- Email: [email protected]
The forkhead gene FH1 is involved in evolutionary modification of the ascidian tadpole larva
Mechanisms of Development 85 (1999) 49±58
The forkhead gene FH1 is involved in evolutionary modi®cation of the ascidian tadpole larva Catherine L. Olsen a,c, Jeanette E. Natzle a, William R. Jeffery a,b,c,* a
Graduate Program in Cell and Developmental Biology, Section of Molecular and Cellular Biology, University of California, Davis, CA 95616, USA b Department of Biology, University of Maryland, College Park, MD 20742, USA c Station Biologique, Roscoff, France Received 8 January 1999; received in revised form 8 March 1999; accepted 10 March 1999
Abstract The forkhead gene FH1 encodes a HNF-3b protein required for gastrulation and development of chordate features in the ascidian tadpole larva. Although most ascidian species develop via a tadpole larva, the conventional larva has regressed into an anural (tailless) larva in some species. Molgula oculata (the tailed species) exhibits a tadpole larva with chordate features (a dorsal neural sensory organ or otolith, a notochord, striated muscle cells, and a tail), whereas its sister species Molgula occulta (the tailless species) has evolved an anural larva, which has lost these features. Here we examine the role of FH1 in modifying the larval body plan in the tailless species. We also examine FH1 function in tailless species £ tailed species hybrids, in which the otolith, notochord, and tail are restored. The FH1 gene is expressed primarily in the presumptive endoderm and notochord cells during gastrulation, neurulation, and larval axis formation in both species and hybrids. In the tailless species, FH1 expression is down-regulated after neurulation in concert with arrested otolith, notochord, and tail development. The FH1 expression pattern characteristic of the tailed species is restored in hybrid embryos prior to the development of chordate larval features. Antisense oligodeoxynucleotides (ODNs) shown previously to disrupt FH1 function were used to compare the developmental roles of this gene in both species and hybrids. As described previously, antisense FH1 ODNs inhibited endoderm invagination during gastrulation, notochord extension, and larval tail formation in the tailed species. Antisense FH1 ODNs also affected gastrulation in the tailless species, although the effects were less severe than in the tailed species, and an anural larva was formed. In hybrid embryos, antisense FH1 ODNs blocked restoration of the otolith, notochord, and tail, reverting the larva back to the anural state. The results suggest that changes in FH1 expression are involved in re-organizing the tadpole larva during the evolution of anural development. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Forkhead genes; Chordate body plan; Ascidian development; Antisense inhibition; Evolution of larval development
1. Introduction Developmental processes are controlled by conserved regulatory genes (McGinnis and Krumlauf, 1992; Littlewood and Evan, 1995; Herr and Cleary, 1995; Jun and Desplan, 1996), but little is known about how these genes generate novel modes of development. Ascidians have served as a model system for studying the evolution of developmental mechanisms (Jeffery, 1997). Most ascidian species pass through a tadpole (or urodele) larval stage (Satoh, 1994). The tadpole larva exhibits a dorsal central nervous system, a notochord, and striated tail muscle cells, and thus is considered a prototype of the ancestral chordate (Satoh and Jeffery, 1995; Di Gregorio and Levine, 1998). * Corresponding author. Tel.: 11-301-405-5454; fax: 11-301-3149358. E-mail address: [email protected] (W.R. Jeffery)
However, a few ascidian species have evolved a tailless (or anural) larva (Berrill, 1931; Jeffery and Swalla, 1990). The anural larva has lost the typical chordate features, although their progenitors are present during embryogenesis (Whittaker, 1979; Swalla and Jeffery, 1990). Anural species have appeared multiple times from urodele ancestors in different ascidian lineages (Had®eld et al., 1995), indicating that reorganization of the chordate body plan has occurred repeatedly during ascidian evolution. The ascidians Molgula oculata (the tailed species), a urodele developer, and Molgula occulta (the tailless species), an anural developer, are capable of interspeci®c hybridization (Swalla and Jeffery, 1990; Jeffery and Swalla, 1992). When tailed species eggs are fertilized with tailless species sperm the resulting hybrid larvae are identical to those of the tailed species. In contrast, when tailless species eggs are fertilized with tailed species sperm, the otolith,
0925-4773/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0925-477 3(99)00061-1
Fig. 1. The nucleotide and deduced amino acid sequences of MoccFH1 and MocuFH1. In the MocuFH1 sequence (lower), dots represent identical nucleotides and letters represent amino acid substitutions. The amino acids differing between the two cDNA clones are listed side by side with the MoccFH1 amino acid listed ®rst (e.g. S/G). The putative winged helix DNA-binding domain is shaded, and the region used to design antisense and sense ODNs has a line drawn above it. The Genbank accession number for MoccFH1 is AF082992.
50 C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
51
Fig. 2. FH1 mRNA expression in embryos of the tailed species, tailless species, and hybrids. (A) Temporal expression of FH1 mRNA during development of the tailless species. Each lane contains RNA isolated from (1) unfertilized eggs, (2) fertilized eggs, (3) gastrulae, (4) neurulae, (5) 6 h embryos, (6) 8 h embryos, and (7) 10 h embryos of the tailless species. Top, MocuFH1 probe. The 2.3 kb FH1 transcript is indicated by the arrow. Bottom, 18S rRNA loading control. (B± J) Tailed species (B±D), tailless species (E±G), and hybrid (H±J) embryos subjected to in situ hybridization with an MocuFH1 antisense RNA probe. (B,C,E,F,H,I) Whole mounts. (D,G,J), sections of whole mounts. (B) A tailed species gastrula showing staining in presumptive notochord (N) cells at the anterior lip of the blastopore and endoderm (E) cells moving inside the blastopore. (C±D) Tailed species neurula (C) and early tailbud stage (D, 6 h) showing staining in the notochord (N) and endoderm (E) cells. (E) A tailless species gastrula showing staining in notochord (N) and endoderm (E) cells. (F,G) Tailless species neurula (G) and 6 h embryo (G) showing staining in notochord (N) and endoderm (E) cells. (H±J) Hybrid gastrula (H), neurula (I) and 6 h embryo (J) showing staining in notochord (N) and endoderm (E) cells. Embryos are shown with anterior pole at the top except in D, where it is shown to the left. Scale bar, 20 mm for each frame.
notochord, and tail are restored in the anural larva. These results imply that loss of function mutations in zygotic genes are responisble for modifying the tadpole larva in the tailless species. A subtractive screen involving the two species has been done to identify some of these genes (Swalla et al., 1993). The screen resulted in the isolation and characterization of the zinc ®nger gene Manx, the leucine zipper gene lynx, and the tyrosine kinase gene cymric, which are downregulated in the tailless species (Jeffery, 1997). In addition, the RNA helicase gene bobcat, which is also down-regulated in the tailless species, was identi®ed by its proximity to Manx in the tailed species genome (Swalla et al., 1999). Disruption of zygotic Manx and bobcat expression by anti-
sense oligodeoxynucleotides (ODNs) prevents restoration of chordate features in hybrids, showing that these genes are necessary for tadpole larval development (Swalla and Jeffery, 1996; Swalla et al., 1999). Genes involved in axis formation in other chordates could also play a role in the evolution of anural development. The HNF-3/forkhead genes are prime candidates for such a role because of their involvement in notochord development in vertebrate embryos (Sasaki and Hogan, 1994; Dirksen and Jamrich, 1992). The putative transcription factors encoded by these genes share the winged helix DNA-binding domain and are expressed in embryonic organizer regions (Kaufmann and Knochel, 1996). For example, HNF-3b is
52
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
expressed in the dorsal lip of the blastopore during Xenopus development and in the node region during mouse development. Later, HNF-3b is expressed in the notochord and ¯oor plate of the neural tube, where it functions in dorsoventral patterning the CNS, and in the endoderm, where it plays a role in gut differentiation. Mouse embryos homozygous for a mutated HNF-3b gene fail to form a notochord or gut and show defects in node, paraxial mesoderm, and neural tube organization (Ang and Rossant, 1994; Weinstein et al., 1994). We previously identi®ed FH1, an ascidian forkhead gene encoding a protein related to HNF-3b (Olsen and Jeffery, 1997). Disruption of FH1 expression by antisense ODNs showed that this gene is required for gastrulation and tadpole larval development in the tailed species. Here we describe the orthologous FH1 gene in the tailless species and show that expression is down-regulated during anural development and restored in hybrids. We also demonstrate that, although FH1 expression is unnecessary for posterior development of the anural larva, which has lost chordate features, it is required for restoration of chordate features in hybrids. These results suggest that changes in FH1 expression are involved in the re-organization of the tadpole larva during the evolution of anural development. 2. Results 2.1. Isolation of an FH1 gene in the tailless species We previously isolated and characterized MocuFH1, a cDNA clone encoding the FH1 gene in the tailed species (Olsen and Jeffery, 1997). To identify the orthologous gene in the tailless species, we screened a gastrula cDNA library with MocuFH1 and isolated MoccFH1 (Molgula occulta ForkHead 1). The 2.2 kb MoccFH1 cDNA exhibits a fulllength 1703 bp ORF, which is preceded by a 109 bp 5' UTR and followed by a 387 bp 3' UTR (Fig. 1). The 3' UTR contains a putative polyadenylation signal but no poly (A) tail is present. MoccFH1 and MocuFH1 are 94% similar at the nucleotide level, exhibit highly conserved 5' and 3' UTRs, and the deduced FH1 proteins are 96% similar (Fig. 1). Amino acid substitutions are distributed throughout the N- and C-terminal regions of the FH1 proteins, but their winged helix DNA-binding domains are identical. A `BLAST' search indicated that MoccFH1 is more closely related to MocuFH1 (Olsen and Jeffery, 1997) than to other ascidian HNF/forkhead genes ( Corbo et al., 1997; Shimauchi et al., 1997), suggesting that these cDNA clones encode orthologous FH1 genes in the tailless and tailed species. 2.2. Changes in FH1 expression during embryogenesis The accumulation of FH1 mRNA during embryogenesis was followed by northern and in situ hybridizations. In the tailed species, it was shown previously that the 2.3 kb FH1 mRNA appeared by the beginning of gastrulation, peaked
