* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Adaptive Evolution of 5#HoxD Genes in the
Group selection wikipedia , lookup
Dual inheritance theory wikipedia , lookup
Quantitative trait locus wikipedia , lookup
Site-specific recombinase technology wikipedia , lookup
Designer baby wikipedia , lookup
Artificial gene synthesis wikipedia , lookup
Population genetics wikipedia , lookup
Koinophilia wikipedia , lookup
Genome (book) wikipedia , lookup
Genomic imprinting wikipedia , lookup
Minimal genome wikipedia , lookup
Genome evolution wikipedia , lookup
Ridge (biology) wikipedia , lookup
Biology and consumer behaviour wikipedia , lookup
Gene expression programming wikipedia , lookup
Gene expression profiling wikipedia , lookup
Adaptive evolution in the human genome wikipedia , lookup
Epigenetics of human development wikipedia , lookup
Adaptive Evolution of 5#HoxD Genes in the Origin and Diversification of the Cetacean Flipper Zhe Wang,* Lihong Yuan, Stephen J. Rossiter,à Xueguo Zuo,* Binghua Ru,* Hui Zhong,* Naijian Han,§ Gareth Jones,k Paul D. Jepson,{ and Shuyi Zhang* *School of Life Science, East China Normal University, Shanghai, China; Guangdong Entomological Institute, Guangzhou, China; àSchool of Biological and Chemical Sciences, Queen Mary, University of London, London, United Kingdom; §Institute of Zoology, Chinese Academy of Sciences, Beijing, China; kSchool of Biological Sciences, University of Bristol, Bristol, United Kingdom; and {Institute of Zoology, Zoological Society of London, Regent’s Park, London, United Kingdom The homeobox (Hox) genes Hoxd12 and Hoxd13 control digit patterning and limb formation in tetrapods. Both show strong expression in the limb bud during embryonic development, are highly conserved across vertebrates, and show mutations that are associated with carpal, metacarpal, and phalangeal deformities. The most dramatic evolutionary reorganization of the mammalian limb has occurred in cetaceans (whales, dolphins, and porpoises), in which the hind limbs have been lost and the forelimbs have evolved into paddle-shaped flippers. We reconstructed the phylogeny of digit patterning in mammals and inferred that digit number has changed twice in the evolution of the cetacean forelimb. First, the divergence of the early cetaceans from their even-toed relatives coincided with the reacquisition of the pentadactyl forelimb, whereas the ancestors of tetradactyl baleen whales (Mysticeti) later lost a digit again. To test whether the evolution of the cetacean forelimb is associated with positive selection or relaxation of Hoxd12 and Hoxd13, we sequenced these genes in a wide range of mammals. In Hoxd12, we found evidence of Darwinian selection associated with both episodes of cetacean forelimb reorganization. In Hoxd13, we found a novel expansion of a polyalanine tract in cetaceans compared with other mammals (17/18 residues vs. 14/15 residues, respectively), lengthening of which has previously been shown to be linked to synpolydactyly in humans and mice. Both genes also show much greater sequence variation among cetaceans than across other mammalian lineages. Our results strongly implicate 5#HoxD genes in the modulation of digit number, web forming, and the high morphological diversity of the cetacean manus. Introduction Modifications in the autopod and in the number of digits have occurred recurrently during tetrapod evolution and are generally viewed as adaptive locomotory responses to changes in lifestyle and habitat use (Lull 1904; Tabin 1992; Shubin and Marshall 2000). In mammals, the ancestral amniote manus is typically characterized by five separate digits (pentadactyly), with each digit consisting of carpals, metacarpals, and phalanges running from proximal to distal axes. In general, each digit ray consists of one metacarpal and either two or three phalanges. Although deviations from this pentadactyl state are seen in several groups (Lull 1904; Golley and Whitmoyer 1959), major evolutionary reorganization of the manus is restricted to two clades, Cetartiodactyla (Artiodactyla þ Cetacea) and Perissodactyla. The ungulate manus shows several cursorial adaptations for increased speed in a terrestrial environment. In modern perissodactyls, digit loss has occurred in horses, which