Chick Pcl2 regulates the left-right asymmetry by repressing Shh expression in Hensen's node

The left-right (LR) asymmetry of an embryo, which is characterized by asymmetric structures and/or asymmetric placement of internal organs, is a vital feature of vertebrate embryogenesis. The rightward looping of the developing heart represents the first morphological indication of LR asymmetry. A failure in the normal development of the LR axis is associated with laterality defects, including isomerism, heterotaxia and situs inversus. Recent studies demonstrate that the initial positional information that specifies the LR axis originates in/around the node, a disc-shaped structure at the anterior end of the primitive streak in developing chick and mouse embryos. In the chick, the asymmetric expression of Activin-βB and ActRIIA in the right side of Hensen's node induces the expression of Bmp4, via chick Mid1, which then induces right-sided Fgf8 expression(Levin et al., 1995; Levin et al., 1997; Monsoro-Burg and Le Douarin,2001; Granata and Quaderi,2003). The asymmetrically expressed BMP4 inhibits Shhexpression in the right side of the node, thereby restricting Shhexpression to the left side of the node(Monsoro-Burg and Le Douarin,2001). This asymmetric pattern of gene expression in the node is translated into asymmetric gene expression in the lateral plate mesoderm(LPM), which further regulates the LR axis development. The asymmetrically expressed Shh is responsible for the asymmetric expression of Nodal in the left LPM. Caronte (Car), a member of Cerberus/Dan gene family, was shown to mediate Nodal induction in the left LPM by antagonizing symmetrically expressed BMP activity (Rodriguez-Esteban et al., 1999; Yokouchi et al.,1999; Zhu et al.,1999). However, it has recently been demonstrated that BMPs are also involved in the activation of Nodal in the left LPM through the activation of CFC, a member of the EGF-CFC family, and ActRIIA(Schlange et al., 2001; Schlange et al., 2002; Piedra and Ros, 2002). Meanwhile, Nodal expression around the node is also independently regulated by the Wnt/β-catenin signaling pathway, a protein kinase A-dependent pathway and Notch signaling pathway(Garcia-Castro et al., 2000; Kawakami and Nakanishi, 2001; Rodriguez-Esteban et al.,2001; Raya et al.,2004). Studies in other vertebrate species, including the mouse and Xenopus, indicate that the asymmetrical expression of Nodal in the left LPM is conserved across species and is crucial for establishing initial LR asymmetry(Collignon et al., 1996; Hyatt et al., 1996; Lowe et al., 1996; Sampath et al., 1997). Downstream from this left side signaling cascade reside Nkx3.2 and Pitx2 (Campione et al.,1999; Logan and Tabin,1998; Piedra et al.,1998; Ryan et al.,1998; St Amand et al.,1998; Yoshioka et al.,1998; Schneider et al.,1999). Pitx2, a homeodomain transcription factor, is possibly involved in the morphogenetic execution of the LR asymmetry. In the right side of Hensen's node, FGF8 is required to upregulate SnR expression in the right LPM and to prevent the left-sided pathway from becoming inappropriately activated in the right LPM(Boettger et al., 1999; Isaac et al., 1997). Lefty1, a downstream gene from Shh, is expressed in the prospective floor plate and functions as a midline barrier to prevent induction of Nodal and Pitx2 on the right side(Tsukui et al., 1999). Although significant progress has been made towards the understanding the molecular mechanism of LR axis development in vertebrates(Capdevila et al., 2000), the precise mechanism of how the LR axis is specified and established remains largely elusive. A central puzzle in the existing pathways that regulate LR axis development is how the initial symmetric expression of Shh in Hensen's node later becomes restricted to the left side of the node.

During embryonic development, once cell fate is specified, cell identity is maintained by epigenetic functions. The Trithorax-group (Trx-G) and Polycomb group (PcG) genes are part of the widely conserved cell memory system that maintains both the active and silenced states of transcription patterns (Kennison,1995). The PcG proteins are encoded by about 40 genes in Drosophila, which include Polycomb, Polyhomeotic,Polycomblike (Pcl) and posterior sex comb. PcG mutants exhibit posterior homeotic transformation because of the ectopic expression of the HOM-C genes. Structural homologs of the Drosophila PcG proteins have been identified in mammals, and the mechanisms by which these proteins silence target genes and stabilize developmental decisions are likely to be conserved between Drosophila and vertebrates. Similar to Drosophila, mutations in different mammalian PcG genes cause posterior axial transformations of the mouse skeleton and anterior shifts of Hox expression boundaries (Akasaka et al.,1996; Schumacher et al.,1996; Coré et al.,1997; Takihara et al.,1997; Suzuki et al.,2002). In addition to the regulation of Hox gene expression, PcG gene members have been shown recently to regulate developmentally important genes, such as hedgehog(Maurange and Paro, 2002), and cell cycle regulation genes, such as Rb(Dahiya et al., 2001). In Drosophila, Polycomblike (Pcl), a member of the PcG, plays an important and perhaps central role in PcG function(Landecker et al., 1994). The Pcl protein contains two Cys₄-His-Cys₃ motifs, known as the plant homeodomain (PHD) type zinc finger(Aasland et al., 1995). The PHD motif, also called the leukemia-associated protein domain, is found in more than 400 eukaryotic proteins. Most of the PHD domain proteins are thought to be involved in transcription regulation, possibly acting through chromatin remodeling and histone acetylation (Aasland et al., 1995; Yochum and Ayer,2001; Kalkhoven et al.,2002). More specifically the PHD finger appears to act as a protein-protein interaction domain to mediate the regulation of gene expression (Jacobson and Pillus,1999). Although the function of the Pcl genes in vertebrate development remains unclear, studies in Xenopus indicate that they negatively regulate, or repress, gene expression in the developing anterior central nervous system (Yoshitake et al.,1999; Kitaguchi et al.,2001).

