Evx1 is required for joint formation in zebrafish fin dermoskeleton


  • Claus J. Schulte,

    1. Department of Physiology, Development and Neuroscience, Anatomy Building, Cambridge, United Kingdom
    Search for more papers by this author
  • Claire Allen,

    1. MRC Centre for Developmental and Biomedical Genetics, Department of Biomedical Science, University of Sheffield, Sheffield, United Kingdom
    Search for more papers by this author
    • Drs. Allen, England, and Juárez-Morales contributed equally to this manuscript and are listed alphabetically.

  • Samantha J. England,

    1. Department of Physiology, Development and Neuroscience, Anatomy Building, Cambridge, United Kingdom
    Search for more papers by this author
    • Drs. Allen, England, and Juárez-Morales contributed equally to this manuscript and are listed alphabetically.

  • José L. Juárez-Morales,

    1. Department of Biology, Syracuse University, Syracuse, New York
    Search for more papers by this author
    • Drs. Allen, England, and Juárez-Morales contributed equally to this manuscript and are listed alphabetically.

  • Katharine E. Lewis

    Corresponding author
    1. Department of Physiology, Development and Neuroscience, Anatomy Building, Cambridge, United Kingdom
    2. Department of Biology, Syracuse University, Syracuse, New York
    • Department of Biology, Syracuse University, 107 College Place, Syracuse, NY 13244
    Search for more papers by this author


The transcription factor Evx1 is expressed in the joints between individual lepidotrichia (bony ray) segments and at the distal tips of the lepidotrichia in developing zebrafish fins. It is also expressed in the apical growth zone in regenerating fins. However, so far there is no functional evidence that addresses whether Evx1 is required for any aspect of fin development or regeneration. In this study, we use a novel mutation in evx1 to address this. We find that Evx1 is not required for either fin outgrowth or regeneration. All of the fins form normally in evx1 mutants, and there are no significant changes in fin length. In contrast, Evx1 is required for lepidotrichia joint formation during both fin development and regeneration. This is a very specific phenotype as both lepidotrichia hemisegment separations and lepidotrichia bifurcations still form normally in evx1 mutant fins, as do joints in the more proximal endoskeletal radials. Developmental Dynamics 240:1240–1248, 2011. © 2011 Wiley-Liss, Inc.


The homeobox transcription factor Evx1 is expressed in multiple tissues during vertebrate development (Dush and Martin,1992; Niswander and Martin,1993; Thaeron et al.,2000; Borday et al.,2001; Moran-Rivard et al.,2001). In zebrafish (Danio rerio), evx1 is expressed in the central nervous system, the proctodeum and the fins (Thaeron et al.,2000; Borday et al.,2001). In fins, evx1 is expressed in the distal tips of the growing lepidotrichia (bony rays of the dermoskeleton) and in the distal blastema in regenerating fins (Borday et al.,2001). It is also expressed in a segmental pattern along the lepidotrichia in the main body of the fin, which corresponds to the scleroblasts around the intersegmental joints (Borday et al.,2001). This expression pattern suggested to us that Evx1 might play a role in the formation of lepidotrichia segmental joints (also suggested by Borday et al.,2001) and/or fin outgrowth and regeneration. To test these hypotheses, we examined zebrafish fins in evx1 homozygous mutants. We find that fin outgrowth and regeneration are normal in the absence of Evx1. However, all evx1 mutant fins (anal, caudal, dorsal, pectoral, and pelvic) lack lepidotrichia segmental joints.

