Medaka strain

The medaka (Oryzias latipes) bbRR-strain was obtained from a local pet shop and maintained at 28°C under a 14 hr light / 10 hr dark regime at the National Research Institute of Aquaculture (Mie, Japan). The eggs were harvested at room temperature (24–26°C). The developmental stage of the embryo is described in terms of the stage number according to Iwamatsu (1993).

Cloning of medaka Fox genes

We amplified the forkhead box fragments from medaka genomic DNA using the polymerase chain reaction method with primer-1, AAG CCC CC (TA) TAC TC (CG) TAC AT and primer-2, GCG (GC) AT AGA GTT CTG CCA. After cloning the 150 bp fragments into pBluescript II (Stratagene), we used one of the clones as a probe to screen the medaka embryonic cDNA library. Prior to use, we determined that this probe fragment was not homologous to both MeHNF3β and Mefkh1. Poly-A RNA was isolated from 500 medaka embryos from the mid-blastula to the late-gastrula stage (total RNA extraction kit and mRNA purification kit, Pharmacia). The embryonic cDNA was synthesized using an oligo-dT primer (Timesaver cDNA synthesis kit, Pharmacia). After screening 500,000 clones from the medaka cDNA library, two types of Fox gene were identified by DNA sequencing of both strands (ABI PRISM, Applied Biosystems).

For Mefkh1, we determined the sequence of the 3′-flanking region up to the poly A signal using the 3′RACE method (Marathon cDNA amplification kit, Clontech)

Southern blotting and Northern blotting

Southern blotting and Northern blotting were conducted according to ‘Current Protocols in Molecular Biology’ (Wiley, USA). We used Hybond-N, -N+, and -NX (Amersham, UK) membranes for DNA or RNA blotting. We labeled the DNA probe using the BcaBEST^TM labeling kit (Takara, Tokyo).

We extracted whole genomic DNA from adult medaka in using 4M urea TNES buffer. DNA extraction was performed according to the method of Nakayama (1995). The extracted DNA was digested with nine restriction enzymes (Apa I, BamH I, EcoR I, Hae III, Hind III, Kpn I, Pst I, Sac I, Xho I) at 37°C overnight. Southern hybridization was performed at 65°C, and the blotted membranes were washed at high stringency, twice with 2×SSC /1 % SDS and twice with 0.2×SSC / 0.1 % SDS, at 65°C for 15 min each time.

To obtain total RNA from each tissue, we dissected an adult medaka that had been paralyzed with FA100 (Tanabe, Osaka), and eluted the RNA from each tissue using Tri-sol (Gibco-BRL, USA). We also used Tri-sol to extract total RNA from medaka embryos at various development stages. Northern hybridization was performed at 42°C overnight. Northern blotted membranes were washed at high stringency using the same buffers and for the same number of times as in Southern hybridization.

We used the entire sequence of MeHNF3β as a template for DNA labeling. For Mefkh1 labeling, we used positions #142-#771 of the sequence of Mefkh1 as a template. The autoradiogram signals were detected with a BioMax film (Kodak, USA).

Whole-mount in situ hybridization

Whole-mount in situ hybridization was carried out essentially according to the protocol of Hauptmann and Gerster (1994). Prehybridization and hybridization were performed at 55°C. The digoxigenin(DIG)-UTP-labelled RNA probes (Boehringer Mannheim, Germany) were generated from deletion clones retaining the N-terminus coding region (D59 for MeHNF3β; positions #0-#572 and R17 for Mefkh1; positions #11-#471) of the medaka Fox genes as a template (Fig. 1). For synthesizing the RNA probes, we used a DIG RNA labeling kit (SP6/T7)(Boehringer Mannheim), and a DIG nucleic acid detection kit for its detection (Boehringer Mannheim). The RNA probes were incubated overnight with embryos from the blastula to the hatching stage. The hatching enzyme gene (choriolysin H, HCE) was provided by Dr. Inohaya (Tokyo Institute of Technology).

