|Home | About | Journals | Submit | Contact Us | Français|
In zebrafish, the endoderm originates at the blastula stage from the most marginal blastomeres. Through a series of complex morphogenetic movements and differentiation events, the endodermal germ layer gives rise to the epithelial lining of the digestive tract as well as its associated organs such as the liver, pancreas and swim bladder. How endodermal cells differentiate into distinct cell types such as hepatocytes or endocrine and exocrine pancreatic cells remains a major question. In a forward genetic screen for genes regulating endodermal organ development, we identified mutations at the shiri locus that cause defects in the development of a number of endodermal organs including the liver and pancreas. Detailed phenotypic analyses indicate that these defects are partially due to a reduction in endodermal expression of the hairy/enhancer of split-related gene, her5, at mid to late gastrulation stages. Using the Tg(0.7her5:EGFP)ne2067 line, we show that her5 is expressed in the endodermal precursors that populate the pharyngeal region as well as the organ-forming region. We also find that knocking down her5 recapitulates some of the endodermal phenotypes of shiri mutants, further revealing the role of her5 in endoderm development. Positional cloning reveals that shiri encodes Med12, a regulatory subunit of the transcriptional Mediator complex recently associated with two human syndromes. Additional studies indicate that Med12 modulates the ability of Casanova/Sox32 to induce sox17 expression. Thus, detailed phenotypic analyses of embryos defective in a component of the Mediator complex has revealed new insights into discrete aspects of vertebrate endoderm development, and provide possible explanations for the craniofacial and digestive system defects observed in humans with mutations in MED12.
Fate mapping experiments in zebrafish have revealed that endodermal progenitors derive from the four tiers of blastomeres closest to the blastoderm margin at the late blastula stage (Warga and Nüsslein-Volhard, 1999). Analyses of zebrafish mutants and overexpression studies have identified members of the Nodal family and their transcriptional effectors as key components of the pathway leading to early endoderm formation (reviewed by Stainier, 2002; Grapin-Botton and Constam, 2007; Zorn and Wells, 2007). Two Nodal-related factors, Cyclops (Cyc) and Squint (Sqt), are essential for endoderm formation since their absence results in lack of all endoderm, and most mesoderm (Feldman et al, 1998). These factors relay their signals through One-eyed pinhead, an EGF-CFC family member and required coreceptor for Nodal molecules (Gritsman et al., 1999), and a type-I TGFβ receptor such as Taram-a (Tar) (Aoki et al., 2002). Three transcription factors, Bonnie and Clyde (Kikuchi et al., 2000), Casanova/Sox32 (Dickmeis et al., 2001; Kikuchi et al, 2001), and Faust/Gata5 (Reiter et al., 2001), which are transcriptional targets as well as effectors of Nodal signaling, are critical for endoderm formation. cas/sox32 mutants lack all endodermal precursors without displaying other overt differentiation defects, indicating the specificity of cas/sox32 in endoderm formation (Alexander et al., 1999). Furthermore, cas/sox32 is sufficient to convert mesodermal precursors into endoderm when it is misexpressed near the margin (Kikuchi et al., 2001), placing Cas/Sox32 as a key regulator of endoderm formation. Pou5f1, a homolog of mammalian OCT4, also plays an important role during endoderm formation (Lunde et al., 2004; Reim et al., 2004). It is required to maintain cas/sox32 expression during gastrulation and functions with Cas/Sox32 to activate sox17 expression.
Fox transcription factors and the hairy/enhancer of split-related gene, her5, are also expressed by endodermal precursors during gastrulation. Three fox genes-foxa1, foxa2, and foxa3- appear to be expressed in endodermal cells in a sequential and overlapping pattern in zebrafish (Odenthal and Nüsslein-Volhard, 1998). Expression of her5 becomes restricted to a subpopulation of anterior endodermal precursors during gastrulation (Bally-Cuif et al., 2000). Ectopic misexpression of wild-type and mutant forms of Her5 has shown that Her5 directs endodermal cells to populate the pharynx, liver, pancreas, and gut (Bally-Cuif et al., 2000).
During gastrulation, endodermal cells converge towards the dorsal side of the embryo and become distributed along the AP axis. The position of these cells predicts the topographic arrangement of the future digestive system; dorsal-most cells will give rise to the pharynx, lateral cells will give rise to the digestive organs, and ventral cells will give rise to the posterior part of the alimentary canal (Warga and Nüsslein-Volhard, 1999; Bally-Cuif et al, 2000; Ward et al., 2007). During liver development, several signaling molecules, such as Wnt2bb (Ober et al., 2006), as well as members of the Bone Morphogenic Protein (BMP) and Fibroblast Growth Factor (FGF) families (Jung et al., 1999; Rossi et al., 2001; Shin et al., 2007) are required for hepatoblast specification as assessed by hhex and prox1 expression, which appear in the hepatic primordium at approximately 22 hours post-fertilization (hpf) (Ober et al., 2006).
