Placental rescue reveals a sole requirement for c-Myc in embryonic erythroblast survival and hematopoietic stem cell function

During the development of the vertebrate `conceptus' (comprising all the products of conception, including the embryo, placenta and all extra-embryonic membranes), hematopoiesis initiates after gastrulation, when a subset of specialized mesodermal precursor cells become committed to the hematopoietic lineage. The establishment of the hematopoietic system occurs in two apparently independent waves: primitive yolk sac (YS) hematopoiesis followed by intra-embryonic definitive hematopoiesis. In the mouse, primitive blood formation begins at embryonic day (E) 7.0-7.5 in the YS where mesodermal clusters form morphological identifiable `blood islands', whereby cells at the periphery differentiate into early endothelial cells and the cells in the centre form the first erythroid cells(Cumano and Godin, 2007; Ferkowicz and Yoder, 2005). Primitive pro-erythroblasts, characteristically large nucleated cells, enter the circulation at E9 where they undergo further maturation. Primitive YS hematopoiesis exists only transiently in the conceptus and although it does not contribute to the hematopoietic system long term, it ensures the rapid production of differentiated blood cells (primitive erythrocytes, macrophages,megakaryocytes) for the immediate needs of the embryo before the definitive system is sufficiently developed (Cumano and Godin, 2007; Palis and Yoder, 2001).

The first definitive hematopoietic cells to emerge in the conceptus are hematopoietic stem cells (HSCs). The main site of HSC emergence occurs autonomously within the aorta-gonad-mesonephros (AGM) region of the embryo proper, where hematopoietic clusters are observed along the ventral wall of the dorsal aorta from E10.0 (Dzierzak and Speck, 2008; Medvinsky and Dzierzak, 1996). Extra-embryonic tissues, including the YS(Lux et al., 2008; Samokhvalov et al., 2007),vitelline and umbilical arteries (de Bruijn et al., 2000) and placenta(Gekas et al., 2005; Ottersbach and Dzierzak, 2005; Zeigler et al., 2006), have also been attributed with definitive hematopoietic potential. AGM derived HSCs isolated at E10.0-11.5 bear long-term repopulating HSC activity and can be identified by a combination of cell-surface markers, including the pan-hematopoietic marker CD45, AA4.1, Kit and CD34(Bertrand et al., 2005; Kumaravelu et al., 2002; Sanchez et al., 1996; Taoudi et al., 2005).

Newly generated HSCs are subsequently thought to migrate to the placenta,which has recently been identified to act as a major (although temporary)embryonic HSC niche, where nascent HSCs from the AGM or umbilical arteries mature and expand (Gekas et al.,2005; Ottersbach and Dzierzak,2005). HSCs are then thought to migrate from the placenta via the umbilical arteries to the second major hematopoietic organ, the fetal liver,where they further mature, expand and undergo differentiation along erythroid,myeloid and lymphoid lineages. Just before birth, definitive HSCs begin to migrate to the bone marrow where they populate niches located at the endosteum and from where adult hematopoiesis is controlled and maintained during adulthood (Wilson and Trumpp,2006). A number of genes have been shown to be important for primitive and/or definitive hematopoiesis. Many genes affecting primitive YS hematopoiesis encode transcription factors such as Scl, Lmo2 and Gata1, and embryos lacking primitive hematopoiesis die before midgestation (E10.5). By contrast, embryos containing mutations in genes that affect the proliferation and/or differentiation of early definitive stem/progenitor cells or more mature myeloid lineages, such as Myb, Rel, Aml/Runx1, Cbfβ and Lhx2,survive significantly longer and die between E11.5 and E16(Godin and Cumano, 2002).

The proto-oncogene c-myc plays an important role during normal adult hematopoiesis and is overexpressed in numerous hematopoietic malignancies. It encodes a nuclear basic helix-loop-helix leucine zipper protein c-Myc, which coordinates expression of a large number of diverse genes involved in processes necessary for cell expansion, cell growth, metabolism,ribosome biogenesis, proliferation, differentiation and stem cell function(Adhikary and Eilers, 2005; Hurlin and Dezfouli, 2004). Recent data also suggest an involvement of c-Myc in the control of various stem/progenitor populations, including those found in the epidermis, intestine and bone marrow (Arnold and Watt,2001; Benitah et al.,2005; Bettess et al.,2005; Muncan et al.,2006; Murphy et al.,2005; Oskarsson et al.,2006; Waikel et al.,2001; Wilson et al.,2004). In addition, c-Myc is a member of a group of four genes sufficient to convert both mouse tail tip fibroblasts and human fibroblasts into cells with pluripotent stem cell like activity, indicating that c-Myc may be a component to supply cells with self-renewal activity(Knoepfler, 2008; Okita et al., 2007; Takahashi et al., 2007; Wernig et al., 2007). Many primary and established cell lines require c-Myc to maintain proliferation,suggesting that Myc activity is indispensable for cell cycle progression(Grandori et al., 2000). This assumption appeared to hold true in vivo, as embryos in which the c-myc gene has been eliminated by gene targeting in ES cells(c-myc^ΔORF/ΔORF) fail to thrive and die before midgestation. c-Myc-deficient embryos are much smaller and are often delayed in development. They exhibit multiple defects, including an enlarged pericardial sac, failure to turn, a wavy neural tube, hypoplastic branchial arches and lack of blood cells (Davis et al., 1993; Trumpp et al.,2001). However, in this report, we show that embryos derived from epiblast-restricted deletion of c-Myc appear morphologically normal and exhibit no obvious proliferation defects, and we attribute most of the developmental defects of c-myc^ΔORF/ΔORFembryos to placental insufficiency. Most interestingly, the only epiblast-derived lineage that absolutely requires c-Myc activity through to E12 is the hematopoietic system.

