Bone Morphogenetic Protein-4 (BMP-4) is a potent epidermal inducer and inhibitor of neural fate. We have used differential screening to identify genes involved in epidermal induction downstream of BMP-4 and report here evidence of a novel translational mechanism that regulates the division of the vertebrate ectoderm into regions of neural and epidermal fate. In dissociated Xenopus ectoderm, addition of ectopic BMP-4 leads to an increase in the expression of translation initiation factor 4AIII (eIF-4AIII), a divergent member of the eIF-4A gene family until now characterized only in plants. In the gastrula embryo XenopuseIF-4AIII (XeIF-4AIII) expression is elevated in the ventral ectoderm, a site of active BMP signal transduction. Moreover, overexpression of XeIF-4AIII induces epidermis in dissociated cells that would otherwise adopt a neural fate, mimicking the effects of BMP-4. Epidermal induction by XeIF-4AIII requires both an active BMP signaling pathway and an extracellular intermediate. Our results suggest that XeIF-4AIII can regulate changes in cell fate through selective mRNA translation. We propose that BMPs and XeIF-4AIII interact through a positive feedback loop in the ventral ectoderm of the vertebrate gastrula.

During vertebrate gastrulation, the cells of the dorsal ectoderm give rise to the central nervous system; ventral ectoderm differentiates into epidermis. Over the past two years, a large body of evidence has come to support a ‘default’ model of neural induction (see Weinstein and Hemmati-Brivanlou, 1997). This model argues that the differentiation of epidermis requires inductive signals, while the neuralization of the dorsal ectoderm requires only an inhibition of this signaling. Recent work from several labs suggests that BMP-4, a ligand of the Transforming Growth Factor-β (TGF-β? superfamily, is an endogenous neural inhibitor and epidermal inducer (Wilson and Hemmati-Brivanlou, 1995; Sasai et al., 1995; Hawley et al., 1995).

We have uncovered evidence of selective translational regulation in the BMP-mediated differentiation of the ectoderm into neural tissue and epidermis. In a number of studies, peptide growth factors have been shown to enhance translation rates through the modification of cytoplasmic proteins involved in translation initiation (Sonenberg, 1996). Interestingly, stimulation of the initiation machinery does not always result in a general increase in the rate of translation: various growth factor treatments can dramatically elevate the translation rates of specific mRNAs (Sonenberg, 1996). Enhanced activity of translation initiation factors may preferentially lead to the expression of mRNAs with a complex secondary structure in their 5 ′untranslated regions (UTR) (Sonenberg, 1996; Brown and Schreiber, 1996). This selective translation can have profound consequences for cell fate. For example, the overexpression of translation initiation factor eIF-4E in Xenopus embryos induces mesoderm in cells that would otherwise develop as epidermis (Klein and Melton, 1994). Ectopic eIF-4E preferentially elevates the translation of mRNA encoding activin, a mesoderm-inducing growth factor.

eIF-4E, along with eIF-4A and eIF-4G, forms the cap-binding complex eIF-4F. eIF-4F functions to unwind secondary structure in the 5 ′ UTR of mRNA to allow ribosome binding and thus initiate translation (Sonenberg, 1996). The helicase activity of eIF-4F is thought to be conferred by the eIF-4A subunit, in conjunction with eIF-4B (Rozen et al., 1990; Pause et al., 1994). Although other RNA helicases have been implicated in specific interactions with mRNA (Lasko and Ashburner, 1988; Hay et al., 1988; Liang et al., 1994), no role has thus far been demonstrated for the helicase component of the translation initiation machinery in binding specificity.

We report here the isolation of Xenopus translation initiation factor eIF-4AIII (XeIF-4AIII), a divergent member of the eIF-4A gene family whose expression, at gastrula stages, is elevated in the ventral ectoderm, a site of active BMP signaling. We demonstrate that XeIF-4AIII, like BMP-4, can induce epidermis in a dissociated ectoderm assay, without a general increase in protein translation. Our results suggest that this induction is mediated through a positive feedback loop, linking XeIF-4AIII and BMP, in the ventral ectoderm of the gastrula embryo.

Isolation of XeIF-4AIII

20 animal caps explants were dissociated and cultured for 4 hours in the presence or absence of 100 ng/ml BMP-4 (Genetics Institute), as described in Wilson and Hemmati-Brivanlou (1995). Parallel cultures were incubated until control embryos reached stage 19, and assayed for the expression of neural and epidermal markers, with results as expected (data not shown). RNA was extracted as previously described (Wilson and Hemmati-Brivanlou, 1995). One-twentieth of this RNA was used as source material for a subtractive screen. Library construction and screening was performed using a modified protocol originally described in Dulac and Axel (1995). Briefly, RNA was derived from cells cultured in the presence or absence of BMP-4. Reverse transcription and Polymerase Chain Reaction (PCR) was used to generate first-strand cDNAs, under conditions favoring uniformly sized molecules. cDNA libraries were constructed from the two pools and plated at 500 plaques/plate. Two radiolabelled probes were made from the remainder of the PCR reactions and used to screen one set of lifts from each library. A total of 5,000 plaques from each pool were screened. Ten clones were isolated that hybridized more strongly to radiolabelled probes generated from the BMP-4 pool. Two of ten cDNAs showed high sequence identity to distinct regions of Nicotiana plumbaginifolia eIF-4A3 (Owttrim et al., 1991).

106 plaques from a λ ZapII stage 28 Xenopus head cDNA library (Hemmati-Brivanlou et al., 1991) were screened at high stringency using radiolabelled probes generated from the two isolates described above and a partial clone was obtained (4A-3). PCR amplification of an oocyte cDNA library, using an internal 4A-3 oligo and an oligo derived from the Zap plasmid, was used to isolate the remaining 5 ′ sequences. cDNAs were sequenced by the dideoxy-chain termination method (Sanger et al., 1977). Sequence analysis was carried out using the DNA Strider and DNA Star software packages and the NIH BLAST program.

