|Home | About | Journals | Submit | Contact Us | Français|
Foxc1a is a member of the forkhead transcription factors. It plays an essential role in zebrafish somitogenesis. However, little is known about the molecular mechanisms underlying its controlling somitogenesis. To uncover how foxc1a regulates zebrafish somitogenesis, we generated foxc1a knock-out zebrafish using TALEN (transcription activator-like effector nuclease) technology. The foxc1a null embryos exhibited defective somites at early development. Analyses on the expressions of the key genes that control processes of somitogenesis revealed that foxc1a controlled early somitogenesis by regulating the expression of myod1. In the somites of foxc1a knock-out embryos, expressions of fgf8a and deltaC were abolished, whereas the expression of aldh1a2 (responsible for providing retinoic acid signaling) was significantly increased. Once the increased retinoic acid level in the foxc1a null embryos was reduced by knocking down aldh1a2, the reduced expression of myod1 was partially rescued by resuming expressions of fgf8a and deltaC in the somites of the mutant embryos. Moreover, a chromatin immunoprecipitation assay on zebrafish embryos revealed that Foxc1a bound aldh1a2 promoter directly. On the other hand, neither knocking down fgf8a nor inhibiting Notch signaling affected the expression of aldh1a2, although knocking down fgf8a reduced expression of deltaC in the somites of zebrafish embryos at early somitogenesis and vice versa. Taken together, our results demonstrate that foxc1a plays an essential role in early somitogenesis by controlling Fgf and Notch signaling through restricting the expression of aldh1a2 in paraxial mesoderm directly.
Vertebrate somitogenesis is reiterated subdivision of paraxial mesoderm that produces bilaterally symmetrical segmented somites beside the notochord during development. As in other vertebrates, zebrafish somitogenesis is initiated at the early gastrulation stage when the stemlike presomitic mesoderm (PSM)3 progenitor cells are specified in the mesoderm by expressing tbx6. With development, molecular oscillator genes like her1, her7, and deltaC, the component genes of Notch signaling, act by dynamically and periodically turning on and off their expression across the PSM in a posterior to anterior fashion to generate the segmentation clock. Along the expressions of molecular oscillator genes, front determination begins the segmentation program characterized by antagonizing the rostrocaudal retinoic acid (RA) morphogen gradient with caudorostral Fgf8a/Wnt3a gradient (1). Following front determination, somite boundary formation begins by the expression of mesp genes (1). Clustering of integrin α5 and deposition of fibronectin at the new furrow leads to the creation of somite boundary. Then somites maturate and differentiate into definitive somitic derivatives, including sclerotome, dermatome, and myotome, by expressing key regulatory genes, such as myod1 for myotome (1). In zebrafish embryos at the 1-somite through 6-somite stage, only adaxial cells abundantly express myod1. After the 6-somite stage, lateral paraxial mesodermal cells of the embryos express more myod1 gradually (2). Adaxial cells differentiate into slow muscle fibers of the adult fish, whereas lateral paraxial mesodermal cells differentiate into fast muscle fibers in the future (2, 3). These processes are finely regulated by multiple signaling pathways, including Notch, RA, Fgf, Wnt, Shh, and Nodal (1).
Notch signaling plays a crucial role in somitogenesis. Zebrafish her1 mutants exhibit disruption of the three anterior-most somite borders, whereas her7 mutants display somite border defects restricted to somites 8 ± 3 to 17 ± 3 along the anterior-posterior axis (4). In the Notch ligand deltaC mutant fish beamter, the expression of myod1 after the fourth somite is disrupted (4). RA is an important morphogen. Formed somites and anterior PSM express RA-synthesizing enzyme aldh1a2, which generates RA to diffuse into the somite region. Excess RA signaling in the gastrula stage leads to more somites in PSM, whereas insufficiency of RA, on the other hand, leads to fewer somites (5). Fgf signaling functions by antagonizing RA in the determination front. With Fgf inhibitor treatment in chick or zebrafish embryos, larger somites are formed (6, 7). In the zebrafish wnt3a mutant, clock and segmentation process are impaired, and somite length increases (8, 9). Mutations in zebrafish integrin α5 disrupt anterior somite formation. Double mutants of the Notch pathway gene and integrin α5 display somite defects along the entire body axis, with a complete loss of the mesenchymal-to-epithelial transition and fibronectin matrix assembly in the posterior (10). In oep;spt double mutants (Nodal receptor cofactor gene oep and T-box transcription factor gene spt), the somite is not formed (11). Ntl also plays an essential role in somitogenesis. Zebrafish ntl (homologous to the mouse Brachyury gene) mutants display a shortened body. Its tail somite is missing (12). In the zebrafish smoothened (smu, Shh signaling pathway genes) mutant, the expression of myod1 is reduced at the adaxial cells, but its expression in the somite mesoderm is unaffected (5, 13).