c-myc^flox/flox mice as previously described (Trumpp et al.,2001) were crossed with Sox2Cre transgenic mice(Vincent and Robertson, 2003), Tie2Cre transgenic mice (Braren et al., 2006) or HNF4αCre transgenic mice(Vincent and Robertson, 2004)to generate epiblast-, vascular and hematopoietic-, and liver-specific c-myc-deficient mice, respectively. Controls are defined as Sox2Cre;c-myc^flox/+,c-myc^flox/+ or c-myc^ΔORFrec/+. No difference was observed between these embryos in any of the experiments described. For embryo analysis, timed matings were set up and the morning of the vaginal plug was considered as embryonic day 0.5 (E0.5). Analysis of the Cre-mediated deletion efficiency at the c-myc^flox locus was assessed by Southern blot analysis as previously described (Trumpp et al.,2001).

Freshly isolated embryos were fixed in 10% buffered formaldehyde overnight at 4°C, then dehydrated to 70% ethanol. After embedding in paraffin, 4μm sections were cut by using a microtome. For routine histology, sections were stained with Hematoxylin and Eosin.

Histological slides were rehydrated and treated with 3% hydrogen peroxide for 20 minutes. Antigen retrieval was performed by boiling samples in Tris buffer (pH 9) for 20 minutes (c-Myc, BrdU, HNF4α) or proteinase K(laminin, cytokeratin). Blocking was carried out in 1% BSA in PBS for 30 minutes. Slides were incubated overnight at 4°C with the following primary antibodies: c-Myc (Upstate, 06-340) 1:250 or (Santa Cruz, N262) 1:250; laminin(Dako Cytomation, Z0097) 1:600; cytokeratin (Dako Cytomation, Z0622) 1:1200;BrdU (Oxford Biotechnology, OBT 0030) 1:100; PECAM (CD31) (Pharmingen,cat:01951D) 1:100; and HNF4α (Santa Cruz) 1:100. Slides were washed and incubated for 1 hour with secondary HRP antibodies (Dako, K4000/K4002 or Amersham, LA9350W) or Alexa488 (Molecular Probes, A11034).

CFU assays were performed as previously described(Delassus and Cumano, 1996). Single cell suspensions of E10.5 control and Sox2Cre;c-myc^flox2 embryos (devoid of head and limbs) or YS were grown in methylcellulose containing a mixture of rIL3, rSCF, rGM-CSF (2ng/ml) and human rEpo (4 U/ml) (StemCell Technology,Vancouver, MethoCult M2231). Colonies were counted and analyzed after 8 days.

Embryos were collected at E10.5-11.0 and dissected (to remove head and limbs) under a stereomicroscope. Tissues were mechanically disrupted to obtain single-cell suspensions by passing them through a 26 G needle. Cells were incubated for 30 minutes on ice with the following mAb conjugates:Ter119-APC-Cy7, Kit (CD117)-PE-Cy7, CD34-PE, CD71-PE (eBioscience) and BrdU-FITC (BD Bioscience). For the TUNEL assay the `in situ cell death detection kit (fluorescein)' from Roche was used according to the manufacturer's instructions. Five- and six-color FACS analysis was performed using a FACS Canto Flow Cytometer (Becton Dickinson) and data were analyzed using FlowJo (Tree Star, USA) software. Cell sorting of CD45⁺ cells was performed on a FACS Aria Flow Cytometer (Becton Dickinson).

RNA from control and Sox2Cre;c-myc^flox2 embryos was isolated using the RNeasy mini kit (Qiagen). cDNA was generated using Stratascript Reverse Transcriptase (Stratagene) followed by gene-specific PCR with primers for c-myc, N-myc and L-myc as previously described (Wilson et al.,2004). All samples were normalized usingβ2-microglobulin.