Fig. 1

The cDNA sequences and the comparison of the forkhead boxes of medaka Fox genes. (A) The Mefkh1 cDNA length is 2,061 bp long. The coding region is from positions #308 to #1,666, and the forkhead box is surmised to be between positions #728 and #1,057. (B) The MeHNF3β cDNA length is 1,715 bp long. The coding region is from positions #220 to #1,464, and the forkhead box is surmised to be between positions #640 and #969. The forkhead box is underlined. The transactivating domains are region-II (double underline), region-III (dot lined square) and region-IV (thick lined square). The serine clusters are indicated in thin lined square. The in frame stop codons are indicated in thick underline. (C) The forkhead box, specific DNA-binding domain, is comprised of 110 amino acids. The medaka Fox genes are classified in class-I or FoxA. The class-I has four subclass-a-d (FoxA1-4). Subclass specific-amino acids are indicated in square. Am; Amphioxus, Ax; axolotl, H; human, M; mouse, Me; medaka, R; rat, S; salmon, X; Xenopus, Z; zebrafish.

Zoo blotting

The medaka and zebrafish were purchased from a pet shop and bred in our laboratory. The rainbow trout (Oncorhynchus mykiss) were bred at our field station. Japanese frogs (Hyla arborea japonica) and turtles (Clemmys japonica) were obtained from the field. A chicken was purchased from a grocery store. Rockhopper penguin (Eudyptes chrysocome) and short-fin pilot whale (Globicephala melaena) DNA samples were provided by Dr. Kage (Mie University). Mouse DNA was provided by Dr. Kudo (Tokyo Institute of Technology). All DNA samples were digested with Dra I and applied to a gel at 5 μg/lane. DNA probes were synthesized using the 5′- specific region of each medaka Fox gene (see whole-mount in situ hybridization). The probes were designed so as to delete the forkhead-box-encoding region. Southern hybridization was conducted according to the method of Church and Gilbert (1984). Prehybridization was carried out for 1.5 hr at 68°C, and the hybridization was performed at 65°C overnight. The washing procedure was the same as that used in Sourthern and Northern hybridization.

Sequencing of medaka Fox genes

To clone the Fox genes, we screened about 500,000 medaka cDNA clones using the PCR-amplified DNA fragment. We identified two types of Fox gene among the seven positive clones. After DNA sequencing, the sequence of the probe fragment was not found in these two Fox genes and the probe fragment thus belonged to another Fox gene.

As a result of the DNA homology search, we found that one clone was a novel Fox gene, Mefkh1. The Mefkh1 cDNA was 2,061 bp long and the predicted coding region was from positions #308 to #1,666 (DDBJ; accession No.  AB001573) (Fig. 1A). The region encoding the forkhead box was surmised to be between positions #728 and #1,057 (underlined in Fig. 1A). The predicted molecular weight of this gene product was estimated to be 48,973 daltons and the protein consisted of 453 amino acids. The other gene, MeHNF3β (medaka- HNF3β), was 1,715 bp long which included an open reading frame from positions #220 to #1,464 (DDBJ; accession No.  AB001572) (Fig. 1B). In this coding region, there is a conserved region encoding the forkhead box between positions #640 and #969 (underlined in Fig. 1B). The amino acid sequence deduced from the DNA sequence is 415 amino acids long and the molecular weight of the protein was estimated to be 45,720 daltons.

Sequence comparison of the forkhead box

The Fox genes have been classified into various types and identified in various organisms (Kaufmann and Knöchel, 1996, Odenthal and Nüsslein-Volhard, 1998). We compared the conserved DNA-binding domain, the forkhead box, of the medaka gene products with those of other species. MeHNF3β is classified into subclass-Ia or Fox A2, while Mefkh1 is grouped into subclass-Id, FoxA3 (Fig. 1C). The product of MeHNF3β consists of subclass-Ia-specific amino acids (squared column in Fig. 1C) and this gene is most closely related to the zebrafish axial/fkd1. The MeHNF3β product showed 99% identical to the axial/fkd1 product in the forkhead box region and 88 % identity in the full-length amino acid sequence. The Mefkh1 product is composed of subclass-Id-specific amino acids, and exhibits the highest similarity to the zebrafish fkd2/zffkh1 and Atlantic salmon forkhead products (Fig. 1C). The Mefkh1 product is 95% identical to the fkd2/zffkh1 product in the forkhead box region and 58% identity in the full-length predicted amino acid sequence.