RT-PCR was performed as described in Wilson and Hemmati-Brivanlou (1995). Primers constructed for this study are as follows:

XeIF-4AIII:U:5′-GGGTGTTGCCATTAACTTTGTC-3′ D: 5′-CCTGCCACATGAAATCCTGG-3′

cSRC:U:5′-GAGTCGCAGTCTGGATATTGC-3′ D: 5′-TGGAATGTAGCCAGTCTGCC-3′

All other primer sequences are listed in Hemmati-Brivanlou and Melton (1994), Hemmati-Brivanlou et al. (1994) and Wilson and Hemmati-Brivanlou (1995). All reactions were performed for 25 cycles except for EF1-α muscle actin and XeIF-4AIII, which were performed for 21 cycles.

Quantitative phosphorimage analysis was performed with a Molecular Dynamics Phosphorimager, using the ImageQuant software package.

Whole-mount in situ hybridization

Whole-mount in situ hybridization was performed as described in Harland (1991). BM purple AP substrate (Boehringer Mannheim) replaced BCIP/NBT. Some embryos were cleared in 2:1 benzyl benzoate: benzyl alcohol (BB:BA). XeIF-4AIII antisense probes were generated using T7 RNA polymerase in the presence of rUTP-digoxigenin (Boehringer Mannheim Biochemicals).

RNA preparation, microinjection, explant dissection and cell culture

mRNA was synthesized in vitro in the presence of cap analog using the mMessage mMachine kit (Ambion). The coding regions of XeIF-4AIII and MeIF-4AI (Nielson et al., 1985) were subcloned into pCS2 (Rupp et al., 1994; Turner and Weintraub, 1994). The coding regions of XeIF-4E (Klein and Melton, 1994) and the tBR construct (Suzukiet al., 1994), were in pSP64T (Krieg and Melton, 1984). RNA from all constructs was synthesized using the Sp6 promoter. Microinjection, explant dissection and dissociation cultures were performed as described in Wilson and Hemmati-Brivanlou (1995).

For oocyte translation, 10 ng of RNA was injected into the marginal zone of stage VI oocytes. Oocytes were cultured, and protein was harvested and analyzed as described previously (Hemmati-Brivanlou et al., 1994).

Isolation of Xenopus eIF-4AIII

Ectodermal (‘animal cap’) explants form epidermis when cultured intact; these explants neuralize if subjected to prolonged dissociation (Grunz and Tacke, 1989; Godsave and Slack, 1989). Soluble BMP-4 can induce epidermis in dissociated ectoderm, substituting for the epidermal inducer presumably lost by dilution (Wilson and Hemmati-Brivanlou, 1995). To identify factors involved in epidermal induction down-stream of BMP-4, we dissociated late blastula animal cap explants in the absence or presence of 4 μm BMP-4 protein. Libraries were constructed from dissociated cells cultured with or without BMP-4, using a reverse-transcription PCR-based strategy (Dulac and Axel, 1995). Ten clones whose expression was elevated in the presence of BMP-4 were selected for sequence analysis. Two of the ten showed highest homology to distinct regions of Nicotiana plumbaginifolia (tobacco) eukaryotic translation Initiation Factor 4A3 (NeIF-4A3) (Owttrim et al., 1991). We used these partial cDNAs as probes in subsequent library screens to isolate a full-length cDNA. This clone shares higher identity with the tobacco gene than with other vertebrate 4A family members (Nielson et al., 1985; Nielson and Trachsel, 1988); therefore, we refer to it as XeIF-4AIII (Fig. 1, Table 1).

Table 1.

Percentage identity between XeIF-4AIII and selected members of the eIF-4A gene family

Percentage identity between XeIF-4AIII and selected members of the eIF-4A gene family
Percentage identity between XeIF-4AIII and selected members of the eIF-4A gene family
Fig. 1.

Comparison of the deduced amino acid sequences of XeIF-4AIII (GenBank accession no., AF020432) and NeIF-4A3. Conserved amino acids are shown in black boxes.

Fig. 1.

Comparison of the deduced amino acid sequences of XeIF-4AIII (GenBank accession no., AF020432) and NeIF-4A3. Conserved amino acids are shown in black boxes.

In order to confirm that BMP-4 enhances XeIF-4AIII expression in dispersed ectoderm, we performed additional dissociation experiments and assayed for XeIF-4AIII expression by RT-PCR. Phosphorimaging analysis indicated that XeIF-4AIII transcripts were enriched approximately 3-fold in cells dissociated in the presence of BMP-4. (data not shown).

Distribution of XeIF-4AIII in Xenopus embryos

To analyze the expression of XeIF-4AIII during the period in which ectoderm is competent to differentiate into either neural tissue or epidermis, RT-PCR analysis was performed on embryos harvested between blastula and neural plate stages (Fig. 2). While expression can be detected at all stages assayed, XeIF-4AIII expression increases dramatically after the mid-blastula transition (MBT, stage 9.5), the initiation of zygotic transcription.

Fig. 2.

Temporal expression of XeIF-4AIII. RT-PCR analysis of XeIF-4AIII expression during development. cSRC and EF1-α are used as loading controls. The ‘RT’ lane contains all reagents except reverse transcriptase and was used as a negative control.

Fig. 2.

Temporal expression of XeIF-4AIII. RT-PCR analysis of XeIF-4AIII expression during development. cSRC and EF1-α are used as loading controls. The ‘RT’ lane contains all reagents except reverse transcriptase and was used as a negative control.

Whole-mount in situ hybridization was used to further study the expression of XeIF-4AIII at several stages of Xenopus development. At blastula and gastrula stages, XeIF-4AIII is expressed throughout the animal pole and marginal zone (data not shown). RT-PCR analysis revealed that XeIF-4AIII expression was enriched in the ventral versus dorsal animal pole at gastrula stages (see below). By early neural plate stages, XeIF-4AIII expression begins to show a more restricted pattern of expression. Transcript is relatively abundant around the remnant of the blastopore and is also enriched in the anterior neural plate (Fig. 3A). While the posterior domain of XeIF-4AIII is transient, anterior neural expression is maintained following closure of the neural tube (Fig. 3B). XeIF-4AIII expression persists in anterior neural structures throughout neurula stages. At late neurula stages, staining can be seen throughout the developing brain, to the approximate caudal boundary of the otic vesicle (Fig. 3C-F). Diffuse expression is also observed throughout the craniofacial structures. Finally, we observe a curious, mid-trunk domain of XeIF-4AIII expression during early tailbud stages (Fig. 3C,D). This expression is first visible at stages 23-24, and is gone by stage 27. In section, this stain appears to be concentrated in lateral regions of the somites (data not shown).

