Decision letter: Gap junctions composed of connexins 41.8 and 39.4 are essential for colour pattern formation in zebrafish

Abstract

Translate Article

Take Notes

Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods References Decision letter Author response Article and author information Metrics Abstract Interactions between all three pigment cell types are required to form the stripe pattern of adult zebrafish (Danio rerio), but their molecular nature is poorly understood. Mutations in leopard (leo), encoding Connexin 41.8 (Cx41.8), a gap junction subunit, cause a phenotypic series of spotted patterns. A new dominant allele, leotK3, leads to a complete loss of the pattern, suggesting a dominant negative impact on another component of gap junctions. In a genetic screen, we identified this component as Cx39.4 (luchs). Loss-of-function alleles demonstrate that luchs is required for stripe formation in zebrafish; however, the fins are almost not affected. Double mutants and chimeras, which show that leo and luchs are only required in xanthophores and melanophores, but not in iridophores, suggest that both connexins form heteromeric gap junctions. The phenotypes indicate that these promote homotypic interactions between melanophores and xanthophores, respectively, and those cells instruct the patterning of the iridophores. https://doi.org/10.7554/eLife.05125.001 eLife digest The colour patterns that mark an animal's skin, hair, or feathers—called the pigmentation pattern—can be very important for its survival and fitness, helping it to hide from predators or to attract a mate. As a result, there is considerable interest in understanding how genes, proteins, and cells work together to produce the many different pigmentation patterns that exist in the animal world. Adult zebrafish have a characteristic pigmentation pattern of horizontal dark and light stripes on their bodies and fins. There are three types of pigment cell that create this pattern. Xanthophores and iridophores are found all over the body, and the dark stripes also contain melanophore cells. The silvery, reflective iridophores are the first of the cells to populate the skin, giving rise to the first light stripe. They then form a dense network of cells that breaks up to form the darker stripes. However, iridophores are not required to form stripes in the fins, suggesting that patterning occurs differently in the fins and the body. Mutations to a gene called leopard, or leo for short, cause spots to form on the skin of the zebrafish in place of the usual stripes. This gene encodes a member of the connexin family of proteins, which form channels in the membranes that surround cells. These channels—known as gap junctions—allow neighbouring cells to communicate with each other. Each gap junction is made up of two half channels, with one half coming from each neighbouring cells. If the two half channels are identical, the gap junction is known as 'homomeric'; 'heteromeric' gap junctions are made from two different half channels, each consisting of a different connexin protein. The connexin encoded by leo is required for both types of gap junction to form between melanophores and xanthophores. Irion et al. discovered a new mutation to the leo gene that completely disrupts the patterning of the zebrafish. A technique called a genetic screen revealed that the same patterning defects are also seen in the body of zebrafish with mutations to another gene called luchs, which encodes a different connexin protein to the one produced by leo. However, the fins of zebrafish with mutant versions of luchs remain striped. The findings of Irion et al. suggest that heteromeric gap junctions formed from the connexins produced by leo and luchs are important for xanthophores and melanophores to communicate with each other and so form the stripy patterning seen on the body of the zebrafish. The signals transmitted through the gap junctions may also make the iridophores adopt the looser arrangement that is required for the dark stripes to form. As a next step, it will be important to identify the signals that pass through these gap junctions that allow the cells to communicate with their neighbours and establish the pigmentation pattern. https://doi.org/10.7554/eLife.05125.002 Introduction Adult zebrafish (Danio rerio) display a characteristic pattern of horizontal dark and light stripes on their bodies as well as on their anal fins and tailfins (Figure 1A). Three types of pigment cells (chromatophores) are