Maintenance of blastemal proliferation by functionally diverse epidermis inregenerating zebrafish fins
Yoonsung Lee a, Danyal Hami a, Sarah De Val b, Birgit Kagermeier-Schenk c, Airon A. Wills a, Brian L. Black b,Gilbert Weidinger c, Kenneth D. Poss a,⁎a Department of Cell Biology, Duke University Medical Center, Durham, North Carolina 27710, USAb Cardiovascular Research Institute, Department of Biochemistry and Biophysics, University of California, San Francisco, San Francisco, California 94158, USAc Biotechnology Center and Center for Regenerative Therapies, 01307 Dresden, Germany
Appendage regeneration in salamanders and fish occurs through formation and maintenance of a mass ofprogenitor tissue called the blastema. A dedicated epidermis overlays the blastema and is required for itsproliferation and patterning, yet this interaction is poorly understood. Here, we identified molecularly andfunctionally distinct compartments within the basal epidermal layer during zebrafish fin regeneration.Proximal epidermal subtypes express the transcription factor lef1 and the blastemal mitogen shh, whiledistal subtypes express the Fgf target gene pea3 and wnt5b, an inhibitor of blastemal proliferation. Ectopicoverexpression of wnt5b reduced shh expression, while pharmacologic introduction of a Hh pathway agonistpartially rescued blastemal proliferation duringwnt5b overexpression. Loss- and gain-of-function approachesindicate that Fgf signaling promotes shh expression in proximal epidermis, while Fgf/Ras signaling restrictsshh expression from distal epidermis through induction of pea3 expression and maintenance of wnt5b. Thus,the fin wound epidermis spatially confines Hh signaling through the activity of Fgf and Wnt pathways,impacting blastemal proliferation during regenerative outgrowth.
Regeneration is a dynamic developmental process in which adultanimals reconstruct body parts lost or damaged by injury. Because ofits spectacular nature and therapeutic implications, the regenerationof major appendages in non-mammalian vertebrates like urodeleamphibians and teleost fish has been a subject of scientific inquiry forcenturies. Critical to appendage regeneration is formation andmaintenance of the blastema, a mass of highly proliferative, undiffer-entiated tissue fromwhich new structures derive (Brockes and Kumar,2005; Stoick-Cooper et al., 2007a).
To replace complex tissues of appropriate shape, size, and function,the urodele or teleost blastema is subject to exquisite regulation bymultiple surrounding influences. First, the amputation injury stump iscovered rapidly by a multi-layered epithelium that interacts with andguides the blastema during growth, patterning, and differentiation. Thisregeneration epidermis synthesizes many secreted factors to mediatethis communication, including Fgfs, Shh, Bmps, Activin-βA, anteriorgradient (AG), andWnts (Becket al., 2003; Jazwinskaet al., 2007;Kumaret al., 2007; Lin and Slack, 2008; Poss et al., 2000a,b; Quint et al., 2002;Schnapp et al., 2005; Smith et al., 2006; Stoick-Cooper et al., 2007b).Second, the extracellular environment, important not only for cellular
.
l rights reserved.
scaffolding but also for tethering and release of growth factors, is alteredin a manner conducive to regeneration (Vinarsky et al., 2005). Third,regeneration requires innervation, in particular as a source of blastemalmitogens (Geraudie and Singer, 1985; Singer, 1952; Singer and Craven,1948). In newts, AG protein is released by nerves, stimulates prolifera-tion of cultured blastemal cells, and is sufficient when introduced byplasmid electroporation to rescue regeneration of the limb afterdenervation and amputation (Kumar et al., 2007). Finally, the blastemais neovascularized soon after its formation, a process necessary tosupport its growth (Bayliss et al., 2006; Huang et al., 2003; Rageh et al.,2002). Thus, the appendage blastema is rapidly cradledwithin a diverseenvironment that cultivates regenerative morphogenesis.
A favorable system for dissecting this regulatory niche is theregenerating zebrafish fin(s), complex structures comprised of bone,connective tissue, nerves, blood vessels, epidermis, and pigmentation.Experimentally useful features of zebrafish fin regeneration includethe rapid and reliable nature of regenerative events, the relativesimplicity of epidermal and mesenchymal structures, and the avail-ability of genetic approaches (Poss et al., 2003; Stoick-Cooper et al.,2007a). A highly proliferative blastema is maintained at the end ofeach bony ray and drives regenerative events.
Close inspection of the zebrafish fin blastema has revealed that it iscompartmentalized into a very small, distal region comprised ofmostly non-proliferating cells, and a larger, more proliferativeproximal domain. Trailing closely proximal to and surrounding the
271Y. Lee et al. / Developmental Biology 331 (2009) 270–280
blastema, new osteoblasts, usually referred to as scleroblasts, activelyalign and begin to deposit bone matrix (Nechiporuk and Keating,2002). Because the heterogeneity and lineage decisions of finblastemal cells have not been elucidated, it is possible that proliferat-ing scleroblasts comprise a cellular subpopulation within the blas-tema. There is evidence that mesenchymal compartmentalization iscritical for regeneration, with the adjacent epidermis suspected toinfluence position, size, and mitotic activity of the blastema asregeneration proceeds (Lee et al., 2005; Poss et al., 2002a). Under thismodel, appropriate epidermal guidance factors are maintained indomains that must be continually shifted and/or reestablished duringthe course of regeneration, to influence proliferation distally whilealso facilitating proximal scleroblast patterning. Recent studies haveidentified several developmental regulators synthesized within theregeneration epidermis, with functional data attributing positive ornegative effects to these factors (Kawakami et al., 2006; Laforest et al.,1998; Lee et al., 2005; Poss et al., 2000a; Quint et al., 2002; Stoick-Cooper et al., 2007b). However, it is unknown how such factorsintegrate spatially and temporally to instruct regenerative events.
Here, we demonstrate the emergence of two spatially andfunctionally distinct cellular subtypes within the basal layer of theepidermis during zebrafish fin regeneration. A consequence of theseepidermal territories is to establish a precise domain of Shh synthesisin a proximal portion of the epidermis, guiding events in adjacentmesenchyme. Fgfs provide both positive and negative regulation tohelp localize Hh signaling, activating and maintaining shh in proximalepidermis while restricting its expression from distal epidermis viaRas signaling. Our data indicate that wnt5b is maintained in distalepidermal cells by Fgf/Ras signaling, where its presence has inhibitoryeffects on shh expression. These findings represent a new cellular andmolecular mechanism by which epidermal signals for proliferationand patterning are positioned in regenerating adult tissue.