In this report, we show that a novel chick Polycomblike gene,chick Pcl2, encodes a transcription repressor and exhibits asymmetric expression in the right side of Hensen's node. Protein-soaked bead implantation studies indicate that chick Plc2 resides downstream of Activin-βB and Bmp4 in the node. Inhibition of chick Pcl2 expression in the early embryos led to randomization of cardiac looping direction. Using gain-of-function and loss-of-function approaches, we demonstrate that chick Pcl2 is both necessary and sufficient for the repression of Shh expression in the node. The repression of Shh expression by chick Pcl2 seems to be conserved in other developing organs and even across species. Overexpression of chick Pcl2 by RCAS retroviral infection in the developing chick limb bud and feather bud inhibited Shh expression in the ZPA of the limb bud and the epithelia of the feather bud. Transgenic overexpression of chick Pcl2 in the mouse limb bud also inhibited Shhexpression in the ZPA. We further demonstrated that chick Pcl2 can repress the activities of the mouse Shh promoter in cell culture assays. Pull-down assays indicate that chick Pcl2 might function as a repressor by recruiting EZH2 via its PHD finger domain. These results indicate that chick Pcl2 plays an essential role in the LR axis specification by silencing Shh expression in the node.

A differential gene expression screen was performed using the PCR-Select cDNA subtraction kit from Clontech (Palo Alto, CA), according to the manufacturer's protocol, using mRNAs extracted from the left and right halves of about 600 stage 5-9 chick embryos(Hamburger and Hamilton,1951). A 280 bp chick Pcl2 cDNA fragment was obtained and labeled with ³²P by random priming using Red Primer II labeling system (Amersham Pharmacia Biotech) to screen a stage 23 chick embryonic cDNA library at high stringency as described previously(St Amand et al., 1998). Two independent clones were obtained and sequenced. One contained a 2299 bp open reading frame encoding a putative 595 amino acid residue protein and was named chick Pcl2. The sequence was deposited into Genbank (Accession Number AY251284).

For the detection of chick Pcl2 expression, the full-length chick Pcl2 containing plasmid was linearized with XhoI. Non-radioactive RNA probes were generated and labeled with digoxigenin (DIG)using T3 RNA Polymerase. Other cDNAs used for gene expression studies include:a 1.1 kb cDNAs of the chick Nodal(Levin et al., 1995), a 0.5 kb Shh cDNA (Ogura et al.,1996), a 0.9 kb Caronte cDNA(Rodriguez-Esteban et al.,1999), a 1.8 kb Pitx2 cDNA(St Amand et al., 1998), a 1.9 kb SnR cDNA (Isaac et al.,1997), a 0.8 kb Fgf8 cDNA(Meyers and Martin, 1999), a 1.4 kb Lefty1 cDNA (Schlange et al., 2001) and a 0.6 kb mouse Shh cDNA(Echelard et al., 1993). DIG-labeled riboprobes were generated according to the manufacturer's instruction (Boehringer Mannheim, Indianapolis, IN). Whole-mount and section in situ hybridization analyses were performed as described previously(St Amand et al., 1998; St Amand et al., 2000). Briefly, samples were fixed in freshly made 4% paraformaldehyde/PBS at 4°C overnight. For whole-mount in situ hybridization, samples were bleached with 6% H₂O₂ prior to dehydration through a graded methanol series. For section in situ hybridization, samples were dehydrated through a graded ethanol series, embedded in paraffin wax and sectioned at 10 μm. Hybridization signals were visualized using the BM purple AP substrate at 4°C.

Bead implantation experiments were performed on stage 4 chick embryos explanted in New cultures (New,1955). Briefly, Affigel-Blue beads (BioRad) were soaked in Activin-A (500 ng/μl in PBS), Follistatin (500 ng/μl in PBS), BMP4 (1μg/μl in PBS) or 1% BSA in PBS (as a control), and were implanted on the left or right side of Hensen's node. All proteins were purchased from R&D Systems (Minneapolis, MN). Embryos were cultured to desired stages and collected for whole-mount in situ hybridization analyses. Oligonucleotide treatment was performed on stage 4-6 chick embryos explanted in New culture as described previously (Isaac et al.,1997; Srivastava et al.,1995; Yu et al.,2001). The sequence for the 20 base antisense oligonuleotide that targets the first PHD domain of chick Pcl2 is 5′-CTCCTCCTGACATATTGTAC-3′. The sequence for the random control oligonucleotide is 5′-GACTATCTAGATAGCTACGT-3′. The oligonuleotides were synthesized as phosphorothioate derivatives and were purified by HPLC(IDT, Corralville, IA). About 10 μl oligonucleotide mixed with lipofectAMINE (GibcoBRL) at a concentration of 40 μM was applied onto cultured embryos constrained by a plastic ring in New cultures. Embryos were collected at appropriate stages for analysis of gene expression and for scoring the direction of cardiac looping.

