Potential conflict of interest: Nothing to report.
During vertebrate embryogenesis, the liver develops at a precise location along the endodermal primitive gut tube because of signaling delivered by adjacent mesodermal tissues. Although several signaling molecules have been associated with liver formation, the molecular mechanism that regulates liver specification is still unclear. We previously performed a screen in medaka to isolate mutants with impaired liver development. The medaka hio mutants exhibit a profound (but transient) defect in liver specification that resembles the liver formation defect found in zebrafish prometheus (prt) mutants, whose mutation occurs in the wnt2bb gene. In addition to their liver abnormality, hio mutants lack pectoral fins and die after hatching. Positional cloning indicated that the hio mutation affects the raldh2 gene encoding retinaldehyde dehydrogenase type2 (RALDH2), the enzyme principally responsible for retinoic acid (RA) biosynthesis. Mutations of raldh2 in zebrafish preclude the development of pectoral fins. Interestingly, in hio mutants, expression of wnt2bb in the lateral plate mesoderm (LPM) directly adjacent to the liver-forming endoderm was completely lost. Conclusion: Our data reveal the unexpected finding that RA signaling positively regulates the wnt2bb gene expression required for liver specification in medaka. These results suggest that a common molecular mechanism may underlie liver and pectoral fin specification during piscine embryogenesis. (HEPATOLOGY 2009.)
Embryonic liver development occurs in multiple stages that are governed by hormonal factors as well as by intercellular and matrix–cellular interactions. In mice, liver ontogeny initiates on approximately embryonic day 9 (E9), when epithelial cells of the foregut endoderm interact with the cardiogenic mesoderm and commit to becoming the liver primordium. The liver primordium proliferates and invades the mesenchyme of the septum transversum to give rise to the hepatic codes and bud at E9.5.1, 2 Over the last decade, studies in rats and mice have greatly expanded the list of molecules known to contribute to liver development; however, it is likely that many more factors are involved in this complex process. In particular, the mechanism underlying the local induction of liver formation remains poorly understood. This gap in our knowledge is reflected in the dearth of reports on rodent mutations that specifically interfere with the initial specification of the liver anlage.
Small fish are particularly suitable for mutational investigations because they are easy to rear in a relatively compact space, their generation times are reasonably short, and they produce transparent embryos. In many fish species, embryos develop outside the mother's body, making it easy to inspect them visually and to manipulate their tissues and cells. Our group has previously used systematic mutagenesis in medaka to generate numerous mutations affecting various aspects of liver development and function.3–5 The focus of this paper is the recessive mutation hiohgi (hio). In wild-type (WT) medaka, the hepatic bud forms from the endoderm rod at stage 25 (50 hours post-fertilization at 27°C; 18–19 somite stage).6 In medaka hio embryos, the liver does not appear until stage 29 and is small and malformed. In addition to this liver defect, hio mutant embryos lack pectoral fins and die after hatching. These phenotypes suggested to us that the study of hio mutants might allow the dissection of various aspects of embryonic specification and perhaps the linking of liver formation to fin formation.
The signaling pathway of vertebrate limb formation has been studied in detail.7 Limbs arise from regions of the lateral plate mesoderm (LPM) at specific positions along the main anteroposterior (AP) body axis. A number of studies have shown that the limb-inducing signal originates in the axial mesoderm and is relayed from there to the LPM. In mouse, chick, and zebrafish, this signal is thought to be retinoic acid (RA), the bulk of which is synthesized by retinaldehyde dehydrogenase type2 (RALDH2) in early somites and the LPM.8–15 With respect to downstream effectors, molecular studies have clearly shown that RA signaling from the zebrafish somitic mesoderm leads to the expression of the wnt2ba gene in the intermediate mesoderm, which then signals to the LPM and triggers tbx5 expression. Tbx5 is required for Fgf signaling in the fin bud that leads to prdm1 expression, which in turn triggers fgf10 and bmp2b expression.7, 16
In contrast, the identity of an initial hepatic inducer in vertebrates has yet to be validated genetically. In the first report to isolate a single gene regulating vertebrate liver specification, Ober et al.17 characterized an interesting zebrafish mutant called prometheus (prt). In prt embryos, the liver is absent or greatly reduced in size at 50 hours post-fertilization but may start to develop and “catch up” to normal size at a later stage. Positional cloning and further analysis revealed that the prt mutation altered the wnt2bb gene (the second wnt2b gene) and that prt/wnt2bb was expressed in restricted bilateral domains in the LPM directly adjacent to the liver-forming endoderm. Subsequently, Shin et al.18 reported that Fgf and Bmp signaling pathways play important roles in zebrafish liver specification and raised the possibility that these molecules act downstream of Wnt2bb. However, the molecules that act upstream of Wnt2bb during liver specification remain to be identified.
