through the direct coordinate regulation of Tbx3 and Tcl1. Furthermore, the ESC-specific expression of Tbx3 and Tcl1 may partially explain the requirement for Zfx in ESC but not in their differentiated progeny. Although none of the known HSC regulators were affected by the loss of Zfx (Suppl. Table III and data not shown), Zfx may similarly regulate a yet unknown HSC-specific self-renewal pathway. At present, the ESC system provides a proof of principle that Zfx can control self-renewal pathways specific for a particular SC type.
In conclusion, our study has identified Zfx as a shared transcriptional regulator of two prototypical yet highly different murine SC types, the embryonic and hematopoietic SC. These results provide genetic evidence for a common molecular basis of self-renewal in pluripotent embryonic SC and adult tissue-specific SC. The identification of Zfx should facilitate the molecular dissection of self-renewal mechanisms in ESC, HSC and possibly in other SC types, including tumor-initiating cancer SC. Finally, the high evolutionary conservation of Zfx raises the possibility that it may regulate SC functions in other vertebrate species, including humans.
Conditional targeting, deletion and re-expression of Zfx in ESC were performed in male 129/SvEv ESC as described in Suppl. Fig. 1 and Suppl. Methods. The presence of Zfx protein was analyzed with an affinity-purified polyclonal antibody against the N-terminal domain. Zfxflox and Zfxnull ESC clones were injected into C57BL/6 blastocysts, and the resulting chimeric males were crossed to wild-type 129/SvEv and Black Swiss females, respectively.
To obtain CKO mice (Zfxflox/y, Cre+), Zfxflox/flox females (pure 129/SvEv background) were crossed with males carrying Mx1-Cre (129/SvEv) or Tie2-Cre (mixed 129/B6) transgenes. Control mice in both strains included Cre+ wild-type and Cre−Zfxflox/y animals, which were indistinguishable by all assays. Mice carrying a tamoxifen-inducible Cre recombinase inserted into Rosa26 locus (R26-CreER, C57BL/6) were kindly provided by Dr. Thomas Ludwig, Columbia University. The recombination in adult Mx1-Cre mice was induced with polyinosinic:polycytidylic acid (pI·C, Amersham Biosciences) administered i.p. three times (0.25 mg every third day). The deletion status of all analyzed CKO cells and tissues was monitored by genomic PCR. All animal studies were performed according to the investigator's protocol approved by the Institutional Animal Care and Use Committee.
Undifferentiated ESC were cultured in complete DMEM containing 15% FCS and recombinant LIF (Chemicon) on mitomycin-treated embryonic fibroblast feeders. For gene expression studies and ChIP, ESC were grown on gelatin-coated plates in 5–15% Knockout SR serum replacement (Invitrogen) and LIF. Serum- and feeder-free culture in the presence of LIF and 10 ng/ml recombinant BMP4 (R&D Systems) was performed as described (Ying et al., 2003). To measure cell growth or colony formation, ESC were seeded into triplicate wells of 24-well plates and cultured as indicated. Apoptosis of ESC was measured by gel fractionation of low molecular weight DNA or by staining live cells with Annexin V-FITC conjugate (BD Pharmingen) and 7-amino-actinomycin D (7-AAD). For competitive growth assay, Zfxflox-neo ESC were transiently transfected with Cre expression vector and grown as indicated. For the batch culture of Zfxnull ESC shortly after Zfx deletion, Zfxflox-neo ESC were transiently transfected with Cre-GFP expression vector, and GFP+ cells were sorted 24 hrs later and cultured for 3 days with LIF on gelatin. Control ESC were generated in parallel by transfection with an empty expression vector.
For differentiation, ESC were grown in 15% FCS without LIF/feeders on gelatin-coated plates. Where indicated, ESC were cultured in suspension for 4 days to form primitive embryoid bodies (EB), and then returned to adherent culture or allowed to form mature EB in suspension for additional 4–9 days. For teratoma formation, undifferentiated ESC grown on gelatin were injected s.c. into syngeneic 129/SvEv mice, and teratomas were harvested 3 weeks later, sectioned and stained with hematoxylin-eosin.
Mouse embryonic fibroblasts (MEF) were derived from 14.5 d.p.c. R26-CreER+, Zfxflox/y embryos. After the first passage, half of the culture was incubated with 4-hydroxytamoxifen to induce Zfx deletion, and then equal numbers of treated and untreated cells were mixed and propagated by the 3T3 protocol.
