SEARCH

SEARCH BY CITATION

Keywords:

  • morpholino;
  • Xenopus tropicalis;
  • microcephaly;
  • head development;
  • apoptosis

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

The diploid frog X. tropicalis has recently been adopted as a model genetic system, but loss-of-function screens in Xenopus have not yet been performed. We have undertaken a pilot functional knockdown screen in X. tropicalis for genes involved in nervous system development by injecting antisense morpholino (MO) oligos directed against X. tropicalis mRNAs. Twenty-six genes with primary expression in the nervous system were selected as targets based on an expression screen previously conducted in X. laevis. Reproducible phenotypes were observed for six and for four of these, a second MO gave a similar result. One of these genes encodes a novel protein with previously unknown function. Knocking down this gene, designated pinhead, results in severe microcephaly, whereas, overexpression results in macrocephaly. Together with the early embryonic expression in the anterior neural plate, these data indicate that pinhead is a novel gene involved in controlling head development. Developmental Dynamics 289–299, 2004. © 2003 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

The frog X. laevis is an excellent organism for studying developmental processes but its allotetraploid genome makes it unsuitable as a genetic model system. The related frog X. tropicalis retains the advantages of external, fast development and short generation time but has a diploid genome (Amaya et al., 1998). Therefore, X. tropicalis is increasingly used as an alternative to X. laevis (Offield et al., 2000; Nutt et al., 2001; Hirsch et al., 2002b; Khokha et al., 2002; Polli and Amaya, 2002; Carruthers et al., 2003; D'Souza et al., 2003; as reviewed in Amaya et al., 1998, and Hirsch et al., 2002a). X. tropicalis has also been the choice for a genomic and a large-scale expressed sequence tag (EST) sequencing project (http://genome.jgi-psf.org/xenopus0/ and http://www.sanger.ac.uk/Projects/X_tropicalis/). However, the function of the vast majority of the genes sequenced in these projects remains to be determined.

Several gain-of-function screens in X. laevis have been reported (e.g., Grammer et al., 2000), but to date, screens based on gene disruption in Xenopus have not been described. Efforts to apply forward genetics in X. tropicalis through both chemical and insertional mutagenesis (Bronchain et al., 1999; reviewed in Hirsch et al., 2002a) are promising, but they are still under development. RNAi does not appear to reliably inhibit gene function in Xenopus embryos (unpublished data), but there are now many reports of using MO oligos to interfere with protein production from single gene transcripts in frog and fish embryos (reviewed in Heasman, 2002). These oligos are usually designed to anneal to the region of mature RNA containing the initiator methionine where they interfere with protein translation. They are resistant to endonucleases and, after injection into single-cell frog embryos, can suppress protein translation until late embryogenesis (i.e., stage 43: Nutt et al., 2001). In the fish, many known mutations have been phenocopied by MO injections (Nasevicius and Ekker, 2000; Feldman and Stemple, 2001; Lele et al., 2001). MO oligos complementary to pre-mRNA splice sites can also suppress protein function by interfering with RNA processing (Draper et al., 2001). Here, we describe a pilot screen using MO oligos in X. tropicalis, to identify novel influences on nervous system development.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Selection of Target Genes and Sequences

Target genes were selected using a large expression database generated by in situ hybridisation in Xenopus laevis (Gawantka et al., 1998; and www.dkfz-heidelberg.de/abt0135/axeldb.htm). From this database, we selected 57 genes that showed specific, primary, or regionalised expression in the nervous system. Of these 57 genes, 41 were found to have orthologues in the X. tropicalis EST database (www.sanger.ac.uk/Projects/X_tropicalis/) in January 2002. For 25 of these (Fig. 1), 5′ sequences contained a potential start ATG, as evidenced either by the presence of an upstream termination codon or by comparison with known orthologues in other species. The X. tropicalis EST sequences have been clustered (www.sanger.ac.uk/Projects/X_tropicalis/ and Gilchrist et al., manuscript in preparation), and all the clones selected in this work were part of a cluster. Therefore, an additional criterion was that the candidate ATG and the MO target sequence must be the same in the majority of the clones in a cluster. The final selection contained genes of several functional classes as well as genes of unknown function (see Fig. 1).

thumbnail image

Figure 1. Targeted genes are grouped according to a functional class. The expressed sequence tag (EST) name for each sequence as assigned by Gawantka et al., (1998) and http://www.dkfz-heidelberg.de/abt0135/axeldb.htm for Xenopus laevis is given next to the best protein match for each translated sequence. Where a protein may not represent an orthologue but is homologous, it is suffixed by “like.” Red, reproducible between experiments; blue, not reproducible between experiments; green, early lethality; black, no phenotype.

Download figure to PowerPoint

MOs (25mers) were designed within the region surrounding the start ATG (−80 bp to +30 bp relative to the A of the start codon, Supplementary Table 1) according to the criteria described on the Gene Tools Web site (www.gene-tools.com). For an additional EST (25H12.1), we designed MOs to bind to the 3′ splice site of an intron.