required to create this pattern. In the dark stripes of the trunk, a net of loose or blue iridophores and pale, stellate xanthophores cover the melanophores, the light stripes are composed of dense silvery iridophores covered by compact orange xanthophores (Hirata et al., 2003, 2005; Frohnhofer et al., 2013; Mahalwar et al., 2014; Singh et al., 2014). Figure 1 Download asset Open asset The leo mutant phenotype. Wild-type zebrafish (A) show a pattern of dark and light stripes on the body and on anal- and tail-fins. At higher magnification (A′), dark melanophores in the stripe regions and orange xanthophores in the light stripe regions are discernible. In mutants homozygous for leot1 (B) and heterozygous for leotK3 (C), the stripes are dissolved into spots. Clusters of melanophores are still visible (B′ and C′). Fish homozygous for leotK3 (D) or trans-heterozygous for leotK3 over leot1 (E) show an identical phenotype of a completely dissolved pattern. Individual melanophores that hardly cluster together are still present, mostly associated with blue iridophores (D′ and E′). In (F) a cartoon of Connexin 41.8 is depicted showing the positions of the leo mutations. Gap junctions are composed of two hemi-channels in adjacent cells. Each hemi-channel is made of six connexin subunits, they can be identical (homomeric) or different (heteromeric). In (G) a heteromeric/heterotypic gap junction is schematically shown. An alignment of the amino acid sequence from zebrafish Cx41.8 with its human orthologue, GJA5, is shown in (H). The transmembrane regions are shaded in grey. Newly identified point mutations in Connexin 41.8 are highlighted in red: leotNR16: Y66S; leotNZ: V79M; leot3OJ022: V85M; leotK3: I152F. Already known alleles are highlighted in blue: leotw28: I31F; leot1: R68X; leotq270: I202F. Polymorphisms found in sequences from wild-type fish are highlighted in green: 106 R/K, 136 G/R; 149 V/I. https://doi.org/10.7554/eLife.05125.003 The adult pigmentation pattern is formed during metamorphosis, a period between approximately 3 and 6 weeks of development. At the onset of metamorphosis, iridophores appear in the skin at the region of the horizontal myoseptum that provides a morphological pre-pattern (Frohnhofer et al., 2013). They proliferate and spread as densely connected cells to form the first light stripe. While spreading further ventrally and dorsally, into the regions where the first two dark stripes will form, they change their appearance and become more loose, then they aggregate again at a distance to form the next light stripes (Singh et al., 2013). Larval xanthophores, covering the flank of the fish, start to proliferate and re-organize into densely packed compact cells above the dense iridophores of the light stripe and into more loosely organized, stellate cells in the dark stripe regions (Mahalwar et al., 2014). Melanoblasts migrate along spinal nerves into the skin in the presumptive stripe regions where they finally differentiate and expand to fill the space (Budi et al., 2011; Dooley et al., 2013; McMenamin et al., 2014; Singh et al., 2014). In the fins, stripe formation does not require iridophores, suggesting that the patterning mechanisms in the body and fins are different. A number of mutants are known in which the pattern is not formed normally. In one class of them, one type of pigment cell is absent; in mutants for nacre/mitfA melanophores are missing (Lister et al., 1999), mutations in pfeffer/csf1rA, lead to the lack of xanthophores (Odenthal et al., 1996; Parichy et al., 2000b), and in shady/ltk, rose/ednrb1b, and transparent/mpv17 mutants iridophores are absent or strongly reduced (Parichy et al., 2000a; Lopes et al., 2008; Krauss et al., 2013). In all these cases, the remaining two types of chromatophores form an irregular residual striped pattern. These genes are autonomously required in the respective cell types indicating that interactions among all three chromatophore types are necessary to generate the striped pattern on the trunk of the fish (Maderspacher and Nusslein-Volhard, 2003; Parichy and Turner, 2003; Frohnhofer et al., 2013; Krauss et al., 2014). Based on the analysis of these mutants and on ablation experiments, several attractive and repulsive signals acting over long or short ranges between the chromatophores have been postulated (Maderspacher and Nusslein-Volhard, 2003; Yamaguchi et al., 2007; Nakamasu et al., 2009; Frohnhofer et al., 2013; Patterson and Parichy, 2013; Krauss et al., 2014). In another class of