1. Developmental Biology and Stem Cells
  2. Genomics and Evolutionary Biology
Download icon

Molecular shifts in limb identity underlie development of feathered feet in two domestic avian species

Research Article
Cited
5
Views
2,992
Comments
0
Cite as: eLife 2016;5:e12115 doi: 10.7554/eLife.12115

Abstract

Birds display remarkable diversity in the distribution and morphology of scales and feathers on their feet, yet the genetic and developmental mechanisms governing this diversity remain unknown. Domestic pigeons have striking variation in foot feathering within a single species, providing a tractable model to investigate the molecular basis of skin appendage differences. We found that feathered feet in pigeons result from a partial transformation from hindlimb to forelimb identity mediated by cis-regulatory changes in the genes encoding the hindlimb-specific transcription factor Pitx1 and forelimb-specific transcription factor Tbx5. We also found that ectopic expression of Tbx5 is associated with foot feathers in chickens, suggesting similar molecular pathways underlie phenotypic convergence between these two species. These results show how changes in expression of regional patterning genes can generate localized changes in organ fate and morphology, and provide viable molecular mechanisms for diversity in hindlimb scale and feather distribution.

https://doi.org/10.7554/eLife.12115.001

eLife digest

Animals ranging from fish to birds display dramatic diversity within and among species; yet remarkably little is known about the genetic and developmental mechanisms that underlie this variation. In birds and their extinct dinosaur relatives, the distribution of scales and feathers on the feet is a highly variable trait.

Different breeds of domestic pigeon all belong to the same species but have feet that can be feathery or scaly to different extents. Classical genetics experiments suggested that only a few genes are involved in the transition from scaled to feathered skin on the feet of pigeons. However, the molecular basis for this transition was unknown.

Domyan et al. set out to identify the genes involved in the transition from scaled to feathered feet by mating different breeds of pigeon in the laboratory and then sequencing the birds’ DNA. They also surveyed the entire DNA sequences of many additional pigeon breeds with and without feathered feet. This combined approach showed that two regions of the pigeon genome have a profound effect on the number and size of feathers on the feet of domestic pigeons. These regions contain genes that are known to play key roles in controlling the growth of a limb and whether it develops into a leg or a wing. In developing pigeon embryos, Domyan et al. found that a gene called Pitx1, which is typically considered a hindlimb gene, is expressed at lower levels in the developing legs of breeds with feathered feet than in a breed with scaled feet. The experiments also showed that Tbx5, a gene that is expressed in the forelimbs of many animals, is expressed abnormally in the embryonic hindlimbs of breeds of pigeon and chicken with feathery feet.

Together, these findings suggest that the hindlimbs of domestic birds with feathery feet are more like wings at the molecular level, which results in them being covered in feathers rather than scales. Future work will now aim to discover the specific DNA sequences that alter the expression of Pitx1 and Tbx5 in feather-footed breeds, and whether the same genes control the foot feathers of other species of birds.

https://doi.org/10.7554/eLife.12115.002

Introduction

In birds, the genetic and developmental mechanisms that control the decision between scale and feather development remain poorly understood. Most birds possess scales on the foot (tarsometatarsus and toes) and feathers elsewhere. Exceptions to this pattern can provide insights into the evolutionary and developmental basis of skin appendage diversity. Some raptors and boreal birds evolved foot feathers instead of scales ('ptilopody'; Danforth, 1919; Lucas and Stettenheim, 1972), but the lack of appendage variation within these species precludes their use as genetic models. Likewise, paravians (birds and their close non-avian theropod dinosaur relatives) and other dinosaurs evolved diverse feather coverings on their legs and feet that sometimes resemble flight-like feathers (Xu et al., 2003; Hu et al., 2009; Turner et al., 2012; Zheng et al., 2013; Foth et al., 2014; Godefroit et al., 2014), but the absence of living specimens preclude mechanistic molecular studies.

In contrast, domestic pigeons (Columba livia) exhibit stunning variation within a single extant species (Shapiro and Domyan, 2013). Most breeds have feet covered by scaled epidermis (wild-type), but scales are replaced by small feathers in grouse (gr) mutants, and by larger feather 'muffs' in birds that also carry mutant alleles at the Slipper (Sl) locus (Doncaster, 1912; Wexelsen, 1934; Hollander, 1937; Levi, 1986) (Figure 1A). In muffed breeds, scutellate scales are generally absent or poorly developed on the feathered epidermis covering the tarsometatarsus and toes, and feathers are surrounded by soft integument. The molecular identities of both gr and Sl are unknown, and additional loci probably control quantitative variation in the muff phenotype. Because both scale-footed and feather-footed pigeon breeds belong to the same species, we can use traditional genetic crosses and whole-genome resequencing to map the genes that control this striking variation (Shapiro et al., 2013; Domyan et al., 2014). Therefore, we can study diversity of the magnitude usually observed among different species without the roadblock of hybrid incompatibility that often eliminates the possibility of genetic mapping studies.

Figure 1 with 1 supplement see all
Two QTL differentiate scale- and feather-footed domestic pigeons.

(A) Common phenotypes of domestic rock pigeon, in order of increasing foot feathering (left to right): scaled, groused, small- and large-muffed. (B-F) QTL scans and effect plots: proportion of tarsus feathered (B,C), number of toe feathers (D,E), and length of toe feathers (F,G). Mean phenotypes ± S.E. are plotted in (C,E,G). For (E) and (G), genotypes at the QTL with the higher LOD score are on the x-axis, and genotypes at the other QTL are inset. See Tables 1 and 2 for detailed QTL statistics. S, allele from scale-footed grandparent; F, allele from feather-footed grandparent.

https://doi.org/10.7554/eLife.12115.003

During development in vertebrates, skin appendages form through interactions between the ectoderm-derived epidermis and the mesoderm-derived dermis, and signals from the dermis determine epidermal appendage fate (Dhouailly, 2009; Hughes et al., 2011). Previous analyses of mutants and gene misexpression in chickens suggest candidates for feathered feet in the Hedgehog (Harris et al., 2002), BMP (Zou and Niswander, 1996; Harris et al., 2002; 2004), Delta-Notch (Crowe et al., 1998), and Wnt (Chang et al., 2004) pathways. Our study of genetic variation and embryonic development in pigeons, however, reveals a surprisingly different mechanism with broad implications for limb identity and patterning.

Results

Two loci of major effect control foot feathering in pigeons

To identify chromosome regions that contribute to feathered feet, we generated an F2 intercross between muffed (Pomeranian pouter) and scaled (Scandaroon) breeds (Figure 1—figure supplement 1). F1 hybrids displayed intermediate foot feathering, demonstrating a semi-dominant inheritance pattern. Scaled, muffed, and intermediate phenotypes were recovered in the F2 population, confirming that a small number of major-effect loci contribute to this trait. Among F2 offspring, digit 3 bore the largest and greatest number of feathers (digit 1, 7.04 ± 5.08 feathers; digit 2, 7.64 ± 7.54; digit 3, 15.46 ± 10.43; digit 4, 6.13 ± 5.70). Using quantitative trait locus (QTL) mapping with 130 F2 offspring genotyped at 3803 polymorphic markers (Broman et al., 2003), we identified two linkage groups (LG11 and LG20) that had significant effects on three different aspects of foot feathering (log10 odds ratio (LOD) > 4.6; Figure 1B–G, Table 1). LG11 had the largest effects on the proportion of the tarsometatarsal epidermis that was transformed from scaled to feathered (LOD = 7.69, percent variance explained (PVE) = 28.4%) and toe feather number (LOD = 6.72, PVE = 21.3%), and a smaller effect on toe feather length (LOD = 8.51, PVE = 15.8%). LG20 had the largest effect on toe feather length (LOD = 20.9, PVE = 52.2%), and a smaller effect on toe feather number (LOD = 5.36, PVE = 16.5%). When toe feather number was analyzed for each digit individually, the same two QTL were identified and had the most pronounced effects on digits 3 and 4 (Table 2). In summary, two major QTL have marked and separable effects on qualitative and quantitative variation in epidermal appendages.

Table 1

Summary of foot feathering QTL.

https://doi.org/10.7554/eLife.12115.005
TraitLGLoc (cM)ScaffoldPositionLODPVEMean ± S.D.
SSSFFF
Proportion tarsus feathered11117799,205,2867.6928.40.46 ± 0.040.58 ± 0.03**0.80 ± 0.04***
Number of toe feathers111247912,325,9776.7321.343.6 ± 8.967.3 ± 6.5*105.3 ± 8.1***
Number of toe feathers2015951,451,1275.3616.545.3 ± 8.178.2 ± 6.5**100.3 ± 9.9***
Toe feather length (mm)20070136,74620.8952.25.2 ± 2.319.9 ± 2.0***37.3 ± 2.5***
Toe feather length (mm)111247912,325,9778.5115.811.4 ± 3.318.7 ± 2.4*28.5 ± 3.0**
  1. LG, linkage group; Loc, genetic location of peak LOD score in centimorgans; S, allele from scaled parent; PVE, percent variance explained; F, allele from feathered parent; LOD, log10 odds ratio. (Welch two sample t-test of means compared to SS homozygote; *p≤ 0.05, **p≤0.005, ***p≤0.0005.)

Table 2

Summary of QTL for numbers of feathers on individual toes.

https://doi.org/10.7554/eLife.12115.006
DigitLGLoc (cM)ScaffoldPositionLODPVEMean ± S.D.
SSSFFF
Digit 2, left foot111487911,624,7015.2020.243.75 ± 1.136.95 ± 1.12*12.19 ± 1.16***
Digit 3, right foot20070136,7467.7124.19.06 ± 1.7017.35 ± 1.48**19.86 ± 1.83***
Digit 3, right foot111487911,624,7016.4719.59.98 ± 1.6015.60 ± 1.62*21.02 ± 1.65***
Digit 3, left foot20070136,74610.7929.848.66 ± 1.5917.06 ± 1.41**21.09 ± 1.86***
Digit 3, left foot111487911,624,7019.4825.529.37 ± 1.5216.26 ± 1.52**21.30 ± 1.57***
Digit 4, right foot2032952,464,7887.3321.362.44 ± 0.907.24 ± 0.75**10.14 ± 1.10***
Digit 4, right foot11118795,475,4746.7819.53.17 ± 0.896.91 ± 0.92*9.95 ± 1.00***
Digit 4, left foot2030952,464,7889.3726.832.44 ± 0.796.63 ± 0.66**9.65 ± 0.96***
Digit 4, left foot111487911,624,7017.5820.833.75 ± 1.146.95 ± 1.13*12.19 ± 1.17***
  1. LG, linkage group; Loc, genetic location of peak LOD score in centimorgans; S, allele from scaled parent; PVE, percent variance explained; F, allele from feathered parent; LOD, log10 odds ratio. (Welch two sample t-test of means compared to SS homozygote; *p≤0.05, **p≤0.005, ***p≤0.0005.)

In parallel to our experimental cross, we performed probabilistic whole-genome scans of allele-frequency differentiation (pFst; see Kronenberg et al., 2014) across a genetically and phenotypically diverse panel of breeds by comparing 15 feather-footed birds (4 groused and 11 muffed) to 28 scale-footed birds (Shapiro et al., 2013). Using this independent approach across breeds, the two most-highly differentiated pFst signals implicate the same genomic regions as the QTL study: genomic scaffold 79 is located on LG11 (p=4.44 x 10–16, genome-wide significance threshold = 2.11 x 10–9), and scaffolds 70 and 95 are adjacent to one another on LG20 (p=9.81 x 10–13) (Figure 2A, Figure 2—figure supplement 1A,B)

Figure 2 with 3 supplements see all
Two regions of genomic differentiation and H3K27ac enrichment distinguish scale- and feather-footed pigeons.