Previous studies detailing the expression of Myc transcripts during gastrulation of the mouse conceptus revealed abundant levels of c-mycRNA in the following extra-embryonic structures: ectoplacental cone,extra-embryonic ectoderm, primary/secondary giant cells, allantois and chorion(Downs et al., 1989). In order to circumvent a putative requirement of c-Myc signaling activity in the extra-embryonic tissues of the conceptus and specifically dissect the role of c-Myc in the development of the embryo proper, we induced deletion of the conditional c-myc^flox allele(Fig. 1A) exclusively in the epiblast using a Sox2Cre mouse strain(Hayashi et al., 2002; Trumpp et al., 2001; Vincent and Robertson, 2003). Sox2Cre;c-myc^flox/floxembryos were observed in the expected Mendelian ratio up to E11.5; however, no viable mutant embryos were observed beyond E12.5 (see Table S1 in the supplementary material). Analysis of the recombination efficiency revealed virtually complete deletion of the c-myc^flox allele within Sox2Cre;c-myc^flox/floxembryo proper (Fig. 1B),whereas the extra-embryonic placental tissue comprised mostly of the unrecombined c-myc^flox allele (extra-embryonic ectoderm-derived chorionic trophoblast and spongiotrophoblast cells) and, as expected, few c-Myc-deficient epiblast-derived cells (chorionic mesoderm and allantois components) (Fig. 1B). The level of c-myc transcripts within the Sox2Cre;c-myc^flox/floxembryo proper was decreased about 600-fold, although no significant change in the expression of N-myc (also known as Mycn) and L-myc was observed (Fig. 1C).

Most noticeably, Sox2Cre;c-myc^flox/floxmutant embryos exhibit an entirely different phenotype to that of c-myc^ΔORF/ΔORF-null embryos(Fig. 1D)(Trumpp et al., 2001). Sox2Cre;c-myc^flox/floxmutants die during a very tight window between E11.25 and E11.75, and appear morphologically normal, with no apparent defects, delay in development or significant difference in overall embryo size up to E10.5(Fig. 1D, parts i,ii). The only defect evident in both the embryo proper and YS was their pale appearance,indicative of a hematopoietic/erythroid deficiency(Fig. 1D, part ii). Close to their time of death at E11.5, mutant embryos appear slightly smaller, probably due to the severe anemia (see below) and consequently a failure to thrive(Fig. 1D, part iii). Although extra-embryonic rescue of c-Myc activity prolongs survival of the conceptus for an additional 2 days, loss of c-Myc activity within the embryo proper remains fatal.

To address a putative role of c-Myc in extra-embryonic/placental tissues,IHC analysis of wild-type placentas from E9.5-E11.5 embryos was performed. This revealed c-Myc protein expression in all compartments of the chorioallantoic placenta, namely the trophoblast giant cells,spongiotrophoblast and chorionic/allantoic cells of the labyrinth layer(Fig. 2A), similar to what has previously been reported by RNA in situ hybridization(Downs et al., 1989). However,no significant c-Myc expression was detected in the deciduum, the maternal compartment of the placenta (Fig. 2A). Circulating hematopoietic cells of embryonic origin, as defined by their morphology (round cells with a high nucleus to cytoplasm ratio) and localization within the fetal blood spaces, also express high levels of c-Myc (Fig. 2A′).

Next, we examined the histology of the chorioallantoic placenta of c-myc^ΔORF/ΔORF embryos. The chorioallantoic placenta is formed by the attachment of the epiblast-derived allantois to the extra-embryonic ectoderm derived chorionic trophoblast cells (at E8.5) and the subsequent invasion and branching of the fetal vessels into a highly vascularized structure referred to as the labyrinth (E9.5-E10.5)(Watson and Cross, 2005). Histological analysis of E10.5 c-myc^ΔORF/ΔORFmutants revealed severe structural defects in the chorioallantoic placenta(Fig. 2B). Immunohistochemical(IHC) analysis further revealed a severe reduction in the number of epithelial derived (cytokeratin positive) chorionic-trophoblast cells(Fig. 2C) and a failure of the fetal blood vessels and allantoic mesenchyme (laminin positive) to invade the chorion compartment, resulting in a strongly hypoplastic labyrinth layer(Fig. 2C, right panel). Hence, c-myc^ΔORF/ΔORF mutants appear to exhibit normal chorioallantoic attachment, but a failure in the initiation of branching morphogenesis. In addition, c-Myc deficient placentas as a whole are severely hypoplastic and exhibit a dramatic loss in the number of proliferating cells, as assayed by 5-bromodeoxyuridine (BrdU) incorporation(Fig. 2D). These results therefore indicate that c-Myc activity within the extra-embryonic chorionic ectoderm or trophoblast cells is crucial for correct chorioallantoic placentation.

