ABSTRACT
The Pax-6 genes of vertebrates and invertebrates encode transcription factors with both a paired domain and a homeodomain. They are expressed in the developing eye and in the central nervous system. Loss-of-function mutations in mammals and in flies result in a reduction or absence of eyes and targeted expression of the Drosophila and the mouse Pax-6 genes induces ectopic eye structures in Drosophila. These findings lead to the proposal that the morphogenesis of the different types of eyes is controlled by a Pax-6-dependent genetic pathway and that the various eye types are of monophyletic origin. We have isolated a Pax-6 homologous gene from the ascidian Phallusia mammillata, because ascidians occupy an important position in early chordate evolution. Furthermore, the Phallusia larva has a simple photosensitive ocellus. Phallusia Pax-6 shares extensive sequence identity and conserved genomic organization with the known Pax-6 genes of vertebrates and invertebrates. Expression of Phallusia Pax-6 is first detected at late gastrula stages in distinct regions of the developing neural plate. At the tailbud stage, it is expressed in the spinal cord and the brain vesicle, where the sensory organs (ocellus and otolith) form, suggesting an important function in their development. Ectopic expression of the ascidian Pax-6 gene in Drosophila leads to the induction of supernumerary eyes indicating a highly conserved gene regulatory function for Pax-6 genes.
INTRODUCTION
The urochordates are lower chordates and as such occupy an important position in the study and the understanding of the evolutionary history of higher chordates, including humans (Satoh, 1994). Ascidians represent the largest class of the subphylum urochordata. Their embryos exhibit a bilateral cleavage and a constant cell lineage (Conklin, 1905; Nishida, 1987). Within 24 hours after fertilization, a free swimming tadpole larva develops. These larvae have a motile tail containing a notochord, muscle cells and a dorsal nerve cord. The central nervous system consists of a brain with two sensory organs: a single photosensitive ocellus and a single gravity sense organ, the otolith (=statocyst). Ocellus and otolith arise from two precursor cells that are located symmetrically in the anterior half of the blastula and form an equivalence group (Nishida and Satoh, 1989). Either one of the precursor cells can differentiate into the ocellus or the otolith, depending on their final position in the brain vesicle. After one to two days, these larvae undergo metamorphosis to form juveniles. Upon metamorphosis, larval organs such as sensory organs, spinal cord, tail muscle and notochord regress and adult organs differentiate (Cloney, 1978).
The body plan of the tadpole-like larva largely resembles a primitive chordate, preserving the morphological characteristics of the ancestor from which the vertebrates may have arisen (reviewed in Crowther and Whittacker, 1992). Thus, urochordates occupy a central position in most theories of chordate evolution (Satoh and Jeffery, 1995, and references therein). Accordingly, the development of the ascidian tadpole may provide clues concerning the origin and evolution of chordates. We would like to determine whether the molecular mechanisms governing brain and sensory organ development in vertebrates are also utilized for the development of brain and sensory organs of ascidians. Because of the crucial function of Pax-6 in eye and nervous system development in vertebrates (reviewed in Callaerts et al., 1996), we characterized the Pax6 gene of the ascidian Phallusia mammillata to identify a relationship between the simple ocellus of urochordates and the complex image forming eye of vertebrates.
The sequence and genomic organization of the Pax-6 homologs of vertebrates and invertebrates analyzed to date are highly conserved. The murine and the human Pax-6 proteins are identical over the entire length of 422 amino acids (Ton et al., 1991; Walther and Gruss, 1991) and the zebrafish Pax-6 is 97% identical at the amino acid level to the mammalian Pax6 homologs (Krauss et al., 1991; Püschel et al., 1992). Pax-6 genes have also been identified in other vertebrates including quail (Martin et al., 1992), chicken (Goulding et al., 1993), rat (Matsuo et al., 1993) and axolotl (Epstein et al., 1994a). The first invertebrate Pax-6 homolog to be isolated is from Drosophila melanogaster and is encoded by eyeless (ey), which shares 94% sequence identity in the paired domain and 90% in the homeodomain with the mammalian Pax-6 genes (Quiring et al., 1994). Subsequently, Pax-6 homologs have been identified in other invertebrate species including the sea urchin Paracentrotus lividus (Czerny and Busslinger, 1995), the cephalopod Loligo opalescens (Tomarev, S. I., Callaerts, P., Kos, L., Zinovieva, R., Halder, G., Gehring, W. and Piatigorsky, J.,unpublished data), the nemertean Lineus sanguineus (Loosli et al., 1996) and the nematode Caenorhabditis elegans (Chisholm and Horvitz, 1995; Zhang and Emmons, 1995). The human, mouse and Drosophila Pax-6 genes are expressed in a comparable pattern in the developing central nervous system and the eye, and their respective loss-of-function mutations, Aniridia, Small eye (Sey) and ey, lead to strong eye defects or the complete loss of eyes (Bridges, 1935; Hill et al., 1991; Hoge, 1915; Nelson et al., 1984; Ton et al., 1991). Ectopic expression of Drosophila or mouse Pax-6 in imaginal discs of Drosophila produces supernumerary eye structures, indicating that Pax-6 acts as a developmental switch of Drosophila eye development in the context of the imaginal discs (Halder et al., 1995a,b).