during neurulation, declined to the level seen at gastrulation by the mid-tailbud stage, and was reduced to low levels by the late tailbud stage (Olsen and Jeffery, 1997). The FH1 mRNA is also about 2.3 kb in the tailless species (Fig. 2A), consistent with the size of the MoccFH1 cDNA (Fig. 1). In general, the temporal pattern of FH1 accumulation was similar in the tailed and tailless species. However, two notable differences were observed in the tailless species (Fig. 2A). First, although FH1 transcripts ®rst appeared during gastrulation, there was no peak at the neurula stage. Instead, FH1 transcripts remained at about the same level from gastrulation through the 6 h stage, which is roughly equivalent to the early to mid tailbud stage in the tailed species. Second, FH1 transcripts appeared to decline earlier after neurulation in the tailless species. The in situ hybridization results are shown in Fig. 2B±J. These experiments were carried out using the same probe and identical hybridization conditions for each kind of embryo. In addition, the tailless species and hybrid embryos used in these experiments were derived from the same clutch of eggs. Therefore, the intensity of FH1 mRNA staining is directly comparable between the different kinds of embryos at the same stage of development. FH1 transcripts were detected primarily in the presumptive endoderm and notochord during gastrulation, neurulation, and tailbud formation in both species and hybrids (Fig. 2B±J). However, in the tailless species FH1 mRNA began to decrease in the endoderm cells at the neurula stage (Fig. 2C,F), and notochord and endoderm cells showed reduced transcript levels at 6 h of development (Fig. 2D,G). The staining intensity was variable between different clutches of tailless species embryos but in no case matched the higher levels observed in the tailed species. In hybrid embryos, FH1 staining during gastrulation was similar to that observed in the tailed and tailless species (Fig. 2B,E,H). However, hybrids showed enhanced FH1 transcript levels in endoderm and notochord cells relative to tailless species embryos beginning at the neurula stage (Fig.2F,I) and extending through 6 h of development (Fig. 2G,J). The proportion of hybrid embryos with enhanced FH1 expression was about 70%, the same as the percentage of hybrids that restored the otolith and tail in this experiment. The results show that (1) FH1 is expressed in the same embryonic tissues in the tailed species, tailless species, and hybrids, (2) the intensity and duration of intense FH1 expression is reduced in the tailless species and (3) the FH1 expression period characteristic of the tailed species is restored in hybrid embryos. 2.3. FH1 is not required for development of posterior features in the anural larva Previous studies showed that antisense ODNs reduce FH1 mRNA levels and prevent gastrulation and development of the tadpole larva (Olsen and Jeffery, 1997). Thus, we have used the antisense approach to determine whether FH1 is required for development of the anural larva. In these
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
53
Fig. 3. The effects of FH1 ODNs on gastrulation and larval development. (A,D,G,J) Untreated control embryos. (B,E,H,K) Sense ODN-2 treated embryos. (C,F,I,L,M±O) Antisense ODN-2 treated embryos. (A±F) Sagittal sections of tailed (A±C) and tailless (D±F) species mid-gastrulae. Untreated and sense ODNtreated gastrulae in (A,B,D,E) show normal morphology with notochord cells (N) involuted at the anterior lip and muscle cells (M) involuted at the posterior lip of the blastopore (B). (C,F) Antisense ODN-treated tailed (C) and tailless (F) gastrulae showing the mass of endoderm cells (EM) at the vegetal pole. Transverse sections of 11 h tailless species (G±I) and hybrid (J±L) embryos. (G,H,J,K) Untreated (G,J) and sense ODN-treated tailless species (G,H) and hybrid (J,K) embryos showing normal morphology, including an archenteron (Ar) surrounded by endoderm cells (E). Hybrid embryos also exhibit an otolith (O). (I,L) Antisense ODN-treated tailless species (I) and hybrid (L) embryos. Tailless species embryos exhibit normal morphology. Hybrid embryos are missing the otolith, exhibit disorganized endoderm cells, and lack an archenteron. (M±O) Antisense ODN-treated tailed species (M), tailless species (N), and hybrid (O) embryos. The tailed species and hybrid embryos exhibit large endodermal masses (arrowheads), whereas the tailless species embryo shows normal morphology. Scale bar, 20 mm for each frame.
experiments, antisense ODNs were designed to be effective in both species and hybrids. Antisense ODN-2 corresponds to the region immediately downstream of the start site in
MocuFH1 (Olsen and Jeffery, 1997), whereas antisense ODN-3 matches the corresponding region in MocuFH1, which differs in two nucleotides from MocuFH1 (Fig. 1).
54
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
Fig. 4. The effect of FH1 ODNs on expression of the CA1 cytoskeletal actin gene, a marker of notochord and muscle development. Tailless species (A,B) and hybrid (C,D) embryos are shown at 11 h. (A,C) Sense ODN-2 treated embryos. (B,D) Antisense ODN-2 treated embryos. N, notochord cells; M, muscle cells. Embryos are shown with the anterior poles at the top of each frame. Scale bar, 50 mm for each frame.
Embryos were treated with antisense and control (sense and scrambled sequence) ODNs just prior to ®rst cleavage and the effects were determined by morphological examination (Fig. 3A±L) and expression of the CA1 cytoskeletal actin gene (Fig. 4A,B), a marker of posterior larval (presumptive notochord and tail muscle) development in the tailed and tailless species (Jeffery et al., 1998). The same results were obtained with antisense ODN-2 and ODN-3. As described previously (Olsen and Jeffery, 1997), after antisense ODN treatment tailed species embryos cleaved normally, but did not undergo complete endoderm invagination at gastrulation, resulting in the protuberance of a lobe of endoderm cells at the vegetal pole (Fig. 3C). The embryos with the large endodermal mass did not develop into tadpole larvae with otoliths or tails (see Fig. 3M). In contrast, most tailed species embryos Table 1 The effect of ODNs on hatching in tailless species embryos. The results of an experiment involving 100±200 embryos are shown. The percent hatched embryos are indicated relative to untreated controls. Hatched larvae (% control) ODN-2 Sense control Scrambled control Antisense
90 84 33
ODN-3 Sense control Antisense
77 25
treated with sense (Fig. 3B) and scrambled sequence (data not shown) ODNs showed normal gastrulation and developed into tadpole larvae. Treatment of tailless species embryos with antisense but not control ODNs also affected gastrulation, however, a much smaller endodermal protuberance was formed (Fig. 3F). The antisense ODN-treated embryos eventually developed into anural larvae with normal-appearing posterior features (Fig. 3G±I). As shown in Fig. 4A,B, changes in CA1 expression in the presumptive notochord or muscle cells were not detected in antisense or sense ODN-treated tailless species embryos, con®rming that posterior development was not suppressed. However, the majority of tailless species embryos treated with antisense ODNs did not hatch (Table 1). The results suggest that FH1 is necessary for normal gastrulation and larval hatching but is not required for development of posterior features in the anural larva. 2.4. FH1 is required for restoration of chordate features in the hybrid larva The role of FH1 expression in restoration of chordate features was determined by treating hybrid embryos with antisense ODNs. Treatment with antisense ODN-2 or ODN3 reduced the number of hybrid and tailed species embryos with restored otoliths and tails to low levels relative to those treated with control ODNs (Table 2). The phenotypes of antisense ODN-treated and control hybrids are shown in Fig. 3J±L. Since the results were the same for ODN-2 and ODN-3 only the former is shown. Following treatment with antisense but not control ODNs, hybrid embryos cleaved but
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
55
Table 2 The effect of ODNs on development of hybrid and tailed species embryos a Hybrid
Tailed species
Otolith (% control)
Tail (% control)
Otolith (% control)
Tail (% control)
ODN-2 Sense control Scrambled control Antisense
53 68 12
62 74 37
96 91 2
98 91 0
ODN-3 Sense control Antisense
78 18
78 8
86 26
90 25
a The results of an experiment involving 100±200 embryos for each ODN are shown. The percent otolith and tail development are indicated relative to development in untreated controls.