Single-cell suspensions from hematopoietic organs were stained with the indicated fluorochrome-conjugated antibodies (BD Pharmingen and eBiosciences) for 4-or 5-color analysis. The lineage cocktail included antibodies to CD11b, Gr-1, Ter119, B220, CD3 and DX5. For apoptosis detection, enriched lineage-negative BM cells were stained with antibody conjugates, Annexin V TACS reagent (Trevigen) and 7-AAD. The samples were acquired using FACSCalibur or LSR II flow cytometers or sorted on FACSAria flow sorter (BD Immunocytometry Systems), and analyzed using FlowJo software (Treestar Inc.). Statistical significance was estimated using unpaired, two-tailed Student's t-test. For competitive reconstitution, BM cells from control (Mx1-Cre+) or CKO (Zfxflox/y, Mx1-Cre+) mice 7 days after pI·C injection was mixed 2:1 with the competitor BM cells from B6.SJL mice (C57BL/6 congenic for CD45.1), and 2x106 total cells were injected i.v. into lethally irradiated B6.SJL recipients. For Zfx deletion in pre-established BM chimeras, BM cells from untreated Mx1-Cre+ control or CKO mice were mixed 2:1 with competitor B6.SJL BM and injected i.v. into lethally irradiated (129/SvEv x B6.SJL)F1 recipients. Four weeks later, one half of each recipient group was treated with pI·C. For progenitor assays, total BM was plated in triplicates (105 cells/35 mm dish) into cytokine-supplemented methylcellulose medium (Methocult, Stem Cell Technologies), and the number and morphology of colonies were scored 10 days later.
For ESC, Zfxnull and control Zfxflox clones were grown on gelatin in medium with 15% Knockout SR and LIF. For HSC, BM cells were pooled from 4–6 Mx1-Cre+ Zfxflox/y CKO and control Zfxflox/y Cre− littermates 4 days after the last pI·C injection, and Lin−Sca-1+ c-Kit+ cells were flow-sorted. RNA was amplified, labeled and hybridized to Mouse Genome 430 2.0 microarrays containing probes for ~37,000 genes (Affymetrix). Details of the sample preparation and analysis are described in Suppl. Methods. Microarray data have been deposited in the GEO database under accession number GSE7069.
Total cell RNA was reverse-transcribed and assayed by quantitative real-time PCR using SYBR Green incorporation. The expression of all genes was normalized to that of β-actin and expressed relative to the indicated reference sample (average ± S.D. of triplicate reactions). For competition experiments (Fig. 2A), Zfxflox and Zfxnull alleles were amplified separately and normalized using qPCR reaction that detects all targeted Zfx alleles but not the feeder cell-derived wild-type allele. The measurements of Zfx expression (Fig. 1A) were performed using the calibration curve method. All other cDNA and genomic DNA measurements were performed using the ΔΔCT method on MX3000P instrument (Stratagene).
An ESC clone expressing FLAG-HA epitope-tagged Zfx protein (Zfx-TAG) was generated as described in Suppl. Methods. Zfx-TAG cells or parental Zfxnull cells were crosslinked with formaldehyde and subjected to ChIP using anti-FLAG M2 antibody (Sigma) or control IgG as described (Boyer et al., 2005). After crosslink reversal, precipitated chromatin was analyzed by qPCR for genomic fragments of interest. For experiments shown in Fig. 7E, ChIP was performed using anti-Zfx antibody or control IgG, followed by whole genome ligation-mediated PCR amplification as described (Boyer et al., 2005). All results were normalized using a control genomic DNA fragment from murine Snai3 gene, which was not enriched in Zfx ChIP.
We thank Philip Leder for his invaluable support of these studies, David C. Page and Mary Goodheart for encouragement and helpful discussions, T. Palomero and A. Ferrando for advice on ChIP, and T. Ludwig, C. Troy, H. Wichterle, C. Schindler, C. Henderson and H. Gu for reagents and access to instrumentation. Supported in part by the NIH grant HL084353, American Cancer Society grant RSG-06-149-01 and the March of Dimes Foundation Basil O'Connor Starting Scholar Award 5-FY04-23 to B.R., and by the NRSA awards AI007525 and AI066459 to T.L.A.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.