Here we describe the isolation of a Pax-6 gene of the ascidian Phallusia mammillata (PPax-6). The extent of the amino acid sequence identity within the paired domain and the homeodomain and the conserved genomic organization suggest that PPax-6 is orthologous to the known vertebrate and invertebrate Pax-6 genes. In addition, the expression pattern of PPax-6 indicates a function during the development of the central nervous system and of the ocellus and otolith. Like the Drosophila and mouse Pax-6 genes, ectopic expression of PPax-6 in Drosophila imaginal discs induces supernumerary eye structures, suggesting conserved biochemical activities and gene regulatory functions for PPax-6 in Phallusia mammillata as compared to that of the Pax-6 genes in Drosophila and vertebrates.
MATERIALS AND METHODS
General methods
DNA manipulations were performed according to standard techniques (Sambrook et al., 1989). Genomic DNA from Phallusia mammillata sperm was prepared by adding 3 ml homogenization buffer (7 M Urea, 2% SDS, 50 mM Tris pH7.5, 10 mM EDTA, 0.35 M NaCl) and 3 ml phenol-chloroform to 1 ml of frozen sperm (− 70°C). This mix was incubated at room temperature for 30 minutes with gentle rotation and centrifuged for 10 minutes at 20°C at 12000 g, followed by phenol-chloroform extraction of the DNA-containing aquaeous phase. Poly(A)+ RNA was isolated (Oligotex-dT Kit, Qiagen) from neurula and tailbud stage embryos of P. mammillata and then two cDNA libraries constructed in the λ ZAP vector (Stratagene). Libraries were screened with random primed 32P-labeled probes (Megaprime Amersham) by hybridization in 35% deionized formamide, 5× SSC, 5× Denhardt’s, 0.5% SDS, 100 μg/ml denatured Salmon sperm DNA at 42°C for 20 hours. Afterwards, the filters were washed in 2× SSC, 0.5% SDS for 5 minutes at RT, in 2× SSC, 0.1% SDS for 15 minutes at RT and twice in 1× SSC, 0,5% SDS for 30 minutes at 60°C.
Amplification by Polymerase Chain Reaction (PCR)
Paired box fragments were PCR-amplified from genomic DNA of Phallusia mammillata, with two different sets of degenerate PCR primers essentially as described by Loosli et al. (1996). A first set of primers was directed against two highly conserved regions found in all paired domains corresponding to amino acid sequences GCVSK and WEIRD. An additional sense primer corresponding to amino acids YYETG was also employed, to circumvent the problem of a potential intron present at a conserved splice site between codons of R56 and Y57 of the paired box. This primer contains an XhoI site at the 5′ end and has the following sequence: CAG CTC GAG (A/C/G/T) TA (T/C) GA (A/G) AC (A/C/G/T) GG. The PCR reaction mix contained 0.5-1 μg P. mammillata genomic DNA, 0.1 μM (each) primer, 2 mM (each) dNTP, 1 unit of Taq polymerase and 10 μl of 10× PCR buffer in 100 μl. PCR reactions with the first set of primers led to the isolation of two different paired box fragments from the genome of P. mammillata, both members of the Pax gene class III constituted by Pax-2, Pax-5 and Pax-8 (Walther et al., 1991). PCR amplification with the YYETG 5′ primer resulted in the isolation of two additional paired box fragments, one homologous to Pax-1, the other to Pax-6 (PCR-PPax-6).
For reverse transcription PCR (RT-PCR), 1 μg of total RNA or 100 ng of Poly(A)+ RNA was reverse transcribed as reported by Wittbrodt and Rosa (1994). For the PCR reaction, gene-specific primers were designed: a sense primer corresponding to amino acids GHSGVNQLG of the paired domain (GGC TCT AGA GGG CAT AGC GGA GTA AAC CAA CTT GG) containing an XbaI site and an antisense primer corresponding to amino acids NRRAKWRRE of the homeodomain (CCG CTC GAG CCC GTC TCC ATT TTG CTC GAC GAT TCG) containing an XhoI site. Cycling conditions were 1 minute 94°C; 30 cycles of 30 seconds 94°C, 1 minute 62°C, 2 minutes 72°C; and 3 minutes 72°C.
To confirm the genomic structure, PCR was performed on genomic DNA with the following gene-specific primers: the sense primer 5′ of the first exon (GGC TCT AGA CGA CAT AAT TTC CAT GTC TGG G) containing an XbaI site and an antisense primer corresponding to the amino acids GGSKPRVA of the paired domain (CCG CTC GAG GCT ACA CGA GGT TTG CTT CCA CC) containing an XhoI site, the two primers used for RT-PCR, the sense primer corresponding to the amino acids NRRAKWRRE of the homeodomain (GGC AAG CTT CGA ATC GTC GAG CAA AAT GGA GAC GG) containing a HindIII site, and an antisense primer 3′ of the last exon (CCG CTC GAG GCG TAA AAC ATT TCG TCT ACA TGT CGG) containing an XhoI site. Cycling conditions were the same as for RT-PCR. PCR fragments were cloned into pBluescript (Stratagene), and nucleotide sequences were determined in part and further analyzed by restriction mapping.