when they reached gastrulation they exhibited a larger endodermal mass than tailless species embryos (data not shown, but see Fig. 3N). At later stages of development, the hybrid embryos showed disorganized endoderm cells and lacked a distinct archenteron (Fig. 3L). However, the most prominent feature of hybrid embryos treated with antisense (Fig. 3L) but not control (Fig. 3K) ODNs was the absence of otoliths and tails (Fig. 3N). There was no affect on CA1 expression in the muscle cells in antisense ODN-treated hybrid embryos, although CA1 mRNA levels were reduced in the notochord cells (Fig. 4C,D), consistent with the observation that FH1 functions upstream of CA1 in notochord development (Jeffery et al., 1998). Similar to anural larvae (Table 1), most hybrid larvae failed to hatch. Fig. 3M±O shows antisense ODN-treated larvae that developed from the three kinds of embryos. The tailed species and hybrid embryos exhibit large endodermal masses (Fig. 3M,N). In contrast, tailless species embryos showed normal morphology (Fig. 3O, also see Swalla and Jeffery, 1990). The results show that FH1 expression is necessary for the restoration of chordate features in hybrid embryos. 3. Discussion The HNF-3b gene has been implicated in gastrulation, notochord, CNS, and gut development in vertebrate embryos (Ang and Rossant, 1994; Sasaki and Hogan, 1994; Weinstein et al., 1994). Similarly, FH1, a forkhead gene related to HNF-3b, is required for gastrulation, notochord, and CNS development in ascidian embryos (Olsen and Jeffery, 1997). Here we show that the FH1 gene is also involved in the axial re-organization of the tadpole larva during ascidian evolution. The results suggest that forkhead genes may play important roles in the generation of novel modes of development. 3.1. Changes in FH1 expression Our results show that FH1 expression is altered in the tailless species. In the tailed species, FH1 is expressed
primarily in presumptive endoderm and notochord cells during the period from just prior to gastrulation through the late tailbud stage. This expression pattern is primitive and ancestral in urodele developers because it is mirrored by homologous forkhead genes in distantly related ascidian species exhibiting tadpole larvae (Ciona intestinalis, Corbo et al., 1997; Halocynthia roretzi, Shimauchi et al., 1997). FH1 expression in the tailless species is similar to the ancestral urodele pattern until neurulation when FH1 mRNA levels begin to be reduced in the tailless species, and continue to decline during the period when morphogenetic events leading to tail formation begin in the tailed species. These results imply that the modi®ed FH1 expression pattern of the tailless species is derived and correlates with tail-forming processes that have been altered during the evolution of anural development. The possible causes of modi®ed FH1 expression in the tailless species range from changes in FH1 mRNA transcription to FH1 mRNA turnover, and have not been elucidated in this study. We have demonstrated, however, that the prolonged FH1 mRNA accumulation period characteristic of the tailed species can be restored in hybrid embryos, in concert with the reappearance of the otolith, notochord, and tail (Swalla and Jeffery, 1990; Jeffery and Swalla, 1992). It has been previously shown that paternal genes introduced into tailless species eggs by tailed species sperm are expressed in hybrid embryos (Kusakabe et al., 1996). Accordingly, the FH1 mRNA expression pattern seen in hybrids may be mediated by (1) maintaining the longer expression period of the paternally derived FH1 allele, (2) lengthening the expression period of the maternally-derived FH1 allele, or (3) both of these processes. Unfortunately, the FH1 mRNAs of the tailed and tailless species do not diverge suf®ciently either in coding or noncoding regions to be discriminated in hybrids by in situ hybridization using gene speci®c probes. The restoration of the ancestral FH1 expression pattern in the hybrids, however, suggests that the paternal FH1 allele is dominant over or at least autonomous of its maternal counterpart, con®rming that anural development evolves by loss-of-function mutations (Jeffery, 1997).
56
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
3.2. Changes in FH1 function It was previously reported that FH1 is required for gastrulation and formation of the tadpole larva in the tailed species (Olsen and Jeffery, 1997). The phenotype of some of the affected embryos, which failed to develop an otolith and undergo posterior notochord cell movements resulting in tail formation, was suggestive of the normal appearance of the anural larva. As shown in the present study, embryos of the tailless species gastrulate abnormally after antisense ODN treatment, but recover in later embryonic stages, and eventually form anural larvae. These anural larvae exhibit normal posterior tissues, as indicated by their morphology and expression of the CA1 cytoskeletal actin gene. However, the anural larvae do not hatch from the chorion. The absence of hatching was also observed in embryos of the tailed species arrested in larval development by FH1 antisense ODNs (Olsen and Jeffery, 1997). It is known that the hatching process is initiated in the trunk region of molgulid ascidian larvae, probably by the secretory activity of anterior ectodermal cells (Berrill, 1931; Swalla and Jeffery, 1990). Therefore, FH1 expression in underlying endoderm cells may be required for development of hatching capacity in the anterior trunk ectoderm. The results suggest that FH1 is necessary for normal gastrulation and anterior larval development but is not required for posterior larval development in the tailless species. Our results show that the ancestral requirement for FH1 function in posterior larval development of the tailed species is recovered in hybrids. The failure of hybrids treated with FH1 antisense ODNs to develop an otolith and tail at the frequencies observed in controls supports the hypothesis that the FH1 gene is necessary for the restoration of these chordate features. Hybrid embryos contain about the same number of notochord precursor cells as tailless species embryos, yet these cells are able to rearrange themselves to form an extended notochord compared to that of the tailed species, which possess about four times as many notochord cells (Jeffery, 1997). Thus, FH1 appears to be involved in cell movements extending the notochord (and tail), rather than in producing the correct number of notochord cells. In addition, rescue in hybrids suggests FH1 is also required for the induction of the otolith by presumptive notochord, endoderm, or spinal cord cells (Reverberi et al., 1960; Nishida, 1991). Taken together, the effects of antisense ODNs on both species and hybrids suggest that FH1 is involved in four developmental processes: (1) gastrulation; (2) anterior movement of the endoderm cells resulting in archenteron formation and larval hatching; (3) posterior movement of notochord cells resulting in tail formation and (4) development of the capacity to induce the otolith in the developing CNS. The third and fourth functions of FH1 appear to be modi®ed or absent in the tailless species. How do changes in FH1 expression mediate the failure of posterior development in anural larvae? The HNF-3b protein is likely to serve as an essential transcription factor
in the genetic cascade leading to notochord development in chordates. This suggestion is supported by the phenotypes of HNF-3b knockout mice, which do not form notochords (Weinstein et al., 1994; Ang and Rossant, 1994). We propose that notochord development in ascidians requires a threshold concentration of FH1. Reduction of the FH1 concentration to below threshold values by treatment of tailed species or hybrid embryos with antisense ODNs, or during normal development in the tailless species, results in the abbreviation of notochord (and tail) development. Likewise, augmentation of FH1 levels above the threshold by increased expression in hybrids results in the rescue of notochord development and tail formation. We consider the alternative possibility that the orthologous FH1 genes of the tailed and tailless species have diverged in function to be unlikely because the two Molgula species have been separated for only a few million years (Had®eld et al., 1995), the winged helix DNA binding domains of the FH1 proteins are identical, and the regions ¯anking the DNA binding domains are also highly conserved in the FH1 proteins. 3.3. Gene hierarchies It has been shown that FH1 activity is required for CA1 expression in the notochord of ascidian embryos (Jeffery et al., 1998). We suspect that the genetic circuitry containing FH1 includes other genes known to be required for tail development in ascidians, such as Manx (Swalla et al., 1993; Swalla and Jeffery, 1996) and bobcat (Swalla et al., 1999), as well as the homologs of previously identi®ed genes necessary for notochord development, such as Brachyury (T) (Herrmann et al., 1990). The phenotypes produced by antisense inhibition of Manx (Swalla and Jeffery, 1996) and bobcat (Swalla et al., 1999) are similar to those produced by FH1 antisense ODNs in hybrid embryos (although not in the tailed species, Olsen and Jeffery, 1997), supporting the hypothesis that these genes may function in the same pathway leading to notochord and tail restoration. Weinstein et al. (1994) proposed that HNF-3b is located upstream of T in the pathway leading to notochord formation in the mouse. The ascidian T homolog is also expressed in the notochord (Yasuo and Satoh, 1994; Corbo et al., 1997). Therefore, future experiments will address the possible functional relationships between FH1, T, Manx and bobcat in the tailed and tailless species. Elucidation of the gene hierarchy involved in notochord development will be necessary to ascertain the molecular basis underlying the switch form urodele to anural development and also to understand the origin of the chordate body plan. 4. Experimental procedures 4.1. Biological materials Molgula oculata and Molgula occulta were collected and
