Forced Expression of the Homeobox-Containing Gene Pem

Developmental Biology 210, 481–496 (1999) Article ID dbio.1999.9279, available online at http://www.idealibrary.com on

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

Forced Expression of the Homeobox-Containing provided by Elsevier - Publisher Connector Gene Pem Blocks Differentiation of Embryonic Stem Cells

Yong Fan,* Mona F. Melhem,†,‡ and J. Richard Chaillet*,1 *Department of Biological Sciences, University of Pittsburgh, Pittsburgh, Pennsylvania 15260; †Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania 15261; and ‡Veterans Administration Medical Center, Pittsburgh, Pennsylvania 15240

Similarities in the differentiation of mouse embryos and ES cell embryoid bodies suggest that aspects of early mammalian embryogenesis can be studied in ES cell embryoid bodies. In an effort to understand the regulation of cellular differentiation during early mouse embryogenesis, we altered the expression of the Pem homeobox-containing gene in ES cells. Pem is normally expressed in the preimplantation embryo and expressed in a lineage-restricted fashion following implantation, suggesting a role for Pem in regulating cellular differentiation in the early embryo. Here, we show that the forced expression of Pem from the mouse Pgk-1 promoter in ES cells blocks the in vitro and in vivo differentiation of the cells. In particular, embryoid bodies produced from these Pgk-Pem ES cells do not differentiate into primitive endoderm or embryonic ectoderm, which are prominent features of early embryoid bodies from normal ES cells. This Pgk-Pem phenotype is also different from the null phenotype, as embryoid bodies derived from ES cells in which endogenous Pem gene expression has been blocked show a pattern of differentiation similar to that of normal ES cells. When the Pgk-Pem ES cells were introduced into subcutaneous sites of nude mice, only undifferentiated EC-like cells were found in the teratomas derived from the injected cells. The Pem-dependent block of ES cell differentiation appears to be cell autonomous; Pgk-Pem ES cells did not differentiate when mixed with normal, differentiating ES cells. A block to ES cell differentiation, resulting from the forced expression of Pem, can also be produced by the forced expression of the nonhomeodomain region of Pem. These studies are consistent with a role for Pem in regulating the transition between undifferentiated and differentiated cells of the early mouse embryo. © 1999 Academic Press Key Words: ES cell; Pem; homeobox; differentiation; placenta.

INTRODUCTION ing the placenta and yolk sac (Hogan et al., 1994). These early cellular transitions between pluripotent cells of the embryo Cellular differentiation during preimplantation and early and the differentiated cells of extraembryonic lineages are postimplantation development results in speciﬁc cell lineages critical events in mammalian development. Little is known, (Pedersen, 1986). During the later stages of preimplantation however, about the molecular and genetic mechanisms un- development, the fate of the outer cells of the mouse embryo derlying such lineage-speciﬁcation events in mammals is restricted to trophectoderm (TE), while the inner cells retain (Latimer and Pedersen, 1993). One example of a probable their pluripotency as inner cell mass (ICM) cells. Following molecular determinant of an extraembryonic tissue type is implantation, ICM cells differentiate into primitive (extraem- Hand1, a basic-helix-loop-helix-containing protein implicated bryonic) endoderm and embryonic ectoderm. Derivatives of in trophoblast and heart development (Cross et al., 1995; primitive endoderm, together with cells from trophectoder- Cserjesi et al., 1995; Hollenberg et al., 1995; Srivastava et al., mal and mesodermal lineages, eventually collaborate in form- 1995). Overexpression of mouse Hand1 in trophoblast stem cells produces trophoblast giant cells, while microinjection of 1 To whom correspondence should be addressed. Fax: (412) 624- Hand1 into blastomeres of preimplantation mouse embryos 4759. E-mail: [email protected]. directs them to a trophoblast giant cell fate (Cross et al., 1995).

The early differentiation events found in the mouse 1997). Taken together, these observations suggest that the embryo are also seen in cultures of embryonic stem (ES) transition from pluripotent stem cells to specific differen- cells. ES cells grown in suspension culture aggregate and tiated lineages during early mouse embryogenesis requires a form embryoid bodies (Martin, 1981). Under extended cul- sophisticated genetic network comprising genes expressed ture conditions, embryoid bodies undergo spontaneous dif- in the preimplantation embryo and expressed in a restricted ferentiation, mimicking the sequential differentiation pattern following implantation. events of the early mouse embryo (Martin et al., 1977). Murine Pem is an X-linked homeobox-containing gene. Embryonal carcinoma (EC) cells also show similar patterns The structure of its homeodomain is distinct from the of differentiation when grown as embryoid bodies (Martin structures of all major classes of homeoproteins, although it and Evans, 1975a). However, unlike the pluripotent ES cell is most closely related to the paired-like class of homeopro- lines, different EC cell lines display widely varying, and teins (Sasaki et al., 1991; Lin et al., 1994; Maiti et al., 1996). frequently restricted, patterns of cellular differentiation. A The Pem protein is detected in the late morula stage of few EC lines are similar to ES cells, exhibiting an ability to preimplantation development in both TE and ICM cells. differentiate into many different embryonic lineages (Mar- After implantation, Pem is highly expressed in ectoplacen- tin, 1980). Most EC cell lines, however, exhibit a restricted tal cone, in parietal endoderm, and in visceral endoderm, pattern of differentiation, ranging from differentiation into whereas there is no detectable Pem expression in primitive only primitive endoderm in F9 cells to differentiation into ectoderm derivatives (Lin et al., 1994). Pem mRNA is also specific embryonic cell types in C17-S1 cells (Jakob and present in pluripotent embryonic stem cells, primordial Nicolas, 1987). Interestingly, a few nullipotent EC lines germ cells, teratocarcinoma cell lines, and a variety of cannot differentiate spontaneously and differentiate only transformed cell lines (Wilkinson et al., 1990; Sasaki et al., following exposure to certain chemical inducing agents 1991; Pitman et al., 1998). Such patterns of expression (Bernstine et al., 1973; Martin, 1980). Notably, these nulli- strongly suggest that Pem might play important roles in the potent EC cells retain many of the molecular features of determination of early somatic and germ cell lineages. undifferentiated ES cells, including a high alkaline phos- Recently, two other homeobox-containing genes, Esx1/ phatase activity and expression of the cell surface SSEA-1 Spx1 and Psx, whose expression patterns are also largely antigen (Bernstine et al., 1973; Solter and Knowles, 1982). restricted to extraembryonic lineages, have been isolated The cause of such restriction on the differentiation poten- (Branford et al., 1997; Li et al., 1997; Han et al., 1998). tial of EC cell lines is unknown. Interestingly, the homeodomains of Esx1/Spx1 and Psx can A shared feature of early embryonic development and the also be classified as paired-like, sharing important struc- differentiation of ES and EC cells into embryoid bodies is tural features with Pem. These findings suggest that the the irreversible transition between undifferentiated stem homeobox-containing gene Pem may play a role in early cells and differentiated cells. Once differentiated, cells are lineage specification and formation of extraembryonic tis- committed to a particular cell lineage and do not reappear sues. as undifferentiated embryonic stem cells. Although little is In an effort to determine the potential function of Pem in known about the genetic control of such transitions from mouse development, Pitman et al. (1998) mutated the Pem undifferentiated, pluripotent cells to differentiated cells, gene in mouse ES cells and generated homozygous Pem the early embryonic expression patterns of a number of knockout mice. However, these studies provided little genes suggest their roles in these critical differentiation insight into the function of Pem, as no abnormal phenotype events (Latimer and Pedersen, 1993). These genes, which was apparent in such Pem-deficient animals. In this report, include known transcription factors, are expressed before we explore the possibility that the homeobox-containing and after the transition from pluripotent cells to specific gene Pem is involved in the determination of early extraem- cell lineages (Rogers et al., 1991; Guillemot et al., 1994; Lin bryonic lineages by studying the effect of forced Pem et al., 1994; Palmieri et al., 1994; Yeom et al., 1996; Rossant expression on the differentiation of ES cells. When Pem is et al., 1998). Oct3/4 is expressed throughout the early constitutively expressed from a murine Pgk-1 promoter, the preimplantation embryo, but later becomes restricted to transfected Pgk-Pem ES cells are morphologically identical ICM cells and epiblast derivatives (Palmieri et al., 1994). It to undifferentiated wild-type ES cells grown in the presence is not expressed in extraembryonic lineages. This expres- of leukemia inhibitory factor (LIF). However, when Pgk- sion pattern of Oct3/4 is consistent with a dual role in Pem ES cells are grown as embryoid bodies in the absence of maintaining the pluripotency of ICM cells (Nichols et al., LIF or grown as subcutaneous teratomas in vivo, they do 1998) and in directing their early embryonic differentiation. not differentiate. In contrast, PemϪ/Y ES cells, in which In contrast, the basic helix-loop-helix transcription regula- endogenous Pem expression has been blocked, show a tor Mash-2 is also expressed in all cells of the preimplanta- nearly normal pattern of embryoid body differentiation. tion embryo, but is expressed in a trophoblast-restricted Such an inhibitory effect on ES cell differentiation by the pattern after implantation (Rossant et al., 1998). Studies on forced expression of Pem appears to be an autonomous mice homozygous for a null allele of Mash-2 suggest that it cellular process, as Pgk-Pem ES cells cannot be induced to is indispensable for the formation of the diploid spongiotro- differentiate when cultured with normal ES cells. More- phoblast lineage (Guillemot et al., 1994; Tanaka et al., over, the forced expression of the N-terminal, nonhomeodo-