Whole-mount in situ hybridization
Adult fertile Phallusia mammillata were obtained from the Laboratoire Arago, Banyuls-sur-Mer, France. In vitro fertilizations were performed at 18°C in sterile filtered sea water (SFSW) provided by the Zoological Gardens of Basel. One hour before reaching the desired developmental stage, embryos were dechorionated by incubating them in 2 mg/ml Serva 2× crystallized trypsin, 10 mM Hepes, pH 8.0 in SFSW for 60 minutes at room temperature and then washed twice in SFSW. Embryos for RNA isolation were frozen at −70°C. For in situ hybridization, embryos were fixed at 4°C overnight in 5% formaldehyde, 0.42 M NaCl, 2 mM MgSO4, 2 mM EGTA, 0.1 M Hepes, pH 7.4 and then dehydrated stepwise in 30%, 50%, 70%, 100% methanol. The embryos can be stored at −25°C. After rehydration of the embryos in 50%, 30%, 0% methanol, in situ hybridization was performed as described in Wada et al. (1995). Embryos were mounted in Canada Balsam. Full-length PPax-6 cDNA and genomic clones were used to generate DIG-labeled sense and antisense RNA probes (Boehringer Mannheim). The alkaline phosphatase detection reaction was allowed to proceed for 24 to 48 hours.
Cytokinesis was arrested by the addition of 1 μg/ml Cytochalasin D (Sigma) to blastula to early gastrula stage embryos (4 hours 45 minutes to 5 hours 30 minutes at 18°C), which then were allowed to develop for the amount of time, corresponding to the age of untreated embryos shortly before hatching. Dechorionization and fixation as above.
Induction of ectopic eye structures in Drosophila
Ectopic expression of PPax-6 in Drosophila melanogaster was performed as described by Halder et al. (1995a) with the GAL4 system (Brand and Perrimon, 1993). PPax-6 full-length cDNA was inserted as an XhoI-XbaI fragment into the pUAST vector. A total of 12 independent pUAST-PPax-6 transformant lines were crossed to the dppblink GAL4 line (kindly provided by M. Hoffman). Fine structure analysis by scanning electron microscopy and sectioning was performed according to Halder et al. (1995a). Imaginal discs were stained with anti-ELAV antibodies (Robinow and White, 1991) and a fluorescein-conjugated secondary anti-rat antibody (Cappel).
RESULTS
Isolation of PPax-6
In order to isolate a Pax-6 homolog from the ascidian Phallusia mammillata, we used a low-stringency PCR approach using degenerated primers directed against two regions highly conserved in all known paired domains corresponding to the amino acid sequences YYETG and WEIRD. Among others, we isolated a clone designated PCR-PPax-6, which is 154 bp long and contains an uninterrupted open reading frame showing 89% sequence identity at the amino acid level to the corresponding part of the paired domain of the human and mouse Pax-6 genes (see Methods, Fig. 1C). Screening of a tailbud stage λ ZAP cDNA library with clone PCR-PPax-6 resulted in the isolation of three clones (gPPax-6-1, gPPax-62 and gPPax-6-3), which not only contained part of the paired box and part of the homeobox but also contained untranslated genomic sequences. To isolate an uninterrupted cDNA containing the paired box and the homeobox, total RNA isolated from neurula stage Phallusia mammillata embryos was amplified by RT-PCR using two gene-specific primers (see Methods) A single PCR product of 778 bp (clone RT1, Fig. 1C) was amplified, and found to contain complete paired box and homeobox sequences. With the 778 bp PCR probe, eight cDNA clones containing both boxes were isolated from a newly constructed neurula stage λ ZAP cDNA library. The sequences of the four largest clones (PPax-6-3, PPax-6-7, PPax-6-10 and PPax-6-11) were analyzed. Clone PPax-6-7 is 1673 bp long (Fig. 1A,C) with an open reading frame containing a putative translation initiation site at position 97 and terminating at position 1488, a polyadenylation signal followed by a Poly(A)+ tail and two in frame stopcodons preceding the translation initiation site. The four clones show variation with respect to the site of polyadenylation. PPax-6-7 and PPax-6-3 are polyadenylated 18 bp downstream from the polyadenylation signal (Fig. 1A, open arrowhead); however, PPax-6-10 and PPax-6-11 are polyadenylated at a distance of 82 bp. PPax-67 encodes a protein of 464 amino acids containing a N-terminal paired domain, a linker region of 65 amino acids and a paired type homeodomain. The paired domain shows 87% and the homeodomain 95% amino acid identity to the respective domains of the human (Aniridia) and mouse (Small eye) Pax6 proteins. The C-terminal region comprises 170 amino acids and is rich in proline (9%), serine (16%) and threonine (9%). Thus, the sequence comparison demonstrates that the cloned Phallusia mammillata gene is homologous to the known Pax6 genes of vertebrates and invertebrates.
Molecular characterization of PPax-6. (A) Nucleotide and deduced amino acid sequence of the PPax-6 cDNA. Boxed amino acid sequences correspond to the paired domain (light gray) and the homeodomain (dark gray). Splice sites are indicated by arrowheads. The open arrowhead indicates the polyadenlation site in clone PPax-6-7. In frame stop codons and the putative polyadenylation signal are underlined. (B) Genomic organization and restriction map of the PPax-6 locus. The paired box is indicated in light gray and the homeobox in dark gray. Genomic PCR clones gPCR1, gPCR2 and gPCR3 are shown. E, EcoRI; Ea, EagI; N, NheI; P, PstI; S, SacI; X, XbaI. (C) Schematic structure of the PPax-6 cDNA. The open reading frame is boxed and the paired and homeobox are indicated as before. The PCR clone PCR-PPax-6 and the RT-PCR clone RT1 as well as the two cDNA clones PPax-6-7 and PPax-6-11 are shown underneath.