The skeleton of the zebrafish anal and dorsal fins consists of not just the distal forming lepidotrichia, but also more proximal endoskeletal radials that also segment, in this case into a proximal and a distal radial. These endoskeletal elements form in a distinct manner from the dermoskeletal lepidotrichia. Most notably, the former progress through a cartilaginous stage whereas the latter do not (Geraudie and Landis,1982; Landis and Geraudie,1990; Smith,1994; Hinchliffe,2002; Suzuki et al.,2003; Mari-Beffa et al.,2007). The separation of proximal and distal radials in the endoskeleton of these fins is thought to be homologous to joint formation in amniote limbs (where the bones also progress through a cartilaginous stage) and similar factors are expressed during both of these processes (Crotwell et al.,2001; Crotwell and Mabee,2007). However, before our analysis of the evx1 mutant, it was not clear whether this genetic homology also extended to the lepidotrichia joints. Interestingly, and in contrast to the joints in the dermoskeleton, we find that the proximal and distal endoskeletal radials are separated normally in evx1 mutant fins. This is consistent with the observation that Evx1 is not expressed during joint formation in mouse limbs (Niswander and Martin,1993). We also examined the expression of gdf5. Gdf5 is expressed in the joints of both amniote limbs and zebrafish fin radials and it is required for correct joint formation in mouse limbs (Storm and Kingsley,1996,1999; Francis-West et al.,1999; Settle et al.,2003; Masuya et al.,2007). In contrast, we did not find any expression of gdf5 in lepidotrichia joints. Taken together, these observations are consistent with the idea that the segmentation of lepidotrichia may use a distinct genetic mechanism to the segmentation of the proximal and distal endoskeletal radials and that these two mechanisms of joint formation may have evolved separately.

Two other mutations that affect joint formation and/or positioning in zebrafish lepidotrichia have been described and these both have an effect on fin length. shortfin (sof) mutants have fewer joints and the fins are approximately half the length of equivalent wild-type (WT) fins (Iovine and Johnson,2000; Iovine et al.,2005; Hoptak-Solga et al.,2008; Sims et al.,2009). In contrast, in another-long-fin (alf) mutants most of the joints are positioned further apart and the fins are longer than in equivalent WT fish (van Eeden et al.,1996; Sims et al.,2009). These phenotypes suggest that there may be a close relationship between joint formation and fin length. Therefore, we also examined fin length and the rate of fin growth in our homozygous evx1 mutant fish that lack all lepidotrichia joints. We found that there was no significant effect on fin length and that evx1 mutant fins grow isometrically at a similar rate to WT fins, suggesting that joint formation is not required for the correct control of fin growth.


evx1 Homozygous Mutants Are Viable

We identified a novel mutation in zebrafish evx1 as part of a TILLING project to isolate ENU-induced mutations of specific genes (see Experimental Procedures). The mutant locus (allele evx1i232) has a point mutation (thymine to adenine) in the second exon, which changes a tyrosine codon into an ochre stop codon (Fig. 1). If translated, this would produce a truncated protein that would prematurely terminate after the third amino acid of the DNA-binding homeodomain. Zebrafish evx1 consists of 3 exons and the homeobox is encoded by part of exon 2 and exon 3 (135 and 42 base pairs (bp), respectively). This makes it extremely unlikely that alternative splicing could circumvent the mutation while preserving any DNA binding activity. Therefore, we conclude that this mutation probably eliminates Evx1 function.

Figure 1.

evx1 mutation probably causes a complete loss of function. A: Schematic of evx1 cDNA indicating exon boundaries. Red arrow shows approximate position of point mutation that converts a Thymine to Adenine and, hence, the Tyrosine (red box in B) into a stop codon. B: Amino acid sequence of Evx1, homeobox in green. C:evx1 sequencing results (reverse complement) for wild-type (WT; left), heterozygous (middle) and homozygous mutant (right) zebrafish. Green arrows indicate Thymine (reverse complement = A; WT allele); Red arrows indicate Adenine (reverse complement = T; mutant allele). Black box outlines the codon that produces Tyrosine in WTs and an ochre stop codon in mutants. Note the change in sequence (SYR [AGT TAT CGA] instead of RYR [AGG TAT CGA]) deliberately introduced by mismatch primers during polymerase chain reaction (PCR; see the Experimental Procedures section). D: Schematic of evx1 mutant genotyping. WT allele left, mutant right. Thymine (purple) is substituted for Guanine by means of PCR. This creates a HincII site (underlined) in mutant DNA. Subsequent enzymatic digestion results in two fragments (158 bp and 28 bp) in mutants while leaving WT DNA uncut (186 bp). E: Genotyping results visualized by gel electrophoresis. The 28-bp fragment generated by HincII digest of mutant DNA is not visible.

Evx1 mutant mice are viable and survive to adulthood (Moran-Rivard et al.,2001). To determine if homozygous evx1 mutant zebrafish might be viable, we raised progeny from an incross of identified heterozygous carriers. While the number of homozygous mutant offspring that survive to adulthood is lower than the expected Mendelian ratio (57 [expected 106.75] out of 427; P = 0.00002 [Binominal Proportions Test]), many homozygous mutant fish are fully viable. These mutants appear to have a normal life span (they live for more than 18 months) and both male and female fish are fertile.