main portion alone inhibits differentiation, whereas expres- cassette was released from the PGEM4/Pgk-1/HA vector by SalI sion of the homeodomain itself has no effect on embryoid and EcoRI and was then cloned into pBK-DCMV to create the body differentiation. These findings are consistent with a expression vector PBK-DCMV/Pgk-1/HA. All Pem cDNAs were role of the Pem gene product in regulating the early transi- PCR amplified by Pfu DNA polymerase (Stratagene), cloned, and tion of undifferentiated pluripotent stem cells to specific sequenced. Oligomer pairs used in PCR for full-length or a portion of Pem cDNA are listed below (all contain EcoRI sites): Oligo 1, (5Ј) extraembryonic lineages. GAATTCGAAGCTGAGGGTTCCA (3Ј); Oligo 2, (5Ј) TGAAT- TCTTGAAAAGTAAGGGC (3Ј); Oligo 3, (5Ј) GAATTCTCATC- CATTCTCCTGCG (3Ј); Oligo 4, (5Ј) GAATTCGGTAGGCAGAT- MATERIALS AND METHODS GCCCC (3Ј). For full-length Pem cDNA Oligo 1 and Oligo 2 were used. For the N-terminal portion of Pem cDNA Oligo 1 and Oligo RNA Analysis 3 were used. For the C-terminal portion of Pem cDNA Oligo 2 and Total RNA from ES cells or embryoid bodies was isolated using Oligo 4 were used. All the cDNA fragments were cloned into the Ultraspec RNA isolation kit (Biotecx, Houston, TX), following pBK-DCMV/Pgk-1/HA as EcoRI fragments. In some cases, an SV40 the manufacturer’s instructions. Antisense RNA probe A (Fig. 1) for NLS was added to an expression plasmid and inserted in frame at Ј Ј measuring Pem mRNA levels protects nt 1–244. Ribonuclease the 5 EcoRI site of pBK-DCMV/Pgk-1/HA, immediately 3 of HA protection probes and Northern blot probes for Oct 3/4, ␨-globin, tag sequences. H-19, ribosomal protein L32 (RPL32), ␣-fetoprotein, and Laminin To assemble the Pem knockout construct, the BamHI fragment B1 have been described (Shen and Leder, 1992). containing the Pem coding exons was linearized at the EcoRV site near the 5Ј end and treated with exonuclease BAL-31 to delete exons 3 and 4. This fragment (ϳ5 kb) was blunt-ended and a ClaI Cell Culture linker ligated to the end. The resulting fragment was subcloned into a pBluescript KS(ϩ) vector (Stratagene) at ClaI and BamHI sites D3 ES cells were maintained in Dulbecco’s modified Eagle’s to produce the plasmid pKS/CB5. To subclone the 5Ј homologous medium supplemented with 15% fetal bovine serum, 2 mM region into the targeting construct, a 3-kb SalI/XbaI Pem genomic glutamate, 2 mM sodium pyruvate, and 1000 Units/ml LIF on fragment containing exons 1–4 was linearized at the unique gelatin-coated tissue culture plates. For ES clones stably trans- internal EcoRV site, and a ClaI linker was ligated to the end. The fected with Pem expression plasmid, 200 ␮g/ml G418 was added to SalI/ClaI 1.8-kb fragment was subcloned into pKS/CB5 to get the ES cell medium. To grow embryoid bodies in suspension pKS/CB5/SC1.8. An LTNL cassette was inserted into pKS/CB5/ culture, 106 ES cells were seeded to 10-cm bacterial culture plate SC1.8 at the ClaI site to produce the final targeting construct. containing 10 ml of ES medium without LIF. Half of the medium Pem expression and knockout constructs were introduced into was changed every other day during the period of suspension D3 ES cells via electroporation with a Bio-Rad Electron Pulsar. The culture. Embryoid bodies were harvested every 4 days up to day 12. Pgk-Pem expression plasmid was linearized at the SalI site, while For the cell proliferation assay, 105 D3 or 105 Pgk-Pem ES cells were all cDNA expression constructs were linearized at the MluI site, seeded into gelatin-treated six-well tissue culture dishes in LIF- located at the 3Ј end of the SV40 polyadenylation site. Twenty-five containing ES medium. Cells were trypsinized and counted every micrograms of linearized expression plasmid and 2.5 ␮g of pPNT 24 h. plasmid, which provides a neor resistant gene (Tybulewicz et al., 1991), were cotransfected in each electroporation reaction. After 2 Construction and Transfections of Pem Plasmids weeks of G418 selection (250 ␮g/ml), single colonies were picked, expanded, and analyzed by Southern blot analysis. The Pem expression plasmid containing a genomic portion of the Pem gene was constructed from a BamHI genomic fragment containing all the coding exons of Pem (exons 3–6), as well as all 3Ј Histological Analysis noncoding sequences. The mouse phosphoglycerate kinase (Pgk-1) promoter was cloned 5Ј to the Pem coding exons in the 6-kb BamHI Embryoid bodies were collected, washed three times with ice- ϫ fragment as an EcoRI/PstI fragment. This expression plasmid is cold 1 PBS, and fixed for 24 h in 2.5% glutaraldehyde at 4°C. After ϫ designated Pgk-Pem. fixation, the embryoid bodies were washed twice with 1 PBS, Expression plasmids containing full-length or partial-length Pem treated with 1% tetraoxide osmium solution, and embedded in ␮ cDNAs were also constructed. The expression plasmid pBK-CMV plastic resins. Sections of 1 m were made and stained with 10% (Stratagene, La Jolla, CA) was partially digested with NsiIat toluidine blue from Tousimis Research Corporation. For cell- position 1900 and totally digested by PstI to remove the CMV mixing experiments, equal numbers of ROSA26 and Pgk-Pem ES ϫ 5 promoter and the LacZ prokaryotic expression promoter. This cells (5 10 each) were mixed and grown as aggregates in a plasmid (designated pBK-DCMV) was then used to generate a series bacterial culture plate. Embryoid bodies were collected, fixed in 3% of expression plasmids. To facilitate the detection of Pem protein paraformaldehyde for 10 min, and permeabilized in 1% Triton expressed from these expression plasmids, we amplified an HA X-100 for 15 min. Embryoid bodies were stained in X-Gal solution triple tag by PCR using the 5Ј oligomer AGGTATGGTTTAC- containing 50 mM Fe3[CN]4 and 50 mM Fe2[CN]3 overnight at 37°C CCATACGAT, and the 3Ј oligomer GAATTCCTGAGCAGCG- and were further fixed in 2.5% glutaraldehyde at 4°C before TAATCT, and cloned the amplified DNA fragment into the SmaI sectioning. site of pGEM4 (Promega, Madison, WI). The cloned HA fragment Ј Ј was digested at the KpnI and EcoRI sites (within the 5 and 3 Teratoma Analysis oligomer sequences, respectively) and cloned into pGEM4 to create pGEM4/HA. The Pgk-1 promoter was cloned into pGEM4/HA as a Pgk-Pem ES cells and D3 ES cells were harvested at 106 cells/ml SalI/SmaI fragment to create pGEM4/Pgk-1/HA. The Pgk-1/HA and 0.5 ml was injected subcutaneously into the right flank of NIH