Molecular characterization of PPax-6. (A) Nucleotide and deduced amino acid sequence of the PPax-6 cDNA. Boxed amino acid sequences correspond to the paired domain (light gray) and the homeodomain (dark gray). Splice sites are indicated by arrowheads. The open arrowhead indicates the polyadenlation site in clone PPax-6-7. In frame stop codons and the putative polyadenylation signal are underlined. (B) Genomic organization and restriction map of the PPax-6 locus. The paired box is indicated in light gray and the homeobox in dark gray. Genomic PCR clones gPCR1, gPCR2 and gPCR3 are shown. E, EcoRI; Ea, EagI; N, NheI; P, PstI; S, SacI; X, XbaI. (C) Schematic structure of the PPax-6 cDNA. The open reading frame is boxed and the paired and homeobox are indicated as before. The PCR clone PCR-PPax-6 and the RT-PCR clone RT1 as well as the two cDNA clones PPax-6-7 and PPax-6-11 are shown underneath.
Determination of the genomic structure of PPax-6
The genomic organization of PPax-6 was determined by sequencing clones gPPax-6-1, gPPax-6-2 and gPPax-6-3 and defining exon-intron boundaries by comparison with the PPax6 cDNA sequence. The splice junctions were confirmed by PCR amplification of three overlapping fragments (gPCR1, gPCR2 and gPCR3; Fig. 1B) from genomic DNA using specific primers (see Methods). Analysis of terminal sequences of the PCR fragments and restriction analysis showed that clones gPPax-6-1 and gPPax-6-2 correspond to the PPax-6 genomic sequence. The isolated transcription unit spans 4 kb and contains 10 exons. There are three splice sites in the paired box, one in the linker region, two in the homeobox and another three in the C-terminal coding region (Fig. 1A,B). The introns of the gene are remarkably short, most just 60-120 bp in length. There are only two larger introns with lengths of 577 and 1214 bp.
Many of the splice sites in PPax-6 are conserved in other Pax-6 genes. The first splice site in codon 1 of the paired box is common to the Pax-6 genes of vertebrates, Drosophila and C. elegans (Fig. 2A). This splice site is also found in the vertebrate Pax-4, Pax-5 and Pax-8 genes. Pax-2 has a splice site in the first codon of the paired box as well, but it is shifted by one nucleotide (Walther et al., 1991) (Fig. 2A). The second splice site in the paired box in codon 56 is only conserved in the Drosophila Pax-6 gene ey, but absent in all other known Pax-6 genes. A splice site at identical position is also found in the paired box of vertebrate Pax-2, Pax-5 and Pax-8 genes. The third splice site in the paired box between codons 116 and 117 is present in all known Pax-6 genes of which the genomic structure has been determined. In Pax-3 and Pax-7, this splice site is shifted by one nucleotide and in Pax-5 and Pax-8 by 17 nucleotides.
Comparison of the amino acid sequence and the genomic organization of the paired domain (A), the homeodomain (B) and a conserved motif in the C terminus (C) of the ascidian Phallusia mammillata PPax-6 protein with Pax-6 proteins of the vertebrates Homo sapiens and Mus musculus (Glaser et al., 1992; Ton et al., 1991; Walther and Gruss, 1991), the sea urchin Paracentrotus lividus (Czerny and Busslinger, 1995), the cephalopod Loligo opalescens (Tomarev, S. I., Callaerts, P., Kos, L., Zinovieva, R., Halder, G., Gehring, W. and Piatigorsky, J., unpublished data), the fly Drosophila melanogaster (Quiring et al., 1994), the nemertean Lineus sanguineus (Loosli et al., 1996), and the nematode Caenorhabditis elegans (Chisholm and Horvitz, 1995). Dots represent identical amino acids, dashes represent introduced gaps. For the vertebrate Pax-6 proteins only the identical sequence of human (homo) and mouse (mus) are shown because all other vertebrate sequences are very similar or identical. Filled arrowheads behind the species name and in the sequence indicate the corresponding Pax-6 genes for which the genomic organization has been determined and where the splice sites are located. (A) Sequence comparison for the paired domain (light gray) and the conserved motif found within the linker region. The open arrowhead in the vertebrate sequence shows the location of an alternative splice site where a 42 bp fragment is inserted into the paired box. Numbers behind the sequences give percent identity compared to Phallusia (first row) and to vertebrates (second row). The paired domain of PPax-6 differs in ten positions (M11, I22, F29, N32, A55, Q79, N82, M86, N105, A111 and N128) from all other known Pax-6 proteins with only one of these substitutions (M11) representing a conservative exchange. (B) Sequence comparison of the homeodomain (dark gray) and motifs flanking the domain. There are amino acid substitutions differing from all other known Pax-6 proteins in three positions (S9, V13, S36) and two of them represent conservative amino acid exchanges (S9, V13). Percent identity as in (A). (C) Sequence comparison of the conserved motif found at the C terminus of the Pax-6 protein in vertebrates, Paracentrotus and Loligo. Asterisks represent the stop codon.
