The VT 1 and VT 2 Isoforms of the Fibroblast Growth Factor Receptor Type 1 Are Differentially Expressed in the Presumptive Mesoderm of Xenopus Embryos and Differ in Their Ability to Mediate Mesoderm Formation*

Previously, we cloned a variant form of the type 1 fibroblast growth factor receptor (FGFR1), FGFR-VT 2 , from Xenopus embryos (Gillespie, L. L., Chen, G., and Paterno, G. D. (1995) J. Biol. Chem. 270, 22758–22763). This isoform differed from the reported FGFR1 sequence (FGFR-VT 1 ) by a 2-amino acid deletion, Val 423 Thr 424 , in the juxtamembrane region. This deletion arises from the use of an alternate 5 * splice donor site, and the activity of the VT 1 and VT 2 forms of the FGFR1 was regulated by phosphorylation at this site. We have now investigated the expression pattern and function of these two isoforms in mesoderm formation in Xenopus embryos. Cells within the marginal zone are induced to form mesoderm during blastula stages. RNase protection analysis of blastula stage embryos revealed that the VT 1 isoform was expressed throughout the embryo but that the VT 2 isoform was expressed almost exclusively in the marginal zone. The ratio of VT 1 :VT 2 transcripts in the marginal zone indicated that the VT 1 form was predominant throughout blastula stages except for a brief interval, coinciding with the start of zygotic transcription, when a dramatic increase in VT 2 expression levels was detected. This increase could be mimicked in part by treatment of animal cap explants with FGF-2.

This isoform differed from the reported FGFR1 sequence (FGFR-VT؉) by a 2-amino acid deletion, Val 423 -Thr 424 , in the juxtamembrane region. This deletion arises from the use of an alternate 5 splice donor site, and the activity of the VT؉ and VT؊ forms of the FGFR1 was regulated by phosphorylation at this site. We have now investigated the expression pattern and function of these two isoforms in mesoderm formation in Xenopus embryos. Cells within the marginal zone are induced to form mesoderm during blastula stages. RNase protection analysis of blastula stage embryos revealed that the VT؉ isoform was expressed throughout the embryo but that the VT؊ isoform was expressed almost exclusively in the marginal zone. The ratio of VT؉:VT؊ transcripts in the marginal zone indicated that the VT؉ form was predominant throughout blastula stages except for a brief interval, coinciding with the start of zygotic transcription, when a dramatic increase in VT؊ expression levels was detected. This increase could be mimicked in part by treatment of animal cap explants with FGF-2. Overexpression of the VT؉ isoform in Xenopus embryos resulted in development of tadpoles with severe reductions in trunk and tail structures, while embryos overexpressing the VT؊ isoform developed normally. A standard mesoderm induction assay revealed that a 10fold higher concentration of FGF-2 was required to reach 50% induction in VT؉-overexpressing animal cap explants compared with those overexpressing the VT؊ isoform. Furthermore, little or no expression of the panmesodermal marker Brachyury (Xbra) was detected in VT؉-overexpressing embryos, while VT؊-overexpressing embryos showed normal staining. This demonstrates that VT؉ overexpression had a negative effect on mesoderm formation in vivo. These data are consistent with a model in which mesoderm formation in vivo is regulated, at least in part, by the relative expression levels of the VT؉ and VT؊ isoforms.
Fibroblast growth factors (FGFs) 1 represent a family of related polypeptides known to stimulate a variety of cellular activities (reviewed in Ref. 1), including mesoderm differentiation in the Xenopus embryo (2). Their effects are mediated by high affinity transmembrane FGF receptors (FGFRs) containing intrinsic tyrosine kinase activity (reviewed in Ref. 3). Like other receptor tyrosine kinases, FGFR signal transduction is initiated by ligand binding and results in activation of several well characterized intracellular signaling pathways, including the protein kinase C (PKC) and Ras/mitogen-activated protein kinase pathways (4,5).