Sequence-specific transcription factors require cofactors to recruit RNA polymerase II and the basal transcriptional machinery. One commonly used cofactor is the Mediator complex, first described in yeast (reviewed by Malik and Roeder, 2000). A number of transcriptional activators, as well as some repressors, require the Mediator complex to regulate transcription in vitro. The Mediator complex can be divided into functional submodules. One of these submodules, called positive cofactor 2 (PC2), is sufficient for co-activator activity in vitro and directly interacts with RNA polymerase II, indicating that it represents a core complex (Ryu et al., 1999). In contrast, some subunits, such as the MED1/TRAP220 subunit, may act as adaptors for specific transcription factors such as nuclear receptors (Ge at al., 2002). MED12/TRAP230 and MED13/TRAP240 subunits are absent from the core complex and form a cyclin-dependent kinase (CDK) module with CDK8 and Cyclin C (Janody et al., 2003). MED12 belongs to a submodule that has been implicated in transcriptional repression in yeast and transcriptional activation in Drosophila (reviewed by Malik and Roeder, 2005). Recently, kohtalo/med12 was shown to have a determinative role in brain (Hong et al., 2005; Wang et al., 2006) as well as neural crest and cartilage (Rau et al., 2006) development in zebrafish.
Here we analyze two med12 mutants identified in a genetic screen for endodermal regulators. We show that these med12 mutants exhibit multiple endodermal defects including delayed differentiation of specific cell types in the liver and pancreas. Gain-of-function experiments with Cas/Sox32 reveal that Med12 modulates the activity of this key regulator of endoderm formation, while loss-of-function experiments reveal genetic interactions between Med12 and Foxa2. We propose a model whereby Med12 interacts with various transcriptional factors at distinct steps of endoderm development to regulate cell differentiation. Defects in these steps may lead to the craniofacial and digestive system phenotypes observed in humans carrying MED12 mutations.
Embryos and adult fish were raised and maintained under standard laboratory conditions. We used the following mutant and transgenic lines: shr/med12s432, shr/med12s435, Tg(gutGFP)s854 (Field et al., 2003a), Tg(fabp1a:dsRed, elastase:GFP; insulin:dsRed) (Dong et al., 2007), and Tg(−0.7her5:EGFP)ne2067 (Tallafuss and Bally-Cuif, 2003).
To search for mutations affecting endodermal organ morphogenesis, we screened 3578 F3 clutches from 652 ENU-mutagenized F2 families in the Tg(gutGFP)s854 background (E.A.O., H.V., H.A.F., D. Dong, P. Aanstad, T. Sakaguchi, M. Bagnat, C. Munson, W.-S. C., C. H. S., S. Curado, R. Anderson, J. Frantsve, D. Beis, T. Bartman and D.Y.R.S., unpublished observations). Based on the number of crosses per F2 family, we calculate that our screen surveyed 1035 mutagenized genomes. The specific locus test, using the golden mutation, indicated a mutation rate of approximately 0.3% per gene per mutagenized genome.
We mapped the mutation to chromosome 14 using a standard set of SSLP markers (Knapik et al., 1998). For fine mapping, 1400 single med12 mutant embryos were tested with SSLP markers in the critical interval. The primer sequence for z markers can be found at http://www.zfin.org/ and the following primers were designed against CA-repeats within the determined interval: 3e19, 112e1, 10h23, 119p14, 18a21. For cloning and sequencing of the med12 gene, total RNA was extracted from 48 hpf wild-type and med12 mutant embryos using Trizol (Invitrogen). cDNAs were generated by performing RT-PCR using the Superscript Kit (Invitrogen). PCR products from at least two independent PCR reactions per mutant allele were sequenced and analyzed.