Importantly, Sox2Cre;c-myc^flox/floxembryos exhibit no defects in the formation of the chorioallantoic placenta(Fig. 2E). The only obvious difference between control and Sox2Cre;c-myc^flox/floxplacentas was a lack of hematopoietic cells within the placental vessels of mutants (Fig. 2E, insets). Hence, the extra-embryonic derivatives of the epiblast that contribute to placentation (allantois and chorionic mesoderm) remain functional in the absence of c-Myc activity. In conclusion, these data strongly suggest that the majority of developmental abnormalities observed in c-myc^ΔORF/ΔORF-null embryos are secondary defects owing to the failure to develop a functional placenta.

The pale appearance of Sox2Cre;c-myc^flox/floxembryos suggests a defect in the generation and/or maintenance of the embryonic hematopoietic system. Histological analysis of E11.0 Sox2Cre;c-myc^flox/floxembryos and YS revealed an almost complete absence of hematopoietic cells(Fig. 3A; data not shown). The majority of blood cells present during midgestation consist of primitive nucleated erythroblasts, generated predominately by YS derived precursors,which enter the embryo at the onset of circulation. Primitive erythroblasts express high levels of the transferrin receptor CD71 and increasing levels of the glycophorin A-associated TER119 during maturation(Fraser et al., 2007). FACS analysis of the erythroid lineage of E11.0 Sox2Cre;c-myc^flox/floxembryos revealed a 11.5-fold reduction (wild type 51,089±10,951, mutant 4430±1797) in the number of more mature CD71⁺Ter119⁺ early basophilic erythroblasts (PE, Fig. 3B,C). Even more apparent was the striking loss of erythroblast precursor cells(CD71⁺Ter119^- primitive proerythroblasts; 1500-fold decrease, wild type 141,439±7844, mutant 899±123)(Fig. 3B,C).

In order to address whether the extremely low numbers of erythroid cells present in Sox2Cre;c-myc^flox/floxembryos is due to aberrant cell death, we assayed TUNEL reactivity via FACS,which revealed a dramatic increase (19-fold) in the apoptosis rate of E9.5 Sox2Cre;c-myc^flox/floxCD71⁺ cells compared with controls (wild type 0.98±1.2%,mutant 18.65±0.07%) (Fig. 3D). This apoptotic phenotype became even more pronounced at E10.5, whereby the CD71⁺ population of Sox2Cre;c-myc^flox/floxembryos exhibit a 41-fold increase in apoptosis (wild type 0.49±0.08%,mutant 20.28±5.03%) (Fig. 3D). This strong increase in the apoptosis rate was specific to the cells of the erythroid lineage, as there was no significant overall increase in cell death within Sox2Cre;c-myc^flox/floxembryos (E9.5 wild type 6.35±1.55%, mutant 10.05±0.21%; E10.5 wild type 2.08±0.15%, mutant 4.51±0.28%). The existence of erythroblasts in Sox2Cre;c-myc^flox/floxembryos (albeit decreased numbers) suggests the initial presence of proerythroblast precursor cells. However, our data show that these cells undergo massive apoptosis and are no longer present in sufficient numbers to maintain hematopoiesis, probably leading to the embryonic death of Sox2Cre;c-myc^flox/floxembryos.

At E11.0 of gestation, the pan-hematopoietic marker CD45 detects both YS-derived primitive macrophages and hematopoietic progeny derived from definitive HSCs (Bertrand et al.,2005; Taoudi et al.,2005). We therefore examined the expression of CD45 and observed a striking reduction in the number of CD45⁺ blood cells in both the dorsal aorta and fetal liver of E11.0 Sox2Cre;c-myc^flox/floxembryos (Fig. 4A). Quantification by FACS analysis revealed a sixfold decrease in the absolute number of CD45⁺ hematopoietic cells(Fig. 4B and C). In order to determine whether the c-myc^flox allele was indeed recombined in the few remaining CD45⁺ cells, these cells were isolated from Sox2Cre;c-myc^flox/floxembryos by flow cytometry and their RNA analyzed by real-time RT-PCR. This analysis revealed a greater than 1000-fold downregulation of c-myctranscripts compared with control cells, demonstrating efficient elimination of c-myc mRNA within the small remaining CD45⁺hematopoietic population.