In this study, we carried out a detailed characterization of our medaka hio mutants, whose signature phenotypes are a small liver and no pectoral fins. Our results define hio as a missense mutation of the raldh2 gene, the expression of which likely results in a nonfunctional RALDH2 protein that cannot support fin development. We also show that the hio mutation causes a retardation of liver budding that resembles that observed in zebrafish prt mutants, and that wnt2bb expression is undetectable in hio LPM. Our data suggest that the role of RA signaling in the specification of both liver and fins is to induce expression of wnt2b family genes.
Medaka were raised and maintained under standard laboratory conditions at approximately 27°C. Heterozygous carriers of the hio mutation were identified by random intercrosses. To obtain homozygous hio mutant embryos, heterozygous carriers of the hio mutation were mated. Typically, the eggs were spawned synchronously every morning. Embryos were raised at 30°C, and embryonic stages were determined based on morphological features, as previously described.6
The hio mutation was induced in the Cab-Kyoto line of medaka.3 The Kaga-Kyoto line of medaka was used for polymorphic marker-based genetic mapping.3 Genetic mapping and chromosome walking were performed essentially as described.19
Reverse-Transcription Polymerase Chain Reaction and Gene Segment Alignment.
Partial or full-length complementary DNAs of the raldh2 (Accession number AB439727), tbx5 (AB439834), wnt2bb (AB439835), wnt2ba, cp, prox1, insulin, and tbx3 genes were generated by reverse-transcription polymerase chain reaction of messenger RNAs (mRNAs) from various stages of medaka embryos (Supporting Table 1). Alignment was performed using MultAlin (http://prodes.toulouse.inra.fr/multalin/multalin.html).
Injection of mRNA.
WT raldh2 mRNA (400 pg), obtained by in vitro transcription of a pBS-KS(−)-raldh2 clone, was injected into the cytoplasm of one-cell stage embryos that were the progeny of intercrossed hio heterozygotes.
Gene Knockdown by Morpholinos.
Morpholino oligonucleotides (MOs) were synthesized by Gene-Tools, LLC (Corvallis, OR). MOs (0.8 pmol) were injected into the cytoplasm of one-cell stage WT medaka embryos. The sequences of MOs used were as follows:
raldh2 MO, 5′-ATGACTGCCGTGGCTGCGCTGCTGT-3′;
wnt2bb MO, 5′-ATATACCTGAGAGTGTCCAGAACAG-3′.
Retinoic Acid Treatments.
Embryos resulting from hio heterozygote intercrosses were incubated in the dark from stage 21 onward in various dilutions of a 10−2 M all-trans RA (Sigma) stock solution in dimethylsulfoxide. The diluent was 1× balanced salt solution composed of 110 mM NaCl, 5 mM KCl, 1 mM CaCl2, and 2.2 mM MgSO4, pH7.5. Teratogenic effects (such as disrupted heart and AP axis) were observed at 10−8 M all-trans RA and above.
Whole-Mount In Situ Hybridization.
Whole- mount in situ hybridization was performed as previously described,3 using antisense DIG-labeled riboprobes generated from medaka tbx5, wnt2ba, prox1, cp, insulin, wnt2bb, tbx3, or raldh2 partial or full-length complementary DNAs. Probes used to detect gata6, foxA3, ck19, and pdx1 expression were as previously described.4
N-([6-(2,4-Dinitro-Phenyl) Amino] Hexanoyl)-1-Palmitoyl-2-BODIPY-FL-Pentanoyl-sn-Glycero-3-Phosphoethanolamine–Mediated Tracking of Lipid Metabolism.