Zebrafish Fin Outgrowth, Fin Length, and Fin Regeneration Do Not Require Evx1

As discussed in the introduction, zebrafish evx1 is expressed at the distal ends of developing lepidotrichia and in the distal blastema in regenerating fins (Borday et al.,2001). This suggested to us that evx1 might have a role in fin growth and/or regeneration. However, we find that evx1 homozygous mutants still develop all of their fins (anal, caudal, dorsal, pectoral, and pelvic; Fig. 2B). We also did not observe any obvious differences in fin length between adult evx1 mutants and their WT siblings on either an AB WT background or the lofd2 background, although the evx1 mutant fins often had extensive structural damage, which sometimes makes the caudal fin appear shorter (e.g., see Fig. 2A–D and Supp. Fig. S1E,F, which is available online). When we measured fin length at 35 days postfertilization (dpf; before any structural damage had occurred), we did not find any statistically significant differences between WTs and evx1 mutants in either absolute caudal fin length (WTs = 3.5 ± 0.3 mm; evx1 mutants = 3.4 ± 0.7 mm; P = 0.602; N = 6 [WT], N = 6 [mutant]) or the ratio of caudal fin length to body length (fin/body ratio: WTs = 0.27 ± 0.02; evx1 mutants = 0.26 ± 0.02; P = 0.188; N = 6 [WT], N = 6 [mutant]). Furthermore, when we analyzed the ratio of caudal fin length to body length in evx1 mutants of different ages and sizes, we found that the ratio stays pretty much constant as the fish grow (Fig. 3). This shows that evx1 mutant fins grow isometrically, just as has been previously described for WT fins (Iovine and Johnson,2000; Fig. 3). This suggests that neither Evx1 nor lepidotrichia joint formation are required for normal isometric growth of the fins or correct fin length.

Figure 2.

Evx1 is required for joint formation in dermoskeleton but not bony ray bifurcation. A–L: Lateral views of heterozygous sibling (A,C,E,G,I) and evx1 mutant (B,D,F,H,J–L) adult zebrafish (A–H,K,L) and 28 days postfertilization (dpf) larvae (I,J). A–F: Adult zebrafish in the TL background (which includes the longfin mutation). For a similar analysis in the AB background, please see Supp. Figure S1. C–L: Alcian blue/Alizarin red staining of caudal fins. Anterior left and dorsal top. B,D: evx1 mutants still form fins, but these fins often exhibit structural damage. F,J: evx1 mutants lack dermoskeletal joints. Size differences in (E) and (F) reflect different ray thicknesses along the proximo-distal fin axis. I,J: By 28 dpf, WT larvae have developed 4–5 joints (I) while joints are completely absent in mutants (J). H: Fin ray bifurcations still form in evx1 mutants. G: Note that bifurcations of WT rays start at joints (arrow). K,L: Bone fractures and irregular bone structures in evx1 mutants. Scale bar = 1.2 cm in A,B, 1 mm in C,D, 140 μm in E,F, 100 μm in G–J, 200 μm in K, 260 μm in L.

Figure 3.

evx1 mutant fins grow isometrically. Graph showing the ratio of caudal fin length to total body length for 18 separate evx1 homozygous mutants (blue squares) and 23 wild-type (WT) fish (red triangles) of different sizes. Each blue square or red triangle represents an individual fish. All of these fish were in the AB background. The ratio of fin length to body length stays constant as the evx1 mutants grow, showing that the caudal fin grows isometrically, just like in WT embryos (also see Iovine and Johnson,2000 for WT data).

To determine if fin regeneration is impaired by the absence of Evx1, we amputated (“fin clipped”) caudal fins of evx1 mutants and WTs at 28 dpf. At 7 days postamputation, there was no statistically significant difference between WTs and evx1 mutants in either the absolute length of the regenerated caudal fins (WTs = 2.5 ± 0.2 mm; evx1 mutants = 2.6 ± 0.2 mm; P = 0.493; N = 6 [WT], N = 6 [mutant]) nor in the ratio of caudal fin length to body length (fin/body ratio: WTs = 0.185 ± 0.02, evx1 mutants = 0.184 ± 0.02, P = 0.952; N = 6 [WT], N = 6 [mutant]). Taken together, our results demonstrate that Evx1 is not required for either fin outgrowth or regeneration.