Four FGFR genes have been described to date, FGFR1-FGFR4, along with a number of alternately spliced variants (reviewed in Refs. 3 and 6). Previously, we cloned from Xenopus embryos an alternately spliced isoform of the FGFR1 that contains a deletion of Val 423 -Thr 424 in the juxtamembrane region (7). We demonstrated that this site could be phosphorylated by PKC. Furthermore, in a functional assay, activation of PKC by the phorbol ester phorbol 12-myristate 13-acetate significantly reduced the activity of the VT-containing isoform (VTϩ) in Xenopus oocytes while having little effect on the deletion isoform (VTϪ). We speculated that differential expression of these two isoforms might represent an important mechanism for regulating FGFR activity in the Xenopus embryo (7).
In the blastula stage Xenopus embryo, cells located in the equatorial region (marginal zone) are induced to differentiate into mesoderm in response to signal(s) from neighboring vegetal cells (reviewed in Refs. 8 and 9). Use of a dominant negative form of the FGFR1 to block FGF activity has provided evidence that FGF/FGFR signaling is required for normal development of the mesoderm (10). The current view is that FGF does not act as the initial inducing signal(s) but rather as a competence factor in the responding cells and that its activity is required for the full range of responses leading to mesoderm formation (9,11). In this report, we examine the role of the VTϩ and VTϪ isoforms in mesoderm formation in Xenopus and show that the two isoforms are differentially expressed in the presumptive mesoderm and that they differ in their ability to mediate mesoderm formation in vitro and in vivo.

EXPERIMENTAL PROCEDURES
Embryos and Microinjections-Xenopus laevis were purchased from Nasco. Eggs were artificially inseminated, and embryos were cultured as described under Godsave et al. (12); embryonic stages were determined according to Nieuwkoop and Faber (13). Stage 8 blastulae were dissected into animal, vegetal, and marginal zone regions as described (14). cRNA was transcribed from FGFRSP64T constructs (7) using the * This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada (to L. L. G. and G. D. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
SP6 Ribomax system (Promega). 4.6 nl containing diethyl pyrocarbonate-treated H 2 O or 650 pg/nl cRNA was microinjected into stage 1 embryos. Embryos were cultured at room temperature until they reached the stage required for the assays described below.
RNase Protection and Reverse Transcription-Polymerase Chain Reaction (RT-PCR)-Probe preparation, RNA extraction, and RNase protection analysis were performed as described (7). RT-PCR analysis was performed as described in Ref. 15 using forward (5Ј-GGGCTGCTTTT-GTGTCCGCAAT-3Ј) and reverse (5Ј-CATTGATGAGCTGGAGTCCCC-TG-3Ј) primers that bracket the VT region and generate 156-and 162-bp fragments for the FGFR-VTϪ and FGFR-VTϩ gene products, respectively. Histone H4 was used as an input control with forward and reverse primers as described (16). EF1␣ was amplified using 5Ј-CCTG-AATCACCCAGGCCAGATTGGTG-3Ј and 5Ј-GAGGGTAGTCTGAGAA-GCTCTCCACG-3Ј, as forward and reverse primers, respectively. The [ 32 P]CTP-labeled PCR products were analyzed in the linear range for amplification, determined empirically (16) to be 19 cycles for histone H4, 22 cycles for EF1␣, and 25 cycles for FGFR, and visualized on a 6% polyacrylamide/6 M urea gel by autoradiography. Quantitation by densitometry was performed as in Ref. 17.
Protein Analysis-For expression analysis of injected FGFR cRNA, 0.5 Ci of [ 35 S]methionine was co-injected into each embryo. Protein extraction, immunoprecipitation, and SDS-polyacrylamide gel electrophoresis analysis were performed as in Ref. 18. The anti-Xenopus FGFR1 used for immunoprecipitation was a polyclonal antibody raised against a synthetic C-terminal peptide (18).
Mesoderm Induction Assays-Recombinant Xenopus FGF-2 was expressed and purified according to Kimelman et al. (19). Animal cap explants were excised from injected embryos and treated with FGF-2 as described (18). Explants were cultured for various times then extracted for RNA analysis or scored for mesoderm induction as in Ref 2. Whole Mount in Situ Hybridization-Whole mount in situ hybridiza-tion with digoxygenin (Roche Molecular Biochemicals)-labeled Xenopus Brachyury (Xbra) or Xenopus chordin cRNA was performed as described (20), using maleic acid buffer. The cDNAs for Xenopus Brachyury and chordin were kindly provided by Dr. R. T. Moon (University of Washington).