Early definitive HSCs in the mid-gestation embryo express the cell-surface markers AA4.1 (CD93), Kit (CD117) and CD34(Cumano and Godin, 2007; Dzierzak and Speck, 2008; Mikkola and Orkin, 2006; Sanchez et al., 1996). FACS analysis of E11.0 Sox2Cre;c-myc^flox/floxembryos revealed an increase in the percentage of c-Kit⁺AA4.1⁺ cells within the total CD45⁺population (Fig. 4D and E). However, the absolute number of CD45⁺c-kit⁺AA4.1⁺ cells per mutant embryo was similar to that of control embryos (Fig. 4F). This was further confirmed using the additional stem cell marker CD34 (see Fig. S1 in the supplementary material). By contrast, Sox2Cre;c-myc^flox/floxembryos exhibit a dramatic decrease in the number of all other definitive hematopoietic cells (CD45⁺cKit⁺AA4.1^-,CD45⁺AA4.1⁺cKit^- and CD45⁺cKit^-AA4.1^-)(Fig. 4D, Fig. 3). In summary, these data suggest that although c-Myc is not required for the generation of CD45⁺AA4.1⁺c-kit⁺CD34⁺ cells,which are highly enriched for definitive HSCs, c-Myc activity is essential to efficiently generate differentiated progeny from HSCs.

In order to address whether the phenotypic Sox2Cre;c-myc^flox/flox HSCs are functional, we determined the number of cells with colony-forming unit(CFU) activity in E10.5 mutant embryos. In control embryos, BFU-E(blast-forming unit-erythroid) and CFU-G/M (colony-forming unit-granulocyte/macrophage) were found at expected frequencies(93±17.9) (Fig. 4G,H)(Trumpp et al., 2001). By contrast, cells derived from the embryo proper of Sox2Cre;c-myc^flox/floxmutants formed no or only very few colonies (2.3±3.3)(Fig. 4G,H), with abnormal morphology (composed of extremely small cells), probably representing aberrantly differentiated cells (Fig. 4G, bottom right panel). Very similar results were also obtained in CFU assays derived from YS cells (data not shown). Thus, although the absolute number of phenotypic HSCs is maintained in the absence of c-Myc activity (Fig. 4F),c-Myc-deficient HSCs are defective and unable to generate a significant number of differentiated progeny.

Following the emergence of definitive HSCs within the AGM, hematopoietic cells are thought to migrate via the placenta (recently shown to contribute to HSC expansion) en route to the fetal liver, which serves as the major intra-embryonic hematopoietic organ to support HSC expansion and maturation. Although scarce, the presence of CD45⁺Kit⁺ cells in the fetal liver of Sox2Cre;c-myc^flox/floxmutants suggests migration of mutant stem/progenitor cells to the fetal liver is unimpaired (data not shown). This is further supported by the fact that the expression of a number of adhesion molecules overexpressed in c-Myc-deficient adult HSCs (Wilson et al.,2004) such as CD29 (β1 integrin), CD18 (β2 integrin),CD49b (α2 integrin), CD49e (α5 integrin) and CD11a (αL integrin), are not significantly changed in Sox2Cre;c-myc^flox/floxmutant stem/progenitor cells (data not shown). Consistent with the observation that c-Myc-deficient HSCs are unable to generate colonies in vitro, all CD45⁺ cells reside in the fetal liver of Sox2Cre;c-myc^flox/floxmutants exclusively as single cells, suggesting that c-Myc deficient HSCs(which constitute ∼25% of the CD45⁺ population, Fig. 4E) are unable to expand and generate progeny in the mutant fetal liver micro-environment(Fig. 4A bottom right panels). Thus, despite the fact that HSC generation is unaffected in the absence of c-Myc activity, these cells are functionally defective in vitro and in vivo. Together these results show that c-Myc is an essential factor necessary for embryonic HSC function.

In order to elucidate the molecular mechanism responsible for the observed hematopoietic defect, we determined the expression of the c-Myc target p21^CIP1 (Oskarsson et al.,2006) and critical regulators of embryonic hematopoiesis,including Scl (Shivdasani et al.,1995), Runx1(Castilla et al., 1996), Gata1 (Pevny et al.,1991), Gata2 (Tsai et al., 1994), Klf4 and Klf6(Matsumoto et al., 2006) in CD45⁺ cells by real-time RT PCR. As has been observed in other c-Myc-deficient cells (Oskarsson et al.,2006), p21^CIP1 was upregulated (2.1-fold), but among the others, only Klf4 was statistically significantly upregulated (1.6-fold)in c-Myc-deficient CD45+ cells (see Fig. S4 in the supplementary material).

In search of non-hematopoietic defects in Sox2Cre;c-myc^flox/floxembryos, a detailed analysis of E10.5-E11.5 mutant embryos was performed. For tissues difficult to distinguish by standard histological analysis alone,specific markers such as Pdx1 (pancreas) and alkaline phosphatase (primordial germ cells) were included (Kim and MacDonald, 2002; Lawson et al., 1999). Surprisingly, these studies revealed no obvious developmental abnormalities (see Fig. S2 in the supplementary material) apart from fetal liver hypoplasia (discussed in more detail below). In addition,BrdU incorporation analyses revealed no significant difference in the proliferative capacity of c-Myc-deficient organs/tissues(Fig. 5A). Similarly, no major change in the rate of apoptosis (caspase 3 and TUNEL) was observed in mutant embryos (data not shown and see above). Moreover, PECAM (CD31) staining of Sox2Cre;c-myc^flox/flox and c-myc^ΔORF/ΔORF embryos revealed the presence of an extensive vascular network, suggesting that c-Myc is not required for the establishment of the vascular system as has recently been reported(Fig. 5B; data not shown)(Baudino et al., 2002). These results indicate that c-Myc is not directly required to maintain cell proliferation and survival outside the hematopoietic system in vivo.