Nu/Nu nude mice. After 4–6 weeks, teratomas were harvested and Pem cells, we performed ribonuclease protection assays on histological sections were prepared as described previously. For RNA isolated from D3 ES cells and from Pgk-Pem cells. A cell-mixing experiments, equal numbers of Pgk-Pem and ROSA26 ribonuclease protection probe comprising Pem exons 1, 2, 6 ES cells (Friedrich and Soriano, 1991) were mixed (10 total cells/ and 3 and a portion of exon 4 protects 244 nt of the ml) and 0.5 ml was injected into a nude mouse. To facilitate the endogenous transcript and 141 nt of the Pgk-Pem transcript X-Gal staining, teratomas harvested after 6 weeks were chopped (Fig. 1A, probe A). The larger protected fragment derived into small cubes of approximately 2 mm/side, before the staining procedure described above was followed. from the endogenous Pem transcript is due to additional protection of the probe by the first two untranslated exons, which are not present in the Pgk-Pem construct. As shown Immunofluorescence Assay in Fig. 1D, both the wild-type D3 ES cells and a Pgk-Pem ES cells stably transfected with Pem cDNA expression plasmid clone have similar endogenous Pem expression levels were seeded onto slides pretreated with 0.1% polylysine, washed (244-nt fragment), while there is a much greater amount of three times with 1ϫ PBS, and fixed in 3% paraformaldehyde for 5 Pgk-Pem mRNA expression (the lower 141-nt band). From min, followed by 0.1% Triton X-100 treatment for 90 s. After this analysis, it appears that the 1.1-kb band is a mixture of ϫ permeabilization, cells were blocked in blocking solution (1 PBS an endogenous Pem transcript and a similar-sized transcript containing 3% BSA, 1 mM glycine) for 30 min at room temperature from the Pgk-Pem expression construct. We conclude from and incubated in primary anti-HA monoclonal antibody (Boehr- inger Mannheim) (1:250 dilution) at 4°C overnight. Goat anti- this that the level of endogenous Pem expression is unaf- mouse IgG–FITC from Sigma (1:500 dilution) was used as second- fected by the presence of the Pgk-Pem expression construct. ary antibody. When cultured in ES cell medium supplemented with LIF, Pgk-Pem ES cells had the same colony morphology as normal, undifferentiated D3 ES cells (Figs. 2A and 2B). Both RESULTS cell types formed tightly compacted round colonies, with no evidence of cellular differentiation. Moreover, the Pgk- Isolation and Characterization of Pgk-Pem ES Cells Pem ES cells exhibited a cellular proliferation rate nearly To address the role of Pem in mouse development, we identical to that of normal D3 ES cells, when cultured examined the effect of the forced, enhanced expression of under identical growth conditions in the presence of LIF the Pem gene in murine ES cells. Pem expression was (Fig. 2C). These results suggest that the forced expression of increased in ES cells by introducing an expression plasmid Pem does not change the overall character of ES cells grown that contains the Pem gene under the regulatory control of on tissue culture plates in the presence of LIF. the mouse Pgk-1 gene promoter (Adra et al., 1987; McBur- ney et al., 1991). The Pem gene in the expression plasmid is Forced Expression of Pem Inhibits ES Cell a 6-kb BamHI fragment containing the terminal four exons Differentiation in Vitro (exon 3–6). The translation initiation AUG codon is located at the 5Ј end of exon 3 and the translation termination As indicated above, the forced expression of Pem in codon in exon 6 (Fig. 1A). Pgk-Pem ES cells has no apparent effect on undifferentiated Individual ES clones stably transfected with the Pgk-Pem ES cells. To study the effect of Pem overexpression on ES expression plasmid (Pgk-Pem ES cells) were isolated and the cell differentiation, we grew Pgk-Pem ES cells in suspen- integration of the expression construct was ascertained by sion culture as embryoid bodies. After 4 days in culture, D3 Southern blotting. Using a Pgk-1 promoter fragment as a ES cells developed an outer layer of endoderm, which hybridization probe, the analyzed clones (Nos. 8, 14, and 19) surrounded a central core of undifferentiated cells (Fig. 3A). showed a strongly hybridizing 3-kb EcoRI fragment, repre- The inner core of cells is analogous to the embryo’s ICM senting multiple copies of the integrated Pgk-Pem expres- cells. The early development of a fluid-filled cyst is also sion construct. The endogenous Pgk-1 locus was identified apparent. With further growth and cellular differentiation as a 6-kb fragment (Fig. 1B). These clones were expanded of a normal embryoid body, a large central cavity forms, and Pem expression was analyzed on Northern blots (Fig. lined by columnar ectoderm. This internal ectoderm layer 1C). The endogenous Pem transcript in D3 ES cells was is separated from the outer endoderm layer by a thin evident as a 1.1-kb hybridization band. In Pgk-Pem clones 8, basement membrane. Some embryoid bodies can develop 14, and 19, Pem transcripts of 1.1 and 1.6 kb were seen. The large cystic structures, resembling visceral yolk sac (Martin 1.6-kb transcription product is likely to be from the Pgk- et al., 1977). An example of this morphology is evident in Pem expression construct, as it is not present in the D3 ES D3 ES cell embryoid bodies grown for 8 days (Fig. 3B). Cells cells, and a transcript of 1.6 kb is predicted from the from all three primary cell lineages, resembling ectoderm, structure of the Pgk-Pem expression construct (Fig. 1A). endoderm, and mesoderm, can be found at this late stage of The 1.1-kb transcript present in Pgk-Pem ES cell clones embryoid body development. may be another transcript from the Pgk-Pem expression When Pgk-Pem ES clones 8, 14, and 19 were grown in construct or may represent an increase in endogenous Pem suspension culture to induce spontaneous differentiation, expression in Pgk-Pem clones. no morphological changes characteristic of wild-type ES To determine the origin of the 1.1-kb transcript in Pgk- cell embryoid bodies were observed. Histological section of

FIG. 1. Characterization of Pgk-Pem ES clones. (A) Genomic structure of the mouse Pem gene and Pem expression construct. Shaded boxes are Pem exons (E1 to E6). The full-length Pem mRNA and the Pem-speciﬁc ribonuclease protection probe used in measuring endogenous and Pgk-Pem transcripts are depicted below the genomic locus. The arrow in the Pem expression construct represents the approximate location of transcription initiation from the Pgk-1 promoter. The predicted Pgk-Pem mRNA is depicted below the expression construct. HD, Pem homeodomain. PGK-1, promoter region of mouse housekeeping gene Pgk-1. B, BamHI site. (B) Southern blot analysis of expression plasmid integration in Pgk-Pem ES clones. 8, 14, and 19 are independent Pgk-Pem ES clones. The 6-kb EcoRI band is the endogenous Pgk-1 genomic fragment, and the 3-kb band represents the integrated Pgk-Pem expression plasmid. The Southern blots were hybridized with the Pgk-1 promoter sequences. (C) Northern blot analysis of Pem gene expression. The hybridization probe is a full-length Pem cDNA. (D) Ribonuclease protection analysis of Pem gene transcripts in Pgk-Pem ES cells. The ribonuclease protection probe is from the 5Ј end of Pem mRNA, and 244 bp represents full-length probe protection.

FIG. 2. Comparison of morphology and proliferation rate of Pgk-Pem ES clone 8 with normal D3 ES cells. (A) Phase-contrast image of normal D3 ES cells. (B) Phase-contrast image of Pgk-Pem ES clones. (C) Proliferation rate assay of D3 ES cells and Pgk-Pem ES cells.

embryoid bodies derived from Pgk-Pem ES cell clone 8 genes. Compared to D3 ES cells, Pgk-Pem ES cell clone 8 revealed a uniform cluster of cells, without any evidence of showed prolonged expression of the Oct3/4 gene during overt segregation of cells into different cell types (Figs. 3C growth as embryoid bodies (Fig. 4). Oct3/4 is a POU-domain and 3D). Indeed, except for the increase in size of the homeobox-containing gene that is highly expressed in the embryoid bodies, the morphology of the Pgk-Pem embryoid preimplantation embryo, most notably in undifferentiated, bodies after 8 days in culture was the same as the morphol- pluripotent ICM cells (Palmieri et al., 1994). After 8 days of ogy following 4 days in culture. Moreover, no differentiated culture in medium without LIF, high levels of transcripts cell type emerged following reattachment and further normally abundant in visceral and/or parietal endoderm, growth of the Pgk-Pem clone 8 embryoid bodies on tissue such as Laminin B1, H19, and ␣-fetoprotein, are present in culture dishes (data not shown). Embryoid bodies from D3 ES cell embryoid bodies (Fig. 4 and Shen and Leder, clones 14 and 19 displayed the same morphological pheno- 1992). However, the levels of H19, ␣-fetoprotein, and Lami- type as embryoid bodies from clone 8 (data not shown). nin B1 expression in embryoid bodies derived from the Properties of the Pgk-Pem embryoid bodies were exam- Pgk-Pem ES cells are similar to levels found in undifferen- ined in detail by analyzing the expression of lineage-speciﬁc tiated ES cells. Speciﬁcally, H19 and ␣-fetoprotein expres-

nude mice, tumors of size similar to that of the D3 ES cell-derived teratomas were obtained. However, rather than being composed of a variety of typical differentiated embryonic tissues, the Pgk-Pem teratomas were composed en- tirely of undifferentiated, EC-like cells (Fig. 5B). These results are also consistent with a block in ES cell differentiation due to the forced expression of Pem.