We used microelectroporation to transfer plasmid DNA into early chick embryonic tissues. To generate expression plasmids, the coding region of chick Pcl2 (amino acids 1-595) was amplified from the full-length chick Pcl2 cDNA plasmid using Pfu DNA polymerase (Stratagene, La Jolla, CA) and was cloned into the pMES expression vector in front of IRES-Gfp-coding sequence. The resulting plasmid, pMES-Pcl2,expresses both the transgene and Gfp simultaneously under the control of the chick β-actin promoter. To perform microelectroporation,a gold-plated cathode was fixed to the bottom of a 60 cm dish with a thin layer of Ringer's saline. Stage 4 chick embryos were prepared for New culture and embryo constrained by a ring was placed onto the cathode. About 1 μl of expression plasmids (1 μg/μl), mixed with fast green to visualize DNA,was injected to the target site between the blastoderm and the vitelline membrane using a glass capillary. A platinum anode (0.5 mm in diameter) was then placed on the hypoblast of the node. The distance between the two electrodes was maintained within 2 mm. Electroporation was performed using BTX electroporator (Electro Square Porator™ ECM 830 Model, BTX, San Diego,CA) with five pulses of 5 V for duration of 25 mseconds and intervals of 454 mseconds. Embryos were then placed onto the agar media in New cultures and were incubated at 38°C to the appropriate stages. Gfp expression in targeted tissues was monitored before embryos were harvested for gene expression analysis.

RCAS retroviral construction and infection was performed as described previously (Yu et al., 2001). To make the RCAS-Pcl2 construct, the coding region of chick Pcl2 (amino acids 1-595) was amplified from chick Pcl2 cDNA plasmid and cloned into the Cla12 vector. The insert was then released by ClaI digestion and cloned into the RCASBP retroviral vector. Chicken embryonic fibroblast (CEF) cells expressing RCAS-Pcl2or RCAS-Gfp were pelleted according to a protocol described previously (Logan and Francis-West,1999). To infect the chick developing limb bud, virus-free chick eggs (CBT farms, Chestertown, MD) were incubated to approximately stage 12. RCAS-Pcl2-expressing CEF cells were centrifuged briefly and incubated as a pellet for 2 hours at 37°C to allow tight cell aggregates to form. Cell pellets were cut into small pieces and were implanted into the right LPM of the prospective forelimb-forming region of chick embryos as described previously (Logan and Tabin,1998). The infected embryos were then cultured in ovo and harvested from stage 19 to stage 24 for morphological and gene expression analyses. For infection of feather buds, dorsal skin tissues from stage 31 chick embryos were dissected in PBS and transferred to Trowell type organ cultures in DMEM supplemented with 10% fetal calf serum. About 10 μl RCAS retroviruses were injected with a microcapillary needle at multiple sites in the dorsal skin explants. The infected explants were incubated at 37°C in a 5% CO₂ atmosphere for 4 days, and were then collected for gene expression assays.

Construction of the chick Pcl2 transgenic construct and pronuclear injection were performed as described previously(Zhang et al., 2000). Briefly,the chick Pcl2 full-length cDNA was cloned into the PCI mammalian expression vector (Promega, Madison, WI). The chick Pcl2 full-length cDNA flanked by a chimeric intron and the SV40 late Poly(A) sequence was released by PstI/BamHI, This fragment was cloned downstream of the 3.6 kb Hoxb6 promoter(Schughart et al., 1991). The orientation of the insert was confirmed by restriction digestion and sequencing. Preparation of DNA fragments for injection, collection of zygotes and pronuclear injection was carried out as described previously(Hogan et al., 1994; Zhang et al., 2000). Embryos were collected at E10.75 for analysis of transgene expression and Shhexpression. The integration of the Hoxb6-Pcl2 transgene was determined by PCR using genomic DNA from the head of each embryo. The PCR primers used are as follows: 5′-TTTTGGTGCAGCAGGTAGAATAGC-3′ (upper primer) and 5′-CTCCCCCTGAACCTGAAACATAAA-3′ (lower primer).

To map the repressive domain of chick Pcl2 protein, the DNA fragments encoding the N terminus (amino acids 1-329), the C terminus (amino acids 331-595) and the PHD fingers domain (amino acids 103-238) of chick Pcl2 were amplified from the full-length chick Pcl2 cDNA plasmid and cloned into the pBXG1 vector that contains the GAL4 DNA-binding domain under the control of SV40 enhancer/promoter(Lillie and Green, 1989). The derived constructs, named GAL4/Pcl2-N, GAL4/Pcl2-C and GAL4/PHD, were confirmed by sequencing. The reporter plasmid, pG5tkCAT, contains the chloramphenicol acetyltransferase(CAT) reporter gene directed by the herpes simplex virus thymidine kinase (TK)promoter with five GAL4 binding sites upstream of the TATA-box. Transfection and CAT assays were carried out in cultured P19 cells, as described previously(Yu et al., 2001). A CMV-β-gal plasmid was included as an internal control for transfection efficiency. Transfected cells were cultured for 36 hours and then CAT activities were determined by thin layer chromatography (TLC) and scintillation counting. Each experiment was repeated at least three times to ensure consistent results.