Comparison of the amino acid sequence and the genomic organization of the paired domain (A), the homeodomain (B) and a conserved motif in the C terminus (C) of the ascidian Phallusia mammillata PPax-6 protein with Pax-6 proteins of the vertebrates Homo sapiens and Mus musculus (Glaser et al., 1992; Ton et al., 1991; Walther and Gruss, 1991), the sea urchin Paracentrotus lividus (Czerny and Busslinger, 1995), the cephalopod Loligo opalescens (Tomarev, S. I., Callaerts, P., Kos, L., Zinovieva, R., Halder, G., Gehring, W. and Piatigorsky, J., unpublished data), the fly Drosophila melanogaster (Quiring et al., 1994), the nemertean Lineus sanguineus (Loosli et al., 1996), and the nematode Caenorhabditis elegans (Chisholm and Horvitz, 1995). Dots represent identical amino acids, dashes represent introduced gaps. For the vertebrate Pax-6 proteins only the identical sequence of human (homo) and mouse (mus) are shown because all other vertebrate sequences are very similar or identical. Filled arrowheads behind the species name and in the sequence indicate the corresponding Pax-6 genes for which the genomic organization has been determined and where the splice sites are located. (A) Sequence comparison for the paired domain (light gray) and the conserved motif found within the linker region. The open arrowhead in the vertebrate sequence shows the location of an alternative splice site where a 42 bp fragment is inserted into the paired box. Numbers behind the sequences give percent identity compared to Phallusia (first row) and to vertebrates (second row). The paired domain of PPax-6 differs in ten positions (M11, I22, F29, N32, A55, Q79, N82, M86, N105, A111 and N128) from all other known Pax-6 proteins with only one of these substitutions (M11) representing a conservative exchange. (B) Sequence comparison of the homeodomain (dark gray) and motifs flanking the domain. There are amino acid substitutions differing from all other known Pax-6 proteins in three positions (S9, V13, S36) and two of them represent conservative amino acid exchanges (S9, V13). Percent identity as in (A). (C) Sequence comparison of the conserved motif found at the C terminus of the Pax-6 protein in vertebrates, Paracentrotus and Loligo. Asterisks represent the stop codon.
The first splice site of the homeobox in codon 19 is found in all analyzed Pax-6 genes whereas the second splice site between codon 46 and 47 of the homeobox is present in all cases except Drosophila ey (Fig. 2B).
PPax-6 expression
Embryos at different developmental stages from the unfertilized egg up to hatched larvae were examined for PPax-6 expression by whole-mount in situ hybridization. The first signal was detected at late gastrula stages, when weak PPax-6 expression is detected in the animal hemisphere in two bilaterally arranged domains of four cells at the anterior lip of the gastrocoel (Fig. 3A,B). According to the description of the development of the central nervous system of the ascidian Ciona intestinalis by Nicol and Meinertzhagen (1988), this region forms the neural plate which consists at this stage of 20 to 40 cells. The symmetrically located cells (Fig. 3A,B) are the brain lineage cells including the two cells (a8.25 line) forming an equivalence group that will give rise to the sensory organs ocellus and otolith, respectively. In between the two groups of four cells there are two (Fig. 3B) to four (Fig. 3A) rows of unlabeled cells. This indicates that the a8.17 line expresses slightly later than the adjacent a8.25 line. In early neurula stage, the expression domain elongates to six cells along the neural folds on each side of the embryo (Fig. 3C,D side view). Little later two spots consisting of two cells each appear posteriorly (Fig. 3E). In late neurula stage embryos (10 hours of development at 18°C) PPax-6 is strongly expressed in the neural tube. There is a region of weaker expression between the strongly expressing cells of the future brain and the two posterior expression domains. In early tailbud stages (Fig. 3GI), strong expression is maintained in the cells forming the sensory vesicle and the spinal cord. Immediately behind the future sensory vesicle remains a region where no or very weak expression is detectable. In the mid tailbud stage (Fig. 3J), melanization of the otolith starts to appear and PPax-6 is expressed in the whole sensory vesicle (prosencephalon) and along the spinal cord (deuterencephalon) (CNS anatomy according to Katz, 1983). No PPax-6 expression is detected in the brain stem. At the stage shortly before hatching (18 hours of development at 18°C), the anterior otolith remains next to the ocellus before migrating to a more anterior and ventral location (Fig. 3K,L). At this stage, PPax-6 expression is no longer detectable in the spinal cord. Weak expression is detected in the whole sensory vesicle (Fig. 3K), where it becomes restricted anteriorly and ventrally (Fig. 3L). There is clear expression of PPax-6 around the pigment cells. Specificity of the staining pattern was confirmed by using a sense probe of PPax-6 that did not give any specific signal.
PPax-6 expression in developing Phallusia embryos by whole-mount in situ hybridization. (A-C; E-G) Dorsal views; (D, I-L) lateral views; (H) dorsolateral view; anterior is to the left. (A,B) Late gastrula/neural plate stage; (C-F) neurula stage; (G-I) early tailbud stage; (J) mid tailbud stage where onset of melanization of the otolith is visible; (K,L) late tailbud stage. Pigmented otolith and ocellus are clearly adjacent to one another just prior to migration of the otolith to a more anterior ventral position.
PPax-6 expression in developing Phallusia embryos by whole-mount in situ hybridization. (A-C; E-G) Dorsal views; (D, I-L) lateral views; (H) dorsolateral view; anterior is to the left. (A,B) Late gastrula/neural plate stage; (C-F) neurula stage; (G-I) early tailbud stage; (J) mid tailbud stage where onset of melanization of the otolith is visible; (K,L) late tailbud stage. Pigmented otolith and ocellus are clearly adjacent to one another just prior to migration of the otolith to a more anterior ventral position.