Pem Inhibition of ES Cell Differentiation Is a Cell- Autonomous Effect Failure of Pgk-Pem ES cell differentiation could result from either a cell autonomous or a nonautonomous mechanism. To study the cellular mechanism of inhibition of Pgk-Pem ES cell differentiation, we attempted to differentiate Pgk-Pem ES cells in the presence of normal ES cells. Both in vitro (embryoid body) and in vivo (teratoma) assays for differentiation were used (see above). To distinguish the derivatives of normal ES cells from the Pgk-Pem clones, the ROSA26 ES cell line was employed (Friedrich and Soriano, 1991; Chen and Behringer, 1995). ROSA26 contains a constitutively expressed bacterial ␤-galactosidase gene, which is ubiquitously expressed during mouse embryogenesis and FIG. 3. Histological analysis of Pgk-Pem ES cell clone 8 embryoid permits identiﬁcation of these cells after X-Gal staining. bodies. (A) D3 ES embryoid body cultured for 4 days. (B) D3 ES Histological sections of both embryoid bodies and terato- embryoid body cultured for 8 days. (C) Pgk-Pem ES cell embryoid mas derived from mixtures of Pgk-Pem clone 8 and body cultured for 4 days. (D) Pgk-Pem ES cell embryoid body cultured for 8 days. The dark circular spots in the sections are from osmium precipitates, used to localize the embryoid bodies following embedding in plastic resin.

sion is negligible and there is no increase in the level of Laminin B1 expression compared to undifferentiated ES cells. Moreover, ␨-globin, which is normally expressed after 8–10 days of embryoid body culture, was not found in Pgk-Pem ES cells. Identical patterns of gene expression were obtained upon growth of Pgk-Pem ES clones 14 and 19 as embryoid bodies (data not shown). These results are consistent with the homogeneous morphology of Pgk-Pem embryoid bodies and the inability of Pgk-Pem ES cells to undergo spontaneous differentiation upon aggregation.

Forced Expression of Pem Inhibits ES Cell Differentiation in Vivo To further analyze the differentiation potential of Pgk- Pem ES cells, we injected normal and Pgk-Pem ES cells subcutaneously into nude mice. Six weeks after subcutaneous injection of normal ES cells, we obtained teratomas FIG. 4. Lineage-speciﬁc gene expression in differentiating D3 and containing mesodermal, ectodermal, and endodermal tis- Pgk-Pem ES cell clone 8 embryoid bodies. Expression of Oct3/4 and H19 was determined by Northern blot analysis. Expression of sues, as expected from such ectopic anatomical placement ␨-globin, ␣-fetoprotein, and laminin B1 was determined by ribo- of ES cells (Evans and Kaufman, 1981; Martin, 1981). A nuclease protection assays. The numbers near the top are the histological section of a representative teratoma derived numbers of days of embryoid body growth in suspension culture. from D3 ES cells is shown in Fig. 5A. Differentiated RNA samples at day 0 of embryoid body growth were collected embryonic epithelia are evident in this section. Six weeks from undifferentiated ES cells, grown attached to tissue culture after injection of Pgk-Pem ES cells (Pgk-Pem clone 8) into dishes in the presence of LIF.

FIG. 5. Histological analysis of teratomas derived from D3 ES cells and from Pgk-Pem ES cell clone 8. (A) Section of teratoma derived from normal D3 ES cells, showing well-differentiated endodermal (single arrow) and ectodermal (double arrow) structures with intervening mesodermal cells. (B) Section of teratoma derived from Pgk-Pem ES cell clone 8, showing uniformly undifferentiated and immature EC-like cells with active mitosis and high nucleus–cytoplasm ratio. FIG. 6. Histological analysis of teratomas and embryoid bodies derived from a mixture of normal ES cells and Pgk-Pem ES cell clone 8. (A) Teratomas derived from a mixture of Pgk-Pem ES cells and ROSA26 ES cells constitutively expressing bacterial ␤-galactosidase (stained blue). Well-differentiated epithelial structures containing only blue cells derived from ROSA26 ES cells are shown, surrounded by undifferentiated Pgk-Pem cells. (B) Immature undifferentiated region of the same teratoma showing only Pgk-Pem ES cell derivatives without any blue cells of a ROSA26 origin. (C) Histological sections of embryoid bodies derived from Pgk-Pem and ROSA26 cell mixture. The outer endoderm cells contain only blue (X-Galϩ) cells of ROSA26 origin, while a central cluster of undifferentiated Pgk-Pem ES cells were stained pink (X-GalϪ).

ROSA26 cells showed extensive proliferation of both Pgk- in the surface layer of differentiated visceral endoderm or Pem ES cells and ROSA26 cells. Close examination of found lining embryoid body cavities. Unstained Pgk-Pem cells teratoma sections showed numerous highly differentiated comprised the undifferentiated core of embryoid bodies (Fig. structures, in which all cells were derived from ROSA26 6C). The absence of Pgk-Pem ES cells in highly differentiated (X-Galϩ) ES cells (Fig. 6A). Pgk-Pem cells (X-GalϪ) were tissues suggests that the block to differentiation in Pgk-Pem found only in undifferentiated, immature areas (Fig. 6B). ES cells cannot be rescued by the presence of normal ES cells. Sections of the day 8 embryoid bodies showed similar Therefore, the block to the differentiation of Pgk-Pem cells patterns of cell distribution. Only ROSA26 cells were found appears to be a cell-autonomous effect.

FIG. 7. Examination of the effect of Pem domains on ES cell differentiation. On the left are six Pem cDNA expression constructs, engineered to determine the part of Pem responsible for the block in ES cell differentiation. HA-TAG is a triple hemagglutinin epitope; NLS is the SV40 large T-antigen nuclear localization signal, and N-terminal is the portion of Pem N-terminus of its homeodomain. The homeodomain in constructs 1–5 is the Pem homeodomain, and the homeodomain in construct 6 is the Esx1/Spx1 homeodomain. In the middle, the numbers near the top are the numbers of days of embryoid body growth in suspension culture. RNA samples at day 0 of embryoid body growth were collected from undifferentiated ES cells, grown attached to tissue culture dishes in the presence of LIF. The levels of ␣-fetoprotein (AFP) expression in embryoid bodies containing constructs 1–6 are shown. ␣-fetoprotein expression was determined by a ribonuclease protection assay. On the right, embryoid body differentiation was determined by phase-contrast microscopy and by microscopic examination of histological sections. ϩ indicates normal embryoid body differentiation. Ϫ indicates a nullipotent embryoid body morphology.

Overexpression of the Nonhomeodomain Portion of stably transfected with a Pem cDNA construct that con- Pem Blocks ES Cell Differentiation tains an in-frame stop codon following the HA tag undergo normal differentiation in suspension culture (data not To further address the mechanism whereby forced ex- shown). Moreover, an ES cell clone stably transfected with pression of the Pem gene blocks ES cell differentiation, we a different HA triple-tagged expression construct (Fig. 7, examined the effect of increasing the expression of portions construct 6), containing the coding region of Esx1/Spx1 of the Pem protein in ES cells (Fig. 7). We constructed a Pem expression plasmid containing a full-length Pem cDNA, cDNA, differentiated normally. These results suggest that transfected this into ES cells, and obtained stably trans- the observed block to ES cell differentiation is speciﬁc to fected ES cell lines (Fig. 7, construct 1). To facilitate the the forced expression of the full-length Pem gene. detection of Pem from the transfected construct, a triple To determine the portion of the Pem protein responsible hemagglutinin (HA) epitope tag was cloned in frame to the for the block to ES differentiation, we evaluated the effect of 5Ј end of the Pem cDNA. We used ␣-fetoprotein expression overexpressing different regions of the Pem cDNA in ES ␣ as an indicator of the differentiation of the transfected ES cells. Using an increase in -fetoprotein expression as an cells, because ␣-fetoprotein has been shown to be an excel- indication of embryoid body differentiation, the forced lent marker of early differentiation during culture of em- expression of the N-terminal nonhomeodomain portion of bryoid bodies (Fig. 4 and Shen and Leder, 1992). As shown in Pem protein alone, in three of three ES cell clones exam- Fig. 7, constitutive expression of the full-length Pem cDNA ined, blocked ES cell differentiation (Fig. 7, construct 2). In has the same inhibitory effect on ES cell differentiation as contrast, all three independently derived ES cell clones the genomic Pem expression construct (Fig. 1A). Only one overexpressing the remaining portion of Pem, which in- such modiﬁed ES clone was examined. In addition, the cludes the C-terminal homeodomain, differentiated nor- morphology of embryoid bodies derived from these ES cells mally (Fig. 7, construct 3). In cells of the early embryo, Pem is the same as embryoid bodies from Pgk-Pem ES cells (data is found predominantly in the nucleus, consistent with a not shown). The triple HA epitope tag attached to Pem has nuclear function (Lin et al., 1994). To exclude the possibil- no apparent effect on ES cell differentiation since ES clones ity that the different effects of forced expression of the

localization of the HA-tagged protein (Figs. 8C and 8D). Moreover, cells expressing an HA-tagged full-length Pem protein (from construct 1, Fig. 7) or an HA-tagged C-terminal portion of Pem (from construct 5, Fig. 7) also contained the tagged proteins in the nucleus (data not shown).