A 1 kb and a 3 kb mouse Shh upstream regions, were amplified using EXL™ Taq (Stratagene, La Jolla, CA) from the mouse genomic clone (RPCI RP23 429M20), and cloned as BamHI-XhoI fragments to replace the TK promoter in pG5tkCAT vector, respectively. The 1 kb Shh upstream region did not give reasonable promoter activities (data not shown). Only the plasmid containing the 3 kb Shh upstream region was used in this study. The pShh-CAT plasmids were cotransfected with pMES-Pcl2, pMES-PHD or pMES (as a control) into P19 cells. Transfection efficiency was monitored by inclusion of CMV-β-gal plasmid. Transfected cells were cultured for 36 hours prior to CAT assays by both TLC and scintillation counting, which were normalized by protein concentrations. Again, each experiment was carried at least three times.

To make the constructs for co-immunoprecipitation, the chick Pcl2full-length cDNA and chick Pcl2 PHD finger domains (amino acids 103-238) were cloned in frame into the pIRES-hrGFP-1α vector(Stratagene, La Jolla, CA), while the mouse EZH2 sequence was amplified by PCR from the mouse EZH2 cDNA and cloned into the pCMV-Tag 3A vector (Stratagene). The resultant expression plasmids, pFLAG-Pcl2, pFLAG-PHD and pMyc-EZH2, were transiently transfected or co-transfected into 293T cells with Lipofectamine™ 2000 (Invitrogen, Carlsbad, CA). Cells were harvested 48 hours after transfection in a lysis buffer consisting of 0.05 M HEPES, 1%Triton X-100, 0.15 M NaCl, 10% glycerol, 1 mM EDTA, 1.5 mM MgCl₂·6H₂O, Trypsin inhibitor 0.01 mg/ml,aprotinin 0.01 mg/ml, phenylmethanesufonyl fluoride 1 mM, leupeptin 0.02 mg/ml, sodium orthovanadate 1.25 mM and sodium fluoride 0.1 mM. Lysates were incubated with either a monoclonal anti-Myc antibody (Abcam, Cambridge, MA) or an anti-FLAG M2 antibody (Stratagene), and then with Protein A Sepharose beads(Amersham Pharmacia, Piscataway, NJ). After extensive washing, the eluted proteins were resolved on PAGE gels, and immunoblotted with an anti-FLAG M2 antibody or a monoclonal anti-Myc antibody, and the binding was detected using the ECL™ plus Western Blotting Detection system (Amersham Biosciences).

In an effort to identify new components in the genetic pathways regulating LR axis development, a PCR-select cDNA subtraction hybridization screen was performed using mRNAs extracted from the left and right halves of stage 5-9 chick embryos (Hamburger and Hamilton,1951). Thirty-four independent cDNA fragments varying from 200-800 bp in length were isolated. Nineteen of these clones exhibited a certain level of homology (higher than 60%) to known genes, while 15 clones showed no homology to known genes. A second round of screening by whole-mount in situ hybridization was then conducted to examine the expression pattern of these clones. Among 10 clones checked, two clones show asymmetric expression in the Hensen's node (data not shown). We focused on one clone that exhibited a high level of homology to the human Polycomblike (PCL) gene. A stage 23, chick whole embryo cDNA library was screened with this chick Pcl fragment as described previously(St Amand et al., 1998). A 2299 bp nucleotide acid cDNA encoding a putative protein of 595 amino acid residue was obtained. The putative amino acid sequence of the chick Pcl is most homologous to the Xenopus Pcl2 when compared with other Pcl family members. It was therefore named chick Pcl2 (chick Pcl2). Similar to other Pcl2 proteins, the predicted amino acid sequence of chick Pcl2 contains two PHD fingers, a well characterized feature of the PcG proteins(Aasland et al., 1995) and a Tudor domain (Ponting, 1997)(Fig. 1A). The amino acid sequence of chick Pcl2 exhibits 86% and 87% identity to the human and mouse putative Pcl2 (M96) proteins, respectively (ID, XP_002013.5; ID, XP_132195.1);and 74% identity with the Xenopus Pcl2 protein (ID, BAB43943.1). Phylogenetic tree analysis (Gap penalty, 10; Gap length penalty, 10) clearly groups chick Pcl2 into a cluster distinct from the Pcl1 proteins(Fig. 1B). This analysis indicates that rather than falling into the Pcl1 group, the Drosophila Pcl protein which contains two PHD fingers more resembles Pcl2 subfamily. Pcl2 proteins are evolved earlier and are more conserved than Pcl1 proteins. We propose a conserved function for Pcl2 among different species.