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
maintained at Station Biologique, Roscoff, France. Hybrids were made by fertilizing M. occulta eggs with M. oculata sperm. Methods of gamete preparation and embryo culture have been described by Swalla and Jeffery (1990). 4.2. Library screening The MocuFH1 cDNA (Olsen and Jeffery, 1997) was labeled with 32P by random priming and used to screen a tailless species gastrula cDNA library prepared in the UniZap vector (Stratagene, La Jolla, CA). The probe was hybridized to phage lifts on Biodyne A nylon ®lters (Pall Biosupport, East Hills, NY) at high stringency. Positive clones were in vivo excised to obtain the pBluescript phagemid with the cDNA insert (ExAssist helper phage kit, Stratagene). One phagemid, designated AF-1, contained a 2.2kb insert, which was sequenced and found to encode MoccFH1. 4.3. Hybridizations The probe for northern and in situ hybridizations was prepared from a plasmid designated af-3, which contains the MocuFH1 insert (Olsen and Jeffery, 1997). The MocuFH1 probe is expected to react with the both FH1 RNAs because they are 94% similar in nucleotide sequence. RNA for northern blots was isolated from embryos at different stages as described previously (Swalla et al., 1993), electrophoresed through formaldehyde gels, and transferred to nylon membranes (MSI; Fisher Scienti®c, San Francisco, CA). The af-3 phagemid was linearized with Eco RI to serve as the template for synthesis of an antisense FH1 RNA probe using T7 RNA polymerase (Stratagene, La Jolla, CA) and [ 32P]UTP (800 Ci/mmol, Amersham, Arlington Heights, IL). Probes were hybridized to blots and washed at high stringency (Swalla et al., 1993). The whole-mount in situ hybridization method described by Olsen and Jeffery (1997) was followed using antisense and sense RNA probes synthesized from af-3 DNA. The section in situ hybridizations were carried as described by Jeffery et al. (1998) using MoccCA1 RNA probes. 4.4. Oligodeoxynucleotide treatment The following 18-mer phosphorothiolate-substituted ODNs were synthesized by Oligos, Etc., Inc. (Wilsonville, OR). As described previously (Olsen and Jeffery, 1997), antisense ODN-2 (5'-AGAAGGTGGCGACGAAAG-3') and sense ODN-2 (5'-CTTTCGTCGCCACCTTCT-3') span nucleotides 46 to 63 of the MocuFH1 cDNA sequence. The scrambled control ODN was 5'-GATGGAAACAGAAGGGCG-3'. Antisense ODN-3 (5'-AGAAGGTGGCGACGATAA-3') and sense ODN-3 (5'-TTATCGTCGCCACCTTCT-3') span nucleotides 115 to 132 of the MoccFH1 cDNA sequence, which corresponds to the region of MocuFH1 spanned by ODN-2. The ODNs were stored
57
lyophilized at 2208C, and a 30 nmol/ml stock solution was prepared in water prior to use. Treatment of embyros with ODNs was performed as described by Swalla and Jeffery (1996). Brie¯y, embryos (100±150/ml) were suspended in Millipore-®ltered seawater containing 30 mM ODN beginning just after ®rst cleavage (about 60 min post-fertilization) and incubated at 16±208C until hatching (10±12 h post-fertilization). The morphology of ODN-treated embryos was examined by light microscopy (Swalla and Jeffery, 1990). A subset of the ODN-treated embryos was ®xed in 4% paraformaldehyde or Bouin's ®xative, embedded in paraplast, and sectioned, and the sections were stained with hematoxylin-eosin. Acknowledgements We thank the staff at Station Biologique, Roscoff, France for collection of experimental animals, Dr. B. J. Swalla for providing RNA samples, and J. Machula for technical assistance. This research was supported by NIH (HD-13970) and NSF (IBN 94-17799 and IBN 98-07899) grants to WRJ. References Ang, S.L., Rossant, J., 1994. HNF3b is essential for node and notochord formation in mouse development. Cell 78, 561±574. Berrill, N.J., 1931. Studies in tunicate development. II. Abbreviation of development in the Molgulidae. Philos. Trans. R. Soc. Lond. B 219, 281±346. Corbo, J.C., Erives, A., Di Gregorio, A., Chang, A., Levine, M., 1997. Dorsoventral patterning of the vertebrate neural tube is conserved in a protochordate. Development 124, 2334±2344. Dirksen, M.L., Jamrich, M., 1992. A novel, activin inducible, blastopore lip speci®c gene of Xenopus laevis contains a fork head DNA-binding domain. Genes Dev. 6, 599±608. Di Gregorio, A., Levine, M., 1998. Ascidian embryogenesis and the origins of the chordate body plan. Curr. Opin. Genet. Dev. 8, 457±463. Had®eld, K.A., Swalla, B.J., Jeffery, W.R., 1995. Multiple origins of anural development in ascidians inferred from rDNA sequences. J. Mol. Evol. 40, 413±427. Herr, W., Cleary, M.A., 1995. The POU domain: A large conserved region in the mammalian pitl, octl, oct2, and Caenorhabditis elegans unc86 gene products. Genes Dev. 9, 1679±1693. Herrman, B.G., Labeit, S., Poustka, A., King, T.R., Lehrach, H., 1990. Cloning of the T gene required in mesoderm formation in the mouse. Nature 343, 617±622. Jeffery, W.R., 1997. Evolution of ascidian development. BioScience 47, 417±425. Jeffery, W.R., Swalla, B.J., 1990. Anural development in ascidians: evolutionary modi®cation and elimination of the tadpole larva. Sem. Dev. Biol. 1, 253±261. Jeffery, W.R., Swalla, B.J., 1992. Factors necessary for restoring an evolutionary change in an anural ascidian embryo. Dev. Biol. 153, 194±205. Jeffery, W.R., Ewing, N., Machula, M., Olsen, C.L., Swalla, B.J., 1998. Cytoskeletal actin genes function downstream of HNF-3b in ascidian notochord development. Int. J. Dev. Biol. 42, 1085±1092. Jun, S., Desplan, C., 1996. Co-operative interactions between paired domain and homeo-domain. Development 122, 2639±2650. Kaufmann, E., Knochel, W., 1996. Five years on the wings of fork head. Mech. Dev. 57, 3±20. Kusakabe, T., Swalla, B.J., Satoh, N., Jeffery, W.R., 1996. Mechanism of
58
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
an evolutionary change in muscle differentiation in ascidians with different modes of development. Dev. Biol. 174, 379±392. Littlewood, T.D., Evan, G.I., 1995. Transcription factors II, helix-loophelix. Protein Pro®le 2, 621±702. McGinnis, W., Krumlauf, R., 1992. Homeobox genes and axial patterning. Cell 68, 283±302. Nishida, H., 1991. Induction of brain and sensory pigment cells in the ascidian embryo analyzed by experiments with isolated blastomeres. Dev. 112, 389±395. Olsen, C.L., Jeffery, W.R., 1997. A forkhead gene related to HNF-3b is required for gastrulation and axis formation in the ascidian embryo. Development 124, 3609±3619. Reverberi, G., Ortolani, G., Farinella-Ferruzza, N., 1960. The causal formation of the brain in the ascidian larva. Acta Embryol. Morphol. Exp. 3, 296±336. Sasaki, H., Hogan, B.L.M., 1994. HNF-3b as a regulator of ¯oor plate development. Cell 76, 103±115. Satoh, N., 1994. Developmental Biology of Ascidians. Cambridge University Press, New York. Satoh, N., Jeffery, W.R., 1995. Chasing tails in ascidians. Developmental insights into the origin and evolution of chordates. Trends Genet. 11, 354±359. Shimauchi, Y., Yasuo, H., Satoh, N., 1997. Autonomy of ascidian fork head/HNF-3 gene expression. Mech. Dev. 69, 143±154.
Swalla, B.J., Jeffery, W.R., 1990. Interspeci®c hybridization between an anural and urodele ascidian, differential expression of urodele features suggests multiple mechanisms control anural development. Dev. Biol. 142, 319±334. Swalla, B.J., Jeffery, W.R., 1996. Requirement of the Manx gene for restoration of chordate features in a tailless ascidian embryo. Science 274, 1205±1208. Swalla, B.J., Makabe, K.W., Satoh, N., Jeffery, W.R., 1993. Novel genes expressed differentially in ascidians with alternate modes of development. Development 119, 307±318. Swalla, B.J., Just, M.E., Pederson, E.L., Jeffery, W.R., 1999. A multigene locus containing the Manx and bobcat genes is required for development of chordate features in the ascidian tadpole larva. Development 126, 1643±1653. Weinstein, D.C., Ruiz i Altaba, A., Chen, W.S., Hoodless, P., Prezioso, V.R., Jessell, T.M., Darnell, J.E., 1994. The winged helix transcription factor HNF3b is required for notochord development in the mouse embryo. Cell 78, 575±588. Whittaker, J.R., 1979. Development of vestigial tail muscle acetylcholinesterase in embryos of an anural ascidian species. Biol. Bull. 156, 393± 407. Yasuo, H., Satoh, N., 1994. An ascidian homolog of the mouse Brachyury (T) gene is expressed exclusively in notochord cells at the fate restricted stage. Dev. Growth Diff. 36, 9±18.