(A) Whole-genome pFst comparisons between genomes of feather-footed and scale-footed pigeons. Scaffolds are ordered by genetic position in a linkage map from an F2 cross (see Figure 1). Dashed line, genome-wide significance threshold. (B) pFst and extended haplotype homozygosity (EHH) plots for region of high differentiation on scaffold 79. Feather-footed birds (n=10, red in EHH plot) homozygous for a 44-kb deletion are differentiated from scale-footed birds (n=28, black) and show extended haplotype homozygosity in this region. Smoothed lines follow a generalized additive model (Wickham, 2009). (C) H3K27ac ChIP-seq enrichment differed significantly between embryonic wing and leg buds of the scale-footed racing homer (RH) in several regions (blue shading), including within a 44-kb interval that is deleted in the muffed Indian fantail (IF; blue arrowheads). This deleted region is orthologous to a known human limb enhancer (hs1473). (D) Selection scans show similar patterns of differentiation on scaffold 70 between muffed (n=11, red in EHH plot) and scale-footed birds (n=28, black). (E) H3K27ac ChIP-seq enrichment differed significantly between leg buds of racing homer and Indian fantail embryos in regions immediately 5’ of Tbx5 (blue shading). Foot and wing drawings modified after Levi (1986).

https://doi.org/10.7554/eLife.12115.007

The peak pFst region on scaffold 79 contained a 44-kb deletion (from 6.719–6.763 Mb) that was homozygous in 10, and heterozygous in 2 of the 15 feather-footed birds (Figure 2B,C; Figure 2—figure supplement 2A). Birds homozygous for the deletion showed elevated levels of haplotype homozygosity relative to scaled birds, a signature of positive selection on this region (Figure 2B). This deletion spans an element orthologous to a known human limb enhancer, hs1473 (Spielmann et al., 2012), which contains active chromatin marks (Cotney et al., 2012) and is bound by the hindlimb-specific transcription factor Pitx1 in the developing mouse hindlimb (Infante et al., 2013) (Figure 2C). The locus was homozygous for the deletion in 35 of 54 additional feather-footed birds from 21 breeds, but was never homozygous in 96 scale-footed birds from 56 breeds (Chi-square, p<0.0001; Figure 2—figure supplement 2B). The 3 feather-footed birds from our whole-genome panel that did not have this deletion (including the male founder of the aforementioned genetic cross implicating this same region) also showed allelic differentiation from scale-footed birds over this interval, suggesting that an additional feathered-foot allele may also exist at this locus (Figure 2—figure supplement 2A).

In contrast to the differentiation signal we observed between scale-footed and all feather-footed birds on scaffold 79, only the muffed birds (more heavily feathered) showed signatures of selection and shared similar haplotypes on scaffold 70 (higher pFst signal than the adjacent scaffold 95) (Figure 2D, Figure 2—figure supplements 2C, 3). Thus, both QTL analyses and whole-genome scans show that mutation on scaffold 79 alone is sufficient for the grouse phenotype (gr locus), and point to scaffold 70 as the major-effect locus for longer toe feathers in birds with muffs (Sl locus) (Figure 1F,G).

Expression of limb outgrowth and identity genes differs between scale-footed and feather-footed breeds

Next, we examined scaffolds 79 and 70 for candidate genes that might control the scale-to-feather transition. The highest pFst peak on scaffold 79 – corresponding to the major-effect QTL on LG 11 for the proportion of tarsometatarsal feathering – was approximately 200 kb upstream of Pitx1, a gene that encodes a homeobox-containing transcription factor that is normally expressed in the vertebrate hindlimb but not the forelimb (Figure 2B). The highest pFst peak on scaffold 70 – corresponding to the major-effect QTL on LG 20 for toe feather length – was 40 kb upstream of Tbx5, a gene that encodes a T-box transcription factor that is normally expressed in the vertebrate forelimb but not the hindlimb (Figure 2D). These regions were especially intriguing because these two genes encode key transcriptional regulators of forelimb (Tbx5) and hindlimb (Pitx1) identity and development (Logan et al., 1998; Logan and Tabin, 1999; Rodriguez-Esteban et al., 1999; Szeto et al., 1999; Takeuchi et al., 1999). For example, misexpression of Pitx1 in the embryonic chick forelimb blocks feather development (Logan and Tabin, 1999), while misexpression of Tbx5 in the early hindlimb field of embryonic chickens is sufficient to induce a partial wing-like transformation, including feather formation on the feet (Takeuchi et al., 1999). In mouse, Pitx1 (but not Tbx5) plays a role in determining limb-type identity (Szeto et al., 1999; Minguillon et al., 2005; DeLaurier et al., 2006), suggesting that the roles of Tbx5 in limb outgrowth and identity have diversified during amniote evolution (Horton et al., 2008).

We did not identify any fixed non-synonymous coding changes in Pitx1 or Tbx5 between scale-footed and feather-footed breeds of pigeon. However, we found striking differences in embryonic hindlimb expression of these genes among three different breeds – racing homer (scale-footed), Indian fantail (small-muffed), and English trumpeter (large-muffed) – at Hamburger-Hamilton stage 25 (HH25; Hamburger and Hamilton, 1951). Pitx1 expression was reduced in both muffed breeds (expression relative to racing homer: Indian fantail 0.75 ± 0.06, p=0.0007; English trumpeter 0.40 ± 0.05, p=0.0007; n = 6 each) and was more severely reduced in the large-muffed English trumpeter (p=0.002) (Figure 3A). Conversely, Tbx5, the forelimb-specific transcription factor, was ectopically expressed in the hindlimbs of both muffed breeds (hindlimb expression relative to racing homer forelimb: racing homer 0.001 ± 0.0004; Indian fantail 0.01 ± 0.008, p=0.0007; English trumpeter 0.14 ± 0.05, p=0.0007; n = 6 each), and was higher in the large-muffed English trumpeter (p=0.002) (Figure 3B). Forelimb expression of Tbx5 was indistinguishable among the three breeds, indicating that upregulation of Tbx5 in feather-footed breeds is restricted to the hindlimb (Figure 3—figure supplement 1A).

Figure 3 with 3 supplements see all
Limb-type gene expression varies among feathered and scaled pigeons and chickens.

(A,B,F,G) qRT-PCR analyses of Pitx1 and Tbx5 expression in HH25 hindlimbs of pigeon (A,B) and chicken (F,G). Boxes span 1st to 3rd quartiles, bars extend to minimum and maximum observed values if within 1.5 times the interquartile range of the box, circles indicate values outside of this range, black line indicates median. **=p<0.01, ***=p<0.001. (C-E, H-J) RNA in situ hybridization for Tbx5 expression in HH25 embryos of racing homer (C), Indian fantail (D), and English trumpeter (E) pigeons; and white leghorn (H), Cochin (I), and silkie (J) chickens. Arrowheads indicate sites of ectopic Tbx5 expression. Scale bar = 1 mm.

https://doi.org/10.7554/eLife.12115.011

We examined expression of additional genes within the two candidate regions at HH25, and found that the Tbx5 paralog Tbx3 was also differentially expressed in both feather-footed pigeon breeds relative to racing homer (Figure 3—figure supplement 1B). This could be due to the fact that Tbx3 is a target of Tbx5 (Mori et al., 2006; Postma et al., 2008), and additional experiments confirm that cis-regulatory changes do not drive this expression difference (see below). The hindlimb-specific transcription factor Tbx4 is not contained in the candidate regions defined by our QTL mapping and genome-wide association studies, but this gene is a downstream transcriptional target of Pitx1 (Logan and Tabin, 1999; Takeuchi et al., 1999; Duboc and Logan, 2011). We therefore compared expression levels of Tbx4 among scaled and feathered breeds at HH25, but found no significant differences at this stage (racing homer 1.00 ± 0.21; Indian fantail 1.00 ± 0.17, p=0.85; English trumpeter 1.05 ± 0.19, p=0.66; n = 6 each) (Figure 3—figure supplement 1C). Thus, the embryonic hindlimbs of muffed pigeons show quantitative expression changes in transcription factors with reciprocal limb expression domains, including the striking downregulation of a key hindlimb identity gene (Pitx1), and the novel expression of a key forelimb-specific gene (Tbx5).

We next analyzed the patterns of Tbx5 and Pitx1 expression in scaled and muffed pigeon embryos at HH25. Interestingly, ectopic hindlimb expression of Tbx5 in muffed embryos was markedly different than its normal forelimb pattern in wild-type pigeons and other vertebrates (Gibson-Brown et al., 1998a; 1998b; Logan et al., 1998; Tamura et al., 1999; Ruvinsky et al., 2000). Tbx5 is typically expressed throughout the mesoderm of the forelimb, but ectopic Tbx5 expression was largely localized to the mesoderm of the proximal and posterior-dorsal hindlimb of the small-muffed Indian fantail (Figure 3D, Figure 3—figure supplement 3B). This domain was further expanded in the large-muffed English trumpeter (Figure 3E, Figure 3—figure supplement 3C), consistent with the quantitative differences in expression between the two breeds (Figure 3B). This domain shows a striking correlation with regions of epidermal transformation, as foot feathers are usually longest and most numerous on the posterior digits (Darwin, 1868; Levi, 1986) (Figure 3—figure supplement 3D–N). In contrast, and despite quantitative expression differences among breeds, Pitx1 had a qualitatively similar expression domain in embryos of scale-footed and feather-footed breeds at this stage (Figure 3—figure supplement 2). Therefore, consistent with the critical role of mesoderm in determining ectodermal fate (Hughes et al., 2011), regionalized ectopic expression of Tbx5 is correlated with enhanced local transformation of epidermal appendages.

Cis-regulatory changes contribute to expression differences in Pitx1 and Tbx5 in muffed pigeons

If cis-acting regulatory mutations are responsible for the differences in Pitx1, Tbx5, and/or Tbx3 expression between embryos of scale-footed and feather-footed pigeons, then differential expression of scaled-foot and feathered-foot alleles should persist in a common trans-acting cellular environment. To test this prediction, we generated F1 hybrid pigeon embryos by crossing an Old Dutch Capuchine (scale-footed) to a fairy swallow (muffed), and measured parent-of-origin allele expression in the hybrid embryonic hindlimb at HH25 (Figure 4A) (Domyan et al., 2014). Consistent with expression differences we observed among breeds (Figure 3A,B), expression of the feathered-foot allele of Pitx1 was approximately 20% lower than the scaled-foot allele (expression of feathered-foot relative to scaled-foot allele: 0.807 ± 0.039, p=0.003, n = 6 embryos), and expression of the feathered-foot allele of Tbx5 was nearly 1600% higher than the scaled-foot allele (relative expression of feathered-foot allele: 15.75 ± 4.69, p=0.002, n = 7 embryos) (Figure 4B,C). In contrast, expression levels of feathered-foot and scaled-foot alleles of Tbx3 were indistinguishable (relative expression of feathered-foot allele: 0.99 ± 0.05, p=0.68, n = 7 embryos) (Figure 4D). These results directly show that cis-acting genetic changes alter expression of feathered-foot alleles of Pitx1 and Tbx5, but not Tbx3, in the embryonic pigeon hindlimb.

Allele-specific expression (ASE) assays demonstrate cis-regulatory changes in Pitx1 and Tbx5.

(A) Schematic of ASE assay using Tbx5 expression as an example. Differences in Tbx5 expression between scale-footed and feather-footed breeds could be due to trans- and/or cis-acting mutations. If expression differences between parental breeds are due to trans changes only (stars), then expression of the two Tbx5 alleles in hybrid embryos will be the same (top right). In contrast, if cis-regulatory changes underlie differences in Tbx5 expression between the parental breeds, then expression of the two Tbx5 alleles in hybrid embryos will be different (bottom right). (B-D) ASE assay in hybrid hindlimb buds indicate cis-regulatory divergence between scale-footed (Old Dutch Capuchine) and muffed (fairy swallow) pigeon breeds in Pitx1 (B) and Tbx5 (C), but not in Tbx3 (D). Dashed blue line indicates null hypothesis of equal expression of alleles. (E) ASE assay in hybrid hindlimb buds indicate cis-regulatory divergence in Tbx5 between feather-footed (silkie) and scale-footed (white leghorn) chicken breeds. Boxes in (B-E) span 1st to 3rd quartiles, bars extend to minimum and maximum observed values if within 1.5 times the interquartile range of the box, circles indicate values outside of this range, black line indicates median. **p≤0.01 ***p≤0.005.

https://doi.org/10.7554/eLife.12115.016

Patterns of open chromatin differ between scale-footed and feather-footed pigeon breeds

We next performed genome-wide comparisons of differential enhancer activity in embryonic limbs of racing homers and Indian fantails, using H3K27ac as a marker for open chromatin (Creyghton et al., 2010). Strikingly, two regions of significantly different enrichment in racing homer hindlimbs relative to forelimbs were within the 44-kb genomic region that is deleted in the Indian fantail, and one of these regions (log10 likelihood ratio = 4.93) is adjacent to the hs1473 limb enhancer (Figure 2C). Furthermore, the most significant differentially enriched region in Indian fantail hindlimbs relative to racing homer hindlimbs was directly upstream of Tbx5 (log10 likelihood ratio = 10.3) (Figure 2E). The overlapping patterns of enrichment in Indian fantail hindlimbs and wild-type forelimbs suggest that ectopic hindlimb expression of Tbx5 could be due to de-repression of forelimb-specific enhancers. In summary, the differential expression of Pitx1 and Tbx5 among pigeon breeds (Figure 3) and between alleles (Figure 4) is also reflected by differential chromatin activation at these genes.