Scoring Criteria

MO oligos were injected into fertilised X. tropicalis embryos before the first cleavage, and embryos were primarily scored for any alterations in morphology at neurula and tadpole (up to ∼ stage 30) stages. Morphologic scoring included microcephaly/macrocephaly, other head and eye defects, cell death, duplicated axes, neural tube closure problems, tumours, and pigmentation defects. To visualise the developing nervous system, surviving embryos were also processed by in situ hybridisation with Sox3, which stains undifferentiated neuronal precursors, and either N-tubulin or ElrC, which mark cells at different stages of neuronal differentiation (Carruthers et al., 2003). Each MO was injected on at least two occasions. Morphologic defects or changes in staining pattern were scored as “reproducible” phenotypes if they were present in two experiments in over 50% of embryos and “inconsistent” if they were not reproduced between experiments. Several studies have shown that, while treatments with some MOs reproduce mutant phenotypes with high penetrance, others result in variable severity or can induce nonspecific effects, including developmental delay, microcephaly, and shortening (Nutt et al., 2001; reviewed in Heasman, 2002). Indeed, in the present study, degrees of developmental delay and shortening of the anterior/posterior axis were frequently seen. To gain additional evidence for specificity in the cases where targeting with one MO (moA) had produced a “reproducible” phenotype, a second MO (moB) was designed. In addition, the expression patterns of these genes were investigated by in situ hybridisation in X. tropicalis embryos.

Overview of the Results

In the majority of cases (15 of 26 MOs), we were not able to ascribe a phenotype to an MO, because embryos appeared normal under our scoring criteria up to ∼ stage 30 (Fig. 1; Supplementary Table 1). MOs for five ESTs (Fig. 1; Supplementary Table 1, which is available at www.mrw.interscience.wiley.com/suppmat/1059-8388/suppmat) were found to produce very variable effects between experiments. These genes are Sox3 (EST = 30F5.2), Sox2 (19F1.1), Pitx2 (17A1), Hes5 (8C9), and Marcks related protein-like (6A5). MOs to Sox3 and Sox2 also gave variable phenotypes within one experimental batch of embryos that did not reproduce. Targeting Hes5 did not show any anomalies in neuronal differentiation that we could detect, but suffered from a neural closure defect over the hindbrain in one of three experiments. MOs to Pitx2 and Marcks related protein-like gave some craniofacial dysmorphy that was not reproducible. It is possible that these phenotypes are a true reflection of the biological activity of these genes, but we did not pursue these possibilities further. A MO against Tbx2 (25H3.1moA) caused strong embryonic lethality between blastula and neurula stages and was not analysed further. Of the 26 targeted genes, MOs to 6 did show a reproducible phenotype, and of those, 4 were further reproduced by a second MO (Supplementary Table 1; Fig. 2A). These phenotypes, therefore, are very likely to be specific to the gene in question and are presented in more detail below.

thumbnail image

Figure 2. A: Reproducible phenotypes induced by morpholino (MO) injection. Control uninjected embryos and embryos injected with each MO, as indicated. All are hybridised with a neural marker (blue staining), apart from (p–s) which are unstained, as follows: a–d, ElrC; e,f, Sox3; g–o, N-tubulin. The main features of the phenotypes caused by the MOs were as follows: 16G8 causes shortening of the anterior–posterior axis, 26A10.2 causes a reduction of ElrC staining particularly in the cranial nerve region (arrowhead in d) and broadening of the hindbrain (arrowhead in f), 17D4 causes posterior truncation (h) and/or bifurcated axis (i), 26E1.1 results in enhanced posterior development/open proctodeum (arrowhead in k and l) and hypodorsalisation, 18E6 causes cell death that starts anteriorly (n) and spreads posteriorly (o), and 25H12.1 causes extreme microcephaly. Most are lateral views apart from e,f,m,n,o,r,s, which are dorsal views. Anterior is to the left in all panels. Each set of control and experimental embryos was photographed under the same magnification (×40–×60). B: Expression patterns of targeted genes by in situ hybridisation. a: Expression of 16G8 (blue) along the neural tube of a neurula embryo. b: Expression of 16G8 (magenta) in a broad band around the equator of a gastrulating embryo, blastopore side down. c: 26A10.2 expression is high in neural tissue of neurula embryos. d: 17D4 expression in a neurula embryo. e: 26E1.1 expressed at the anterior neural tube. f: 18E6 expression at the anterior end of the neural tube. g: 25H12.1 expression in mid-anterior neural plate in two bilateral patches (black arrowhead) and the anterior neural ridge (white arrowhead). h,i: 25H12.1 is expressed around the blastopore of gastrulating embryos but excludes the dorsal mesoderm highlighted by in situ hybridisation to chordin (in light blue). All neurula embryos are viewed from the dorsal side with anterior to the front.