mutants, an abnormal pattern is formed with all three chromatophore types present (Haffter et al., 1996); in these animals, the communication between the cells might be affected. The genes identified in this group encode integral membrane proteins, for example, obelix/Kir7.1, a rectifying potassium channel (Iwashita et al., 2006), seurat/Igsf11, a cell-adhesion molecule of the immunoglobulin superfamily (Eom et al., 2012), or dali/Tetraspanin 3c, a transmembrane-scaffolding protein (Inoue et al., 2012). The best-known example for this class of mutants is leopard (leo), where the stripes are wavy or broken up into a series of dark spots (Figure 1B). The original mutant has been regarded as a separate Danio species (Kirschbaum, 1975; Kirschbaum, 1977; Frankel, 1979). Subsequently several dominant alleles were identified in Danio rerio (Haffter et al., 1996), and it has been shown that the leo phenotype is caused by a mutation in connexin 41.8 (Watanabe et al., 2006), which codes for a subunit of gap junctions (gap junction protein α5, GJA5). Gap junctions are intercellular channels that allow the passage of small molecules and ions between neighbouring cells, and thus are responsible for their chemical and electrical coupling (Kar et al., 2012). They are formed by the juxtaposition of two hemi-channels (connexons), composed of six connexin subunits, in adjacent cells (Unwin and Zampighi, 1980). Connexins are integral membrane proteins with four transmembrane domains, two extracellular loops, one intracellular loop, and intracellular N- and C-termini (Milks et al., 1988) (Figure 1F,G). Hemi-channels can be composed of different subunits; the resulting gap junctions are homotypic, if identical connexons from neighbouring cells pair, or heterotypic, if different hemi-channels come together. In addition to the subunit composition, gap junction conductivity is regulated by a number of different factors, for example, by the intracellular levels of Ca2+, by polyamines, by the membrane potential, or by phosphorylation (Thévenin et al., 2013). It has been reported that the function of the leo gene is required in two of the chromatophore types, the melanophores and the xanthophores, for homotypic and heterotypic cellular interactions (Maderspacher and Nusslein-Volhard, 2003). Wild-type and mutant forms of Cx41.8 artificially expressed in melanophores can lead to different variations of the stripe pattern (Watanabe and Kondo, 2012). This could indicate that gap junctions are responsible for some of the short range signals postulated to occur within and between these two chromatophore types (Parichy and Turner, 2003; Inaba et al., 2012; Frohnhofer et al., 2013). Three alleles of leo have been described so far. The original allele, leot1, is recessive; it has a premature stop codon at position 68 of the coding sequence and is most likely a functional null-allele. Two dominant alleles, leotq270 and leotw28, carry missense mutations; they lead to a stronger phenotype (Haffter et al., 1996; Watanabe et al., 2006). The dominance of these alleles led to the suggestion that heterotypic as well as homotypic connexons containing Cx41.8 could be involved in pigment patterning, postulating the existence of other connexin partner(s) in the potential heterotypic channels (Watanabe et al., 2006). In this study, we identified several additional alleles of leo. All of them are dominant, and the strongest of them, leotK3, leads to a complete loss of the pigmentation pattern in homozygous carriers; the number of melanophores is reduced, and they appear as small groups or individual cells in the skin, embedded in an expanded light region of dense iridophores covered by xanthophores. In a genetic screen, we found two dominant alleles of a gene we named luchs (luc) as enhancers of the leo loss-of-function phenotype. We show that the luc gene codes for another connexin, Cx39.4. A loss-of-function allele for luc displays a phenotype similar to leot1; however, the fins are not affected. Further we show that both genes, leo and luc, are required in the trunk in xanthophores and in melanophores, but not in iridophores. Our results suggest that Cx41.8 and Cx39.4 form heteromeric gap junction channels in the plasma membranes of melanophores and xanthophores. In the complete absence of the channels, iridophores take over and almost fully fill the space normally occupied by alternating light and dark stripes, whereas melanophores are suppressed. This suggests that the heteromeric gap junctions