Medaka embryos at stage 36 were placed in 0.5 mL 1× balanced salt solution containing 0.3 mg/mL N-([6-(2,4-dinitro-phenyl)amino]hexanoyl)-1-palmitoyl-2-BODIPY-FL-pentanoyl-sn-glycero-3-phosphoethanolamine (PED6) and incubated in the dark for 4 hours at 28°C. The treated embryos were rinsed with 1× balanced salt solution and placed in a glass depression slide. PED6 fluorescence was detected using a Zeiss Axioplan 2 microscope.
The hio Mutation Alters the raldh2 Gene.
Using bulked segregation analysis, we performed positional cloning and mapped hio between restriction fragment length polymorphisms OLc2806f and Scaf21_1.0M on LG3 (Fig. 1A). This region includes a sequence with homology to the mammalian and zebrafish raldh2 genes. Because the “missing fin” phenotype of medaka hio mutants was similar to that of the zebrafish raldh2 mutants neckless (nls) and no-fin (nof),8, 10raldh2 appeared to be a good candidate for the gene affected by the hio mutation. We compared the sequence of a genomic fragment encoding the WT medaka raldh2 gene with the sequences of the corresponding fragments from four independent homozygous hio embryos. We found an A to G transversion in hio alleles that would cause the threonine 468 residue in the WT RALDH2 enzyme to be replaced by alanine (Fig. 1B). A comparison of the predicted WT RALDH2 amino acid sequences among medaka, human, xenopus, and zebrafish revealed an overall amino acid sequence identity of 81% (between medaka and human or xenopus) and 84% (between medaka and zebrafish) (Fig. 1C). The threonine 468 residue was conserved among all species examined. Moreover, threonine 468 lies within the catalytic domain of WT RALDH2 (Fig. 1C). These results suggest that the mutant RALDH2 protein produced in hio mutants is inactive.
It has been well established that the defects of RA signaling lead to the impairment of fin development in zebrafish.7, 8, 10, 16 We showed that the injection of RALDH2-MO into WT embryos results in the impairment of fin development, and the injection of raldh2 mRNA or exogenous RA rescued the defects of fin development of hio mutant (Supporting Fig. 1). These results indicate that RALDH2 and RA regulate fin development in medaka. In addition, hio embryos lacked tbx5 and wnt2ba expression, which acted downstream of RA during fin development (Supporting Fig. 2). Taken together, we concluded that RA signaling plays important roles in fin development in medaka.
Loss-of-Function of RALDH2 Reduces Liver Size in Medaka.
We have previously reported that the medaka hio mutation results in a small and malformed liver.3 To examine the role of raldh2-dependent signaling in liver formation in medaka, we employed three approaches. First, to investigate whether loss-of-function of raldh2 could account for this liver defect, we injected raldh2-MO into WT embryos and inspected the developing liver. We found that the raldh2 morphants had the same undersized livers as the hio mutants (Fig. 2A). Estimation of liver size via in situ hybridization using a gata6 probe confirmed the reduced liver size in the raldh2 morphants (Fig. 2B). Second, to determine whether the hio/raldh2 mutation was responsible for the small livers of these mutants, we injected in vitro transcribed raldh2 mRNA into the cytoplasm of one-cell stage embryos that were the progeny of intercrossed hio heterozygotes and used gata6 in situ hybridization to assay these embryos for rescue of liver size. As expected, 25% of the progeny of intercrossed hio heterozygotes (uninjected controls) had small livers. In contrast, the percentage of progeny with decreased liver size was reduced to 14% after injection of raldh2 mRNA (Fig. 2C). Finally, we investigated whether treatment with exogenous RA, the bulk of which is synthesized by RALDH2, could rescue the liver defects caused by the hio mutation. We treated the progeny of intercrossed hio heterozygotes with all-trans retinoic acid (atRA) and monitored liver development. Whereas 25% of the untreated progeny of intercrossed hio heterozygotes had small livers, the percentage of progeny with a small liver was reduced to 13% after exposure to 5 × 10−9 M atRA (Fig. 2D). Thus, treatment with either WT raldh2 mRNA or exogenous RA can rescue the small liver phenotype in at least some hio mutants, although the efficiency of such rescue is much lower for the liver than for the pectoral fin. When the livers of hio mutants with treatment with either WT raldh2 mRNA or exogenous RA became as large as that of WT medaka, we judged it to be rescued. Therefore, we may have underestimated the recovery rate of liver phenotype. In any case, the loss of raldh2 function in hio mutants causes a defect not only in pectoral fin development but also in liver formation.