Embryoid Bodies from Pem؊/Y ES Cells Differentiate To further study the role of the Pem gene during ES cell differentiation, we also established D3 ES cell lines containing a null allele of Pem (PemϪ/Y ES cells). The gene encoding Pem was mutated by homologous recombination using a targeting vector in which exons 3 and 4 were replaced by a loxP/Pgk-tk/Pgk-neo/loxP (LTNL) cassette. Because the Pem gene is located on the X chromosome and the D3 ES cell line was derived from a male mouse, targeted mutagen- esis of the Pem allele will produce a PemϪ/Y ES cell line. D3 ES cells transfected with the targeting construct (Fig. 9) were screened by Southern blot analysis with a probe from the deleted region (exons 3–4) (Figs. 9A and 9B). DNA from FIG. 8. Immunofluorescence analysis of an HA-tagged Pem pro- two clones (Nos. 19 and 66) did not hybridize to the probe, tein. (A and B) Normal D3 ES cells as viewed with differential indicating the targeted disruption of Pem in these two interference contrast (DIC) microscopy (A) or as viewed with clones. To verify the gene’s disruption in clones 19 and 66, fluorescence microscopy after immunostaining with an anti-HA- a probe outside of the targeting construct (probe B) was used tag primary antibody and a FITC-conjugated secondary antibody to detect the expected restriction fragment length variation (B). (C and D) ES cells expressing an HA-tagged, N-terminal portion created by introducing an EcoRI site in the mutated Pem of Pem as viewed with DIC microscopy (C) or as viewed with allele (Figs. 9A and 9B). The normal wild-type Pem allele is fluorescence microscopy (D). represented as an 8.5-kb band on EcoRI-digested genomic DNA, whereas the mutated allele is seen as a 6.5-kb band following EcoRI digestion (Fig. 9B). In both mutant clones, N-terminal and the C-terminal halves of Pem were due to a the expression of Pem mRNA was analyzed by Northern difference in intracellular localization of the proteins, we blotting using either the 5Ј or the 3Ј half of the Pem cDNA targeted both portions of Pem to the nucleus by incorporat- as a hybridization probe (Fig. 9C). As expected, no endoge- ing an SV40 signal for nuclear localization (NLS) into Pem nous Pem mRNA transcript of 1.1 kb was detected using cDNA constructs. The addition of an SV40 NLS to both the either probe. Using the 3Ј half of the Pem cDNA as a probe, N-terminal (three clones examined) and the C-terminal a weakly hybridizing band of 1.3 kb was seen. Although the (three clones examined) HA-tagged Pem proteins did not precise etiology of this hybridization band is not known, it alter the outcome of the differentiation experiments (Fig. 7, may represent a transcript initiated from the promoters of constructs 4 and 5). A block to differentiation was apparent the upstream selectable marker genes and including se- only when proteins containing the N-terminal, nonhomeo- quences of the remaining downstream Pem exons (exons 5 domain portion of Pem were overexpressed in ES cells (Fig. and 6). It is unlikely that such a transcript would produce a 7, constructs 1, 2, and 4). protein product with Pem activity, as the Pem coding To verify the nuclear localization of the proteins ex- sequences remaining in the mutated Pem allele encode pressed from the different Pem cDNA constructs, we exam- amino acids for only helix 3 of the Pem homeodomain. ined the intracellular localization of the HA-tagged full- To analyze the differentiation potential of PemϪ/Y ES length Pem protein, as well as the HA-tagged (and NLS- cells, PemϪ/Y ES clones 19 and 66 were expanded and grown tagged) N-terminal and C-terminal portions of Pem as embryoid bodies in the absence of LIF. Both PemϪ/Y ES (constructs 1, 4, and 5 in Fig. 7). These localization studies clones showed no morphological difference from wild-type were performed using an immunofluorescence assay with ES cells either as attached colonies in the presence of LIF or an anti-HA monoclonal antibody. One example of these as embryoid bodies grown in the absence of LIF (data not localization studies is shown in Fig. 8. As expected, normal shown). The differentiation potentials of both PemϪ/Y ES D3 ES cells, without a transfected Pem expression plasmid, clones were further analyzed by examining the expression did not show any evidence of HA antigen (Figs. 8A and 8B). of early embryonic and extraembryonic lineage-specific However, cells expressing the HA-tagged N-terminal por- genes. For PemϪ/Y clone 66, notable differences were seen tion of Pem from construct 4 (Fig. 7) showed nuclear between normal and PemϪ/Y ES cells in the expression of

FIG. 9. Targeted disruption of the Pem gene in mouse ES cells. (A) Restriction maps of the mouse Pem genomic locus, targeting construct, and predicted structure of targeted Pem allele. Restriction enzymes: B, BamHI; E, EcoRI. LTNL, loxP/Pgk-tk/Pgk-neo/loxP cassette, providing the selectable marker. Pem exons are displayed as dark boxes. Dashed lines denote the borders of homology between the wild-type Pem gene and the targeting construct. The locations of hybridization probes A and B are shown. The predicted sizes of EcoRI restriction fragments from wild-type and mutant alleles are 8.5 and 6.5 kb, respectively. (B) Southern blot analysis of ES clones. Genomic DNA isolated from wild-type D3 ES cells and PemϪ/Y ES clones (19 and 66) was digested with EcoRI and hybridized with either probe A (left) or probe B (right). (C) Northern blot analysis of Pem mRNA expression in PemϪ/Y ES clones. Total RNA was collected from attached ES cells cultured with 1000 Units/ml of LIF. 5Ј Pem cDNA probe is from 1 to 433 nt of Pem cDNA, and 3Ј Pem cDNA probe is from 435 to 781 nt.

Oct3/4, ␨-globin, ␣-fetoprotein, and laminin B1 (Fig. 10). DISCUSSION The level of Oct3/4 transcripts did not decline as rapidly in Ϫ Pem /Y cells as in normal ES cells. Furthermore, after 12 Pgk-Pem Phenotype days of embryoid body culture, the levels of ␨-globin, ␣-fetoprotein, and laminin B1 in PemϪ/Y cells were signifi- Murine ES cell lines and EC cell lines have been exten- cantly lower than the levels of these transcripts found in sively used in both in vitro and in vivo studies to understand the processes underlying differentiation of early em- normal ES cells. In particular, after 12 days of embryoid bryonic cell types (Martin, 1980; Coucouvanis and Martin, body culture, ␨-globin expression was evident in D3 ES Ϫ 1995; Keller, 1995). In this study, we modified the D3 ES cells, but absent in Pem /Y cells. Identical time courses and cell line by introducing a Pgk-Pem expression plasmid, levels of gene expression were observed for PemϪ/Y clone 19 which forces the expression of the mouse homeobox- PemϪ/Y (data not shown). Although cells differentiated as containing gene Pem from the promoter of the mouse Pgk-1 embryoid bodies, the pattern of gene expression in these gene. This genetic alteration of D3 ES cells into Pgk-Pem ES embryoid bodies suggests that their differentiation is al- lines results in a profound inhibition of cellular differentia- tered compared to normal ES cells. Based on the profiles of tion. Specifically, the transition from undifferentiated ES gene expression in embryoid bodies derived from normal ES cells to differentiated cell types, recognized by their distinc- Ϫ/Y cells, Pem cells, and Pgk-Pem cells (Figs. 4 and 10), we tive morphology and their lineage-specific gene expression, Ϫ conclude that the Pem /Y ES cells exhibit an intermediate is blocked when the expression of Pem is increased. This phenotype between those of normal ES cell embryoid bod- block to cellular differentiation is evident when Pgk-Pem ies and Pgk-Pem embryoid bodies. ES cells are grown as embryoid bodies. Pgk-Pem embryoid

are unable to undergo the normal series of embryoid body differentiation, producing instead embryoid bodies with the same morphology as is found in embryoid bodies derived from nullipotent EC cell lines (Martin and Evans, 1975b; Schindler et al., 1984). Furthermore, when Pgk-Pem ES cells are transplanted into ectopic anatomical locations in nude mice, where normal ES cells and pluripotent EC cells readily proliferate and differentiate, they do not differentiate. Although the subcutaneously implanted Pgk-Pem ES cells readily grew into tumors (teratomas), these tumors comprised solely immature, undifferentiated cells. Because ES cells with a forced level of Pem expression are also incapable of differentiation when cocultured with normal ES cells, it appears that Pem overexpression acts at the level of individual cells (cell-autonomous effect). Inter- estingly, the effect of the forced expression of the Pem homeoprotein can be reproduced by the forced expression of

Ϫ just the N-terminal, nonhomeodomain portion. Therefore, FIG. 10. Lineage-speciﬁc gene expression during Pem /Y ES cell the genetic manipulation of ES cells via the forced expres- clone 66 embryoid body differentiation. Embryoid bodies were grown without LIF for 4, 8, and 12 days. RNA samples from sion of the Pem homeoprotein from the Pgk-Pem expres- undifferentiated ES cells were sources of samples designated day 0. sion plasmid produces a phenotype similar, if not identical, 15 ␮g of total RNA was used in ribonuclease protection assays to the phenotype observed in nullipotent EC cell lines. The (␨-globin, ␣-fetoprotein, and Laminin B1), and 7.5 ␮g of total RNA phenotype of the Pgk-Pem ES is a consequence of the forced was used for Northern blot analysis (Oct3/4). RPL32 was used as an expression of Pem, while the nullipotency of certain EC internal loading control. cells is due to unknown genetic or epigenetic mechanisms (Martin, 1980).