Foxc1a is a member of the forkhead box (Fox) transcription factor family. Knocking down foxc1a using morpholino (MO) leads to missing somites in zebrafish embryos (14). However, little is known about the molecular mechanisms underlying its controlling somitogenesis. To uncover how foxc1a regulates zebrafish somitogenesis, we generated two lines of foxc1a knock-out zebrafish using TALEN (transcription activator-like effector nuclease) technology. Unlike foxc1a morphants reported previously that completely missed somites (14), we found that foxc1a null embryos displayed a reduced size of the first six somites in the embryos at the 9-somite stage. Further analyses revealed that Foxc1a affected the expression of myod1 by controlling Fgf and Notch signaling through directly restricting the expression of aldh1a2 in the paraxial mesoderm of embryos at early somitogenesis.
Zebrafish were housed in the zebrafish facility of the Model Animal Research Center, Nanjing University, in accordance with the Institutional Animal Care and Use Committee-approved protocol. The embryos were staged as described previously (15).
RA, 4-diethylamino benzaldehyde (DEAB; inhibitor of Aldh1a), N-[N-(3,5-difluorophenacetyl-l-alanyl)]-(S)-phenylglycine t-butyl ester (DAPT; Notch signaling inhibitor), and DMSO were purchased from Sigma-Aldrich. 1.0 × 10−6 m RA and 1.0 × 10−4 m DAPT were used to treat zebrafish embryos from 8 hpf to the 9-somite stage, respectively. 2.0 × 10−6 m DEAB was used to treat zebrafish embryos from 0 hpf to the 9-somite stage. The final vehicle concentration of DMSO in each treatment, and control was 0.1%.
RT-qPCR was performed to examine the relative expression level of foxc1a versus actb1 (for control), as described previously (16). The sequences of primers for amplifying foxc1a cDNA were TTTACTACCCCGTGGTGGAC (forward) and CGTCTGACGCATTTCAACAC (reverse) (17).
To make knock-out zebrafish, we designed two pairs of TALENs (T-1 and T-2) that target the exon encoding functional domains of Foxc1a using online tools (TAL Effector Nucleotide Targeter version 2.0) as we reported previously (18, 19). The target sequence of T-1 was ATAACACCCACGTTGTCCCTGAATTATTCTCCCAATCAGTCGTCCGTGT-3′, and that of T-2 was CTCACGCGGCCCACGACCAGTACCCCGCCAGCATGGCGAGGGCATATGGGCCATAC (left and right arms are underlined). The expression plasmids of the TALENs were constructed using the “unit assembly” method (20). The resulting plasmids were named pT-1L, pT-1R, pT-2L, and pT-2R, respectively. mRNAs were synthesized in vitro using the expression plasmids as templates with the Message mMACHINE SP6 kit (Ambion), purified with the RNeasy Mini Kit (Qiagen, Germany), and dissolved in RNase-free nanopure water (Ambion).
To test the activities of the TALENs, we microinjected 1 nl of solution containing 300 ng/μl mRNA of pT-1L and pT-1R or pT-2L and pT-2R into zebrafish embryos at the 1–2-cell stage, respectively. The method used to examine TALEN activity in zebrafish embryos was the same as described previously (21). The primers for amplifying the foxc1a fragment containing the potential TALEN targeting site were CCGTTTTGGAGAGCAGTCA (forward) and GTCTCCGGCCTGGTTCAG (reverse). The PCR conditions were 94 °C for 2 min; 35 cycles of 30 s at 94 °C, 30 s at 60 °C, and 1 min 50 s at 72 °C; and a final extension of 10 min at 72 °C. The 1,255-bp PCR product was then cloned into pGEM-T easy vector (Promega). 22 and 11 of the PCR-positive transformants were further sequenced to examine the activity of T-1 and T-2, respectively.
To generate heritable foxc1a knock-out zebrafish, the embryos microinjected with T-1 and T-2 were raised to adults as founders. The mutant zebrafish were screened using the method as we reported recently (22). Briefly, the sexual mature founders were allowed to mate ad libitum, producing F1 embryos. When F1 reached 5 weeks, they were genotyped by fin clipping using the PCR method as described above. Two lines of mutant zebrafish carrying mutated alleles of foxc1a, foxc1anju18/+ (generated by T-1) and foxc1anju19/+ (generated by T-2), were selected to pass the null alleles to their progeny, respectively. When the heterozygous mutants were sexually mature, they were used to produce embryos for performing experiments.
The primers for genotyping the foxc1anju18 allele were CTATTCCGTCTCCAGTCCCAACT (forward) and GCCTCTTTAACACCTCGGTCCTC (reverse). The PCR conditions were 94 °C for 2 min; 35 cycles of 30 s at 94 °C, 30 s at 60 °C, and 50 s at 72 °C; and a final extension of 10 min at 72 °C. The 549-bp (WT allele) or 542-bp (nju18 allele) PCR product was then digested with BglI (Fermentas). WT allele could be digested into two fragments, namely 401 and 148 bp. In contrast, the nju18 allele could not be digested by BglI.
The primers for genotyping foxc1anju19 allele were GTCCCTGAATGATTTTTCTA (forward, nju19 allele-specific), ACGTTGTCCCTGAATTATTC (forward, WT allele-specific), and GACGAAAAGAACGGAGGAAA (reverse). The PCR conditions were 94 °C for 2 min; 35 cycles of 30 s at 94 °C, 30 s at 50 °C for nju19 allele or 60 °C for WT allele, and 50 s at 72 °C; and a final extension of 10 min at 72 °C. The 576-bp (nju19 allele) or 571-bp (WT allele) PCR product was then subjected to 1% agarose gel electrophoresis separation. The genotyping method was also applied for genotyping individual embryos that were either mounted or not mounted.