Using the full-length chick Pcl2 cDNA as a probe, the expression of chick Pcl2 was examined during early embryonic development by whole-mount in situ hybridization. At HH stage 4, chick Pcl2expression was found mainly in the posterior one-third of the primitive streak, and was not expressed in Hensen's node(Fig. 2A). Beginning at HH stage 5, chick Pcl2 expression was found to shift the anterior region of the primitive streak, and was present asymmetrically in the ectoderm of the right side of Hensen's node (Fig. 2B,C). This asymmetric pattern of chick Pcl2 expression is complementary to that of Shh, which initially appears symmetrically in the Hensen's node, but becomes restricted to the left side of Hensen's node beginning at stage 5 (Levin et al., 1995). Chick Pcl2 expression remained restricted in the right side of Hensen's node until HH stage 8(Fig. 2D). Activin signaling,which is mediated by Activin-βB, has been known to set up the asymmetric pattern of Shh in the node, by repressing Shh on the right side of the node (Levin et al.,1997). Because of the PcG family of proteins act as transcriptional repressors, it is possible that chick Pcl2 could act downstream of Activin signals to regulate gene expression in the node. To begin to test this hypothesis, we asked whether Activin-βB regulates chick Pcl2 expression in the right side of Hensen's node. Activin-A-soaked beads (500 ng/μl) were implanted on the left side of Hensen's node at stage 4. Six hours after bead implantation, embryos were processed for in situ hybridization. The activity of the Activin-A in these assays was confirmed by the repression of Shh in the left side of Hensen's node (data not shown). Activin-A beads induced ectopic chick Pcl2 expression in the left side of Hensen's node (7/12; Fig. 2E). Control beads (soaked in BSA) had no effect on the expression of either chick Pcl2(Fig. 2H) or Shh (data not shown). To test whether an Activin-like signal is necessary for the restriction of chick Pcl2 expression in the right side of Hensen's node, beads soaked with Follistatin, an antagonist of Activin and BMP signaling (Hemmati-Brivanlou et al.,1994; Yamashita et al.,1995; Iemura et al.,1998), were implanted into the right side of Hensen's node at stage 4. chick Pcl2 expression was repressed not only on the right side of node (6/11) but also in the primitive streak (3/10) following Follistatin treatment (Fig. 2F). The effect of BMP4 signaling on chick Pcl2expression was also examined. Similar to Activin-A, BMP4-soaked beads were able to induce ectopic chick Pcl2 expression in the left side of Hensen's node and the primitive streak (6/7; Fig. 2G). Taken together, the data indicate that chick Pcl2 is asymmetrically expressed in Hensen's node during developmental stages that are crucial for LR axis development, and both Activin and BMP4 can positively regulate chick Pcl2expression.

To study the effect of a gain-of-function of chick Pcl2, we used microelectroporation to deliver this gene to chick embryos in New cultures. Although retrovirally mediated gene delivery has been routinely used to ectopically express target genes in chick embryos, the delay of target gene expression after viral infection makes it inappropriate for use in this study. Electroporation is a rapid and efficient strategy to introduce ectopic gene expression (Yasuda et al.,2000; Uchikawa et al.,2003). We adopted and modified a microelectroporation method(Momose et al., 1999) to introduce ectopic gene expression in early chick embryos in New cultures. By optimizing the electrode size, DNA concentration and injection method, as well as electroporation parameters, we detected strong expression of the target gene within 6 hours after electroporation with minimal side effects to the embryos (Fig. 3A,B). In these assays, pMES-Pcl2 or pMES-Gfp control vectors were targeted to Hensen's node, and embryos were placed in New cultures until the desired stages of development were reached. Whole-mount in situ hybridization was performed using riboprobes for Shh, Nodal, chick SnR, Pitx2and Lefty1 genes that are known to be involved in LR axis determination. Ectopic chick Pcl2 expression to the left side of Hensen's node significantly repressed Shh expression in the node(10/12, Fig. 3D), expression of its downstream genes Nodal (5/9, Fig. 3F) and Pitx2(7/16, Fig. 3H) in the left LPM, and expression of Lefty1 (5/6, Fig. 3J) in the left prospective floor plate. However, chick SnR, a gene known to be negatively regulated by Nodal (Patel et al., 1999), was ectopically expressed in the left LPM (4/10, Fig. 3L). In the control embryos targeted with the pMES-Gfp vector, the expression patterns of these genes were unaltered (Fig. 3C,E,G,I,K). These results indicate that chick Pcl2expression is sufficient to repress Shh in the node. The repression of Shh by chick Pcl2 ectopic expression was also seen in the notochord (Fig. 3D), suggesting that chick Pcl2 could repress Shh in other developing organs and tissues.

To evaluate the possibility that chick Pcl2 might repress Shh expression in other tissues, we first analyzed chick Pcl2 expression in several developing chick organs where the expression of Shh has been well characterized. In these assays that focused on the limb and feather buds, chick Pcl2 showed a dynamic expression patterns. chick Pcl2 was found expressed in the medial limb mesenchyme at HH stage 18 (Fig. 4A), when Shh is barely detectable(Pearse et al., 2001). Chick Pcl2 expression remained in the medial mesenchyme at the time when Shh expression is initiated in the posterior mesenchyme to mediate the polarizing activity of the ZPA (Riddle et al., 1993). By HH stage 24, the chick Pcl2 expression domain split into two regions in the dorsal and ventral limb mesenchyme, and overlapped with regions of muscle mass (data not shown). These results indicated that, although chick Pcl2 expression is not completely complementary to Shh, it does not overlap with the Shh-expressing domain in the developing limb bud. In developing feathers, Shh expression was first detected at the center of the epithelia placode (Fig. 4D). In the long feather bud, Shh expression is known to shift to the distal and posterior side of the feather epithelia(Ting-Berreth and Chuong,1996). Interestingly, chick Pcl2 shows a complementary expression pattern to that of Shh in the feather bud. Chick Pcl2 expression was first detected in the feather placode mesenchyme in the middle region, and then became more strongly expressed in the posterior mesenchyme in the feather bud (Fig. 4C). To test the effect of ectopic chick Pcl2 on Shh expression in these organs, we proceeded to misexpress chick Pcl2 to the developing limb bud and feather bud using RCAS-Pcl2 retroviruses. Infection efficiency was determined by whole-mount in situ hybridization using a probe against RCAS gag RNA sequences (data not shown). Overexpression of chick Pcl2 in these assays caused an unambiguous downregulation of Shh expression in the ZPA of infected limb buds (Fig. 4B), and in feather buds (Fig. 4E). In controls, infection of either limb or feather buds with the RCAS-Gfp control retroviruses did not alter Shhexpression (data not shown). These results indicate that ectopic expression of chick Pcl2 indeed is able to repress Shh expression in the limb and feather buds.