Evx1 Is Required for Dermoskeleton Joints in the Fins

Evx1 is also expressed in thin stripes in scleroblasts that are located at the joints/segmental separations in the fin dermoskeleton (Borday et al.,2001; Fig. 4A). Therefore, we examined whether Evx1 is required for correct formation of these joints. When we examined the skeleton with differential interference contrast (DIC) microscopy or with Alcian blue and Alizarin red staining, we found that homozygous evx1 mutants lack intersegmental joints in the dermoskeleton of all of their fins (anal, caudal, dorsal, pectoral, and pelvic fins; N = 23; Fig. 2D,F,H,J; Supp. Fig. S1B,D; and data not shown). Consistent with these results, we also observed a lack of joints in the regenerated caudal fins of evx1 homozygous mutants (N = 21; Supp. Fig. S1F,H). In contrast, evx1 heterozygotes have normal joints (N = 6; Fig. 2A,C,E,G,I; Supp. Fig. S1A,C,E,G). In homozygous mutants, the individual bony units of the lepidotrichia are fused into long single rays that span the entire fin. Occasionally, we see breaks in the lepidotrichia. These are more common in older fins and we assume that they are fractures that have formed as a result of the decreased flexibility of the “jointless” fins (e.g., see Fig. 2D,K,L). In many cases, these fractures are associated with irregular bone structures where the bones have fused back together (Fig. 2D,K,L).

Figure 4.

Endoskeletal joints and hemisegment separation do not require Evx1. A–D,G,H: Lateral views of wild-type (WT) or heterozygous siblings (A,C,G,H) and evx1 mutants (B,D). E,F: Cross sections through WT (E) and evx1 mutant (F) lepidotrichia. A,B,G,H: Anal fins. C,D: Dorsal fins. E,F: Caudal fins. A,B: 43 days postfertilization (dpf). C,D: 28 dpf. E,F: Adult. G: 22 dpf. H: 35 dpf. A,B: evx1 expression. C–F: Alcian blue/Alizarin red staining. G,H: gdf5 expression. A,B: evx1 is expressed at intersegmental joints in WT siblings (A), but it is not expressed in equivalent regions in evx1 mutant fins (B). Fin endoskeleton is not affected by the evx1 mutation. C,D: The distal radials of dorsal fins separate from the proximal radials (arrows) in both WT siblings (C) and evx1 mutants (D). E,F: Dermoskeletal hemisegments are separated and not fused in WT (E) and evx1 mutant (F) fins. G,H: gdf5 is expressed during joint formation in the endoskeleton (G) but not in the dermoskeleton (H). Arrow in H indicates a joint clearly visible by differential interference contrast microscopy. Scale bar in panel E = 120 μm in A,B, 160 μm in C,D, 34 μm in E,F, 55 μm in G, 100 μm in H.

To determine whether the lack of joints in evx1 mutants is due to a failure of joint formation or joint maintenance, we analyzed homozygous mutants during the initial phase of joint formation (28 dpf; N = 6). In WT caudal fins at this stage, the lepidotrichia have developed 4–6 segments and associated joints. In contrast, evx1 homozygous mutants completely lack joints. This suggests that Evx1 is required for the formation of joints.

In zebrafish fins, individual lepidotrichia segments are further divided into two hemisegments. Therefore, to check that the lack of joints was not due to a general problem of inappropriate bone fusion, we examined these hemisegments in homozygous mutants. We found that, in contrast to the intersegmental joints, the hemisegments of the lepidotrichia are not fused in evx1 mutants (Fig. 4E,F). This suggests that the lack of joints is a specific phenotype.