Materials and methods
Zebrafish and fin amputations
Wildtype or transgenic zebrafish 4–6 months of age of the outbredEkkwill (EK) strain were used for fin amputation studies. Foramputation experiments, fish were anesthetized in either tricaine or2-phenoxyethanol, and one-half of the caudal finwas amputated witha razor blade. After the surgery, animals were returned to waterheated to 26 °C or 33 °C as previously described (Lee et al., 2005).
In situ hybridization
Whole-mount in situ hybridization (ISH) of fin regenerates andembryos was performed as described previously (Poss et al., 2000b),using digoxygenin-labeled probes produced with published DNAtemplates. For quantification of ISH expression domains, signals fromthe second and third rays with respect to the most lateral ray weremeasured using Openlab software, as described (Lee et al., 2005). ISHon cryosections of 4% paraformaldehyde-fixed fins was performed asdescribed (Poss et al., 2002a). To obtain sections of fin regenerates,fins were mounted in 1.5% agarose/5% sucrose and then saturated in30% sucrose overnight at 4 °C. Frozen blocks were consecutivelysectioned at 10 μm and ISH was performed with serial sections. Atleast 8 fin regenerates were assessed for each marker in eachexperiment described.
BrdU incorporation and immunofluorescence
BrdU solution (2.5 mg/ml) was injected intraperitoneally 30 minprior to collection of fin regenerates, which were then fixed inCarnoy's fixative. After fixation, whole-mount fins were stainedwith arat anti-BrdU antibody (Accurate), a rabbit polyclonal anti-H3P
antibody (Upstate Biotechnology), a mouse zn-3 antibody (ZebrafishInternational Resource Center (ZIRC)) or a rabbit polyclonal anti-Lef1antibody (Abmart) to visualize scleroblasts, as previously described(Lee et al., 2005; Poss et al., 2002b). For costains of epidermal markersand BrdU labeling, whole-mount or section ISH was performed first,after which sections were stained for BrdU immunoreactivity asdescribed (Poss et al., 2002a). For visualization of scleroblasts incryosections of fin regenerate samples, the mouse monoclonal zns-5antibody (ZIRC) was used for staining as described (Johnson andWeston, 1995; Poss et al., 2002b).
Transgenic animals
The hsp70:ca-fgfr1 has been described recently (Marques et al.,2008). To generate hsp70:dn-ras and hsp70:v-ras lines, we subcloned adominant negative N17 H-Ras (dn-ras) or constitutively active viral H-Ras (v-ras) behind the zebrafish heat shock protein 70 (hsp70)promoter (Ras cassettes kindly provided by Malcolm Whitman)(Whitman and Melton, 1992). To allow for rapid identification oftransgenic animals, we included a second gene:promoter cassettecomprised of a lens-specific alpha crystallin promoter fused to an EFGP(hsp70:dn-ras) or DsRed-Ex (hsp70:v-ras) reporter gene, as describedpreviously (Marques et al., 2008). The two constructs, hsp70:dn-ras;α-crystallin:egfp and hsp70:v-ras; α-crystallin:dsred, were injected intozebrafish embryos using standard techniques (Higashijima et al.,1997). Heat-shock protocols were as described previously (Lee et al.,2005). For heat-shocks of transgenic embryos, 24 hours post-fertilization (hpf) embryos were placed in room temperature waterwithin a 250 ml flask and transferred to a 38 °C water bath for 40 min.Embryos were collected at 30 hpf for further analysis.
Cyclopamine and SAG treatment
Cyclopamine (Toronto Research Chemicals) was dissolved in 95%ethanol at 70 °C to produce a 10 mM stock. Animals were transferredat 4 days post-amputation (dpa) into 100 ml of 0.5% ethanol or 50 μMcyclopamine solution for 5 or 28 h. BrdU was injected 30 min prior tocollection. To quantify the length between the distal limits of shhexpression and scleroblasts (see Fig. 3G), three serial sections fromeach of six untreated or cyclopamine-treated fins were used. SAG(Calbiochem) was dissolved in aquariumwater and the animals wereincubated in 100 ml of 5 μM SAG solution in a 150 ml beaker for 2 h.Transgenic animals (hsp70:wnt5b, hsp70:dn-fgfr1 and hsp70:v-ras)were given a heat-shock at 4 dpa and transferred to SAG solution at 3 hpost heat-shock. The fish were incubated in the solution for 2 h andBrdU was injected 30 min prior to collection of fins.
Electrophoretic mobility shift assays
Electrophoretic mobility shift assays (EMSAs) were performed asdescribed previously (De Val et al., 2004). A cDNA encoding solely theDNA-binding domain of zebrafish Pea3 protein was transcribed andtranslated from the plasmid pCITE2A-zf-Pea3, using the TNT Quick-coupled Transcription/Translation System (Promega). The Pea3control binding site oligo was adapted as described (Brown andMcKnight, 1992). The sense strand sequences of the wnt5b oligonu-cleotide used for EMSAwerewnt5bWT, 5′-GGGCTAAGAACAGGAAACT-GAGGCTA-3′; wnt5bMT, 5′-GGGCTAAGAACACCAAACTGAGGCTA-3′.
Results
shh, lef1, and wnt5b mark distinct cellular subtypes within the basallayer of the regeneration epidermis
Recent studies of zebrafish fin regeneration have assignedfunctions to a number of factors expressed in the basal epidermal
272 Y. Lee et al. / Developmental Biology 331 (2009) 270–280
layer of regenerating zebrafish fins. First, Lef1 is a transcriptionaltarget of canonical Wnt signaling and a downstream transcriptionalactivator of other Wnt-responsive genes. Wnt/Beta-catenin signalingwas recently shown to have stimulatory effects on fin regeneration(Kawakami et al., 2006; Poss et al., 2000a; Stoick-Cooper et al., 2007b).Second, Shh is an important embryonic morphogen that was reportedto promote both blastemal proliferation and scleroblast patterning(Laforest et al., 1998; Quint et al., 2002). Third, Wnt5b is a presumednon-canonical Wnt ligand with inhibitory effects on blastemalproliferation and regenerative growth, based on gain-of-functionand loss-of-function studies (Poss et al., 2000a; Stoick-Cooper et al.,2007b). Its paralog wnt5a is expressed in a similar pattern (Stoick-Cooper et al., 2007b).