Download figure to PowerPoint

16G8 (Stunted)

The first MO directed against the translational start site of 16G8 (16G8moA) resulted in embryos that were severely truncated along the anterior/posterior axis (A/P) and had small heads (100%; n = 106; Fig. 2A,b). Neural staining with ElrC (Fig. 2B) and Sox3 (not shown) were present but reduced. A/P shortening has been reported as a nonspecific effect of MO injection (reviewed in Heasman, 2002), and indeed, we observed some degree of shortening with several other MOs. Therefore, to test whether 16G8moA-injected embryos are unusually short, we compared the mean lengths of these embryos with those of embryos injected with other test MOs. This histogram in Figure 3 shows that A/P shortening accompanies all five reproducible phenotypes tested and is seen for two other MOs (13B4moA and 6A5moA) of otherwise wild-type or inconsistent appearance, respectively. Some A/P shortening in these cases may be accounted for, at least partly, by developmental delay. By contrast 16G8moA causes over 40% A/P reduction without significant developmental delay. Injection with a second MO directed against the initiator ATG region of this gene (16G8moB) resulted in similarly microcephalic embryos with marked A/P shortening (38%). These embryos also possessed numerous cutaneous growths (Supplementary Fig. 1b).

thumbnail image

Figure 3. Percentage of anterior/posterior (A/P) length of morpholino (MO) -injected embryos. At least 30 injected embryos were measured at tadpole stage, and the average values normalised against their respective controls, represented here by the 100% gridline. MOs are categorised as those producing inconsistent, reproducible, or no phenotype. The fluorescein isothiocyanate (FITC) MO is a GeneTools standard directed against a sequence with no counterpart in Xenopus. Light bars indicate that standard deviations overlapped with those of control embryo, whereas dark bars indicate no overlap.

Download figure to PowerPoint

16G8 shows high expression along the neural tube of late neurula X. tropicalis embryos, including the eye anlage (Fig. 2B,a). While this finding may be relevant to the microcephaly and reduction of neural staining, the A/P-short phenotype is most likely to be due to interference with an earlier function, and indeed, 16GB is expressed around the equator of the gastrula stage embryos in X. tropicalis (Fig. 2B,b).

16G8 is highly similar to Xenopus laevis genes XIdx (97%; Wilson and Mohun, 1995; accession no. S76880) and XId3 (95%; Zhang et al., 1995; accession no. AJ292558), which encode basic helix-loop-helix (HLH) proteins of the Id family of transcription factor inhibitors that are involved in negatively regulating HLH proteins. Although XIdx inhibits MyoD (Wilson and Mohun, 1995) and Xid3 may be activated by Notch (Zhang et al., 1995), their precise function in Xenopus development is not known. In the mouse, ablation of Id3 alone has no observable neural phenotype, whereas simultaneous elimination of Id1 and Id3 results in abnormalities of neurectodermal angiogenesis and reduction in brain size due to premature neural differentiation (Lyden et al., 1999). Of interest, these embryos also show an overall 30% reduction in size (Lyden et al., 1999).

26A10.2 (Paralysed)

At first inspection, embryos that had been injected with 26A1.1moA appeared morphologically normal, apart from a small delay, slight A/P truncation and a broad hindbrain by stage 25 (Fig. 2A,e,f). The first indication of major neural abnormality was a reduced staining of ElrC over the cranial ganglia region (Fig. 2A,c,d), although Sox3 appeared normal (Fig. 2A,f). Further inspection revealed that the main defect was the lack of hatching response and of any response to touch even at stage 35, when primary neurons readily mediate an escape response in control embryos (100%, n = 71, supplementary movie at http://www.welc.cam.ac.uk/∼amayalab/magomovies.htm). A second MO (26A10.2moB) had milder morphologic effects (slight microcephaly and oedematous heart chamber by stage 35, reduced eye pigment, no hindbrain malformation; Supplementary Fig. 1d), but all embryos were paralysed (100%, n = 121). Sox3 staining was normal, as for 26a10.2moA, but ElrC staining was diffuse (Supplementary Fig. 1d).

The 26A10.2 sequence corresponds to the X. tropicalis orthologue of mago nashi (accession no. AF007860). mago nashi was first identified in Drosophila as a gene involved in mRNA transport. Subsequently, it was found to be highly conserved in all vertebrates (98–100%), but its developmental function is unknown. mago nashi has been shown to be a component of the exon–exon junction complex associated with spliced mRNA in higher eukaryotes. This complex is exported with mRNA to the cytoplasm and is involved in nuclear export, mRNA localization, and nonsense-mediated mRNA decay (reviewed in Palacios, 2002). mago nashi RNA is ubiquitously expressed in X. tropicalis gastrulae (not shown) and widely expressed at the neurula and tadpole stages particularly in the developing nervous system (e.g., Fig. 2B,c).