Lepidotrichia Branching Does Not Require Intersegmental Joints

Individual lepidotrichia branch into bony sister rays at several points along their final length (Quint et al.,2002). We observed that, in at least most cases, the most proximal split (the starting point of the bifurcation) occurs at the proximal end of a lepidotrichia segment (i.e., just after an intersegmental joint; Fig. 2G). This raises the possibility that joints might be required either for bifurcation events to occur, or for their correct positioning. Consistent with this hypothesis, fish heterozygous for the alf mutation have both irregular lepidotrichia segment length and irregular bifurcation (van Eeden et al.,1996). However, we found that, even though evx1 mutants have no joints, bifurcations of individual lepidotrichia are still present. These bifurcations sometimes occur after break points but often they do not (Fig. 2 H,L). This demonstrates that joints are not required for lepidotrichia branching/bifurcations to occur.

To determine whether bifurcations occur in their normal locations in evx1 homozygous mutants, we examined evx1 expression in mutant fins, to see if the bifurcations localized to sites of evx1 expression (sites where joints would normally occur). However, we found that evx1 is not expressed in joint regions in evx1 mutants (Fig. 4B; N = 6 [WT]; N = 6 [mutant]; 37–43 dpf). This further confirms that this mutation leads to a complete loss of Evx1 function, at least in the fin joint regions, and it suggests that either evx1 mutant RNA is degraded or that Evx1 regulates this particular aspect of its own expression. Arguing against the former explanation we still see evx1 expression in other regions of evx1 mutant embryos (data not shown), making it likely that Evx1 is required to maintain its own expression in the joint regions of the dermal fin skeleton.

To determine if Evx1 or joints are required for the correct spacing of bifurcation events we, therefore, measured the distance between the proximal end of the caudal fin and the first bifurcation in WTs and evx1 mutants. No significant difference could be detected (70 dpf; P = 0.6365; N = 6 [WT]; N = 6 [mutant]). At this stage, secondary bifurcations are yet to be established but we have also seen no obvious differences in secondary bifurcations between WTs and evx1 mutants in older adults (data not shown). These results demonstrate that intersegmental joints and/or Evx1 function are also not required for the spacing of lepidotrichia branching events.

Joint Formation in Fin Endoskeleton Does Not Require Evx1

As mentioned in the introduction, the separation of proximal and distal radials in the anal and dorsal fin endoskeleton is thought to be homologous to joint formation in tetrapod limbs (Crotwell and Mabee,2007). However, before this analysis, it was unclear whether dermoskeletal joint formation also uses the same genetic mechanisms. Our discovery that Evx1 is required for lepidotrichia joint formation suggested that this may not be the case, as Evx1 RNA is not expressed during murine limb joint formation (Niswander and Martin,1993). Therefore, we examined whether endoskeletal joints were affected in evx1 homozygous mutants. We analyzed radials of evx1 mutant and WT fins and could not detect any obvious differences in their appearance. Furthermore, all of the radials that we analyzed had undergone separation of their proximal and distal parts in the appropriate manner (Fig. 4D; 100%; N = 6; 28 dpf). This suggests that endoskeletal segmental separation events do not require Evx1.

Joint Marker gdf5 Is Not Expressed in Intersegmental Joints in the Dermoskeleton

To further determine whether dermoskeletal and endoskeletal joints use different genetic mechanisms we examined a key player in endoskeletal joint formation. GDF5 is required for the formation of at least some tetrapod limb joints (Storm and Kingsley,1996,1999; Francis-West et al.,1999; Settle et al.,2003; Masuya et al.,2007) and consistent with the above-mentioned homology, gdf5 expression can be observed in endoskeleton joints in the fins during zebrafish larval development (Crotwell et al.,2001; Fig. 4G). Therefore, we examined gdf5 expression during intersegmental joint formation of the dermoskeleton. We performed in situ hybridization at both 28 dpf and 35 dpf, two stages where larvae have already formed some, but not all, of their joints. However, we did not observe gdf5 expression in the distal (dermoskeletal) part of any of the WT fins (Fig. 4H). This further supports the hypothesis that dermoskeletal joint formation may use different genetic mechanisms to endoskeletal joint formation.