Fig. 3.

Localization of XeIF-4AIII transcripts during Xenopus development. (A) Dorsal view of a neural plate stage embryo (stage 14); anterior is to right. XeIF-4AIII transcript is enriched in a region around the blastopore (arrowhead) and in the anterior neural plate (arrow). (B) Dorsal view of a neural plate stage embryo (stage 16); anterior is to right. High levels of XeIF-4AIII transcript can be seen in the anterior neural tube (arrow). (C) Lateral view of a stage 23 embryo; anterior is at right, dorsal is at top. (D) Dorsal view of embryo shown in C. Line in C and D demarcates region of mid-trunk eIF-4AIII expression. (E) Lateral view of a stage 27 embryo; anterior is at right, dorsal is at top. (F) Detail showing head of embryo in E. ov: otic vesicle. Embryos in C-F were cleared in 2:1 benzyl benzoate/benzyl alcohol.

Fig. 3.

Localization of XeIF-4AIII transcripts during Xenopus development. (A) Dorsal view of a neural plate stage embryo (stage 14); anterior is to right. XeIF-4AIII transcript is enriched in a region around the blastopore (arrowhead) and in the anterior neural plate (arrow). (B) Dorsal view of a neural plate stage embryo (stage 16); anterior is to right. High levels of XeIF-4AIII transcript can be seen in the anterior neural tube (arrow). (C) Lateral view of a stage 23 embryo; anterior is at right, dorsal is at top. (D) Dorsal view of embryo shown in C. Line in C and D demarcates region of mid-trunk eIF-4AIII expression. (E) Lateral view of a stage 27 embryo; anterior is at right, dorsal is at top. (F) Detail showing head of embryo in E. ov: otic vesicle. Embryos in C-F were cleared in 2:1 benzyl benzoate/benzyl alcohol.

XeIF-4AIII expression is elevated in the ventral ectoderm

In the ectoderm of the Xenopus gastrula, BMP signaling, and consequently epidermal induction, is thought to occur on the ventral side only; in the dorsal ectoderm, BMP signaling is inhibited by secreted factors from Spemann’s organizer (see Weinstein and Hemmati-Brivanlou, 1997). In order to more precisely examine the localization of XeIF-4AIII expression in the ectoderm during gastrula stages, we isolated midgastrula dorsal and ventral ectoderm explants (Fig. 4A), and assayed for XeIF-4AIII expression by RT-PCR (Fig. 4B). Epidermal keratin, a marker of epidermis, was detected in the ventral explants only, as expected (Jonas et al., 1985). While XeIF-4AIII is expressed throughout the midgastrula ectoderm, phosphorimaging data indicated that XeIF-4AIII transcripts are approximately 2-fold more abundant in the ventral explants (Fig. 4B and data not shown). This result is consistent with our data from dissociated cells: active BMP signaling correlates with elevated levels of XeIF-4AIII transcript, both in cell culture and in vivo.

Fig. 4.

XeIF-4AIII expression is elevated in the ventral versus dorsal ectoderm. (A) Schematic diagram of stage 11.5 embryo, showing regions isolated for expression analysis in B. (B) RT-PCR analysis of XeIF-4AIII expression in midgastrula dorsal and ventral ectoderm explants (stage 11.5). Epidermal keratin expression is restricted to the ventral ectoderm; Xbra, a pan-mesodermal marker at this stage, is absent from the explants. XeIF-4AIII levels in the dorsal and ventral ectoderm were obtained by quantitative phosphorimaging, with normalization to EF1- levels (Krieg et al., 1989).

Fig. 4.

XeIF-4AIII expression is elevated in the ventral versus dorsal ectoderm. (A) Schematic diagram of stage 11.5 embryo, showing regions isolated for expression analysis in B. (B) RT-PCR analysis of XeIF-4AIII expression in midgastrula dorsal and ventral ectoderm explants (stage 11.5). Epidermal keratin expression is restricted to the ventral ectoderm; Xbra, a pan-mesodermal marker at this stage, is absent from the explants. XeIF-4AIII levels in the dorsal and ventral ectoderm were obtained by quantitative phosphorimaging, with normalization to EF1- levels (Krieg et al., 1989).

XeIF-4AIII can induce epidermis in dissociated ectoderm

In order to examine the function of XeIF-4AIII, dissociated cell assays were performed on embryos injected with synthetic XeIF-4AIII RNA at early cleavage stages. Cells were reaggregated at midgastrula stages and cultured until midneurula stages. RNA was extracted and assayed for the expression of cell-type-specific molecular markers by RT-PCR. Cells from control embryos dissociated in this manner show strong expression of the general neural marker NCAM, and low expression of epidermal keratin, indicating that these cells have neuralized (Fig. 5A, lane 3) (Jonas et al., 1985; Kintner and Melton, 1987). Cells dissociated in the presence of BMP-4 express high levels of epidermal keratin, and low levels of NCAM (Fig. 5A, lane 4), indicating their differentiation into epidermis. Like cells treated with BMP-4, cells that express ectopic XeIF-4AIII do not express NCAM and express high levels of epidermal keratin (Fig. 5A, lane 1). Thus, XeIF-4AIII can induce epidermis in dissociated cells and inhibit the neural fate. Mesoderm is not present or induced in these cultures: neither muscle actin, a marker of dorsolateral mesoderm, nor Xbra, a marker of both notochord and ventral/posterior mesoderm at this stage, are expressed (Mohun et al., 1984; Smith et al., 1991).