Expression of MeHNF3β and Mefkh1 in adults and embryos

We determined which adult tissues express the Fox genes. Northern blotting analysis was performed using total RNA from the various tissues. The results showed that Mefkh1 was clearly expressed in the liver, gut and skin, but the expression levels were low compared to those of MeHNF3β. We did not find differences in the Mefkh1 expression level between males and females (Fig. 2A). On Northern blots, the Mefkh1 signal was observed at about 2.0 kb. MeHNF3β was expressed in the gills, liver and gut (Fig. 2B), and we also found a weak total RNA signal from the brain (data not shown). These signals appeared to be around 1.7 kb. No major difference in the expression levels was found between the tissues of males and females.

Fig. 2

Expression of Mefkh1 and MeHNF3β in adults and embryos. (A) Mefkh1 probe hybridized with the total RNA that was extracted from liver, gut and skin. (B) MeHNF3β probe hybridized with the total RNA that was extracted from gill, liver and gut. Total RNA was applied (40 mg) in each lane. M; male, F; female. (C) The expression of Mefkh1 in embryos. (D) The expression of MeHNF3β in embryos. 2-16; 2-16 cell stage, Mor; morula, Bla; blastula, Gas; gastrula, HF; head-forming stage, 2s; twosomite stage, 16s; 16-somite stage, 18s; 18-somite stage, 30s; 30-somite stage, Soc; somite-forming completion stage, Vc; vessels-forming completion stage and Hatch; hatching fry. Arrowheads indicate positions of transcripts of Fox genes. 28S; 28S ribosomal RNA (3.6kb), 18S; 18S ribosomal RNA (1.7kb). Each lane was applied at 30 μg of total RNA from whole embryos.

We also determined when these genes begin to be expressed in the embryo. The Mefkh1 signal was detected from the gastrula to the vessel-forming completion stage (Fig. 2C). The autoradiograms were overexposed to intensify the weak signals which were observed at around 2.0 kb. The Mefkh1 expression peaked around the somite-forming completion stage. Mefkh1 mRNA was not detected in the maternal unfertilized egg. The MeHNF3β signal was detected from the gastrula to the hatching stage (Fig. 2D). The maximum level of expression of MeHNF3β was observed around the two-somite stage to the 16-somite stage. This MeHNF3β expression peaked at an earlier stage compared with that of Mefkh1, and included the nine-somite stage at which the notochord tissue is visible (Iwamatsu, 1993). This implies that the MeHNF3β product plays a role in notochord development.

Whole-mount in situ hybridization of Fox genes

We analyzed the expression pattern of the medaka Fox genes in embryos using whole-mount in situ hybridization. We used specific RNA probes to prevent cross hybridization among the Fox genes (see Materials and Methods).

Mefkh1 was expressed in the dorsal marginal zone and organizer at stages 12–13 (from the late blastula to early midgastrula stage). In stage-17 embryos (head-forming stage), a group of Mefkh1-positive cells appeared at the rostral end (Fig. 3A, arrow) which is the peak of a mass of cells known as the polster. This tissue includes precursor cells of hatching enzyme grand cells. Mefkh1 expression was also observed along the body axis, but it was very weak. Notably, this Mefkh1 expression pattern was different from that of the zebrafish fkd2/zffkh1. The axial tissue expression of fkd2/zffkh1 was very similar to that of MeHNF3β. MeHNF3β expression was detected as a sharp stripe along the midline of the body (Fig. 3E, arrow). In stage-19 embryos (two-somite stage), the group of Mefkh1-positive cells was located at the tip of the head (Figs. 3B and 3C, arrow). At more advanced stages, the expression signal was not detectable in cells which were derived from the polster and which were moving into the inner pharyngeal duct. At stages 27–30 (24-35-somite stage), we also observed Mefkh1 expression in the rudiments of the liver (Fig. 3 D, arrow).

Fig. 3

Whole-mount in situ hybridization of Mefkh1 and MeHNF3β. (A) Mefkh1, at the head-forming stage, the expression was observed in the polster (arrow). (B, C) At two-somite stage, the polster cells expressed Mefkh1 migrate into head region (arrow). (D) At 24-35-somite stage, Mefkh1 was expressed in rudiments of liver (arrow). (E) MeHNF3β is expressed along body axis at head formation stage (dorsal view). (F, G) At two-somite stage, MeHNF3β was expressed in the primordium of notochord, ventral mesoderm (arrow and arrowhead). (H) At 24-35-somite stage, MeHNF3β was expressed in the otic vesicle (arrow) and the floor plate (arrowhead). Dorsal views (A, B, E, F) and lateral views (C, D, G, H), anterior to the left, posterior to the right. (I) The scheme of the expression pattern of FoxA2 and FoxA3. Both medaka and zebrafish FoxA2 express in axial mesoderm of their embryos. Whereas medaka FoxA3 restrictedly expresses in polster (and very slightly in posterior axial mesoderm), zebrafish FoxA3 expresses in polster and axial mesoderm like FoxA2 expression. Strong expression is indicated in black, and weak expression is indicated in gray.