The hio Mutation Retards Liver Specification.
Although the molecular mechanism by which RA signaling initiates fin development is well established,7, 20 the molecular regulation of liver development by RA signaling remains to be elucidated. To address this issue, we used in situ hybridization with a probe specific for the endodermal marker foxA3 to monitor liver development in hio embryos. Whereas hepatic buds were observed in WT medaka at stage 25, these structures did not form in hio mutants until stage 29 (Fig. 3A). By stage 32, hepatic buds were noticeably smaller in hio embryos compared with the WT. These data indicate that the medaka hio mutation retards hepatic bud formation.
Next, we determined whether the hio mutation interferes with the initial specification of liver anlage in medaka. We carried out in situ hybridization using a probe for the hepatic specification marker prox1 to monitor liver specification. In WT medaka embryos, prox1 was induced in the hepatic bud starting at stage 25 (Fig. 3B, upper panel), and by stage 29, prox1-positive cells were observed only in the hepatic region. In hio embryos, the formation of the hepatic bud was delayed until stage 29 (Fig. 3A), so that prox1-positive cells were not observed in the hepatic region until this stage (Fig. 3B, bottom panel). These results indicate that the hio mutation compromises the signaling pathway required for initial hepatic fate specification.
The Small Livers in hio Embryos Exhibit Normal Hepatic Cell Differentiation and Function.
The most important cell types in the vertebrate liver are cholangiocytes (bile duct cells) and hepatocytes. To determine whether hio livers were capable of normal hepatic cell differentiation, we subjected WT and hio embryos to in situ hybridization with a probe for the cholangiocyte marker cytokeratin19 (ck19) and the hepatocyte marker ceruloplasmin (cp). At stage 28, although WT embryos showed a few ck19-positive cells in the hepatic region, hio embryos did not (Supporting Fig. 3). However, by stage 32, ck19 expression was comparable in WT and hio livers (Fig. 4A, left panel). Furthermore, cp expression was comparable in WT and hio livers at stage 34 (Fig. 4A, right panel). Thus, although liver formation is delayed in hio embryos, the small livers of these mutants can give rise to differentiated liver cells. To ascertain whether the small hio liver was functional, we took advantage of a reporter system based on PED6, a fluorescent phospholipid.21 When WT medaka ingest PED6, endogenous lipase activity and the rapid transport of cleavage products results in intense gallbladder fluorescence.4 We observed equivalent levels of green fluorescence in the gallbladders of WT and hio embryos treated with PED6 at stage 36 (Fig. 4B), indicating that hio livers have a normal capacity to metabolize lipids. Taken together, our results show that loss of raldh2 function initially impairs liver specification and retards liver development but does not impair hepatic cell differentiation or liver functions at later stages of embryogenesis.
Stafford and Prince22 have reported that hepatic and pancreatic cell markers are undetectable in zebrafish nls embryos. To investigate whether the hio mutation affected pancreas development in medaka, we carried out in situ hybridization using probes for the pdx1 and insulin genes. We observed pdx1-expressing cells in the pancreatic primordium region in both WT and hio embryos at stage 28 (Fig. 4C). Furthermore, insulin-expressing cells were present in both WT and hio embryos at stage 30 (Fig. 4D). Thus, unlike its effects on liver development, the medaka hio mutation does not appear to affect pancreas development. This result stands in contrast to the zebrafish nls mutation, which severely impairs the development of both the liver and the pancreas.
wnt2bb Expression in the LPM Is Lost in hio Embryos.
It has been shown in zebrafish that mesodermal wnt2bb expression promotes liver specification.17 It is also known that the wnt2ba gene acts downstream of RA signaling and regulates pectoral fin development in zebrafish.7, 20 The wnt2ba and wnt2bb genes are both members of wnt2b gene family that exists in both zebrafish and medaka. These observations suggested to us that Wnt2bb might be a good candidate for the key molecule regulating piscine liver specification downstream of RA signaling. To explore this hypothesis, we examined wnt2bb expression in hio embryos. At stage 22, no wnt2bb expression in the LPM was observed in either WT or hio embryos (Supporting Fig. 4). However, by stage 24, wnt2bb expression in the LPM directly adjacent to the liver-forming endoderm was induced in WT embryos but not in hio embryos (Fig. 5A, left panel). These results suggest that the hio mutation causes a loss of wnt2bb gene expression. Interestingly, wnt2bb still had not been expressed in the hio LPM at stage 29, when the liver bud forms (Fig. 5A, right panel). Thus, the small livers that eventually appear in medaka hio mutants seem to form independently of Wnt2bb signaling, just as occurs in zebrafish prt mutants.
To investigate whether wnt2bb positively regulates liver specification in medaka as it does in zebrafish, we injected wnt2bb-specific morpholino antisense oligonucleotides (wnt2bb-MO) into the cytoplasm of one-cell stage WT embryos and evaluated the outcome by in situ hybridization using a prox1 probe. We found that, like hio embryos, WT medaka embryos that had been injected with wnt2bb-MO lacked prox1 expression (Fig. 5B). These results suggest that Wnt2bb signaling is responsible for liver specification in medaka.
In conclusion, our study has shown that the hio mutation in medaka impairs liver specification by abrogating wnt2bb expression. Our data are thus the first genetic evidence that RA signaling positively regulates liver specification by inducing wnt2bb expression.
Function of RA Signaling in Pectoral Fin and Liver Development in Medaka.
In this study, we examined the role of RA signaling during embryogenesis by characterizing medaka hio mutants. These mutants bear an alteration to the raldh2 gene (Fig. 1) that encodes the enzyme principally responsible for RA synthesis, and we interpret that this is a nearly null mutation because the phenotypes of hio mutant are similar to that of RALDH2 morphants (Fig. 2 and Supporting Fig. 1). The hio mutants exhibit two prominent phenotypes: missing pectoral fins and a small liver (Fig. 2 and Supporting Fig. 1). Work in mouse, chick, and zebrafish has shown that RA signaling from the somitic mesoderm is essential for limb induction and is mediated by the expression of downstream factors such as wnt2ba and tbx5.7–14 We show that the hio mutation in medaka leads to defects in pectoral fin development and tbx5 and wnt2ba expression (Supporting Fig. 2). Thus, our results indicate that RA signaling is crucial for fin specification in medaka and show that limb induction signaling is conserved across a broad range of species (Fig. 6, right part). Significantly, our work has also uncovered a role for RA signaling in liver development. We have demonstrated that the hio mutation retards the formation of hepatic buds from the foregut (Fig. 3A) and causes a profound defect in liver specification (Fig. 3B). In addition, we have shown that the wnt2bb expression required for the regulation of liver specification is undetectable in the LPM of hio embryos (Fig. 5A). Our data constitute the first genetic evidence that RA signaling regulates vertebrate liver specification by inducing wnt2bb gene expression (Fig. 6, left part). Previously, Wang et al.23 reported that liver growth is severely affected in RALDH2-deficient mouse embryos. Thus, RA signaling in liver specification may be conserved among other species.
Molecular Mechanisms Regulating the Development of Pectoral Fins and Liver.
There are several similarities in the signaling pathways governing pectoral fin and liver organogenesis. During zebrafish pectoral fin development, RA signaling induces wnt2ba expression, which in turn induces tbx5 expression. Tbx5 is a key molecule that regulates the expression of downstream effectors such as the fgf and bmp family members fgf24, fgf10, and bmp2b.7, 16 Thus, limb induction requires a sequential RA → Wnt → Tbx → Fgf + Bmp signaling cascade. A parallel situation may exist for liver specification in medaka. Wnt2bb, Tbx3, Fgf, and Bmp have all been shown to positively regulate the development of this organ in mice or zebrafish.17, 18, 24 In this study, we showed that RALDH2 drives wnt2bb expression during liver specification in medaka (Fig. 5). Based on the proposal of Shin et al.18 that Fgf and Bmp act downstream of Wnt2bb during liver specification, the sum total of all these results suggests that liver specification also requires a sequential RA → Wnt → Fgf + Bmp signaling cascade. Intriguingly, we found that RA signaling induced tbx3 expression in medaka (Supporting Fig. 5). However, our morpholino studies showed that RA signaling associated with liver formation can regulate tbx3 expression without involving Wnt2bb (Supporting Fig. 5). These data indicate that Tbx3 can act downstream of RA signaling, but it is likely that other T-box family members are involved in the putative RA → Wnt → Tbx → Fgf + Bmp signaling cascade that drives liver development. We are continuing our search for the identity of this transcription factor.
Differential Requirements for Liver and Pectoral Fin Specification During Embryogenesis.
A sequential RA → Wnt → Tbx → Fgf + Bmp signaling cascade is indispensable for the limb induction process that underlies pectoral fin development. Alterations in raldh2 such as the medaka hio and zebrafish nls and nof mutations lead to an absence of pectoral fins, as does knockdown of wnt2ba using MO in WT zebrafish.8, 10, 16 Notably, these mutants and morphants never form pectoral fins during the entire course of embryogenesis. Conversely, a sequential RA → Wnt → Fgf + Bmp signaling cascade is not indispensable for liver specification, because medaka hio mutants and zebrafish prt mutants are able to form a functional liver at an abnormally late stage of development. A molecule that may be able to partially compensate for a loss of RALDH2 is Fgf10, which is also induced downstream of RA signaling and involved in limb and liver formation. Loss of fgf10 prevents fin development in zebrafish,7 and Fgf10-deficient mouse embryos lack limbs and have an abnormally small liver.25, 26 Thus, fgf10 and raldh2 functions may cooperate during embryogenesis such that their mutation results in similar phenotypes. Moreover, in zebrafish fgf10 mutants, the hepatopancreatic ductal epithelium is severely dysmorphic, and cells of the hepatopancreatic ductal system and adjacent intestine misdifferentiate and adopt a hepatic or pancreatic fate.27 These results indicate that Fgf10 functions to repress the differentiation of hepatopancreatic ductal epithelium into hepatic or pancreatic cells and thus demarcates developing organs and tissues. In our hio mutants, it may be that the observed lack of liver specification leads not only to impaired liver development but also to misdifferentiation in the hepatopancreatic ductal system that results in the formation of a small liver. Such misdifferentiation could obscure an absolute requirement of raldh2 for liver specification, and might create an obstacle to finding mutations that specifically interfere with the initial specification of the liver anlage. Further analysis is needed to substantiate this hypothesis.
Comparison of raldh2 Alterations in Medaka hio and Zebrafish nls Mutants.
The nls mutation in zebrafish is a loss-of-function allele of the raldh2 gene that was generated by the ENU approach. Originally, nls was isolated in an in situ hybridization screen and was detected by its effects on neural AP patterning.8 The nls embryos lack pectoral fin buds and fins. A similar phenotype has been reported for a natural loss-of-function raldh2 mutation in zebrafish called no-fin.10 In addition to their lack of fins, nls embryos do not express the hepatocyte and pancreatic cell markers that are detectable in WT zebrafish embryos.22 Stafford and Prince22 also showed that exogenous RA treatment of WT zebrafish embryos resulted in the anterior expansion of the pancreatic anlage. Thus, RA signaling is a determinant of the regionalization of both neuroectoderm and endoderm, and defects in raldh2 function prevent the development of the endodermal region in which liver and pancreatic cells would normally appear. In contrast, our medaka hio mutation does not have severe effects on neuroectoderm and endoderm regionalization, and the liver in hio embryos, although reduced in size and delayed in appearance, eventually forms in the normal location. Thus, hio is a unique mutation affecting liver organogenesis, and continued study of this mutation should yield new insights into the involvement of RA signaling in liver specification. It remains to be elucidated how medaka hio mutants escape the defect in endodermal regionalization associated with zebrafish nls mutations.
The availability of two closely related fish model systems, medaka and zebrafish, for studies in genetics, experimental embryology, and molecular biology is unique among vertebrates and advantageous for two reasons. First, the evolutionary distance between these two species is particularly well suited for comparative functional genomics. Second, and more importantly, the parallel existence of medaka and zebrafish transforms the perceived weakness of studying genetics in fish, namely, the many analogous groups of genes formed because of genomic duplications, into an advantage: the study of a gene in one species may shed light on a gene function that is hidden in the other species.28 For example, RALDH2's function in AP patterning is not apparent in medaka hio mutants, and RALDH2's function in liver specification is not apparent in zebrafish nls mutants. Our results clearly demonstrate that a comparison of two related species can be a powerful means of dissecting genetic and molecular mechanisms underlying vertebrate development.
The authors thank numerous members of the Nishina and Katada laboratories for excellent fish care, technical assistance, and helpful discussions.