Muffed pigeon breeds incur musculoskeletal patterning changes

In mouse and chicken embryos, experimental manipulation of Pitx1 and Tbx5 expression can result in muscular and skeletal abnormalities. Experiments in both chick and mouse consistently demonstrate that Pitx1 plays a necessary (but not sufficient) role in determining hindlimb-type morphology of the skeleton, muscles, and tendons (Logan and Tabin, 1999; Takeuchi et al., 1999; Minguillon et al., 2005; DeLaurier et al., 2006; Duboc and Logan, 2011). Experimentally induced ectopic expression of Tbx5 in the hindlimbs of chick embryos can also induce muscular and skeletal anomalies, although Tbx5 does not directly control limb skeletal patterning or determine forelimb-type morphology in mice (Rodriguez-Esteban et al., 1999; Takeuchi et al., 1999; Minguillon et al., 2005Hasson et al., 2007) However, normal patterning of limb muscles and tendons is dependent on Tbx5 and Tbx4 in mice (Hasson et al., 2010). These apparent discrepancies between mammalian and avian systems point to subtle differences in limb development in different lineages (Horton et al., 2008).

Given the dramatic musculoskeletal defects observed in other organisms with experimentally altered Pitx1 and Tbx5 expression, we compared the hindlimb morphology of adult feral pigeons (scale-footed, n=2) to that of the English trumpeter (muffed, n=2) and the Pomeranian pouter (muffed, n=1). We found consistent soft-tissue patterning defects in both feather-footed pigeon breeds: the fibularis longus (FL) tendon inserts on the dorsal rather than ventral surface of the proximal tarsometatarsus, the flexor perforans et perforatus (FPP3) muscle adopts a longitudinal rather than pennate fiber orientation, and a slip of the FPP3 fuses with the FL tendon (Figure 5A). These changes are aberrations of normal patterning, although they are not necessarily clear transformations to a more forelimb-like configuration. We also found that the fibula, which is normally splint-like and shorter than the tibia in pigeons, was enlarged (Figure 5D,E) and two phalanges of digit 4 were fused in feather-footed breeds (not shown). These are not necessarily limb-type transformations, either. However, experimental ectopic expression of Tbx5 in the hindlimbs of chick embryos produces an enlargement of the fibula reminiscent of extreme pigeon phenotypes, and Takeuchi et al., 1999 compared this morphology to a forelimb-like condition (the fibula “makes a joint at its distal end like a normal ulna [the corresponding postaxial zeugopod bone of the forelimb],” p. 810). Notably, all of the modified structures of ptilopodous pigeons develop in the posterior (lateral in the adult) and dorsal hindlimb, which are the primary sites of ectopic Tbx5 expression. Thus, the morphological changes to the hindlimbs of feather-footed pigeon breeds are considerably more than skin deep.

Muffed pigeons have re-patterned hindlimb musculoskeletal system and wing-like foot feathers.

(A,B,C) Gross muscle morphology of scale-footed (feral) and muffed (Pomeranian pouter) left hindlimbs, dorsal view. (A) Skin and superficial muscles have been removed to reveal re-patterning of the fibularis longus (FL, red). Dashed black line, approximate position of ankle joint axis. (B,C) The FPP3 is a pinnate muscle in scale-footed pigeons (B), but a slip of fibers fuses with the adjacent FL in muffed pigeons (arrowhead in C). (D,E) X-ray computed tomography images of scale-footed (feral, right leg) and muffed (English trumpeter, left leg) hindlimbs. Arrowheads mark the proximal and distal ends of the fibula. The wild-type pigeon fibula (D) is short and splint-like. In the muffed bird (E), the fibula extends from the knee to the ankle. We observed distal elongation of the fibula in another muffed breed (fairy swallow) but the fibula did not completely extend to the ankle (not shown). t, tibia; tmt, tarsometatarsus. (F) Toe and wing (flight) feathers of a muffed pigeon (English trumpeter), highlighting vane width asymmetries. Blue bar, inner vane; red bar, outer vane. Scale bar = 2 cm.

https://doi.org/10.7554/eLife.12115.018

Tbx5 is ectopically expressed in the hindlimb buds of feather-footed chickens

Other bird species, including domestic chickens, independently evolved foot feathers. Similar to pigeons, Tbx5 was ectopically expressed at HH25 in hindlimb buds of two feather-footed chicken breeds, the Cochin and the silkie (hindlimb expression relative to white leghorn forelimb: white leghorn 0.005 ± 0.004; Cochin 0.066 ± 0.050, p=0.002; silkie 0.034 ± 0.014, p=0.009; n = 6 white leghorns, 6 Cochins, 4 silkies) (Figure 3G). Ectopic Tbx5 expression in feathered-foot chicken embryos had a similar domain to that of feathered-foot pigeon embryos at HH25 (Figure 3I,J) and, as in pigeons, cis-acting changes contributed to this expression (expression of feathered-foot allele relative to scaled-foot allele in HH25 silkie x white leghorn F1 hybrid hindlimbs: 1.80 ± 0.41, p=2.55 x 10–5, n = 11 hybrid embryos) (Figure 4E). Hence, Tbx5-related developmental mechanisms may, in part, underlie the evolution of foot feathering in two species that last shared a common ancestor over 80 million years ago (Claramunt and Cracraft, 2015).

Classical genetic studies implicate at least two loci in heavy foot feathering in chickens (Punnett and Bailey, 1918; Lambert and Knox, 1929; Warren, 1948; Somes, 1992), although the molecular genetic origins of the trait remain unknown. Previously, a chromosome region containing Pitx1 was implicated in foot feathering in silkie chickens (Dorshorst et al., 2010). However, we did not detect statistically significant changes in Pitx1 expression between scaled-foot (white leghorn) and feathered-foot (silkie and Cochin) chicken embryos at HH25 (expression relative to white leghorn: Cochin 0.92 ± 0.24, p=0.93; silkie 0.71 ± 0.18, p=0.18; n = 6 each) (Figure 3F). This apparent conflict could be because the causative gene in silkies is not actually Pitx1 but rather a gene closely linked to it, or because Pitx1 expression differences are more pronounced and consistent at developmental stages that we did not assay. Furthermore, different populations of breeds such as silkies appear to have different constellations of ptilopody loci and alleles, and it is possible that we used strains that do not have Pitx1 mutations (Wexelsen, 1934; Somes, 1992). Also in contrast to our results in feather-footed pigeons, Tbx3 was not upregulated in ptilopodous chicken breeds (white leghorn 1 ± 0.17, silkie 0.48 ± 0.18, p=0.004; Cochin 0.86 ± 0.44, p=0.40; silkie vs. Cochin p=0.07; n = 6 samples each) (Figure 3—figure supplement 1D). In all, these results suggest that both shared and distinct mechanisms regulate foot feathering among avian species.

Discussion

Genetic architecture of ptilopody in pigeons

Extensive classical breeding experiments in pigeons demonstrate that complex derived traits can often be parsed into component parts (Sell, 1994, 2012). Thus, while traits are not always simple, they are often genetically tractable when using an informed breeding strategy (Domyan et al., 2014). Equivalent insights about the genetic architecture of phenotypic divergence between wild vertebrate species are often considerably more difficult to acquire. With pigeons, however, we have documentation for specific breed selection criteria and direct evidence for the resulting genetic architecture of derived traits (Levi, 1965; 1986; Sell, 1994; National Pigeon Association, 2010; Sell, 2012). This information offers a crucial advantage because it informs how we design genetic crosses and choose breeds for whole-genome resequencing to identify causal genes and mutations. Thus, we can combine classical breeding strategies and genomics to identify the molecular basis of both simple and oligogenic traits, as well as dissect different components of a complex phenotype, and define functional interactions among genes (Shapiro and Domyan, 2013; Domyan et al., 2014; Vickrey et al., 2015).

Classical studies in pigeon suggest two major-effect loci – grouse (gr) and Slipper (Sl) – are responsible for most of the variation in foot feathering (Doncaster, 1912; Wexelsen, 1934; Hollander, 1937; Levi, 1986). Through a combination of genetic, genomic, and developmental approaches, our data implicate regulatory mutations in the limb outgrowth and identity genes Pitx1 and Tbx5 as the molecular identities of the gr and Sl locus, respectively (Figure 6). Each locus has significant and separable effects on qualitative and quantitative variation in epidermal appendages: derived alleles of Pitx1 increase the extent of foot feathering, while a derived allele of Tbx5 is associated with the more elaborate muffed phenotype (Figure 1). Notably, these feathers are most numerous on the central forward-facing toe (digit 3; Table 2; Figure 3—figure supplement 3), just as forelimb feathers are most numerous on the central forelimb digit in birds and their dinosaurian relatives (Yalden et al., 1985; Gishlick et al., 2001; Hieronymus, 2015). Further, we also find that muscular and skeletal morphology are altered in muffed pigeons.

Model describing link between Pitx1 and Tbx5 expression levels and foot epidermal appendage morphology.

Darker colors indicate higher expression levels. Decreased expression of Pitx1 and ectopic expression of Tbx5 are associated with foot feathering (and other morphological transformations) in domestic pigeons.

https://doi.org/10.7554/eLife.12115.019

Collectively, these findings point to a partial alteration of the identity of the developing hindlimb, rather than localized changes to individual epidermal placodes. These alterations do not represent a complete transformation of limb type, as the hindlimbs of feather-footed pigeons are still readily recognizable as legs. This suggests that limb-type identity is not a simple binary choice between two global fates. For example, feather-footed pigeons have a radical transformation of the distal hindlimb dermis, yet changes to other hindlimb mesoderm derivatives (muscle, skeleton) are subtler and largely restricted to lateral structures in the adult. Therefore, we propose that different aspects of fore- and hindlimb morphology could have different dosage- and/or stage-dependent requirements for exposure to identity cues. Our ongoing analyses of musculoskeletal phenotypes in our F2 cross, which includes individuals with different combinations of feathered-foot alleles of Pitx1 and Tbx5, will allow us to understand the separate and epistatic effects of these loci on musculoskeletal anatomy. We note that, although genetic manipulations indicate that Tbx5 does not specify forelimb identity in mice, the divergence time between mammals and birds is deep (>300 million years) and subtly different roles for this transcription factor in limb outgrowth and identity might have evolved in these lineages (Minguillon et al., 2005; Horton et al., 2008).