Interestingly, examination of published whole-mount ISH experi-ments performed by us and others suggested to us that there are twodistinct expression domains for lef1, shh, and wnt5b within the basalepidermal layer (Fig. 1A) (Laforest et al., 1998; Poss et al., 2000a). Toconfirm these data in a quantitative manner, we measured the lengthsof expression domains from new whole-mount ISH experiments, inwhich we could readily identify specific rays for measurement (Lee etal., 2005). Through quantification of expression domains from many
Fig. 1. Expression of epidermal regulators in the regenerating zebrafish fin. (A) Whole-mouviolet). (B,C) Quantification of epidermal expression domains measured from either 3 dpalength from distal tip of regenerate to proximal end of ISH signal; colored bars: total length ofa longitudinal section of 3 dpa (D) or 6 dpa (E) fin regenerates. (F,G)Whole-mount staining findicates the proliferative blastema.
animals, we found that distally localized epidermal cells were wnt5b-positive and negative for shh and lef1, while more proximal cellsexpressed mainly the latter two factors (Figs. 1B, D). To determinewhether this epidermal profile is maintained throughout regenera-tion, we also measured expression domains from 6 dpa regenerates, astage at which the blastema has decreased in size and mitotic activityand regeneration begins to slow to completion (Figs. 1F, G) (Lee et al.,2005). Similar epidermal compartments, albeit compacted, were seenin these later-stage regenerates (Figs. 1C, E), indicating that thisorganization is maintained throughout the regenerative process.
To characterize these domains, we performed ISH on adjacent,serial sections from caudal fin regenerates at 3 dpa. This analysisconfirmed two compartments with little overlap, a proximal regionexpressing shh and lef1, and a distal wnt5b-positive compartment(Fig. 2A). To authenticate colocalization of shh and lef1 and theirrestriction from the distal epidermis, we generated an antibodyagainst Lef1 and assessed its localization in fin regenerates from shh:EGFP transgenic zebrafish (Wills et al., 2008). These experimentsconfirmed ISH data, although Lef1 protein localization extendedfurther proximally than EGFP fluorescence driven by the shh promoter(Fig. 2B). Sections of fin regenerate stained for BrdU and lef1, shh, or
nt ISH of 3 dpa fin regenerates (33 °C) for lef1, shh, wnt5b and pea3 (Black arrowheads,(B) or 6 dpa (C) whole-mount ISH-processed fins, plotted as mean±SEM. White bars:ISH expression domain. (D,E) Cartoon representation of data from (B, C), represented asor BrdU (green) and scleroblasts (red) in 3 dpa (F) and 6 dpa (G) fin regenerates. Bracket
Fig. 2. The regenerating zebrafish fin has two epidermal cellular subtypes. (A) ISH forlef1, shh, wnt5b and pea3 in serial sections from a single 3 dpa (33 °C) fin regenerate.Arrows indicate the distal end of the lef1, shh or the proximal end of the wnt5b, pea3signals. Note that section ISH (A) gives slightly different domain representation thanwhole-mount ISH as in Fig. 1A, based on differences in probe penetration and signaldevelopment (Smith et al., 2008). (B) Antibody staining for Lef1 in a section from 3 dpa(33 °C) shh:EGFP fin regenerates. The distal limit of the Lef1 staining (red) is alignedwith the distal limit of Shh expression (green). (C) ISH for shh, wnt5b and antibodystaining for zns-5 tomark scleroblasts, using serial sections from single 3 dpa (33 °C) finregenerate. Black and white arrows indicate the distal end of the shh and zns-5 signal,or the proximal end of the wnt5b signal. Red arrows indicate distal limits of weaksignals, still similar between shh and zns-5, and non-overlapping with wnt5b.
Fig. 3. Pharmacologic manipulation of Hh signaling in ongoing regenerates affectsblastemal proliferation and scleroblast patterning. (A, B) Analysis of BrdU incorporationin 5 dpa fin regenerates (26 °C) of animals treated with 50 μM cyclopamine for 28 h.Cyclopamine treatment dramatically reduced blastemal BrdU incorporation (B), ascompared to vehicle-treated control animals (A; yellow arrowheads indicate blastema).(C, D) Serial sections of vehicle- (C) and cyclopamine-treated fin regenerates (D)stained for shh mRNA localization or scleroblasts (zns-5). Mesenchymal scleroblastsalignwith epidermal shh signals in vehicle-treated regenerates, while they significantlytrail these signals after 28 h cyclopamine treatment. Arrows indicate the distal limit ofdetectable shh or scleroblasts. (E) Quantification of distance between distal limits of shhexpression and patterned scleroblasts. White bar = vehicle; gray bar = cyclopamine(n=11; mean±SEM; Student's t-test, ⁎Pb0.05). (F, G) Images of H3P staining of 4 dpafin regenerates (26 °C) of animals treated with 5 μM SAG or vehicle for 2 h. Activation ofHh signaling by SAG treatment enhanced the number of mitotic blastemal cells (G). (H)Quantification of mitotic counts in SAG- and vehicle-treated animals. White bar =vehicle; black bar = SAG (n=12; mean±SEM; Student's t-test, ⁎Pb0.05).
273Y. Lee et al. / Developmental Biology 331 (2009) 270–280
wnt5b after a 30-minute BrdU labeling period placed the majority ofproliferating blastemal cells adjacent to epidermis positive for each ofthese markers (see Fig. S1 in the Supplementary data). Furthermore,the distal limit of aligned mesenchymal scleroblasts was typicallyadjacent to the distal limits of lef1 and shh expression, as previouslydescribed (Laforest et al., 1998; Poss et al., 2000a), while the proximaldetectable limit of wnt5b expression was distal to aligned scleroblasts(Fig. 2C).