Interestingly, another component of eIF-4F, unrelated to eIF-4A, has also been shown to affect cell fate in Xenopus ectoderm: eIF-4E overexpression induces mesoderm in intact caps (Klein and Melton, 1994; Fig. 5B, lane 2). Thus, it was important to determine if epidermal induction could also be mediated by other members of the translation initiation complex. This does not appear to be the case: our results indicate that the inductive capacities of XeIF-4AIII and XeIF-4E are distinct. XeIF-4E does not induce epidermal keratin in dissociated cells, (Fig. 5A, lane 2), while XeIF-4AIII does not induce mesoderm in intact animal caps (Fig. 5B, lane 1).

To further assess the specificity of XeIF-4AIII, we examined the inductive ability of a related eIF-4A molecule in the dis-sociated ectoderm assay. Mouse eIF-4AI (MeIF-4AI) (Nielson et al., 1985) and XeIF-4AIII were subcloned into the same expression vector, and produced similar levels of protein in a rabbit reticulocyte lysate translation system (data not shown). MeIF-4AI, however, did not induce epidermis (Fig. 5C, lane 2), nor did it block induction by XeIF-4AIII when injected at equimolar amounts (Fig. 5C, lane 3).

eIF-4A has been extensively characterized as a component of the general translation initiation machinery. To determine if XeIF-4AIII overexpression results in a general increase in translation rates, we injected stage VI oocytes with XeIF-4AIII and cultured the cells in the presence of [35S]methionine (Fig. 5D). Exogenous XeIF-4AIII can be seen as a dark band (of approximately 50 kDa) in the cell fraction from the injected cells. Overexpression of XeIF-4AIII did not lead to an increase in the overall levels of protein synthesis; moreover, no increase was seen in the levels of any specific protein band.

Epidermal induction by XeIF-4AIII requires extracellular signaling

Previous work has suggested that enhanced activity of translation initiation factors preferentially leads to the translation of mRNAs with a complex secondary structure in their 5 ′untranslated regions (UTR) (see Sonenberg, 1996; Brown and Schreiber, 1996); thus, XeIF-4AIII could mediate the preferential translation of a factor in the epidermal induction pathway. As a first step towards identification of RNA targets for XeIF-4AIII, we performed dissociation experiments with and without subsequent reaggregation. If ectopic XeIF-4AIII mediates the increased production of an intracellular protein, the resulting epidermal induction should not be dependent on reaggregation. If, however, XeIF-4AIII overexpression leads to the increase of a secreted protein, reaggregation might be required to avoid diluting this factor to non-functional levels. In dissociated cultures harvested without reaggregation, ectopic XeIF-4AIII did not induce epidermis or inhibit the neural fate (Fig. 6, lanes 1-8). Neither neuralization nor epidermal induction, per se, requires cell-cell contact: epidermal induction by BMP-4 protein is not dependent on reaggregation (Fig. 6, compare lanes 6,14), while dissociated cells do not require reaggregation to express neural markers (Fig. 6, compare lanes 5, 13). These results suggest that XeIF-4AIII overexpression induces epidermis through an extracellular intermediate.

Fig. 5.

XeIF-4AIII inhibits neuralization and induces epidermis. (A) Induction of epidermis by XeIF-4AIII, but not XeIF-4E, in dissociated ectoderm cultures. Synthetic RNA, as listed, was injected into both blastomeres of 2-cell-stage embryos; 2 ng mRNA was injected into each blastomere. Animal caps were dissected at late blastula stages (stage 9), dissociated for 4 hours, then reaggregated and cultured until midneurula stages (stage 20). BMP-4 protein was added to the sample in lane 4 at 2 μM (50 ng/ml) immediately after dissociation. EF1-α is used as a loading control. NCAM is a pan-neural marker and epidermal keratin is a marker of epidermis. Xbra is expressed in both the notochord and ventral/posterior mesoderm at this stage. muscle actin is a marker of mediolateral mesoderm. The ‘—RT’ lane contains all reagents except reverse transcriptase and was used as a negative control. (B) XeIF-4E, but not XeIF-4AIII, induces mesoderm in intact ectoderm. Synthetic RNA, as listed, was injected as in A. Animal caps were dissected at late blastula stages (stage 9) and cultured intact until midgastrula stages (stage 11.5). Xbra is a marker of both notochord and of all noninvoluted mesoderm at this stage. (C) Induction of epidermis by XeIF-4AIII, but not MeIF-4AI, in dissociated ectoderm cultures. RNA injection, cell culture, markers and controls are as in A. For single RNAs, 1 ng was injected into each blastomere; for double RNAs, 2 ng was injected into each blastomere. (D) General translation rates are not affected by XeIF-4AIII overexpression. 10 ng XeIF-4AIII mRNA was injected into stage VI oocytes and cultured in the presence of [35S]methionine for 3 days. Oocytes were lysed and this cell fraction was analyzed on 10% acrylimide gels by SDS-PAGE. All samples were treated with β-mercaptoethanol.

Fig. 5.

XeIF-4AIII inhibits neuralization and induces epidermis. (A) Induction of epidermis by XeIF-4AIII, but not XeIF-4E, in dissociated ectoderm cultures. Synthetic RNA, as listed, was injected into both blastomeres of 2-cell-stage embryos; 2 ng mRNA was injected into each blastomere. Animal caps were dissected at late blastula stages (stage 9), dissociated for 4 hours, then reaggregated and cultured until midneurula stages (stage 20). BMP-4 protein was added to the sample in lane 4 at 2 μM (50 ng/ml) immediately after dissociation. EF1-α is used as a loading control. NCAM is a pan-neural marker and epidermal keratin is a marker of epidermis. Xbra is expressed in both the notochord and ventral/posterior mesoderm at this stage. muscle actin is a marker of mediolateral mesoderm. The ‘—RT’ lane contains all reagents except reverse transcriptase and was used as a negative control. (B) XeIF-4E, but not XeIF-4AIII, induces mesoderm in intact ectoderm. Synthetic RNA, as listed, was injected as in A. Animal caps were dissected at late blastula stages (stage 9) and cultured intact until midgastrula stages (stage 11.5). Xbra is a marker of both notochord and of all noninvoluted mesoderm at this stage. (C) Induction of epidermis by XeIF-4AIII, but not MeIF-4AI, in dissociated ectoderm cultures. RNA injection, cell culture, markers and controls are as in A. For single RNAs, 1 ng was injected into each blastomere; for double RNAs, 2 ng was injected into each blastomere. (D) General translation rates are not affected by XeIF-4AIII overexpression. 10 ng XeIF-4AIII mRNA was injected into stage VI oocytes and cultured in the presence of [35S]methionine for 3 days. Oocytes were lysed and this cell fraction was analyzed on 10% acrylimide gels by SDS-PAGE. All samples were treated with β-mercaptoethanol.