17D4 (Truncated)

Injection with 17D4moA resulted in 42% (n = 36) of surviving embryos having a gastrulation defect presenting as bifurcated axis (spina bifida; Fig. 2A,i). The rest were abruptly truncated at the posterior end, with little or no tail bud (Fig. 2A,h). Of interest, the shape of these truncated embryos was otherwise normal as was head development and neuronal staining with Sox3, ElrC, and N-tubulin. A second MO to this gene (17D4moB) produced morphologically normal embryos; therefore, we cannot draw any conclusions yet about the specificity of this phenotype. 17D4 is the X. tropicalis orthologue of O-linked N-acetylglucosamine (GlcNAc) transferase (98% identity, accession no. AAH14434). This enzyme modifies a wide variety of proteins, including cytoskeletal proteins and signal transduction and transcription factors. It mediates a novel glycan-dependent signalling pathway that has a dynamic interaction with phosphorylation (Wells et al., 2001), and although it is found in many organisms, its role in early development has not yet been investigated. 17D4 RNA is detected weakly throughout X. tropicalis neurula embryos with enhanced staining in the neural tube (Fig. 2B,d). At tadpole stages the gene is expressed primarily in the head (not shown).

26E1.1 (Enhanced Posterior)

Targeting of 26E1.1 with 26E1.1moA resulted in high preneural mortality (75%). Surviving embryos exhibited a complex phenotype consisting of varying degrees of hypodorsalisation up to complete lack of axis and neural staining (Fig. 2A,l) posteriorisation (small head and enhanced posterior–ventral development) as well as enlarged and open proctodeum (Fig. 2A,k,l). However, a second MO directed against this gene did not produce a phenotype. Therefore, no conclusion can yet be drawn about the specificity of the phenotype.

26E1.1 encodes HMG4 (92% identity, accession no. 015347; Wilke et al., 1997), a member of the High Mobility Group of chromatin-associated proteins. These proteins can act as modulators of transcription through their ability to influence chromatin architecture (reviewed in Wolffe, 1999). 26E1.1 RNA is ubiquitous in X. tropicalis gastrulae (not shown) but in late neurula embryos is seen only in the anterior neurectoderm (Fig. 2B,e). Later on, in late neurula and tadpole stages, expression is present throughout the developing nervous system (not shown).

18E6 (Anterior Cell Death)

After injection with MO, 18E6moA embryos underwent apparently normal gastrulation and began neuronal differentiation and neural tube closure. However, at approximately stage 19 to 21, cells of the anterior neuroectoderm underwent lysis, revealing the underlying tissue (N-tubulin–stained embryos are shown in Fig. 2A,n,o). From the anterior end, cell death spread posteriorly, and these embryos died without further development. A second MO (18E6moB) resulted in an almost identical phenotype (Supplementary Fig. 1f,g). In X. tropicalis neurula embryos, the gene is expressed in the anterior portion of the neuroectoderm where MO-induced cell death is first seen (Fig. 2B,f).

The gene targeted by this MO encodes a X. tropicalis homologue (73% identity) of a mammalian protein that has structural similarity to armadillo family proteins (nuclear associated protein or NAP, accession no. AAK27389.1). Recently, this protein has been renamed CTNNBL1, and, of interest, it has been found to induce apoptosis when expressed in Chinese hamster ovary cells (Jabbour et al., 2003). Our data suggest that this protein also has an important role as a negative regulator of cell death at least in Xenopus development.

25H12.1 (Pinhead)

MOs to 25H12.1 reduce anterior development.

For 25H12.1, no upstream termination codon existed before the first ATG, therefore we could not be certain that this was the true initiator ATG. However, a single sequence (TGas046h20) of eight found in the database included an insertion, flanked by GT and AG sequences, indicative of an unspliced intron. As MOs have also been reported to effectively knockdown gene function by interfering with hnRNA splicing, we designed MO oligos to bind to the 3′ splice site of this putative 25H12.1 intron.

Injection with 25H12.1moA produced a marked reduction in anterior structures, appearing as a narrowing of the head (hence pinhead) in most embryos (85%, n = 100; Fig. 2A,q,s). The narrowing of the head structures was accompanied by an increase in posterior–ventral development and was first evident at late neurulation. A second MO to the splice junction (25H12.1moB) had a similar effect (81%, n = 48; Supplementary Fig. 1i,k). Because abnormal splicing may mimic hypomorphic alleles, we also targeted the initiator ATG. The 5′ end of the cDNA was first defined by 5′ rapid amplification of cDNA ends (RACE), which added only 17 bp to the existing sequence (Clontech SMART system). Furthermore, Northern blot analysis gave a band of approximately the same size as the clone insert (2.5 kb, not shown), suggesting the correct 5′ end had been isolated. Although no upstream stop codon is present, the most 5′ ATG in the 25H12.1 sequence has a reasonable “Kozak” transcription start consensus sequence (cagacATGg, Supplementary Fig. 2). An MO directed against this site produced exogastrulation in all embryos (n = 45) with spina bifida (32%) or microcephaly (68%) in survivors. Therefore, suppression of translation of this gene reproduced the microcephalic phenotype but also produced an earlier, more severe phenotype.