Roles of Pitx1 and Tbx5 in diversity and disease

Another important avenue of future research will be to determine the downstream molecular consequences of Pitx1 and Tbx5 misregulation, and how this ultimately results in the transformation of scaled into feathered epidermis. Mutations at these genes can cause congenital limb deformities in humans, including clubfoot and Liebenberg syndrome (Pitx1; Gurnett et al., 2008; Spielmann et al., 2012), and Holt-Oram syndrome (Tbx5; Basson et al., 1997). Notably, haploinsufficiency causes these human syndromes, and clubfoot is partially penetrant in Pitx1+/- mice (Alvarado et al., 2011), collectively pointing to an exquisite sensitivity of limb morphology to levels of Pitx1 and Tbx5 gene products. Pitx1 is also involved repeatedly in the evolution of adaptive pelvic fin loss in stickleback fish (Cresko et al., 2004; Shapiro et al., 2004; 2006; Coyle et al., 2007; Chan et al., 2010; Shikano et al., 2013). Threespine sticklebacks that are homozygous for a Pitx1 pelvic enhancer deletion have severely reduced or absent pelvises, but heterozygotes also have smaller pelvises, thereby allowing natural selection to act on fish carrying one mutant allele (Shapiro et al., 2004; Chan et al., 2010). Similarly, we observed increased foot feathering in pigeons with just one derived allele of Pitx1 or Tbx5 (Figure 1). Ancient pigeon breeders could have rapidly selected for ptilopodous pigeon lines starting with birds that were heterozygous for mutations at either locus, and later generated the extreme muffed phenotype by hybridization. Together, studies of diversity and disease indicate that modest changes to the amount and location of Pitx1 and Tbx5 gene expression can cause dramatic alterations to limb development and morphology.

In addition to implicating cis- acting mutations in Pitx1 and Tbx5 driving transformation of scales into feathers in domestic pigeon, our results suggest that additional, as yet unidentified, mutations contribute to the muff phenotype. Although all feathered-foot embryos examined in our gene expression experiments contained derived Pitx1 and Tbx5 haplotypes, misregulation of each gene was more severe in the large-muffed English trumpeter than in the small-muffed Indian fantail. The English trumpeter may therefore contain additional cis-acting mutations at one or both loci, and/or mutations in upstream regulators of Pitx1 and Tbx5. Additional studies will be required to discriminate between these possibilities. Our findings suggest a quantitative link between transcription factor abundance and skin appendage fate and morphology, thereby highlighting foot-feathering in pigeons as a model for studying the regulatory interactions that govern expression of these two important determinants of limb morphology.

Evolution of epidermal appendage distribution

How might pigeons help us understand the evolution of epidermal appendage distribution and limb morphology in other species? Our findings suggest the mechanistic basis for the development of feathered feet in two distantly related domestic bird species is due to a partial transformation of limb identity, through cis-acting regulatory mutations in limb-type specific transcription factors. Most modern wild birds have a scaled metatarsus and toes, although some species (e.g., ptarmigan, snowy owl, and golden eagle) have extensive foot feathering. However, recent paleontological evidence suggests that feathers – not scales – might be the ancestral hindlimb skin appendages in birds and some of their close non-avian dinosaur relatives (Hu et al., 2009; Zheng et al., 2013). Thus, the epidermis of feather-footed modern birds might actually represent a reversion to the ancestral avialan skin condition. In some cases, the large, asymmetric-vaned, pennaceous metatarsal feathers of basal birds and their non-avian dinosaur relatives are so extensive that their hindlimbs have been interpreted as 'hind wings', although they clearly retain hindlimb skeletal identity (Zheng et al., 2013). Furthermore, these long foot feathers are directed laterally from the foot, and they display vane width asymmetries reminiscent of flight feathers; we find a similar morphology in muffed pigeons (Figure 5F, Figure 3—figure supplement 3). Perhaps not coincidentally, Darwin, 1868 noted of the muffed English trumpeter pigeon, “Their feet are so heavily feathered, that they almost appear like little wings” (p. 155).

Building on classical breeding experiments in both pigeons and chickens, we find that a relatively small number of genetic changes account for a large proportion of the variation in epidermal appendage morphology and distribution. Thus, major determinants of dramatic phenotypic variation can be mechanistically simple and therefore potentially evolve rapidly. In pigeons, these mechanisms can generate wing-like feathers on a hindlimb that is not used for powered flight or gliding. This, in turn, suggests that wing-like foot and leg feathers in other species, such as non-avian dinosaurs, might result from developmental constraints on the morphology of large limb feathers, rather than from functional adaptations for flight (Gould and Lewontin, 1979; Foth et al., 2014).

Materials and methods

Animal husbandry and phenotyping of F2 offspring

Animals were housed in accordance with the University of Utah Institutional Animal Care and Use Committees of University of Utah (protocols 10–05007 and 13–04012). 130 F2 offspring were generated by mating a male Pomeranian pouter to two female Scandaroons, and DNA samples extracted (DNeasy Blood and Tissue Kit, Qiagen, Valencia, CA). 114 F2 offspring survived to 6 months of age, at which time they were euthanized and phenotypic measurements taken. Proportion of the tarsus was measured by dividing the length of the dorsal tarsus that was feathered by the total length of the tarsus (measured from the tibia-tarsometatarsus joint to the distal aspect of tarsometatarsal-phalangeal joint of digit 3), and averaged between the two tarsi. Toe feathers were counted on each toe, and summed across all 8 toes. The length of each of the longest three toe feathers on digit 3 (the central forward-directed toe), which bore the longest toe feathers on each foot, was measured to the nearest 1 mm and averaged for each bird.

Whole-genome genotyping by sequencing (GBS)

For genotyping, we used a previously published approach (Elshire et al., 2011) with minor modifications. Briefly, for each founder parent and 130 F2 offspring, 50 ng of DNA was digested with ApeKI, ligated to barcoded adapters, and then 10 ng of each barcoded sample was pooled in batches of 26 individuals and purified (Qiagen PCR Purification Kit). DNA fragments 550–650 bp in size were selected using Pippin Prep (Sage Science, Beverly, MA), and amplified by 10–12 cycles of PCR using custom indexed primers. Libraries were purified with Ampure beads (Sigma-Aldrich, St. Louis, MO) and sequenced using 100- or 125-bp, paired-end sequencing on the Illumina HiSeq2000 platform at the University of Utah Genomics Core Facility. Reads were trimmed to 90 bp, filtered for quality, and de-multiplexed using Stacks (Catchen et al., 2011). Reads were mapped to the pigeon reference genome (Shapiro and Domyan, 2013) using Bowtie2 (Langmead and Salzberg, 2012), filtering for MAPQ < 20. The average number of mapped reads among F2 individuals was 3,397,598, with a mean depth of 6.3x. Genotypes were called using Stacks (Catchen et al., 2011), with a minimum read-depth cutoff of 5. Markers that were genotyped in ≥ 70 of the 130 F2 individuals were retained.

Genetic map construction and QTL mapping

Genetic map construction and QTL mapping was performed using R/qtl (www.rqtl.org) (Broman et al., 2003). Markers showing segregation distortion (Chi-square, p<0.05) were removed. 3803 markers were assembled into linkage groups using the parameters (max.rf = 0.15, min.lod = 6). Linkage groups were numbered in descending order, based on the number of markers. Linkage group 11 and 20 QTL were initially mapped using the scanone function using Haley-Knott regression. Probable false-homozygote genotyping errors resulting from the low read-depth cutoff used (5x), identified as closely-spaced double-crossover events, were manually corrected on these linkage groups. Subsequently, the stepwiseqtl function was used to identify additional QTL, and the fitqtl function used to account for the effect of one linkage group while calculating the LOD scores and percent variance explained (PVE) of the other. Significance thresholds of α = 0.05 were calculated with 1000 permutations of each phenotype across all linkage groups. The peak markers for each phenotype were used to test for the effect of each QTL.

Genomic analyses

BAM files generated previously for a whole-genome resequencing panel (Shapiro and Domyan, 2013) were combined with BAM files for two new Pomeranian pouter whole-genome sequences to call genomic variants (SNVs and small indels) using the Genome Analysis Toolkit (Unified Genotyper and LeftAlignAndTrimVariants functions, default settings) (McKenna et al., 2010). We removed variant sites that were called in two or fewer genomes (i.e., all other genomes were no-calls) or that had variant alleles on only two or fewer chromosomes, as these categories of low-frequency variants would be uninformative to our analyses. The resulting variant call format (VCF) file was used for subsequent analyses.

Individual birds from different breeds were binned into the following phenotypic classifications:

Groused: Berlin long-faced tumbler, Lahore, Oriental frill, Shaksharli.

Muffed: English long-faced muffed tumbler, English pouter, English trumpeter, frillback, ice pigeon, Indian fantail (2 individuals), Pomeranian pouter (2 individuals), Saxon monk, Saxon pouter. The English pouter is an unusual breed that is sometimes classified by breeders as slipper only. Its foot feathering is far more extensive than groused breeds, which led us to include it in the muffed group for the purposes of the genomic analyses.

Scale-footed: African owl, archangel, Birmingham roller, carneau, Chinese owl, cumulet, Egyptian swift, English carrier, fantail, feral (2 individuals), Iranian tumbler, Jacobin, king, Lebanon, Marchenero pouter, mookee, Oriental roller, parlor roller, runt, Scandaroon, Spanish barb, starling, Syrian dewlap, Thai laugher.

pFst, a modified likelihood ratio test that accounts for genotype uncertainty, extended haplotype homozygosity (EHH), and haplotype network analyses were implemented using the GPAT++ software library (Kronenberg et al., 2014; see https://github.com/vcflib/vcflib for software updates).

Genotyping assays

Primers for genotyping the scaffold 79 deletion are listed in Supplementary file 1. Breeds used for association testing were as follows:

Feather-footed (21 breeds total): Berlin long-faced tumbler, Berlin short-faced tumbler, Bokhara trumpeter, classic Oriental frill, crested Saxon field color, English trumpeter, fairy swallow, frillback, German double-crested trumpeter, ice pigeon, Indian fantail, Lahore, Mindian fantail, Oriental frill, Persian roller, Pomeranian pouter, Russian tumbler, saint, Schmalkaldner moorhead, Uzbeck tumbler, West of England.

Scale-footed (56 breeds total): African owl, Altenburg trumpeter, American flying tumbler, American giant homer, American mini crest, American show racer, archangel, Bohemian pouter, Brunner cropper, Budapest tumbler, Cauchois, Chinese owl, cumulet, Danzig highflier, domestic show flight, dragoon, English baldhead long-faced clean-legged tumbler, English carrier, English magpie, English short-faced tumbler, exhibition homer, fantail, Franconian trumpeter, French mondaine, giant runt, Holle cropper, horseman pouter, Italian owl, Jacobin, Jiennesse pouter, king, Lebanon, Spanish little friar tumbler, Maltese, medium-faced crested helmet, Modena, mookee, Norwich cropper, nun, Old Dutch Capuchine, Old German owl, Oriental roller, parlor roller, Portuguese tumbler, Scandaroon, showtype racing homer, Spanish barb, starling, Syrian Baghdad, Texas pioneer, Thai laugher, Thuringer clean leg, Vienna medium-faced tumbler, Voorburg shield cropper, zitterhals.

High-throughput chromatin immunoprecipitation and sequencing (ChIP-seq)

Forelimb and hindlimb buds from HH25 racing homer and Indian fantail embryos were collected and placed in 1% formaldehyde for 20 min at room temperature, then washed 3x in ice-cold PBS and stored at -80°C until chromatin extraction. ChIP was performed on 200 micrograms of chromatin isolated from embryonic pigeon limbs. Control libraries were prepared using 100 ng of input chromatin. A total of 16 libraries were created (8 ChIP and 8 input controls for each breed and limb combination). A validated monoclonal antibody against H3K27ac (Millipore #05–1334, Billerica, MA) was used to perform ChIP, and sequencing libraries were prepared using NEBNext Ultra DNA Library Prep Kit for Illumina with NEBNext Multiplex Oligos for Illumina (Index Primers Set 1; New England BioLabs, Ipswich, MA). All libraries were size selected using SPRI magnetic beads to eliminate adapter dimers. All 8 ChIP libraries showed enrichment for a positive control site relative to input libraries (tested by qPCR). Single-end, 50-bp read sequencing was performed on Illumina HiSeq2000 platform at the University of Utah Genomics Core Facility.

Fold-enrichment plots were generated using MACS (Zhang et al., 2008; Feng et al., 2012) and visualized in IGV (Robinson et al., 2011; Thorvaldsdóttir et al., 2013). Regions of differential enrichment between racing homer and Indian fantail hindlimbs were identified using function bdgdiff in MACS2 (https://pypi.python.org/pypi/MACS2/2.0.10.20130522). Regions with a log10 likelihood ratio ≥ 3 were considered to have differential enrichment between the two groups.