function in the communication between xanthophores and melanophores and potentially in the transduction of signals to the dense iridophores to induce the transition into the loose shape required for dark stripe formation. Results New alleles of leo In a number of experiments designed to find mutants with adult phenotypes, we identified four additional alleles of leo. All of them are dominant. Three of our new alleles result in a phenotypic series of patterns from undulating stripes to breaking up of the dark stripe regions into small spots. As in wild-type (Figure 1A–A′), the dark stripe regions of melanophores are covered with loose blue iridophores. The light stripe regions composed of dense iridophores covered by compact xanthophores are expanded and ingress into the stripe regions. Depending on the genotypes, they display varying strengths, similar to the previously described dominant alleles leotw28 and leotq270 (Haffter et al., 1996; Watanabe et al., 2006). However, one allele, leotK3, is considerably stronger. In heterozygous fish, it leads to an intermediate phenotype similar to the one seen in homozygotes for the loss-of-function allele leot1. Fish homozygous for leotK3, or trans-heterozygous for leotK3 and leot1, show a much stronger phenotype with a loss of any striped arrangement (Figure 1B–E). Instead of dark and light stripe regions, they display variable numbers of small groups of melanophores in islands of blue iridophores, or even single melanophores distributed in an almost even background of dense iridophores covered by xanthophores characteristic for the light stripe regions (Figure 1D′,E′). The stripes in the anal and caudal fins are also affected in leo mutants, weak phenotypes show some short residual melanophore stripes, in homozygous leotK3 animals the pattern is completely lost and only few melanophores are present at the margins of the fins. The newly identified leo alleles carry mis-sense mutations in the coding sequence for Connexin 41.8. The affected amino acid residues are at highly conserved positions in the N-terminal half of the protein (Figure 1F,H); the mutation in leotK3, Ile152Phe, lies within the third transmembrane domain at a position where all connexins invariably have a small hydrophobic residue. Dominant mutations in luc enhance the phenotype of leot1 The fact that the mis-sense mutation in leotK3 leads to a much stronger phenotype than the one seen in fish homozygous for the loss-of-function allele leot1 argues for a dominant negative effect of the mutant protein on an additional component involved in pigment patterning. To identify this component, we performed a genetic screen for dominant mutations that enhance the phenotype of leot1. We mutagenized males homozygous for leot1 with ENU and crossed them to homozygous mutant females. In their offspring, we screened for fish with a stronger phenotype, similar to the one observed in leotK3 mutants. Among 4469 F1 fish, we found three individuals with a strongly enhanced phenotype. One of them turned out to be an allele of obelix (Haffter et al., 1996), which has been shown before to enhance the leo phenotype (Maderspacher and Nusslein-Volhard, 2003). The other two mutants, tXA9 and tXG1 (Figure 2B,E), carry alleles of the same gene, which we named luchs (luc) after the German name for lynx. Both have almost identical phenotypes, heterozygous fish show wavy stripes with some gaps, in homozygotes and in trans-heterozygotes, the stripes are completely dissolved into very small spots and individual melanophores (Figure 2C,D,F,G). Mutants of luc display patterns that are not significantly different from those of the leo phenotypic series, however, the fins are much less affected in our new mutants. To identify the responsible mutations, we used a candidate approach and sequenced the coding sequences of several connexin genes. For both alleles we identified a mis-sense mutation in the gene encoding Connexin 39.4 (cx39.4). In luctXA9 an A to C transversion leads to an amino acid exchange T29P, in luctXG1 a G to T transversion results in W47L, both residues are highly conserved and lie within the first transmembrane domain and the first extracellular loop, respectively (Figure 3A). Figure 2 Download asset Open asset Dominant enhancers of the leot1 phenotype. The spotted pattern of fish homozygous for leot1 (A) is lost when they carry an additional mutant allele for luctXA9 (B) or luctXG1 (E). Fish heterozygous for luctXA9 (C) and luctXG1 (F) have undulating and disrupted