The results presented thus far indicate that the only epiblast-derived tissue perturbed in the absence of c-Myc activity is the hematopoietic system. In order to genetically prove this hypothesis, we generated Tie2Cre;c-myc^flox/floxembryos, whereby c-Myc is specifically deleted in the hemogenic precursor of hematopoietic and endothelial lineages(Braren et al., 2006). As expected, Tie2Cre;c-myc^flox/floxembryos are lethal at E11.5 and exhibit a hematopoietic phenotype identical to that described above for Sox2Cre;c-myc^flox/floxembryos (Fig. 6A; data not shown). FACS analysis of E11.0 Tie2Cre;c-myc^flox/floxembryos revealed a severe decrease in the total number of CD45⁺hematopoietic cells (Fig. 6B),and the typical increase in the relative percentage of CD45⁺c-Kit⁺AA4.1⁺ stem/progenitors and concomitant decrease in progenitors and mature cells(Fig. 6C). Similarly, we observed no major vascular defects in E11.0 Tie2Cre;c-myc^flox/flox YS or embryos (data not shown). The virtually identical phenotypes of Tie2Cre;c-myc^flox/flox and Sox2Cre;c-myc^flox/floxembryos provide strong genetic evidence that in the embryo proper, c-Myc is uniquely required for the development of the hematopoietic lineage.

Although the location of fetal liver niches, the cell types they are comprised of or the molecules required for its activity are not known, it is likely that hepatoblastic cells are at least in part involved in the fetal liver niche (Cumano and Godin,2007). To address whether c-Myc activity is important for the generation, maintenance or function of hepatoblasts and putative HSC micro-environment, E10.5 Sox2Cre;c-myc^flox/floxfetal livers were examined and observed to be dramatically reduced in size compared with control liver samples (a decrease which is numerically larger than the loss of hematopoietic cells alone). Histological analysis revealed severe morphological defects, whereby mutant livers exhibited a mesenchymal rather than typical epithelial organization of this organ(Fig. 7A,B). Nevertheless, Sox2Cre;c-myc^flox/floxliver cells expressed normal levels of the hepatocyte nuclear factor 4α(HNF4α), confirming their identity as hepatoblasts(Fig. 7C). To address genetically whether c-Myc is directly required for early liver development,the c-myc^flox allele was specifically deleted in fetal liver hepatoblasts from E9.0 using the Hnf4a-Cre mouse line(Vincent and Robertson, 2004). Unexpectedly, Hnf4aCre;c-myc^flox/floxembryos were viable and histological analysis of E10.5-E16.5 embryos revealed no apparent difference in liver size or morphology (Fig. S3). Moreover,detailed analysis of all major hematopoietic cell types including HSCs were present in normal numbers in hepatoblast-restricted c-Myc null embryos (data not shown). These data suggest that c-Myc is neither directly required for embryonic liver development nor for the establishment or maintenance of the hematopoietic liver microenvironment. Moreover, these genetic results confirm that the severe liver phenotype observed following epiblast-restricted deletion of c-Myc(Sox2Cre;c-myc^flox/floxembryos) is not due to loss of c-Myc within the hepatic cells, but rather represents a secondary defect due to defective hematopoiesis. This hypothesis is further supported by the observation that the fetal livers of Tie2Cre;c-myc^flox/floxembryos, the hepatoblasts of which continue to express c-Myc, show the same reduction in fetal liver size and hematopoietic defects observed in Sox2Cre;c-myc^flox/floxembryos (Fig. 7D). These data therefore show that in the absence of hematopoietic cells, hepatoblasts fail to expand and are unable to establish a normal liver morphology. To our knowledge, these results provide the first genetic evidence that hematopoietic cells are crucial for the early expansion and structural establishment of the embryonic liver.

