Expression patterns of Fgf-8 during development and limb regeneration of the axolotl

Authors


Abstract

Fgf-8 is one of the key signaling molecules implicated in the initiation, outgrowth, and patterning of vertebrate limbs. However, it is not clear whether FGF-8 plays similar role in development and regeneration of urodele limbs. We isolated a Fgf-8 cDNA from the Mexican axolotl (Ambystoma mexicanum) through the screening of an embryo cDNA library. The cloned 1.26-kb cDNA contained an open reading frame encoding 212 amino acid residues with 84%, 86%, and 80% amino acid identities to those of Xenopus, chick, and mouse, respectively. By using the above clone as a probe, we examined the temporal and spatial expression patterns of Fgf-8 in developing embryos and in regenerating larval limbs. In developing embryos, Fgf-8 was expressed in the neural fold, midbrain-hindbrain junction, tail and limb buds, pharyngeal clefts, and primordia of maxilla and mandible. In the developing axolotl limb, Fgf-8 began to be expressed in the prospective forelimb region at pre–limb-bud and limb bud stages. Interestingly, strong expression was detected in the mesenchymal tissue of the limb bud before digit forming stages. In the regenerating limb, Fgf-8 expression was noted in the basal layer of the apical epithelial cap (AEC) and the underlying thin layer of mesenchymal tissue during blastema formation stages. These data suggest that Fgf-8 is involved in the organogenesis of various craniofacial structures, the initiation and outgrowth of limb development, and the blastema formation and outgrowth of regenerating limbs. In the developing limb of axolotl, unlike in Xenopus or in amniotes such as chick and mouse, the Fgf-8 expression domain was localized mainly in the mesenchyme rather than epidermis. The unique expression pattern of Fgf-8 in axolotl suggests that the regulatory mechanism of Fgf-8 expression is different between urodeles and other higher species. The expression of Fgf-8 in the deep layer of the AEC and the thin layer of underlying mesenchymal tissue in the regenerating limbs support the previous notion that the amphibian AEC is a functional equivalent of the AER in amniotes. © 2001 Wiley-Liss, Inc.

INTRODUCTION

Urodeles have been used extensively to study the mechanism of regeneration because of their remarkable capacity to restore missing parts during the entire span of their lives. Classic studies showed that if nerve or wound epidermis is removed during early phases of regeneration, regeneration is blocked (Singer, 1952; Thornton, 1957). Many kinds of factors, including growth factors, are suggested to be produced from nerve, wound epidermis, or mesenchymal tissue of blastema during the process of regeneration, and they are presumed to play crucial roles for successful limb regeneration. However, the precise role of growth factors in limb regeneration is still obscure. Nevertheless, some growth factors are linked to limb regeneration as has been demonstrated by the presence of FGF-1 in the regenerating blastema and the expression of FGFRs associated with regeneration (Boilly et al., 1991; Poulin et al., 1993; Poulin and Chiu, 1995).

FGF, a member of the heparin-binding growth factor family (HBGF), is known to control the proliferation and differentiation of cells and to be involved in the embryonic induction, angiogenesis, and healing of damaged tissue (Basilico and Moscatelli, 1992; Pandit et al., 1998). So far, at least 20 members of FGFs have been reported in vertebrates (reviewed in Szebenyi and Fallon, 1999; Nishimura et al., 1999; Kirikoshi et al., 2000). Many kinds of FGFs such as FGF-2, -4, -8, and -10 have been reported to play important signaling roles in the vertebrate limb development. Furthermore, it is well known that the spatial and temporal expression profile of each Fgf is unique. For example, Fgf-2 is expressed in the mesenchymal tissue and epidermis of the developing chick limb (Fallon et al., 1994), and Fgf-4 is expressed in the posterior half of apical ectodermal ridge (AER) of mouse limb bud (Niswander and Martin, 1992). Fgf-8 begins to be expressed at the presumptive AER of the pre–limb-bud ectoderm and expressed continuously in the AER during limb bud outgrowth in mouse and chick (Heikinheimo et al., 1994; Crossley et al., 1996b). Moreover, these FGFs can substitute for the AER's role, as has been well demonstrated by the restoration of limb outgrowth and gene expression in AER-removed limb buds (Niswander et al., 1993; Fallon et al., 1994). Interestingly, when the FGF-2 or FGF-8 protein-soaked bead or the FGF-4 expressing cells are implanted into the flank region of chick embryo, an extra limb is induced (Cohn et al., 1995; Ohuchi et al., 1995; Vogel et al., 1996). It indicates that the FGFs are crucial signaling molecules in limb initiation and development.

Receptor binding specificity is an essential mechanism for regulating FGF activity during limb induction in the mouse (Xu et al., 1998). Generally, genes of FGF receptors are expressed differentially in embryonic tissues of mouse and regenerating newt limb. Fgfr1 is expressed only in the mesenchyme of limb buds and regenerating blastema, whereas the KGFR variant transcript of Fgfr2 is expressed predominantly in the epithelial cells of embryonic skin and of developing organs and in the basal layer of wound epithelium during limb regeneration and the bek variant transcript of Fgfr2 is expressed predominantly in the mesenchymal cells of the developing mouse organs and in the blastemal mesenchyme during newt limb regeneration (Peters et al., 1992; Orr-Urtreger et al., 1993; Poulin and Chiu, 1995). The differential expression of Fgfr suggests that these genes are independently regulated and that they have unique functions during development and regeneration.

In the regenerating Xenopus tadpole limb, Xenopus FGF-10 (xFGF-10) expression in the limb mesenchymal cells corresponds to the regenerative capacity and that Fgf-10 and Fgf-8 are synergistically reexpressed in regenerating blastemas (Christen and Slack, 1997; Yokoyama et al., 2000). Previous studies on the expression and the role of FGF-8 as stated above suggest that the FGF signal transduction pathway activated by FGF-8 may play a similar role in the development and regeneration of vertebrate limb as well as in embryogenesis. Because the apical epithelial cap (AEC) of urodele limb regenerate derived from the wound epidermis after amputation is speculated to be a functional equivalent of AER, it is intriguing to know whether the Fgf-8 expression pattern in a urodele is similar to amniote and anuran counterparts. To obtain aforementioned information, we cloned an Fgf-8 homologue of Mexican axolotl, Ambystoma mexicanum, and analyzed the expression pattern of Fgf-8 in embryonic development and in developing and regenerating limbs. In the developing embryo, Fgf-8 was expressed in various regions involved in patterning such as limb bud, tail bud, the surface ectoderm covering facial primordia, and midbrain-hindbrain junction. These expression patterns are similar to those found in other vertebrate species such as chick and mouse (Crossley and Martin, 1995; Mahmood et al., 1995). However, unlike Xenopus or amniotes such as chick and mouse, the Fgf-8 expression in developing axolotl limbs was localized mainly in mesenchymal rather than epidermal tissue. In the regenerating limb, Fgf-8 expression was found in the basal layer of AEC and the underlying thin layer of mesenchymal tissue. These data support that the AEC of urodele amphibians can be a homologous to the AER of developing limbs of amniotes (Christensen and Tassava, 2000).

RESULTS

Axolotl Fgf-8 cDNA Cloning

An axolotl Fgf-8 cDNA clone was isolated by screening of cDNA library constructed with RNA from stage 35 embryos of Mexican axolotl (Ambystoma mexicanum) by using a chicken Fgf-8 cDNA as a probe. As a consequence of nucleotide sequence analysis of four positive clones, they were confirmed to be 1.26 kb in size and consisted of 5′ untranslated region (UTR) of 531 bp, an open reading frame of 639 bp encoding 212 amino acids and the termination codon, and 3′ UTR of 90 bp (Fig. 1A,B). In addition, axolotl FGF-8 had two putative N (Asn)-linked glycosylation sites (NFT and NYT at amino acid position 32-34 and 138-140, respectively; Fig. 1A).

Figure 1.

Sequence analysis of Fgf-8 cDNA and FGF-8 protein of Mexican axolotl, Ambystoma mexicanum. A: Nucleotide and its deduced amino acid sequences of axolotl FGF-8. The predicted signal peptide is underlined. Bold amino acids represent putative N-linked glycosylation sites (N-X-T/S). Asterisk represents stop codon. GenBank accession number for axolotl Fgf-8 is AF190448. B: Comparison of the deduced amino acid sequences of axolotl FGF-8 with Xenopus (Christen and Slack, 1997), zebrafish (Fürthauer et al., 1997), chick (Crossley et al., 1996b), mouse (Tanaka et al., 1992), and human (Gemel et al., 1996) homologs. Dashes represent gaps inserted to align homologous residues. Identical amino acids and conservative substitutions are indicated by black boxes. Amino acids under asterisks represent consensus amino acids (G-x-L-(x)9∼10-C-x-F-x-E-(x)6-Y) conserved in most of members of vertebrate FGF family. C: Phylogenetic tree of FGF-8 by using clustral method of MegAlign (DNASTAR). D: Hydrophobicity profile of axolotl FGF-8 protein. The vertical scale (Score) indicates Kyte Hydrophobicity values and the numbers in the horizontal line represent amino acid position of FGF-8 protein. Hydrophilic signal peptide can be seen at N-terminal.

Cloned axolotl FGF-8 showed a high sequence similarity with 84%, 86%, and 80% identities in deduced amino acid sequence to those of Xenopus, chick, and mouse, respectively. Moreover, in the 130 amino acid conserved core region, the sequence similarity was more than 90%, and the axolotl FGF-8 contained three conserved cysteine residues (20, 110, and 128) as in other vertebrates (Fig. 1B). The analysis of deduced amino acid sequence revealed that axolotl FGF-8 contained typical consensus amino acid sequence, ′G-x-L-(x)9∼10-C-x-F-x-E-(x)6-Y′, which is the conserved core region of all members of FGF family from Xenopus to human (Fig. 1B). Phylogenetic analysis shows clustering of FGF-8 among several species (Fig. 1C). Hydrophobicity profile of axolotl FGF-8 protein showed that FGF-8 had a potential signal peptide sequence consisting of 22 amino acids at N-terminal, and the remaining part of FGF-8 was mostly composed of hydrophilic amino acids (Fig. 1A,D).

Fgf-8 Expression Pattern in Embryos

The Fgf-8 expression in the developing axolotl embryo showed differences as well as similarities when compared with the expression patterns in other vertebrate species such as Xenopus, chick, and mouse (Crossley and Martin, 1995; Mahmood et al., 1995; Christen and Slack, 1997). Similarities include the early expression of Fgf-8 in the neural folds, posterior blastopore, and future midbrain-hindbrain junction (isthmus; Fig. 2A) in the neurula stage (stage 15) embryo. In particular, the expression of Fgf-8 in the isthmus was detected in the neural plate as two bands beginning at the neurula stage and continuing throughout embryogenesis until hatching (Fig. 2A–D and data not shown). At the beginning of tail bud stage (stage 24), Fgf-8 expression was detected in the gill swelling, facial primordia, isthmus, and tail bud (Fig. 2B). In the continuing middle tail bud stage (stage 32), Fgf-8 expression was strong in the regions of facial primordial (mandibular arch), gill swelling, isthmus, and tail bud (Fig. 2C). At stage 35 (late tail bud), Fgf-8 signal was also detected in the tail bud and the primordia of facial structure such as maxilla and mandible (Fig. 2D and data not shown). These results suggest that the role that FGF-8 plays during axolotl embryogenesis is similar to that of other amphibians and amniotes.

Figure 2.

The expression profiles of Fgf-8 in whole embryos. All embryos are lateral view except A (dorsal view). A: Stage 15 (neurula stage). Fgf-8 transcript is expressed in the prospective midbrain-hindbrain junction (isthmus; white arrowheads), anterior neural fold (arrowhead), and posterior blastopore (asterisk). B: Stage 24 (early tail bud stage). Fgf-8 is expressed lightly in the vaguely recognizable tail bud (asterisk), gill swelling (arrow), facial primodia (arrowhead) and isthmus of brain (white arrowhead). C: Stage 32 (middle tail bud stage). D: Stage 35 (late tail bud stage). Fgf-8 is expressed in gill swelling (arrow), tail bud (asterisks), mandibular arch (arrowheads), and isthmus (white arrowhead). Scale bars = 500 μm.

Fgf-8 Expression Pattern in Developing Limbs

The expression pattern of Fgf-8 in the developing axolotl limb is unique when compared with other vertebrates. Like other vertebrates, Fgf-8 was initially expressed in the epidermis of the prospective forelimb region at prelimb bud stage before hatching (stage 42, data not shown). At 1 day after hatching, in which limbs form tiny buds, Fgf-8 was expressed in the mesenchyme as well as epidermis (Fig. 3A,B). However, as limbs develop, Fgf-8 transcript was localized mainly in the mesenchyme of the bud (Fig. 3C,D). The sectioned tissues of predigit stages limb buds (4 days after hatching) showed its expression only in the mesenchymal and subepidermal tissue of distal region of forelimb bud (Fig. 3E,F). Interestingly, Fgf-8 expression signal was more intense in the anterior region of the limb bud than the posterior region (Fig. 3E). In the dorsoventral plane, Fgf-8 expression signal was especially intense at the distal end of the limb bud, but polarized expression pattern was not distinguishable as in the anteroposterior axis (Fig. 3F). This expression pattern was continued until 8 days after hatching, but the level of expression starts to decline from this time (Fig. 3G). As digits formed at 11 days and 14 days after hatching, the Fgf-8 transcripts disappeared gradually in the limb (Fig. 3H,I). In hindlimb buds at predigit stage (14 days after hatching), Fgf-8 transcripts were also detected at the mesenchymal tissue of the limb buds (data not shown). Thus, the expression pattern of Fgf-8 in the developing axolotl limb was somewhat different from amniotes in which it is expressed only in the AER during limb bud outgrowth.

Figure 3.

The expression profiles of Fgf-8 in the developing limbs of axolotl larvae. A: At 1 day after hatching, when limb buds start to form, Fgf-8 is expressed in the epidermis and mesenchyme of the budding limb (arrowhead). Gills were removed for better viewing of the limb bud. B: Longitudinal section of forelimb bud at 1 day after hatching. The Fgf-8 transcript is detected in the mesenchyme as well as epidermis. Anterior is to the top and distal is to the right in the image. C,D: At 2 days (C) and at 4 days (D) after hatching, a strong expression is detected in the mesenchymal and subepidermal tissue of forelimb bud (arrowheads). The expression at 4 days after hatching is more intense than that at 2 days after hatching, and the expression domain is somewhat anteriorly polarized. E,F: Sections of limb bud at 4 days after hatching in the anterior-posterior plane (E) and dorsal-ventral plane (F). Note the anteriorly polarized expression of Fgf-8 in the anterior-posterior axis (E), but not as much as pronounced in the dorsal-ventral axis (F). Anterior (in E) and dorsal (in F) are to the top and distal is to the right in both images. G: This expression pattern is continued until 8 days after hatching, but the level of expression is clearly reduced (arrowhead). H,I: At 11 days (H) and 14 days (I) after hatching, when the digits are formed, only the trace level (H) or background level (I) of expression can be detected. C–I; ventral views except in E and F., Scale bars = 200 μm in A; 20 μm in B; 100 μm in C–I (except E and F in which bars are 50 μm).

Fgf-8 Expression Pattern in Regenerating Limbs

To examine the expression pattern of Fgf-8 in the regenerating limb, forelimbs of axolotl larvae were amputated at the distal stylopodium (elbow) level. We could not observe the Fgf-8 expression until the dedifferentiation stage (data not shown and Fig. 4A; 2 days after amputation). The expression of Fgf-8 began to be detected in the regenerating blastema even though a very low level of signal was detected at the early bud stage (Fig. 4B; 4 days after amputation). Axolotl Fgf-8 was expressed mainly during the blastema formation stage showing a highest intensity in the medium bud stage (Fig. 4C,D; 6–8 days after amputation). After the medium bud stage, the Fgf-8 signal disappeared gradually (Fig. 4E; 10 days after amputation, late bud stage). Beyond the palette stage, only background levels of expression were detected (Fig. 4F). Thus, in the regenerating limb, Fgf-8 expression was not apparent from the onset of the regeneration process, rather, it expressed as the blastema forms.

Figure 4.

The expression profiles of Fgf-8 in the regenerating forelimbs of axolotl larvae. Forelimbs were amputated at the level of distal stylopodium (white lines). A: Two days after amputation (dedifferentiation stage). No expression of Fgf-8 is detected until this stage. B: Four days after amputation (early bud stage). Very low level of signal is detected in the blastema. C: Six days after amputation. D: Eight days after amputation. Medium bud stage. Fgf-8 signal is detected mainly in the blastema (white arrowheads). Fgf-8 transcripts are localized in the basal layer of the apical ectodermal cap and underlying thin layer of mesenchymal tissue (refer Fig. 5). E: Ten days after amputation (late bud stage). Weak signal of Fgf-8 is seen in the blastema (white arrowhead). F: Twelve days after amputation (palette stage). Only background level of expression is detected. Scale bars = 1 mm.

The Fgf-8 signals of whole-mount regenerates seem to be detected in the mesenchymal tissue of blastema such as in the developing limbs. To study the expression of Fgf-8 in detail, histologic examination was done in the sectioned regenerating tissues after whole-mount in situ hybridization. The result showed that Fgf-8 transcripts were localized at the basal layer of AEC and the underlying thin layer of mesenchymal tissue (Fig. 5). This pattern of expression gave a false impression, as if it is expressed in the mesenchymal blastemal tissue in the whole-mount samples, because the distal part of blastema is covered with the basal layer as a cap. Interestingly, the Fgf-8 expression in the blastema showed a somewhat polarized profile like developing limbs. This impression was more pronounced in the medium bud stage blastema in which Fgf-8 signal was more intense in the anterior region rather than in the posterior region (Fig. 5A,B). However, in the dorsoventral plane, polarized signal distribution was not evident (Fig. 5C,D). Therefore, there is some similarity in that Fgf-8 is expressed more strongly in the anterior region of both the developing limb bud and the regenerating limb blastema.

Figure 5.

Histologic examination of Fgf-8 expression in sectioned regenerating tissues. Tissues were prepared by sectioning the 8-days after amputation samples. A,B: Anteroposteriorly sectioned plane. C,D: Dorsoventrally sectioned plane. B and D are magnified images of boxes in A and C, respectively. a: anterior, p: posterior, d: dorsal, v: ventral. The Fgf-8 expression shows somewhat polarized pattern in the anterior-posterior plane. However, in the dorsal-ventral plane, polarized expression pattern is not shown. H, humerus; m, mesenchyme; AEC, apical epithelial cap. Scale bars = 100 μm.

DISCUSSION

Fibroblast growth factor-8 (FGF-8) has been implicated to play crucial signaling roles in many vertebrate developmental processes. Especially, in several previous studies, the role of FGF-8 in limb development has been studied extensively in chick and mouse. In the developing limb, Fgf-8 is expressed in AER from pre–limb-bud stage and has been shown to be essential for outgrowth (Crossley et al., 1996b; Vogel et al., 1996). However, it has not been known whether Fgf-8 is expressed in the regenerating limb of urodele that has remarkable regenerative power. Thus, the present study was carried out to determine whether FGF-8 shows similar a expression pattern in limb development and regeneration of urodele.

Cloning and Characterization of Axolotl Fgf-8 Gene

Although many Fgf genes have been cloned in many vertebrate species, including mouse, human, chick, Xenopus, and fish (Tanaka et al., 1992; Gemel et al., 1996; Crossley et al., 1996b; Christen and Slack, 1997; Fürthauer et al., 1997), among urodele amphibians only Fgf-1 has been previously cloned in the newt (Patrie et al., 1997). Thus, our cloning of Fgf-8 in the Mexican axolotl represents only the second member of the FGF family to be cloned from a urodele, and the first characterization of a urodele Fgf-8 gene.

The deduced amino acid sequence of axolotl FGF-8 showed more than 80% of similarities with the FGF-8s of other vertebrates. Moreover, in 130 amino acids of the consensus core region except N- and C-termini, it showed more than 90% sequence similarity among the species. In addition, the axolotl FGF-8 contained the typical consensus amino acid sequences shown in the conserved core region of all FGF family members of vertebrates. Thus, the axolotl FGF-8 appears to be a highly conserved molecule evolutionarily (Coulier et al., 1997). Some FGF members such as FGF-1, FGF-2, FGF-9, and FHFs are devoid of consensus secretory signal peptides. However, the presence of a potential signal peptide and two putative glycosylation sites suggest that axolotl FGF-8 is a secreted protein.

FGF-8 in Embryogenesis

We have examined whether the expression profile of axolotl Fgf-8 is similar to that of other species during embryogenesis. In axolotl embryo, Fgf-8 was found to be expressed in isthmus, neural fold, pharyngeal clefts, craniofacial primordia (maxilla and mandible), tail, and limb buds. These expression patterns suggest that FGF-8 is involved in the formation of midbrain, facial structures, pharyngeal derivatives, tail, and limbs. Moreover, the similar Fgf-8 expression pattern in the developing axolotl embryo with other species such as Xenopus, chick, and mouse suggests that FGF-8 plays similar signaling roles for the formation of various organs as has been proposed in Xenopus and amniotes (Crossley and Martin, 1995; Vogel et al., 1996; Christen and Slack, 1997).

It is worthwhile to note the expression pattern of Fgf-8 in the developing midbrain-hindbrain junction (isthmus) and developing limbs. Isthmus has been proposed to be an organizing center regulating the midbrain development (Joyner, 1996). FGF-8 signaling from the midbrain-hindbrain junction (isthmus) is suggested to play a key role in coordinately regulating growth and polarity in the developing mesencephalon of chick and mouse (Crossley et al., 1996a; Lee et al., 1997) and recently in zebrafish (Reifers et al., 1998). In chick embryo, FGF-8 induces mirror-imaged duplication of midbrain and ectopic expression of Engrailed-2 (En-2) and Wnt1 in the forebrain, which are normally expressed in the isthmus (Lee et al., 1997). In addition, FGF-8 is suggested to be required for the induction of En-2 that controls the pattern of midbrain-hindbrain region and for the sustained expressions of Pax2, Wnt1, and En-1 that establish the isthmus in rat embryo (Ye et al., 1998). In axolotl, Fgf-8 expression was detected in the presumptive isthmus from as early as neural fold stage and maintained during the whole span of embryogenesis until hatching. Thus, the result suggests that axolotl FGF-8 plays a similar role in the development of axolotl midbrain as in other species.

FGF-8 in Limb Development

In urodele embryo, the prospective forelimb field appears to be specified at tail bud stage (stage 30), and limb development begins through the proliferation of cells in the somatic layer of the limb field of lateral plate mesoderm. In sequence, a mass of mesodermal cells proliferated in the forelimb field gives rise to the limb bud, which can be observed before hatching (Stocum and Fallon, 1982).

In the present study, we have found that the Fgf-8 expression domain is translocated from epidermis to mesenchyme gradually as limb development proceeds. Although the role of FGF-8 is not well studied during the initiation and in the maintenance of urodele limb outgrowth, based on the expression pattern in the epidermis during the early phase of limb development, axolotl FGF-8 is likely to play a role in limb initiation that is similar to that in chick or mouse (Lewandoski et al., 1997). However, further studies are needed to know what role FGF-8 expression in the mesenchyme of the axolotl limb bud is playing in limb outgrowth. In the limbless mutant of chick, AER does not form due to dorsalized ectoderm and Fgf-8 is not expressed consequently (Grieshammer et al., 1996; Ros et al., 1996). However, exogenous supply of FGF can rescue the defect. Moreover, ectopic application of FGF-8 in the flank tissue induces additional limb (Crossley et al., 1996b). Thus, FGF-8 was proposed to be a signaling molecule regulating the outgrowth of limb (Crossley et al., 1996b; Vogel et al., 1996). Ohuchi et al. (1997) proposed that FGF-10 rather than FGF-8 is a key factor responsible for the initiation and outgrowth of chick limb bud. Their study shows that Fgf-10 expression in the prospective limb mesenchyme precedes Fgf-8 expression in the nascent AER, and ectopic application of FGF-10 to the flank region can induce Fgf-8 expression in the adjacent ectoderm, resulting in the formation of an additional complete limb. Therefore, it would be interesting to know whether Fgf-10 is expressed in axolotl limb development and how it is related to the expression and function of Fgf-8.

Notably, there are differences in the expression pattern of Fgf-8 between the developing limbs of two representative amphibian species, axolotl and Xenopus. First, in axolotl, like in chick and mouse, Fgf-8 expression begins before limb bud formation, whereas, it begins from stage 49 of Xenopus tadpole when the limb bud was clearly visible (Christen and Slack, 1997). Second, Fgf-8 is expressed in the distal mesenchymal tissue of the axolotl limb bud, whereas it is expressed in the distal ectodermal tip of the Xenopus hindlimb bud as a broad stripe along the dorsoventral boundary (Christen and Slack, 1997). Similarly, FGFR shows also a differential expression pattern between the regeneration processes of two systems. In the regenerating hindlimb of Xenopus, FGFR-1 (flg) is expressed both in the mesenchyme and the wound epithelium (D'Jamoos et al., 1998), but it is expressed only at the blastemal mesenchymal cells in newt limb regenerate (Poulin et al., 1993). These differences suggest FGF signaling mechanism can be differential between anuran and urodele.

In chick and mouse, Fgf-8 is also expressed in the ectoderm of the limb field before the outgrowth of the limb buds and localized later only to the AER until regression (Mahmood et al., 1995; Vogel et al., 1996). In our study, the axolotl Fgf-8 probe could cross-hybridize the mRNA in the chick limb bud. In whole-mount in situ hybridization of chick limb bud with the axolotl Fgf-8 probe, the signal was seen only in the AER of limb bud (data not shown). There is no possibility Fgf-8 probe is hybridized with other members of FGF family under high stringent condition because FGF-8 sequence show less than 50% of similarities with them. It seems highly unlikely that the peculiar spatial expression pattern shown in axolotl limb bud is the result of artifact due to the cross-reactivity of Fgf-8 probe with other members of the FGF family. Thus, the Fgf-8 expression pattern in the developing axolotl limb is certainly different from amniotes and Xenopus.

FGF-8 in Limb Regeneration

Wound epidermis or apical epithelial cap (AEC, a new nomenclature, as has been proposed recently by Christensen and Tassava in Dev Dyn, 2000) is crucial in limb regeneration, which cannot proceed properly without it (Thornton, 1957). Therefore, it can be analogous to the role that the AER plays in limb development (Saunders, 1948). If the notion that AEC in amphibian limb regenerates is a functional equivalent of the amniote AER is correct, it would be reasonable to anticipate Fgf-8 expression in the AEC of limb regenerates. The result was not one-sided because the Fgf-8 expression was localized as a cap-like shape encompassing both the basal layer of the AEC and the underlying thin layer of mesenchymal tissue. XenopusFgf-8 was also reported to be expressed in the inner basal layer of the AEC of regenerating limbs (Yokoyama et al., 2000). The Fgf-8 expressed in the basal layer of regenerating amphibian AEC is supposed to be involved in the regeneration process by the mesenchymal-epidermal interaction like in the vertebrate limb development. Christensen and Tassava (2000) suggested recently that the basal layer of the AEC rather than whole AEC is regarded as a functional equivalent to AER of developing amniote limbs by showing a strong expression of fibronectin in this inner layer and its distinct notch/groove shape-like stripe along the dorsoventral boundary. The Fgf-8 expression profile in our study partly support their suggestion, because Fgf-8 transcripts are present both in the basal layer of the AEC with additional transcripts seen in the distal most blastema. Moreover, Fgf-8 is expressed mainly when the blastema forms. Thus, this spatial and temporal expression pattern of Fgf-8 suggests that FGF-8 is a molecule specifically involved in the cell proliferation of the regenerating blastema.

There is evidence, several indirect and direct, that FGF signaling is essential for amphibian limb regeneration. FGF is able to promote mitotic activity in the denervated blastemas of regenerating urodele limbs and to induce morphogenetic events in amputated frog limb and stimulates proliferation of cultured blastema cells (Mescher and Gospodarowicz, 1979; Gospodarowicz and Mescher, 1980). FGF is also capable of sustaining proliferation in regenerating limbs deprived of the wound epidermis and of exerting a positive influence on the synthesis and the extracellular accumulation of GAGs (Chew and Cameron, 1983; Mescher and Munaim, 1986). FGF-1 is present in the blastema of the regenerating axolotl limbs (Boilly et al., 1991), and FGFR1 and two types of splicing variants of FGFR2 are expressed in the regenerating newt limbs (Poulin et al., 1993; Poulin and Chiu, 1995). Especially, it is noteworthy that the KGFR-specific probes hybridize with the transcripts mainly in the AEC, whereas the bek-specific probes hybridize with the transcripts mainly in the blastema mesenchyme during the blastema growth period (Poulin and Chui, 1995). It is generally known that FGF-8 specifically activates the bek variant of FGFR2 (FGFR2IIIc) and, thus, may act as a paracrine inducer of underlying mesenchyme (MacArthur et al., 1995, Ornitz et al., 1996). Therefore, our result fits well with the findings of Poulin and Chiu and strengthens the view that the AEC produce mitogenic factor(s). Moreover, Fgf-8 is expressed in the epidermis of regenerating Xenopus tadpole hindlimbs and FGFR inhibitors cause limb deformities in the regenerating Xenopus tadpole limbs (Christen and Slack, 1997; D'Jamoos et al., 1998). The results obtained from our studies reinforce the previous studies on the FGF signaling in the amphibian limb regeneration and will be an asset in understanding the regeneration mechanism of higher vertebrates, including human, which have a limited regeneration capacity.

EXPERIMENTAL PROCEDURES

Experimental Animals and Manipulation

Mexican axolotl (Ambystoma mexicanum) larvae were obtained from the Sogang University Amphibian Colony. Embryos were staged according to Bordzilovskaya et al. (1989) and regenerating limbs of larvae were staged referring to Stocum (1979).

Animals were anesthetized in 0.02% (w/v) ethyl p-aminobenzoate (benzocaine; Sigma) before amputating limbs or collecting limb regenerates. To perform whole-mount in situ hybridization in regenerating limbs, both forelimbs of 3- to 4-cm axolotl larvae were amputated at the elbow. After amputation, animals were returned to water and allowed to regenerate to the desired stage. Regenerates were collected at 1 day, 2 days (dedifferentiation stage), 4 days (early bud stage), 6 days (medium bud stage), 8 days (medium bud stage), 10 days (late bud stage), 12 days (palette stage), and 16 days (digital outgrowth stage) after amputation. For whole-mount in situ hybridization of developing limbs, embryos before hatching (stages 38–42) and between 1 to 14 days after hatching were used. To examine the expression pattern of Fgf-8 during embryogenesis, whole embryos were also collected at neurula stage (stage 15), early tail bud stage (stage 24), middle tail bud stage (stage 32), and late tail bud stage (stages 35). Three to five whole embryos or limb regenerates were sampled at each stage.

Fgf-8 cDNA Cloning and Sequencing

One million plaques of a stage 35 axolotl embryo λ Uni ZAP-XR cDNA library were screened on Hybond N+ nylon membranes (Amersham) in a solution of 50% formamide, 5× Denhardt's solution, 0.5% SDS, 100 μg/ml herring sperm DNA by using the 32P-labeled chicken Fgf-8 cDNA as a probe (kindly provided by Dr. G. Martin). After screening, positive clones were excised and reconstructed to pBluescript recombinant phagemid by in vivo excision according to manufacturer's protocol (Stratagene). Isolated cDNA was sequenced manually or by automated sequencing. The full-length coding sequence of axolotl cDNA was deposited in GenBank (accession number AF190448) and will be released on publication.

Whole-Mount In Situ Hybridization

Digoxigenin-labeled antisense riboprobe was prepared from the 1.26-kb axolotl Fgf-8 cDNA template linearized with EcoRI by using T7 RNA polymerase (Boehringer Mannheim). Whole-mount in situ hybridization was performed essentially as described by Wilkinson (1992) with minor modification. Briefly, hybridization was carried out by using 1 μg/ml of riboprobe in a solution of 50% formamide, 5× SSC (pH 4.5), 1% SDS, 50 μg/ml yeast tRNA, 50 μg/ml heparin at 70°C for 18 hr. Then, the samples were washed under high-stringency condition and incubated with alkaline phosphatase-conjugated anti-digoxigenin antibody (Boehringer Mannheim) at 4°C for 16 hr. The bound probe and antibody complex was detected by using BM purple (Boehringer Mannheim) as substrate of alkaline phosphatase. The limb buds were initially subjected to whole mount in situ hybridization with the DIG-labeled Fgf-8 riboprobe, and then mounted in Paraplast for sectioning.

Acknowledgements

We thank David L. Stocum for helpful suggestions on this study, and Ken Muneoka and Jai-Seob Kim for critical reading and valuable comments on the manuscript. W.-S. Kim was supported by grants from the Korea Science and Engineering Foundation.

Ancillary