MOs were purchased from Gene Tools. The sequences of foxc1a MO were CCTGCATGACTGCTCTCCAAAACGG (14). The sequences of aldh1a2 MO were GCAGTTCAACTTCACTGGAGGTCAT (23). The sequences of fgf8a MO were GAGTCTCATGTTTATAGCCTCAGTA (24). The sequences of control MO were CCTCTTACCTCAGTTACAATTTATA. MOs were dissolved in nanopure water and microinjected into the embryos at the 1–2-cell stage. The microinjected amount/embryo was about 1 nl of solution containing 3 or 6 ng of foxc1a MO, 2 ng of aldh1a2 MO, or 3.2 ng of fgf8a MO and an equal amount of control MO.
To synthesize mRNA in vitro, full-length coding sequences of Notch1a intracellular domain (NICD) (25) and fgf8a (26) were amplified by RT-PCR with high fidelity DNA polymerase. The PCR products were then subcloned into pXT-7 vector under T7 promoter direction as we reported previously (16). fgf8a or NICD mRNA was synthesized in vitro using the mMESSAGE mMACHINE T7 Ultra Kit (Ambion). About 1 nl of 0.2 ng/μl fgf8a and 200 ng/μl NICD mRNA combined with 1% (w/v) Texas Red dextran (D-1864, 70 kDa, Life Technologies, Inc.) (27) were co-microinjected into a zebrafish embryo at the 1–2-cell stage.
Whole mount in situ hybridizations were performed as described previously (28). RNA probes for detecting the expressions of myod1 and aldh1a2 were prepared as described previously (29). The cDNA templates for making antisense RNA probes of fn1b, foxc1a, erm1, fgf8a, fgf17b, deltaC, her1, her7, oep, ntla, shha, tbx6, mespba, paraxis, mylpfa, smyhc1, and ckma were RT-PCR-amplified fragments. The sequences of primers were CCTCCTCATGTGTCGTTTCA (forward) and GATGTTTTTGCATGGTGTC (reverse) for fn1b (30), CCGTTTTGGAGAGCAGTCA (forward) and GACGAAAAGAACGGAGGAAA (reverse) for foxc1a (17), GCGACATGGATGGGTTTTATG (forward) and CACACCCTGCTCATTGTAAC (reverse) for erm1 (31), TCTTCGTGGTGTGGTGTCAT (forward) and GGGGTAAGATGGAGGAGGAG (reverse) for fgf17b (32), TGCTTCTGAGCAGCAGCAGATC (forward) and TGCTCCTGTGTAACCAAATACAGC (reverse) for fgf8a (26), ACTGTAGGGTTCGCTCTG (forward) and CTCCACAGTCACAGTTCC (reverse) for deltaC (33), AGGTGTATCGTCTTCTTC (forward) and GACAACGCTGGTTATTCT for her1 (34), CATATCCTCATTGATATC (forward) and AGAGATTACACAAGGCCC (reverse) for her7 (35), GAACGTGCATTTTGGGAAGT (forward) and CACGGCAAATGAAAGTTGTG (reverse) for oep (36), GAATTCCCGCTGTCAAAGCA (forward) and CCTCCTGAAGCCAAGATCAA (reverse) for ntla (37), AGGAGACGGCAAGGACATCG (forward) and TTGTCCGGCTCTGACACTGC (reverse) for shha (38), TGACCAGCAGGCTACGAAAT (forward) and CACGGCACGAACATGACAAT (reverse) for tbx6 (39), ACACGACATGCAAACCTCAA (forward) and GTATCCGCCCTCAGTTTTTG (reverse) for mespba (40), ATCCCATCCAGCTCTTCTCA (forward) and TCGAAATTGTCGCTTGTCTG (reverse) for paraxis (41), GAGGAGAGGGTTCCTCCAAC (forward) and GCAGAAGAAGCAGCGACTTT (reverse) for mylpfa (42), AGGACTGTTCTGTGTCACTG (forward) and GACTTAGCCAGAGCACCAAT (reverse) for smyhc1 (42), and GGCAACTGACAAGCACAAGA (forward) and TCAGCAACAGGCACTCACTC (reverse) for ckma (42).
Whole mount double in situ hybridization was performed following the methods as reported (28). The two antisense probes were labeled with fluorescein-12-UTP (for aldh1a2; Roche Applied Science) and digoxigenin-11-UTP (for fgf8a; Roche Applied Science), respectively. The first color was developed with 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium for 8 h, and the second color was developed with 2-(4-iodophenyl)-3-(4-nitrophenyl)-5-phenyl-2H-tetrazolium chloride/nitro blue tetrazolium (Roche Applied Science) for 2 h at room temperature. The embryos were flat mounted in 3% methylcellulose.