The expression pattern of MeHNF3β was very similar to that of mouse HNF3β. MeHNF3β was expressed in the organizer and dorsal marginal zone. In stage-17 embryos (head-forming stage), MeHNF3β expression was observed in the ventral mesoderm of the trunk and tail (Fig. 3E). This region is thought to be the primordium of the notochord. In stage-19 embryos (two-somite stage), the mRNA signal was also observed in the head region (Fig. 3G, arrowhead). At stages 27–30 (24-35-somite stage), the signals in the ventral mesoderm extended to the floor plate of the predicted mid-brain region and to the tail bud (Fig. 3H, arrowhead). We also observed signals in the otic vesicles (Fig. 3H, arrow). Our results indicate that MeHNF3β was expressed in the notochord and the precursor cells of the floor plate. At more advanced stages, we observed MeHNF3β expression in the primordium of the liver and a transient signal around the anus (stage 32, somite-forming completion stage). In Caenorhabtis elegans, the product of lin-31, a Fox gene, controls the fate of vulval cells (Miller et al., 1993). Thus, we speculate that the MeHNF3β product also plays a role in anus formation in vertebrates.

Southern blotting and zoo blotting

To determine the copy number of Mefkh1 and MeHNF3β in haploids, we performed genomic Southern blotting using nine restriction enzymes (see Materials and Methods). MeHNF3β and Mefkh1 were found to exist in a single copy in the haploid genome (data not shown).

To elucidate the conservation of the Fox genes among various vertebrate species, we performed zoo blotting analysis using medaka Fox gene probes. Hybridization of the Mefkh1 probe with the digested genome was not clear for most of the species except turtle, chicken and penguin (Fig. 4A). The Mefkh1 probe did not hybridize with genomic DNA of the zebrafish, fkd2/zffkh-1, which indicates that medaka FoxA3 has a molecular structure different from that of zebrafish FoxA3. We used the same digested genomic DNA on the agarose gels for zoo blotting of Mefkh1 and MeHNF3β. The washing procedure for Mefkh1 hybridization was carried out under high-stringency conditions as was that for MeHNF3β hybridization (see Materials and Methods). We concluded that the DNA samples were not degraded as shown in Fig. 4B. This result suggests that Mefkh1 is conserved in all organisms, but the molecular structure of its protein product differs among species. Regarding the medaka genomic DNA, two bands were observed with estimated molecular sizes of approximately 850 bp and 6,500 bp. This indicates that Mefkh1 was digested into two separate fragments, because the results of Southern blotting using the nine restriction enzymes showed that there is one copy in the genome. MeHNF3β is conserved in all species examined, from fish to mammals (Fig. 4B). This result shows that the MeHNF3β product has a universal biological function in all chordates and thus may be conserved in each species. For all of the DNA samples derived from the various animals used, one or two clear bands were observed except for the rainbow trout for which at least six bands were observed for the haplotype DNA. This result may indicate that rainbow trout may have originated from a tetraploid ancestor.

Fig. 4

Zoo blotting of Mefkh1 (A) and MeHNF3β(B). The medaka FoxA3 does not hybridize with the zebrafish genome DNA (A), although the medaka FoxA2 hybridizes with it (B). Me; medaka, Zf; zebrafish, Tr; rainbow trout, Fr; frog, Tu; turtle, Ch; chicken, Pe; penguin, Wh; whale, Mo; mouse. Each lane was applied 5 μg of the genome DNA.