In summary, our data demonstrate two distinct epidermalterritories along the proximodistal axis, surrounding the blastemaand associated with the distal limits of patterned scleroblasts.
274 Y. Lee et al. / Developmental Biology 331 (2009) 270–280
Wnt5b restricts expression of the proliferation and patterning factor shh
Recently, Akimenko and colleagues determined that cyclopamine, anantagonist of Hh signaling, reduced mesenchymal proliferation andstunted regeneration when treated immediately after amputation(Quint et al., 2002). We initiated cyclopamine (50 μM) treatment at4 dpa and found that it alsomarkedly reduced BrdU incorporation in anestablished blastema (Figs. 3A, B). In addition, Quint et al. (2002) foundthat cyclopamine treatment immediately after amputation led toabnormally patterned bone in the stunted regenerates, and that Shhdelivery into regenerates by plasmid injection caused ectopic boneformation. We directly examined the effects of ∼1 day of cyclopaminetreatment during ongoing scleroblast patterning, and found that shhmRNAexpressionwasgrossly unchanged in level and location.However,the distal limit of aligned scleroblasts shifted ∼20 μmproximally duringthis treatment (Figs. 3C–E), suggesting that newpatterning events weredelayed and uncoupled from epidermal shhmRNA expression.
To test the effects of increasing Hh signaling during regeneration,we treated 4 dpa fin regenerateswith 5 μMSmoothened agonist (SAG)for 2 h (Meloni et al., 2006). This treatment increased by ∼98% thenumber of blastemal cells in mitosis, including metaphase, anaphase,and telophase nuclei (Figs. 3F–H). The published and these newfunctional data with Hh pathway modulators do not distinguishbetween potential roles of shh and other Hh ligands; in particular, ihhis detectable in scleroblasts during fin regeneration (Avaron et al.,
Fig. 4. Wnt5b restricts epidermal expression of shh. (A) Whole-mount ISH and representativand hsp70:wnt5b fin regenerates. Ectopic overexpression ofwnt5b diminishes expression of etarget gene pea3. (B, C) Images of BrdU incorporation in 4 dpa fin regenerates of hsp70:wnpartially rescued proliferation in hsp70:wnt5b regenerates. (D) Quantification of BrdU-positivt-test, ⁎Pb0.05).
2006). Nevertheless, they reinforce the idea that Hh signaling iscritical for both proliferation and patterning, and that Shh releaserequires precise regulation during regeneration.
Because the distal epidermal factor wnt5b showed little colocaliza-tionwith shh, and is known to negatively impact blastemal proliferation(Stoick-Cooper et al., 2007b), we hypothesized that it may help restrictshh to the proximal epidermis. To test this idea, we assessed shh and lef1expression 5 h after a single heat-shock in 4 dpa wildtype and hsp70:wnt5b fin regenerates, a protocol that induceswnt5b in basal epidermisandmesenchyme (data not shown). Ectopicwnt5b expressionmarkedlyreduced expression of both lef1 and shh, but had no detectable effect ona more distal basal epidermal marker, the Fgf target gene pea3 (seebelow; Fig. 4A). Remarkably, a brief SAG treatment was able to increaseBrdU incorporation during wnt5b overexpression by 36% compared tovehicle, indicating a partial rescue of blastemal proliferation (Figs. 4B–D). This result implicates shh as a significant target of the inhibitoryeffects of Wnt5b presence in distal epidermal tissue.
Fgf signaling is required for expression of both proximal and distalepidermal regulators
To further investigate the regulation and significance of epidermalsubdivision during regeneration, we examined participation by Fgfsignaling. We and others have demonstrated requirements for Fgfsignaling in formation and proliferation of the zebrafish fin blastema
e sections of epidermal gene expression 5 h after a single heat-shock in 4 dpa wildtypepidermal markers lef1 and shh (arrowheads), while it has no detectable effect on the Fgft5b animals given a heat-shock, followed by SAG or vehicle treatment. SAG treatmente cells in SAG- and vehicle-treated hsp70:wnt5b animals (n=9; mean±SEM; Student's
275Y. Lee et al. / Developmental Biology 331 (2009) 270–280
(Lee et al., 2005; Poss et al., 2000b; Whitehead et al., 2005). The Fgfreceptor fgfr1 is expressed in basal epidermal cells during regenerativeoutgrowth, as are several Fgf target genes like mkp3, sef, and spry4(Lee et al., 2005). Because these target genes enable negative feedbackof the pathway, their potential functions may be difficult to interpret.Therefore, we examined the expression at 3 and 6 dpa of pea3, an Ets-related transcription factor downstream of Fgfr activation and notimplicated in negative feedback. Qualitative and quantitative assess-ment of pea3 revealed expression in distal epidermal tissue, over-lapping mainly with wnt5b in the distal epidermis and less with lef1and shh expression domains (Figs. 1A–E and 2A). Because of theexpression pattern of pea3, we suspected that Fgf signaling helps tomaintain epidermal cell territories.
Fig. 5. Transgenic Fgfr blockadedepletes both positive and negative epidermal regulators. (A)Wshock in 4 dpa wildtype and hsp70:dn-fgfr1 regenerates. The expression of Fgf target genes peashh, and wnt5b (arrowheads). (B, C) Images of BrdU incorporation in 4 dpa fin regenerateproliferation in hsp70:dn-fgfr1 animals. (D) Quantification of BrdU-positive cells in SAG- and v
To determine requirements for Fgfr activation in epidermalspecification, we used a transgenic line that facilitates heat-inducibleexpression of a dominant negative zebrafish Fgfr1 (hsp70:dn-fgfr1).After regeneration was allowed to occur normally for 4 days, a singleheat-shock was applied 5 h prior to collection of fins. This protocol hasbeen shown to strongly inhibit blastemal proliferation and associatedregenerative events (Lee et al., 2005; Yin et al., 2008). In theseexperiments, it severely reduced expression of pea3, as well as that ofa second Ets transcription factor and Fgf target gene with similarexpression, erm. Thus, although pea3 and erm can be regulated bypathways other than Fgf in certain developmental contexts, expres-sion of these genes is largely or wholly dependent on Fgf signalingduring fin regeneration (Fig. 5A).
hole-mount ISH for epidermalmarkers, and representative sections, 5 h after a single heat-3 and erm is markedly reduced during Fgfr inhibition, as are other epidermal markers lef1,s of heat-shocked hsp70:dn-fgfr1 animals treated with SAG. SAG augmented blastemalehicle-treated hsp70:dn-fgfr1 animals (n=12; mean±SEM; Student's t-test, ⁎Pb0.005).