FIG. 1. FGFR-VT؊ expression is spatially and temporally restricted during blastula stage.
A, schematic illustrating the 261-nt probe that corresponds to the sequence of the VTϩ isoform as well as the VTϩ and VTϪ protected fragments produced by RNase digestion for the analysis shown in B. Digestion of probe:VTϩ hybrids results in a 162-nt protected fragment, while digestion of probe:VTϪ hybrids results in digestion of the 6-nt single strand loop encoding the VT (black box), producing two protected fragments of 107 and 49 nt. The size in nt is listed below each fragment. B, RNase protection analysis of FGFR-VTϩ and FGFR-VTϪ mRNA in Xenopus blastulae. Stage 8 blastulae were dissected into animal, vegetal, and marginal zone regions (as illustrated in the schematic diagram shown above lanes 5-7), as described (14). Total RNA was isolated from each region, and RNase protection analysis was performed using a 32

RESULTS AND DISCUSSION
Previously, we demonstrated that the activity of the VTϩ and VTϪ isoforms could be differentially regulated by phosphorylation (7). One obvious question is whether these two isoforms function differently in mesoderm formation in Xenopus embryos. Induction to form mesoderm takes place in the marginal zone cells of the blastula stage embryo, so we began by examining the spatial expression pattern of the VTϩ and VTϪ isoforms during this stage of development.
Embryos were dissected into three regions representing the three germ layers: animal (presumptive ectoderm), marginal zone (presumptive mesoderm), and vegetal (presumptive endoderm). RNase protection assays of these three regions revealed that the VTϪ isoform was expressed predominantly in the marginal zone with very little detectable message in the animal and vegetal regions (Fig. 1B). In contrast, only small differences in VTϩ expression were observed in the three regions (Fig. 1B).
Although we had previously reported that the VTϩ isoform was the major form expressed throughout development and that no change in the ratio of VTϩ to VTϪ isoforms was detected (7), the time intervals used in that study were large in order to cover a broad range of developmental stages. In light of the results in Fig. 1B, we decided to reexamine the temporal VTϩ and VTϪ expression patterns in the marginal zone, using shorter time intervals and focusing on the stages when mesoderm induction is known to take place. Blastula stage embryos were collected at 0.5-h time intervals, and the VTϩ and VTϪ expression levels in the marginal zone were analyzed by RT-PCR. For this purpose, we employed primers that bracket the VT region and generate VTϩ and VTϪ products that can be distinguished on a sequencing gel. Our analysis revealed that in early blastulae (4.5 h postfertilization; late stage 7), the VTϩ isoform was the major form in marginal zone cells (Fig. 1C, lane  3), consistent with our previous findings (7). However, 30 min later (stage 8), a dramatic increase in the level of VTϪ relative to VTϩ was observed, such that the VTϪ isoform became predominant (Fig. 1C, lane 4). This was quickly followed by a decrease in VTϪ expression to initial levels, with the VTϩ isoform remaining predominant at all subsequent time points examined (Fig. 1C, lanes 5 and 6).
Our previous work demonstrated that these two FGFR1 isoforms arise by alternate use of a 5Ј splice donor site during transcription (7). In Xenopus embryos, however, zygotic transcription does not begin until midblastula transition (21). This occurs during stage 8, but the precise timing of this developmental event cannot be determined by either the number of cell divisions or the time after fertilization (Ref. 21; reviewed in Ref. 22). Instead, an increase in elongation factor 1-␣ expression, one of the earliest transcripts to be expressed by the embryonic genome (23), has been frequently used to indicate that zygotic transcription has begun. We measured elongation factor 1-␣ levels in our marginal zone samples and determined that expression levels began to increase as early as 5 h (Fig. 1C,  lane 4). This demonstrates that the increase in VTϪ expression takes place concurrent with the onset of zygotic transcription.
We investigated the possibility that FGF itself was involved in this switch in expression pattern, since Musci et al. (24) and Friesel and Dawid (25) have reported that FGFR1 mRNA levels in animal cap explants were regulated by FGF. We cultured

FGFR1 Isoforms in Mesoderm Formation
blastula stage animal cap explants with FGF-2 in a standard mesoderm induction assay (2) and determined VTϩ and VTϪ expression levels at various times after the addition of FGF-2. VTϪ expression levels in FGF-2-treated explants increased within 0.5 h, compared with untreated control explants (Fig. 2,  lanes 1 and 2). At all subsequent time intervals tested, VTϪ expression levels in FGF-treated explants remained the same as control levels (Fig. 2, lanes 3-8). VTϩ expression levels, on the other hand, remained relatively constant throughout. These data demonstrate that a temporary increase in the expression level of the VTϪ isoform can be triggered by FGF. It is clear, however, that FGF induction alone cannot account for the large increase in VTϪ levels observed in vivo (compare Fig.  1C, lane 4, with Fig. 2, lane 2). The additional factor(s) responsible for the increase in VTϪ expression in the marginal zone remain to be determined.
The rapid and transient switch to VTϪ expression in the presumptive mesoderm suggests that these two FGFR1 isoforms have distinct roles in early development. We investigated this possibility by examining the effects of overexpressing each isoform. cRNA was microinjected into fertilized eggs, and FGFR-injected embryos were compared with control H 2 O-injected embryos for their ability to develop into normal tadpoles. While embryos overexpressing the VTϪ isoform developed normally (Figs. 3A and 4C), less than 10% of those overexpressing the VTϩ isoform developed into normal tadpoles (Figs. 3A and 4B). VTϩ-injected embryos began to develop abnormally at 10 -12 h postinjection: gastrulation was incomplete, leaving an enlarged blastopore with protruding yolk plug (not shown). Despite this, embryos continued to develop, albeit abnormally. The obvious effect was a reduction in trunk and tail structures in the resulting tadpoles (Fig. 4B). This differential effect was not due to differential stability or translation of the cRNAs, since equivalent levels of VTϩ and VTϪ RNA (Fig. 3B) and protein (Fig. 3C) were detectable at 24 h, long past the stage when abnormalities first became apparent in the VTϩ-overexpressing embryos.
The VTϩ abnormalities were similar to those reported by Amaya et al. (10) in embryos overexpressing a dominant negative FGFR1 (XFD). These authors showed that XFD inhibited endogenous FGFR signaling and that overexpression in embryos caused severe posterior truncations resulting from a reduction in posterior mesoderm development, as well as deficiencies in gastrulation movements. This similarity suggested that the VTϩ phenotype may result from a deficiency in mesoderm formation. To test this, we investigated the effect of VTϩ and VTϪ overexpression on mesoderm formation in vitro and in vivo.
First, we measured the FGF-2 dose-response curve in explants from embryos microinjected with either VTϩ or VTϪ cRNA and compared it with the curve for explants from H 2 Oinjected embryos. Our results demonstrate that overexpression of the VTϩ isoform reduced the level of mesoderm induction by FGF-2; overexpressing explants required a 2-fold higher concentration of FGF than control explants to achieve 50% induction (Fig. 5A). Explants overexpressing the VTϪ isoform, on the other hand, reached 50% induction at a 5-fold lower concentration than control explants (Fig. 5A). Thus, overexpression of VTϩ decreased sensitivity to FGF, while overexpression of VTϪ dramatically increased sensitivity. Examination of the VTϪ and VTϩ expression levels in these explants revealed that the sensitivity to FGF was directly correlated with the relative expression levels of the two isoforms (Fig. 5B). The ratio of VTϪ to VTϩ in explants from VTϩ-injected embryos was half that of control explants (Fig. 5B, lanes 1 and 3), as was the proportion of induced VTϩ explants at virtually every concentration of FGF tested (Fig. 5A). Explants from VTϪ-injected embryos, on the other hand, had the highest ratio of VTϪ to VTϩ (Fig. 5B, lane 5) and the highest sensitivity to FGF (Fig. 5A). In fact, 30% of the latter were induced in the absence of added FGF-2 (Fig. 5A). This autoinduction may result from interaction of overexpressed VTϪ with maternally derived FGF present in animal cap explants (9).
From the results of the dose-response curves, one would predict that overexpressing the VTϩ isoform would result in a decreased level of mesoderm formation in vivo. To test this hypothesis, we examined the expression of an early panmesodermal marker Xbra, which is normally expressed throughout the presumptive mesoderm of the early gastrula stage embryo (26). Furthermore, FGFR signaling is required for Xbra expression (reviewed in Ref. 27). Staining for Xbra was not detectable or was very faint in VTϩ-injected embryos (Fig. 6A). VTϪinjected embryos, on the other hand, expressed levels similar to those of H 2 O-injected controls (Fig. 6A). Expression of chordin, a dorsal lip marker involved in neural development (28,29), was unaffected in VTϩ embryos (Fig. 6B), demonstrating that lack of Xbra expression in VTϩ-injected embryos was not the result of a nonspecific inhibition of transcription. Thus, the VTϩ isoform can function to negatively regulate mesoderm formation in Xenopus embryos.
In this study, we have shown that while the VTϩ isoform was predominant in the presumptive mesoderm during most of

FGFR1 Isoforms in Mesoderm Formation
blastula stages, a brief but dramatic increase in VTϪ mRNA expression occurred during this time period (Fig. 1C), coinciding temporally with mesoderm induction in vivo (9). It would be important to determine how this transient burst in VTϪ expression affects known FGF signaling pathways in the embryo, such as mitogen-activated protein kinase and phospholipase C␥. Labonne and Whitman (30) reported that mitogen-activated protein kinase activity was first detectable during midblastula and peaked by midgastrula. Phosphorylation of phospholipase C␥, on the other hand, occurred over a period of 1.5 h during early blastula to midblastula stages (18). While activation of these pathways persisted over a longer time interval than that reported in this study, it is possible that the VTϪ protein has a longer half-life than its cognate mRNA. We are attempting to generate antibodies that can distinguish between the VTϩ and VTϪ isoforms and would enable investiga-tion of isoform-specific signaling pathways.
We have also shown that the VTϩ and VTϪ isoforms differ significantly in their ability to mediate mesoderm induction. This difference in isoform function could be due to PKC activity. PKC is known to be activated during mesoderm induction by FGF (4), and activated PKC caused a substantial reduction in FGF signaling through the VTϩ, but not the VTϪ, isoform (7). The differential activity of these two FGFR isoforms means that two tissues in the embryo, one expressing predominantly VTϪ and the other predominantly VTϩ, could be exposed to the same, low concentration of FGF, and only the tissue expressing high levels of VTϪ would be induced. Our data suggest then that mesoderm formation in vivo is dependent not only on the local concentration of FGF but also on the relative expression levels of the VTϩ and VTϪ isoforms in the responding tissue. This hypothesis provides a possible mechanism for Values from 10 individual experiments were plotted; the bars represent S.E. B, total RNA was extracted from stage 8 animal cap explants (five per sample) from injected embryos and analyzed by RT-PCR for VTϩ, VTϪ, and histone H4 expression levels, as described in the legend to Fig. 1C. A representative experiment is shown. The positions of the VTϩ, VTϪ, and H4 PCR products are indicated. Quantitation by densitometry of the VTϩ and VTϪ expression levels in each sample was performed as described in Ref. 17, and the ratio of VTϪ to VTϩ obtained from these measurements is indicated below the appropriate lane.
restricting mesoderm induction to marginal zone cells, although all cells in the blastula stage embryo have been shown to express FGFRs (11,14). While much of the work on mesoderm induction has focused on analysis of potential inducers, such as eFGF, activin, and Vg1 (reviewed in Refs. 9, 31, and 32), our results demonstrate that regulated expression of receptors and receptor isoforms also plays a critical role.