stripes on their flanks. In mutants homozygous for luctXA9 (D) or luctXG1 (G) the pattern is almost as strongly disrupted as in leotK3 mutants (compare to Figure 1D). https://doi.org/10.7554/eLife.05125.004 Figure 3 Download asset Open asset luc loss-of-function leads to patterning defects. A comparison of the amino acid sequences of Cx39.4 of zebrafish, encoded by the luc gene, and human Cx37 (GJA4) is shown in (A). The transmembrane domains are shaded in grey, the residues mutated in luctXA9 (T29 P) and luctXG1 (W47 L) are highlighted in red. Note the extension of the N-terminal cytoplasmic domain of Cx39.4 by two amino acid residues. In (B) the beginning of the coding sequence for Cx39.4 is shown, the mutations in luctXA9 and luctXG1 are indicated in red above the sequence; the CRISPR target site is underlined. The sequence traces of wild-type fish and two fish homozygous for deletions of 2 and 22 bp are shown in (C). The predicted amino acid sequences for these mutants are indicated. In (D) a fish homozygous for the 2 bp deletion is shown, the striped pattern on the flank of the fish is disrupted and partly dissolved into spots, the fins are almost normal. Double mutants for leot1 and luc loss-of-function (E) have only very few melanophores on the flank and show an almost uniform pattern of iridophores and xanthophores. https://doi.org/10.7554/eLife.05125.005 luc function in pigment patterning To investigate the role of Cx39.4 in pigment pattern formation in zebrafish, we created loss-of-function mutations in the luc gene using the CRISPR/Cas9 system (Hwang et al., 2013). We targeted the 5′ region of the coding sequence (Figure 3B) and found a high incidence of small deletions leading to frame shift mutations. Most often we found a 3 bp (CTC) deletion accompanied by a 1 bp (G) insertion resulting in a premature stop codon and leading to a truncation of the translated protein after 18 amino acids. We also frequently found a 22 bp deletion resulting in a truncated protein of only 13 amino acids followed by seven unrelated residues (Figure 3C). Fish homozygous or trans-heterozygous for these knock-out alleles (luck.o.) develop an irregular pattern with interrupted stripes and spots very similar to leo mutants. Again, like in the case of the dominant alleles, the patterning of the fins is less affected in the mutants and they show an almost normal arrangement of stripes (Figure 3D). We also reverted the phenotype of heterozygous luctXA9 mutants to wild-type by inducing loss-of-function mutations with the CRISPR/Cas system on the chromosome carrying the dominant allele. This confirms that the luc alleles carry mutations in the Cx39.4 gene. Interestingly, double mutants homozygous for loss-of-function alleles of leo and luc display a phenotype which is indistinguishable from that of homozygous mutants of the dominant leotK3 allele and slightly stronger than that of the homozygous dominant alleles of luc (Figure 3E). This indicates that the dominant negative effect observed in leotK3 is mediated through connexin 39.4, and that in leotK3 both the leo and luc functions are abolished. Our results are most easily explained if the two connexins form heteromeric gap junctions. The development of the leo and luc phenotypes To establish the time point when the phenotype of the leo and luc mutants becomes first apparent, we followed wild-type and mutant fish during the course of metamorphosis. The first signs of the phenotype become visible at stage PR (8.6 mm SSL). When in wild-type, melanophores gather immediately dorsally and ventrally to the dense iridophores of the first light they more in the mutants (Figure iridophores covered with xanthophores in an irregular in the mutants to form a first light into the dark stripe region and melanophores dorsally and ventrally to form irregular (Figure At stage when in wild-type three dark and three light stripes are in the luc and leo loss-of-function mutants the spotted pattern is (Figure and the dominant mutants show of dense iridophores and xanthophores with melanophores (Figure This analysis that the function is required during development and the mutant patterns are not caused by pattern Figure Download asset Open asset The leo and luc phenotypes during metamorphosis. In wild-type dense iridophores of formed become visible during PR and appear in the stripe regions. In luc loss-of-function leot1 luctXG1 and leotK3 mutants iridophores of the first form normally at stage PR however, the melanophores to the first