276 Y. Lee et al. / Developmental Biology 331 (2009) 270–280
Notably, we found that brief, transgenic Fgfr inhibition alsodiminished the level of additional epidermal markers lef1, shh andwnt5b (Fig. 5A). Treatment with SAG during Fgfr inhibition partiallyrescued blastemal proliferation defects, to similar extents as it had inhsp70:wnt5b animals (∼27%; Figs. 5B–D). These findings indicate thatFgf signaling, a pathway essential for blastemal proliferation, controlsepidermal gene expression and helps define epidermal territories bymaintaining expression of distal wnt5b as well as proximal shh andlef1. Moreover, they suggest that one of the means by which Fgfsstimulate blastemal proliferation is through maintenance of shhexpression.
Fig. 6. Transgenic Ras activation induces Fgf target genes in fin regenerates. (A) Whole-moushock in 4 dpa wildtype and hsp70:v-ras regenerates. Brief Ras activation increases and expagenes mkp3 and spry4 (arrowheads). (B) Images of whole-mount ISH and representative sfgfr1; hsp70:v-ras animals, collected 5 h after a 38 °C heat-shock. While Fgfr inhibition depletexpression proximally (third from left). Double-transgenic hsp70:dn-fgfr1; hsp70:v-ras regesingle transgenics.
Transgenic increases in Fgf/Ras signaling distalize theregeneration epidermis
Paradoxically, our loss-of-function experiments indicated that Fgfsignaling promotes expression of both positive (shh, lef1) and negative(wnt5b) effectors of blastemal function. To clarify this finding, wesought to test the impact of increased levels of Fgf signaling duringregeneration. First, we created a transgenic line that would facilitateinducible expression of a ligand-independent, constitutively activefgfr1 cassette (hsp70:ca-fgfr1) (Marques et al., 2008; Neilson andFriesel, 1996). Two additional transgenic lines were generated that
nt ISH and representative sections of epidermal gene expression 5 h after a single heat-nds pea3 and erm expression in the basal epidermal layer, as well as the other Fgf targetections of 4 dpa regenerates from hsp70:dn-fgfr1, wildtype, hsp70:v-ras, and hsp70:dn-es pea3 expression from the basal epidermal layer (far left), Ras activation expands pea3nerates (far right) display similar ectopic induction of pea3 expression as hsp70:v-ras
277Y. Lee et al. / Developmental Biology 331 (2009) 270–280
facilitate heat-inducible expression of either a dominant negative(hsp70:dn-ras) or constitutively active (hsp70:v-ras) Ras, a commondownstream component of Fgf signaling in diverse developmentalprocesses (Neuhaus et al., 2003; Poulain et al., 2006; Shinya et al.,2001; Whitman and Melton, 1992). To determine the ability of thesetransgenic strains to modulate Fgf signaling, we gave a single heat-shock to 24 hpf embryos and examined expression 6 h later of the Fgftarget genes pea3 and erm. At this stage, pea3 and erm mRNAsnormally co-localize with fgf8 mRNA in the midbrain–hindbrainboundary, the pharyngeal arches, and the tailbud (Munchberg et al.,1999). As expected, Ras inhibition diminished pea3 and erm expres-sion in a manner comparable to direct Fgfr inhibition. Conversely,
Fig. 7. Transgenic Fgf/Ras distalizes the regeneration epidermis, blocking blastemal prolifewildtype and hsp70:v-ras animals, collected 5 h after a heat-shock. lef1 and shh expressionincorporation (red) and phosphorylated Histone 3 (H3P; green) assessment in wildtype (proliferation (arrowheads) is markedly reduced by brief Ras activation in transgenic regenerfin regenerates from heat-shocked hsp70:v-ras zebrafish treated with SAG or vehicle. SAG paand vehicle-treated hsp70:v-ras animals (n=12; mean±SEM; Student's t-test, ⁎Pb0.05). (Gelement upstream of the zebrafish wnt5b transcription start site. (Left) Excess addition of ntranslated Pea3 DNA-binding domain and a radiolabeled canonical binding site (Pea3 concomplex. A wnt5b oligo containing two nucleotide changes (wnt5bMT) fails to compete. (domain, an interaction competed by excess unlabeled wnt5bWT oligo.
ectopic Fgfr and Ras activation each elevated and expanded pea3 anderm expression (Fig. S2A in the Supplementary data). Thus, we havegenerated three new strains with which to inducibly manipulate Fgf/Ras signaling in zebrafish.
Further experiments focused on the hsp70:v-ras line, as transgeneinducibility was absent in adult fin tissues of hsp70:ca-fgfr1 animals(data not shown). These effects most likely reflect epigenetic silencingas have been reported with other hsp70-driven constructs (Thummelet al., 2006). To examine the impact of ectopic Ras activation duringregeneration, we first assessed Fgf target genes by whole-mount ISHafter only a brief (5-hour) induction of v-ras. The expression of pea3and erm, as well as additional Fgf target genes mkp3 and spry4, was
ration. (A) Whole-mount ISH and representative sections of 4 dpa regenerates fromare abolished during Ras activation, while wnt5b expression is expanded. (B, C) BrdUB) and hsp70:v-ras (C) 4 dpa fin regenerates 5 h after a single heat-shock. Blastemalates, as compared to wildtype regenerates. (D, E) Images of BrdU incorporation in 4 dpartially rescued blastemal proliferation. (F) Quantification of BrdU-positive cells in SAG-) Electrophoretic mobility shift assays of binding by zebrafish Pea3 protein to a bindingon-labeled wnt5b oligo (wnt5bWT) effectively competes for binding between in vitrotrol; Ct). Arrow indicates unbound radiolabeled oligo; double asterisks marks shiftedRight) Radiolabeled wnt5bWT mobility is shifted by the zebrafish Pea3 DNA-binding
278 Y. Lee et al. / Developmental Biology 331 (2009) 270–280
elevated and also extendedmany cell diameters proximally within thebasal epidermis (Fig. 6A). To confirm that these effects on pea3 anderm were downstream of Fgfrs, we heat-shocked hsp70:dn-fgfr1;hsp70:v-ras animals at 4 dpa. In these experiments, double-transgenicregenerates had enhanced and extended pea3 expression that wasindistinguishable from single-transgenic hsp70:v-ras regenerates (Fig.6B). These findings indicate that several Fgf target genes like pea3 anderm are normally regulated through Ras during fin regeneration.
To define the effects of ectopic Ras activation on epidermalterritories, we assayed the subtype-specific markers lef1, shh, andwnt5b. Strikingly, wnt5b expression was expanded many cell dia-meters proximally by this treatment, mimicking effects of Rasactivation on pea3 and erm. By contrast, ectopic Ras activationabolished expression of the proximal epidermal markers lef1 andshh (Fig. 7A). These results indicate that brief Ras activation distalizedthe epidermal profile, with several distal markers expanded at theexpense of proximal markers. To gauge the impact of this transforma-tion, we assessed BrdU incorporation after 5 h of increased Rasactivity. Consistent with what would be predicted based on the loss oflef1/shh and expansion of wnt5b, blastemal proliferation was sharplyreduced (Figs. 7B, C). When hsp70:v-ras regenerates were treatedbriefly with SAG, BrdU incorporation was partially rescued (∼32%),pointing again to shh as an important mitogenic regulatory targetsynthesized in the proximal epidermis (Figs. 7D–F).
Colocalization, loss-of-function, and gain-of-function experimentsdescribed above indicate that wnt5b is regulated by Fgf/Ras signalingduring fin regeneration, a relationship that to our knowledge has notbeen reported in any developmental system. To identify this regulationin a different developmental context, we looked at 30 hpf zebrafishembryos. First, we noted expression of both pea3 and wnt5b in thetailbud and branchial arches, consistent with co-regulation. Then, weexamined the impact onwnt5b expression of brief increases in Fgf andRas signaling in hsp70:ca-fgfr1 and hsp70:v-ras embryos, respectively.In both strains, like pea3 expression, wnt5b expression was stronglyaugmented in rostral and caudal structures compared to wildtypeembryos (Fig. S2B in the Supplementary data). Thus, brief activation ofFgfr or Ras stimulates wnt5b expression in embryonic tissue.
Taken together, our data support a working model in which Fgfsignaling modulates epidermal influences through two importantmechanisms (Fig. 8). First, Fgfs activate expression of the criticalproliferation and patterning factor shh in the proximal epidermis,
Fig. 8. Model for control of epidermal compartments during fin regeneration. Fgfsignaling helps establish expression of lef1 and shh in proximal regions of the basalepidermal layer, supporting morphogenesis in this area. Distally, a second arm of Fgfsignaling acts through Ras to promote wnt5b expression, regulatory events that restrictlef1 and shh expression from distal-most areas of the regeneration epidermis. Gray =bone; red circles = blastema.
through a mechanism that is ostensibly independent of Ras and Pea3/Erm. Second, Fgfs act in distal epidermis to limit shh expression in thisterritory, through a mechanism that involves Ras, Pea3, and thetypically non-canonical Wnt ligand Wnt5b. Because multiple pre-dicted Pea3 recognition sequences are present 5′ of the wnt5b startcodon (data not shown), the latter regulation might involve directregulation of wnt5b transcription by Pea3. Indeed, electromobilityshift assays using in vitro translated zebrafish Pea3 demonstratedefficient binding to a labeled putative binding element ∼2.3 kbupstream of the predicted transcriptional start site, containingnucleotide sequence highly conserved among vertebrates (see Fig.7G and Fig. S3 in the Supplementary data). Future studies to solidify adirect regulatory relationship during regenerationwill require a seriesof new transgenic reporter strains and functional studies. In summary,the Fgf signaling pathway employs both positive and negativeregulation to control important epidermal gene expression duringfin regeneration.
Discussion
Distinct compartments within the fin regeneration epidermis
Amputation of an adult fin or limb causes massive trauma anddisorganization, followed by rapid formation of a wound epidermis. Inorder to regenerate structures of correct pattern and function,intricate spatial and temporal regulation of epidermal niche signalsis necessary. Here, we have identified new cellular and molecularaspects of regulation exerted on the zebrafish fin blastema by itsepidermal covering.
Our experiments reveal new diversity in cells of the basal layer ofthe wound epidermis directly adjacent to the blastema. While distalepidermal cells express the effectors pea3 and wnt5b, more proximalsubtypes express shh and lef1. We postulate that this compartmenta-lization is important for the process of regenerating patterned bonefrom the adjacent mesenchyme. By confining shh to a proximalportion of the regeneration epidermis, its mitogenic effects can betargeted to adjacent mesenchyme, helping to localize the blastema. Inaddition, effects of Shh on scleroblast patterning are restrictedproximally, where bone deposition trails the proliferative blastema.The initial cellular targets of epidermal Shh are thought to be the basalepidermal layer and scleroblasts, areas where the receptor Patched1 isdetectable (Laforest et al., 1998; Quint et al., 2002); tissue-specifictools for gene manipulation in adults will be necessary to dissectfunctions in the different cell types.
A second consequence of positional specification in epidermal cellsis the placement of an inhibitor of blastemal proliferation, Wnt5b, inthe distal portion of the basal epidermal layer. It is likely that thisnegative influence contributes to the non-proliferative properties ofthe very distal portion of the blastema (Nechiporuk and Keating,2002); however, the purpose of this region of mesenchyme isunknown. Interestingly, our data indicate that Wnt5b exerts negativeeffects at least in part through restriction of effectors like lef1 and shhto proximal epidermal cells. Thus, fin regeneration proceeds byestablishing a domain of inhibitory epidermis at the distal tip of theregenerate, regulatory tissue that can help ensure that stimulatoryfunctions of Shh, and possibly Lef1, are maintained in the correctepidermal compartment.