Pinhead MOs cause aberrant splicing.

It was possible to check whether the splice junction MOs were causing aberrant splicing of 25H12.1 transcripts by reverse transcriptase-polymerase chain reaction (RT-PCR). Figure 4 shows that both MOs cause abnormal splicing. In the untreated sample, the expected band of 187 bp is seen. Two predominant bands of ∼350 bp and <100 bp are seen after MO treatment in addition to a small amount of wild-type 187-bp product. An ∼ 600-bp genomic band was also seen for all cDNA preparations. Sequencing this product revealed two introns of 179 bp (intronA) and 228 bp (intronB) in this interval. Intron A is identical to the targeted intron detected in TGas046h20. The aberrant splice products were sequenced. The larger (366 bp) was found to result from retention of intron A and the smaller (91 bp) from skipping of the 96-bp exon B (Fig. 4A). The former would result in a shift of the translational reading frame, whereas the latter would give a protein with a 32 amino acid deletion.

thumbnail image

Figure 4. Morpholinos (MOs) directed against 25H12.1 (pinhead) splice sites cause aberrant splicing and an increase in detectable pinhead mRNA. A: Diagram of part of the 25H12.1 gene in X. tropicalis with exonic sequences represented as boxes and intronic sections as black lines. I: The target sites of 25H12.1moA and 25H12.1moB at the boundary of a 179-bp intron and a 96-bp exon are shown as moA and moB. Primers f2 and r1 used for reverse transcriptase-polymerase chain reaction (RT-PCR) are depicted as arrows. II: Wild-type cDNA yields a 187-bp RT-PCR product. III,IV: Aberrantly spliced cDNAs yield 366-bp and 91-bp RT-PCR products with f2 and r1 due to inclusion of the 179-bp intron or skipping of the 96-bp exon. The gel shows RT-PCR (primers f2 and r1) of 25H12.1 cDNA from uninjected (WT, wild-type) neurula embryos or embryos injected at one-cell stage with 10 ng of 25H12.moA (moA) or 25H12.1moB (moB). Amplified DNA fragments are labelled (I–IV) according to the diagram shown above. Both MOs induce abnormal splicing. M, marker lane; C, no cDNA control; G, Xenopus tropicalis genomic DNA; P, plasmid DNA from TGas046h20 that includes insertion of the 179-bp intron. B:X. tropicalis embryos uninjected (a) or injected at the one-cell stage with 10 ng of 25H12.1moB (b,c) and hybridised with an 25H12.1 antisense RNA probe. As well as the pinhead phenotype, injected embryos show increased amounts of 25H12.1 mRNA in anterior regions (arrowheads in b) and around the tail bud (arrowhead in c).

Download figure to PowerPoint

Pinhead is a novel gene.

The entire pinhead cDNA was determined by sequencing 4 ESTs and 5′-RACE (Supplementary Fig. 2) and verified by Northern blotting (not shown). No homologous expressed sequences for species other than X. laevis and X. tropicalis were found by BLAST searching. However, TBLASTX searching with the pinhead open reading frame (ORF) revealed homologous genomic sequences in tetraodontoid fish Fugu rubripes (scaffold 33, 415,000 to 418,000 nt). Exon prediction programme genescan predicts a 7-exon gene corresponding to Fugu pinhead coding sequence (http://fugu.hgmp.mrc.ac.uk/). Interestingly, Fugu pinhead is located immediately upstream of the Fugu orthologue of the gene for antidorsalising morphogenetic protein (admp), a TGF-β homologue expressed in the Spemann organiser (Moos et al., 1995). We do not know whether the close proximity of dorsalising (pinhead) and antidorsalising (admp) genes has any functional significance. A cDNA clone similar to pinhead is also found in the invertebrate Ciona intestinalis (accession no. AK113761) and a short segment of genomic sequence corresponding to part of pinhead is also found in the zebrafish genomic sequence (ensembl contig no. NA4458.1), but homologous sequences have not been found in any mammalian genomic databases. Although the overall conservation is low, amino acid alignment of the Fugu, X. tropicalis, and Ciona predicted peptides (Fig. 5) shows three copies of a conserved motif (consensus V/I/L,DVGXC) of unknown functional significance.

thumbnail image

Figure 5. Comparison of pinhead amino acid sequence across species. Alignment of predicted sequences of pinhead proteins from Xenopus tropicalis, Fugu rubripes, and Ciona intestinalis. The amino acid sequence identity is 46.7% between X. tropicalis and Fugu and 26.8% between X. tropicalis and Ciona. Highly conserved regions include a repeated motif (V/I/L,D,V,G,X,C, underlined). [Color can be viewed at the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Overexpression of pinhead causes enlarged heads.