Whole mount double fluorescent in situ hybridizations were performed following the protocol reported previously (43). Briefly, digoxygenin-labeled foxc1a probe and fluorescein-labeled aldh1a2 probe were hybridized with embryos, followed by incubating with anti-fluorescein-peroxidase antibody (1:500; Roche Applied Science) along with TSA (tyramide signal amplification) Plus fluorescein solution (1:500; PerkinElmer Life Sciences). After inactivating the first peroxidase, anti-digoxigenin-peroxidase antibody (1:500; Roche Applied Science) along with TSA Plus Cy5 solution (1:500; PerkinElmer Life Sciences) were added. Then embryos were stained with propidium iodide (Life Technologies). They were then removed yolk and flat mounted on a microscope slide. Images were photographed with a TCS SP5 confocal microscope (Leica, Heidelberg, Germany).
To perform the ChIP assay, we first recombined zebrafish foxc1a coding sequence into pMyc-N1 vector (modified from pEGFP-N1, Clontech). The coding sequence of Myc-tagged Foxc1a (Foxc1a-Myc) was then subcloned into the pcDNA3.1(−) vector (Invitrogen) under T7 promoter direction. mRNA of Foxc1a-Myc was synthesized in vitro using the mMESSAGE mMACHINE T7 Ultra Kit (Ambion). About 2 nl of 10 ng/μl Foxc1a-Myc mRNA was microinjected into zebrafish embryos at the 1–2-cell stage. The embryos were grown to the 9-somite stage for the ChIP assay. ChIP analyses were performed using the EZ-ChIP chromatin immunoprecipitation kit (Millipore) following the manufacturer's instructions as we previously reported (44). The immunoprecipitated DNA was amplified by real-time PCR with primers (AAGTTGTCATGTTTTCAAGACGTG and CCTTCAGAACATGATAAATATGACG) for detecting the aldh1a2 promoter p1 (151 bp) or with primers (AAGCTCCAGATTTGGCTTCA and AGGTGCTATGTATGGCCTGC) as a negative control pNTC (165 bp). The relative enrichment of Foxc1a on aldh1a2 p1 or pNTC was calculated using the -fold enrichment method (Invitrogen) by normalizing the PCR signals obtained from ChIP with anti-Myc tag antibody to those from control ChIP with mouse IgG. Results were subjected to Student's t test.
The lengths of somite size were measured in the dorsal viewed embryos at the 9-somite stage in bright field with the software UTHSCSA ImageTool and shown in mean ± S.D. The length unit was arbitrary. Results were subjected to Student's t test.
To knock out the foxc1a gene (AF219949) in zebrafish, we constructed TALENs using the unit assembly method (20). Two potential target sites and their corresponding TALEN pairs (T-1 and T-2) were selected. T-1 recognizes the DNA region of nucleotides 134–189 (the translation initiation site is marked as +1), and T-2 recognizes the DNA region of nucleotides 922–970 (Fig. 1A). Sequence analyses of the molecules amplified from the embryos microinjected with T-1 or T-2 mRNAs revealed that the mutated rate of foxc1a alleles created by T-1 was 45.4% (5 of 11), and that created by T-2 was 72.8% (16 of 22), respectively. The mutated types of the molecules created by T-1 included deletion (80%, 4 of 5) and complex (20%, 1 of 5) (Fig. 1B), and those created by T-2 were insertion (12.5%, 2 of 16), deletion (81.25%, 13 of 16), and complex (6.25%, 1 of 16) (Fig. 1C).
To generate foxc1a knock-out embryos, we raised the embryos microinjected with T-1 and T-2 as founders. In total, 39 (T-1) and 46 (T-2) F1 juvenile zebrafish derived from the founders mated ad libitum were genotyped. Among the 39 offspring of the founders created by T-1, 11 were mutants carrying 11 mutated alleles of foxc1a. The mutant rate was 28.2% (11 of 39). The types of mutated alleles included insertion (9.1%, 1 of 11), deletion (36.4%, 4 of 11), and complex (54.5%, 6 of 11) (Fig. 1D). Among the 46 offspring of the founders created by T-2, 18 were mutants carrying 21 mutated alleles of foxc1a (three carrying two mutated alleles). The mutant rate was 39.1% (18 of 46). The types of mutated alleles included insertion (14.28%, 3 of 21), deletion (61.90%, 13 of 21), and complex (23.81%, 5 of 21) (Fig. 1E). The mutated allele (foxc1anju18) with a complex mutation of 9-bp deletion plus 2-bp insertion created by T-1 was predicted to encode a truncated Foxc1a protein with only the first 50 aa of Foxc1a plus 20 mutated aa (Fig. 1, F and G). The mutated allele (foxc1anju19) with a 10-bp insertion created by T-2 was predicted to encode a truncated Foxc1a protein with only the first 315 aa of Foxc1a plus 61 mutated aa (Fig. 1, F and G). The two mutant lines (foxc1anju18/+ and foxc1anju19/+) were therefore bred for further experiments.