PPax-6 expression in cytokinesis-arrested embryos
Ascidian embryos have a mosaic development with a constant cell lineage. If their cytokinesis is blocked at early stages development by cytochalasin, nuclear division still continues and the arrested cells continue to differentiate and express of histospecific enzymes in the appropriate cell lineages (Whittaker, 1973). This type of cytokinesis arrest experiment and the detailed knowledge of cell lineage in ascidians allows cells expressing genes of interest to be identified. We examined PPax-6 expression in cleavage-arrested embryos to clarify the relationship between PPax-6-expressing cells and the cells developing into the pigment cells, of the ocellus and otolith, respectively. Cytokinesis was blocked at late blastula/early gastrula stages (4 hours 45 minutes to 5 hours 30 minutes of development at 18°C) (Fig. 4A), and embryos were fixed at a stage comparable to that of control embryos just before hatching. At this time, the multinucleated cells of cytochalasin-treated embryos are much larger than normal larval cells and obvious pigment granules have developed in the melanocyte lineage cells (Fig. 4B). PPax-6 expression is detected in most cases in two single cells immediately anterior to the pigmented cells (Fig. 4C). According to the cell lineage studies in Halocynthia roretzi by Nishida (1987), these two bilaterally located cells are of brain developmental fate. We sometimes detected diffuse expression in more than two single cells, but always anterior to the pigmented cells (data not shown).
Cytochalasin arrest of cytokinesis in Phallusia embryos and PPax-6 expression by whole-mount in situ hybridizations. (A) Blastula stage of a Phallusia embryo. At this and slightly later stages of development, cytochalasin D was added to block cytokinesis. (B) Cytokinesis-arrested embryos at the age comparable to that of control embryos shortly before hatching. The two pigment cells that would have given rise to the ocellus and the otolith, respectively, are clearly visible. At this stage, embryos were fixed after dechorionization for whole-mount in situ hybridization. (C) Whole-mount in situ hybridization on a cytokinesis-arrested Phallusia mammillata embryo as shown in B. The two bilaterally arranged PPax-6-expressing cells are immediately anterior to the two darkly pigmented cells.
Cytochalasin arrest of cytokinesis in Phallusia embryos and PPax-6 expression by whole-mount in situ hybridizations. (A) Blastula stage of a Phallusia embryo. At this and slightly later stages of development, cytochalasin D was added to block cytokinesis. (B) Cytokinesis-arrested embryos at the age comparable to that of control embryos shortly before hatching. The two pigment cells that would have given rise to the ocellus and the otolith, respectively, are clearly visible. At this stage, embryos were fixed after dechorionization for whole-mount in situ hybridization. (C) Whole-mount in situ hybridization on a cytokinesis-arrested Phallusia mammillata embryo as shown in B. The two bilaterally arranged PPax-6-expressing cells are immediately anterior to the two darkly pigmented cells.
Targeted expression of the ascidian PPax-6 cDNA in Drosophila melanogaster
Ectopic expression of the ey cDNA in various imaginal discs of Drosophila under the control of the GAL4-UAS system (Brand and Perrimon, 1993) leads to the induction of ectopic eye structures (Halder et al., 1995a). The mouse Pax-6 gene can substitute for the Drosophila ey gene in ectopic eye induction experiments in Drosophila. We tested whether the less conserved ascidian PPax-6 cDNA give similar results in this functional assay. Several UAS-PPax-6 transgenic lines were generated. Flies carrying this UAS-PPax-6 transgene where crossed with flies of the dppblink GAL4 driver line. The progeny of these crosses thus ectopically expresses PPax-6 in various imaginal discs. As a consequence, ectopic eye structures were induced on legs and on wings of adult flies.
The fine structure of the ectopic eyes was analyzed by scanning electron microscopy (Fig. 5A,B). Distinct ommatidia with lenses and interommatidial bristles were seen, although irregular spacing of bristles and fusion of facets was sometimes observed. Microscopic analysis of sections (Fig. 5C,D) showed that the ommatidia of the ectopic eyes consisted of all different cell types present in a normal compound eye. Cornea, pseudo-cone, cone cells, pigment cells and photoreceptors with rhabdomeres were distinguishable. Irregular spacing of ommatidial clusters often resulted, perhaps due to the abnormal three-dimensional organization and the relatively small size of the ectopic eyes. Neuronal differentiation of the photoreceptors in the various imaginal discs was analyzed by means of anti-ELAV antibody staining (Robinow and White, 1991). The differentiating ommatidial clusters have an organization similar to the ommatidial clusters in the normal eye imaginal disc (Fig. 5E,F). These data demonstrate that ey activity can be provided by PPax-6 and that it can induce the formation of complete and morphologically normal eyes by switching on the eye developmental pathway in imaginal discs.
Ectopic eye structures in Drosophila melanogaster induced by overexpression of the ascidian PPax-6 cDNA. (A,B) Scanning electron micrograph of an ectopic eye on the prothoracic leg. (A) Overview of the fly; (B) higher magnification of A. The ectopic eye contains hexagonal ommatidia and interommatidial bristles. (C,D) Micrograph of a section through an ectopic eye next to wing tissue. (C) Overview of the section. The wing tissue is visible to the right of the ectopic eye. (D) Higher magnification of the same section showing all cell types present in a normal compound eye. (E,F) Micrograph of a wing imaginal disc stained with an antibody against the neuronal marker ELAV and a secondary fluorescein-labeled antibody. (E) Differentiating photoreceptors are detected along the anteroposterior axis of the wing imaginal disc. (F) Higher magnification of E showing the photoreceptor clusters in the developing ectopic eye.
Ectopic eye structures in Drosophila melanogaster induced by overexpression of the ascidian PPax-6 cDNA. (A,B) Scanning electron micrograph of an ectopic eye on the prothoracic leg. (A) Overview of the fly; (B) higher magnification of A. The ectopic eye contains hexagonal ommatidia and interommatidial bristles. (C,D) Micrograph of a section through an ectopic eye next to wing tissue. (C) Overview of the section. The wing tissue is visible to the right of the ectopic eye. (D) Higher magnification of the same section showing all cell types present in a normal compound eye. (E,F) Micrograph of a wing imaginal disc stained with an antibody against the neuronal marker ELAV and a secondary fluorescein-labeled antibody. (E) Differentiating photoreceptors are detected along the anteroposterior axis of the wing imaginal disc. (F) Higher magnification of E showing the photoreceptor clusters in the developing ectopic eye.
DISCUSSION
Sequence conservation and genomic organization of PPax-6
Pax-6 of the ascidian Phallusia mammillata (PPax-6) codes for a protein containing two sequence-specific DNA-binding domains, a paired domain and a homeodomain, which show high sequence identity to the respective domains of the known vertebrate and invertebrate Pax-6 genes (Fig. 2A,B). The high sequence conservation and the highly conserved genomic organization indicates that the PPax-6 gene is orthologous to the vertebrate and invertebrate Pax-6 genes.
The paired domain of PPax-6 shows 87% amino acid identity to the respective domains in the human and mouse Pax-6 genes. This is in contrast to the paired domains of Drosophila, squid and sea urchin Pax-6, which are all more than 90% identical when compared to human and mouse Pax 6. In contrast to the PPax-6 paired domain, the homeodomain is closely related to the Pax-6 homeodomain of vertebrates. The homeodomain of PPax-6 shows 95% amino acid identity to the respective domain of vertebrate Pax-6 genes (Fig. 2B). High sequence similarity is also found in the amino acids flanking the homeodomain. The paired domain is separated from the paired-type homeodomain by a 65 amino acid linker region. This is, at present, the shortest linker region of all known Pax-6 genes. The linker regions of all other Pax-6 proteins, except Drosophila ey, vary in length between 72 and 92 amino acids (Callaerts et al., 1996; Czerny and Busslinger, 1995; Quiring et al., 1994). In the linker region, a short conserved motif with the consensus sequence MYDKLRMLNGQ (Fig. 2A) has been found in all known Pax-6 genes and is absent in all other Pax genes (Loosli et al., 1996). This conserved motif is not present in the linker region of PPax-6. Another conserved motif has been identified at the C terminus of the Pax-6 genes of vertebrates, sea urchin and squid (Tomarev, S. I., Callaerts, P., Kos, L., Zinovieva, R., Halder, G., Gehring, W. and Piatigorsky, J., unpublished data) (Fig. 2C). The ascidian Pax-6 gene lacks this motif as well. Despite the close evolutionary relationship of urochordates and vertebrates, the sequence of PPax-6 is more diverged in comparison with Pax-6 of other invertebrate species.
An alternatively spliced Pax-6 transcript has been identified in vertebrates (Glaser et al., 1992; Goulding et al., 1993; Püschel et al., 1992; Walther and Gruss, 1991). These transcripts contain an additional exon of 42 bp giving rise to a 14 amino acid insertion between Q44 and V45 of the paired domain (Fig. 2A open arrowhead). A different function for the two forms is suggested by the different expression levels of the short and the extended form of Pax-6 in the eye and the brain of mouse (Epstein et al., 1994b; Richardson et al., 1995). A similar exon has been found neither in the PPax-6 gene nor in any invertebrate Pax-6 gene analyzed so far. Hence, this insertion seems to have occurred in evolution after the branching of the urochordate lineage from the cephalochordate and vertebrate lineages. It will be interesting to determine if Pax-6 in cephalochordates such as Amphioxus contains this additional exon or whether it is specific to the vertebrate lineage.
The fact that some of the exon-intron boundaries in the paired box of PPax-6 are shared with Pax-6 genes and others with Pax-2, Pax-5 and Pax-8 genes is consistent with the proposal that Pax-4 and Pax-6 represent an evolutionary sister group of Pax-2, Pax-5 and Pax-8 (Noll, 1993). Most likely the ancestral paired box had three introns at the positions of the splice sites of PPax-6, some of which were lost by different Pax-6 genes. Similar losses and shifting of splice sites may have occurred within the other Pax gene classes.
Overall the ascidian PPax-6 gene is one of the most diverged Pax-6 genes found to date, despite the phylogenetic position of the ascidians within the phylum of the chordates. This observation could reflect the speed by which larval development in ascidians seems to evolve (Satoh and Jeffery, 1995). However, the degree of conservation is remarkably high, when compared to the highly diverged Hox-cluster genes of various ascidian species (di Gregorio et al., 1995; Holland et al., 1994; Ruddle et al., 1994). This points to a strong selective pressure acting on PPax-6 and an essential and specialized developmental function of its gene product (Halder et al., 1995b; Noll, 1993).