Our results identify a novel function for the transcription factor Evx1 in dermoskeletal joint formation in zebrafish fins. In this study, we demonstrate that Evx1 is not required for fin outgrowth or regeneration but it is required for the formation of dermoskeletal joints. The specificity of this fin phenotype may seem surprising, given that evx1 is also expressed at the distal end of developing and regenerating fins. However, it is possible that evx1 acts redundantly with evx2 during fin outgrowth and regeneration as evx2 is also expressed in these structures (Sordino et al.,1996; Brulfert et al.,1998). In contrast, evx2 is not expressed in the joints (Brulfert et al.,1998). The role of Evx1 in lepidotrichia joint formation is intriguing, given the segmental nature of these structures and the role of the orthologous gene even-skipped in segmentation in Arthropods. This suggests that Evx1 may be required to establish segmental boundaries in zebrafish lepidotrichia, although we cannot rule out the possibility that it functions downstream of this process, directly in joint formation.

Our analysis provides the first example of a mutant that lacks dermoskeletal joints. However, mutations in two other genes are known to affect the positioning of these joints. In shortfin (sof) mutants, which have a partial loss of function of the gap junction protein Connexin43, the fins are approximately half the length of equivalent WT fins (Iovine and Johnson,2000). In sof mutants, joints are positioned closer together, there are fewer joints (and consequently fewer segments) and cell proliferation is also reduced (Iovine and Johnson,2000; Iovine et al.,2005; Hoptak-Solga et al.,2008; Sims et al.,2009). In contrast in another-long-fin (alf) mutants most of the joints are positioned farther apart and the fins are longer (van Eeden et al.,1996; Sims et al.,2009). [Both of these mutations are distinct from lof mutants, where the joints form and are positioned normally, but the mutants grow a larger number of segments (Iovine and Johnson,2000)]. The alf mutant locus has not yet been identified but connexin43 is ectopically expressed in alf mutant fins and knock-down of Connexin43 in alf mutants results in shorter fins (Sims et al.,2009). Of interest, we did not observe any obvious differences in fin length in evx1 mutants, suggesting that joint formation can be regulated independently of fin size. Taken together, this suggests that Alf and Connexin43 both act to regulate joint location and fin length (but neither is required for joint formation). In contrast evx1 is not required for correct fin length but it is required for joint formation.

During the process of fin outgrowth, individual lepidotrichia bifurcate to form additional bony rays. These bifurcations usually occur at the beginning of a segment, i.e. just after a joint. Therefore, we also examined lepidotrichia bifurcations in evx1 mutants, to determine whether bifurcations require either joints or Evx1. We found that bifurcations still occur in evx1 mutants and there is no significant difference in their spacing compared with WT fins. This clearly demonstrates that bifurcation and segmentation of the lepidotrichia are independently regulated.

Our analysis demonstrates that evx1 homozygous mutants are adult viable. By the time that they are adult, the fins of evx1 mutants often have a “ragged” appearance as the result of multiple fractures. These fractures are probably a consequence of the decreased flexibility of “jointless” fins. Consistent with this, these sorts of fractures have also been observed in alf mutants that, as discussed above, often have very long fin ray segments (Sims et al.,2009). Despite these damaged fins, evx1 mutants can still swim, they appear to be as healthy as their WT siblings and both males and females are fertile.

Our results also strongly suggest that joint formation in the fin dermal and endoskeleton use different genetic mechanisms. The endoskeleton in zebrafish does not require Evx1 for joint formation and tetrapod limbs do not express evx1 at the stages that joints form (Niswander and Martin,1993). These homologous joints, instead use a mechanism that requires GDF5. In contrast, joints in the fin dermoskeleton do not express gdf5, but instead require Evx1 for their formation. This is, therefore, a striking case of similar developmental events within the same structure utilizing different genetic pathways.


Fish Lines

Zebrafish (Danio rerio) were maintained on a 14 hr light/10 hr dark cycle at 28.5°C. Age is indicated as days postfertilization (dpf). evx1i232 mutants have a point mutation in the evx1 gene that probably causes a complete loss of function (for more information see Results). We analyzed the evx1 mutant phenotype on different WT backgrounds (AB, TL, and AB/TL hybrids). The TL background is widely considered a WT strain (e.g., see http://zfin.org/cgi-bin/webdriver?MIval=aa-genotypeview.apg&OID=ZDB-GENO-990623-2), even though it contains both the leot1 recessive mutation, which produces spotted pigment patterns in adult fish and the dominant longfind2 (lofd2) mutation. lofd2 zebrafish grow a larger number of fin segments, and consequently lofd2 adult fins are up to 60% longer than fins in similarly sized AB adults (Iovine and Johnson,2000). We found that adult evx1 homozygous mutants could be more easily identified in the lofd2 background, as the longer fins often acquired a “ragged” appearance. However, we did not observe any obvious differences in the fin phenotype of evx1 mutants on either the longer fin (lofd2) or the normal fin length (AB) backgrounds (e.g., cf Supp. Figs. S1E–H and Fig. 2A–F). Notably, in both cases, mutant fins were a similar length to WT siblings of the same background, and all of the fins lacked lepidotrichia joints (for more information see Results).

