fgf20 Is Essential for Initiating Zebrafish Fin Regeneration

See allHide authors and affiliations

Science  23 Dec 2005:
Vol. 310, Issue 5756, pp. 1957-1960
DOI: 10.1126/science.1117637


Epimorphic regeneration requires the presence or creation of pluripotent cells capable of reproducing lost organs. Zebrafish fin regeneration is mediated by the creation of blastema cells. Here, we characterize the devoid of blastema (dob) mutant that fails fin regeneration during initial steps, forms abnormal regeneration epithelium, and does not form blastema. This mutation has no impact on embryonic survival. Dob results from an fgf20a null mutation, Y148S. Fgf20a is expressed during initiation of fin regeneration at the epithelial-mesenchymal boundary and later overlaps with the blastema marker msxb. Thus, fgf20a has a regeneration-specific requirement, initiating fin regeneration, and controlling blastema formation.

Vertebrate regeneration is of scientific and medical interest. Although acute tissue regeneration in humans is limited, other vertebrates possess extraordinary regenerative capabilities. Zebrafish are amenable to genetic analyses and regenerate an impressive array of structures, including spinal cord, optic nerve, heart, and fins (13). Zebrafish fin regeneration is marked by five stages: regeneration epithelialization, mesenchymal disorganization, blastema formation, regenerative outgrowth, and termination. Although genetic analyses have enhanced our understanding of fin regeneration (1, 46), the specific signaling factor(s) that initiate regeneration and blastema formation are unknown.

To discover genes that initiate regeneration, we treated zebrafish with N-ethyl-N-nitrosourea (ENU) and screened adults for mutants (1, 46). We looked for temperature-sensitive (ts) effects on regeneration, because many genes involved in regeneration also function during embryogenesis (1, 47). The dob mutant displayed an early, genetically recessive regeneration block at 2 days post-amputation (dpa) at 33°C (Fig. 1). Mutant fins were covered only by epithelium, whereas wild-type fins grew beyond the amputation plane and later fully regenerated (Fig. 1). Most (72%, 34/47) mutants showed an identical regeneration defect at 25°C (fig. S2A). dob also failed to regenerate pectoral, dorsal, and anal fins. Thus, a regenerative block in dob is observed at both temperatures.

Fig. 1.

dob, a mutant with early fin regeneration block. Regenerating wild-type (wt) and dob fins at 33°C following amputation. Arrows mark amputation plane. At 1 dpa, both wt and dob have covered the amputation with epithelium. By 2 dpa, wt fins have regenerated beyond the amputation plane, whereas dob mutants have not. At 10 dpa, wt has undergone full reconstitution, but dob fails to regenerate. Scale bars, 400 μm.

We expected that dob would also disrupt embryogenesis (1, 46). However, at 33°C, dob viability was comparable with wild type (fig. S1A). Half (23/46) of dob adults developed asymmetric caudal fin lobes when heat-shocked as embryos, yet all wild type (41/41) developed symmetric fin lobes (fig. S1B). The total size of wild-type and dob caudal fins was comparable (fig. S1B). Therefore, there appears to be an incompletely penetrant ts patterning defect in dob (8). Survival of dob adults at 33°C was also comparable to wild type (fig. S1C). These data suggest a regeneration-specific requirement for dob.

To determine the cellular nature of dob regenerative failure, we examined histology of regenerates at 33°C. The first stage of regeneration, formation of regeneration epithelium, appeared abnormal. At 6 and 12 hours post-amputation (hpa), dob regenerates demonstrated a thickened regeneration epithelium (Fig. 2A). Epithelial proliferation levels in dob at 6 and 12 hpa were similar to wild type (fig. S3A). Therefore, thickened regeneration epithelium likely results from aberrant epithelial migration (9, 10).