Fgfs regulate epidermal expression domains
A surprising aspect of this model is how Fgfs direct the process. Ourdata indicate that Fgfs have both positive and negative impact onexpression of shh and lef1 in epidermal cells. Initially paradoxical, thisresult can be explained by integrating Pea3 and Wnt5b into theregulatory model. In proximal epidermal cells, Fgfs maintain shhexpression through a mechanism that does not appear to involve Ras
279Y. Lee et al. / Developmental Biology 331 (2009) 270–280
and Pea3. SAG rescue events in this study indicate that maintenance ofshh expression contributes to the positive effects of Fgfs on blastemalproliferation. This is unlikely to be the only mechanism by which Fgfsmaintain blastemal proliferation (Yin et al., 2008). At the same time,Fgfs function in a negative manner within the distal epidermal cells.Here, by maintaining pea3 and erm transcription through Rassignaling, inhibitory signals like Wnt5b and possibly other factorsare induced. Our data indicate that Fgf and Ras signaling inducewnt5btranscription in both embryonic and adult tissue, and suggest thatPea3 can directly regulate wnt5b transcription. Thus, under thismodel, Fgfs released in the regenerate provide both the initiatingproximal signal for shh expression and the distal cutoff. This is likely toresult in the observed tight epidermal domain of shh as regenerativeoutgrowth proceeds, and explains how Hh agonist treatment canpartially rescue the effects of either overabundant or insufficient Fgfsignaling. It is unclear without genetic cell labeling studies whetherthis epidermal domain is maintained during rapid growth throughtranslocation of a committed subpopulation of epidermal cells, orthrough recurrent reprogramming of epidermal cells into newproximal and distal zones.
Wepostulate that differentexpressiondomainsof Fgfrs or Fgf ligandsenable this apparent dual regulation of shh during fin regeneration.Indeed, ISH on serial sections revealed different proximodistal expres-sion domains of fgfrs1–4, with fgfr1 throughout the mesenchymal andepidermal compartments, fgfr2 in distal basal epidermal cells, and fgfr3and fgfr4 in proximal scleroblasts (see Fig. S4A in the Supplementarydata). Also, fgf20a and fgf24, ligands previously localized in regeneratingfins (Poss et al., 2000b; Whitehead et al., 2005), showed differentialexpression in basal epidermal cells. fgf24 was detectable in a restricteddistal domain, while fgf20a occupied a broader proximodistal domain ofbasal layer cells (see Figs. S4B, C in the Supplementary data). These dataawait conditional mutagenesis for functional confirmation, but areconsistent with differential roles for ligand/receptor pairs duringregeneration. Notably, similar dual effects on shh by Fgfs have beenrecently observed during limb development in vertebrate embryos(Mao et al., 2009; Zhang et al., 2009). There, Fgfs have positive effects onshh expression within the zone of polarizing activity at the posteriormargin, but also restrict its expression from distal mesenchyme via ETSfactors Etv4 and Etv5. It is thus likely that this form of bi-directionalregulation of shh by Fgfs is utilized in other examples of embryonic andregenerative development, perhaps with differing intermediary factors.
Acknowledgments
We thank M. Whitman and R. Friesel for plasmids, L. Saunders forhelp with embryo experiments, X. Meng (Abmart, Shanghai) for thezebrafish Lef1 antibody, ZIRC for antibodies, J. Burris and A. Eastes forexcellent animal care, and Poss lab members for comments on themanuscript. We thank J. Klingensmith and E. Tanaka for suggesting Hhpathway agonist experiments. Y.L. was supported by a predoctoralfellowship from the American Heart Association. This work wassupported by grants from NIGMS, NHLBI, Pew Charitable Trusts, andthe Whitehead Foundation to K.D.P.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.ydbio.2009.05.545.
References
Avaron, F., Hoffman, L., Guay, D., Akimenko, M.A., 2006. Characterization of two newzebrafish members of the hedgehog family: atypical expression of a zebrafishindian hedgehog gene in skeletal elements of both endochondral and dermalorigins. Dev. Dyn. 235, 478–489.
Chemical modulation of receptor signaling inhibits regenerative angiogenesis inadult zebrafish. Nat. Chem. Biol. 2, 265–273.
Beck, C.W., Christen, B., Slack, J.M., 2003. Molecular pathways needed for regenerationof spinal cord and muscle in a vertebrate. Dev. Cell 5, 429–439.
Brockes, J.P., Kumar, A., 2005. Appendage regeneration in adult vertebrates andimplications for regenerative medicine. Science 310, 1919–1923.
Brown, T.A., McKnight, S.L., 1992. Specificities of protein–protein and protein–DNAinteraction of GABP alpha and two newly defined ets-related proteins. Genes Dev. 6,2502–2512.
De Val, S., Anderson, J.P., Heidt, A.B., Khiem, D., Xu, S.M., Black, B.L., 2004. Mef2c isactivated directly by Ets transcription factors through an evolutionarily conservedendothelial cell-specific enhancer. Dev. Biol. 275, 424–434.
Geraudie, J., Singer, M., 1985. Necessity of an adequate nerve supply for regeneration ofthe amputated pectoral fin in the teleost Fundulus. J. Exp. Zool. 234, 367–374.
Higashijima, S., Okamoto, H., Ueno, N., Hotta, Y., Eguchi, G., 1997. High-frequencygeneration of transgenic zebrafish which reliably express GFP in whole muscles orthe whole body by using promoters of zebrafish origin. Dev. Biol. 192, 289–299.
Huang, C.C., Lawson, N.D., Weinstein, B.M., Johnson, S.L., 2003. reg6 is required forbranchingmorphogenesis during blood vessel regeneration in zebrafish caudal fins.Dev. Biol. 264, 263–274.
Jazwinska, A., Badakov, R., Keating, M.T., 2007. Activin-betaA signaling is required forzebrafish fin regeneration. Curr. Biol. 17, 1390–1395.
Johnson, S.L., Weston, J.A., 1995. Temperature-sensitive mutations that cause stage-specific defects in zebrafish fin regeneration. Genetics 141, 1583–1595.
Kawakami, Y., Rodriguez Esteban, C., Raya, M., Kawakami, H., Marti, M., Dubova, I.,Izpisua Belmonte, J.C., 2006. Wnt/beta-catenin signaling regulates vertebrate limbregeneration. Genes Dev. 20, 3232–3237.