Mechanism of Pem Effect in Blocking ES Cell bodies are morphologically distinct from those derived from Differentiation normal D3 ES cells grown under identical conditions. Notably, primitive endoderm does not form in Pgk-Pem ES What could be the potential molecular mechanism of the embryoid bodies, even though parietal and visceral effect of forced Pem expression on ES cell differentiation? It endoderm readily form in normal D3 ES cells and in PemϪ/Y has been known that cytokines such as LIF play important ES cells. We conclude from these observations that the roles in maintaining the undifferentiated state of ES cells forced level of Pem gene expression results in an efﬁcient (Rathjen et al., 1990; Smith et al., 1992). Upon LIF binding, and complete block to cellular differentiation. LIF receptors form heterodimers with glycoprotein gp130, a The inhibitory phenotype of Pgk-Pem ES cells is reminis- common cytokine signal transducer, and activate both the cent of the phenotype seen in a variety of nullipotent EC JAK/STAT pathway and the ras/MAP kinase pathway to cell lines (Stevens, 1958; Martin, 1980; Jakob and Nicolas, support ES cell survival and proliferation (Ernst et al., 1996; 1987). Early studies of EC cells showed that a profound loss Niwa et al., 1998). Withdrawal of LIF from ES cultures will of differentiation potential can occur in certain EC cell lines deactivate the JAK/STAT signaling cascade (Hocke et al., during their in vitro culture or in vivo passage as teratomas 1995). In both of these signal transduction systems, direct (Stevens, 1958; Martin, 1980). Not surprisingly, these nul- activation of factors downstream of LIF signaling could lipotent EC cell lines have a number of features that effectively maintain the pluripotency of ES cells (Yoshida et resemble the undifferentiated state of pluripotent ES cells al., 1994). Interestingly, the phenotype of Pgk-Pem embry- and the undifferentiated states of EC cell lines that are oid bodies is similar to the normal ES cell phenotype seen in pluripotent or have a more restricted developmental poten- the presence of LIF. Speciﬁcally, in the absence of LIF, the tial. These nullipotent features include a high level of Pgk-Pem ES cell embryoid bodies are unable to differenti- alkaline phosphatase activity and the expression of certain ate. Therefore, forced expression of Pem in ES cells might cell surface antigens (Bernstine et al., 1973; Fox et al., 1981; exert its effect on downstream factors of LIF signaling Jakob and Nicolas, 1987; Pease et al., 1990). Although such pathways, maintaining the undifferentiated state of ES cells nullipotent EC cells can aggregate as embryoid bodies in in a LIF-independent manner. suspension culture or form teratomas in vivo, they do not The effect on embryoid body differentiation produced differentiate into any differentiated cell types. Rather, they by the overexpression of Pem appears to be a stronger proliferate as a homogeneous cluster of undifferentiated block to differentiation than the block observed in the cells (Stevens, 1958; Martin and Evans, 1975b). In a fashion presence of LIF. In the presence of 1000 Units/ml LIF, similar to that of nullipotent EC cells, Pgk-Pem D3 ES cells embryoid bodies from normal ES cells will form primi-

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. Pem Blocks ES Differentiation 493 tive endodermal cells, although mesodermal cell types role in maintaining the undifferentiated state of embryo will not develop (Shen and Leder, 1992). A similar in vivo cells or in regulating the transition between undifferenti- effect of LIF on ES cell differentiation was observed by ated and differentiated cells. In this regard, observations on Conquet et al. (1992). Upon the introduction of ES cells Pem gene expression in a variety of tumor cell lines and in constitutively expressing LIF into blastocysts, the ES the normal mouse embryo are instructive. A general char- cells were able to undergo normal extraembryonic devel- acteristic of Pem expression in tumor-derived cell lines is opment, but were unable to participate normally in seen when the level of expression in each line is compared gastrulation. These studies suggest that the phenotype of to the level in normal tissues (Wilkinson et al., 1990). Pem embryoid bodies from Pgk-Pem ES cells does not appear expression is significantly higher in a variety of murine simply to be the consequence of activation of a LIF- lymphoma-derived cell lines than in normal lymphoid cells dependent signaling pathway (Gendall et al., 1997). (Wilkinson et al., 1990). As well, Pem is expressed in many Regardless of the specific molecular mechanism pro- other cell lines derived from nonlymphoid adult tumors, ducing the Pgk-Pem phenotype, it is clear from the but not expressed in the parent tissue type. These observa- described experiments that the phenotype is mediated by tions suggest a role for Pem in generating or maintaining the N-terminal Pem sequences. A block in ES cell differ- immortalized or transformed versions of normal adult cell entiation as a result of the forced expression of the types. nonhomeodomain portion of Pem protein alone suggests The observed patterns of Pem expression in mouse em- that the inhibitory effect on differentiation is due to bryos and embryoid bodies also suggest that it is unlikely protein–protein interactions. It has been well known that that Pem functions solely in defining or maintaining a interaction with other proteins is crucial for a homeo- specific differentiated cell type. Pem protein is expressed in domain-containing protein to exert its function (Zhang et all cell types found in the preimplantation embryo, includ- al., 1997). In addition to its extensively studied role in ing ICM cells (Lin et al., 1994). After implantation, Pem binding to specific DNA sequences, homeodomain- expression during embryogenesis is restricted to cells of containing proteins can interact directly with other pro- extraembryonic lineages, including trophoblast cells and teins, including other homeoproteins. Some of these primitive endodermal derivatives. Pem expression in undif- protein–protein interactions are known to occur outside ferentiated and primitive endodermal derivatives can also of the homeodomain. For example, Hox gene products be seen during in vitro embryoid body differentiation. In have the ability to interact with either Extradenticle or embryoid bodies grown from D3 ES cells, Pem expression Pbx gene products and to cooperatively enhance binding dramatically increases after 2 days of growth, well before to specific DNA sequences. The region in Hox proteins morphological signs of cellular differentiation appear required for this reaction is located N-terminal of the (Sasaki et al., 1991). Pem is also expressed in undifferenti- homeodomain and contains a highly conserved hexapep- ated F9 EC cells and in visceral or parietal endoderm tide (Chang et al., 1996). Likewise, the developmental developing in F9 embryoid bodies following exposure to and transcriptional effects of a number of mammalian retinoic acid (Lin et al., 1994). Notably, Pem is absent from homeoproteins are mediated through their nonhomeodo- the core cells of the embryoid bodies, which correspond to main regions. For example, a highly conserved region in embryonic ectoderm of the mouse embryo (Lin et al., 1994). the N-terminus of the Drosophila homeodomain- Based on the localization of Pem in mouse embryos and containing gene Fushi tarazu (Ftz) protein product can in F9 embryoid bodies, on the mouse phenotype resulting interact with ␣Ftz-F1, which is important for regulating from knocking out Pem (Sasaki et al., 1991; Lin et al., 1994; proper expression of Engrailed and Wingless gene prod- Pitman et al., 1998), and on ES cell phenotypes resulting ucts (Florence et al., 1997; Gulchet et al., 1997; Yu et al., from altering Pem levels in ES cells, we can develop a model 1997). Deletion of the homeodomain of Ftz has no effect for Pem’s function in mouse embryos. We propose that Pem on this activity. Another well-characterized example is is involved in regulating the transition between undifferen- Hox11. Its N-terminal nonhomeodomain region is tiated and differentiating cells of the early embryo. We known to interact with cell cycle regulatory phospha- hypothesize that this regulation is a two-step process, both tases, which are known to be important for cell-cycle steps involving Pem. Pem helps to maintain the undiffer- transitions (Kawabe et al., 1997). In light of these types of entiated stem cell state, and Pem also is involved in helping protein–protein interactions, it is not surprising that Pem to direct early differentiation to particular cell lineages. In function can be mediated through the nonhomeodomain this model, the effect of forced Pem expression is most portion. evident in the first step, resulting in maintenance of the undifferentiated state. Such a proposed effect of Pem may be occurring normally in early germ cell development, in The Biological Significance of Blocking ES Cell which Pem expression appears to be exclusively in the Differentiation by the Forced Expression of Pem undifferentiated primordial germ cells, but not in differen- Currently, the function of the Pem gene in normal mouse tiating derivatives (Pitman et al., 1998). Pem’s second development is not known. Our observations on the phe- action is to promote stem cell differentiation into extraem- notype of Pgk-Pem ES cells suggest that Pem may play a bryonic lineages, possibly by helping to define a population