Fig. 6.

Epidermal induction by XeIF-4AIII acts though an extracellular intermediate. Epidermal induction by XeIF-4AIII requires reaggregation. Synthetic RNA, as listed, was injected into both blastomeres of 2-cell-stage embryos. Animal caps were dissected at late blastula stages (stage 9) and dissociated. Left panel: cells were dissociated for 6 hours to late gastrula stages (stage 12.5), at which point RNA was harvested. Cultures maintained as dissociated until midneurula stages (stage 20) have the same expression profile as cultures harvested at late gastrula stages (data not shown). Right panel: cells were dissociated for 4 hours, then reaggregated and cultured until midneurula stages. XeIF-4AIII induces epidermal keratin and inhibits NCAM expression at RNA concentrations down to 1 ng (lanes 9-16). Controls are as listed in Fig. 5A.

Fig. 6.

Epidermal induction by XeIF-4AIII acts though an extracellular intermediate. Epidermal induction by XeIF-4AIII requires reaggregation. Synthetic RNA, as listed, was injected into both blastomeres of 2-cell-stage embryos. Animal caps were dissected at late blastula stages (stage 9) and dissociated. Left panel: cells were dissociated for 6 hours to late gastrula stages (stage 12.5), at which point RNA was harvested. Cultures maintained as dissociated until midneurula stages (stage 20) have the same expression profile as cultures harvested at late gastrula stages (data not shown). Right panel: cells were dissociated for 4 hours, then reaggregated and cultured until midneurula stages. XeIF-4AIII induces epidermal keratin and inhibits NCAM expression at RNA concentrations down to 1 ng (lanes 9-16). Controls are as listed in Fig. 5A.

Epidermal induction by XeIF-4AIII requires signaling through the BMP receptor

Our data suggest that XeIF-4AIII may induce epidermis through the preferential translation of an mRNA encoding an extracellular factor. BMP-4 has been described as a soluble epidermal inducer (Wilson and Hemmati-Brivanlou, 1995); thus, BMP-4 mRNA is a candidate target for preferential translation by XeIF-4AIII. To address this possibility, we co-injected XeIF-4AIII RNA with RNA for a truncated type I BMP receptor (tBR), shown to inhibit signaling of BMP-4 and BMP-2 (Graff et al., 1994; Suzuki et al., 1994). Overexpression of tBR neuralizes intact caps (Xu et al., 1995; Sasai et al., 1995; Suzuki et al., 1995). Dissociated cells expressing tBR are indistinguishable from uninjected controls (Fig. 7, compare lanes 2, 4), but are unresponsive to addition of BMP-4 protein, again expressing only neural, and not epidermal, markers (Fig. 7, compare lanes 5, 6). Dissociated cells from animal caps injected with XeIF-4AIII express epidermal keratin (Fig. 7, lane 1). Dissociated cells that co-express XeIF-4AIII and tBR neuralize (Fig. 7, lane 3), indicating that overexpression of tBR blocks XeIF-4AIII-mediated epidermal induction. This result suggests that XeIF-4AIII requires active signaling through the wild-type BMP receptor to induce epidermis in dissociated cells.

Fig. 7.

Epidermal induction by XeIF-4AIII requires an active BMP signaling pathway. Co-injection of a truncated BMP receptor (tBR) blocks epidermal induction by XeIF-4AIII. Synthetic RNA, as listed, was injected into both blastomeres of 2-cell-stage embryos. 1 ng of each RNA was injected per blastomere. Animal caps were dissected at late blastula stages (stage 9) and dissociated. Cells were dissociated for 4 hours, then reaggregated and cultured until midneurula stages.

Fig. 7.

Epidermal induction by XeIF-4AIII requires an active BMP signaling pathway. Co-injection of a truncated BMP receptor (tBR) blocks epidermal induction by XeIF-4AIII. Synthetic RNA, as listed, was injected into both blastomeres of 2-cell-stage embryos. 1 ng of each RNA was injected per blastomere. Animal caps were dissected at late blastula stages (stage 9) and dissociated. Cells were dissociated for 4 hours, then reaggregated and cultured until midneurula stages.

In an attempt to identify factors involved in the epidermal induction pathway, we have isolated the translation initiation factor XeIF-4AIII. XeIF-4AIII transcripts are elevated in dissociated cells cultured in the presence of BMP-4, a soluble epidermal inducer; furthermore, XeIF-4AIII transcripts are elevated, albeit modestly, in the ventral ectoderm of the gastrula, the region fated to give rise to epidermis and a site of active BMP signaling. Ectopic XeIF-4AIII acts as an epidermal inducer and neural inhibitor in dissociated ectoderm via an extracellular intermediate. This induction occurs without a Our results imply a positive feedback loop between a soluble epidermal inducer and XeIF-4AIII (Fig. 8). Ectopic XeIF-4AIII selectively mediates the production of a secreted factor in the epidermal induction pathway that acts, directly or in-directly, upstream of the BMP receptor. This factor may be BMP-4 or a related protein. Conversely, ectopic BMP-4 stimulates expression of XeIF-4AIII. The newly synthesized XeIF-4AIII protein may differ from preexisting molecules in the type or levels of post-transcriptional modification, or in the degree of association with putative, inhibitory 4A-binding proteins, as have been described for eIF-4E (Sonenberg, 1996). In maize, eIF-4A phosphorylation occurs during oxygen starvation and correlates with translational suppression (Webster et al., 1991). The proposed autoregulatory loop offers a means by which XeIF-4AIII could enhance the signal for epidermal induction. Epidermal induction is inhibited in the dorsal ectoderm by direct binding of neuralizing factors in the dorsal ectoderm to BMPs (see Weinstein and Hemmati-Brivanlou, 1997). The loop would only be maintained, then, in the ventral ectoderm, sharpening the boundary between the neurogenic ectoderm and the future epidermis.