Donor embryos were injected with 5% rhodamine-dextran, and cells were transplanted into wild-type hosts at the 1000 cell stage. Transplanted embryos were visualized daily with a fluorescence microscope to monitor rhodamine positive cells over a span of 12 days. Live host embryos were embedded in 3% methylcellulose and images were obtained using a Zeiss LSM5 Pascal confocal microscope. Whole-mount in situ hybridization was performed as previously described (Alexander et al., 1998), using the following probes: hhex (Ho et al., 1999), prox1 (Glasgow and Tomarev, 1998), cp (Korzh et al., 2001), sepp1b (Field et al., 2003a), her5 (Bally-Cuif et al., 2000), bon (Kikuchi et al., 2000), cas/sox32 (Kikuchi et al., 2001), foxa2 (Odenthal and Nüsslein-Volhard, 1998), gata5 (Reiter et al., 2001), sox17 (Alexander et al., 1999), and foxa3 (Odenthal and Nüsslein-Volhard, 1998). We used the following antibodies: polyclonal antibody against Prox1 (rabbit, 1:1000; Chemicon), Insulin (guinea pig, 1:200; Biomeda), Somatostatin (rabbit, 1:200; ICN Biomedicals, INC.), pan-Cadherin (rabbit, 1:1000; Sigma), ABCB11 (rabbit, 1:200; Komiya), monoclonal antibody against Islet1/2 (1:15; Developmental Studies Hybridoma Bank, clone 39.4D5), Glucagon (mouse, 1:200; Sigma), Alcam (mouse, 1:10; Zebrafish International Resource Center), 2F11 (mouse, 1:400; generous gift from J. Lewis, Cancer Research, UK), and fluorescently-conjugated antibodies from Molecular Probes. Manual removal of the yolk was followed by whole-mount antibody staining in PT (0.3% Triton in PBS pH 7.3) with 4% BSA. For BrdU pulse-chase experiments, embryos were incubated with 10mM BrdU (Sigma-Aldrich) at 33 hpf and 47 hpf for one hour and fixed at 34 hpf and 48 hpf, respectively. Manual removal of the yolk was followed by five-minute Pronase and one-hour 2M HCl incubation. Rat anti-BrdU antibody (1:200; Accurate Chemical and Scientific Corporation) was used. Images were obtained using a Zeiss LSM5 Pascal confocal microscope. For the TUNEL assay, embryos were fixed in 3% formaldehyde, sectioned (250µm thickness), preincubated in PBST, and then labeled with the TUNEL kit (Roche) for 1 hr at 37°C. Sections were washed with PBT (0.1% Triton X-100 in PBS pH 7.3) and visualized by confocal imaging. For coimmunostaining with Prox1, sections were first incubated with primary antibodies, then with TUNEL solutions, and finally with secondary antibodies.
Image analysis was performed using ImageJ. Data stacks that had been acquired on the confocal microscope and saved in LSM format were imported using the “LSM reader” plugin (by P. Pirrotte et al.). The RGB overlay was examined and the liver was outlined and highlighted using the “dotted line” plugin (by K. Balder). The green channel (Tg(gutGFP)s854) was cropped to only the outlined volume. After applying a median filter to reduce noise, the green stack was thresholded and the liver volume was estimated using the “voxel counter” plugin (by W. Rasband). The binary stack was then used as a mask on the red channel (BrdU-positive cells) and proliferating cells were manually counted using the “point picker” plugin (by P. Thévenaz) and the number was normalized by the volume.
Morpholinos against med12, her5 (Geling et al., 2003), and foxa2 (Norton et al., 2005) were purchased from GeneTools LLC (med12: 5’-CAAGTGAAGTCATACCTGTTGGTTA-3’; her5: 5’-TTGGTTCGCTCATTTTGTGTATTCC-3'; foxa2: 5'-CCTCCATTTTGACAGCACCGAGCAT - 3'). Wild-type (AB or TL), Tg(gutGFP)s854, Tg(fabp1a:dsRed, elastase:GFP; insulin:dsRed) or Tg(−0.7her5:EGFP)ne2067 embryos were injected at the one-cell stage with 1–2ng (med12), 4–8ng (her5) or 4–5ng (foxa2) of morpholino and assayed between 90% epiboly and 96 hpf. 5–10 pg of cas, 60 pg of gata5 or 100 pg of bon mRNA was injected into one-cell stage embryos with or without med12 morpholino. For genotyping of med12s432 allele, the following primers were used: 5’- TGTGAGTGGGCTGTTTCTTG-3’ and 5’-GAGTGTGGCAGCAGACAGAGG-3’, followed by Fnu4HI digestion.
To identify genes involved in endodermal organ morphogenesis, we carried out a forward genetic screen using the Tg(gutGFP)s854 line, a transgenic zebrafish line expressing GFP throughout the developing endoderm (Field et al., 2003a). We identified the recessive mutant shiri (shr) in which, at 54 hpf, the size of the liver and pancreas is reduced (Fig. 1A, A’). Two mutant alleles (shrs432 and shrs435) were identified and could not be distinguished phenotypically. To determine when defects in liver formation can first be detected in shr mutants, we examined the expression of several established liver markers. The earliest known genes expressed in the zebrafish liver anlage are hhex and prox1 at approximately 22 hpf (Ober et al., 2006). The early expression of these genes is detected only at very low levels in shrs435 mutants (Fig. 1B, B’, S1A, A’), indicating that hepatic specification is affected. However, Prox1 (Burke and Oliver, 2002) immunostainings at 54 hpf show that a small liver bud eventually forms in shrs435 mutants and that all the liver cells express Prox1 (Fig. 1C, C’). Therefore, hepatic specification appears to be delayed or partially affected in shrs435 mutants. The endodermal cells between the liver and intestinal bulb primordium are thought to give rise to the extra-hepatic duct, while the extra-pancreatic duct comes from the anterior (ventral) pancreatic bud (Field et al., 2003b). shrs435 mutants exhibit malformations of the extra-hepatic and extra-pancreatic ducts (white brackets in Fig. 1A, A’, C, C’) and often show multiple anterior pancreatic buds (white arrowheads in Fig. 1A, A’, C, C’, S1C, C’, S2A’). Gene expression studies in mammals indicate that hepatoblasts give rise to cholangiocytes (or bile-duct cells) as well as hepatocytes (Germain et al., 1988). In zebrafish, Alcam (formerly known as Dm-grasp), a cell surface adhesion molecule of the immunoglobulin superfamily (Fashena and Westerfield, 1999), is initially localized along the entire cell surface of hepatoblasts, then becomes restricted by 54 hpf (Fig. 1C). In shrs435 mutants, Alcam remains distributed along the entire cell surface of liver cells (Fig. 1C’, D’, E’). ABCB11 is a bile salt transporter localized in the bile canaliculi (Lam et al., 2005) which can be used to assess hepatic differentiation. A dramatic reduction of ABCB11 expression is observed in shrs435 mutant larvae (Fig. 1D, D’). Pan-Cadherin staining shows a very similar pattern as Alcam staining during liver development and also reveals a clear difference between wild-type and shrs435 mutant larvae at 72 hpf (Fig. S1C, C’): Cadherin remains distributed along the entire cell surface of liver cells in shrs435 mutants (Fig. S1C, C’; data not shown).