Conserved expression pattern of PPax-6 and functional implications in sensory organ development
PPax-6 expression in the urochordate Phallusia mammillata is first detected in the lateral regions of the neural plate, which will form the central nervous system. Later, expression is seen in a broad domain in the entire prosencephalon (sensory vesicle) and in the spinal cord along the whole anteroposterior axis. This broad expression domain in the early to late tailbud stage (Fig. 3I-K) encompasses all of the developing structures of the ocellus, which in the larva consist of a pigmented cell, several lens and photoreceptor cells, which have not yet been studied at the ultrastructural level. However, in the closely related species Ciona intestinalis, the presence of one pigment cell, three lens cells and 15 to 20 photoreceptor cells has been demonstrated (Eakin and Kuda, 1971). In larvae shortly before hatching, PPax-6 becomes restricted to the anterior portion of the prosencephalon including the neurons associated with the pigment cells. This expression pattern is comparable to that of vertebrate Pax-6 where expression is seen in the presumptive prosencephalon and rhombencephalon, in the neural tube along the entire anteroposterior axis and in the developing eye where Pax-6 is expressed in all structures giving rise to the adult eye. In vertebrates, as in ascidians, the initially broad expression domains in the central nervous system become resolved during further development into more discrete areas (Gérard et al., 1995; Grindley et al., 1995; Krauss et al., 1991; Li et al., 1994; Püschel et al., 1992; Walther et al., 1991; reviewed in Callaerts et al., 1996).
The expression pattern of Pax-6 in Phallusia and in those invertebrates where Pax-6 expression has been investigated, are comparable as well. In Drosophila, ey is expressed throughout development in the central nervous system and in the developing eye prior to the onset of differentiation. In the squid Loligo opalescens, Pax-6 expression is observed in the region of the rudimentary eye primordia and in the developing eye as well as in the olfactory organ, arms and suckers (Tomarev, S. I., Callaerts, P., Kos, L., Zinovieva, R., Halder, G., Gehring, W. and Piatigorsky, J., unpublished data). In all species investigated so far, Pax-6 is expressed in the respective photoreceptive organs and in the CNS, suggesting a function in their development.
The observed PPax-6 expression in cleavage-arrested Phallusia embryos in the two cells immediately anterior to the pigment cells of the future otolith and ocellus, implies an involvement of PPax-6 in the development of these sensory organs. Although PPax-6 is expressed in early and late tailbud stages in the complete developing brain, including the pigment cells (Fig. 3I-K), no signal is detected in the pigment cells of cytokinesis-arrested embryos. One possible, albeit unlikely, explanation is that pigmentation interferes with the detection of the PPax-6 in situ hybridization signal. An alternative hypothesis, which we favour, is that PPax-6 is no longer expressed in the pigment cells of arrested embryos, which correspond to control embryos just prior to hatching. This situation would be comparable to the downregulation of Pax6 expression in the pigment epithelium of the retina in mice (Grindley et al., 1995).
In the ascidian Halocynthia roretzi, it has been shown that otolith and ocellus form an equivalence group where the dominant fate is the ocellus pathway and the decision to develop into either an ocellus or otolith occurs by the midtailbud stage (Nishida and Satoh, 1989), which is a time point after our treatment of the ascidian embryos with cytochalasin. The expression pattern of PPax-6 in two single cells immediately anterior to the pigmented cells supports this equivalence group idea.
In a functional assay for PPax-6 in Drosophila melanogaster, the ascidian gene induces ectopic eye structures (Fig. 5). This has the following implications for a functional conservation of the PPax-6 protein. First, because the PPax-6 protein sequence does not contain the conserved motifs in the linker region and the C terminus, these motifs are not required for ectopic eye induction in Drosophila. Second, because PPax-6 can induce ectopic eyes, in Drosophila, at least some of the target sequences of the EY protein appear to be recognized by the PPax-6 protein. It is possible that the primary function of PPax-6 in Drosophila is the induction of the resident Drosophila Pax-6 gene, which appears to be auto-regulated. But the high degree of sequence conservation in the two DNA-binding domains suggests that also downstream target genes may well be regulated. Therefore, it will be necessary to determine whether the target sequences of PPax6 and EY are indeed conserved and to identify the genes that they control. This in turn will further our understanding of the genetic interactions that are modulated during evolution giving rise to morphologically different organs.
ACKNOWLEDGEMENTS
We are grateful to Drs J. Marthy and A. Guille of the Laboratoire Arago, Banyuls-sur-Mer for their hospitality and for providing ascidian animals, and to Dr T. Jermann of the Zoological Gardens of Basel for supplying sea water. We thank P. Baumgartner for his excellent technical assistance. We are indepted to M. Affolter, M. Berry and J. Groppe for their helpful discussions and critical reading of the manuscript and to Erika Marquardt-Wenger for processing the manuscript. This work was supported by the Collen Foundation and Sandoz Foundation to P. C., the Janggen Pöhn Stiftung to G. H., the Swiss National Science Foundation and the Kantons of Basel.
REFERENCES
Note added in proof
The nucleotide sequence data reported in this paper will appear in the EMBL, GenBank and DDBJ Nucleotide Sequence Databases under the accession number YO9975.