To gain additional evidence for the involvement of pinhead in head formation, a construct containing the ORF of the gene was overexpressed in X. laevis embryos (Fig. 6). In contrast to the microcephaly seen by inhibiting pinhead expression, overexpression induced marked macrocephaly. Posterior–ventral development was correspondingly attenuated, and embryos appeared slimmer in their posterior end. Pinhead-injected embryos also developed cystic loosening of the skin around the trunk (Fig. 6).

thumbnail image

Figure 6. Overexpression of 25H12.1 (pinhead) causes enlarged heads. Xenopus laevis embryos were injected in each blastomere after the first cleavage with 1 ng of 25H12.1 sense mRNA containing the coding section of the gene together with 100 ng/blastomere of nuclear β-gal mRNA as an injection tracer (blue stain). Pinhead-injected embryos show enhanced anterior development (d–f) compared with embryos injected with nuclear β-gal alone (a–c), as well as some epidermal blistering in the trunk.

Download figure to PowerPoint

Pinhead expression.

Expression of pinhead in wild-type X. tropicalis embryos is restricted to gastrula and very early neurula stages (Fig. 2B,g,h,i), consistent with the presence of pinhead sequences in gastrula and neurula but not egg EST databases. In gastrulating embryos, pinhead is expressed in an arc around the blastopore with a distinct gap corresponding to dorsal mesoderm, as shown by staining with chordin (Fig. 2B,h,i). In the neurula, it is expressed in the anterior neural plate, in two transverse stripes in the area, which is likely to correspond to the diencephalic or mesencephalic primordium and in the anterior neural ridge (Fig. 2B,g). We have not been able to detect expression in later stage embryos, but in X. laevis, there is some expression over the mid–hindbrain region (Gawantka et al., 1998). However, in MO-injected tadpole stage embryos, expression is detected in the anterior endoderm, in a stripe posterior to the eyes, and in the posterior ventral mesoderm (Fig. 4B). This expression is consistent with the expression of the gene in gastrula and neurula embryos and may be due to stabilisation of the message or up-regulation of the endogenous transcripts in MO-injected embryos.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Loss-of-function screens are a powerful method to uncover gene function. So far, gene inactivation screens have not been feasible in the frog, an otherwise powerful model organism (reviewed in Hirsch et al., 2002a). However, the emergence of large scale EST and genome sequencing projects, together with MO technology, provides an opportunity to perform large scale loss-of-function screens in the frog by inhibiting translation or splicing of genes with known sequence but unknown function. Here, we have conducted the first such MO screen, on a relatively small scale to access the feasibility of this approach for larger scale knock-down screens. We have selected 26 genes, which showed neural expression in X. laevis and for which we could find a X. tropicalis orthologue with a convincing start of translation or splice site. By using stringent criteria, we have found that MOs to 6 of these 26 genes (23%) had a reproducible phenotype. Of those, 4 were reproduced with a second MO, increasing our confidence that these phenotypes are specific for the inactivated genes (15%).

Why is the percentage of phenotypes not higher? The simplest explanation is that many MOs are not effective in blocking translation. In a screen, it is difficult to check the efficacy of each MO, because it is not feasible to generate antibodies to each targeted gene. However, there may also be biological reasons for the lack of phenotype in some cases. Some genes may be difficult to knock-down with MOs, due to inhibitory RNA secondary structure or if there is a large store of stable maternal protein. Some genes, such Id3/X and Hes, may be members of multigene families and have redundant functions. Furthermore, some phenotypes will require longer development times or additional assays for detection. However, it is interesting to note that the percentage of phenotypes that we have uncovered with this screen (23% with 1 MO or 15% with 2 MOs) is quite similar to the percentage found in a recent large scale RNAi screen in Caenorhabditis elegans (10–14%; Fraser et al., 2000; Kamath et al., 2003). Furthermore, in a modified gene trap screen in mice, only one third of 60 trapped genes were lethal. Therefore, a percentage of 15–23% phenotypes is not unrealistically low compared with other loss of function screens.

Although the efficacy of ATG-directed MOs is difficult to address in a screen, it is by contrast relatively easy to check the efficacy of MOs designed to inhibit splicing, as we have shown here for one of the targeted genes. We suggest that future screens should take advantage of the genomic sequence, currently under way, to design splice junction MOs whenever possible. Another issue to be considered in an MO is the issue of specificity. As with antibodies, the time and effort required for rescue experiments makes them prohibitory for a large scale screen. As an alternative, the issue of specificity can be addressed by designing a second, unrelated MO for the subset of genes that had a reproducible phenotype in the first round of screening. In this work, we have found that only two of six phenotypes did not reproduce with a second MO.