To explore the role of Foxc1a in zebrafish early development, we produced homozygous foxc1a mutant embryos using the mutants carrying foxc1anju18 and foxc1anju19 allele, respectively. It turned out that the foxc1a null embryos exhibited normal somites like their wild type siblings when they were examined at the 2- or 5-somite stage (Fig. 2, A–D and A′–D′; data for foxc1anju19/nju19 not shown). When they were observed at the 9-somite stage, all foxc1anju18/nju18 and foxc1anju19/nju19 embryos had obvious defects in the somites, especially the first six somites, with significantly reduced size (p < 0.01), although the boundaries between somites were still distinguishable (Fig. 2, E–G, E′–G′, and G″). The size of the first six somites in foxc1anju18/nju18 embryos (145.40 ± 15.26, n = 8) was about 80% that of their wild type siblings (180.64 ± 6.09, n = 11), and that of the first six somites in foxc1anju19/nju19 embryos (154.28 ± 11.05, n = 5) was about 85% that of their wild type siblings. The defective somitogenesis in mutant embryos then underwent a gradual recovery with development. The mutant embryos displayed somites morphologically similar to those of their wild type siblings except for mild cardiac edema when they were observed at 24 hpf (Fig. 2, H–J). At 60 hpf, both foxc1anju18/nju18 and foxc1anju19/nju19 embryos exhibited severe cardiac edema and truncated body length (Fig. 2, K–M), and they finally died around 9 days postfertilization. The results revealed that foxc1a plays a crucial role in zebrafish somitogenesis and cardiac development.
To investigate when foxc1a affected zebrafish somitogenesis, we first examined the expression of tbx6, the presomitic precursor marker (1), in the mutated zebrafish embryos at bud stage. Similar to their wild type siblings, foxc1anju18/nju18 embryos displayed normal expression of tbx6 (data not shown). We then detected the expressions of segmentation clock genes including her1, her7, and deltaC (45) in foxc1anju18/nju18 embryos at the 6-somite stage, and no expression changes were found in the PSM of mutant embryos (data not shown). We next detected the expressions of aldh1a2 and fgf8a, the main genes responsible for front determination of somites (1), in foxc1anju18/nju18 embryos at the 3-somite stage. Again, the two genes were expressed normally in PSM (data not shown). We continued to examine the expression of mespba, the marker gene for the somite boundary formation, in the foxc1anju18/nju18 embryos at the 9-somite stage. The results showed that the expression pattern of mespba was not altered (data not shown). Consistently, both the boundary marker gene fn1b and the epithelialization marker gene paraxis were normally expressed in the somite boundary, although their expression levels were slightly reduced (data not shown). Finally, we analyzed the expression of myod1, the marker gene for somite differentiation. The results revealed that the expression of myod1 was significantly decreased in the lateral paraxial mesoderm region, especially in the first six somites, but not in adaxial cells of both foxc1anju18/nju18 and foxc1anju19/nju19 embryos (Fig. 3, A–C).
Unexpectedly, the abnormal somitogenesis in foxc1a null embryos was found to be different from the previous findings that knocking down foxc1a using MO completely blocked formation of morphological somites, formation of segment boundaries, segmented expression of genes normally transcribed in anterior and posterior somites, and expression of paraxis implicated in somite epithelialization (14). The differences could be ascribed to the maternal expression of foxc1a because MO was expected to inhibit both maternal and zygotic transcripts. To clarify this, we examined the expression of foxc1a in the embryos at early development by performing quantitative RT-PCR. The results revealed that no maternal transcripts of foxc1a was found in zebrafish embryos (Fig. 3G).
To distinguish the differences, we reanalyzed the expression of myod1 in the embryos microinjected with the MO that was used previously (14). The results showed that the embryos microinjected with different MOs exhibited different defective phenotypes. Low dose (3 ng) of foxc1a MO could phenocopy the foxc1a knock-out embryos somehow (Fig. 3, B, C, and E), whereas a high dose (6 ng) of foxc1a MO led to completely depleted expression of myod1 in the somite region (Fig. 3F), as reported previously (14).
To investigate whether the somite differentiation was affected in foxc1a mutants at late somitogenesis, we examined the expressions of myod1 and three terminal differentiation makers for skeletal muscle (42, 46), including mylpfa (marker gene for fast muscle), smyhc1 (maker gene for slow muscle), and ckma (maker gene for both slow and fast muscle), in the embryos at 24 hpf. It turned out that the foxc1a null embryos exhibited similar expression patterns of myod1 (Fig. 3, H, I, H′, and I′), mylpfa (Fig. 3, J, K, J′, and K′), smyhc1 (Fig. 3, L, M, L′, and M′), and ckma (Fig. 3, N, O, N′, and O′) to their wild type siblings at 24 hpf. The results were consistent with the morphological recovery of the somites in the foxc1a null embryos at 24 hpf (Fig. 2, I and J). Taken together, our results demonstrated that foxc1a plays an essential role in early somitogenesis by controlling the expression of myod1 in somites.
To uncover how foxc1a affects early somitogenesis in zebrafish embryos, we first examined the expressions of ntla, shha, and oep in foxc1a knock-out embryos at the 9-somite stage. The results showed that no expression changes of ntla, shha, and oep were found in the mutant embryos at the 9-somite stage (data not shown).
We then checked the expression of deltaC, the gene mediating Notch signaling, in the paraxial mesoderm of foxc1a null embryos at the 9-somite stage. Interestingly, the expression of deltaC was almost completely abolished in the paraxial mesoderm of somite region, although its expression in the PSM was similar to that in wild type siblings (Fig. 4, A and B). Like foxc1a knock-out embryos, embryos treated with Notch signaling inhibitor DAPT exhibited significantly reduced expression of myod1 in the somite region, especially in the anterior somite region (47) (data not shown). The result indicates that foxc1a controls the expression of myod1 at early somitogenesis by regulating Notch signaling.