We further asked whether repression of Shh by chick Pcl2is also conserved across different species. A transgenic construct containing the full-length chick Pcl2 cDNA under the control of the mouse Hoxb6 promoter was generated. The 3.6 kb mouse Hoxb6promoter drives transgene expression specifically to the posterior mesenchyme of the forelimb and in all mesenchymal tissue in the hindlimb of mouse embryos at E9.5 and E10.5 when Shh is expressed in the ZPA(Schughart et al., 1991; Zhang et al., 2000). Hoxb6-Pcl2 transgenic mice were generated by pronuclear injection. Out of 28 embryos collected on E10.5-10.75, five were positive for transgenicity. Mid-sagittal cut was made to separate these embryos into left and right half embryos that were processed for whole-mount in situ hybridization with chick Pcl2 probe for the transgene expression and the mouse Shh probe for Shh expression in the ZPA,respectively. Two embryos that showed the transgene expression in a correct pattern (Fig. 4G) exhibited strong repression of Shh expression in the ZPA of both the forelimb and hindlimb (Fig. 4I). Non-transgenic littermates showed normal Shh expression(Fig. 4H).

Our results established that chick Pcl2 is sufficient for Shh repression in the node and in other developing organs. We further examined the function of chick Pcl2 in the development of chick LR axis by a loss-of-function approach using antisense oliogonucleotides(Srivastava et al., 1995; Yu et al., 2001). A 20 base antisense oligonucleotide targeting to the first PHD domain of chick Pcl2 and random control oligonucleotides were synthesized as phosphorothioate derivatives and were purified by HPLC. Stage 4 chick embryos were treated with oligonucleotides in New cultures as described previously(Yu et al., 2001), and were allowed to develop to stage 11 for an examination of heart looping direction. In the studies, 46% (12/26) of embryos treated with antisense oligonucleotides to chick Pcl2 at stage 4 exhibited reversed cardiac looping(Fig. 5B), while only 8% (1/12)of embryos treated with control oligonucleotides exhibited a reversed cardiac looping (Fig. 5A). However,when antisense oligonucleotides were applied at stage 5, only 39% (7/18) of embryos showed reverse cardiac looping; and when applied at stage 6, only 14%(1/7) showed that phenotype. This stage-dependent effect of antisense oligonucleotides treatment is consistent with the timing of the asymmetrical chick Pcl2 expression in the node. Our results indicate that chick Pcl2 plays a crucial role in the LR axis development in the chick.

We next asked whether the inhibition of chick Pcl2 expression led to randomization of heart looping by causing ectopic Shh expression in the node and its downstream genes in the right LPM. Embryos treated with oligonucleotides at stage 4 were allowed to develop to stage 6 for the examination of Shh, Fgf8 and chick Pcl2, and to stage 8-10 for other genes. chick Pcl2 expression was found significantly reduced in the chick Pcl2 antisense oligonucleotide treated embryos(Fig. 5C), indicating the efficiency of antisense treatment. In antisense-treated embryos, Shhexpression was seen in the right side of the node in about 30% (4/14) of embryos (Fig. 5G). Furthermore,ectopic expression of Caronte (2/8, Fig. 5I), Nodal (3/11, Fig. 5K) and Pitx2(3/9, Fig. 5M) was also observed in the right LPM, possibly resulting from ectopic Shhexpression in the right side of the node. However, Fgf8 (8/9)expression was repressed in the right side of the node following antisense treatment (Fig. 5D,E). In control embryos treated with random oligonucleotides, the expression of all these genes appeared normal (Fig. 5F,H,J,L). These results established a crucial role for chick Pcl2 in the LR axis development by repressing Shh in the right side of the node.

Products of PcG family genes in Drosophila are required for the epigenetic repression of homeotic genes and other key developmental regulatory genes such as hedgehog, through direct transcription effects(Kennison, 1995; Maschat et al., 1998; Maurange and Paro, 2002). In addition, overexpression of the Pcl subfamily members has been shown to repress target gene expression in vertebrates(Yoshitake et al., 1999; Kitaguchi et al., 2001). We next performed in vitro assays to map the repression domain of the chick Pcl2 protein. Constructs containing DNA fragments encoding the chick Pcl2N terminus (amino acids 1-329), C terminus (amino acids 331-595), and the region encompassing the two PHD finger domains (amino acids 103-238) which mediates the association of Drosophila Pcl with ESC/E(Z) repression complexes (O'Connell et al.,2001), fused in frame to the 147 amino acid yeast GAL4 DNA-binding domain were generated. A construct expressing the N terminus of the chick Pitx2a transcription activator, which was previously shown to have no effect on reporter gene expression (Yu et al.,2001), was included as control. The transcription activity of the fusion proteins was assessed on the GAL4-responsive pG5tkCATreporter. These assays showed that the fused protein containing the PHD finger domains (GAL4-Pcl2^aa103-238) significantly repressed reporter gene expression (Fig. 6A). GAL4-Pcl2N which contains the PHD domains also exhibited a repressive effect on reporter gene expression. The GAL4-Pcl2C had no repressive effect and neither did the control vector GAL4-Pitx2aN(Fig. 6A). It was therefore concluded that chick Pcl2 indeed encodes for a transcription repressor and the repression is mediated by PHD finger domain.