RNA isolation and cDNA synthesis

To assay gene expression, limb buds from HH25 embryos were harvested and placed in RNAlater (Qiagen, Valencia, CA) at 4°C overnight. Total RNA was extracted, cleaned and DNase-treated (Qiagen RNeasy Kit). mRNA was reverse-transcribed to cDNA using oligo-dT and M-MLV RT (Invitrogen, Carlsbad, CA) according to the manufacturer’s protocol.

qRT-PCR analyses

cDNA was amplified using intron-spanning primers for each target using a CFX96 qPCR instrument and iTaq Universal Sybr Green Supermix (Bio-Rad, Hercules, CA). Results were compared by Mann-Whitney U test. Two technical replicates of each sample were performed, and the mean value determined. Differences were considered statistically significant if p<(0.05 / # genes assayed) to control for multiple-testing. Each experiment was performed three times, and the results presented are from one representative experiment. Primers used for each assay are listed in Supplementary file 1.

Allele-specific expression assay

SNPs in Pitx1 and Tbx5 transcripts were identified by Sanger sequencing in the parents of a cross between an Old Dutch Capuchine (scale-footed) and a fairy swallow (muffed) that was homozygous for the 44-kb deletion upstream of Pitx1, and PyroMark Custom Assays (Qiagen) for each SNP were designed using the manufacturer’s software. Pyrosequencing was performed on cDNA and gDNA derived from HH25 limb buds using a PyroMark Q24 instrument (Qiagen). The signal intensity ratio of feathered allele to scaled allele from cDNA samples was normalized to ratios obtained from gDNA samples from the same embryos to control for allele-specific amplification bias. Normalized ratios were analyzed by Mann-Whitney U test, and considered significant if p<(0.05 / # genes assayed) to control for multiple-testing. Each experiment was performed twice, and the results presented are representative. Primers used for each assay are listed in Supplementary file 1.

Whole-mount in situ hybridization

Linear templates for probe synthesis were amplified from cDNA by PCR using primers listed in Supplementary file 1. Binding sites for T3 and T7 polymerase were incorporated into the forward and reverse primers to facilitate subsequent transcription of sense and antisense probe, respectively.

Embryos used for RNA in situ hybridization were dissected from eggs, and fixed overnight in 4% paraformaldehyde at 4°C on a shaking table, then dehydrated into 100% MeOH and stored at -20°C. RNA in situ hybridization was performed as described (Abler et al., 2011). Hybridization with sense probe was performed as negative control.

References

  1. 1
    A high throughput in situ hybridization method to characterize mRNA expression patterns in the fetal mouse lower urogenital tract
    1. LL Abler
    2. V Mehta
    3. KP Keil
    4. PS Joshi
    5. CL Flucus
    6. HA Hardin
    7. CT Schmitz
    8. CM Vezina
    (2011)
    Journal of Visualized Experiments, 10.3791/2912.
  2. 2
  3. 3
  4. 4
    R/qtl: QTL mapping in experimental crosses
    1. KW Broman
    2. H Wu
    3. S Sen
    4. GA Churchill
    (2003)
    Bioinformatics 19:889–890.
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
    A new role for Notch and Delta in cell fate decisions: patterning the feather array
    1. R Crowe
    2. D Henrique
    3. D Ish-Horowicz
    4. L Niswander
    (1998)
    Development 125:767–775.
  14. 14
    An Hereditary Complex in the Domestic Fowl
    1. CH Danforth
    (1919)
    Genetics 4:587–596.
  15. 15
    The Variation of Animals and Plants Under Domestication
    1. CR Darwin
    (1868)
    London: John Murray.
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
    Involvement of t-box genes Tbx2-Tbx5 in vertebrate limb specification and development
    1. JJ Gibson-Brown
    2. SI Agulnik
    3. LM Silver
    4. L Niswander
    5. VE Papaioannou
    (1998a)
    Development 125:2499–2509.
  26. 26
  27. 27
    New Perspectives on the Origin and Early Evolution of Birds: Proceedings of the International Symposium in Honor of John H. Ostrom
    1. AD Gishlick
    (2001)
    The function of the manus and forelimb of Deinonychus antirrhopus and its importance for the origin of avian flight , New Perspectives on the Origin and Early Evolution of Birds: Proceedings of the International Symposium in Honor of John H. Ostrom, New Haven, Peabody Museum of Natural History, Yale University.
  28. 28
  29. 29
    The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme
    1. SJ Gould
    2. RC Lewontin
    (1979)
    Proceedings of the Royal Society of London. Series B, Biological Sciences of the United States of America 205:581–598.
  30. 30
  31. 31
    A series of normal stages in the development of the chick embryo
    1. V Hamburger
    2. HL Hamilton
    (1951)
    Journal of Morphology 88:49–92.
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
    Department of Genetics
    1. WF Hollander
    (1937)
    p 67, Hereditary Interrelationships of Certain Factors in Pigeons, Department of Genetics, University of Wisconsin.
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
    The inheritance of of leg-feathering in the chicken
    1. WV Lambert
    2. CW Knox
    (1929)
    Iowa Acad Sci Proc 36:383–384.
  44. 44
  45. 45
    Encyclopedia of Pigeon Breeds
    1. WM Levi
    (1965)
    Sumter: Levi Publishing Co., Inc.
  46. 46
    The Pigeon
    1. WM Levi
    (1986)
    Sumter: Levi Publishing Co., Inc.
  47. 47
    Differential regulation of T-box and homeobox transcription factors suggests roles in controlling chick limb-type identity
    1. M Logan
    2. HG Simon
    3. C Tabin
    (1998)
    Development 125:2825–2835.
  48. 48
    Role of Pitx1 upstream of Tbx4 in specification of hindlimb identity
    1. M Logan
    2. CJ Tabin
    (1999)
    Science 283:1736–1739.
  49. 49
    Avian Anatomy - Integument
    1. AM Lucas
    2. PR Stettenheim
    (1972)
    Washington, DC.
  50. 50
  51. 51
  52. 52
  53. 53
    National Pigeon Association Book of Standards
    1. National Pigeon Association
    (2010)
    Goodlettsville: Purebred Pigeon Publishing.
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
    The evolution of paired appendages in vertebrates: T-box genes in the zebrafish
    1. I Ruvinsky
    2. AC Oates
    3. LM Silver
    4. RK Ho
    (2000)
    Development Genes and Evolution 210:82–91.
  59. 59
    Breeding and Inheritance in Pigeons
    1. A Sell
    (1994)
    Germany: Schober Verlags-GmbH, Hengersberg.
  60. 60
    Pigeon Genetics: Applied Genetics in the Domestic Pigeon
    1. A Sell
    (2012)
    Germany: Verlag Karin und A. Sell, Achim.
  61. 61
    Parallel genetic origins of pelvic reduction in vertebrates
    1. MD Shapiro
    2. MA Bell
    3. DM Kingsley
    (2006)
    Proceedings of the National Academy of Sciences of the United States of America 103:13753–13758.
    https://doi.org/10.1073/pnas.0604706103
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
    Identifying the ptilopody (feathered shank) loci of the chicken
    1. RG Somes
    (1992)
    The Journal of Heredity 83:230–234.
  67. 67
  68. 68
    Role of the Bicoid-related homeodomain factor Pitx1 in specifying hindlimb morphogenesis and pituitary development
    1. DP Szeto
    2. C Rodriguez-Esteban
    3. AK Ryan
    4. SM O'Connell
    5. F Liu
    6. C Kioussi
    7. AS Gleiberman
    8. JC Izpisúa-Belmonte
    9. MG Rosenfeld
    (1999)
    Genes & Development 13:484–494.
  69. 69
  70. 70
    Differential expression of Tbx4 and Tbx5 in Zebrafish fin buds
    1. K Tamura
    2. S Yonei-Tamura
    3. JC Izpisúa Belmonte
    (1999)
    Mechanisms of Development 87:181–184.
  71. 71
  72. 72
  73. 73
    Convergent Evolution of Head Crests in Two Domesticated Columbids Is Associated with Different Missense Mutations in EphB2
    1. AI Vickrey
    2. ET Domyan
    3. MP Horvath
    4. MD Shapiro
    (2015)
    Molecular Biology and Evolution, 32, 10.1093/molbev/msv140.
  74. 74
    Linkage relations of autosomal factors in the fowl
    1. DC Warren
    (1948)
    Genetics 34:333–350.
  75. 75
    Types of leg feathering in pigeons
    1. H Wexelsen
    (1934)
    Hereditas 18:192–198.
  76. 76
    ggplot2: Elegant Graphics for Data Analysis
    1. H Wickham
    (2009)
    New York: Springer.
  77. 77
  78. 78
    Forelimb function in Archaeopteryx
    1. DW Yalden
    (1985)
    In: M. K Hecht, J. H Ostrom, G Viohl, P Wellnhofer, editors. The Beginnings of Birds: Proceedings of the International Archaeopteryx Conference, 1984. Eichstätt, Germany: Freunde des Jura-Museums Eichstätt. pp. 91–97.
  79. 79
  80. 80
  81. 81

Decision letter

  1. Marianne E Bronner
    Reviewing Editor; California Institute of Technology, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your work entitled "A molecular shift in limb identity underlies the convergent evolution and development of feathered feet" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor, Marianne Bronner, and Detlef Weigel as the Senior Editor.

The reviewers have discussed the reviews with one another and the Reviewing editor has drafted this decision to help you prepare a revised submission.

This work represents an elegant use of pigeon genetics to address an important question of evolutionary morphology: the origin of feathered feet. A molecular understanding of this issue would open the door to advances in our knowledge of the scale to feather transition, the origin of feathered appendages, and other problems of significance. Using the tractable pigeon system, with a bevy of tools available for genetic and anatomical analysis, the authors reveal a correlation between feathered feet and a shift in expression of two regulatory factors, Pitx1 and Tbx5. The novelty is slightly dampened by the fact that the genes they found have been known to alter limb identify for more than a decade, but the work still represents a logical approach and valuable documentation, showing us how different approaches can lead scientists to the same important genes. While the genetic approach in pigeon is strong, the chicken variant part needs improvement as detailed below. Importantly, some of the discussion regarding the evolutionary implications seems over-interpreted and needs to be toned down.

1) A major problem that must be clarified and rephrased is the relationship to dinosaur feather/scale evolution. While this is certainly an interesting topic, the "four wing dinosaurs" found by Xing Xu and his associates clearly have leg skeleton in the hind feet. Thus, a "feathered limb" can have wing or leg skeleton inside. Consistent with this, Figure 4 nicely shows that the muffed pigeon foot exhibits some wing-like skeleton and muscle patterns. In dinosaur evolution, the change of feather and scale occurs in the dermis, and the skeleton remains foot-like. In contrast, the mutation studied here is a limb identity decision, which is a different matter and this must be clarified. It is an inappropriate analogy particularly to be in the first and last sentence of the manuscript.

2) "Convergent Evolution" used in the title is too strong and misleading. It should be rephrased in the title, Abstract and Introduction.

3) It is claimed that misexpression of a Pitx1-Enr construct disrupts scale formation and that this is consistent with phenotypes obtained following Pitx1 misexpression in the wing. What is shown in the two panels is a severely truncated and abnormal limb. It is not possible to assess any effect on scale formation from this part of the data. It seems to go too far to say that Pitx1-Enr is dominant-negative, when the luciferase assay merely indicates that the construct can repress transcription in a heterologous system. It is incorrect to state that “[…]verify that Pitx1EnR inhibited wild-type Pitx1 activity".

4) While the figure title states that "Tbx5 alters development of the foot epidermal appendages" the main text goes much further suggesting "misexpression of Tbx5[…] induced ectopic development of foot feathers". This is not at all clear from the panels in Figure 2—figure supplement 4B. Normal developing feathers are clear in other parts of the embryos at these stages and the structures highlighted do not look similar. The authors should be much more cautious about exactly what has occurred in these embryos.