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. 494 Fan, Melhem, and Chaillet of undifferentiated, Pem-expressing cells that are commit- cell differentiation. In molecular terms, such an effect could ted to differentiation. This second function of Pem is result from the sequestration or depletion of cellular pro- affected by the removal of Pem in PemϪ/Y ES cells, resulting teins by the elevated concentrations of Pem. These Pem- in a partial block or delay in forming extraembryonic interacting proteins would then not be available for inter- lineages. actions with other cellular proteins involved in the Such a dual role during preimplantation and postimplan- transition between undifferentiated and differentiating tation development has been suggested for other genes cells. This interpretation of our findings suggests that both expressed throughout the preimplantation embryo and later full-length Pem and the N-terminal portion of Pem, when expressed in lineage-restricted patterns (Rossant et al., overexpressed, function in a dominant-negative fashion. 1998). For example, Oct3/4, a member of the POU-domain Other examples in which overexpression of the normal, class of homeobox-containing genes, has been proposed to full-length form of a protein can lead to a dominant- be critical for development of the ICM lineage. Notably, it negative inhibition of function can be found. For example, is expressed throughout the early preimplantation embryo in the case of Drosophila E-cadherin, overexpression of and becomes ICM-specific at the late blastocyst stage either full-length E-cadherin or a truncated form leads to a (Palmieri et al., 1994). In contrast, Mash2, which is also sequestration of the Armadillo protein and a dominant- expressed in preimplantation embryo, becomes specifically negative block to the wingless intracellular signaling path- expressed in trophoblast cells only after implantation. way (Sanson et al., 1996). Despite our inability at this time These complementary patterns of Oct3/4 and Mash 2 to determine how Pem works, the phenotype in ES cells expression suggest that commitment of cells to the tropho- associated with the forced expression of Pem affords an blast lineage requires the expression of Mash2, as well as opportunity to investigate its biochemical function. repression of Oct3/4 (Rossant et al., 1998). Assuming that certain genes, expressed in the preimplantation embryo, function in regulating the commitment of ACKNOWLEDGMENTS cells to specific lineages following implantation, how do we explain the phenotype found in the Pgk-Pem ES cells? A We thank Dr. Phil Soriano for ROSA26 cells and Dr. Michael general explanation is that a balance in the relative expres- Shen for lineage-specific ribonuclease protection and Northern blot sion of certain regulatory genes is important for preimplan- probes. We express our gratitude to Tom Harper for his assistance tation development. Following implantation, and coinci- with embryoid body fixation, sectioning, and staining. We are very grateful to Dr. Richard Carthew and Dr. Julie A. DeLoia for their dent with the differentiation of stem cells into specific helpful discussions and comments on the manuscript. This study lineages, this balance changes. The activity of some genes is was supported by PHS Grant CA73811 to J.R.C. elevated, whereas the activity of others is repressed. Acti- vated genes assume a lineage-specific role, whereas the repression of other genes makes the lineage determination REFERENCES event permissive. In this formulation, an introduced alteration in the level of activity of one of the regulatory genes Adra, C. H., Boer, P. H., and McBurney, M. W. (1987). Cloning and would affect the normal program of embryonic differentia- expression of the mouse Pgk-1 gene and the nucleotide sequence tion. In this manner, the forced expression of Pem in of its promoter. Gene 60, 65–74. undifferentiated ES cells would preclude their differentia- Bernstine, E. G., Hooper, M. L., Grandchamp, S., and Ephrussi, B. tion, including their differentiation into cells of extraem- (1973). Alkaline phosphatase activity in mouse teratoma. Proc. bryonic lineages. Natl. Acad. Sci. USA 70, 3899–3903. An alternative model for Pem’s function is strongly Branford, W. W., Zhao, G. Q., Valerius, M. T., Weinstein, M., Birkenmeier, E. H., Rowe, L. B., and Potter, S. S. (1997). Spx1, a suggested by a comparison of the profiles of gene expression novel X-linked homeobox gene expressed during spermatogen- Pgk-Pem PemϪ/Y in normal D3 ES cells, cells, and cells. The esis. Mech. Dev. 65, 87–98. ␨ ␣ rates of change in Oct3/4, -globin, -fetoprotein, and Chang, C. P., Brocchieri, L., Shen, W. F., Largman, C., and Cleary, Ϫ/Y Laminin B1 transcript levels during Pem embryoid body M. L. (1996). Pbx modulation of Hox homeodomain amino- differentiation are between the rates seen in normal ES cells terminal arms establishes different DNA-binding specificities and Pgk-Pem cells (Figs. 4 and 10). This intermediate across the Hox locus. Mol. Cell Biol. 16, 1734–1745. phenotype of the PemϪ/Y cells suggests that Pem’s normal Chen, Z. F., and Behringer, R. R. (1995). Twist is required in head function is in directing early differentiation to particular mesenchyme for cranial neural tube morphogenesis. Genes Dev. lineages. Studies of Pem expression in embryoid bodies and 9, 686–699. early embryos (Sasaki et al., 1991; Lin et al., 1994) suggest Conquet, F., Peyrieras, N., Tiret, L., and Brulet, P. (1992). Inhibited gastrulation in mouse embryos overexpressing the leukemia that Pem may fulfill this role by helping to define cells that inhibitory factor. Proc. Natl. Acad. Sci. USA 89, 8195–8199. appear undifferentiated, yet are committed to differentia- Coucouvanis, E., and Martin, G. R. (1995). Signals for death and tion. In this one-step model, Pem would have no direct survival: A two-step mechanism for cavitation in the vertebrate action in maintaining embryo cells undifferentiated. The embryo. Cell 83, 279–287. phenotype of Pgk-Pem cells would be an enhancement of Cross, J. C., Flannery, M. L., Blanar, M. A., Steingrimsson, E., Ϫ the Pem /Y phenotype, leading to a complete block in ES Jenkins, N. A., Copeland, N. G., Rutter, W. J., and Werb, Z.