Despite the limitations, there are many advantages to a MO-based screen that make it a worthwhile approach. The gene sequences are known a priori, and once an interesting phenotype has been identified, the expression pattern can be easily determined. In fact, the gene sequences and/or the expression patterns can be used to select for inactivation candidate genes of particular specifications, as we have done here for neurally expressed genes that were previously identified in an expression screen (Gawantka et al., 1998). Furthermore, because the X. tropicalis cDNAs are already cloned in an expression vector, the phenotype of overexpressing the gene(s) of interest can be determined without delay. The time and cost that it takes to generate a loss-of-function phenotype is minimal compared with other methods, and, although one cannot establish mutant lines, individual MOs can be passed on to interested researchers for further study.

The usefulness of this method is demonstrated by the phenotypes uncovered in this pilot screen. Three of six phenotypes that we observed can be attributed to interference with morphogenetic movements. Although these phenotypes have some common elements (for example, all are shorter than average), the overall shape of the embryo is distinct in each case, demonstrating that a different aspect of morphogenesis has been disrupted in each case. Another phenotype involves a striking case of cell death, which starts at the anterior end of the ectoderm at the late neurula stage. The gene targeted in this case has been implicated recently as a positive regulator of apoptosis through overexpression in tissue culture (Jabbour et al., 2003). Our in vivo data indicate that it may also prevent apoptosis during neural tube formation in Xenopus development. Perhaps the levels of this gene product are critical, and either over- or underexpression can trigger apoptosis.

Of particular interest to neural development are two further phenotypes, one of which involves paralysis and the other microcephaly. The embryonic paralysis phenotype is caused by an MO to mago nashi, a very highly conserved gene which is a component of the exon–exon junction and in Drosophila is involved in RNA localisation (reviewed in Palacios, 2002). This phenotype could be due to incorrect localisation of mRNA pre- or postsynaptically at nerve endings. Of interest, mutations in another protein involved in mRNA biogenesis, survival motor neuron (SMN), result in a neuromuscular defect in humans called spinal muscular atrophy (SMA). In Drosophila, disruption of the homologue of this protein (smn) results in defects of the neuromuscular junction and paralysis (Chan et al., 2003).

Finally, the microcephalic phenotype is due to interference with a gene of previously unknown function, which we have termed pinhead. The expression of pinhead in the anterior neural plate is consistent with the developmental defects caused by the MOs. Furthermore, the effects of under- and overactivity are complementary. Therefore, our results indicate that pinhead is a novel gene that is essential for head development. Several reports have shown that antagonism of the wnt signalling pathway is essential for head development in vertebrates (e.g., Glinka et al., 1997, 1998; Kim et al., 2000; Pera et al., 2001; Houart et al., 2002; Richard-Parpaillon et al., 2002). Whether pinhead is a novel component of this pathway or a component of another pathway, is a subject of on-going investigation. The sequence of pinhead has three copies of a previously unknown motif and no obvious orthologues outside frogs, fish, and the invertebrate Ciona intestinalis. Perhaps pinhead is a rapidly evolving gene, much like Xnot, which is only 30% identical to its zebrafish homologue flh and also has not been found in mammals (Gomperts and E. A., unpublished observation). Loss-of-function screens are invaluable in pinpointing the functional significance of such sequence-divergent genes, which are likely to be missed in any sequence-homology–based work.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

MOs

All MOs (Supplementary Table 1) were purchased from Gene Tools LLC (USA) and designed according to the criteria described at www.gene-tools.com. Stock solutions (10 ng/nl in water) were vortexed and heated at 65°C for 5 min before dilution. Diluted MO solutions were heated to 65°C for 5 min and cooled to room temperature before injection.

Embryo Culture, Injection, and Fixation

X. laevis eggs were obtained, in vitro fertilised, and dejellied; embryos were cultured as described previously (Sive et al., 2000). X. tropicalis embryos were handled with a few modifications from standard X. laevis protocols, as described below (see also Khokha et al., 2002). X. tropicalis females and males were primed 3–4 days before ovulation with an injection of 15 U of pregnant mare serum gonadotropin (PMSG). In the morning of the ovulations, the females and males were injected with 150 U of human chorionic gonadotropin (Sigma). Three to 4 hr later, the females began to ovulate. At this time, a male was killed and the testes were removed with special care not to damage them in the process. The testes were transferred to a medium containing L-15 with 10% fetal calf serum, as described (Khokha et al., 2002). Females were manually stripped of eggs into a dry Petri dish, and the dry eggs were rubbed with a macerated testis. After 3–5 min, the eggs were flooded with 0.1× MMR (for MMR, see Sive et al., 2000). After 10–15 min, the eggs were dejellied in 2% cysteine in 0.1× MMR at pH 7.8. To keep the NaCl concentration relatively low, we used free-base cysteine rather than cysteine:HCl powder. After the eggs begin to pack in the cysteine solution, they were thoroughly washed in 0.1× MMR. X. tropicalis embryos were kept in Petri dishes covered with 1% agarose in 0.01× MMR at all times up to hatching, at a temperature of 22–24°C. The embryos were injected in 0.1× MMR with 1% Ficoll primarily at the one cell stage.