To test whether Fgf signaling mediates the role of foxc1a in early somitogenesis, we examined the expression of fgf8a in foxc1a null embryos. The results showed that the expression of fgf8a was significantly reduced in the somite region of the mutant embryos, although it was expressed normally in other regions, such as the midbrain-hindbrain boundary, rhombomere 2, rhombomere 4, and tail bud, and its expression in the dorsal telencephalon was increased (Fig. 4, C and D). Consistently, the expressions of both erm1 and fgf17b, the downstream target genes of fgf8a signaling, were greatly reduced in the paraxial mesoderm of foxc1a null embryos (Fig. 4, E and H). Like foxc1a knock-out embryos, fgf8a morphants displayed greatly reduced expression of myod1 (26) (data not shown). The result suggests that foxc1a controls the expression of myod1 in the somites of embryos at early somitogenesis by regulating Fgf signaling.
It is known that RA signaling plays a crucial role in somitogenesis by antagonizing Fgf signaling (1). Now that Fgf signaling was significantly reduced in the paraxial mesoderm of somite region in foxc1a knock-out embryos at early somitogenesis, we therefore checked whether RA signaling was increased in the mutants. As we expected, the expression of aldh1a2, the major gene that is responsible for converting retinal to RA in the early development, was significantly increased in the paraxial mesoderm and anterior lateral plate mesoderm of foxc1a knock-out embryos (Fig. 5, A and B). To investigate whether the increased expression of aldh1a2 in the paraxial mesoderm was responsible for the defective somitogenesis, we decreased the expression of aldh1a2 in foxc1a mutants by microinjecting aldh1a2 MO and then examined the expression of myod1. The results revealed that knocking down aldh1a2 did not affect the expression of myod1 in the foxc1a+/+ embryos (Fig. 5, C and F). However, knocking down aldh1a2 in foxc1a mutants could partially rescue the expression of myod1 in the paraxial mesoderm (Fig. 5, D and E). To confirm the findings, we treated foxc1a null embryos with 2 μm DEAB to inhibit endogenous RA signaling. We found that the wild type embryos exhibited slightly reduced expression of myod1 in their somite region (Fig. 5, C and G), but part of their foxc1a mutant siblings resumed their myod1 expression to nearly normal level (Fig. 5, D and H).
It was reported previously that RA signaling plays a positive role in myod1 expression (5). By treating the embryos with 1 μm RA from 8 hpf, we found that the expression of myod1 was significantly increased not only in the newly formed three somites of both wild type embryos and their foxc1a mutant siblings but also in the whole PSM of foxc1a mutant embryos (Fig. 5, I and J). However, RA treatment did not rescue myod1 expression in the first 6-somite region of foxc1a mutant embryos (Fig. 5, D and J). Taken together, the results suggest that foxc1a controls the early somitogenesis by restricting aldh1a2 expression.
To investigate whether Foxc1a inhibits the expression of aldh1a2 directly, we first performed fluorescent in situ hybridization to examine the expressions of aldh1a2 and foxc1a in the developing somites. The results showed that aldh1a2 and foxc1a had co-expression in somites (Fig. 5, K, K′, and K″), and the expression of aldh1a2 was greatly increased and ectopically expanded in the paraxial mesoderm of foxc1a knock-out embryos (Fig. 5, L, L′, and L″). We then analyzed the 4,077-bp sequences upstream of the translation start site of aldh1a2 using MatInspector and found that there was a presumptive core sequence site of the Foxc1a binding site (TGTTTATTT) lying between −3,025 and −3,017 bp of the regulatory region (Fig. 5M). To confirm that Foxc1a could bind the aldh1a2 promoter in vivo, we performed a ChIP assay. The results revealed that the -fold enrichment value of Foxc1a on the aldh1a2 promoter was 6.23-fold higher (p < 0.01) than control (Fig. 5N). In contrast, the region that was not the predicted Foxc1a binding site on aldh1a2 promoter was not enriched with Foxc1a at all (Fig. 5N). Together, these results demonstrated that Foxc1a bound the aldh1a2 promoter in vivo.
Because knocking down aldh1a2 could partially rescue the abnormal somite differentiation (Fig. 5) and Fgf and Notch signaling were significantly reduced in the paraxial mesoderm of foxc1a knock-out embryos (Fig. 4), the question of whether the role of aldh1a2 in regulating the expression of myod1 in somite region was mediated by Fgf and Notch signaling was intriguing. To identify the genetic hierarchy of these genes in early somitogenesis, we examined the expression changes of fgf8a and deltaC in the foxc1a knock-out and aldh1a2 knockdown embryos. As shown in Fig. 6, knocking down aldh1a2 did not affect the expressions of deltaC and fgf8a in the paraxial mesoderm of embryos at the 9-somite stage (Fig. 6, A, D, E, and H), but it could partially rescue the decreased expressions of fgf8a (Fig. 6, A–C) and deltaC (Fig. 6, E–G) in foxc1a null embryos. In contrast, neither knocking down fgf8a nor inhibiting Notch signaling affected the expression of aldh1a2 in paraxial mesoderm (Fig. 6, I–L), although knocking down fgf8a reduced the expression of deltaC (Fig. 6, M and N), whereas inhibiting Notch signaling decreased the expression of fgf8a in the paraxial mesoderm (Fig. 6, O and P). Furthermore, overexpression of Fgf8a and NICD could partially rescue the reduced expression of myod1 in foxc1a null embryos (Fig. 6, Q and R). Taken together, the results demonstrate that foxc1a works upstream of aldh1a2 to control the expression of myod1 in somites by regulating expressions of fgf8a and deltaC in early somitogenesis of zebrafish embryos.