Next, we examined the expression of the differentiation marker, fatty acid binding protein 1a, liver (fabp1a), using the Tg(fabp1a:dsRed, elastase:GFP; insulin:dsRed) line (Dong et al., 2007): shrs435 mutant larvae show a reduced number of DsRed expressing cells in the liver (Fig. S2A, A’; data not shown). We also examined the expression of ceruloplasmin (cp), which is detected in the developing liver and hepatic duct, and selenoprotein P, plasma, 1b (sepp1b), which is only expressed in differentiating hepatocytes. Both genes are greatly reduced in their expression in shrs435 mutants at 52 hpf (Fig. S1B, B’; data not shown). Taken together, these data suggest that shr is required for hepatic specification and differentiation as well as for the development of the ventral pancreas and hepatopancreatic duct.
In zebrafish, the dorsal pancreatic bud, present by 24 hpf on the dorsal aspect of the developing gut, generates endocrine tissue (mainly producing Insulin, Somatostatin, and Glucagon), while the ventral pancreatic bud, present by 40 hpf on the ventral aspect of the gut, gives rise to the pancreatic duct and exocrine cells (mainly producing digestive enzymes) (Field et al., 2003b). To examine the role of shr in dorsal pancreas development, we examined the expression of several endocrine pancreatic markers, such as Islet, Insulin, Somatostatin, and Glucagon (Biemar et al., 2001). The anti-Islet antibody detects all three endocrine cell types: Insulin-producing β-cells, Somatostatin-producing δ-cells, and Glucagon-producing α-cells. The expression of Islet and Insulin in shrs435 mutants at 34 hpf is comparable to wild-type (Fig. 2A, A’, B, B’) while that of Somatostatin appears to be reduced (Fig. 2D, D’). Strikingly, Glucagon expression is almost completely absent in shrs435 mutants at 34 hpf (Fig. 2C, C’), indicating that shr is essential for Glucagon expression at early stages. At 56 hpf, the expression of Islet, Insulin, Somatostatin, and Glucagon is comparable in mutants and wild-type (Fig. 2E – H’), suggesting that there is either a compensation of Glucagon regulation in the absence of shr function or that Glucagon expression is merely delayed in shrs435 mutants. Altogether, these data point to the crucial role of shr in Glucagon expression prior to 56 hpf.
To better understand the molecular basis of the shr phenotypes, we isolated the gene disrupted by the shr mutation. Using a standard set of SSLP markers, we located shr on chromosome 14. Using zebrafish genetic and genomic databases and chromosomal walking, we constructed a contig of BAC clones that spans the shr region (Fig. 3A; data not shown). Gene prediction program analyses revealed three open reading frames within this region, and sequencing those candidate genes showed that both shr mutant alleles of med12 contain significant lesions (Fig. 3B). Med12 is a subunit of the Mediator complex, a key cofactor that acts as a bridge between DNA-binding transcription factors and RNA polymerase II (Malik and Roeder, 2005). The shrs432 mutant allele contains a C to A base change at position 1638 leading to a premature stop codon that deletes the PQL and OPA as well as most of the LS domains. The shr s435 mutant allele contains a C to T base change at position 4383 creating a premature stop codon that truncates the OPA and most of the PQL domains. The PQL and OPA domains are essential for other transcription factors to bind (Zhou et al., 2002). To further investigate whether a reduction in Med12 function causes the liver phenotype observed in shr mutants, we knocked down its function using morpholino antisense oligos (MO). Injection of 2 ng of a med12 splice-site MO into Tg(gutGFP)s854 embryos recapitulated the shr morphological phenotypes at 36 hpf (Fig. 3C, bright field). It also led to a strong reduction of hepatic tissue at 48 hpf (Fig. 3C, confocal projections). We examined the expression pattern of med12 at the eight-cell stage, 24 and 48 hpf. med12 RNA is present as maternal RNA and is widely expressed in the early embryo (data not shown). Overall, the tight genetic linkage, presence of severe molecular lesions, and MO phenocopy indicate that med12 is the gene affected by the shr mutations.