5) An important issue is the localization of the cells that are expressing Tbx5 in the hindlimb, since it is epidermal derivatives that are being analysed largely and these structures are known to be patterned by signals from the mesoderm. It is stated that "ectopic Tbx5 expression was largely localized to the mesoderm of the proximal and posterior hindlimb" based on whole mount in situ hybridization. It would be clearer to show this in a section. It is further stated that this domain "shows a striking correlation with regions of epidermal transformation". This is not clear at all. Also, not too much can/should be read into the expression domain of Pitx1 shown for two breeds of pigeon. Are you certain that these are the correct stages to be analyzing the expression domains of these genes with respect to any possible action they may have on patterning epidermal structures? The same argument applies for the analysis of ectopic Tbx5.

7) The data in Figure 4 comparing two muscles and tibia/fibula in feral and Pomeranian pouter demonstrate a re-patterning of hind limb structures to more wing-like and that this can be attributed to the ectopic expression of Tbx5 described in different breeds (Indian fantail, English trumpeter). In regard to the fibula differences, it would certainly appear to be longer in the muffed pigeon but it remains a fibula in position and morphology. To suggest it is like an ulna is a misrepresentation. It looks nothing like an ulna in shape or articulation. The length of fibula can be different in different species and this need not be considered as a hindlimb to forelimb transformation.

Additional issues with this section are:

The section starts with a misrepresentation of the literature suggesting Tbx5 can result in muscular and skeletal mispatterning.

Any potential contribution of reduced Pitx1 expression to the morphological differences described is not considered.

Is "ulna" in subheading “Muffed pigeon breeds incur musculoskeletal patterning changes” a typo?

Is "lateral" hindlimb really intended. Is this really the primary site of ectopic Tbx5. This is described as 'posterior' in the earlier section.

8) In the Discussion, while it is mentioned that additional mutations also contribute to the muff phenotype it is assumed that these must all converge on the Tbx5 and Pitx1 pathways. This train of thought is continued in the suggestion that "a relatively small number of genetic changes are sufficient for a surprisingly large transformation of epidermal appendage morphology and distribution.” I don't believe the data make the case for 'sufficiency' in their argument since the misexpression data are not convincing and the birds could be harbouring a large number of additional mutations that contribute/are responsible for the feathering phenotype.

9) There is surprisingly little discussion of work on Tbx5 and Pitx1 function in the mouse. The authors must be aware that much of what they show and conclude is inconsistent with analyses of various mouse mutants of Tbx5 and Pitx1. This has been ignored largely, even though it would be exciting, if indeed true. The current interpretations of the results, without reference to the prior work in other systems, could further muddy an already messy literature on the roles of genes in determining the differences between fore limbs and hind limbs. It is therefore essential to clearly understand the functional significance of the QTL and expression analyses they have carried out. It would be interesting to test if elevating the levels of Pitx1 in a feathered breed would be sufficient to reduce feathers/replace with scales and reduce Tbx5 expression in the leg.

10) Have the authors considered whether the specific QTLs identified relate to expression of Tbx5 in the wing or other regions of the embryo or adult? What attempts have been made to demonstrate that the QTL are causative to the expression domain in the leg? Only one possible explanation (out of the many other alternatives) appear to have been explored and the final conclusion hangs on what appear to me to be unconvincing functional data.

11) Ptilopody refers to the presence of feathers in the shank. There are feathers on proximal parts of pigeon legs (both with scaled shank and feathered) that do not correlate with Tbx5 expression or reduction in Pitx1 expression. This is not mentioned or discussed. Is the assumption that the capacity of Tbx5 is localized to distal regions of the limb? There are of course feathers in many other regions of the bird where Tbx5 is not expressed also. Chicken breeds also show variation in feathering in the hindlimb-cochin and silkie chicken are analyzed to some extent. In silkie and some other chicken breeds (silky-feather) the basis of the mutation is known and/or causes other defects such as ploydactyly in Silkies. These do not correlate with the results and conclusions derived from this work in pigeon. These have not been discussed and are relevant to the significance of the proposed mechanisms in bird/dinosaur evolution.

12) The shared genotype for feathered feet between pigeon and chicken may be partial since some chicken with feathered feet have Tbx5 in the leg bud, but no difference in Pitx1, which is on chromosome 13 where the chicken ptilopody locus is located. Can the authors analyze different chicken ptilopody variants to qualify their convergent evolution statement?

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Molecular shifts in limb identity underlie development of feathered feet in two domestic avian species" for further consideration at eLife. Your revised article has been favorably evaluated by Detlef Weigel (Senior editor), a Reviewing editor (Marianne Broner), and two reviewers. The manuscript has been improved but there are some remaining, relatively minor issues that need to be addressed before acceptance, as outlined below. We look forward to receiving a revised version that addresses these points:

Point 1: The point was to distinguish that there can be changes in feather and scale dermis patterns that are distinct from the other mesoderm derivatives (e.g. skeleton) rather than anything about dosage and/or stage requirements.

Point 7: Although the text has been changed, the authors still seem to be implying that the longer fibula is a partial hindlimb to forelimb transformation which it need not be (Minor issue:the ulna does not articulate at the wrist, the radius does.)

Please avoid the term “limb-specific”.

Subheading “Evolution of epidermal appendage distribution”: avialan to avian

https://doi.org/10.7554/eLife.12115.027

Author response

This work represents an elegant use of pigeon genetics to address an important question of evolutionary morphology: the origin of feathered feet. A molecular understanding of this issue would open the door to advances in our knowledge of the scale to feather transition, the origin of feathered appendages, and other problems of significance. Using the tractable pigeon system, with a bevy of tools available for genetic and anatomical analysis, the authors reveal a correlation between feathered feet and a shift in expression of two regulatory factors, Pitx1 and Tbx5. The novelty is slightly dampened by the fact that the genes they found have been known to alter limb identify for more than a decade, but the work still represents a logical approach and valuable documentation, showing us how different approaches can lead scientists to the same important genes. While the genetic approach in pigeon is strong, the chicken variant part needs improvement as detailed below. Importantly, some of the discussion regarding the evolutionary implications seems over-interpreted and needs to be toned down.

We appreciate the positive feedback about our approach and results. We address our results in chicken directly in our responses to specific comments below. We also understand the concern about over-interpretation of the evolutionary implications of our results, and we have toned down our language, especially in the Abstract, Introduction, and Results. We still feel that our work provides exciting new insights about evolutionary changes in epidermal appendage identity and distribution, but we have made this less of a central theme of the paper.

1) A major problem that must be clarified and rephrased is the relationship to dinosaur feather / scale evolution. While this is certainly an interesting topic, the "four wing dinosaurs" found by Xing Xu and his associates clearly have leg skeleton in the hind feet. Thus, a "feathered limb" can have wing or leg skeleton inside. Consistent with this, Figure 4 nicely shows that the muffed pigeon foot exhibits some wing-like skeleton and muscle patterns. In dinosaur evolution, the change of feather and scale occurs in the dermis, and the skeleton remains foot-like. In contrast, the mutation studied here is a limb identity decision, which is a different matter and this must be clarified. It is an inappropriate analogy particularly to be in the first and last sentence of the manuscript.

We agree with this assessment, although we wish to point out a couple of items of potential confusion. First, the leg skeletons of muffed pigeons are clearly still legs, and we did not state otherwise in the original manuscript. To clarify this point, we now state in the Discussion:

“Collectively, these findings point to an alteration of the identity of the developing hind limb, rather than localized changes to individual epidermal placodes. […] Instead, different aspects of hind and forelimb morphology appear to have different dosage- and/or stage-dependent requirements.”

On this same theme, we do not claim that muscle mis-patterning in the hind limb directly resembles the normal forelimb. To emphasize this point, we added the following statement in the Discussion:

“These changes are aberrations of normal patterning, although they are not necessarily clear transformations to a more forelimb-like configuration.”

2) "Convergent Evolution" used in the title is too strong and misleading. It should be rephrased in the title, Abstract and Introduction.

This comment raises an interesting issue. Descent with modification under domestication is widely accepted as evolution, albeit not in a “natural” setting, but we realize that the language we used could lead to confusion. Our interpretation of convergence also seems to have no alternative – inheritance of feathered feet from the most recent common ancestor of pigeons and chickens is rejected by every recent phylogenetic hypothesis of avian evolution. We have altered our verbiage to the more descriptive “genetic mechanisms” and similar terms to avoid confusion and detract from the main messages of our study.

3) It is claimed that misexpression of a Pitx1-Enr construct disrupts scale formation and that this is consistent with phenotypes obtained following Pitx1 misexpression in the wing. What is shown in the two panels (Figure 2—figure supplement 4B) is a severely truncated and abnormal limb. It is not possible to assess any effect on scale formation from this part of the data. It seems to go too far to say that Pitx1-Enr is dominant-negative, when the luciferase assay merely indicates that the construct can repress transcription in a heterologous system. It is incorrect to state that“[…]verify that Pitx1EnR inhibited wild-type Pitx1 activity".

After further reflection, we agree that the direct implications of these functional experiments are difficult to assess. As a result, we chose to remove them from the revised manuscript.

4) While the figure title states that "Tbx5 alters development of the foot epidermal appendages" the main text goes much further suggesting "misexpression of Tbx5[…] induced ectopic development of foot feathers". This is not at all clear from the panels in Figure 2figure supplement 4B. Normal developing feathers are clear in other parts of the embryos at these stages and the structures highlighted do not look similar. The authors should be much more cautious about exactly what has occurred in these embryos.

As mentioned above, we have removed these experiments from the manuscript.

5) An important issue is the localization of the cells that are expressing Tbx5 in the hindlimb, since it is epidermal derivatives that are being analysed largely and these structures are known to be patterned by signals from the mesoderm. It is stated that "ectopic Tbx5 expression was largely localized to the mesoderm of the proximal and posterior hindlimb" based on whole mount in situ hybridization. It would be clearer to show this in a section.

This is an excellent suggestion. We now include images of sections from each of the three pigeon phenotypes studied (scaled, small muff, large muff) in Figure 3—figure supplement 3. These sections helped us detect more dorsal expression than we were able to appreciate from whole embryos, and now describe “dorsal-posterior” expression of Tbx5 in the manuscript.

It is further stated that this domain "shows a striking correlation with regions of epidermal transformation". This is not clear at all.

Figure 3—figure supplement 3 now shows images from adult and embryonic muffed limbs. These images emphasize that the most pronounced epidermal transformations occur in the posterior (lateral digits in the adult) and dorsal aspects of the limb. The enlargement of the fibula (posterior/lateral zeugopod bone, Figure 5) also supports this interpretation.

Also, not too much can/should be read into the expression domain of Pitx1 shown for two breeds of pigeon. Are you certain that these are the correct stages to be analyzing the expression domains of these genes with respect to any possible action they may have on patterning epidermal structures? The same argument applies for the analysis of ectopic Tbx5.

We agree that the stages we chose to analyze might not represent differences (or similarities) that persist throughout development. To make this point clear, we have specified Hamburger-Hamilton stages in the revised manuscript. Further developmental biology work is needed to fully understand the specific time points at which the actions of these two genes are most important for specific aspects of epidermal, muscular, and skeletal development. However, our genetic and genomic results provide clear motivation for studying these two genes, and our allele-specific expression assays demonstrate that cis-regulatory differences specifically at these two loci affect their expression levels.

7) The data in Figure 4 comparing two muscles and tibia/fibula in feral and Pomeranian pouter demonstrate a re-patterning of hind limb structures to more wing-like and that this can be attributed to the ectopic expression of Tbx5 described in different breeds (Indian fantail, English trumpeter). In regard to the fibula differences, it would certainly appear to be longer in the muffed pigeon but it remains a fibula in position and morphology. To suggest it is like an ulna is a misrepresentation. It looks nothing like an ulna in shape or articulation. The length of fibula can be different in different species and this need not be considered as a hindlimb to forelimb transformation.