Kumar, A., Godwin, J.W., Gates, P.B., Garza-Garcia, A.A., Brockes, J.P., 2007. Molecularbasis for the nerve dependence of limb regeneration in an adult vertebrate. Science318, 772–777.
Laforest, L., Brown, C.W., Poleo, G., Geraudie, J., Tada, M., Ekker, M., Akimenko, M.A.,1998. Involvement of the sonic hedgehog, patched 1 and bmp2 genes in patterningof the zebrafish dermal fin rays. Development 125, 4175–4184.
Lee, Y., Grill, S., Sanchez, A., Murphy-Ryan, M., Poss, K.D., 2005. Fgf signaling instructsposition-dependent growth rate during zebrafish fin regeneration. Development132, 5173–5183.
Lin, G., Slack, J.M., 2008. Requirement for Wnt and FGF signaling in Xenopus tadpole tailregeneration. Dev. Biol. 316, 323–335.
Mao, J., McGlinn, E., Huang, P., Tabin, C.J., McMahon, A.P., 2009. Fgf-dependent Etv4/5activity is required for posterior restriction of Sonic Hedgehog and promotingoutgrowth of the vertebrate limb. Dev. Cell. 16, 600–606.
Marques, S.R., Lee, Y., Poss, K.D., Yelon, D., 2008. Reiterative roles for FGF signaling inthe establishment of size and proportion of the zebrafish heart. Dev. Biol. 321,397–406.
Munchberg, S.R., Ober, E.A., Steinbeisser, H., 1999. Expression of the Ets transcriptionfactors erm and pea3 in early zebrafish development. Mech. Dev. 88, 233–236.
Nechiporuk, A., Keating, M.T., 2002. A proliferation gradient between proximal andmsxb-expressing distal blastema directs zebrafish fin regeneration. Development.129, 2607–2617.
Neilson, K.M., Friesel, R., 1996. Ligand-independent activation of fibroblast growthfactor receptors by point mutations in the extracellular, transmembrane, and kinasedomains. J. Biol. Chem. 271, 25049–25057.
Neuhaus, P., Oustanina, S., Loch, T., Kruger, M., Bober, E., Dono, R., Zeller, R., Braun, T.,2003. Reduced mobility of fibroblast growth factor (FGF)-deficient myoblastsmight contribute to dystrophic changes in the musculature of FGF2/FGF6/mdxtriple-mutant mice. Mol. Cell. Biol. 23, 6037–6048.
Poss, K.D., Shen, J., Keating, M.T., 2000a. Induction of lef1 during zebrafish finregeneration. Dev. Dyn. 219, 282–286.
Poss, K.D., Shen, J., Nechiporuk, A., McMahon, G., Thisse, B., Thisse, C., Keating, M.T.,2000b. Roles for Fgf signaling during zebrafish fin regeneration. Dev. Biol. 222,347–358.
Poss, K.D., Keating, M.T., Nechiporuk, A., 2003. Tales of regeneration in zebrafish. Dev.Dyn. 226, 202–210.
Poulain, M., Furthauer, M., Thisse, B., Thisse, C., Lepage, T., 2006. Zebrafish endodermformation is regulated by combinatorial Nodal, FGF and BMP signalling. Develop-ment. 133, 2189–2200.
Quint, E., Smith, A., Avaron, F., Laforest, L., Miles, J., Gaffield, W., Akimenko, M.A., 2002.Bone patterning is altered in the regenerating zebrafish caudal fin after ectopicexpression of sonic hedgehog and bmp2b or exposure to cyclopamine. Proc. Natl.Acad. Sci. U. S. A. 99, 8713–8718.
Rageh, M.A., Mendenhall, L., Moussad, E.E., Abbey, S.E., Mescher, A.L., Tassava, R.A.,2002. Vasculature in pre-blastema and nerve-dependent blastema stages ofregenerating forelimbs of the adult newt, Notophthalmus viridescens. J. Exp. Zool.292, 255–266.
Schnapp, E., Kragl, M., Rubin, L., Tanaka, E.M., 2005. Hedgehog signaling controlsdorsoventral patterning, blastema cell proliferation and cartilage induction duringaxolotl tail regeneration. Development. 132, 3243–3253.
Shinya, M., Koshida, S., Sawada, A., Kuroiwa, A., Takeda, H., 2001. Fgf signalling through
280 Y. Lee et al. / Developmental Biology 331 (2009) 270–280
MAPK cascade is required for development of the subpallial telencephalon inzebrafish embryos. Development. 128, 4153–4164.
Singer, M., 1952. The influence of the nerve in regeneration of the amphibian extremity.Q. Rev. Biol. 27, 169–200.
Singer, M., Craven, L., 1948. The growth andmorphogenesis of the regenerating forelimbof adult Triturus following denervation at various stages of development. J. Exp.Zool. 279–308.
Smith, A., Avaron, F., Guay, D., Padhi, B.K., Akimenko, M.A., 2006. Inhibition of BMPsignaling during zebrafish fin regeneration disrupts fin growth and scleroblastsdifferentiation and function. Dev. Biol. 299, 438–454.
Smith, A., Zhang, J., Guay, D., Quint, E., Johnson, A., Akimenko, M.A., 2008. Geneexpression analysis on sections of zebrafish regenerating fins reveals limitations inthe whole-mount in situ hybridization method. Dev. Dyn. 237, 417–425.
Stoick-Cooper, C.L., Moon, R.T., Weidinger, G., 2007a. Advances in signaling in vertebrateregeneration as a prelude to regenerative medicine. Genes Dev. 21, 1292–1315.
Thummel, R., Burket, C.T., Hyde, D.R., 2006. Two different transgenes to study genesilencing and re-expression during zebrafish caudal fin and retinal regeneration.ScientificWorldJournal 6 (Suppl. 1), 65–81.
Whitehead, G.G., Makino, S., Lien, C.L., Keating, M.T., 2005. fgf20 is essential forinitiating zebrafish fin regeneration. Science 310, 1957–1960.
Whitman, M., Melton, D.A., 1992. Involvement of p21ras in Xenopus mesoderminduction. Nature. 357, 252–254.
Wills, A.A., Kidd III, A.R., Lepilina, A., Poss, K.D., 2008. Fgfs control homeostaticregeneration in adult zebrafish fins. Development. 135, 3063–3070.