Foxc1a plays a crucial role in multiple developmental processes (48,–50). Previously, it was reported that knocking down foxc1a using MO led to a complete loss of somite boundaries, blocked formation of morphological somites, abolished myod1 expression in paraxial mesoderm, and eliminated expression of paraxis implicated in somite epithelialization (14). However, we report in this study that foxc1a null zebrafish embryos from two lines of zebrafish mutants exhibit nearly normal somite boundary and show abnormal somitogenesis only with reduced size of the first six somites in the embryos at early somitogenesis (Fig. 2). Further analyses reveal that the expressions of the key genes that control the processes of somitogenesis, including the marker gene for the formation of presomitic mesoderm (tbx6) (39), segmentation clock genes (her1, her7, and deltaC) (45), and marker genes for front determination of somites (aldh1a2 and fgf8a) (51) are all similar to those of wild type siblings. Unlike the morphants reported previously (that mespba, the marker gene for the somite boundary formation, was abnormally expressed (40), and paraxis lost its expression during somitogenesis), the foxc1a knock-out embryos exhibit normal expression of mespba and slightly decreased expressions of fn1b and paraxis in the paraxial mesoderm of the embryos at early somitogenesis (data not shown). Because the somite boundary is determined by mespb and then epithelialization takes place by expressions of fn1b (30) and paraxis (52), our results are consistent with the normal somite boundary found in the foxc1a knock-out embryos (Fig. 2). Therefore, we conclude that foxc1a is not essential to somite boundary formation. Moreover, unlike the morphants with completely lost myod1 expression in somite region, foxc1a knock-out embryos only decrease its expression in the somite region, especially the first six somites (Fig. 3, A–C). Taken together, our results demonstrate that foxc1a regulates early somitogenesis by controlling the expression of myod1 in the somite region. Because foxc1a is zygotically expressed (Fig. 3G), the discrepancy of somitogenesis phenotypes between our foxc1a null embryos and previous morphants (14) is very likely to be due to the nonspecific or toxic effects of MO (53) (Fig. 3G) used in the previous research (14).
In mouse embryos, the gene dosage of Foxc1/Foxc2 determines the somite formation. Both Foxc1 null and Foxc1−/−;Foxc2+/− embryos exhibit morphologically normal somites. However, the somite shape of Foxc2 null mouse embryos is slightly abnormal, and that of Foxc1+/−;Foxc2−/− mouse embryos is very narrow and irregular (54). When Foxc1 and Foxc2 are double knocked out, the double null embryos display a complete lack of somites (55). In this study, we report that somite sizes in zebrafish foxc1a null embryos are normal at the 2- and 5-somite stages (Fig. 2, A–D and A′–D′). However, the somites of the mutant embryos are abnormal, with not only the reduced size for the first six somites (Fig. 2, E–G, E′–G′, and G″) but also the reduced expression of myod1 in somites (Fig. 3, A–C). Interestingly, the abnormal somitogenesis underwent a gradual recovery with development, and somites became normal at 24 hpf with not only normal morphology (Fig. 1, H–J) but also a similar expression pattern of maker genes, including myod1, mylpfa, smyhc1, and ckma (Fig. 3, H–O and H′–O′). Because the assay on synteny between mouse and zebrafish suggests that both zebrafish foxc1a and foxc1b are orthologues of mouse Foxc1 (mouse Foxc1 linking to Foxq1 and Foxf2; zebrafish foxc1a linking to foxq1a and foxf2a; and zebrafish foxc1b linking to foxq1b and foxf2b), a close look at the mouse Foxc1 null embryos would answer the question of whether they have the same transient abnormal somitogenesis at early development.
Multiple signalings, such as Shh, Nodal, Fgf, Notch, and RA signaling, have been reported to control the expression of myod1 in the embryos at early somitogenesis (1, 56). However, the master regulator of the molecular network remains elusive (45). Shh signaling is essential for its expression in adaxial cells (57, 58). Consistent with the finding that foxc1a null embryos exhibit normal expression of shha (data not shown), we report in this study that the expression of myod1 in the adaxial cells is not changed in foxc1a knock-out zebrafish (Fig. 3, A–C).
Oep/Nodal affects the expression of the mesoderm marker gene ntla (59). In ntla mutant embryos, the onset of myod1 expression is delayed until the end of gastrulation but partially recovered during segmentation stages (60). Here, we report that ntla or oep expression is not altered in foxc1a knock-out embryos, indicating that foxc1a controls early somitogenesis independent of Ntl or Nodal signaling.