Thank you for highlighting these points of potential confusion. We agree that the fibula of muffed pigeons is still a fibula in both morphology and position, and we suspect that a typo (pointed out below) contributed to the impression that we were mistakenly identifying the fibula as an ulna. On the contrary, our intent was to convey that the fibula of the English trumpeter in Figure 4 (now Figure 5) is longer than usual and articulates with both the knee and the ankle, just as the ulna articulates with the elbow and wrist. We agree that fibular morphology varies widely among birds, but the configuration we observe in muffed birds is highly aberrant for a pigeon, and this is the key finding we tried to communicate. To emphasize this point, we modified the following sentence to refer specifically to pigeons instead of “birds” in general:

“We also found that the fibula, which is normally reduced in pigeons relative to the homologous forelimb structure (ulna), was enlarged[…]”

Further confusion might have originated from the typo in the following statement:

“Ectopic expression of Tbx5 in the hind limbs of chick embryos produces a similar enlargement of the ulna.”

In the above sentence, “ulna” should have been “fibula.” We have made this change in the revised manuscript.

Additional issues with this section are:

The section starts with a misrepresentation of the literature suggesting Tbx5 can result in muscular and skeletal mispatterning.

Thank you for this comment. Our effort to be economical with space unfortunately resulted in the following confusing statement:

“In mouse and chicken embryos, experimental manipulation of Pitx1 and Tbx5 expression can also result in muscular and skeletal mispatterning (Logan and Tabin 1999; Hasson, et al. 2010).”

We have now expanded the Discussion at the beginning of this section. We start with a similar overview statement, and provide additional commentary to de-convolute the known roles of these two genes on muscular and skeletal patterning in different vertebrates.

“In mouse and chicken embryos, experimental manipulation of Pitx1 and Tbx5 expression can result in muscular and skeletal abnormalities. […] Given the dramatic musculoskeletal defects observed in other organisms with experimentally altered Pitx1 and Tbx5 expression, we compared the hind limb morphology of adult feral pigeons[…]”

Any potential contribution of reduced Pitx1 expression to the morphological differences described is not considered.

As mentioned above, and as detailed in the Discussion (“Roles of Pitx1 and Tbx5 in diversity and disease” subheading), we now discuss the impact of Pitx1 misexpression and mutant alleles in a bird, two mammals, and a fish. These examples illustrate that decreases in Pitx1 expression can result in skeletal changes, although we are not presently able to determine the specific effects of Pitx1 versus Tbx5 on musculoskeletal morphology in muffed pigeons. Follow-up work will entail detailed phenotyping of the musculoskeletal system of our F2 cross by dissection and CT analysis, and these phenotypes will allow us to separate the effects of these two loci on musculoskeletal changes. To emphasize this point, we now state in the aforementioned subsection of the Discussion:

“Our ongoing analyses of musculoskeletal phenotypes in our F2 cross, which includes individuals with different combinations of feathered-foot alleles of Pitx1 and Tbx5, will allow us to understand the separate and epistatic effects of these loci on musculoskeletal anatomy.”

Is "ulna" in subheading “Muffed pigeon breeds incur musculoskeletal patterning changes”a typo?

Yes, thank you for identifying this error. “Ulna” has been changed to “fibula” in the revised manuscript.

Is "lateral" hindlimb really intended. Is this really the primary site of ectopic Tbx5. This is described as 'posterior' in the earlier section.

Thank you for pointing out this discrepancy. The posterior limb bud of the embryo becomes the lateral side of the adult limb. To clarify this point, we have altered the original statement to read:

“Notably, all of the modified structures of ptilopodous pigeons develop in the posterior (lateral in the adult) and dorsal hind limb, which are the primary sites of ectopic Tbx5 expression.”

8) In the Discussion, while it is mentioned that additional mutations also contribute to the muff phenotype it is assumed that these must all converge on the Tbx5 and Pitx1 pathways. This train of thought is continued in the suggestion that "a relatively small number of genetic changes are sufficient for a surprisingly large transformation of epidermal appendage morphology and distribution.” I don't believe the data make the case for 'sufficiency' in their argument since the misexpression data are not convincing and the birds could be harbouring a large number of additional mutations that contribute/are responsible for the feathering phenotype.

We agree that other genes are probably involved in the foot feathering phenotype. However, multiple lines of evidence demonstrate that only two loci have major effects on the phenotype. First, classical genetic studies implicate major effects of the grouse and slipper loci, the molecular identities of which were previously unknown. Second, our QTL mapping study implicated just two loci of major effect on the phenotype. Importantly, our F2 intercross led to the recombining of muffed and scaled genomes, thereby minimizing the effects of the parental genetic backgrounds on phenotypic variation in the F2 offspring. If other loci had significant effects on the phenotype, we would have observed additional significant QTL. As the size of our F2 population increases, we might have enough power to map additional loci that make small contributions to the phenotype. Some of these loci might converge directly on the Tbx5 and Pitx1 pathways, but we see no reason to assume that they will. However, this does not diminish the result that the two QTL we mapped in this study do indeed control a large proportion of phenotypic variance in the cross. Third, our genome scans comparing scale-footed and feather-footed breeds independently detected the same two chromosome regions as our QTL mapping study. If other loci exerted major effects on the phenotype or were selected repeatedly across breeds, we would have detected them in our selection scans. It remains possible that other loci have minor effects on the phenotype, and we might not have the power to detect them in our whole-genome scans of scale-footed and feather-footed breeds. Our results also do not exclude the possibility of allelic heterogeneity at modifier loci, which would make them harder to detect. However, these possibilities do not diminish the two highly significant associations we detected with our current data set. We agree that our misexpression data do not make a clear case for developmental sufficiency, but our results robustly support genetic sufficiency of these loci for foot feathering.

9) There is surprisingly little discussion of work on Tbx5 and Pitx1 function in the mouse. The authors must be aware that much of what they show and conclude is inconsistent with analyses of various mouse mutants of Tbx5 and Pitx1. This has been ignored largely, even though it would be exciting, if indeed true. The current interpretations of the results, without reference to the prior work in other systems, could further muddy an already messy literature on the roles of genes in determining the differences between fore limbs and hind limbs. It is therefore essential to clearly understand the functional significance of the QTL and expression analyses they have carried out. It would be interesting to test if elevating the levels of Pitx1 in a feathered breed would be sufficient to reduce feathers/replace with scales and reduce Tbx5 expression in the leg.

We agree that this is an important topic in the evolution and development of fore- and hind limb identity. In our response to comment 7, we provide a more in-depth discussion of previous work in both the mouse and chicken experimental systems. In short, we find that regulatory alterations to Tbx5 in pigeon and chicken breeds align nicely with previous functional data in chickens. We believe this work doesn't muddy the waters, but rather bolsters confidence in the original Tbx5 overexpression experiments in chickens. Pitx1 and Tbx5 undoubtedly retain similar functions among tetrapods (and other vertebrates), but we find no reason to presume that gene regulation will work identically in avians and mammals. Moreover, the common ancestor of mammals and birds lived more than 300 million years ago and the ectodermal structures that are the focus of this study (feathers) are not even present in mammals. Horton et al. (2008, Dev Genes Evol) speculate about an FGF-mediated mechanism that could explain part of the difference in the conflicting data for the role of Tbx5 in forelimb identity, and we now cite this paper in the manuscript. For example, in the Discussion, we now state:

“We note that, although genetic manipulations indicate that Tbx5 does not specify forelimb identity in mice, the divergence time between mammals and birds is deep (>300 million years) and subtly different roles for this transcription factor in limb outgrowth and identity might have evolved in these lineages (Minguillon et al. 2005; Horton et al. 2008).”

We also agree that the suggested Pitx1 misexpression experiment is worthwhile. We hope to conduct this experiment soon.

10) Have the authors considered whether the specific QTLs identified relate to expression of Tbx5 in the wing or other regions of the embryo or adult? What attempts have been made to demonstrate that the QTL are causative to the expression domain in the leg? Only one possible explanation (out of the many other alternatives) appear to have been explored and the final conclusion hangs on what appear to me to be unconvincing functional data.

Thank you for pointing out these important issues that need clarification. It is indeed possible that expression of Tbx5 is altered in other regions of the embryo; however, the embryonic wing is not one of these regions, as demonstrated by the expression assays reported in Figure 3—figure supplement 1 (Tbx5 forelimb expression among breeds is statistically indistinguishable). Moreover, whole-mount in situ hybridization of Tbx5 only revealed major Tbx5 expression pattern changes in the hind limbs of muffed embryos. Crucially, the two significant QTL and genomic regions were identified specifically and exclusively based on hind limb feathering phenotypes. These genomic regions might have an effect on development and feathering of other structures, but this issue was not addressed in, and is largely irrelevant to, our study. Two separate methodologies – QTL mapping and whole-genome scans – converged on the same two loci, and only these two loci, that exert quantitative (QTL mapping) and qualitative (genome scans) effects on foot feathering. After identifying these two loci based specifically on hind limb phenotypes, we expected to see coding changes – we did not find any fixed differences – or limb bud expression differences among breeds with different hind limb phenotypes. Therefore, we assayed limb bud expression among breeds to check the viability of strong candidate genes in these regions, and then showed definitively that Pitx1 and Tbx5 have cis-regulatory changes in a hybrid background. The hybrid background is important because alleles inherited from each parent are responding to the same trans-acting factors in a common cellular environment; thus, any differences in expression are attributable to cis-regulatory differences between parental alleles. This allele-specific expression experiment is crucial because it shows that expression differences in Pitx1 and Tbx5 are not just different among phenotypes, but they are also heritable: the feathered-foot allele of Tbx5 is expressed at a higher level in the hybrid hind limb bud, and the feathered-foot allele of Pitx1 is expressed at a lower level. In contrast, similar expression levels of different alleles in the hybrid background would confirm that differences among phenotypes are solely attributable to trans-acting regulatory changes (e.g., differential expression of an upstream transcription factor); however, we did not observe this outcome for either Pitx1 or Tbx5. To emphasize the importance of this experiment, we have moved the explanation of this experiment and its results into a new figure in the main text (Figure 4, formerly part of Figure 3—figure supplement 1). Further supporting our interpretation, the genome-wide H3K27ac ChIP-seq enrichment differences in limb buds from breeds with different limb phenotypes were significant in regions near Pitx1 and Tbx5. All of these analyses were specific to the embryonic (gene expression) or adult (phenotypes) hind limbs. Future experiments could indeed explore the role of Tbx5 in generating variation in feathering or other traits elsewhere in the embryo and adult; however, we intentionally drew genomic data from a wide variety of feather-footed breeds that do not share other obvious traits in common. Therefore, we would not expect to obtain similar results for other traits with our current data set.

11) Ptilopody refers to the presence of feathers in the shank. There are feathers on proximal parts of pigeon legs (both with scaled shank and feathered) that do not correlate with Tbx5 expression or reduction in Pitx1 expression. This is not mentioned or discussed. Is the assumption that the capacity of Tbx5 is localized to distal regions of the limb? There are of course feathers in many other regions of the bird where Tbx5 is not expressed also.

We believe that the presence of feathers throughout the body of a pigeon (including the regions shown in Figure 2, in which Tbx5 is not expressed) is well known, and is also displayed in the phenotypes of interest in Figure 1A. The presence of feathers in other parts of the limb are now also shown in Figure 3—figure supplement 3. Our focus for this paper is on sites of variation in scaled and feathered hind limb epidermis, which include only the distal portion of the limb. We did not investigate the roles of Tbx5 and Pitx1 in feather variation in other parts of the body (please also see our response to the previous comment).

Chicken breeds also show variation in feathering in the hindlimb-cochin and silkie chicken are analyzed to some extent. In silkie and some other chicken breeds (silky-feather) the basis of the mutation is known and/or causes other defects such as ploydactyly in Silkies. These do not correlate with the results and conclusions derived from this work in pigeon. These have not been discussed and are relevant to the significance of the proposed mechanisms in bird/dinosaur evolution.