The forkhead gene FH1 is involved in evolutionary modi®cation of the ascidian tadpole larva Catherine L. Olsen a,c, Jeanette E. Natzle a, William R. Jeffery a,b,c,* a
Graduate Program in Cell and Developmental Biology, Section of Molecular and Cellular Biology, University of California, Davis, CA 95616, USA b Department of Biology, University of Maryland, College Park, MD 20742, USA c Station Biologique, Roscoff, France Received 8 January 1999; received in revised form 8 March 1999; accepted 10 March 1999
Abstract The forkhead gene FH1 encodes a HNF-3b protein required for gastrulation and development of chordate features in the ascidian tadpole larva. Although most ascidian species develop via a tadpole larva, the conventional larva has regressed into an anural (tailless) larva in some species. Molgula oculata (the tailed species) exhibits a tadpole larva with chordate features (a dorsal neural sensory organ or otolith, a notochord, striated muscle cells, and a tail), whereas its sister species Molgula occulta (the tailless species) has evolved an anural larva, which has lost these features. Here we examine the role of FH1 in modifying the larval body plan in the tailless species. We also examine FH1 function in tailless species £ tailed species hybrids, in which the otolith, notochord, and tail are restored. The FH1 gene is expressed primarily in the presumptive endoderm and notochord cells during gastrulation, neurulation, and larval axis formation in both species and hybrids. In the tailless species, FH1 expression is down-regulated after neurulation in concert with arrested otolith, notochord, and tail development. The FH1 expression pattern characteristic of the tailed species is restored in hybrid embryos prior to the development of chordate larval features. Antisense oligodeoxynucleotides (ODNs) shown previously to disrupt FH1 function were used to compare the developmental roles of this gene in both species and hybrids. As described previously, antisense FH1 ODNs inhibited endoderm invagination during gastrulation, notochord extension, and larval tail formation in the tailed species. Antisense FH1 ODNs also affected gastrulation in the tailless species, although the effects were less severe than in the tailed species, and an anural larva was formed. In hybrid embryos, antisense FH1 ODNs blocked restoration of the otolith, notochord, and tail, reverting the larva back to the anural state. The results suggest that changes in FH1 expression are involved in re-organizing the tadpole larva during the evolution of anural development. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Forkhead genes; Chordate body plan; Ascidian development; Antisense inhibition; Evolution of larval development
1. Introduction Developmental processes are controlled by conserved regulatory genes (McGinnis and Krumlauf, 1992; Littlewood and Evan, 1995; Herr and Cleary, 1995; Jun and Desplan, 1996), but little is known about how these genes generate novel modes of development. Ascidians have served as a model system for studying the evolution of developmental mechanisms (Jeffery, 1997). Most ascidian species pass through a tadpole (or urodele) larval stage (Satoh, 1994). The tadpole larva exhibits a dorsal central nervous system, a notochord, and striated tail muscle cells, and thus is considered a prototype of the ancestral chordate (Satoh and Jeffery, 1995; Di Gregorio and Levine, 1998). * Corresponding author. Tel.: 11-301-405-5454; fax: 11-301-3149358. E-mail address: [email protected] (W.R. Jeffery)
However, a few ascidian species have evolved a tailless (or anural) larva (Berrill, 1931; Jeffery and Swalla, 1990). The anural larva has lost the typical chordate features, although their progenitors are present during embryogenesis (Whittaker, 1979; Swalla and Jeffery, 1990). Anural species have appeared multiple times from urodele ancestors in different ascidian lineages (Had®eld et al., 1995), indicating that reorganization of the chordate body plan has occurred repeatedly during ascidian evolution. The ascidians Molgula oculata (the tailed species), a urodele developer, and Molgula occulta (the tailless species), an anural developer, are capable of interspeci®c hybridization (Swalla and Jeffery, 1990; Jeffery and Swalla, 1992). When tailed species eggs are fertilized with tailless species sperm the resulting hybrid larvae are identical to those of the tailed species. In contrast, when tailless species eggs are fertilized with tailed species sperm, the otolith,
0925-4773/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0925-477 3(99)00061-1
Fig. 1. The nucleotide and deduced amino acid sequences of MoccFH1 and MocuFH1. In the MocuFH1 sequence (lower), dots represent identical nucleotides and letters represent amino acid substitutions. The amino acids differing between the two cDNA clones are listed side by side with the MoccFH1 amino acid listed ®rst (e.g. S/G). The putative winged helix DNA-binding domain is shaded, and the region used to design antisense and sense ODNs has a line drawn above it. The Genbank accession number for MoccFH1 is AF082992.
50 C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
51
Fig. 2. FH1 mRNA expression in embryos of the tailed species, tailless species, and hybrids. (A) Temporal expression of FH1 mRNA during development of the tailless species. Each lane contains RNA isolated from (1) unfertilized eggs, (2) fertilized eggs, (3) gastrulae, (4) neurulae, (5) 6 h embryos, (6) 8 h embryos, and (7) 10 h embryos of the tailless species. Top, MocuFH1 probe. The 2.3 kb FH1 transcript is indicated by the arrow. Bottom, 18S rRNA loading control. (B± J) Tailed species (B±D), tailless species (E±G), and hybrid (H±J) embryos subjected to in situ hybridization with an MocuFH1 antisense RNA probe. (B,C,E,F,H,I) Whole mounts. (D,G,J), sections of whole mounts. (B) A tailed species gastrula showing staining in presumptive notochord (N) cells at the anterior lip of the blastopore and endoderm (E) cells moving inside the blastopore. (C±D) Tailed species neurula (C) and early tailbud stage (D, 6 h) showing staining in the notochord (N) and endoderm (E) cells. (E) A tailless species gastrula showing staining in notochord (N) and endoderm (E) cells. (F,G) Tailless species neurula (G) and 6 h embryo (G) showing staining in notochord (N) and endoderm (E) cells. (H±J) Hybrid gastrula (H), neurula (I) and 6 h embryo (J) showing staining in notochord (N) and endoderm (E) cells. Embryos are shown with anterior pole at the top except in D, where it is shown to the left. Scale bar, 20 mm for each frame.
notochord, and tail are restored in the anural larva. These results imply that loss of function mutations in zygotic genes are responisble for modifying the tadpole larva in the tailless species. A subtractive screen involving the two species has been done to identify some of these genes (Swalla et al., 1993). The screen resulted in the isolation and characterization of the zinc ®nger gene Manx, the leucine zipper gene lynx, and the tyrosine kinase gene cymric, which are downregulated in the tailless species (Jeffery, 1997). In addition, the RNA helicase gene bobcat, which is also down-regulated in the tailless species, was identi®ed by its proximity to Manx in the tailed species genome (Swalla et al., 1999). Disruption of zygotic Manx and bobcat expression by anti-
sense oligodeoxynucleotides (ODNs) prevents restoration of chordate features in hybrids, showing that these genes are necessary for tadpole larval development (Swalla and Jeffery, 1996; Swalla et al., 1999). Genes involved in axis formation in other chordates could also play a role in the evolution of anural development. The HNF-3/forkhead genes are prime candidates for such a role because of their involvement in notochord development in vertebrate embryos (Sasaki and Hogan, 1994; Dirksen and Jamrich, 1992). The putative transcription factors encoded by these genes share the winged helix DNA-binding domain and are expressed in embryonic organizer regions (Kaufmann and Knochel, 1996). For example, HNF-3b is
52
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
expressed in the dorsal lip of the blastopore during Xenopus development and in the node region during mouse development. Later, HNF-3b is expressed in the notochord and ¯oor plate of the neural tube, where it functions in dorsoventral patterning the CNS, and in the endoderm, where it plays a role in gut differentiation. Mouse embryos homozygous for a mutated HNF-3b gene fail to form a notochord or gut and show defects in node, paraxial mesoderm, and neural tube organization (Ang and Rossant, 1994; Weinstein et al., 1994). We previously identi®ed FH1, an ascidian forkhead gene encoding a protein related to HNF-3b (Olsen and Jeffery, 1997). Disruption of FH1 expression by antisense ODNs showed that this gene is required for gastrulation and tadpole larval development in the tailed species. Here we describe the orthologous FH1 gene in the tailless species and show that expression is down-regulated during anural development and restored in hybrids. We also demonstrate that, although FH1 expression is unnecessary for posterior development of the anural larva, which has lost chordate features, it is required for restoration of chordate features in hybrids. These results suggest that changes in FH1 expression are involved in the re-organization of the tadpole larva during the evolution of anural development. 2. Results 2.1. Isolation of an FH1 gene in the tailless species We previously isolated and characterized MocuFH1, a cDNA clone encoding the FH1 gene in the tailed species (Olsen and Jeffery, 1997). To identify the orthologous gene in the tailless species, we screened a gastrula cDNA library with MocuFH1 and isolated MoccFH1 (Molgula occulta ForkHead 1). The 2.2 kb MoccFH1 cDNA exhibits a fulllength 1703 bp ORF, which is preceded by a 109 bp 5' UTR and followed by a 387 bp 3' UTR (Fig. 1). The 3' UTR contains a putative polyadenylation signal but no poly (A) tail is present. MoccFH1 and MocuFH1 are 94% similar at the nucleotide level, exhibit highly conserved 5' and 3' UTRs, and the deduced FH1 proteins are 96% similar (Fig. 1). Amino acid substitutions are distributed throughout the N- and C-terminal regions of the FH1 proteins, but their winged helix DNA-binding domains are identical. A `BLAST' search indicated that MoccFH1 is more closely related to MocuFH1 (Olsen and Jeffery, 1997) than to other ascidian HNF/forkhead genes ( Corbo et al., 1997; Shimauchi et al., 1997), suggesting that these cDNA clones encode orthologous FH1 genes in the tailless and tailed species. 2.2. Changes in FH1 expression during embryogenesis The accumulation of FH1 mRNA during embryogenesis was followed by northern and in situ hybridizations. In the tailed species, it was shown previously that the 2.3 kb FH1 mRNA appeared by the beginning of gastrulation, peaked