Fig. 2.

dob fails to initiate fin regeneration and does not form a blastema. Dashed line or arrow marks amputation plane. (A) Hematoxylin-stained sections of caudal fin regenerates. At 6 and 12 hpa, the regeneration epithelium in dob is thicker than wild type (wt). At 18 hpa, disorganization of mesenchyme beneath the regeneration epithelium is apparent in wt fins but was not observed in dob. At 36 hpa, a blastema is seen in wt regenerates, yet dob lacks blastema. b, blastema; emb, epithelial-mesenchymal boundary. (B) lef1 and sparc mark the basal epidermal layer of the regeneration epithelium in wt. At 24 hpa, lef1 and sparc are absent from the basal epithelial layer in dob. No msxb expression is apparent in dob. Violet stain, indicated by red arrow, marks gene expression. (C) 24-hpa fin regenerates stained for BrdU (green) and 4′,6′-diamidino-2-phenylindole (blue). White dashed line marks epithelial-mesenchymal boundary. During blastema formation, mutants had reduced mesenchymal and epithelial proliferation (n = 19). Graph displays the indices of proliferation in mesenchyme and epithelium of wt and dob regenerates. **, P « 0.01; *, P < 0.05. Scale bars, 50 μm (A), 100 μm [(B) and (C)].

To determine whether dob resulted from a primary defect in wound healing, we performed a longitudinal incision along the caudal fin and allowed healing at 33°C. The wild-type response to this injury is nonregenerative, as the wound is covered by epithelium and leaves a slit down the fin. We found no difference in the timing, histochemistry, or bromodeoxyuridine (BrdU) immunohistochemistry of wound-healing between wild type and dob (fig. S3B). The possibility remains that the dob mutant may have a subtle defect in wound-healing not identified by our observations.

To further characterize the dob defect in regeneration epithelialization, we performed in situ hybridization experiments. The wild-type regeneration epithelium at 24 hpa is molecularly and histochemically distinct. Specifically, the basal epithelium in wild-type regenerates consists of ordered linear cuboidal epithelial cells (11). In dob, the basal epithelium lacked the distinctive cuboidal shape and was nonlinear. Lef1, a transcription factor downstream of Wnt (12), and sparc, a matricellular protein (13), both demarcate regeneration basal epithelium. Lef1 and sparc in situ hybridization in dob revealed absent basal epithelial expression (Fig. 2B). Thus, normal formation of a basal regeneration epithelium appears essential for fin regeneration.

At 18 hpa in wild-type regenerates, disorganized mesenchymal cells beneath the amputation plane are considered evidence of dedifferentiation (9, 11). dob mutants did not undergo mesenchymal disorganization (Fig. 2A). In 18 hpa wild-type regenerates, hsp60 is up-regulated in mesenchymal cells destined to form blastema (6). However, dob did not express hsp60 in these cells (fig. S2B). These data suggest a mesenchymal disorganization defect in dob.

Mesenchymal disorganization is followed by cell proliferation, migration, and blastema formation (9, 10). The blastema is a mass of undifferentiated mesenchymal cells that have proliferated beyond the amputation plane to drive fin regrowth. At 36 hpa, wild-type regenerates show proper blastema formation; however, dob is devoid of blastema (Fig. 2A). These data indicate that dob does not initiate fin regeneration and fails to form regeneration epithelium and blastema.

To determine the effect of dob on blastema formation, we performed in situ hybridization experiments. In 24 hpa wild-type fins, msxb marks rudimentary blastema cells, the mesenchyme distal to the amputation plane. At 72 hpa, during regenerative outgrowth, msxb marks distal blastema (7, 9). No msxb expression was apparent in dob at 24 hpa during blastema formation (Fig. 2B). Faint msxb expression was present in dob at 72 hpa at the central tip of mesenchyme (fig. S2C). These msxb-positive cells may represent a later, inadequate attempt at blastema formation. These data demonstrate that dob lacks early msxb expression and does not form blastema.