Transcriptional activation domains and serine clusters

Analysis of the predicted protein structure revealed that the Fox proteins contain conserved motifs except for the DNA-binding domain. Regions II-IV are conserved in the class I, FoxA proteins (Pani et al., 1992). These regions are predicted to interact with transcriptional activation co-factors. The MeHNF3β and Mefkh1 products retain all of regions II-IV (Fig. 5). We found many serine clusters such as SS, SSS and SSSS in the Fox proteins. The serine clusters are mostly located outside of the DNA-binding domain and are marked on a schematic map of the Fox proteins (indicated by red bars) in Fig.5. The HNF3β and HNF3γ proteins have highly similar serine cluster/scatter patterns between humans and rats (subclass-Ia and Id in Fig. 5). The medaka MeHNF3β and zebrafish axial/fkd1 products appear also to have similar scatter patterns (subclass-Id in Fig. 5). This indicates that these are conserved proteins which have retained the same biological roles throughout species evolution. On the other hand, the serine cluster pattern of the Mefkh1 product exhibits no similarity with that of the fkd2/zffkh1 and Atlantic salmon forkhead products. This result suggests that Mefkh1 is related to the zebrafish fkd2/zffkh1 from the viewpoint of molecular evolution, but the biological roles of the respective products have possibly been modified during the long period of evolution after the organisms had diverged into different species. As a result of a Blast search, some of the serine residues in the cluster are considered to belong to the modification site. For example, Ser 269–271 and Ser 285–287 of the MeHNF3β product and Ser 299–300 of the Mefkh1 product are included in the predicted phosphorylating site for casein kinase II. These serine residues possibly play a role in binding the transcriptional co-factor and regulating the gene expression.

Fig. 5

Domain structures and the serine cluster index of the medaka Fox genes. The DNA-binding domain is the forkhead box (narrow diagonal lined). The transactivating domains are region-II (dotted), region-III (wide diagonal lines) and region-IV (chevroned). The serine clusters are indicated in thin (SS) or thick (SSS or SSSS) bars. H; human, Me; medaka, R; rat, S; salmon, Z; zebrafish.

Differences in expression patterns of Fox genes between two fish species

We compared the molecular structures and expression patterns of FoxA2 and FoxA3 between medaka and zebrafish. The medaka and zebrafish FoxA2 have almost the same molecular structures and expression patterns. On the other hand, the medaka and zebrafish FoxA3 have different molecular structures and expression patterns. Why are there differences in the degrees of conservation between FoxA2 and FoxA3? All class I (FoxA type proteins) genes in vertebrates are expressed in the derivatives of the embryonic shield. The fkd2/zffkh1 gene (zebrafish FoxA3) is expressed in the embryonic shield, notochord, floor plate, polster (Dirksen and Jamrich, 1995) and neural crest cells (Odenthal and Nüsslein-Volhard, 1998). On the other hand, Mefkh1 (medaka FoxA3) is expressed only in the embryonic shield and polster, and slightly in axial tissue (Fig. 3I). The expression pattern of fkd2/zffkh1 in the notochord and floor plate is similar to that of the medaka FoxA2 gene. The biological function of the zebrafish FoxA3 might overlap with that of the medaka FoxA2, while the medaka FoxA3 function may have become specialized and differs from that of medaka FoxA2. Further comparative studies are required to elucidate the universal role of FoxA3.

Correlation between expression pattern of Mefkh1 and hatching enzyme in polster

In fish, the polster is differentiated from the rostral end and derived from the embryonic shield (Shih and Fraser, 1996). Cells in the polster are assumed to be precursors of hatching enzyme (HCE) gland cells. As embryogenesis progresses, cells expressing the HCE in the polster migrate to the head region, and HCE expression is continuous until the cells come to reside in the inner wall of the pharyngeal duct (Inohaya et al., 1995, Inohaya et al., 1997). In stage-17 embryos, the group of Mefkh1-positive cells coincided with the cell mass expressing HCE. While HCE expression is continuously observed in the progenitor of the hatching enzyme gland proceeding to the inner wall of the pharyngeal duct, the Mefkh1 expression level decreases as the cells started migrating to the head. This indicates that Mefkh1 may be required for the hatching enzyme gland development.

Suggested role of Mefkh1 product

Mefkh1 is transiently expressed immediately before the start of organogenesis. For example, its expression in the shield occurs before that in the epiboly, its expression in the polster occurs immediately before the development of the hatching enzyme gland, and its expression in the primordium of the liver occurs before development of the liver. This implies that the Mefkh1 product may play a key role in cell differentiation and organogenesis.