are functionally monodactyl (digit III) with vestigial digits II and IV, and also in tapirs and rhinoceroses, which possess four and three digits, respectively. Similar reduction in digit size and number has occurred during the evolution of artiodactyls. Specifically, the reduced first metacarpal of the pentadactyl basal taxon Diacodexis (Rose 1982) is absent in modern lineages, which also show a reduction or loss of digits II and V. Perhaps the most dramatic evolutionary reorganization of the autopod has occurred in the cetaceans (whales, dolphins, and porpoises), the common ancestor of which diverged from an extinct group of deer-like artiodactyls Key words: Hoxd12, Hoxd13, adaptive evolution, cetaceans, digits, flipper. E-mail: [email protected]; [email protected]. Mol. Biol. Evol. 26(3):613–622. 2009 doi:10.1093/molbev/msn282 Advance Access publication December 10, 2008 Ó The Author 2008. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: [email protected] approximately 50 Ma (Thewissen et al. 2007). Changes in limb morphology associated with the return to water followed three main routes. First, cetacean digits became embedded in skin tissue (an interdigital web), a state also seen in other mammals (e.g., bats and seals). Second, cetaceans lost most of their lower limb bones, and some also lost digits in their upper limbs, which evolved into paddle-shaped flippers (Berta et al. 2006; Thewissen et al. 2006). Specifically, odontocetes (toothed whales) and some mysticetes (baleen whales) possess a pentadactyl manus, whereas other mysticetes have lost digit I (Berta et al. 2006; Cooper et al. 2007). Third, cetaceans show considerable variation in their number of phalanges, with digits I and V typically characterized by few phalanges (ranging from zero to five) and digits II–IV by as many as 13 in the second digital ray of some species. Indeed, the carpals of many cetaceans show clear species-specific polymorphisms, and phalangeal count can even vary among conspecifics (Cooper et al. 2007). In recent years, considerable progress has been made in our understanding of the molecular basis of digit pattern control. Many genes expressed in the embryonic limb bud have been shown to play an important role in autopod formation (Zakany et al. 2004; Tickle 2006). The Hox genes Hoxd12 (Homeobox D12) and Hoxd13 (Homeobox D13) of the 5#HoxD cluster both show strong expression in posterior cells of limb buds (Papageorgiou 2007), and comparative studies suggest that their expression (Davis et al. 2007; Freitas et al. 2007), coding sequences, and even noncoding elements (Lee et al. 2006; Lemons and McGinnis 2006) are highly conserved across vertebrates. Much knowledge has been gained by genetic manipulation or by screening deformed individuals. For example, Hoxd12 defects in mice are associated with deformities in carpals, metacarpals, and phalanges (Knezevic et al. 1997), whereas Hoxd13-defective humans exhibit autopod malformations 614 Wang et al. characterized by extra digits (polydactyly), short digits (brachydactyly), and fused digits (syndactyly) (Muragaki et al. 1996; Zhao et al. 2007). Several studies have also linked the autosomal-dominant condition of synpolydactyly (polydactyly with syndactyly), in which digits are webbed and duplicated, to a polyalanine repeat expansion at the 5#-end of the Hoxd13 protein (Johnson et al. 1998). Moreover, Hoxd12 and Hoxd13, together with other limbpatterning genes, have been implicated in more severe forelimb defects (Davis and Capecchi 1996). The Hoxd12 protein directly amplifies the posterior Shh polarizing signal in a reinforcing positive feedback loop and interacts with Gli3, both of which also regulate digit number and identity (Knezevic et al. 1997; Litingtung et al. 2002). More recently, Hoxd12 and Hoxd13 have also been proposed to modulate digit number (Sheth et al. 2007; Zakany and Duboule 2007). Here we use a molecular evolution approach to test whether key evolutionary transitions in the limbs of mammals have involved the Hox genes Hoxd12 and Hoxd13. We conduct a wide survey of mammals, including representatives from Cetartiodactyla and Perissodactyla, and apply codon substitution models to identify lineages and sites under Darwinian selection. Additionally, we examine the number of repeat units in a polyalanine stretch in Hoxd13 that is implicated in major limb defects in humans and mice. Materials and Methods Taxonomic Coverage Tissue samples held at the School of Life Science, East China Normal University, Shanghai, China, and the Institute of Zoology, London, United Kingdom, were obtained for 22 mammal species. These comprised 10 cetaceans (Balaenoptera acutorostrata, Balaenoptera physalus, Megaptera novaeangliae, Physeter macrocephalus, Kogia breviceps, Ziphius cavirostris, Mesoplodon densirostris, Phocoena phocoena, Lagenorhynchus albirostris, and Stenella coeruleoalba), four bats (Rousettus leschenaulti, Hipposideros armiger, Megaderma lyra, and Taphozous melanopogon), two artiodactyls (Sus scrofa and Capra hircus), one perissodactyl (Equus caballus), two carnivores (Felis catus and Arctonyx collaris), one eulipotyphlan (Erinaceus europaeus), one rodent (Petaurista alborufus), and one lagomorph (Oryctolagus cuniculus). From GenBank, we obtained Hoxd12 sequences for three primates(Homo sapiens, NM_021193; Pan troglodytes, XM_515926; and Papio hamadryas, AC116665), two rodents (Rattus norvegicus, XM_345373; Mus musculus, NM_008274), one bat (Carollia perspicillata, AY744676), one carnivore (Canis familiaris, XM_545535), one artiodactyl (Bos taurus, XM_598664), and one perissodactyl (E. caballus, XM_001500102). Similarly, we obtained Hoxd13 sequences for three primates (H. sapiens, NC_000002; P. troglodytes, XM_525967; and P. hamadryas, AC116665), two rodents (R. norvegicus, NM_001105886; M. musculus, NM_008275), one bat (C. perspicillata, AY744676), and one carnivore (C. familiaris, AY308818). We also obtained Ensembl predicted Hoxd12 sequences from published genomes (http://genome.ucsc.edu/and http://www.ensembl .org) for the primates Macaca mulatta, Callithrix jacchus, Microcebus murinus, Pongo pygmaeus, and Otolemur garnettii, and the predicted Hoxd13 sequences for the artiodactyl B. taurus and the primates C. jacchus and P. pygmaeus. Polymerase Chain Reaction (PCR) Amplification and Sequencing Genomic DNA was isolated using DNeasy Tissue kits (Qiagen, Hilden, Germany) and used to amplify the Hoxd12 and Hoxd13 genes via the PCR. For Hoxd12, we designed one pair of primers (forward: 5#-TTG GAG ATG TGT GAG CGC AGT CTC TAC AG-3#; reverse: 5#-TTC ATA CG(A/C) CGG TTC TGG AAC CA-3#) and for Hoxd13, we designed separate pairs of primers to span the gene, with an overlap of 75 bp. We refer to these as primer set 1 (5#-ATG GAC GGG CTG CGG (A/G)C(A/ G) GAC G-3# and 5#-CC(C/T) GAC ACG TCC ATG TAC TTC TCC AC-3#) and primer set 2 (5#-GCA (A/ G)AA TGC TCT (C/T)AA GTC (G/C)TC GCC-3# and 5#-CCA A(G/C)C TGG (T/A/C)CC ACA TCA (A/ G)GA GAC AGT (A/G)TC-3#), respectively. To verify the Hoxd13 sequences and extend the overlap, we also amplified a 317-bp sequence using a set of published primers (5#-CT(G/C) GGC TA(C/T) GGC TAC CAC-3# and 5#TG(G/A) TA(C/G) CCC TCC ATG GAG ATG-3#; Chen et al. 2005). PCR products for both genes were ligated into a pMD19-T vector (TaKaRa, Kyoto, Japan), cloned and sequenced using the Big Dye Terminator kit on an ABI 3730 DNA sequencer (Applied Biosystems, Foster City, CA). Finally, the sequences of both genes were edited to remove an intron and the sections corresponding to the primers, thus yielding 721 bp (89%) and 965–971 bp (96%) of the protein-coding region of Hoxd12 and Hoxd13, respectively. GenBank accession numbers of all new sequences are FJ455455–FJ455496 inclusive. Ancestral State Reconstruction To reconstruct the ancestral digit pattern of the forelimb for different mammalian lineages across species (Murphy et al. 2001; Price et al. 2005; Teeling et al. 2005; O’Leary and Gatesy 2008), we used the parsimony method in Mesquite 2.01 (Maddison WP and Maddison DR 2007), which minimizes the number of steps between character state changes. We recorded the observed distribution of digit and phalanx number among extant taxa, and reconstructed the ancestral states at nodes, for six characters: the total number of digits per manus and the number of phalanges for each of the digits I–V (see fig. 1). Parsimony reconstruction was undertaken assuming both ordered and unordered models of character change. In the former, the cost of change from state i to state j is assumed to equal |i – j| steps, whereas for the latter, one state change is modeled as one step. Ancestral state reconstruction of forelimb digit number was also undertaken using a maximum likelihood (ML) approach in Mesquite 2.01. However, because this approach does not support uncertain or polymorphic character states, it was not used to infer phalanx number. Adaptive Evolution of 5#HoxD Genes in Cetaceans 615 FIG. 1.—Digit and phalanx number for the forelimb of extant and extinct mammals. Character states of ancestral and extant taxa are shown on the branches and after the taxon names, respectively. In each case, numbers follow the format: digit number (phalanx number of digit I, digit II, digit III, digit IV, and digit V). For the extinct taxon Diacodexis, the digit number is five, but the number of phalanges is uncertain. Values above and below the branches correspond to the results of ordered parsimony and ML reconstructions, respectively. The names of fossil species are framed, and the names of extant species which were not included in our analyses of molecular evolution are underlined. An asterisk represents the consensus digit pattern of 5(2,3,3,3,3). The four positions in the tree in which digit number is known to have changed are denoted by I, II, III, and IV. Note, in cases where the number of phalanges is zero, the metacarpal, and thus the digit, can still exist. Lower case letters on the tree illustrate the ancestral branches of the cetaceans (c) and tetradactyl mysticetes (tm). Upper case letters represent taxonomic groupings as follows: P, Perissodactyla; C, Carnivora; E, Eulipotyphla; R, Rodentia; and L, Lagomorpha. We also performed the same methods to reconstruct the ancestral states of the length of the polyalanine stretch in Hoxd13. Here, we considered 35 extant species (see fig. 2), and the single character was the number of alanine residues. Molecular Evolution Analyses To test for evidence of molecular adaptation in Hoxd12 and Hoxd13, we used the CODEML program in PAML 4 (Yang 2006, 2007) to derive ML estimates of the rates of synonymous and nonsynonymous substitutions (dS and dN, respectively) and the dN/dS ratio (omega, x). Positive selection is evident where x . 1, neutral evolution where x 5 1, and purifying selection where x , 1 (Nei and Kumar 2000). Rates were estimated along specific branches in a tree, the topology of which was constrained based on published studies (Murphy et al. 2001; Price et al. 2005; Teeling et al. 2005; O’Leary and Gatesy 2008) and is shown as figure 2. For all PAML-based analyses, alignment gaps were treated as ambiguity characters (setting cleandata 5 0). Branch lengths were estimated simultaneously (iteration setting method 5 0), and codon frequencies were calculated from the average nucleotide frequencies at the three codon positions (setting CodonFreq 5 2 (F3X4)). We tested for positive selection on the branches ancestral to all cetaceans, to tetradactyl mysticetes, to perissodactyls, and to cetartiodactyls using branch models. Specifically, a model in which x was fixed across the tree (one-ratio) was compared with models in which x was allowed to differ between the background and a focal branch (two-ratio model), or between the background and among two focal branches. To test whether the x ratio of each focal branch was significantly greater than both the background ratio and one (Yang 1998), we performed likelihood ratio tests (LRTs). The LRT statistic (2Dl) approximates to a chi-square distribution and can be used to compare nested likelihood models. Where we found evidence of a significant change in selection pressure (x) in an ancestral branch, we explored this further with the free-ratio model, in which x is assumed to be independent on each branch of the tree. To identify the sites under positive selection along the Hox genes, we implemented site models in which x can vary among sites. However, because positive selection often operates episodically at small numbers of sites and is thus difficult to detect (Zhang et al. 2005), we also applied the branch-site model A, in which x can vary among sites along specific lineages (Yang et al. 2005). Separate branchsite models were implemented with branches c and tm set as the foreground branches, and model significance was tested using the LRTs termed branch-site Tests 1 and 2 (Zhang et al. 2005). In the former, branch-site model A is compared with two degrees of freedom to a site model (M1a, ‘‘Nearly Neutral’’) with two site classes: 0 , x0 , 1 and x1 5 1 fixed. In the latter, branch-site model A is compared with one degree of freedom to an equivalent model in which x2 is fixed at 1. Test 1 is unable to distinguish relaxed constraint from positive selection, whereas test 2 is a direct test for positive selection on the lineage of interest, but appears 616 Wang et al. FIG. 2.