During blastema formation, mesenchymal cells reenter the cell cycle and begin to proliferate (9, 10). These cells migrate toward regeneration epidermis and form the rudimentary blastema. To further define the mechanism of the dob regenerative defect, we examined DNA replication, through BrdU labeling. At 24 hpa, dob mesenchymal proliferation levels were one-third of wild type, and epithelial proliferation was slightly lower (Fig. 2C). These data indicate that dob fails to form a blastema through an early defect in mesenchymal proliferation.

To identify the dob gene, we raised 2027 zebrafish from dob–/– × dob+/– mapping crosses to adulthood at 25°C, scored for regenerative defects at 33°C, and genotyped these animals (1, 46). Two markers, bef7 and tof24, flanked the 0.2 centimorgan (cM) dob critical region on chromosome 1. fgf20a was the only transcript within this region, genetically excluding neighboring transcripts (Fig. 3A). Syntenic multicontig alignments demonstrated that no transcripts were located between efha2 and fgf20a in human or fugu databases.

Fig. 3.

fgf20a Y148S missense mutation causes dob. (A) Genetic map of dob on chromosome 1. Refined linkage analysis mapped dob to the 0.2 cM region between bef7 and tof24. The only gene between flanking recombinant markers is fgf20a. Numbers above linear map quantify recombination events between dob and linked markers from 2027 meioses. Bef7, fe12, mt254, tof24, and et6 are polymorphic genetic markers between AB and SJD strains identified by random DNA amplification and sequencing within the dob critical region. (B) DNA sequence chromatograms of wild-type, dob+/–, and dob–/– fish. A443C transversion leads to Y148S amino acid substitution. (C) Tyr-148 is conserved across vertebrates and among most zebrafish Fgfs. (D) Fgf secondary β-trefoil structure. Tyr-148 is located in the β9 strand of Fgf20a. (E) Phenotypic classes and frequencies (%) obtained after injection of wild-type fgf20a, Y148S fgf20a, wild-type fgf3, or Y148C fgf3 mRNAs (10 ng/μl) into wild-type embryos. Y148S fgf20a had no effect on the embryo, suggesting loss of function. Wt, normal; p1, head reduction, loss of tail; p2, lysis.

To identify the dob mutation, we performed DNA sequence analysis of fgf20a. We discovered one missense mutation, an adenine-443 to cytosine (A443C) transversion, in the fgf20a gene of dob that converted tyrosine-148 to serine (Y148S) (Fig. 3B). We genotyped 140 dob mutants, 30 dob heterozygotes, and 20 wild-type controls and verified that the A443C transversion cosegregated with the dob phenotype. Dob was isolated in the SJD background, and DNA sequence of fgf20a in wild-type SJD revealed no mutation. Therefore, the A443C transversion in dob was caused by ENU mutagenesis. This point mutation was not found among five commonly used laboratory strains, indicating that A443C is not a polymorphism. These data indicate that the dob phenotype results from fgf20a Y148S.

Fgf20 is a newly identified member of the Fgf family. Fgf20 is overexpressed in cancer cell lines, promotes proliferation and differentiation of myocardial cells, and enhances survival of dopaminergic neurons in the adult brain, functions consistent with a role in regeneration (1416). Y148 exists in the highly conserved Fgf-core domain in the β9 strand of the 12-strand β-trefoil, adjacent to residues (WYN) (Fig. 3, C and D) that make essential Fgf receptor contacts (17). Y148 is invariant across vertebrate species of Fgf20 and most zebrafish Fgfs (Fig. 3C). Thus, Y148 is evolutionarily conserved among species and within the Fgf family.

To determine the effect of Y148S on the activity of Fgf20a protein, we carried out overexpression studies in zebrafish embryos. Injection of wild-type fgf mRNA leads to dorsalization of the embryo and death (18). Injected fgf20a Y148S mRNA failed to recapitulate this phenotype (Fig. 3E). Similar results have been demonstrated for the null fgf3t21142 allele, a Y148C mutation, the same tyrosine mutated in dob (18). These data indicate that Y148S is likely a null mutation and that Y148 is crucial for Fgf function.