The results of this genetic study have revealed several unexpected roles for c-Myc function during early development of the mouse embryo (see model in Fig. 7E). The first is the importance of c-Myc activity in extra-embryonic tissues, given the severe defects in chorioallantoic placentation that arise in c-myc^ΔORF/ΔORF null embryos. Moreover, we genetically demonstrate that most of the developmental defects present in c-myc^ΔORF/ΔORF-null embryos, including abnormalities of the heart, pericardium, neural tube, delay in embryo turning and embryo size (Davis et al.,1993; Trumpp et al.,2001) are virtually absent in c-Myc-deficient embryos that develop with a wild-type extra-embryonic support system(Sox2Cre;c-myc^flox/flox and Tie2Cre;c-myc^flox/floxembryos). We have therefore shown that c-Myc, initially believed to be crucial for the development of many embryonic structures, is indeed primarily important for placental development, and that this causes severe secondary defects and early lethality. Future studies detailing the spatio-temporal expression of c-Myc in the individual components/structures necessary to generate a functional chorio-vitelline and chorio-allantoic placenta should reveal the exact role of this protein in placentation. However, this initial assessment suggests that c-Myc plays an important role in the chorionic trophoblast cells. c-Myc expression is controlled by a large number of signaling pathways, including the canonical Wnt-β-catenin, Notch and Ras-MAPK pathway or signaling occurring downstream of many receptor tyrosine kinases such as PdgfR, Kit or TpoR (Barone and Courtneidge, 1995; Chanprasert et al., 2006; Grandori et al., 2000; Sansom et al., 2007; Sharma et al., 2007). Given that our study has uncovered c-Myc as a crucial component of placental development, it is likely that other knockout mouse models of genes influencing c-Myc expression may also exhibit thus far undocumented placental defects.

Second, via extensive analysis of Sox2Cre;c-myc^flox/flox and Tie2Cre;c-myc^flox/floxembryos, we have shown that c-Myc activity is dispensable for the development of most embryonic organs/tissues until E11.5. Given the large body of evidence indicating that c-Myc is crucial to maintain proliferation(Adhikary and Eilers, 2005; Grandori et al., 2000; Prathapam et al., 2006; Trumpp et al., 2001; de Alboran, 2001), it is surprising that we did not observe multiorgan hypoplasia as previously characterized in c-myc^ΔORF/ΔORF embryos(Fig. 1D)(Davis et al., 1993; Trumpp et al., 2001). In addition to other genetic studies reporting c-Myc independent proliferation in a subset of cells in adult tissues (Baena et al., 2005; Bettess et al.,2005; Oskarsson et al.,2006; Wilson et al.,2004), it therefore appears that c-Myc activity is not required to maintain at least some highly self-renewing adult tissues nor sustain the high proliferative capacity of most embryonic cell types in vivo. These observations are puzzling given the large quantity of data indicating that c-Myc function is essential for cell cycle progression and cell growth of diverse cell types in vitro (de Alboran et al., 2001; Douglas et al.,2001; Mateyak et al.,1997; Oskarsson et al.,2006; Prathapam et al.,2006; Trumpp et al.,2001). One possible explanation for the continued proliferation of c-myc-deficient cells in vivo is functional redundancy with other Myc family members (N-Myc and L-Myc). Although neither N- nor L-myctranscripts are significantly upregulated in c-Myc-deficient embryos, the phenotype of epiblast-specific deletion of both c-Myc and N-Myc is more severe than ablation of either gene alone (N.C.D., Robert N. Eisenman and A.T.,unpublished). Hence, it remains plausible that in cell types predominantly expressing c-Myc, the remaining low levels of endogenous N-Myc may be sufficient to maintain the proliferation of non-hematopoietic cell types in Sox2Cre;c-myc^flox/floxmidgestation embryos. Such functional redundancy does not appear to occur in most cultured cells, however, given that cell cycle progression of many cell types arrest upon inactivation of c-Myc alone. The only approach to address whether a completely Myc independent proliferation program exists is to eliminate all three Myc family members genetically (c-Myc, N-Myc and L-Myc) in an inducible fashion in vivo. In any case, it is important to note that elimination of c-Myc alone may not necessarily affect the proliferation and survival of cells in vivo. This conclusion has important consequences regarding the use of c-Myc inhibitors designed as anti-cancer drugs in a clinical setting as they may have less severe side effects than so far anticipated.

Previous studies have shown that c-Myc plays an essential role in the de novo formation of the vascular system(Baudino et al., 2002), and both endothelial cells and hematopoietic cells are thought to arise from a common progenitor (the hemangioblast)(Lacaud et al., 2001). These data would suggest that the hematopoietic phenotype we observe in Sox2Cre;c-myc^flox/flox embryos may arise due to defective vasculogenesis or hemangioblast development. However, in contrast to the study performed by Baudino and colleagues, we observed no major vascular defects in the YS or embryo proper of three genetically distinct c-Myc-deficient mouse models (conventional knockout, c-myc^ΔORF/ΔORF;epiblast-specific deletion, Sox2Cre;c-myc^flox/flox; and endothelial/hematopoietic specific deletion, Tie2Cre;c-myc^flox/flox). The reason for this apparent discrepancy remains unclear; however, we have found that c-myc^ΔORF/ΔORF-null embryos exhibit variability in their time of death (E9.0-E10.5) and it is likely that endothelial markers are no longer informative in dying embryos. Nevertheless,we provide strong genetic evidence that c-Myc is not directly required for the formation of the embryonic vascular system. As described in the accompanying paper by He et al. (He et al.,2008) minor vascular abnormalities are present upon c-Myc deletion; however, they arise secondarily owing to the lack of hematopoietic cells (He et al., 2008).