Bone Visualization

Acid-free Alcian blue/Alizarin Red staining was performed according to Walker and Kimmel, 2006, except that fins were fixed for 3 hr in 4% paraformaldehyde (PFA); washed for 30 min in 100 mM Tris pH7.5/25 mM MgCl2 and only bleached for 10 min.

In Situ Hybridization

In situ hybridization was performed following a protocol from P. Laurenti. Whole larval zebrafish (see figure legends and/or main text for age) were fixed in 4% PFA at 4°C and then dehydrated and stored in methanol at −20°C. Specimens were rehydrated through a methanol series (75% methanol / 25% phosphate buffered saline [PBS]; 50% methanol / 50% PBS; 25% methanol / 75% PBS), for 5 min each. They were then washed for 4 × 5 min in PBT (PBS + 0.1% Tween) and 30 min in Proteinase K (10 μg/ml). The Proteinase K reaction was stopped by washing for 2 × 5 min in glycine (2 mg/ml in PBT). Samples were refixed for 20 min in 4% PFA and washed 5 × 5 min in PBT. Subsequently, an acetylation reaction was performed for 10 min (0.1 M TEA [Triethanolamine], 0.25% acetic anhydride in dH2O) and then samples were washed 2 × 10 min in PBT. The specimens were prehybridized in hybridization solution (50% formamide, 5× standard saline citrate [SSC], 0.1% Tween, 1 M citric acid, 50 μg/ml heparin, 500 μg/ml yeast tRNA) for at least 2 hr at 65°C and incubated with appropriate digoxigenin-labeled probes (in hybridization solution) overnight at 65°C. Samples were then washed at 65°C (10 min each: 75% hybridization solution / 25% 2× SSC; 50% hybridization solution / 50% 2× SSC; 25% hybridization solution / 75% 2× SSC; 100% 2× SSC), followed by further washes at 60°C (2 × 30 min: 2× SSC) and at room temperature (5 min each: 75% 0.2× SSC / 25% PBT; 50% 0.2× SSC / 50% PBT; 25% 0.2× SSC / 75% PBT; 100% PBT). This was followed by blocking (2 mg/ml bovine serum albumin, 2% sheep serum in PBT) for at least 2 hr. Samples were incubated for 2 hr in anti–digoxigenin-AP antibody (1:3,000 in block solution, preabsorbed against embryos overnight at 4°C, Roche) and washed 8 × 15 min (room temperature) + overnight [4°C] in PBT. Samples were then incubated for 3 × 10 min in Levamizol solution (1 mM Levamizol, 100 mM Tris pH 9.5, 50 mM MgCl2, 100 mM NaCl, 0.1% Tween in dH2O) and incubated in nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) solution (20 μl/ml NBT/BCIP stock solution; Roche) in Levamizol solution without Levamizol. To stop the color reaction, samples were washed 2 × 15 min in Levamizol solution (without Levamizol), 3 × 10 min in PBS and stored at 4°C.

For the gdf5 experiments, we used 22 dpf WTs as a positive control (Fig. 4G). The expression around the endoskeletal radials is consistent with published data (Crotwell et al.,2001). Antisense RNA probes were prepared using the following templates: gdf5 (Clement et al.,2008); evx1 (Thaeron et al.,2000).