We further asked whether chick Pcl2 represses Shhexpression by acting directly on the Shh promoter. As transgenic expression of chick Pcl2 in the mouse developing limb bud repressed Shh expression and because of the availability of mouse genome information, we decided to use the mouse Shh upstream region in this study. It has been shown previously that a 1 kb mouse Shh upstream region was able to drive reporter gene expression in cell culture(Knezevic et al., 1997). We cloned this 1 kb mouse Shh upstream region and found that this genomic fragment could only drive weak reporter gene expression in P19 cells and in the MPLB-2, a mouse embryonic limb bud cell line(Trevino et al., 1993), making it inappropriate for this study. A bioinformatic approach using rVISTA software was subsequently performed to search for potential PcG protein-binding sites in the 10 kb upstream sequences of the mouse Shh gene (Loots et al.,2002), multiple YY1 (the vertebrate homolog of DrosophilaPcG pleiohomeotic gene) binding sites were found within 3 kb upstream of the mouse Shh gene. Accordingly, this 3 kb mouse Shhupstream region was PCR amplified and linked with the CAT reporter gene,generating the Shh-CAT construct. CAT assays revealed that this 3 kb Shh upstream region drove strong reporter gene expression in P19 cells (Fig. 6B). P19 cells were co-tranfected with expression vectors expressing the full-length chick Pcl2 or the PHD domain construct and the Shh-CAT construct. Shhpromoter activity was significantly reduced (2.4-fold) by the full-length chick Pcl2 construct, but was not affected by the PHD domain construct, when compared with controls (Fig. 6B). Neither the chick Pcl2 nor the PHD domain construct repressed activity of the TK promoter (Fig. 6B). Therefore, we conclude that chick Pcl2 can specifically repress Shh promoter activity, but this repression requires chick Pcl2 protein integrity.

Generally, PcG proteins function by forming protein complexes with other proteins. It was shown that the PHD fingers from the Drosophila Pcl protein interact directly with E(Z) (Enhancer of Zeste), a component of the ESC/E(Z) repressive complex (O'Connell et al., 2001). This interaction between Pcl and E(Z) is also conserved in their human homologs(O'Connell et al., 2001). To test whether chick Pcl2 might also interact with EZH2, the vertebrate homolog of the E(Z) protein, co-immunoprecipitation studies were performed. FLAG-tagged chick Pcl2 (FLAG-Pcl2) or PHD fingers (amino acids 103-238;FLAG-PHD) and Myc-tagged mouse EZH2 (Myc-EZH2) were co-expressed in the 293T cells. As shown in Fig. 7, the PHD fingers and EZH2 proteins were steadily pulled down by antibodies against FLAG or Myc reciprocally, confirming an interaction between EZH2 and the PHD fingers of chick Pcl2. By contrast, chick Pcl2 proteins failed to pull down EZH2 reciprocally using the same approach(Fig. 7). This also happened to the Drosophila Pcl fusion protein which was unable to but its PHD finger domain could interact with E(Z)(O'Connell et al., 2001). This could be explained by the mask of the PHD finger domain in chick Pcl2 fusion protein or instability of chick Pcl2-EZH2 association.

Left-sided Shh expression in Hensen's node is crucial for setting up LR asymmetry in the chick (Levin et al., 1995; Pagan-Westphal and Tabin, 1998). This is achieved by the repressive action of the asymmetrically expressed Activin-βB and Bmp4in the right side of the node, which restricts Shh to the left side of the node. However, little is known about the transcriptional repressor that mediates Shh inactivation in the right side of the node. In this study, we have provided evidence that a novel chick Polycomblikegene, chick Pcl2, which exhibits right-sided expression in the node,is an excellent candidate for this repression. Our data demonstrate that chick Pcl2 encodes a transcription repressor that is both sufficient and necessary for the repression of Shh in the right side of Hensen's node and that the repression of Shh by chick Pcl2 is conserved in different organs and even across species. Furthermore, chick Pcl2 can repress the mouse Shh promoter activity and probably interacts with EZH2 via the PHD fingers. Based on these results, we propose that chick Pcl2 plays a crucial role in regulating LR axis patterning by silencing Shh in the right side of the node.