To determine whether Med12 is required for cell proliferation, which might explain the small size of the liver in med12 mutants, we performed BrdU pulse-chase experiments. Wild-type and med12s435 mutant embryos were pulsed with BrdU at 33 and 47 hpf and harvested one hour later (Fig. 4A–B’). The number of BrdU-positive cells was quantitated in the liver region (white dashed lines in Fig. 4A–B’) and the overall proportion of BrdU incorporation in med12s435 mutants did not appear to be different from wild-type in these experimental conditions (Table 1). In parallel, an analysis of apoptosis was carried out in wild-type and med12s435 mutant embryos at 34 and 48 hpf (Fig. S3C–H). No TUNEL-positive endodermal cells were observed in med12s435 mutants at these stages (Fig. S3C, D, E, F), whereas brain cells were clearly labeled in 48 hpf mutant embryos (Fig. S3G, H). It has been reported that MED1, another subunit of the Mediator complex, is essential for embryonic survival in a cell-autonomous manner (Landles et al., 2003). To determine whether Med12 is similarly required, we transplanted wild-type and med12s435 mutant cells into wild-type embryos. We found that both wild-type and med12s435 mutant cells survive until at least day 12 in wild-type hosts (Fig. 4C–D’). Altogether, these data indicate that Med12 is not essential for cell proliferation or survival, as previously shown during eye-antennal disc development in Drosophila (Treisman, 2001).
As med12 does not appear to regulate cell proliferation, we hypothesized that the reduced liver size in med12 mutants might be due to a defect in endodermal induction. We found that at 90% epiboly, a subset of sox17- positive cells in the anterior region of the embryo appears to be missing in med12s432 mutants (Fig. S4C, C’), whereas other early endodermal genes such as bon, gata5, foxa2, and cas show comparable expression in wild-type and mutants (Fig. S4A, A’, B, B’, D, D’, E, E’). Because her5 has been shown to be expressed in anterior endodermal cells during gastrulation (Bally-Cuif et al., 2000), we further examined the sox17 phenotype by examining her5 expression and observed a pronounced reduction in endodermal her5 expression (Fig. 5A, A’). Based on fate mapping studies (Warga and Nüsslein-Volhard, 1998), anterior endodermal precursors give rise to the endoderm in the pharyngeal and organ (i.e. liver, pancreas, and swim bladder) forming region. Next, we examined the wild-type expression of her5 using the Tg(−0.7her5:EGFP)ne2067 line (Tallafuss and Bally-Cuif, 2003) to further determine the endodermal contribution of her5-positive cells. her5 mRNA is no longer detected after 100% epiboly, but the progeny of her5- expressing cells can be traced with GFP expression for several hours after zygotic expression terminates. Tg(−0.7her5:EGFP)ne2067 expression has been shown to recapitulate her5 endodermal expression, with similar onset and anteroposterior extent (Tallafuss and Bally-Cuif, 2003). We found that GFP is expressed in the endoderm of the pharyngeal region as well as the organ-forming region during late somitogenesis stages (Fig. 5C, D; data not shown). In addition, at 30 hpf, we detected clear GFP expression in the liver-forming region as well as the dorsal pancreatic region (data not shown). We injected the med12 splice-site MO into the Tg(−0.7her5:EGFP)ne2067 line and examined the expression of GFP at several time points. Compared to uninjected controls, med12 MO-injected embryos showed a clear reduction of GFP expression at 10, 15, and 19 hpf, as well as later stages (Fig. 5B’, C’, D’; data not shown; n=20 in each set of MO-injections; the percentage of MO-injected embryos showing a decrease of GFP expression at 10 hpf was 80% (n=16/20), at 15 hpf was 90% (n=18/20), and at 19 hpf was 80% (n=16/20)). Taken together, these data suggest that her5-positive endodermal cells contribute to the formation of the liver and pancreas and that Med12 regulates their number, and/or differentiation.
We hypothesized that a reduction in endodermal her5 expression at 90% epiboly in med12s432 mutants could lead to the endodermal organ defects seen at later stages. To test this hypothesis, we first determined when a defect in liver formation is first detectable in her5 MO-injected embryos (a.k.a. morphants; the her5 MO was previously validated by Geling et al. (2003) using the her5PAC-GFP transgenic line). The expression of hhex and prox1 was detected only at very low levels in 30 hpf her5 morphants (Fig. 6A–B’; n=20 in each set of MO-injections; the percentage of morphants showing the decrease of hhex was 60% (n=12/20) and prox1 was 80% (n=16/20)), indicating that hepatic specification was affected. Immunostainings in the Tg(gutGFP)s854 line with Prox1 and Insulin indicated that the formation of the hepatopancreatic duct and the outgrowth of the ventral pancreas were perturbed in her5 morphants, leading to isolated Insulin-positive endocrine cells (white bracket and arrowheads in Fig. 6C, C’). However, similar to med12s435 mutants, all the liver cells expressed Prox1 in the her5 morphants at 46 hpf (Fig. 6C’), indicating that hepatic specification was only partially defective when her5 function was down regulated. We also examined the expression of ABCB11 in her5 morphants at 76 hpf, and observed a dramatic reduction (Fig. 6D, D’). These data suggest that down-regulation of her5 recapitulates some of the endodermal phenotypes of med12 mutants, leading to defects in the development of the liver, pancreas, and hepatopancreatic duct.