Initial Mutant Identification

ENU-mutagenesis was performed according to van Eeden et al. (1996) using six consecutive treatments of 3.3 mM ENU. Twelve TL mutagenized males survived the treatment and these were used to create a live F1 library called ENU Library SMTL0604. These F1 fish were housed with equal numbers of males and females in a tank and individual DNA samples were isolated from a fin clip of each fish. The tank number and sex of each fish was recorded, so that once a mutation was found in a DNA sample, only fish of the correct sex needed to be re-screened to identify the fish carrying the mutation. The library had a hit rate of 217308 bp / mutation. This was calculated by sampling the library and sequencing several distinct polymerase chain reaction (PCR) amplicons. The evx1i232 mutation was initially identified in a single female by means of nested PCR and subsequent sequencing of a region of exon 2. PCRs were performed as described in Wienholds et al. (2003). Standard conditions were used (30 cycles of 94°C for 20 sec; 58°C for 30 sec, and 72°C for 1 min). Initial screening was conducted using multiplexed PCRs that amplified products from 4 different genes at the same time. Subsequent screening was performed with just the evx1 exon 2 primers. The primers used (M13 tagged; obtained from Sigma-Genosys) were: Evx-1-c1-4 GGTCTCAGAGTTGGTGTTGC; Evx-1-c1-2 TGTAAAACGACGGCCAGT AATCTCTGAACCTGTCATGG; Evx-1-c1-1 TCATACCCTGTCATTTGTTC; Evx-1-c1-3 AGGAAACAGCTATGACCAT ACAGCATTGGCAAGGATTAC.

Genotyping of Mutants

For the analysis in this study, heterozygous and homozygous mutants were identified using a restriction enzyme test that we developed. Fish were fin-clipped and the clipped fins were incubated in Proteinase K solution (100 μg/ml ProK, 10 mM Tris HCl pH8, 200 mM NaCl, 5 mM ethylenediaminetetraacetic acid [EDTA] pH 8) at 55°C for several hours or overnight. ProK was then inactivated by heating to 100°C for 10 min. Tubes were centrifuged at high speed for 2 min. Supernatant was removed and used at a 1:10 dilution in PCR reactions.

Part of the evx1 gene was amplified by PCR using 35 cycles of 94°C for 30 sec; 60°C for 30 sec; 72°C for 45 sec and the following primers: PrimerFW: ACTATGGCTCTGACCAGATGCGAAGTTA; PrimerRV: GTCAAATAGAATCGGACTTGCAAAGCA.

This PCR introduces a base pair change in close proximity to the mutation (Fig. 1D). This creates a HincII recognition site exclusively in mutant DNA. PCR products were digested overnight with HincII and then run with 0.4% sodium dodecyl sulfate and loading buffer on a 2.5% agarose gel in TBE. This allowed DNA from WT and mutant fish to be distinguished (Fig. 1E). WT DNA fragments remain uncut and are 186 bp long, whereas mutant DNA fragments are cut and are 158 bp (and 28 bp) long.

Fin Clipping

Fish were anesthetized using MS-222 (Sigma-Aldrich) at a concentration of 0.168 mg/ml. For genotyping, caudal fins were cut approximately a third of a fin length away from the fin bud using scissors. DNA was then prepared from clipped fins as described above. For the growth/regeneration experiments, 28 dpf caudal fins were cut as close as possible to the fin bud using a razor blade. Length of regenerated caudal fin and total body length (excluding the caudal fin) were measured 7 days after the clipping procedure.

Photography and Measurements

Pictures were taken using an Olympus SZX16 stereo microscope with Multipublisher 5 camera or Zeiss M1 AxioImager compound scope with AxioCam MRc5 camera; and processed using Adobe Photoshop. Measurements were performed using calibrated measurements in QCapture Pro 6.0 and/or directly using a calibrated eye piece graticule.

Statistical Tests

Statistical calculations, as indicated in the results, were performed in Microsoft Excel and R (http://www. r-project.org/).


Funding was provided by a Royal Society University Research Fellowship and Syracuse University start-up funds awarded to K.E.L. and BBSRC, Cambridge Trust, DAAD, Daimler-Benz-Foundation and Trinity Hall funding to C.J.S. The initial identification of the evx1 mutant allele was funded by MRC Centre Development grant number G0400100. We thank the Sanger Institute Zebrafish Mutation Resource, funded by Wellcome Trust grant number WT 077047/Z/05/Z, for help with initial sequencing and isolation of the mutant. We are also grateful to Murray Hargrave for help with the mutagenesis, Michael Akam, Andrew Gillis and Philip Ingham for interesting and helpful discussions and Adrian McNabb, Tomasz Dyl and Kirsty Scott for help in maintaining fish lines.