(1995). Hxt encodes a basic helix-loop-helix transcription factor Lin, T. P., Labosky, P., Grabel, L. B., Kozak, C. A., Pitman, J. L., that regulates trophoblast cell development. Development 121, Kleeman, J., and MacLeod, C. L. (1994). The pem homeobox gene 2513–2523. is X-linked and exclusively expressed in extraembryonic tissues Cserjesi, P., Brown, D., Lyons, G. E., and Olson, E. N. (1995). during early murine development. Dev. Biol. 166, 170–179. Expression of the novel basic helix-loop-helix gene eHAND in Maiti, S., Doskow, J., Sutton, K., Nhim, R. P., Lawlor, D. A., Levan, neural crest derivatives and extraembryonic membranes during K., Lindsey, J. S., and Wilkinson, M. F. (1996). The Pem ho- mouse development. Dev. Biol. 170, 664–678. meobox gene: Rapid revolution of the homeodomain, X chromo- Ernst, M., Oates, A., and Dunn, A. R. (1996). Gp130-mediated somal localization, and expression in reproductive tissue. signal transduction in embryonic stem cells involves activation Genomics 34, 304–316. of Jak and Ras/mitogen-activated protein kinase pathways. Martin, G. R. (1980). Teratocarcinomas and mammalian embryo- J. Biol. Chem. 271, 30136–30143. genesis. Science 209, 768–776. Evans, M. J., and Kaufman, M. (1981). Establishment in culture of Martin, G. R. (1981). Isolation of a pluripotent cell line from early pluripotential cells from mouse embryos. Nature 292, 154–156. mouse embryos cultured in medium conditioned by teratocarci- Florence, B., Guichet, A., Ephrussi, A., and Laughon, A. (1997). noma stem cells. Proc. Natl. Acad. Sci. USA 78, 7634–7638. Ftz-F1 is a cofactor in Ftz activation of the Drosophila engrailed Martin, G. R., and Evans, M. E. (1975a). Differentiation of clonal gene. Development 124, 839–847. lines of teratocarcinoma cells: Formation of embryoid bodies in Fox, N., Damjanov, I., Marinez-Hernandez, A., Knowles, B. B., and vitro. Proc. Natl. Acad. Sci. USA 72, 1441–1445. Solter, D. (1981). Immunohistochemical localization of the early Martin, G. R., and Evans, M. J. (1975b). The formation of embryoid embryonic antigen (SSEA-1) in postimplantation mouse embryos bodies in vitro by homogeneous embryonal carcinoma cell cul- and fetal and adult tissues. Dev. Biol. 83, 391–398. tures derived from isolated single cells. In “Teratomas and Friedrich, G., and Soriano, P. (1991). Promoter traps in embryonic Differentiation,” pp.169–188. Academic Press, New York. stem cells: A genetic screen to identify and mutate developmen- Martin, G. R., Wiley, L. M., and Damjanov, I. (1977). The develop- tal genes in mice. Genes Dev. 5, 1513–1523. ment of cystic embryoid bodies in vitro from clonal teratocarci- Gendall, A. R., Dunn, A. R., and Ernst, M. (1997). Isolation and noma stem cells. Dev. Biol. 61, 230–244. characterization of a leukemia inhibitory factor-independent McBurney, M. W., Sutherland, L. C., Adra, C. H., Leclair, B., embryonic stem cell line. Int. J. Biochem. Cell Biol. 29, 829–840. Rudnicki, M. A., and Jardine, K. (1991). The mouse Pgk-1 Guichet, A., Copeland, J. W., Erdelyi, M., Hlousek, D., Zavorszky, promoter contains an upstream activator sequence. Nucleic P., Ho, J., Brown, S., Percival-Smith, A., Drause, H. M., and Acids Res. 19, 5755–5761. Ephrussi, A. (1997). The nuclear receptor homologue Ftz-F1 and Nichols, J., Zevnik, B., Anastassiadis, K., Niwa, H., Klewe- the homeodomain protein Ftz are mutually dependent cofactors. Nebenius, D., Chambers, I., Scholer, H., and Smith, A. (1998). Nature 385, 548–552. Formation of pluripotent stem cells in the mammalian embryo Guillemot, F., Nagy, A., Auerbach, A., Rossant, J., and Joyner, A. depends on the POU transcription factor Oct4. Cell 95, 379–391. (1994). Essential role of Mash-2 in extraembryonic development. Niwa, H., Burdon, T., Chambers, I., and Smith, A. (1998). Self- Nature 371, 333–336. renewal of pluripotent embryonic stem cells is mediated via Han, Y. J., Park, A. R., Sung, D. Y., and Chun, J. Y. (1998). Psx, a activation of stat3. Genes Dev. 12, 2048–2060. novel murine homeobox gene expressed in placenta. Gene 207, Palmieri, S. L., Peter, W., Hess, H., and Scholer, H. R. (1994). Oct-4 159–166 transcription factor is differentially expressed in the mouse Hocke, G. M., Cui, M. Z., and Fey, G. H. (1995). The LIF response element of the alpha 2 macroglobulin gene confers LIF-induced embryo during establishment of the first two extraembryonic transcriptional activation in embryonal stem cells. Cytokine 7, cell lineages involved in implantation. Dev. Biol. 166, 259–267. 491–502. Pease, S., Braghetta, P., Gearing, D., Grail, D., and Williams, R. L. Hogan, B., Beddington, R., Costantini, F., and Lacy, E. (1994). In (1990). Isolation of embryonic stem (ES) cells in media supple- “Manipulating the Mouse Embryo: A Laboratory Manual,” pp. mented with recombinant leukemia inhibitory factor (LIF). Dev. 98–105. Cold Spring Harbor Laboratory Press, NY. Biol. 141, 344–352. Hollenberg, S. M., Sternglanz, R., Cheng, P. F., and Weintraub, H. Pedersen, R. A. (1986). Potency, lineage, and allocation in preim- (1995). Identification of a new family of tissue-specific basic plantation mouse embryos. In “Experimental Approaches to helix-loop-helix proteins with a two-hybrid system. Mol. Cell. Mammalian Embryonic Development,” pp. 3–34. Cambridge Biol. 15, 3813–3822. Univ. Press, New York. Jakob, H., and Nicolas, J. (1987). Mouse teratocarcinoma cells. Pitman, J. L., Lin, T. P., Kleeman, J. E., Erickson, G. F., and Methods Enzymol. 151, 66–81. MacLeod, C. L. (1998). Normal reproductive and macrophage Kawabe, T., Muslin, A. J., and Korsmeyer, S. J. (1997). Hox11 function in Pem homeobox gene-deficient mice. Dev. Biol. 202, interacts with protein phosphatases PP2A and PP1 and disrupts a 196–214. G2/M cell-cycle checkpoint. Nature 385, 454–458. Rathjen, P. D., Nichols, J., Toth, S., Edwards, D. R., Heath, J. K., and Keller, G. M. (1995). In vitro differentiation of embryonic stem Smith, A. G. (1990). Developmentally programmed induction of cells. Curr. Opin. Cell Biol. 7, 862–869. differentiation inhibiting activity and the control of stem cell Latimer, J. J., and Pedersen, R. A. (1993). Epigenetic interactions populations. Genes Dev. 4, 2308–2318. and gene expression in pre-implantation mouse embryo develop- Rogers, M. B., Hosler, B. A., and Gudas, L. J. (1991). Specific ment. In “Genes in Mammalian Reproduction,” pp.131–172. expression of a retinoic acid-regulated, zinc-finger gene, Rex-1, in Wiley-Liss, New York. preimplantation embryos, trophoblast and spermatocytes. De- Li, Y., Lemaire, P., and Behringer, R. R. (1997). Esx1, a novel X velopment 113, 815–824. chromosome-linked homeobox gene expressed in mouse ex- Rossant, J., Guilllemot, F., Tanaka, M., Latham, K., Gertenstein, traembryonic tissues and male germ cells. Dev. Biol. 188, 85–95. M., and Nagy, A. (1998). Mash2 is expressed in oogenesis and

preimplantation development but is not required for blastocyst Tybulewicz, V. L. J., Crawford, C. E., Jackson, P. K., Bronson, R. T., formation. Mech. Dev. 73, 183–191. and Mulligan, R. C. (1991). Neonatal lethality and lymphopenia Sanson, B., White, P., and Vincent, J. P. (1996). Uncoupling in mice with a homozygous disruption of the c-abl proto- cadherin-based adhesion from wingless signalling in Drosophila. oncogene. Cell 65, 1153–1163. Nature 383, 627–630. Wilkinson, M. F., Kleeman, J., Richards, J., and MacLeod, C. L. Sasaki, A. W., Doskow, J., MacLeod, C. L., Rogers, M. B., Gudas, (1990). A novel oncofetal gene is expressed in a stage-specific L. J., and Wilkinson, M. F. (1991). The oncofetal gene Pem manner in murine embryonic development. Dev. Biol. 141, encodes a homeodomain and is regulated in primordial and 451–455. pre-muscle stem cells. Mech. Dev. 34, 155–164. Yeom, Y., Fuhrmann, G., Ovitt, C. E., Brehm, A., Ohbo, K., Gross, Schindler, J., Hollingsworth, R., and Coughlin, P. (1984). Inducer- M., Hubner, K., and Scholer, H. R. (1996). Germline regulatory dependent phenotypic divergence in an embryonal-carcinoma element of Oct-4 specific for the totipotent cycle of embryonal cell line. Differentiation 27, 236–242. cells. Development 122, 881–894. Shen, M. M., and Leder, P. (1992). Leukemia inhibitory factor is Yoshida, K., Chamber, I., Nichols, J., Smith, A., Saito, M., Ya- expressed by the preimplantation uterus and selectively blocks sukawa, K., Shoyab, M., Taga, T., and Kishimoto, T. (1994). primitive ectoderm formation in vitro. Proc. Natl. Acad. Sci. Maintenance of the pluripotential phenotype of embryonic stem USA 89, 8240–8244. cells through direct activation of gp130 signalling pathways. Solter, D., and Knowles, B. B. (1982). Monoclonal antibody defining Mech. Dev. 45, 163–171. a stage-specific mouse embryonic antigen (SSEA-1). Proc. Natl. Yu, Y., Li, W., Su, K., Yussa, M., Han, W., Perrimon, N., and Pick, Acad. Sci. USA 75, 5565–5569. L. (1997). The nuclear hormone receptor Ftz-F1 is a cofactor for Srivastava, D., Cserjesi, P., and Olson, E. N. (1995). A subclass of the Drosophila homeodomain protein Ftz. Nature 385, 552–555. bHLH proteins required for cardiac morphogenesis. Science 270, Zhang, H., Hu, G., Wang, H., Sciavolino, P., Iler, N., Shen, M., and 1995–1999. Abate-Shen, C. (1997). Heterodimerization of Msx and Dlx Stevens, L. C. (1958). Studies on transplantable testicular homeoproteins results in functional antagonism. Mol. Cell. Biol. teratomas of strain 129 mice. J. Natl. Cancer Inst. 20, 1257– 17, 2920–2932. 1272. Tanaka, M., Gertsenstein, M., Rossant, J., and Nagy, A. (1997). Received for publication December 11, 1998 Mash2 acts cell autonomously in mouse spongiotrophoblast Revised March 22, 1999 development. Dev. Biol. 190, 55–65. Accepted March 22, 1999