Fig. 8.

Model for a positive feedback loop involving BMP and XeIF-4AIII. In the ventral ectoderm of the Xenopus gastrula, binding of BMP-4 (diamond) to its receptor (Y) leads to an increase in the expression of XeIF-4AIII RNA (nuclear wavy line), which presumably results in an increase in cytoplasmic XeIF-4AIII protein (oval). Elevated levels of XeIF-4AIII protein may mediate the increased translation of specific mRNAs (cytoplasmic wavy line). In this model, the resulting protein products (coil) drive the production of a secreted epidermal inducer (diamond), either directly, or indirectly via one or more molecular intermediates. This factor requires active signaling through the BMP receptor and may be BMP-4 or a related protein. n, nucleus; c, cytoplasm. concomitant general increase in protein synthesis and requires an active BMP signaling pathway.

Fig. 8.

Model for a positive feedback loop involving BMP and XeIF-4AIII. In the ventral ectoderm of the Xenopus gastrula, binding of BMP-4 (diamond) to its receptor (Y) leads to an increase in the expression of XeIF-4AIII RNA (nuclear wavy line), which presumably results in an increase in cytoplasmic XeIF-4AIII protein (oval). Elevated levels of XeIF-4AIII protein may mediate the increased translation of specific mRNAs (cytoplasmic wavy line). In this model, the resulting protein products (coil) drive the production of a secreted epidermal inducer (diamond), either directly, or indirectly via one or more molecular intermediates. This factor requires active signaling through the BMP receptor and may be BMP-4 or a related protein. n, nucleus; c, cytoplasm. concomitant general increase in protein synthesis and requires an active BMP signaling pathway.

XeIF-4AIII induces epidermis via a soluble intermediate that signals through the endogenous BMP receptor. There are, at present, several candidates for this factor. The wild-type BMP receptor has been shown to bind both BMP-2 and BMP-4, and may bind other, related factors as well (Graff et al., 1994; Suzuki et al., 1994). BMP-4 induces epidermis in dissociated ectoderm (Wilson and Hemmati-Brivanlou, 1995; this study). BMP-2 and BMP-7 are also active in this assay (Suzuki et al., 1997). BMP-4 RNA is excluded from the cells of the dorsal blastopore lip and a small region of neighboring dorsal ectoderm, but is strongly expressed in the remaining gastrula ectoderm (Fainsod et al., 1994; Hemmati-Brivanlou and Thomsen, 1995; Schmidt et al., 1995). Both BMP-2 and BMP-7 are expressed throughout the animal pole at blastula stages, and throughout the ectoderm and marginal zone at gastrula stages (Hemmati-Brivanlou and Thomsen, 1995; Hawley et al., 1995). Thus, the expression patterns and activities of BMP-2, BMP-4 and BMP-7 are consistent with a role for any of these factors, or a combination of them, in eIF-4AIII-mediated epidermal induction.

Our data suggest that XeIF-4AIII is involved in selective translation. A similar role has been proposed for the vasa gene product in the translation of nanos RNA during Drosophila development (see Curtis et al., 1995). vasa shows homology to eIF-4A and has RNA helicase activity in vitro (Lasko and Ashburner, 1988; Hay et al., 1988; Liang et al., 1994). Other initiation factors, unrelated to eIF-4A, have also been shown to translate a subset of RNAs when overexpressed. For example, ectopic eIF-4E selectively increases the translation of activin mRNA in a Xenopus oocyte system (Klein and Melton, 1994). In similar assays, we failed to detect any increase in translation of either BMP-2 or BMP-4 mRNA when co-injected with XeIF-4AIII (data not shown). There are several simple explanations for these results: (1) our BMP constructs may lack the complete 5′ UTR with which the translation machinery interacts in vivo; (2) XeIF-4AIII may act by increasing translation of BMP-7 or other, as yet uncharacterized, epidermal inducers or (3) XeIF-4AIII may increase the production of the soluble inducer indirectly, employing molecular intermediates not present in the oocyte.

This is the first report of an animal homolog of eIF-4AIII. Two related molecules, eIF-4AI and eIF-4AII, have been reported in mammals (Nielson et al., 1985; Nielson and Trachsel, 1988) and, recently, in Xenopus (Morgan and Sargent, 1997). The expression patterns of the two mammalian genes are distinct: the level of eIF-4AII transcript varies widely among tissues, while eIF-4AI expression is relatively uniform in all tissues examined (Nielson and Trachsel, 1988). In Xenopus, we find that eIF-4AIII is not ubiquitous, but exhibits a dynamic pattern of expression during development. Xenopus eIF-4AI (XeIF-4AI) is expressed at all stages and in all tissues examined, while XeIF-4AII is expressed at high levels from stage 11.5 (Morgan and Sargent, 1997). Interestingly, XeIF-4AII expression at midgastrula stages is enriched in the dorsal versus ventral ectoderm, the converse of what we observe for XeIF-4AIII (Morgan and Sargent, 1997; this study). More surprising, Morgan and Sargent demonstrate that both eIF-4AII and eIF-4AI can induce markers of the neural plate border region in intact ectodermal explants. Cells in this region give rise to tissues that include neural crest and cement gland, fates that can be considered intermediate between neural and epidermal along the dorsoventral axis. Thus, eIF-4AII, enriched in the dorsal gastrula ectoderm, can induce dorsal fates, while eIF-4AIII is enriched in the ventral gastrula ectoderm and can induce ventral fate (epidermis) (Morgan and Sargent, 1997; this study). These results clearly point to an important role for the eIF-4A gene family in the early patterning of the vertebrate ectoderm.