The transcription factor Sox9 is involved in brain, neural crest, and cartilage development and interacts with human and zebrafish Med12 (Zhou et al., 2002; Rau et al., 2006; Wang et al., 2006). Casanova, a novel Sox-related protein and one of the effectors of the Nodal signaling pathway, is necessary for endoderm development in zebrafish (Alexander et al., 1999; Dickmeis et al., 2001; Kikuchi et al., 2001). Therefore, we tested whether Med12 was required for Cas/Sox32 function in early endoderm development in zebrafish. When cas/sox32 mRNA is injected into wild-type embryos at the one-cell stage, it induces a dramatic increase in sox17 expression (Dickmeis et al., 2001; Kikuchi et al., 2001; Sakaguchi et al., 2001). We injected cas/sox32 mRNA into embryos from med12s432 heterozygote incrosses and examined sox17 expression (Fig. 7A, B, C; a total of 48 embryos were analyzed and each of them was categorized as strong induction, intermediate induction, or no/slight induction). Surprisingly, only 25% of the embryos showed strong induction of sox17 expression (n=12, Fig. 7A), while approximately 25% showed almost no induction (n=10, Fig. 7C). The rest of the embryos showed an intermediate level of sox17 induction (n=26, Fig. 7B). Genotyping revealed a strong correlation between the level of sox17 induction and the number of copies of wild-type med12. Of the 48 injected embryos, 16 were homozygous wild-type and 12 of them showed strong sox17 induction while 4 showed intermediate induction; 19 were heterozygous and 18 of them showed intermediate induction while 1 showed no/slight induction; 13 were homozygous mutants and 9 of them showed no/slight induction while 4 showed intermediate induction. These data reveal the essential role for Med12 in enabling Cas/Sox32 to induce sox17 expression in early embryos.
Next, we examined whether Med12 has a specific role in Cas/Sox32 activity. It has been reported that bon and gata5 overexpression can also induce an increase in sox17 expression (Reiter et al., 2001). We injected bon, cas/sox32 or gata5 mRNA with or without med12 MO and examined sox17 expression (Fig. S5A, B, C, G, H, I, M, N, O). Co-injecting the embryos with med12 MO caused a reduction in sox17 expression compared to the embryos injected with mRNA alone (n=20 in each set of MO injections; the percentage of morphants showing a decrease in sox17 expression following bon mRNA injection was 70% (n=14/20), following cas mRNA injection was 80% (n=16/20), and following gata5 mRNA injection was 80% (n=16/20)). We further analyzed whether the decrease in sox17 expression in these co-injected embryos was due to the impairment of Bon or Gata5 activity or the reduction of cas/sox32 expression. We examined cas/sox32 expression in mRNA injected embryos and med12 MO co-injected embryos (Fig. S5D, E, F, J, K, L). Induction of cas/sox32 expression appeared comparable in both sets of embryos (n=20 in each set of experiments; the percentage of embryos showing an increase in cas/sox32 expression following bon mRNA injection was 80% (n=16/20), and following gata5 mRNA injection was 80% (n=16/20), while following co-injection of bon mRNA and med12 MO was 80% (n=16/20), and following co-injection of gata5 mRNA and med12 MO was 70% (n=14/20)). Thus, from these overexpression experiments, it appears that Bon and Gata5 act through Cas/Sox32 to induce sox17 expression and that Med12 is required for the full level of Cas/Sox32 activity.