Fgf8, diffusing from the anterior somite, perhaps acting through either Fgfr1 or Fgfr4, promotes myod1 expression in nearby cells (61). In the absence of Fgf8 signaling, lateral somite cells fail to properly activate myod1 (61, 62). In this study, we report that fgf8a expression is significantly reduced in the paraxial mesoderm of foxc1a null embryos at early somitogenesis (Fig. 4, C and D), suggesting that fgf8a works downstream of foxc1a in regulating myod1 expression. This is also in line with previous observations of the reduced expression of myod1 in zebrafish acerebellar (fgf8a) mutant embryos (26). These results indicate that foxc1a controls myod1 expression in early somitogenesis by regulating fgf8a.
Notch signaling is crucial for both clock gene oscillations and somite formation in mouse embryos (63). It is required to synchronize oscillations between neighboring cells (64, 65). Zebrafish mesp genes are involved in anteroposterior specification within the presumptive somites by regulating the essential signaling pathways mediated by Notch signaling (40). In our study, we report that mespba expression is not altered in foxc1a null embryos (data not shown). However, deltaC expression is significantly reduced in the paraxial mesoderm of foxc1a null embryos at early somitogenesis (Fig. 4, A and B). The results suggest that foxc1a does not affect anteroposterior specification within the presumptive somites, and Notch signaling controls myod1 expression at early somitogenesis by working downstream of foxc1a.
RA is known to play important roles in somitogenesis (51, 66). Zebrafish embryos treated with exogenous RA from 8 hpf to the 10-somite stage display up-regulated expression of myod1 in the somite and the anterior presomitic mesoderm (5). However, we report in this study that the increased expression of aldh1a2 due to loss of function of its direct upstream gene (Fig. 5, K″, L″, M, and N), foxc1a, results in significantly decreased expression of myod1 in the paraxial mesoderm of foxc1a null embryos (Fig. 5, B and D), and reducing RA signaling either by partially knocking down aldh1a2 or treating with DEAB partially rescued the decreased expression of myod1 in foxc1a null embryos (Fig. 5, E and H). Moreover, RA treatment is not able to rescue the reduced expression of myod1 in the somite region of foxc1a null embryos (Fig. 5J), although it does increase the expression of myod1 in their wild type siblings (Fig. 5I). Together, our results demonstrate that phenotypes derived from altered RA signaling by administrating RA exogenously do not always phenocopy the changes from altering endogenous RA signaling. This is also true for the expression changes of fgf8a in the embryos. When zebrafish embryos are exposed with exogenous RA, the expression of fgf8a is increased in the anterior presomitic mesoderm and somites, causing the increased expression of myod1 in somites (5). However, we demonstrate in this study that the increased RA signaling in foxc1a null embryos is responsible for the reduced expression of myod1 through decreasing Fgf signaling. Partially knocking down aldh1a2 does not change the expression of fgf8a in either PSM or the paraxial mesoderm (Fig. 6H) but rescues the decreased expressions of fgf8a and myod1 in the paraxial mesoderm of foxc1a null embryos (Figs. 5E and and66G). Consistently, both RA-deficient chick embryos and mouse Raldh2−/− embryos exhibit ectopic Fgf8 expression extending from the caudal progenitor zone into the trunk that disrupts somitogenesis (67, 68). Recently, it has been demonstrated that RA controls mouse somitogenesis by directly repressing Fgf8 transcription through binding to the RARE (RA response element) in the regulatory sequence of Fgf8 (69). By analyzing the 6,000-bp sequences upstream of the fgf8a transcription start site (ENSDARG00000003399) using MatInspector, as we reported previously (44), we find that zebrafish fgf8a has a presumptive DR5 RARE with the core sequences of TTGTTGATCAGGCAATGAGCAACAG located between −3,857 and −3,833 bp, positioned in the reverse direction. Taken together, our results are consistent with the previous observation that endogenous RA signaling regulates somitogenesis by antagonizing Fgf signaling (51, 66).
Proper RA signaling is also known to be essential for the normal expression of Notch signaling in PSM. Either an increase of endogenous RA signaling by knocking down cyp26a1 or a decrease of endogenous RA signaling by knocking down aldh1a2 (16 ng/embryo) leads to asymmetric expressions of deltaC, her1, and her7 in PSM between the 6- and 13-somite stages (47, 70). In this study, we demonstrate that partially knocking down aldh1a2 (2 ng/embryo) does not change the expressions of deltaC in either PSM or the paraxial mesoderm (Fig. 6D) or the expression of myod1 in somites (Fig. 5F) but rescues the decreased expressions of deltaC and myod1 in the paraxial mesoderm of foxc1a null embryos (Figs. 5E and and66C). The results suggest that the increased RA signaling in foxc1a null embryos is responsible for the abnormal somitogenesis through decreasing Notch signaling and therefore reducing myod1 expression in paraxial mesoderm.
In summary, our results provide support for a major role of forkhead transcription factor Foxc1a in controlling the expression of myod1 in paraxial mesoderm of the embryos at early somitogenesis by working on the top of the genetic hierarchy of the RA, Fgf, and Notch signaling network (Fig. 7).
*This work was supported by Ministry of Science and Technology of China Grant 2011CB943804, National Natural Science Foundation of China Grant 31171434, and Specialized Research Fund for the Doctoral Program of Higher Education (SRFDP) Grant 20130091110051.
3The abbreviations used are: