The zebrafish homeobox gene irxl1 is required for brain and pharyngeal arch morphogenesis


  • Han-Ni Chuang,

    1. Department of Biomedical Sciences, Chung Shan Medical University, Taichung, Taiwan
    2. Department of Life Science and Institute of Molecular Biology, National Chung Cheng University, Chia-Yi, Taiwan
    Search for more papers by this author
  • Hsiao-Yun Cheng,

    1. Department of Biomedical Sciences, Chung Shan Medical University, Taichung, Taiwan
    Search for more papers by this author
  • Kuang-Ming Hsiao,

    1. Department of Life Science and Institute of Molecular Biology, National Chung Cheng University, Chia-Yi, Taiwan
    Search for more papers by this author
  • Chieh-Wen Lin,

    1. Department of Biomedical Sciences, Chung Shan Medical University, Taichung, Taiwan
    Search for more papers by this author
  • Mei-Ling Lin,

    1. Department of Biomedical Sciences, Chung Shan Medical University, Taichung, Taiwan
    Search for more papers by this author
  • Huichin Pan

    Corresponding author
    1. Department of Biomedical Sciences, Chung Shan Medical University, Taichung, Taiwan
    2. Department of Medical Research, Chung Shan Medical University Hospital, Taichung, Taiwan
    • Department of Biomedical Sciences, Chug Shan Medical University, 110, Sec.1, Jianguo North Road, Taichung 40201, Taiwan
    Search for more papers by this author


Iroquois homeobox-like 1 (irxl1) is a novel member of the TALE superfamily of homeobox genes that is most closely related to the Iroquois class. We have identified the zebrafish irxl1 gene and characterized its structure. The protein contains a homeodomain that shares 100% sequence identity with other vertebrate orthologs. During embryogenesis, irxl1 is expressed from 18 hours postfertilization onward and prominent expression is detected in the pharyngeal arches. Knockdown of irxl1 by morpholinos results in malformed brain and arch structures, which can be partially rescued by cRNA injection. The heads of the morphants become small and flat, and extensions along the anterior–posterior/dorso–ventral axes are reduced without affecting regional specification. Loss of irxl1 function also causes deficit in neural crest cells which consequently results in partial loss of craniofacial muscles and severe deformation of arch cartilages. These observations suggest that irxl1 may regulate factors involved in brain and pharyngeal arch development. Developmental Dynamics 239:639–650, 2010. © 2009 Wiley-Liss, Inc.


Homeobox genes encode a superfamily of transcription factors that play important roles in the regulation of embryonic development. They are characterized by a 180-bp DNA sequence that encodes a conserved helix-turn-helix DNA-binding domain called the homeodomain. These proteins were first identified in Drosophila and their homologs were later found in a diverse range of species from yeast to humans. They participate in various aspects of developmental processes including axial patterning, regional specification, and cell fate determination (Gehring et al.,1994; Krumlauf,1994). Several families exist within the homeobox gene superfamily, which are classified on the basis of the conservation of their homeodomains as well as additional motifs that contribute to DNA binding and to interactions with other proteins (Banerjee-Basu and Baxevanis,2001).

The Iroquois homeobox genes (Irx/iro) belong to the TALE (three-amino-acid loop extension) superclass of the atypical homeobox gene family. In addition to the extra three amino acids in the loop between helix I and helix II of the homeodomain, which is characteristic of the TALE superclass, the Irx homeodomains share a common alanine at position 50 (A50) of helix III, which participates in DNA binding recognition (Duan and Nilsson,2002). Another unique feature of the Irx proteins is a conserved C-terminal acidic motif of 13 amino acids known as the IRO box (Cavodeassi et al.,2001; Gomez-Skarmeta and Modolell,2002). The iroquois genes were first identified in Drosophila as regulators of proneural gene expression. They specify cell identity within the compartments of the eye and participate in the formation of sensory bristles in insects (Gomez-Skarmeta and Modolell,1996; Leyns et al.,1996; Cavodeassi et al.,1999; Calleja et al.,2002). Six Irx genes have been identified in the mouse and human genomes, and these genes are organized into two clusters (Irx1, 2, 4, and Irx3, 5, 6; Peters et al.,2000; Gomez-Skarmeta and Modolell,2002). In zebrafish, 11 irx genes have been identified, with more clusters, due to an additional genome duplication event in the teleost lineage after its divergence from other vertebrates (Dildrop and Ruther,2004; Feijoo et al.,2004). The vertebrate Irx genes are expressed in a broad range of tissues and are involved in numerous patterning processes such as organizer formation, neural plate specification, subdivision of the anterior–posterior and dorso–ventral axes of the nervous system, and heart chamber specification (Bao et al.,1999; Cavodeassi et al.,2001; Gomez-Skarmeta and Modolell,2002).

Recently, a novel member of the TALE superclass, the Iroquois homeobox-like 1 (Irxl1) or Mohawk (Mkx) gene, was identified (Weinmann et al.,2005; Anderson et al.,2006; Liu et al.,2006; Mukherjee and Burglin,2007; Takeuchi and Bruneau,2007). Sequence analysis revealed that Irxl1 is closely related to the Iroquois class. The greatest homology is within the helix III of the homeodomain, including the conserved A50 residue. However, Irxl1 lacks the typical IRO box found in all Irx genes (Anderson et al.,2006; Liu et al.,2006). The predicted Irxl1 orthologs exist in vertebrate species such as frog, fish, mouse, rat, chimpanzee, and human, as well as in invertebrates such as fly (Weinmann et al.,2005; Anderson et al.,2006; Takeuchi and Bruneau,2007). Comparison of the chromosomal loci of vertebrate orthologs revealed a conserved gene synteny, in that Irxl1 is positioned between Rab18 and ArmC4 (Weinmann et al.,2005). During mouse embryonic development, Irxl1 RNA is highly expressed in craniofacial mesenchyme, including the frontonasal process, otic vesicle, and branchial arch mesenchyme. It is also expressed in the muscle and cartilage progenitors of the somites, limb buds, and tail, and in the metanephrogenic kidney and the testis cords of the gonad (Anderson et al.,2006; Liu et al.,2006; Takeuchi and Bruneau,2007). The conservation of Irxl1 across vertebrate species, together with its specific embryonic expression pattern, suggests an important role for Irxl1 in vertebrate development. However, the function of Irxl1 has not yet been fully characterized.

In this study, we used zebrafish as a model to analyze the function of irxl1 in embryonic development. We first cloned and characterized the sequence of zebrafish irxl1, and determined its RNA expression profile both in adult tissues and during embryonic/larval development. Using morpholino knockdown analysis, we also provide evidence that irxl1 is required for brain and pharyngeal arch morphogenesis in zebrafish.


Cloning and Characterization of the Zebrafish irxl1 cDNA

The zebrafish irxl1 gene was identified by a TBLASTN search of the GenBank database using the homeodomain sequence of the mouse Mkx ortholog (NP_808263). The full-length cDNA was then retrieved by reverse transcriptase-polymerase chain reaction (RT-PCR) and 5′-/3′-rapid amplification of cDNA ends (RACE), followed by cloning and sequencing analyses. Two splicing variants of the zebrafish irxl1 were revealed, which we designated as irxl1a and irxl1b (GenBank accession nos. EF457976 and EF457977). The first four exons are identical in both forms, and irxl1b differs from irxl1a only in its C-terminal 36 amino acids (Fig. 1A). BLAST analysis showed that the zebrafish irxl gene is located on chromosome 12.

Figure 1.

Structure and sequence of the zebrafish irxl1 gene. A: Genomic organization of the two splice forms of irxl1 in zebrafish. Boxes and lines represent exons and introns, respectively. The untranslated regions are denoted in white and the coding regions are denoted in black. B: Alignment of amino acid sequences of irxl1 orthologs from different species. D.r., Danio rerio; H.s., Homo sapiens (NP_775847.1); M.m., Mus musculus (NP_808263.1); R.n., Rattus norvegicus (ENSRNOP00000025623); X.t., Xenopus tropicalis (ENSXETP00000009890); D.m., Drosophila melanogaster (NP_608502.1); C.e., Caenorhabditis elegans (CAA04166.1). Identical amino acids are boxed in black and similar amino acids are in grey.

Zebrafish irxl1a is composed of seven exons (2,945 bp) that span 47,418 bases in the genome and encodes a polypeptide of 349 amino acids. The homeodomain is encoded by exon 3 and exon 4. Translation starts in exon 2 and ends in exon 7, at which a large 3′-untranslated region (1,686 bp) is located. Exon 6 is small (34 bp), while intron 5 is large (27 kb). This genomic organization is highly homologous to both human and mouse orthologs. Comparison of zebrafish irxl1 at the amino acid level with the predicted orthologs in vertebrates revealed a high degree of conservation (66–71% identity) in the entire protein (Fig. 1B; Supp. Table S1, which is available online). Notably, the homeodomain displays 100% homology across all vertebrate orthologs analyzed. In contrast, only 31–32% and 50–44% sequence homology was found in the entire protein and the homeodomain of Drosophila and C. elegans orthologs, respectively (Supp. Table S1).

Expression Pattern of irxl1

To determine the temporal expression and tissue distribution of irxl1 mRNA, we first performed RT-PCR analysis. By 3 hours postfertilization (hpf), which is around the midblastula stage, a low level of irxl1 expression was detected (Fig. 2A). This early expression suggests that irxl1 is maternally transcribed. During embryonic development, zygotic transcription of irxl1 was initially detected at 18 hpf and continued to 5 days postfertilization (dpf; Fig. 2A). Both isoforms were expressed in most adult fish tissues, including brain, eye, muscle, ovary, testis and swim bladder (Fig. 2B), although the expression levels varied between the two isoforms. Among the tissues examined, only heart did not show detectable irxl1 transcripts.

Figure 2.

Expression profiles of irxl1 by reverse transcriptase-polymerase chain reaction (RT-PCR). A: Expression of irxl1 in developing fish embryos at indicated time points. hpf, hours postfertilization. B: Expression of irxl1 in adult zebrafish tissues. irxl1a, 964 bp; irxl1b, 875 bp. M, DNA markers; B, brain; E, eye; H, heart; O, ovary; T, testes; Mu, muscle; Sk, skin; Sb, swim bladder; G, gill. Amplification of elongation factor (Ef) was used as an internal control.

Whole-mount in situ hybridization (WISH) analysis was then used to examine the spatial expression pattern of irxl1 mRNA in zebrafish embryos at various developmental stages. At 18 hpf, ubiquitous irxl1 expression was observed at low levels, with slightly more signals distributed along the notochord (Fig. 3A,B). Clear irxl1 transcription was detected at 24 hpf, at which time the expression was primarily in the central nervous system including the telecephalon (Fig. 3C). Beginning from 48 hpf, prominent staining was observed in the mandibular and hyoid arches (Fig. 3D,E). Double staining with the sox9a probe and sectioning through the arch region revealed partial overlapping of irxl1 signals with sox9a expression domains (Fig. 3F–H), suggesting that irxl1 was expressed in the prechondrogenic cells in the pharyngeal arches. The staining became more intense at 72 hpf (Fig. 3K,L) and persisted to 96 hpf in the arches (Fig. 3M,N), where the signals were detected in the trabeculae, palatoquadrate, ceratohyal, hyosymplectic, basibranchial, and ceratobranchial cartilages. By immunofluorescence staining using an anti-irxl1 antibody, irxl1 proteins were also detected in the mandibular, hyoid, and branchial arches at 48 hpf (Fig. 3I) and in the tissue surrounding the ceratobranchial cartilages at 96 hpf (Fig. 3O). In the trunk region, irxl1 expression was observed but did not show a distinct pattern (Fig. 3J). Expression in this region was further confirmed by RT-PCR analysis of embryos from 24 to 72 hpf (Supp. Fig. S1).

Figure 3.

Spatial expression of irxl1 during zebrafish development. A–H,J–N: Whole-mount in situ hybridization of embryos at 18–96 hours postfertilization (hpf) using a probe corresponding to the first five exons of irxl1 (recognizes both a and b isoforms) (A,B,D–H,J–N) or a probe corresponding to the 3′-untranslated region of irxl1a (C). I,O: Immunofluoresence staining using an anti-irxl1 antibody. I: Whole-mount embryo at 48 hpf. O: Cryosection at 96 hpf. F–H: Double labeling with irxl1 (blue) and sox9a (red) probes. H: Coronal section along the plane indicated by the dashed line in F. A,C: Dorsal views. B,D,F,I,J,K,M,O: Lateral views. E,G,L,N: Ventral views. b, branchial arch; bb, basibranchial; cb, ceratobranchial; ch, ceratohyal; h, hyoid arch; hys, hyosymplectic; m, mandibular arch; nc, notochord; pq, palatoquadrate; te, telencephalon; t, trabeculae. Scale bars = 300 μm in A,B, 100 μm in C–E,K–O, 60 μm in F–J.

Dose-Dependent Morphological Defects Were Observed in irxl1 Morphants

To explore the function of irxl1 in embryogenesis, we injected the antisense morpholino oligonucleotides (MO) to knockdown irxl1 expression. Two MOs were designed, one complementary to 25 bp spanning the translation initiation site of irxl1 mRNA (MOI, −17 to +8) and the other complementary to 25 bp at the boundary of intron 3 and exon 4 (MOII; Supp. Fig. S2A). Both MOs target the two isoforms of irxl1. A nonspecific MO (control MO) was used as a control. The effectiveness of MOI and MOII was confirmed by lack of protein expression (data not shown) and the reduction of correctly-spliced RNA or presence of unspliced RNA in the morphant embryos (Supp. Fig. S2B,C), respectively.

Of interest, extensive cell death was noted in the developing head and trunk along the neural tube in irxl1 MOI morphants at 25 hpf (Supp. Fig. S3C,D). In contrast, there was little cell death in the wild-type (Supp. Fig. S3A,B) and MOII morphant embryos (Fig. 4). Because off-target neural death is often induced by MO knockdown through activation of p53 (Ekker and Larson,2001; Robu et al.,2007), we thus co-injected p53 MO with irxl1 MOI (at a dose of 1.5:1) to determine if the cell death observed in morphants was nonspecific. The p53 activation with concurrent p21 transcription was seen in embryos injected with irxl1 MOI only. However, activation of both genes was repressed with co-injection of the p53 MO (Supp. Fig. S3G). Although p53 knockdown largely ameliorated cell death in the irxl1 MOI morphants, these embryos still displayed similar phenotypes seen in embryos injected with irxl1 MOII (See below). This suggests that the morphological features observed in embryos co-injected with irxl1 MOI and p53 MO are specific for loss of irxl1 function. Therefore, p53 MO was co-injected with all MOI injections in the subsequent experiments.

Figure 4.

Gross morphological phenotypes of the irxl1 MOII morphants. A–H: The embryos were injected with 0.5 pmol MOII and observed at 24 hours postfertilization (hpf; A–D) and 48 hpf (E–H). They are grouped into four classes as defined in the text. The overall pictures of MOII morphants (I) and embryos rescued with 400 pg irxl1(a+b) cRNA (J) at 48 hpf are shown for comparison. MO, morpholino oligonucleotides. Scale bars = 500 μm in A–D, 700 μm in E–J.

Despite that MOI caused some nonspecific cell death, injection of MOI plus p53MO gave similar effects as injection of MOII. The survival rates for embryos injected with irxl1 MOs or the control MO were similar; however, embryos injected with irxl1 MOs displayed significant morphological defects (Table 1; Supp. Table S2). At 24 hpf, the heads of most morphants became small and flat (Fig. 4A–D, Fig. 5). In addition to the head phenotype, irxl1 morphants also showed curved tails and body axes, shrinkage of the yolk stalk, and circle swimming behavior. At 48 hpf, more than half of the morphants displayed abnormal phenotypes (Table 1). The morphants were grouped into four categories according to the severity of gross morphological phenotype: wild-type, phenotypically similar to uninjected wild-type embryos; Mild, embryos with flat heads, reduced yolk stalk, slightly curved tails, but body length similar to the wild-type; Moderate, embryos with reduced heads, small eyes, lack of jaw protrusion, curved and shortened body axis; Severe, embryos that are severely malformed with underdeveloped heads, jaws and body axis (Fig. 4). The percentages of embryos with each phenotype are shown in Table 1. Most deformed larvae only survived for 5 to 7 days. Both the severity and defect rate increase as the dose of MO increases, indicating that the defects are dose-dependent.

Figure 5.

Irxl1 morphants show abnormal brain morphogenesis. Wild-type (A,D,G,J) and irxl1 MOI (B,E,H,K), and MOII (C,F,I,L) morphant embryos at 24 hours postfertilization (hpf). The p53 MO was co-injected with MOI. A–C,G–L: Lateral views. D–F: Dorsal views. A–F: Brain morphology of live embryos. The heads of ixl1 morphants become small and flat, and the ventricles fail to inflate properly (arrows in E,F). G–L: Whole-mount in situ hybridization with wnt1 (G–I) and krox20 (J–L) reveals defective segmentation of the hindbrain (arrows in H,I) and reduced expansion of the brain in the anterior–posterior and dorsal–ventral directions (double arrows in K, L) in morphants. di, diencephalon; e, eye; h, hindbrain; m, midbrain; mhb, mid–hindbrain boundary; r, rhombomere; te, telencephalon. Scale bar = 100 μm.

Table 1. Gross Morphological Phenotypes of irxl1 Morphants at 48 hpf
MorpholinoInjection dose (pmol/embryo)Survival rateaPhenotypebDefect ratec
  • a

    Survival rate is calculated as number of survived embryos at 48 hpf divided by total number of embryos injected.

  • b

    Phenotypes are classified and defined as shown in Figure 4E–H.

  • c

    Defect rate is calculated as number of embryos with mild, moderate, and severe phenotypes divided by number of embryos survived at 48 hpf.

  • d

    Rescue was performed by co-injecting 0.5 pmol MOII with 200 pg each of irxl1a and irxl1b, or 400 pg of irxl1a, or 400 pg of irxl1b.

  • hpf, hours postfertilization; WT, wild-type; MO, morpholino oligonucleotides.

Control MO0.595.3% (162/170)100%0%0%0%0% (0/162)
192.6% (87/94)94.3%0%0%5.7%5.7% (5/87)
irxl1 MOI (+ p53 MO)0.5 (+ 0.75)96.0% (237/247)49.8%35.0%13.5%1.7%50.2% (119/237)
1 (+ 1.5)86.4% (336/389)11.3%36.9%44.4%7.4%88.7% (298/336)
irxl1 MOII0.575.7% (109/144)23.9%25.7%23.9%26.6%76.1% (83/109)
Irxl1 MOII + Irxl1 cRNAdirxl1a + irxl1b (200 pg each)81.9% (236/288)45.8%25.0%17.8%11.4%54.2% (128/236)
400 pg irxl1a95.3% (101/106)58.4%12.9%22.8%5.9%41.6% (42/101)
400 pg irxl1b92.9% (52/56)51.9%28.8%15.4%3.8%48.1% (25/52)

Irxl1 Knockdown Causes Defective Brain Morphogenesis

At 24 hpf, the most obvious morphological features of irxl1 morphants are flat and small heads (Fig. 5A–F) with failure in the ventricle inflation (arrows in E, F). To examine if different brain regions are formed properly, we analyzed the expression of several marker genes by WISH. Wnt1 plays an important role in regulating neurogenesis and hindbrain segmentation in zebrafish (Amoyel et al.,2005). At 24 hpf, wild-type embryos expressed wnt1 in the dorsal midline of the midbrain, the mid–hindbrain boundary and the hindbrain (Fig. 5G). The irxl1 morphant embryos also expressed wnt1 in these regions but with reduced levels (Fig. 5H,I). The midline domain in the midbrain was slightly shortened, and segmentation of the hindbrain was not clear in the morphant embryos (arrows in Fig. 5H,I). To determine if rhombomere formation was affected, in situ hybridization with krox20 which is normally expressed in rhombomeres 3 and 5 (Fig. 5J) was performed. Although the morphant embryos expressed krox20, the expression domain in both the anterior–posterior and dorsal–ventral directions was reduced (double arrows in Fig. 5K,L). Examining with other patterning genes such as flh and pax2.1 also revealed similar results (data not shown), indicating that different brain regions are formed in the irxl1 morphants but with reduced extension along the A–P/D–V axes.

Irxl1 Is Required for Pharyngeal Arch Development

Because irxl1 is expressed in the arch region (Fig. 3D–I), we examined the formation of the pharyngeal arches of irxl1 morphants by analyzing the expression of certain markers. The development of the pharyngeal arches involves coordination of several disparate cell types: the externally covered ectoderm, the internal endoderm, and a mesenchymal core of neural crest and mesodermal cells (Yelick and Schilling,2002; Graham,2003). In the wild-type embryos, the endodermal pouches that separate the body of each arch can be clearly labeled by nkx2.3 (Fig. 6A,C), which is expressed in the endoderm of the pharyngeal arch and gut during embryogenesis (Lee et al.,1996). In the irxl1 morphant embryos, the endodermal lining of the arches was observed at 30 hpf (Fig. 6B), but the structure of the pouches became disorganized at 48 hpf (Fig. 6D), indicating that the endodermal pouches were formed but failed to organize properly. To evaluate pharyngeal muscle formation, we compared the expression of myoD in wild-type and morphant embryos. myoD transcripts are expressed in somites along the trunk in early embryos, and they become less abundant in the trunk and start to be expressed in all of the head muscle precursors at later stages (48–58 hpf; Lin et al.,2006), including the extraocular and pharyngeal arch muscles (Fig. 6E,G). In contrast to the wild-type embryos, the pharyngeal arch muscle and extraocular muscle precursors are largely lost in irxl1 morphants at 48 hpf (Figs. 6F, 9) and 58 hpf (Fig. 6H). Of interest, myoD expression in the trunk of the morphant is up-regulated at 48–58 hpf (see Fig. 9 Type 3 and Fig. 6H), whereas it is diminished at the same stage in the wild-type embryos. The loss of myoD expression in head muscles cannot be simply attributed to developmental retardation in the morphants, because morphants at later stages do not display the expression pattern of earlier stages (compare Fig. 6E,H and data not shown).

Figure 6.

Defects in the formation of pharyngeal arches in irxl1 morphants. A–S: Whole-mount in situ hybridization of nkx2.3 expression at 30 (A,B) and 48 (C,D) hours postfertilization (hpf), myoD expression at 48 (E,F), and 58 (G,H) hpf, dlx2 expression at 30 hpf (I–L), sox9a expression at 48 hpf (M–P), and runx2b expression at 48 hpf (Q–S). Both MOI (plus p53 MO) and MOII morphants are analyzed and shown as indicated. A–D,K–L,O–S: Dorsal views. E–J,M–N: Lateral views. B,D: The endodermal pouches form at 30 hpf (B) but are disorganized at 48 hpf (D). F,H: The pharyngeal arch muscle and some extraocular muscle precursors are missing in morphants (arrows). J,L: The expression domains of dlx2 are reduced in morphants, especially in the branchial arches (arrows). N,P,R,S: Note that expression domains of sox9a in the ceratobranchials (arrows in N,P) and runx2b in ceratobranchials 3–4 (arrows in R,S) and the parasphenoid (asterisks in R,S) are severely reduced in morphants. b, branchial arches; cb, ceratobranchials; ch, ceratohyal; cl, cleithrum; di, diencephalon; em, extraocular muscles; ep, endodermal pouches; h, hyoid arch; m, mandibular arch; MO, morpholino oligonucleotides; ms, Meckel's; pam, pharyngeal arch muscles; pq; palatoquadrate; ps, parasphenoid; te, telencephalon. Scale bars = 100 μm in A–P,Q–S.

Neural crest cells are crucial to vertebrate craniofacial development. They generate the majority of the connective and skeletal elements in the head and arches (Trainor,2005). The migrating crest cells can be visualized by dlx2 expression (Ellies et al.,1997). At 30 hpf, dlx2 expression is seen in the first (mandibular), second (hyoid), and three posterior (branchial) arches (Fig. 6I,K). In morphant embryos, dlx2 expression is slightly reduced in the mandibular and hyoid arches and greatly reduced in posterior branchial arches (Fig. 6J,L). Because defects in craniofacial neural crest cell often lead to malformation or loss of craniofacial cartilage elements, we then examined the expression pattern of a prechondrogenic marker, sox9a (Yan et al.,2002), and a differentiating skeletogenic cell marker, runx2b, in the irxl1 morphants. In the wild-type embryos at 48 hpf, sox9a expression was detected in the cartilage elements of the first two arches, including the Meckel's, palatoquadrate and ceratohyal cartilages, and in the posterior ceratobranchial cartilages (Fig. 6M,O). In contrast, the sox9a expression domains were either fused in the first two arches or greatly reduced in the posterior arches (Fig. 6N), making it difficult to reveal the formation of ceratobranchials in the morphants (arrow in Fig. 6P) Similarly, expression of runx2b in the parasphenoid and ceratobranchials (especially cb3-4) was severely reduced in both MOI and MOII morphants at 48 hpf (Fig. 6Q–S).

To determine if irxl1 already functions at earlier stages of arch formation, we also analyzed the expression of snail2 (snail1b) and sox9b in early neural crest cells at 18–24 hpf. At 18 hpf, snail2 was expressed in the head mesenchyme, neural crest, neural tube, and somites in the wild-type embryos (Fig. 7A). These domains of expression were all reduced in the morphants (Fig. 7B,C). There are three streams of migrating neural crest cells (Fig. 7D). Particularly, expression of snail2 in the second and third streams was severely affected in morphants (arrows in Fig. 7E,F). At 18 hpf, transcripts of sox9b were detected in neural crest, ventral diencephalon and otic vesicles (Fig. 7G), and the expression domains extended to the anterior forebrain, eye, epiphysis and were strong at rhombomere boundaries at 24 hpf (Fig. 7J,K). In the irxl1 morphants, sox9b expression in all these domains was significantly reduced, although not completely absent (arrows in Fig. 7H,I, and Fig. 7L–O). This result suggests that irxl1 already functions at early stages and may affect neural crest cell formation or migration. Thus, defects in neural crest cell formation/migration were followed by failure of pharyngeal arch morphogenesis in irxl1 morphant embryos.

Figure 7.

Expression of early neural crest cell markers is affected in irxl1 morphants. A–O: Expression of snail2 of embryos at 18 hours postfertilization (hpf; A–F) and sox9b at 18 hpf (G–I), and 24 hpf (J–O) are examined by whole-mount in situ hybridization. A,D,G,J,K: Wild-type (WT). B,E,H,L,M:irxl1 MOI (plus p53 MO) morphants. C,F, I, N,O:irxl1 MOII morphants. A–C,G–I,J,L,N: Lateral views. D–F,K,M,O: Dorsal views. Arrows indicate reduced expression domains. I–III indicate the three streams of migrating neural crest cells. ov, otic vesicle. Scale bar = 200 μm.

Severe disruption in pharyngeal cartilage formation was observed by alcian blue staining (Schilling et al.,1996) at 5 dpf. A large proportion (>80%) of morphants injected with 0.5 pmol irxl1 MOI (plus p53 MO) or MOII displayed reduced or malformed cartilage elements. The pharyngeal cartilages are well developed in the wild-type larva at 5 dpf (Fig. 8A,F). However, the Meckel's and palatoquadrate cartilages in all the affected morphants were reduced in size (Fig. 8B–D,G–I). The ceratohyals were either misplaced (Fig. 8B), with an increased angle (Fig. 8H) or did not extend anteriorly (Fig. 8I). The ceratobranchials were also underdeveloped (Fig. 8G,H) or nearly completely absent (Fig. 8I). Both MOI and MOII morphants showed similar types of phenotypes. The phenotype can be rescued by irxl1 cRNA co-injection (Fig. 8E,J; see below).

Figure 8.

The arch cartilage patterning is disrupted in irxl1 morphants. Cartilage elements were stained with Alcian blue at 5 dpf. A–E: Lateral views. F–J: Ventral views. A,F: Wild-type. B–D,G–I: Morphants of MOI (0.5 pmol plus 0.75 pmol p53 MO) and MOII (0.5 pmol) exhibiting different degrees of cartilage defects are shown as indicated. E,J: MOII morphants rescued by co-injection of irxl1(a+b) cRNA. Arrows indicate the malformed ch (B,I) and pq (H) cartilages; double arrows (H) indicate an increased angle between the ceratohyal cartilages. bb, basibranchials; cb, ceratobranchials; ch, ceratohyal; hys, hyosymplectic; m, Meckel's cartilage; pq, palatoquadrate. Scale bar = 100 μm.

Morphological Defects of irxl1 Morphants Can Be Partially Rescued by cRNA Injection

To confirm that the defective phenotypes caused by irxl1 MO were specific, we tested if the phenotypes could be rescued by irxl1 cRNA. Injection of 260 pg of irxl1 cRNA (130 pg each of isoform a and b) into embryos receiving 1 pmol of irxl1 MOI partially rescued the gross phenotypes at 24 hpf (Supp. Table S2). Because MOI partially overlaps with the injected cRNA, we also performed rescue experiments with MOII. The results indicate that there was a tendency of phenotype reverting similar to that observed by using MOI. Phenotypes including smaller head, abnormal jaw, curved tail, and shrinkage of yolk stalk were all reverted to some extent (Fig. 4J). Specific numbers in each category are shown in Table 1. Of interest, both isoforms of irxl1 cRNA can recue the phenotype alone and in a similar manner, suggesting that they are equally functional during embryogenesis.

The classification of gross phenotype at 25–48 hpf did not accurately reflect the severity of cartilage defects observed at 5 dpf. Because morphants with severe phenotypes died before 5 dpf, it is difficult to quantify the rescue effect at 5 dpf. We found that the craniofacial myoD expression pattern at 48 hpf was closely related to the cartilage phenotype at 5 dpf. Therefore, we scored myoD expression in the head muscle precursors at 48 hpf as a basis for phenotypic classification (Fig. 9). Type 1 phenotype refers to partial loss of ventral pharyngeal arch muscle precursors (im, chv, sh); Type 2 refers to additional loss of most extraocular muscle precursors (so, sr, io); and embryos that completely lack head muscle precursors are classified as Type 3. When both forms of irxl1 cRNA was co-injected with 0.5 pmol of irxl1 MOI, the percent of embryos with normal and type 1 phenotypes increased from 48.8% to 75%, while those with type 2 and type 3 phenotypes decreased from 51.2% to 25%. At a higher dose of MOI (1 pmol), a large proportion of embryos (72.6%) completely lost the head muscle precursors. This proportion was significantly reduced to 29.1% by cRNA co-injection, while the proportion of normal and type 1 phenotypes increased from 13.2% to 38.2% (Fig. 9). The rescued larva displayed normal arch cartilage formation at 5 dpf (Fig. 8E,J). Thus, specific craniofacial defects, as well as gross morphological defects, can be rescued by irxl1 cRNA.

Figure 9.

Craniofacial defects in irxl1 MOI morphants can be rescued by cRNA injection. myoD expression in head muscle precursors of 48-hours postfertilization (hpf) embryos was used as a score to classify the craniofacial defects. Wild-type (WT), myoD expression observed in all head muscle precursors; Type 1, partial loss of ventral pharyngeal arch muscle precursors (im, chv, sh); Type 2, loss of some extraocular muscle precursors (so, sr, io, ir) in addition to type 1; Type 3, complete lack of head muscle precursors. Embryos were injected with 0.5 or 1 pmol of irxl1 MOI plus 0.75 or 1.5 pmol of p53 MO. In rescue experiments, 290 pg of irxl1 cRNA (145 pg irxl1a plus 145 pg irxl1b) was co-injected. ah, adductor hyoideus; am, adductor mandibulae; ao, adductor operculi; cd, constrictor dorsalis; chv, constrictor hyoideus ventralis; im, intermandibularis; io, inferior oblique; ir, inferior rectus; lr, lateral rectus; MO, morpholino oligonucleotides; mr, medial rectus; sh, sternohyoideus; so, superior oblique; sr, superior rectus. Scale bar = 100 μm.


Irxl1 was first identified in mouse as a novel member of the TALE superclass of homeobox genes (Anderson et al.,2006; Liu et al.,2006; Takeuchi and Bruneau,2007). Database searching revealed that the predicted irxl1 orthologs exist in the genomes of various species. We have cloned the zebrafish irxl1 gene and have shown that it is highly conserved among vertebrate species. Notably, the homeodomain exhibits 100% homology to the vertebrate orthologs. In addition to the protein sequence, the genomic organization of irxl1 is also homologous. The genes are organized into seven exons and span 73 kb, 70 kb, and 47 kb in the human, mouse and zebrafish genomes, respectively. The exon sizes vary from 34 bp in exon 6 (all three species) to 2,541 bp (human), 2,055 bp (mouse), and 1,882 bp (zebrafish) in exon 7. The intron sizes are relatively large, with intron 5 being the largest (59 kb in human, 55 kb in mouse, and 27 kb in zebrafish; Weinmann et al.,2005; Anderson et al.,2006). The high degree of homology in irxl1 gene structure and sequences in vertebrate species, especially in the homeodomain region, strongly suggests that it is a key regulator of vertebrate development. Interestingly, two splicing variants of irxl1 are found in zebrafish embryos and adult tissues. Because the b isoform has not been reported in other species, it is not clear at this moment if it represents a unique form or plays any specific role in zebrafish.

During mouse embryonic development, expression of Irxl1 RNA was first detected at embryonic day 8.5–9.5 (E8.5–E9.5) in the dorsal region of the dermamyotome of anterior somites (Anderson et al.,2006; Liu et al.,2006; Takeuchi and Bruneau,2007). Later (E10.5–E11.5), the expression domain extended posteriorly to the trunk and tail somites, mainly restrictive to the progenitor cells of skeletal muscle, tendon, and cartilage (Anderson et al.,2006). Consistently, RT-PCR analysis revealed that irxl1 expression in developing zebrafish embryos began at 18 hpf (18 somites), which is comparable to mouse embryonic day (E) 9.5 (20-somite stage). Although WISH analysis failed to reveal a distinct pattern of irxl1 transcripts in developing somites, expression of irxl1 in the trunk region was confirmed by RT-PCR of 24–72 hpf embryos (Supp. Fig. S1). It is possible that low levels of irxl1 are expressed in the trunk region, or alternatively, only a few cells in this region express irxl1. Indeed, mouse Mkx is not transcribed in the myotome compartment of the somites, rather its expression is restricted to the dorsomedial and ventrolateral lips of the dermamyotome where the myogenic progenitors are found (Anderson et al.,2009).

At 24 hpf, clear expression of zebrafish irxl1 was detected in the brain tissue, including the telecephalon. In mouse embryos, a high level of Irxl1 RNA expression was also detected in the forehead region ventral to the telencephalic vesicles (Liu et al.,2006). In addition, strong Irxl1 expression was found in neural crest-derived head mesenchyme, palatal mesenchyme, and preossification mesenchymal cells surrounding the Meckel's cartilage in the mandible (Liu et al.,2006). This pattern of craniofacial expression is similar in both mouse and zebrafish embryos. Furthermore, zebrafish irxl1 is widely transcribed in various adult tissues, including muscle and testis, which were shown to express Irxl1 in mouse. Thus, zebrafish irxl1 may regulate developmental processes primarily at a later stage or only in a subset of cells, and is thus minimally expressed in the trunk during somitogenesis. Recently, mouse Irxl1 (Mkx) was demonstrated to function as a transcriptional repressor. When introduced into 10T1/2 fibroblasts, Irxl1 was capable of blocking the MyoD-driven conversion of fibroblasts to myoblasts (Anderson et al.,2009). This is consistent with our observation that myoD transcription in the trunk somites is up-regulated in irxl1 morphants at 48–58 hpf (Fig,. 6H, 9). In fact, our preliminary data indicate that zebrafish irxl1 can repress myoD promoter activity (unpublished results). Thus, irxl1 may not be necessary during early stages of myogenic differentiation, but its activity is required to down-regulate myogenic factors after completion of myogenesis.

Irxl1 is most closely related to the irx genes, which have been shown to regulate proneural gene expression and specify the neuroectoderm both in Drosophila and in vertebrates. In collaboration with other transcription factors, irx genes also participate in subdivision of the neural plate (Gomez-Skarmeta and Modolell,2002). In zebrafish, the 11 irx genes are expressed in distinct domains of the nervous system and are suggested to participate in neural patterning, neurogenesis and neuronal specification (Lecaudey et al.,2005). Irxl1 is also expressed in the nervous system in both mouse and zebrafish. Our data indicate that knockdown of irxl1 results in defective brain morphogenesis. In the affected embryos, differentiation of all major regions of the brain and regional specification along the anterior–posterior and dorsal–ventral axes of the nervous system appear normal as assessed by the expression of specific marker genes (Fig. 5G–L and data not shown). However, the brain fails to expand normally along the anterior–posterior axis and ventricle formation in the forebrain and midbrain is defective. Thus, unlike the Irx genes, irxl1 does not seem to participate in early neural specification and patterning. Instead, it may function in brain morphogenesis at a later stage. This is also consistent with its relatively late stage of expression during zebrafish embryogenesis.

In addition to brain morphogenesis, the most prominent defect in the irxl1 morphants was observed in the pharyngeal arches. The development of pharyngeal arches involves endodermal, mesodermal, and neural crest derivatives, and all of these components showed defects in the irxl1 morphant embryos. First, the endodermal pouches were present but became fused and disorganized (Fig. 6D). It has been shown that endoderm, as well as the neural crest, plays a prominent role in establishing and patterning the pharyngeal arches (Graham,2003). Thus, disorganization of endodermal pouches may lead to deformation of the arches in the morphants. Second, most of the head muscles originate from cranial paraxial mesoderm. With interactions of the surrounding cranial neural crest, the mesoderm differentiates into the head muscle primordia that further form extraocular muscles and pharyngeal arch muscles (Noden et al.,1999; Kimmel et al.,2001). As evident by myoD expression, the formation of head muscles, most notably the pharyngeal arch muscles, are disrupted in irxl1 morphants (Fig. 6F,H). Lastly, analysis of the neural crest markers sox9b and snail2 at 18–24 hpf revealed a significant reduction in the expression levels and domains in irxl1 morphants (Fig. 7), suggesting that early neural crest cell differentiation and migration are already affected by irxl1 knockdown. This is further supported by examining dlx2 expression at a later stage (30 hpf). The reduction of migratory neural crest cells were clearly observed in the pharyngeal arches, especially in the posterior branchial arches in the morphants (Fig. 6J,L). Subsequently, the neural crest detects lead to reduced condensation of prechondrogenic cells (sox9a+) and differentiation of skeletogenic cells (runx2b+; Fig. 6M–S), resulting in abnormal pharyngeal cartilage formation as observed at 5 dpf (Fig. 8). The defects in gross morphology and in craniofacial muscle formation in morphants can be largely rescued by irxl1 cRNA injection, indicating that the phenotypes are caused by loss of irxl1 function.

Because neural crest cells generate the majority of the cartilage, nerve, and connective tissues in the head, craniofacial abnormalities are largely attributed to defects in the formation, migration, and differentiation of neural crest cells (Trainor,2005). Our data suggest a potential role of irxl1 in regulating neural crest development. Of interest, the mouse Irxl1 gene was identified at the Twirler (Tw) locus and was considered a good candidate for the Tw gene based on its developmental expression pattern (Liu et al.,2006). The Twirler mutation causes cleft lip and cleft palate in homozygous mutant mice. The phenotype of defective cartilage formation in the arches of zebrafish irxl1 morphants strongly supports the notion that mouse Irxl1 gene is responsible for the Tw mutation.

In summary, we have cloned and analyzed the zebrafish irxl1 gene. Inhibition of irxl1 translation results in defective brain morphogenesis without affecting regional specification of the nervous system. It also causes a defect in neural crest cell formation and/or migration and consequently results in loss of craniofacial muscles and deformed arch cartilages. These results suggest that the irxl1 protein may regulate factors that are required for neural crest and chondrocyte differentiation and thus is involved in arch morphogenesis in zebrafish.


Fish Stocks

Zebrafish (AB strain) embryos were obtained from the Zebrafish International Resource Center at the University of Oregon and were raised and maintained according to standard laboratory conditions (Westerfield,2000). The embryos were reared in 0.003% PTU to prevent pigmentation and staged at 28°C according to Kimmel et al. (1995).