foxa2 is also expressed by endodermal precursors during gastrulation (Odenthal and Nüsslein-Volhard, 1999), although foxa2 mutants do not show severe endodermal defects (Norton et al., 2004). In order to investigate the role of Med12 and Foxa2 in early endoderm development, we injected wild-type embryos and embryos from med12s432 heterozygote incrosses with foxa2 MO at the one-cell stage and examined foxa3 expression at 48 hpf (Fig. 7D–H’; the foxa2 MO was validated by Norton et al. (2004) and can phenocopy the foxa2st20 and foxa2tv53a mutations). Compared with wild-type embryos, foxa2 morphants showed a slight perturbation of endodermal organ development (Fig. 7D, F). However, in the absence of one copy of med12, more severe defects in endodermal organ formation were evident (Fig. 7G). Additionally, med12s432 homozygous mutant embryos injected with foxa2 MO showed a nearly complete absence of endodermal organs (Fig. 7H, H’). A total of 40 embryos were analyzed from med12s432 heterozygote incrosses injected with foxa2 MO and each of them was categorized as “foxa2 MO phenotype”, “more severe defects than the foxa2 MO phenotype”, or “complete absence of endodermal organs”. Of the injected 40 embryos, 12 were homozygous wild-type and 8 of them showed the foxa2 MO phenotype while 4 showed more severe defects than the foxa2 MO phenotype; 17 were heterozygous and all of them showed more severe defects than the foxa2 MO phenotype; 11 were homozygous mutants and 9 of them showed a complete absence of endodermal organs while 2 showed more severe defects than the foxa2 MO phenotype. We further compared the expression levels of fabp1a in foxa2 MO injected wild-type and med12s432 heterozygous larvae (Fig. 7I, J, K). Approximately 50% of the foxa2 MO injected larvae from med12s432 outcrosses showed a dramatic down-regulation of fabp1a expression (Fig. 7J, K). A total of 60 larvae were analyzed and each of them was categorized as strong or weak fabp1a expression. Of the 60 injected embryos, 27 were homozygous wild-type and 24 of them showed strong expression while 3 showed weak expression; 33 were heterozygous and 29 of them showed weak expression while 4 showed strong expression. Altogether, these data suggest that Med12 and Foxa2 function together to regulate endoderm development.
Mutations in MED12 can cause Lujan syndrome (Schwartz et al., 2007) and Opitz-Kaveggia syndrome (Risheg et al., 2007) which include multiple clinical features potentially derived from endodermal defects. In this study, we examined the role of Med12 in endoderm and endodermal organ development in zebrafish. We showed a critical role for Med12 in endodermal expression of her5 as well as a modulator of Cas/Sox 32 activity. Med12 is subsequently required for the differentiation of hepatocytes as well as a subset of pancreatic endocrine cells.
We explored the possibility that Med12 functions as a modulator of Cas/Sox32 activity in early endoderm development based on two findings: a subset of sox17 positive endodermal cells in the anterior region of the embryo, which are known to express her5, appears to be missing in med12s432 mutants at 90% epiboly, and Med12 interacts with Sox9 in humans as well as in zebrafish (Zhou et al., 2002; Rau et al., 2006; Wang et al., 2006). Ectopic expression of cas/sox32 induces sox17 in wild-type embryos but has little effect in med12s432 homozygous mutant embryos, indicating that Med12 is a modulator of Cas/Sox32 activity (Fig. S6). However, med12s432 mutants only show a mild sox17 phenotype, which is likely due to the presence of maternal med12 mRNA (data not shown) and possibly Med12 protein.
med12s432 mutants also show a specific reduction of endodermal her5 expression at 90% epiboly. Analysis of the Tg(−0.7her5:EGFP)ne2067 line indicates that her5 expressing endodermal cells contribute to the pharyngeal region as well as the organ-forming region. In the absence of zygotic Med12 function, the expression level of GFP in both populations of endodermal cells is greatly decreased. These data indicate that Med12 regulates endodermal her5 expression as well as the differentiation of the her5 expressing endodermal cells (Fig. S6).
Pancreas development relies on a well-characterized network of transcription factors and coactivators. In med12 mutants, we observed a specific down-regulation of Glucagon at 34 hpf and restoration by 56 hpf. As the expression of Islet and Insulin at 34 hpf is comparable in wild-type and mutant embryos while that of Glucagon is almost completely absent, it is likely that α–cell (Glucagon-producing cell) precursors are present but have not yet fully differentiated. Alternatively, the α–cells could arise from a different source.
Sox4b is a key player in pancreatic α–cell differentiation (Mavropoulos et al., 2005). Sox4b knock-down led to a drastic reduction in glucagon expression, while other pancreatic markers such as insulin and somatostatin were not significantly affected (Mavropoulos et al., 2005). This phenotype is similar to the med12s435 mutant phenotype. Therefore, it is likely that Med12 acts as a coactivator of Sox4b (Fig. S6). The facts that Med12 modulates Sox9 activity during cartilage and neural crest development (Rau et al., 2006), and Cas/Sox32 activity during early endoderm development, and that Med12 is a potential coactivator of Sox4b suggest that Med12 interacts with specific Sox proteins.