Hensen's node, the chick organizer, has been a focus in the study of the molecular mechanism that establishes LR asymmetry. The chick organizer shows striking asymmetries well before the detection of any overt morphological signs of asymmetry. For example, the morphological asymmetry is apparent at the chick node as early as stage 4 (Cooke,1995; Dathe et al.,2002). At the molecular level, a number of genes have been found to be expressed in or near the chick node in an asymmetric fashion. These include Activin βB and its receptor cAct-RIIa(Levin et al., 1995; Levin et al., 1997), Bmp4 (Monsoro-Burq and Le Douarin, 2001), Fgf8(Boettger et al., 1999), N-Cadherin (Garcia-Castro et al.,2000) and chick Mid1(Granata and Quaderi, 2003) in the right side of the node, and Shh(Levin et al., 1995) and Nodal (Levin et al.,1995) in or near the left side of the node. Shh is initially expressed symmetrically in the node and becomes restricted to the left side of the node at stage 5 by an inferred Activin-like signal, possibly Activin-βB. This asymmetrical Shh in Hensen's node is both necessary and sufficient for the left-sided expression of Nodal in the LPM which is conserved in vertebrates(Levin et al., 1995; Pagan-Westphal and Tabin,1998). Interestingly, asymmetric expression of Shh in the node is not observed in mice, zebrafish or rabbit(Levin et al., 1995; Fischer et al., 2002), arguing against the conserved mechanism of the initial symmetry-breaking events. However, Shh mutant mice do display a variety of laterality defects(Tsukui et al., 1999). It is known that the antagonistic interactions between Shh and Bmp4 maintain the restricted Shh expression in the left side of Hensen's node (Monsoro-Burg et al., 2001). Recently, chick Mid1, a microtubule-associated ubiquitin ligase, was shown to act upstream of Bmp4 to mediate the antagonistic interaction (Granata et al., 2003). It is reasonable to suspect that a putative transcription repressor is expressed in the right side of node to mediate the repression of Shhby the Activin-like signals. We identified such repressor, a novel chick Polycomblike gene, chick Pcl2, which acts downstream of Activin and Bmp4 and is expressed in the right side of Hensen's node. In the node region, chick Pcl2 is expressed in the ectoderm as is Shh but in a complementary pattern. Chick Pcl2 encodes a transcription repressor which specifically represses the activity of the Shh promoter. These studies establish chick Pcl2 as a candidate for the putative transcriptional repressor of Shh in the node. In support of this hypothesis, ectopic expression of chick Pcl2in the left side of Hensen's node abolished Shh expression in the node and blocked the expression of its downstream genes in the LPM and the midline. Ablating chick Pcl2 expression by an antisense oligonucleotide approach caused ectopic Shh expression in the right side of Hensen's node, which in turn randomized heart looping. These data indicate that chick Pcl2 is both sufficient and necessary for the repression of Shh in the right side of the node. In the chick, the LR identity is liable at stage 4, but becomes fixed at stage 5, concurrent with the right-sided expression of chick Pcl2 in the node. Chick Pcl2 seems to participate in stabilizing developmental decision that establish the LR asymmetry.

Shh, when acting as a morphogen, is essential for crucial pathways that regulate the differentiation and patterning of a number of tissues. In our studies, overexpression of chick Pcl2 by electroporation to the node of stage 4 chick embryos repressed Shh expression in the node and also in the notochord, a node-derived tissue. Retrovirus-mediated ectopic expression of chick Pcl2 also repressed Shh expression in the developing chick limb bud and feather bud. Moreover, transgenic overexpression of chick Pcl2 in the mouse limb bud similarly downregulated Shh expression. These consistent results indicate that the repression of Shh by chick Pcl2 may represent a general regulatory mechanism for controlling Shh expression in different organs and tissues in the chick and even across species. The repression of Shh by chick Pcl2 seems to be a specific rather than a general repressive effect, because chick Pcl2 represses the mouse Shh promoter activity but not the TK promoter in vitro.

The repressive effect of chick Pcl2 on Shh expression suggests that the products of PcG genes not only maintain the silenced status of gene expression involved in long-term developmental decisions, but also function to regulate/silence the expression of active gene during embryonic development. This hypothesis is supported by the finding that in Drosophila the product of Polyhomeotic (ph) exerts negative transcriptional control on active genes (Randsholt et al.,2000). Furthermore, the Xenopus Pcl genes were also shown to repress gene expression in the developing central nervous system(Yoshitake et al., 1999; Kitaguchi et al., 2001). It was recently shown that the antagonistic functions of the Polycomb group complex and Trithorax complex on a Polycomb response element can govern the transition between the repressed and active status of gene expression(Poux et al., 2002).

Our results demonstrate that chick Pcl2 encodes a transcription repressor, and its repression domain was mapped to the PHD fingers. We demonstrated that the PHD fingers of chick Pcl2 can interact directly with EZH2, a finding that is consistent with previous results showing that PHD fingers from both Drosophila and human Pcl bind with high specificity with EZ in ESC/E(Z) complexes (O'Connell et al., 2001). Drosophila Pcl was also shown to be a component of ESC/E(Z) complexes, which contains histone deacetylase and histone methyltransferase activities (Tie et al., 2003; Kuzmichev et al., 2002; Muller et al.,2002). These enzymatic activities contribute to chromatin remodeling and transcriptional repression(Zhang and Reinberg, 2001). Based on our evidence showing that chick Pcl2 can specifically repress Shh promoter activity in vitro and Shh expression in developing organs, we propose that chick Pcl2 may repress Shh by recruiting the EED/EHZ2 complex [the mammalian homolog of ESC/E(Z)] through the conserved interaction between the PHD fingers and EZH2. The potential DNA-binding domain in chick Pcl2 and chick Pcl2 response element remain to be identified.

We thank Drs Judith Venuti and Michael Lan of Louisiana State University Health Sciences Center for reagent, and members of the Chen laboratory for discussion and technical support. We are also indebted to Dr Carol Bursdal of Tulane University for her critical reading of the manuscript. This work was supported by an Established Investigator Award (0340166N) from the American Heart Association and partially by grants from the NIH to Y.P.C.