Our genetic studies have revealed that during early embryogenesis, only primitive and definitive hematopoietic cells are dependent on c-Myc function(Fig. 7E). At time of death, Sox2Cre;c-myc^flox/flox and Tie2Cre;c-myc^flox/flox mutant embryos present with∼10% of the normal number of erythroblasts and exhibit an almost complete lack of erythroblast precursors, owing to vast apoptosis of CD71⁺cells. Thus, we provide the first genetic evidence that, in contrast to pathologically high Myc expression, as is often found in many tumors(Evan et al., 2005; Pelengaris et al., 2002),endogenous c-Myc activity has a pro-survival role in primitive erythroid cells. Although this is a novel finding, it is not completely unexpected given c-Myc is a downstream target of a number of receptor tyrosine kinases that have known pro-survival function, including Kit, Pdgf and Tpo(Kaushansky, 2005; Levitzki, 2004; Martelli et al., 2006). Future studies will need to determine whether the here uncovered pro-survival role of endogenous c-Myc is specific for hematopoietic cells or whether this is a general feature of endogenous Myc function in other cell types too.

E11.0 Sox2Cre;c-myc^flox/floxembryos exhibit a striking decrease in the absolute number of CD45⁺hematopoietic cells. By contrast, the number of phenotypic definitive HSCs remains constant in the presence or absence of c-Myc, suggesting that the generation, specification and proliferation of HSCs is a c-Myc-independent process. Despite the generation of normal HSC numbers, c-Myc-deficient HSCs fail to give rise to a significant amount of differentiated progeny and lack CFU activity, strongly suggesting that HSCs lacking c-Myc activity are non-functional. These data are in agreement with what we observed following elimination of c-Myc in adult bone marrow HSCs(Murphy et al., 2005; Wilson et al., 2004; Wilson and Trumpp, 2006). Although our analysis of the hematopoietic regulators Scl, Runx1, Gata1 and Klf6 revealed no change in expression levels within the CD45⁺ cell compartment of Sox2Cre;c-myc^flox/floxembryos, it is important to note that the proportion of phenotypic HSCs within this population is five times higher than in control embryos. This may indicate that the expression of each transcription factor is indeed up to five times lower in mutant HSC/progenitors. Unfortunately, isolation of a significant number of highly pure HSC populations from mutant embryos have technical limitations and thus prohibited us from addressing this possibility directly.

In the embryo, HSCs emerge from the AGM and migrate in a β1 integrin-dependent manner to the fetal liver where they mature and expand(Cumano and Godin, 2007; Hirsch et al., 1996). c-Myc-deficient embryonic HSCs express normal levels of β1 integrin and CD45⁺Kit⁺ cells are found in the liver, suggesting that the migration of HSCs to the fetal liver is not c-Myc dependent. However, in contrast to normal embryonic HSCs, which upon reaching the fetal liver undergo massive expansion, c-Myc-deficient hematopoietic cells reside exclusively as single cells, further supporting the conclusion that they are unable to generate hematopoietic progeny. However, unlike in the adult, no accumulation of phenotypic HSCs is observed in c-Myc mutant embryos. This may be attributable to the lack of appropriate fetal liver stromal environment required to support hematopoietic activity, given the severely impaired development of the fetal liver in mutant embryos. Only Sox2Cre (but not Tie2Cre) induces recombination in the hepatoblast lineage, yet both mouse models exhibit hepatic disorganization and hypoplasia, thus it is likely that the liver phenotype arises as a direct result of hematopoietic insufficiency. This is further supported by the fact that restricted elimination of c-Myc in the hepatoblast lineage has no discernable effect on either fetal liver or hematopoietic development. At the onset of fetal liver development (at E9.5),only primitive hematopoietic cells are present, raising the possibility that primitive erythroblasts are required to promote fetal liver development in order to generate an appropriate microenvironment or `niche' for incoming HSCs. In conclusion, these data provide first genetic evidence that proper fetal liver development requires the presence of hematopoietic cells and suggests that hepatic and hematopoietic progenitors develop in a symbiotic relationship in which hematopoietic and hepatic cells exchange signals required to promote the development of each other.

We thank Christelle Dubey for animal husbandry and genetic screening; Dr Hannah Mikkola for helpful suggestions and providing antibodies; and Frédéric Grosjean, Joanna Roberts and Steven Merlin for FACS sorting. This work was supported by a Marie-Curie post-doctoral fellowship to C.A. and by grants to A.T. from the Swiss National Science Foundation and the Swiss Cancer League.