during neurulation, declined to the level seen at gastrulation by the mid-tailbud stage, and was reduced to low levels by the late tailbud stage (Olsen and Jeffery, 1997). The FH1 mRNA is also about 2.3 kb in the tailless species (Fig. 2A), consistent with the size of the MoccFH1 cDNA (Fig. 1). In general, the temporal pattern of FH1 accumulation was similar in the tailed and tailless species. However, two notable differences were observed in the tailless species (Fig. 2A). First, although FH1 transcripts ®rst appeared during gastrulation, there was no peak at the neurula stage. Instead, FH1 transcripts remained at about the same level from gastrulation through the 6 h stage, which is roughly equivalent to the early to mid tailbud stage in the tailed species. Second, FH1 transcripts appeared to decline earlier after neurulation in the tailless species. The in situ hybridization results are shown in Fig. 2B±J. These experiments were carried out using the same probe and identical hybridization conditions for each kind of embryo. In addition, the tailless species and hybrid embryos used in these experiments were derived from the same clutch of eggs. Therefore, the intensity of FH1 mRNA staining is directly comparable between the different kinds of embryos at the same stage of development. FH1 transcripts were detected primarily in the presumptive endoderm and notochord during gastrulation, neurulation, and tailbud formation in both species and hybrids (Fig. 2B±J). However, in the tailless species FH1 mRNA began to decrease in the endoderm cells at the neurula stage (Fig. 2C,F), and notochord and endoderm cells showed reduced transcript levels at 6 h of development (Fig. 2D,G). The staining intensity was variable between different clutches of tailless species embryos but in no case matched the higher levels observed in the tailed species. In hybrid embryos, FH1 staining during gastrulation was similar to that observed in the tailed and tailless species (Fig. 2B,E,H). However, hybrids showed enhanced FH1 transcript levels in endoderm and notochord cells relative to tailless species embryos beginning at the neurula stage (Fig.2F,I) and extending through 6 h of development (Fig. 2G,J). The proportion of hybrid embryos with enhanced FH1 expression was about 70%, the same as the percentage of hybrids that restored the otolith and tail in this experiment. The results show that (1) FH1 is expressed in the same embryonic tissues in the tailed species, tailless species, and hybrids, (2) the intensity and duration of intense FH1 expression is reduced in the tailless species and (3) the FH1 expression period characteristic of the tailed species is restored in hybrid embryos. 2.3. FH1 is not required for development of posterior features in the anural larva Previous studies showed that antisense ODNs reduce FH1 mRNA levels and prevent gastrulation and development of the tadpole larva (Olsen and Jeffery, 1997). Thus, we have used the antisense approach to determine whether FH1 is required for development of the anural larva. In these
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
53
Fig. 3. The effects of FH1 ODNs on gastrulation and larval development. (A,D,G,J) Untreated control embryos. (B,E,H,K) Sense ODN-2 treated embryos. (C,F,I,L,M±O) Antisense ODN-2 treated embryos. (A±F) Sagittal sections of tailed (A±C) and tailless (D±F) species mid-gastrulae. Untreated and sense ODNtreated gastrulae in (A,B,D,E) show normal morphology with notochord cells (N) involuted at the anterior lip and muscle cells (M) involuted at the posterior lip of the blastopore (B). (C,F) Antisense ODN-treated tailed (C) and tailless (F) gastrulae showing the mass of endoderm cells (EM) at the vegetal pole. Transverse sections of 11 h tailless species (G±I) and hybrid (J±L) embryos. (G,H,J,K) Untreated (G,J) and sense ODN-treated tailless species (G,H) and hybrid (J,K) embryos showing normal morphology, including an archenteron (Ar) surrounded by endoderm cells (E). Hybrid embryos also exhibit an otolith (O). (I,L) Antisense ODN-treated tailless species (I) and hybrid (L) embryos. Tailless species embryos exhibit normal morphology. Hybrid embryos are missing the otolith, exhibit disorganized endoderm cells, and lack an archenteron. (M±O) Antisense ODN-treated tailed species (M), tailless species (N), and hybrid (O) embryos. The tailed species and hybrid embryos exhibit large endodermal masses (arrowheads), whereas the tailless species embryo shows normal morphology. Scale bar, 20 mm for each frame.
experiments, antisense ODNs were designed to be effective in both species and hybrids. Antisense ODN-2 corresponds to the region immediately downstream of the start site in
MocuFH1 (Olsen and Jeffery, 1997), whereas antisense ODN-3 matches the corresponding region in MocuFH1, which differs in two nucleotides from MocuFH1 (Fig. 1).
54
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
Fig. 4. The effect of FH1 ODNs on expression of the CA1 cytoskeletal actin gene, a marker of notochord and muscle development. Tailless species (A,B) and hybrid (C,D) embryos are shown at 11 h. (A,C) Sense ODN-2 treated embryos. (B,D) Antisense ODN-2 treated embryos. N, notochord cells; M, muscle cells. Embryos are shown with the anterior poles at the top of each frame. Scale bar, 50 mm for each frame.
Embryos were treated with antisense and control (sense and scrambled sequence) ODNs just prior to ®rst cleavage and the effects were determined by morphological examination (Fig. 3A±L) and expression of the CA1 cytoskeletal actin gene (Fig. 4A,B), a marker of posterior larval (presumptive notochord and tail muscle) development in the tailed and tailless species (Jeffery et al., 1998). The same results were obtained with antisense ODN-2 and ODN-3. As described previously (Olsen and Jeffery, 1997), after antisense ODN treatment tailed species embryos cleaved normally, but did not undergo complete endoderm invagination at gastrulation, resulting in the protuberance of a lobe of endoderm cells at the vegetal pole (Fig. 3C). The embryos with the large endodermal mass did not develop into tadpole larvae with otoliths or tails (see Fig. 3M). In contrast, most tailed species embryos Table 1 The effect of ODNs on hatching in tailless species embryos. The results of an experiment involving 100±200 embryos are shown. The percent hatched embryos are indicated relative to untreated controls. Hatched larvae (% control) ODN-2 Sense control Scrambled control Antisense
90 84 33
ODN-3 Sense control Antisense
77 25
treated with sense (Fig. 3B) and scrambled sequence (data not shown) ODNs showed normal gastrulation and developed into tadpole larvae. Treatment of tailless species embryos with antisense but not control ODNs also affected gastrulation, however, a much smaller endodermal protuberance was formed (Fig. 3F). The antisense ODN-treated embryos eventually developed into anural larvae with normal-appearing posterior features (Fig. 3G±I). As shown in Fig. 4A,B, changes in CA1 expression in the presumptive notochord or muscle cells were not detected in antisense or sense ODN-treated tailless species embryos, con®rming that posterior development was not suppressed. However, the majority of tailless species embryos treated with antisense ODNs did not hatch (Table 1). The results suggest that FH1 is necessary for normal gastrulation and larval hatching but is not required for development of posterior features in the anural larva. 2.4. FH1 is required for restoration of chordate features in the hybrid larva The role of FH1 expression in restoration of chordate features was determined by treating hybrid embryos with antisense ODNs. Treatment with antisense ODN-2 or ODN3 reduced the number of hybrid and tailed species embryos with restored otoliths and tails to low levels relative to those treated with control ODNs (Table 2). The phenotypes of antisense ODN-treated and control hybrids are shown in Fig. 3J±L. Since the results were the same for ODN-2 and ODN-3 only the former is shown. Following treatment with antisense but not control ODNs, hybrid embryos cleaved but
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
55
Table 2 The effect of ODNs on development of hybrid and tailed species embryos a Hybrid
Tailed species
Otolith (% control)
Tail (% control)
Otolith (% control)
Tail (% control)
ODN-2 Sense control Scrambled control Antisense
53 68 12
62 74 37
96 91 2
98 91 0
ODN-3 Sense control Antisense
78 18
78 8
86 26
90 25
a The results of an experiment involving 100±200 embryos for each ODN are shown. The percent otolith and tail development are indicated relative to development in untreated controls.