Our results, together with those of Morgan and Sargent, indicate that members of the eIF-4A gene family can trigger specific developmental responses. In addition, these data suggest that different eIF-4A family members have different activities. How might related eIF-4A molecules confer distinct biological activities? eIF-4AI and 4AII share high sequence identity, and are incorporated into the eIF-4F complex with similar kinetics (Nielson and Trachsel, 1988; Yoder-Hill et al., 1993). XeIF-4AIII is clearly more divergent from 4AI and 4AII at the sequence level than the two are from each other; differences in incorporation kinetics and/or helicase activity between eIF-4AIII and other family members are intriguing possibilities. Distinct activities and expression domains among related eIF-4A genes suggest an important and previously unsuspected role for these factors in the specific regulation of gene expression. Further progress will require a combination of bio-chemical and embryological approaches.

We wish to thank P. Klein, N. Sonenberg and A. Suzuki for providing us with the XeIF-4E, MeIF-4AI and tBR cDNA clones, respectively. We would also like to thank A. Suzuki for help with the microdissection schematic, J. Marden for excellent technical assistance, and P. Wilson and G. Lagna for critical reading of the manuscript. Recombinant human BMP-4 was a gift of Genetics Institute. A. H. B. is a Searle and a McKnight scholar. This work was supported by NIH grant HD 32105-01 and by grants from the Klingenstein and Merck Foundations to A. H. B.