are more In the mutants at stage the to form in the region where dense iridophores appear in irregular at stage irregular stripes and spots become visible in luc loss-of-function and leot1 mutants and in luctXG1 and leotK3 mutants and the number of melanophores is and dense iridophores cover on the flank of the of leo and luc function in chromatophores To the function of leo and luc in the different types of we with mutants one of the three chromatophore types, in which residual striped patterns are (Maderspacher and Nusslein-Volhard, 2003; Frohnhofer et al., 2013). We created double mutants with leotK3, where leo and luc functions are and observed that the residual striped pattern is lost in all three (Figure In mutants, where melanophores are light stripe of dense iridophores covered by xanthophores are and display irregular between regions of blue iridophores characteristic for the dark stripe (Figure Double mutants of with leotK3 show an of dense iridophores covered with xanthophores, in some blue regions are completely lost (Figure In which lack xanthophores, melanophore stripes are broken into of spots by dense iridophores, and melanophores appear in the light stripe regions (Figure If is with leotK3, individual melanophores are over the flank of the fish that are almost covered with dense iridophores (Figure In mutants iridophores, or the first two melanophore stripes are broken into spots, and further stripes are missing Figure for and Figure for In the double mutants with leotK3, only if melanophores are present that are an pattern (Figure We also these double mutant with leot1 or (Figure and observed very slightly double mutant Figure with 1 all Download asset Open asset loss of the pigment pattern in double mutants of and with Mutants one pigment cell and and and and only form of the pigmentation pattern. In mutants of dense iridophores covered with compact xanthophores to light and loose iridophores to dark are as In homozygous double mutants (B) the regions of dense iridophores covered by compact xanthophores expand at the of blue In mutants (C) in the absence of xanthophores the melanophores form in the regions the first light stripe but are more at some distance to it In double mutants of with homozygous leotK3 melanophores are distributed almost as individual cells over the flank of the fish In mutants where iridophores are present, melanophores and xanthophores only partly form the first light and dark stripes In double homozygous double mutants this residual pattern is lost and only very few individual melanophores are on the flank of the fish In the mutants, the strongest phenotypic effect is seen in iridophores, which to appear mostly in the dense form, of the light stripes. However, iridophores not autonomously require the leo or luc function Figure This suggests that is affected xanthophores the phenotype an effect of xanthophores on the of dense melanophores has been in the absence of xanthophores, the dense regions of the light stripe expand and ingress into the stripe breaking it up into spots (Frohnhofer et al., 2013). we that xanthophores mutant for leo and luc have lost their to the of dense iridophores. This the of blue regions in the double mutants, when with Figure 6 with 1 all Download asset Open asset leo gap junctions are required in melanophores and xanthophores but not in iridophores. of animals from are shown. from which can iridophores as their only chromatophore were in which lack most iridophores. The resulting animals in show that mutant iridophores form a wild-type pattern when with wild-type xanthophores and melanophores and When cells are from into with mutant xanthophores in the of the fish, to the in and wild-type iridophores and melanophores, there is wild-type pattern but the chromatophores are almost distributed and in When cells are from into with mutant melanophores and wild-type iridophores and xanthophores in 68 wild-type pattern is but the mutant melanophores form loosely of cells and C′). A for leo and luc in melanophores, as observed in the Figure is by the phenotypes of the double mutants with if leo and luc were only required in xanthophores, the phenotypes of mutants, which lack xanthophores and the double mutants be However, we find an even of melanophores in the double mutants than the series of spots that are still present in (Figure This phenotype suggests a of leo and luc for the homotypic of Cx41.8 and Cx39.4 channels are required in xanthophores and melanophores but not in iridophores