Stages of development were assessed according to Nieuwkoop and Faber (1967). Preliminary titration experiments indicated that 10 ng was the optimal MO dose; therefore, for the MO screen, 10 ng of test MO together with 0.3 ng of fluorescein isothiocyanate (FITC) -labelled control MO (Gene Tools standard, 5′cctcttacctcagttacaatttata3′) were injected in 2 nl into ∼50 uncleaved, fertilised eggs. Control embryos were either uninjected or injected with 10 ng of the FITC-labelled control standard. This MO has no known target in Xenopus and does not give rise to any apparent phenotype other than mild developmental delay. Embryos were fixed in MEMFA (Harland, 1991) for 2 hr at room temperature.

RACE and RT-PCR

For 5′-RACE for the 25H12.1 gene, the SMART system (Clontech) was used. Briefly, first-strand synthesis was performed by using Powerscript reverse transcriptase (Clontech) in the presence of the SMART primer and 25H12.1 gene-specific reverse primer r1 (5′ccacaatgagacctgcacaagcc3′). After removal of primers using QIA (Qiagen) Nested, touchdown PCR (68°C, for 10 cycles, then 65°C for 25 cycles) was then performed by using gene-specific reverse primer r4 (5′cacttactggactgccagaaatatcttgtc3′) and TS2 (5′aagcagtggtatcaacgcagagt 3′) with Advantage 2 (Clontech) enzyme with TS2 added after the first 10 cycles. A single RACE-specific product of ∼400 bp containing the 179-bp intron and 17 bp of additional 5′ sequence was obtained.

For RT-PCR, single embryos were lysed in 200 μl of TRIzol (Invitrogen) and RNA was isolated according to the manufacturer's instructions. Samples were treated with RQ1 DNase (Promega). First-strand cDNA synthesis was performed by standard procedures using AMV reverse transcriptase (Roche) and random hexamers. PCR amplification was achieved using the forward primer sequence f2 (5′atctctgtagcatgttcagctcac 3′) and reverse primer r1 (5′ccacaatgagacctgcacaagcc 3′). PCR products were resolved on 2% agarose gel, excised, and purified for sequencing using QIAquick gel extraction kit (Qiagen).

Whole-Mount In Situ Hybridisation

In situ hybridisation was performed as previously described (Harland, 1991). Antisense RNA probes were transcribed with incorporation of digoxigenin-11-UTP (Harland, 1991) from linearised plasmid templates. All X. tropicalis ESTs are cloned in pCS107 and RNA probes can be made with T7 polymerase (http://www.hgmp.mrc.ac.uk/geneservice/reagents/products/descriptions/XtropEST.shtml). Plasmids were linearised with ClaI, BamHI, or EcoRI at the 5′ end of the insert dependent upon absence of these sites within the inserts. The substrate for the chromogenic reaction was either 5-bromo-6-chloro-3-indoyl-phosphate (Magenta phos: Magenta colour) or 4-nitro blue tetrazolium chloride plus 5-bromo-4-chloro-3-indoyl-phosphate (NBT/BCIP: blue colour). After staining, embryos were washed in methanol and stored in MEMFA (Sive et al., 2000).

RNA Injection for Overexpression of 25H12.1

For overexpression, the ORF of 25H12.1 was PCR cloned into the StuI site of pCS2 expression vector, to remove the long 3′-untranslated region. This construct was linearised with NotI and transcribed by using the SP6 mMessage mMachine (Ambion). RNA was also produced by a NotI linearised nuclear β-galactosidase (Sive et al., 2000) to use as a lineage tracer. RNAs were injected into X. laevis embryos at the two-cell stage with a range of concentrations of 25H12.1 RNA (250 pg–1 ng/blastomere) with 100 pg/blastomere B-gal RNA. β-Gal activity was detected in developed tadpoles by X-gal staining as described in (Sive et al., 2000).

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

We thank Mike Gilchrist for clustering of X. tropicalis ESTs and generating our EST database; Julia Mason for help with in situ hybridization; Greg Elgar for navigating the Fugu rubripes genome; Ian Adams for help with chemiluminescent RNA detection; and Andy Chalmers, Sam Caruthers, Jana Voigt, and Bernhard Strauss for helpful discussion and technical advice. N.P. and E.A. are Wellcome Trust Senior Research Fellows.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

The Supplementary Material referred to in this article can be viewed at www.mrw.interscience.wiley.com/suppmat/1058–8388/suppmat

FilenameFormatSizeDescription
suppmat_figure1_289.gif1260KSupporting Information file suppmat_figure1_289.gif
suppmat_figure2_289.pdf17KSupporting Information file suppmat_figure2_289.pdf
suppmat_table1_289.pdf58KSupporting Information file suppmat_table1_289.pdf

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.