We agree that our results on foot feathering in pigeons conflict somewhat with previous results in chicken models. However, to our knowledge, the specific mutation(s) leading to foot feathering in silkies and other breeds are not known. Indeed, we believe that the identification of specific candidate genes contributes to the novelty of our work in pigeons. A previous genetic mapping study implicated a chromosomal region around Pitx1 in ptilopody in silkies, but specific genes and mutations were not identified. In fact, the Pitx1 gene was not even mentioned as a candidate gene in that study (Dorshorst et al., 2010, J Hered). Polydactyly in silkies is a different genetic issue than ptilopody (these two loci are unlinked), and we do not address this issue in the manuscript. Ptilopodous silkies can be polydactyl, but not all are. Silky feathering throughout the body is also unlinked to ptilopody. We agree that the disparity between chickens and pigeons is important and we address it in some detail in the last paragraph of the Results section:

“Classical genetic studies implicate at least two loci in heavy foot feathering in chickens (Punnett and Bailey 1918; Lambert and Knox 1929; Warren 1948; Somes 1992), although the molecular genetic origins of the trait remain unknown. […] Furthermore, different populations of breeds such as silkies appear to have different constellations of ptilopody loci and alleles, and it is possible that we used strains that do not have Pitx1 mutations (Wexelsen 1934; Somes 1992).”

12) The shared genotype for feathered feet between pigeon and chicken may be partial since some chicken with feathered feet have Tbx5 in the leg bud, but no difference in Pitx1, which is on chromosome 13 where the chicken ptilopody locus is located. Can the authors analyze different chicken ptilopody variants to qualify their convergent evolution statement?

The common ancestor of chickens and pigeons did not have feathered feet, so we believe that the feathered feet of these two species is best described as a case of phenotypic convergence. If the reviewer is referring to breed “variants,” we already examined gene expression in two ptilopodous breeds, the Cochin and silke. None of our gene expression data point to a role for Pitx1 in ptilopody in silkies or Cochins, but as we state in the manuscript, we could be examining the wrong stage to implicate this gene. However, we demonstrate differences in Tbx5 expression between the hind limb buds of wild-type and ptilopodous breeds. We also generated silkie x white leghorn hybrid embryos and showed that ectopic expression of the feathered-foot Tbx5 allele is due (at least in part) to cis-regulatory changes, but we do not have genetic mapping or enough genomic resequencing data from chickens to directly test for involvement of specific sequence variants.

To emphasize that our results might account for some, but not all, of the ptilopody phenotype in chickens, we have added “in part” to the following statement:

“Hence, Tbx5-related developmental mechanisms may, in part, underlie the evolution of foot feathering in two species that last shared a common ancestor over 80 million years ago.”

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Point 1: The point was to distinguish that there can be changes in feather and scale dermis patterns that are distinct from the other mesoderm derivatives (e.g. skeleton) rather than anything about dosage and/or stage requirements.

Thank you for this clarification. We have addressed this comment in the revised manuscript, and the paragraph in question now reads:

“Collectively, these findings point to a partial alteration of the identity of the developing hindlimb, rather than localized changes to individual epidermal placodes. […] Therefore, we propose that different aspects of fore- and hindlimb morphology could have different dosage- and/or stage-dependent requirements for exposure to identity cues.”

We chose to retain a modified version of the last sentence in the revised manuscript because we believe it proposes an important and testable research hypothesis for future work.

Point 7: Although the text has been changed, the authors still seem to be implying that the longer fibula is a partial hindlimb to forelimb transformation which it need not be (Minor issue:the ulna does not articulate at the wrist, the radius does.)

We have altered this passage to highlight a previous interpretation about the enlargement of the fibula in experimentally induced hindlimb expression of Tbx5. We agree that the longer fibula of pigeons need not necessarily be due to a partial hindlimb to forelimb transformation, but given the strong evidence that epidermal identity is transformed, we believe we are justified in raising this issue and would be downplaying the significance of important experimental interpretations from others’ work by not doing so. We realize that not everyone will agree with these interpretations, but we feel that it is our responsibility to credit these experimental and interpretive precedents. The revised passage reads as follows:

“We also found that the fibula, which is normally splint-like and shorter than the tibia in pigeons, was enlarged (Figure 5D, E) and two phalanges of digit 4 were fused in feather-footed breeds (not shown). […] However, experimental ectopic expression of Tbx5 in the hindlimbs of chick embryos produces an enlargement of the fibula reminiscent of extreme pigeon phenotypes, and Takeuchi et al. (1999) compared this morphology to a forelimb-like condition (the fibula “makes a joint at its distal end like a normal ulna [the corresponding postaxial zeugopod bone of the forelimb].”

To tone down our argument further, we replaced “transformation” with “changes” in the following concluding sentence:

“Thus, the morphological changes to the hindlimbs of feather-footed pigeon breeds are considerably more than skin deep.”

We also deleted from the caption for Figure 5 a statement that mentions a comparison between the fibula and ulna.

Regarding the minor issue, we cannot find any description in the comparative anatomy literature that supports the statement about a lack of articulation between the ulna and wrist in avians, nor does our own experience with skeletal preparations of embryonic or adult pigeon material support this statement.

Please avoid the term limb-specific

Without guidance about the reason to modify this term, we are somewhat confused about how best to address this request. “Limb-specific” (including “forelimb-specific” and “hindlimb-specific”) is commonly used throughout the limb development and morphology literature to describe gene expression patterns that are specific to one or another set of limbs. Many of the instances of “limb-specific” in the previous version of the manuscript resulted from an earlier request to change “hindlimb” (the standard British spelling and a widely accepted format in the anatomical and developmental biology literature) to “hind limb”; therefore, “hindlimb-specific” (a term in widespread use) became “hind limb-specific,” which is comparatively unusual and awkward. We note that even the first and second round of reviewer comments on this manuscript use “hindlimb” in every case (e.g., Point 7, above), with the exception of the comment in the first review to make the change from “hindlimb” to “hind limb.” Therefore, with clarity and convention in mind, we elected to change “hind limb” to “hindlimb” in the revised manuscript.

To further clarify what is meant by limb specificity, we have expanded the descriptions of Pitx1 and Tbx5 limb expression domains when these genes are first mentioned:

“The highest pFst peak on scaffold 79 – corresponding to the major-effect QTL on LG 11 for the proportion of tarsometatarsal feathering – was approximately 200 kb upstream of Pitx1, a gene that encodes a homeobox-containing transcription factor that is normally expressed in the vertebrate hindlimb but not the forelimb (Figure 2B). The highest pFst peak on scaffold 70 – corresponding to the major-effect QTL on LG 20 for toe feather length – was 40 kb upstream of Tbx5, a gene that encodes a T-box transcription factor that is normally expressed in the vertebrate forelimb but not the hindlimb (Figure 2D).”

We made several other changes to cut down on the number of times “fore/hindlimb- specific” appears in the text, but we believe this terminology is usually the most meaningful, accurate, and clear way to convey information. We welcome suggestions for alternative terminology.

Subheading “Evolution of epidermal appendage distribution” avialan to avian

This change would alter the meaning of the sentence, so we elected to delete the word entirely. Doing so actually broadens the scope of the question we are asking, which we see as a positive outcome:

“How might pigeons help us understand the evolution of epidermal appendage distribution and limb morphology in other species?”

We also use “avialan” later in the same paragraph to convey an interpretation by Zheng et al. 2013. Changing this term to “avian” would be incorrect and change the meaning of the statement.

https://doi.org/10.7554/eLife.12115.028

Article and author information

Author details

  1. Eric T Domyan

    Department of Biology, University of Utah, Salt Lake City, United States
    Present address
    Department of Biology, Utah Valley University, Orem, United States
    Contribution
    ETD, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  2. Zev Kronenberg

    Department of Human Genetics, University of Utah, Salt Lake City, United States
    Present address
    Department of Genome Sciences, University of Washington, Seattle, United States
    Contribution
    ZK, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
  3. Carlos R Infante

    Department of Genetics, University of Georgia, Athens, United States
    Present address
    Department of Molecular and Cellular Biology, University of Arizona, Tucson, United States
    Contribution
    CRI, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  4. Anna I Vickrey

    Department of Biology, University of Utah, Salt Lake City, United States
    Contribution
    AIV, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  5. Sydney A Stringham

    Department of Biology, University of Utah, Salt Lake City, United States
    Contribution
    SAS, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  6. Rebecca Bruders

    Department of Biology, University of Utah, Salt Lake City, United States
    Contribution
    RB, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  7. Michael W Guernsey

    Department of Biology, University of Utah, Salt Lake City, United States
    Present address
    Department of Genetics, Stanford University, Stanford, United States
    Contribution
    MWG, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  8. Sungdae Park

    Department of Genetics, University of Georgia, Athens, United States
    Contribution
    SP, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  9. Jason Payne

    Poultry Science Department, University of Georgia, Athens, United States
    Contribution
    JP, Acquisition of data, Drafting or revising the article, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
  10. Robert B Beckstead

    Poultry Science Department, University of Georgia, Athens, United States
    Contribution
    RBB, Conception and design, Drafting or revising the article, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
  11. Gabrielle Kardon

    Department of Human Genetics, University of Utah, Salt Lake City, United States
    Contribution
    GK, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  12. Douglas B Menke

    Department of Genetics, University of Georgia, Athens, United States
    Contribution
    DBM, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon 0000-0002-7109-1451
  13. Mark Yandell

    1. Department of Human Genetics, University of Utah, Salt Lake City, United States
    2. Utah Center for Genetic Discovery, University of Utah, Salt Lake City, United States
    Contribution
    MY, Conception and design, Analysis and interpretation of data, Drafting or revising the article, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
  14. Michael D Shapiro

    Department of Biology, University of Utah, Salt Lake City, United States
    Contribution
    MDS, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    shapiro@biology.utah.edu
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon 0000-0003-2900-4331

Funding

National Institutes of Health (F32GM103077 Fellowship)

  • Eric T Domyan

National Institutes of Health (T32HD07491 Fellowship)

  • Eric T Domyan

National Institutes of Health (T32GM007464 Fellowships)

  • Zev Kronenberg
  • Sydney A Stringham

National Science Foundation (EDEN RCN Internship IOS0955517)

  • Anna I Vickrey

National Science Foundation (Graduate Research Fellowship)

  • Rebecca Bruders

National Institutes of Health (R01HD053728)

  • Gabrielle Kardon

National Science Foundation (CAREER IOS1149453)

  • Douglas B Menke

National Institutes of Health (R01GM104390)

  • Mark Yandell

National Institutes of Health (R01GM115996)

  • Michael D Shapiro

Burroughs Wellcome Fund (Career Award in the Biomedical Sciences)

  • Michael D Shapiro

National Science Foundation (CAREER DEB1149160)

  • Michael D Shapiro

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Della Fixsen, Cassandra Garner, Kamala Ganesh, Patrick Miller, the University of Utah Genomics Core, and the University of Utah Small Animal Imaging Core for technical assistance, and members of the Utah Pigeon Club and National Pigeon Association for generously providing samples. We also thank Elena Boer for comments and discussion on the manuscript. Wild-type pigeon image in Figure 5D is courtesy of Dr. M Scott Echols, Grey Parrot Anatomy Project, University of Utah and the Medical Center for Birds (Oakley, California). This work was supported by the National Science Foundation (CAREER DEB1149160 to MDS, CAREER IOS1149453 to DBM, IOS0955517 EDEN internship to AIV), a Burroughs Wellcome Career Award in the Biomedical Sciences (MDS), and the National Institutes of Health (R01GM115996 to MDS, F32GM103077 and T32HD07491 fellowships to ETD, T32GM007464 fellowships to ZK and SAS, R01HD053728 to GK, R01GM104390 to MY). We acknowledge a computer time allocation from the Center for High Performance Computing at the University of Utah. Illumina shotgun reads for two Pomeranian pouter pigeons are deposited in NCBI BioProject ID PRJNA284526. H3K27ac ChIP-seq data are deposited in the Gene Expression Omnibus (GEO accession number GSE67875). Correspondence should be addressed to MDS (shapiro@biology.utah.edu).

Ethics

Animal experimentation: This study was performed in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled and housed according to approved University of Utah institutional animal care and use committee (IACUC) protocols 10-05007 and 13-04012.

Reviewing Editor

  1. Marianne E Bronner, Reviewing Editor, California Institute of Technology, United States

Publication history

  1. Received: October 7, 2015
  2. Accepted: February 19, 2016
  3. Version of Record published: March 15, 2016 (version 1)

Copyright

© 2016, Domyan et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,992
    Page views
  • 573
    Downloads
  • 5
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Comments

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Cell Biology
    Yi Liu et al.
    Research Article Updated