To ensure that fgf20a Y148S is responsible for the dob phenotype, we injected wild-type and dob embryos with wild-type fgf20a mRNA. We found that 87.8% of wild-type embryos showed dorsalization and lethality, compared with 53.5% of dob embryos (P = 0.02) (table S1). These data support the view that the dob phenotype results from reduced fgf20a and that fgf20a Y148S causes dob.

To define the timing and pattern of fgf20a expression during fin regeneration, we performed reverse transcriptase polymerase chain reaction and in situ hybridization experiments. fgf20a was expressed as early as 1 hpa. Expression peaked at 6 hpa, gradually declined, peaked again at 24 hpa, then declined (fig. S2D). During initiation of fin regeneration (6 to 12 hpa), fgf20a expression was localized to mesenchyme at the epithelial-mesenchymal boundary (Fig. 4A). fgf20a was later expressed in blastema during early blastema formation (18 hpa) (Fig. 4A). These data indicate that fgf20a is expressed in key regeneration cells during initiation of fin regeneration, consistent with the dob phenotype.

Fig. 4.

fgf20a expression localizes to epithelial-mesenchymal boundary during initiation of fin regeneration. (A) Whole-mount in situ hybridization and sections showing fgf20a expression. During initiation of fin regeneration (6 to 12 hpa), fgf20a expression is localized to mesenchymal cells directly underneath the regeneration epithelium. During early blastema formation (18 hpa), fgf20a is confined to the blastema. (B) During blastema formation (24 hpa), fgf20a and msxb colocalize in blastema cells. fgf20a and msxb expression domains overlap at 72 hpa, when fgf20a is concentrated at the distal tip of msxb-positive distal blastema. Violet stain indicated by red arrow marks expression. Scale bars, 100 μm.

We further characterized the expression of fgf20a during fin regeneration and compared fgf20a and msxb expression. At 24 hpa, msxb marks rudimentary blastema cells. fgf20a expression colocalized with msxb in early blastema cells (Fig. 4B). During regenerative outgrowth, at 72 hpa, msxb expression marks stem cell–like distal blastema cells (6, 9). fgf20a expression was restricted to a subset of msxb-positive distal blastema cells (Fig. 4B). These data indicate that fgf20a and msxb expression overlap during blastema formation and regenerative outgrowth.

We conclude that the dob phenotype is caused by a Y148S mutation in Fgf20a. Data implicating fgf20a in dob include (i) genetic linkage of dob to a 70-kb critical interval on chromosome 1, the presence of fgf20a in this interval, and the absence of other transcripts; (ii) the presence of an Fgf20a missense mutation (Y148S) in a completely conserved amino acid linked with the dob phenotype (P = 4.22 × 10–42, Fisher's Exact test); (iii) the absence of Y148S in five wild-type strains, indicating that this variant is not a polymorphism; (iv) in situ hybridization showing that fgf20a is expressed in mesenchymal cells adjacent to regeneration epithelium as early as 6 hpa; (v) functional studies showing that the Y148S mutation leads to loss of function; (vi) DNA microarray data demonstrating disrupted Fgf signaling in dob (fig. S4A and table S2); and (vii) insensitivity of dob embryos to wild-type fgf20a mRNA overexpression (table S1). Thus, the early regenerative defect observed in dob results from Fgf20a dysfunction.

We have genetically identified a specific growth factor, Fgf20a, that is essential for initiating fin regeneration, regeneration epithelialization, and blastema formation. These findings provide a genetic foothold on the early signaling events of regeneration that will enable further identification of key regeneration genes. This information will broaden our understanding of regenerative mechanisms and may enable regenerative medicine.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S4

Tables S1 and S2


References and Notes

View Abstract

Navigate This Article