Brown
,
E. J.
and
Schreiber
,
S. L.
(
1996
).
A signaling pathway to translational control
.
Cell
86
,
517
520
.
Curtis
,
D.
,
Lehmann
,
R.
and
Zamore
,
P. D.
(
1995
).
Translational regulation in development
.
Cell
81
,
171
178
.
Dulac
,
C.
and
Axel
,
R.
(
1995
).
A novel family of genes encoding putative pheromone receptors in mammals
.
Cell
83
,
195
206
.
Fainsod
,
A.
,
Steinbeisser
,
H.
and
DeRobertis
,
E. M.
(
1994
).
On the function of BMP-4 in patterning the marginal zone of the Xenopus embryo
.
EMBO J
.
13
,
5015
5025
.
Godsave
,
S. F.
and
Slack
,
J. M. W.
(
1989
).
Clonal analysis of mesoderm induction in Xenopus laevis
.
Dev. Biol
.
134
,
486
490
.
Graff
,
J. M.
,
Thies
,
R. S.
,
Song
,
J. J.
,
Celeste
,
A. J.
and
Melton
,
D. A.
(
1994
).
Studies with a Xenopus BMP receptor suggest that ventral mesoderm-inducing signals override dorsal signals in vivo
.
Cell
79
,
169
179
.
Grunz
,
H.
and
Tacke
,
L.
(
1989
).
Neural differentiation of Xenopus laevis ectoderm takes place after disaggregation and delayed reaggregation without inducer
.
Cell Diff. Devl
.
28
,
211
218
.
Harland
,
R. M.
(
1991
).
In situ hybridization: an improved whole mount method for Xenopus embryos
.
Meth. Cell Biol
.
36
,
675
685
.
Hawley
,
S. H. B.
,
Wunnenberg-Stapleton
,
K.
,
Hashimoto
,
C.
,
Laurent
,
M. N.
,
Watabe
,
T.
,
Blumberg
,
B. W.
and
Cho
,
K. W. Y.
(
1995
).
Disruption of BMP signals in embryonic Xenopus ectoderm leads to direct neural induction
.
Genes Dev
.
9
,
2923
2935
.
Hay
,
B.
,
Jan
,
L. Y.
and
Jan
,
Y. N.
(
1988
).
A protein component of Drosophila polar granules is encoded by vasa and has extensive sequence similarity to ATP-dependent helicases
.
Cell
55
,
577
587
.
Hemmati-Brivanlou
,
A.
,
de la Torre
,
J. R.
,
Holt
,
C.
and
Harland
,
R. M.
(
1991
).
Cephalic expression and molecular characterization of Xenopus En-2
.
Development
111
,
715
724
.
Hemmati-Brivanlou
,
A.
,
Kelly
,
O. G.
and
Melton
,
D. A.
(
1994
).
tFollistatin, an antagonist of activin, is expressed in the Spemann organizer and displays direct neuralizing activity
.
Cell
77
,
283
295
.
Hemmati-Brivanlou
,
A.
and
Melton
,
D. A.
(
1994
).
Inhibition of activin receptor signaling promotes neuralization in Xenopus
.
Cell
77
,
273
281
.
Hemmati-Brivanlou
,
A.
and
Thomsen
,
G. A.
(
1995
).
Ventral mesodermal patterning in Xenopus embryos: expression patterns and activities of BMP-2 and BMP-4
.
Dev. Genetics
17
,
78
89
.
Jonas
,
E.
,
Sargent
,
T. D.
and
Dawid
,
I. B.
(
1985
).
Epidermal keratin expressed in embryos of Xenopus laevis
.
Proc. Natl. Acad. Sci. USA
82
,
5413
5417
.
Kintner
,
C. R.
and
Melton
,
D. A.
(
1987
).
Expression of Xenopus NCAM RNA in ectoderm is an early response to neural induction
.
Development
99
,
311
325
.
Klein
,
P. S.
and
Melton
,
D. A.
(
1994
).
Induction of mesoderm in Xenopus laevis by translation initiation factor 4E
.
Science
265
,
803
806
.
Krieg
,
P.
,
Varnum
,
S.
,
Wormington
,
M.
and
Melton
,
D. A.
(
1989
).
The mRNA encoding elongation factor 1a (EF-1a) is a major transcript at the mid-blastula transition in Xenopus
.
Dev. Biol
.
133
,
93
100
.
Lasko
,
P. F.
and
Ashburner
,
M.
(
1988
).
The product of the Drosophila gene vasa is very similar to eukaryotic initiation factor-4A
.
Nature
335
,
611
617
.
Liang
,
L.
,
Diehl-Jones
,
W.
and
Lasko
,
P.
(
1994
).
Localization of vasa protein to the Drosophila pole plasm is independent of its RNA-binding and helicase activities
.
Development
120
,
1201
1211
.
Mohun
,
T. J.
,
Brennan
,
S.
,
Dathan
,
N.
,
Fairman
,
S.
and
Gurdon
,
J. B.
(
1984
).
Cell type-specific activation of actin genes in the early amphibian embryo
.
Nature
311
,
716
721
.
Morgan
,
R.
and
Sargent
,
M. G.
(
1997
).
The role in neural patterning of translation initiation factor eIF-4AII; induction of neural fold genes
.
Development
124
,
2751
2760
.
Nielson
,
P. J.
,
McMaster
,
G. K.
and
Trachsel
,
H.
(
1985
).
Cloning of eukaryotic protein synthesis initiation factor genes: isolation and characterization of cDNA clones encoding factor eIF-4A
.
Nucl. Acids Res
.
13
,
6867
6880
.
Nielson
,
P. J.
and
Trachsel
,
H.
(
1988
).
The mouse protein synthesis initiation factor 4A gene family includes two related functional genes which are differentially expressed
.
EMBO J
.
7
,
2097
2105
.
Owttrim
,
G. W.
,
Hofmann
,
S.
and
Kuhlemeier
,
C.
(
1991
).
Divergent genes for translation initiation factor eIF-4A are coordinately expressed in tobacco
.
Nucl. Acids Res
.
19
,
5491
5496
.
Pause
,
A.
,
Methot
,
N.
,
Svitkin
,
Y.
,
Merrick
,
W. C.
and
Sonenberg
,
N.
(
1994
).
Dominant negative mutants of mammalian initiation factor eIF-4A define a critical role for eIF-4F in cap-dependent and cap-independent initiation of translation
.
EMBO J
.
13
,
1205
1215
.
Rozen
,
F.
,
Edery
,
I.
,
Meerovitch
,
K.
,
Dever
,
T. E.
,
Merrick
,
W. C.
and
Sonenberg
,
N.
(
1990
).
Bidirectional RNA helicase activity of eucaryotic translation initiation factors 4A and 4F
.
Mol. Cell. Biol
.
10
,
1134
1144
.
Rupp
,
R. A. W.
,
Snider
,
L.
,
Weintraub
,
H.
(
1994
).
Xenopus embryos regulate the nuclear localization of XMyoD
.
Genes Dev
.
8
,
1311
1323
.
Sanger
,
F.
,
Nicklen
,
S.
and
Coulson
,
A. R.
(
1977
).
DNA sequencing with chain-terminating inhibitors
.
Proc. Natl. Acad. Sci. USA
74
,
5463
5467
.
Sasai
,
Y.
,
Lu
,
B.
,
Steinbeisser
,
H.
and
DeRobertis
,
E. M.
(
1995
).
Regulation of neural induction by the Chd and BMP-4 antagonistic patterning signals in Xenopus
.
Nature
376
,
333
336
.
Schmidt
,
J. E.
,
Suzuki
,
A.
,
Ueno
,
N.
and
Kimelman
,
D.
(
1995
).
Localized BMP-4 mediates dorsal/ventral patterning in the early Xenopus embryo
.
Dev. Biol
.
169
,
37
50
.
Smith
,
J. C.
,
Price
,
B. M. J.
,
Green
,
J. B. A.
,
Weigel
,
D.
and
Herrmann
,
B. G.
(
1991
).
Expression of a Xenopus homolog of Brachyury (T) is an immediate-early response to mesoderm induction
.
Cell
67
,
79
87
.
Sonenberg
,
N.
(
1996
).
mRNA 5 ′cap-binding protein eIF-4E and control of cell growth
.
In Translational Control
. (ed.
J. W. B.
Hershey
,
M. B.
Mathews
and
N.
Sonenberg
). pp.
245
269
.
New York
:
Cold Spring Harbor Press
.
Suzuki
,
A.
,
Kaneko
,
E.
,
Ueno
,
N.
and
Hemmati-Brivanlou
,
A.
(
1997
).
Regulation of epidermal induction by BMP2 and BMP7 signaling
.
Dev. Biol
. (in press).
Suzuki
,
A.
,
Shioda
,
N.
and
Ueno
,
N.
(
1995
).
Bone morphogenetic protein acts as a ventral mesoderm modifier in early Xenopus embryos
.
Dev. Growth Differ
.
37
,
581
588
.
Suzuki
,
A.
,
Thies
,
R. S.
,
Yamaji
,
N.
,
Song
,
J. J.
,
Wozney
,
J. M.
,
Murakami
,
K.
and
Ueno
,
N.
(
1994
).
A truncated bone morphogenetic protein receptor affects dorsal-ventral patterning in early Xenopus embryos
.
Proc. Natl. Acad. Sci. USA
91
,
10255
10259
.
Turner
,
D. L.
and
Weintraub
,
H.
(
1994
).
Expression of achaete-scute homolog 3 in Xenopus embryos converts ectodermal cells to a neural fate
.
Genes Dev
.
8
,
1434
1447
.
Webster
,
C.
,
Gaut
,
R. L.
,
Browning
,
K. S.
,
Ravel
,
J. M.
and
Roberts
,
J. K. M.
(
1991
).
Hypoxia enhances phosphorylation of eukaryotic initiation factor 4A in Maize root tips
.
J. Biol. Chem
.
266
,
23341
23346
.
Weinstein
,
D. C.
and
Hemmati-Brivanlou
,
A.
(
1997
).
Neural induction in Xenopus laevis: evidence for the default model
.
Curr. Opin. Neurobiol
.
7
,
7
12
.
Wilson
,
P. A.
and
Hemmati-Brivanlou
,
A.
(
1995
).
Induction of epidermis and inhibition of neural fate by Bmp-4
.
Nature
376
,
331
333
.
Xu
,
R. H.
,
Kim
,
J.
,
Taira
,
M.
,
Zhan
,
S.
,
Sredni
,
D.
and
Kung
,
H. F.
(
1995
).
A dominant negative bone morphogenetic protein 4 receptor causes neuralization in Xenopus ectoderm
.
Biochem. Biophys. Res. Commun
.
212
,
212
219
.
Yoder-Hill
,
J.
,
Pause
,
A.
,
Sonenberg
,
N.
and
Merrick
,
W. C.
(
1993
).
The p46 subunit of eukaryotic initiation factor (eIF)-4F exchanges with eIF-4A
.
J. Biol. Chem
.
268
,
5566
5573
.