when they reached gastrulation they exhibited a larger endodermal mass than tailless species embryos (data not shown, but see Fig. 3N). At later stages of development, the hybrid embryos showed disorganized endoderm cells and lacked a distinct archenteron (Fig. 3L). However, the most prominent feature of hybrid embryos treated with antisense (Fig. 3L) but not control (Fig. 3K) ODNs was the absence of otoliths and tails (Fig. 3N). There was no affect on CA1 expression in the muscle cells in antisense ODN-treated hybrid embryos, although CA1 mRNA levels were reduced in the notochord cells (Fig. 4C,D), consistent with the observation that FH1 functions upstream of CA1 in notochord development (Jeffery et al., 1998). Similar to anural larvae (Table 1), most hybrid larvae failed to hatch. Fig. 3M±O shows antisense ODN-treated larvae that developed from the three kinds of embryos. The tailed species and hybrid embryos exhibit large endodermal masses (Fig. 3M,N). In contrast, tailless species embryos showed normal morphology (Fig. 3O, also see Swalla and Jeffery, 1990). The results show that FH1 expression is necessary for the restoration of chordate features in hybrid embryos. 3. Discussion The HNF-3b gene has been implicated in gastrulation, notochord, CNS, and gut development in vertebrate embryos (Ang and Rossant, 1994; Sasaki and Hogan, 1994; Weinstein et al., 1994). Similarly, FH1, a forkhead gene related to HNF-3b, is required for gastrulation, notochord, and CNS development in ascidian embryos (Olsen and Jeffery, 1997). Here we show that the FH1 gene is also involved in the axial re-organization of the tadpole larva during ascidian evolution. The results suggest that forkhead genes may play important roles in the generation of novel modes of development. 3.1. Changes in FH1 expression Our results show that FH1 expression is altered in the tailless species. In the tailed species, FH1 is expressed
primarily in presumptive endoderm and notochord cells during the period from just prior to gastrulation through the late tailbud stage. This expression pattern is primitive and ancestral in urodele developers because it is mirrored by homologous forkhead genes in distantly related ascidian species exhibiting tadpole larvae (Ciona intestinalis, Corbo et al., 1997; Halocynthia roretzi, Shimauchi et al., 1997). FH1 expression in the tailless species is similar to the ancestral urodele pattern until neurulation when FH1 mRNA levels begin to be reduced in the tailless species, and continue to decline during the period when morphogenetic events leading to tail formation begin in the tailed species. These results imply that the modi®ed FH1 expression pattern of the tailless species is derived and correlates with tail-forming processes that have been altered during the evolution of anural development. The possible causes of modi®ed FH1 expression in the tailless species range from changes in FH1 mRNA transcription to FH1 mRNA turnover, and have not been elucidated in this study. We have demonstrated, however, that the prolonged FH1 mRNA accumulation period characteristic of the tailed species can be restored in hybrid embryos, in concert with the reappearance of the otolith, notochord, and tail (Swalla and Jeffery, 1990; Jeffery and Swalla, 1992). It has been previously shown that paternal genes introduced into tailless species eggs by tailed species sperm are expressed in hybrid embryos (Kusakabe et al., 1996). Accordingly, the FH1 mRNA expression pattern seen in hybrids may be mediated by (1) maintaining the longer expression period of the paternally derived FH1 allele, (2) lengthening the expression period of the maternally-derived FH1 allele, or (3) both of these processes. Unfortunately, the FH1 mRNAs of the tailed and tailless species do not diverge suf®ciently either in coding or noncoding regions to be discriminated in hybrids by in situ hybridization using gene speci®c probes. The restoration of the ancestral FH1 expression pattern in the hybrids, however, suggests that the paternal FH1 allele is dominant over or at least autonomous of its maternal counterpart, con®rming that anural development evolves by loss-of-function mutations (Jeffery, 1997).
56
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
3.2. Changes in FH1 function It was previously reported that FH1 is required for gastrulation and formation of the tadpole larva in the tailed species (Olsen and Jeffery, 1997). The phenotype of some of the affected embryos, which failed to develop an otolith and undergo posterior notochord cell movements resulting in tail formation, was suggestive of the normal appearance of the anural larva. As shown in the present study, embryos of the tailless species gastrulate abnormally after antisense ODN treatment, but recover in later embryonic stages, and eventually form anural larvae. These anural larvae exhibit normal posterior tissues, as indicated by their morphology and expression of the CA1 cytoskeletal actin gene. However, the anural larvae do not hatch from the chorion. The absence of hatching was also observed in embryos of the tailed species arrested in larval development by FH1 antisense ODNs (Olsen and Jeffery, 1997). It is known that the hatching process is initiated in the trunk region of molgulid ascidian larvae, probably by the secretory activity of anterior ectodermal cells (Berrill, 1931; Swalla and Jeffery, 1990). Therefore, FH1 expression in underlying endoderm cells may be required for development of hatching capacity in the anterior trunk ectoderm. The results suggest that FH1 is necessary for normal gastrulation and anterior larval development but is not required for posterior larval development in the tailless species. Our results show that the ancestral requirement for FH1 function in posterior larval development of the tailed species is recovered in hybrids. The failure of hybrids treated with FH1 antisense ODNs to develop an otolith and tail at the frequencies observed in controls supports the hypothesis that the FH1 gene is necessary for the restoration of these chordate features. Hybrid embryos contain about the same number of notochord precursor cells as tailless species embryos, yet these cells are able to rearrange themselves to form an extended notochord compared to that of the tailed species, which possess about four times as many notochord cells (Jeffery, 1997). Thus, FH1 appears to be involved in cell movements extending the notochord (and tail), rather than in producing the correct number of notochord cells. In addition, rescue in hybrids suggests FH1 is also required for the induction of the otolith by presumptive notochord, endoderm, or spinal cord cells (Reverberi et al., 1960; Nishida, 1991). Taken together, the effects of antisense ODNs on both species and hybrids suggest that FH1 is involved in four developmental processes: (1) gastrulation; (2) anterior movement of the endoderm cells resulting in archenteron formation and larval hatching; (3) posterior movement of notochord cells resulting in tail formation and (4) development of the capacity to induce the otolith in the developing CNS. The third and fourth functions of FH1 appear to be modi®ed or absent in the tailless species. How do changes in FH1 expression mediate the failure of posterior development in anural larvae? The HNF-3b protein is likely to serve as an essential transcription factor
in the genetic cascade leading to notochord development in chordates. This suggestion is supported by the phenotypes of HNF-3b knockout mice, which do not form notochords (Weinstein et al., 1994; Ang and Rossant, 1994). We propose that notochord development in ascidians requires a threshold concentration of FH1. Reduction of the FH1 concentration to below threshold values by treatment of tailed species or hybrid embryos with antisense ODNs, or during normal development in the tailless species, results in the abbreviation of notochord (and tail) development. Likewise, augmentation of FH1 levels above the threshold by increased expression in hybrids results in the rescue of notochord development and tail formation. We consider the alternative possibility that the orthologous FH1 genes of the tailed and tailless species have diverged in function to be unlikely because the two Molgula species have been separated for only a few million years (Had®eld et al., 1995), the winged helix DNA binding domains of the FH1 proteins are identical, and the regions ¯anking the DNA binding domains are also highly conserved in the FH1 proteins. 3.3. Gene hierarchies It has been shown that FH1 activity is required for CA1 expression in the notochord of ascidian embryos (Jeffery et al., 1998). We suspect that the genetic circuitry containing FH1 includes other genes known to be required for tail development in ascidians, such as Manx (Swalla et al., 1993; Swalla and Jeffery, 1996) and bobcat (Swalla et al., 1999), as well as the homologs of previously identi®ed genes necessary for notochord development, such as Brachyury (T) (Herrmann et al., 1990). The phenotypes produced by antisense inhibition of Manx (Swalla and Jeffery, 1996) and bobcat (Swalla et al., 1999) are similar to those produced by FH1 antisense ODNs in hybrid embryos (although not in the tailed species, Olsen and Jeffery, 1997), supporting the hypothesis that these genes may function in the same pathway leading to notochord and tail restoration. Weinstein et al. (1994) proposed that HNF-3b is located upstream of T in the pathway leading to notochord formation in the mouse. The ascidian T homolog is also expressed in the notochord (Yasuo and Satoh, 1994; Corbo et al., 1997). Therefore, future experiments will address the possible functional relationships between FH1, T, Manx and bobcat in the tailed and tailless species. Elucidation of the gene hierarchy involved in notochord development will be necessary to ascertain the molecular basis underlying the switch form urodele to anural development and also to understand the origin of the chordate body plan. 4. Experimental procedures 4.1. Biological materials Molgula oculata and Molgula occulta were collected and