Cloning of Zebrafish irxl1 cDNA and Sequence Analysis

A predicted zebrafish irxl1 ortholog (XM_678274) was identified by TBLASTN search of the GenBank using the homeodomain sequence of the mouse Mkx ortholog (NP_808263). Several sets of primers were designed based on this predicted sequence to perform RT-PCR and 5′-/ 3′-RACE PCR. RT-PCR was carried out using 5 μg of total RNA isolated from 48 hpf zebrafish embryos. The PCR products were then cloned into the pGEMT-easy vector (Promega) and sequenced. The obtained sequences were analyzed by GeneWise and compared with the genomic sequences to determine the axon/intron boundaries and confirm the splicing patterns. The sequences of two splicing variants of the zebrafish irxl1 gene were submitted to NCBI under the accession numbers EF457976 (irxl1a) and EF457977 (irxl1b). Multiple sequence alignments of the Irxl1 orthologs were generated using ClustalW and the results are presented using BoxShade.


Total RNA was isolated from 1.5 to 120 hr postfertilization (hpf) wild-type embryos and from various adult tissues according to the manufacturer's protocol of TRI-Reagent (Molecular Research Center, Inc.). After solubilization, the RNA solution was treated with DNase for 15 min followed by phenol/chloroform extraction and ethanol precipitation to remove residual DNA. First-strand cDNA was synthesized from 5 μg of total RNA by oligo (dT)15 and SuperScript II (Life Technologies), according to the manufacturer's protocol. One twentieth of the cDNA was used for PCR amplification. The primers used for irxl1 expression were irxl1-forward: 5′-CAGCCAGGTCTTATTCGAGG-3′, irxl1a-reverse: 5′-TTTCTGGATGATGCAGCTGG-3′ (964 bp) and irxl1b-reverse: 5′-TGATAATGGGCTGAAGAAACG-3′ (875 bp). Thirty-five cycles of PCR reactions (94°C for 30 sec, 55°C for 30 sec, and 72°C for 1 min) were performed. Amplification of the elongation factor 1α (5′-GCTCAAGGAGAAGATCG-3′ and 5′-TCAAGCATTATCCAGTCC-3′) was used as an internal control. RACE-PCR was performed using a SMART RACE cDNA Amplification Kit (Clontech) according to the manufacturer's protocol.

Whole-Mount In Situ Hybridization and Immunofluorescence Staining

Antisense RNA probes were synthesized by in vitro transcription using digoxigenin (Roche) as a label. WISH was performed as previously described (Jowett,2001; Thisse and Thisse,2008) using 20- to 100-ng riboprobes. Hybridization and washing were performed at 65°C. The zebrafish irxl1a was cloned into pET29(b) which was then transformed into BL21(DE3) to express the His-tagged Irxl1 protein. After purification by His-bind resin (Novagen), the recombinant protein was mixed with Visual Protein and injected into rabbits to obtain the antiserum. The antiserum was further purified by ammonium sulfate precipitation and protein A affinity column. Initial characterization of this antibody indicated that it could detect exogenously-expressed Irxl1 protein in cultured C2C12 cells but failed to detect endogenous protein in either embryos or adult tissues by Western blotting (Supp. Fig. S4). For immunofluorescence staining, a 1:200 dilution of this antibody and a FITC anti-rabbit antibody (Santa Cruz, 1:200) were used in either whole-mount embryos or cryosections according to the established protocols (Westerfield,2000).

Injection of Morpholinos and cRNA

Antisense morpholino oligonucleotides (MO) were designed and purchased from Gene Tools (Philomath, OR). The full-length cDNA of irxl1a and irxl1b were amplified by PCR and cloned into pCS2+. Capped RNA was synthesized from pCS2+-irxl1a and pCS2+-irxl1b using the mMessage mMachine kit (Ambion) according to the manufacturer's procedure. Morpholinos were dissolved in Danieau solution (58 mM NaCl, 0.7 mM KCl, 0.4 mM MgSO4, 0.6 mM Ca(NO3)2, and 5 mM HEPES, pH 7.6) and, with 0.05% phenol red, were injected at a dose of 0.5 pmol (∼4 ng), 1 pmol (∼8 ng), or 2 pmol (∼16 ng) per embryo into one- or two-cell embryos. To suppress the off-target effect, p53 MO was co-injected at a dose of 1.5-fold of the MOI used (Robu et al.,2007). For rescue experiments, embryos receiving 0.5 to1 pmol MOI or MOII were co-injected with 290–400 pg of irxl1 cRNA (145–200 pg each of irxl1a and irxl1b), 400 pg of irxl1a, or 400 pg of irxl1b. Control MO: 5′CCTCTTACCTCAGTTACAATTTATA3′; p53 MO: 5′GCGCCATTGCTTTGCAAGAATTG3′; Irxl1-MOI: 5′GTGTTCATCTTTGGTATATTAGTCC3′, Irxl1-MOII: 5′CCAGTTAGACACCTGAAAATAAAAC3′.

Acridine Orange and Alcian Blue Staining

For cell death detection, 25 hpf embryos were dechorionated and incubated in 5 μg/ml acridine orange in E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4, 10 mM HEPES, pH 7.2) for 30 min. The embryos were then washed in E3 medium three times and anesthetized with 0.64 mM tricaine before observation under a fluorescence microscope. Cartilage staining was carried out as previously described (Schilling et al.,1996) with slight modifications. Larvae were fixed in 4% paraformaldehyde in phosphate buffered saline (PBS) overnight at 4°C, rinsed with PBS, and then transferred into 0.1% alcian blue dissolved in 80% ethanol/20% glacial acetic acid at room temperature overnight. The specimens were rinsed in 90% ethanol/10% glacial acetic acid and rehydrated gradually into PBS. Tissues were cleared in 0.05% trypsin dissolved in saturated sodium tetraborate for 3 hr, followed by bleaching in 3% hydrogen peroxide/1% potassium hydroxide for three to five hours. After rinsing in PBS and post-fixing in 4% paraformaldehyde, the specimens were mounted in 50% glycerol /PBS for microscopic observation.


We thank Drs. S-H Wang, J-N Tsai, Y-C Cheng, Y-H Chen, and C-H Hu for helpful discussion and riboprobes. H.P. and K.-M.H. were funded by the National Science Council of Taiwan.