Hepatocyte nuclear factor 4 (HNF-4) is a member of the nuclear receptor superfamily and one of the key regulators of hepatic differentiation in mammals (Li et al., 2000; Parviz et al., 2003). Conditional knock-down of HNF-4 in mouse indicates that it is essential for the accumulation of hepatic glycogen stores, the generation of a hepatic epithelium, and the morphological and functional differentiation of hepatocytes (Parviz et al., 2003). It has been shown that HNF-4 interacts physically with several subunits of the TRAP/SMCC/Mediator complex such as MED1/TRAP220, MED14/TRAP170, and MED12/TRAP230 and that these interactions are required for its transcriptional activity (Malik et al., 2002). In med12s435 mutants, the maturation of hepatocytes is defective based on Alcam, Cadherin, and ABCB11 immunostaining. It is likely that this phenotype is due to the impairment of Hnf-4 function in the absence of Med12 (Fig. S6). In zebrafish, hnf4 starts to be expressed in the liver and intestinal bulb at 30 hpf (Field et al., 2003a), and hnf4 expression in wild-type and med12s435 mutant embryos was comparable at 36 hpf (data not shown). hnf4 MO-injected embryos show hepatoblast specification (as assessed by Prox1 immunostaining) but failure of hepatic differentiation (as assessed by Alcam and Cadherin immunostaining, unpublished observations). Altogether, these data indicate that Med12 regulates hepatocyte differentiation at multiple levels, including by regulating her5 expression at early stages and Hnf-4 function subsequently (Fig. S6).
Lujan and Opitz-Kaveggia syndromes are X-linked mental retardation syndromes caused by mutations in MED12 (Risheg et al., 2007; Schwartz et al., 2007). They have several clinical features in common including maxillary hypoplasia, open mouth, high narrow palate, dental crowding and micrognathia/retrognathia, as well as anal anomalies in Opitz-Kaveggia syndrome. These craniofacial and digestive system defects may derive from primary endodermal defects. Zebrafish med12 mutants show defects in the formation of pharyngeal arches (Hong et al., 2005; Rau et al., 2006), which is essential for craniofacial development. These arches form by the contribution and interaction of tissues derived from the neural crest, mesoderm, and endoderm (reviewed by Yelick and Schilling, 2002). Mutations in genes such as bon (Kikuchi et al., 2000), cas/sox32 (Dickmeis et al., 2001; Kikuchi et al., 2001), and faust/gata5 (Reiter et al., 2001) which are required for early endodermal development cause defects in craniofacial chondrogenesis, providing evidence for a role for the endoderm in craniofacial skeletal development (Piotrowski and Nüsslein-Volhard, 2000; reviewed by Yelick and Schilling, 2002).
The terminal portion of the hindgut contacts the surface ectoderm to form the cloacal membrane, which eventually becomes divided into the posterior anal membrane and the anterior urogenital membrane. Subsequently, an opening is made between the terminal portion of the hindgut and the outside by rupturing the posterior anal membrane (Sadler, 1995). Therefore, anal anomalies such as the imperforate anus observed in Opitz-Kaveggia patients could be derived from hindgut malformations (Qi et al., 2006). These observations and comparisons between human and zebrafish highlight an essential role for Med12 during development and emphasize the usefulness of the zebrafish to elucidate human congenital malformations.
The Mediator is a key cofactor complex that enables transcription factors to regulate transcription by RNA polymerase II. It has been implicated in the regulation of transcriptional initiation from yeast to mammals. The role of each individual subunit of the Mediator, especially the CDK module (Med12, Med13, CDK8, and CycC) remains unclear. In Caenorhabditis elegans, genetic analysis has suggested that MDT-12 (the MED12 ortholog) suppresses the β-catenin-dependent pathway to allow the expression of a caudal homolog, pal-1 (Zhang et al., 2000). During Drosophila eye development, Med12 and Med13 clearly have determinative roles (Treisman, 2001) as zebrafish Med12 appears to have during endoderm development through its regulation of transcription factors such as Her5, Sox32, Sox4b, and Hnf-4. Recently, it was shown that MED13 is a target of miR-208 required for expression of betaMHC in response to stress and hypothyroidism (van Rooij et al., 2007). Therefore, despite the fact that the Med12-Med13-CDK8-CycC module has been known to have an intrinsic potential to repress transcription in yeast and C. elegans, the Mediator complex containing this module could participate in activation in higher eukaryotes. It is clear that many additional roles for the Mediator complex remain to be discovered in its capacity as an effector of diverse developmental and homeostatic signals (Kim et al., 2006; Zhou et al., 2006). Specifically, continuing studies of Med12 in various model systems will further our understanding of its role in human development and disease.
We thank Ana Ayala for excellent fish care, Pia Aanstad, Chantilly Munson, Rima Arnaout, Paul Gerhman, and Donghun Shin for critical readings and discussions of the manuscript, and Chunyue Yin and other Stainier lab members for technical help and discussion. S. –K. H. would like to thank Dr. Igor Dawid for his support and guidance. C.H.S. was supported by an NIH postdoctoral fellowship (F32DK068891), W.-S.C. by a fellowship from the California Institute for Regenerative Medicine (CIRM), E.A.O. by the UCSF Liver Center through an NIH pilot feasibility grant, and H.V. and J. H. by a long-term fellowship of the Human Frontier Science Program (HFSP). This work was supported in part by grants from the NIH (NIDDK) and the Packard Foundation to D.Y.R.S.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.