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
maintained at Station Biologique, Roscoff, France. Hybrids were made by fertilizing M. occulta eggs with M. oculata sperm. Methods of gamete preparation and embryo culture have been described by Swalla and Jeffery (1990). 4.2. Library screening The MocuFH1 cDNA (Olsen and Jeffery, 1997) was labeled with 32P by random priming and used to screen a tailless species gastrula cDNA library prepared in the UniZap vector (Stratagene, La Jolla, CA). The probe was hybridized to phage lifts on Biodyne A nylon ®lters (Pall Biosupport, East Hills, NY) at high stringency. Positive clones were in vivo excised to obtain the pBluescript phagemid with the cDNA insert (ExAssist helper phage kit, Stratagene). One phagemid, designated AF-1, contained a 2.2kb insert, which was sequenced and found to encode MoccFH1. 4.3. Hybridizations The probe for northern and in situ hybridizations was prepared from a plasmid designated af-3, which contains the MocuFH1 insert (Olsen and Jeffery, 1997). The MocuFH1 probe is expected to react with the both FH1 RNAs because they are 94% similar in nucleotide sequence. RNA for northern blots was isolated from embryos at different stages as described previously (Swalla et al., 1993), electrophoresed through formaldehyde gels, and transferred to nylon membranes (MSI; Fisher Scienti®c, San Francisco, CA). The af-3 phagemid was linearized with Eco RI to serve as the template for synthesis of an antisense FH1 RNA probe using T7 RNA polymerase (Stratagene, La Jolla, CA) and [ 32P]UTP (800 Ci/mmol, Amersham, Arlington Heights, IL). Probes were hybridized to blots and washed at high stringency (Swalla et al., 1993). The whole-mount in situ hybridization method described by Olsen and Jeffery (1997) was followed using antisense and sense RNA probes synthesized from af-3 DNA. The section in situ hybridizations were carried as described by Jeffery et al. (1998) using MoccCA1 RNA probes. 4.4. Oligodeoxynucleotide treatment The following 18-mer phosphorothiolate-substituted ODNs were synthesized by Oligos, Etc., Inc. (Wilsonville, OR). As described previously (Olsen and Jeffery, 1997), antisense ODN-2 (5'-AGAAGGTGGCGACGAAAG-3') and sense ODN-2 (5'-CTTTCGTCGCCACCTTCT-3') span nucleotides 46 to 63 of the MocuFH1 cDNA sequence. The scrambled control ODN was 5'-GATGGAAACAGAAGGGCG-3'. Antisense ODN-3 (5'-AGAAGGTGGCGACGATAA-3') and sense ODN-3 (5'-TTATCGTCGCCACCTTCT-3') span nucleotides 115 to 132 of the MoccFH1 cDNA sequence, which corresponds to the region of MocuFH1 spanned by ODN-2. The ODNs were stored
57
lyophilized at 2208C, and a 30 nmol/ml stock solution was prepared in water prior to use. Treatment of embyros with ODNs was performed as described by Swalla and Jeffery (1996). Brie¯y, embryos (100±150/ml) were suspended in Millipore-®ltered seawater containing 30 mM ODN beginning just after ®rst cleavage (about 60 min post-fertilization) and incubated at 16±208C until hatching (10±12 h post-fertilization). The morphology of ODN-treated embryos was examined by light microscopy (Swalla and Jeffery, 1990). A subset of the ODN-treated embryos was ®xed in 4% paraformaldehyde or Bouin's ®xative, embedded in paraplast, and sectioned, and the sections were stained with hematoxylin-eosin. Acknowledgements We thank the staff at Station Biologique, Roscoff, France for collection of experimental animals, Dr. B. J. Swalla for providing RNA samples, and J. Machula for technical assistance. This research was supported by NIH (HD-13970) and NSF (IBN 94-17799 and IBN 98-07899) grants to WRJ. References Ang, S.L., Rossant, J., 1994. HNF3b is essential for node and notochord formation in mouse development. Cell 78, 561±574. Berrill, N.J., 1931. Studies in tunicate development. II. Abbreviation of development in the Molgulidae. Philos. Trans. R. Soc. Lond. B 219, 281±346. Corbo, J.C., Erives, A., Di Gregorio, A., Chang, A., Levine, M., 1997. Dorsoventral patterning of the vertebrate neural tube is conserved in a protochordate. Development 124, 2334±2344. Dirksen, M.L., Jamrich, M., 1992. A novel, activin inducible, blastopore lip speci®c gene of Xenopus laevis contains a fork head DNA-binding domain. Genes Dev. 6, 599±608. Di Gregorio, A., Levine, M., 1998. Ascidian embryogenesis and the origins of the chordate body plan. Curr. Opin. Genet. Dev. 8, 457±463. Had®eld, K.A., Swalla, B.J., Jeffery, W.R., 1995. Multiple origins of anural development in ascidians inferred from rDNA sequences. J. Mol. Evol. 40, 413±427. Herr, W., Cleary, M.A., 1995. The POU domain: A large conserved region in the mammalian pitl, octl, oct2, and Caenorhabditis elegans unc86 gene products. Genes Dev. 9, 1679±1693. Herrman, B.G., Labeit, S., Poustka, A., King, T.R., Lehrach, H., 1990. Cloning of the T gene required in mesoderm formation in the mouse. Nature 343, 617±622. Jeffery, W.R., 1997. Evolution of ascidian development. BioScience 47, 417±425. Jeffery, W.R., Swalla, B.J., 1990. Anural development in ascidians: evolutionary modi®cation and elimination of the tadpole larva. Sem. Dev. Biol. 1, 253±261. Jeffery, W.R., Swalla, B.J., 1992. Factors necessary for restoring an evolutionary change in an anural ascidian embryo. Dev. Biol. 153, 194±205. Jeffery, W.R., Ewing, N., Machula, M., Olsen, C.L., Swalla, B.J., 1998. Cytoskeletal actin genes function downstream of HNF-3b in ascidian notochord development. Int. J. Dev. Biol. 42, 1085±1092. Jun, S., Desplan, C., 1996. Co-operative interactions between paired domain and homeo-domain. Development 122, 2639±2650. Kaufmann, E., Knochel, W., 1996. Five years on the wings of fork head. Mech. Dev. 57, 3±20. Kusakabe, T., Swalla, B.J., Satoh, N., Jeffery, W.R., 1996. Mechanism of
58
C.L. Olsen et al. / Mechanisms of Development 85 (1999) 49±58
an evolutionary change in muscle differentiation in ascidians with different modes of development. Dev. Biol. 174, 379±392. Littlewood, T.D., Evan, G.I., 1995. Transcription factors II, helix-loophelix. Protein Pro®le 2, 621±702. McGinnis, W., Krumlauf, R., 1992. Homeobox genes and axial patterning. Cell 68, 283±302. Nishida, H., 1991. Induction of brain and sensory pigment cells in the ascidian embryo analyzed by experiments with isolated blastomeres. Dev. 112, 389±395. Olsen, C.L., Jeffery, W.R., 1997. A forkhead gene related to HNF-3b is required for gastrulation and axis formation in the ascidian embryo. Development 124, 3609±3619. Reverberi, G., Ortolani, G., Farinella-Ferruzza, N., 1960. The causal formation of the brain in the ascidian larva. Acta Embryol. Morphol. Exp. 3, 296±336. Sasaki, H., Hogan, B.L.M., 1994. HNF-3b as a regulator of ¯oor plate development. Cell 76, 103±115. Satoh, N., 1994. Developmental Biology of Ascidians. Cambridge University Press, New York. Satoh, N., Jeffery, W.R., 1995. Chasing tails in ascidians. Developmental insights into the origin and evolution of chordates. Trends Genet. 11, 354±359. Shimauchi, Y., Yasuo, H., Satoh, N., 1997. Autonomy of ascidian fork head/HNF-3 gene expression. Mech. Dev. 69, 143±154.
Swalla, B.J., Jeffery, W.R., 1990. Interspeci®c hybridization between an anural and urodele ascidian, differential expression of urodele features suggests multiple mechanisms control anural development. Dev. Biol. 142, 319±334. Swalla, B.J., Jeffery, W.R., 1996. Requirement of the Manx gene for restoration of chordate features in a tailless ascidian embryo. Science 274, 1205±1208. Swalla, B.J., Makabe, K.W., Satoh, N., Jeffery, W.R., 1993. Novel genes expressed differentially in ascidians with alternate modes of development. Development 119, 307±318. Swalla, B.J., Just, M.E., Pederson, E.L., Jeffery, W.R., 1999. A multigene locus containing the Manx and bobcat genes is required for development of chordate features in the ascidian tadpole larva. Development 126, 1643±1653. Weinstein, D.C., Ruiz i Altaba, A., Chen, W.S., Hoodless, P., Prezioso, V.R., Jessell, T.M., Darnell, J.E., 1994. The winged helix transcription factor HNF3b is required for notochord development in the mouse embryo. Cell 78, 575±588. Whittaker, J.R., 1979. Development of vestigial tail muscle acetylcholinesterase in embryos of an anural ascidian species. Biol. Bull. 156, 393± 407. Yasuo, H., Satoh, N., 1994. An ascidian homolog of the mouse Brachyury (T) gene is expressed exclusively in notochord cells at the fate restricted stage. Dev. Growth Diff. 36, 9±18.