—Phylogeny of 35 mammals used for evolutionary analysis of Hoxd12 and Hoxd13. The topology is based on Murphy et al. (2001), Price et al. (2005), and Teeling et al. (2005). Note, for analyses of Hoxd13, the taxon Equus caballus was replaced with Equus asinus, and the sequences of three primates (Macaca mulatta, Otolemur garnettii, and Microcebus murinus) and the carnivore Canis familiaris were not available from GenBank. Numbers above the branches refer to the lengths of polyalanine repeats at the 5#-end coding region of Hoxd13 (see Results). Branches c (common ancestors of cetaceans) and tm (common ancestors of tetradactyl mysticetes) are the major lineages of interest in the analysis of this study. Upper case letters represent taxonomic groupings as follows: P, Perissodactyla; C, Carnivora; E, Eulipotyphla; R, Rodentia; and L, Lagomorpha. conservative overall (Zhang et al. 2005). The Bayes Empirical Bayes (BEB) method for identifying positively selected sites was applied to the results of branch-site model A. Finally, to visualize variation in x along each gene, we undertook a sliding window analysis using the software SWAAP1.0.2 (Pride 2000). We set the window size at 60 bp (20 codons) and the step size at 12 bp (4 codons). Values of x were estimated following Nei and Gojobori (1986). Results Digit Pattern in the Mammalian Forelimb Parsimony-based ancestral reconstructions of mammalian forelimb digit pattern yielded broadly similar results based on ordered and unordered models of character change, and ancestral digit numbers also agreed with the results of ML reconstructions (fig. 1). Four major transitions in digit number were detected. Reductions in the number of digits occurred in the ancestral lineages of both the perissodactyls and cetartiodactyls. Conversely, an increase in digit number was observed in the ancestral branch of the cetaceans, whereas a digit was then lost in the ancestral tetradactyl mysticetes. In terms of phalanx number, considerable variation is evident both among ancestral and modern cetaceans. However, digits I and V typically show reduced numbers of phalanges, whereas digits II, III, and IV show increased numbers. Alanine Repeat Length Polymorphism in Hoxd13 A polyalanine tract at the 5#-end of Hoxd13 was expanded in all the cetacean species studied (fig. 3). The number of alanine repeats ranged from 17 to 18 in modern cetaceans compared with 14 to 15 in all other mammals. The nucleotide alignment indicated that two or three alanine residues have become inserted into the middle of the polyalanine stretch in cetaceans. Ancestral reconstructions of the number of alanine repeats were identical for both parsimony (ordered and unordered) and ML models of character change (fig. 2). We found that two alanines were gained in the common ancestor of all cetaceans, which was assigned 17 units. Three independent cases of repeat gain were also detected within the cetaceans, on the branches leading to Adaptive Evolution of 5#HoxD Genes in Cetaceans 617 FIG. 3.—Alignment of polyalanine repeats tract in Hoxd13. Sequences for 32 mammal species were aligned in ClustalX (Thompson et al. 1997) and, where indels were present, by eye based on translated amino acid sequences using MEGA 4.0 (Tamura et al. 2007). Short dashes (–) represent indels and numbers after the sequences indicate the counts of polyalanine residues. Alleles isolated from both a human and mouse exhibiting the morphological defect synpolydactyly are also included (Muragaki et al. 1996; Johnson et al. 1998). In both cases, there is an insertion of seven alanine residues into the middle of the polyalanine stretch. B. acutorostrata, P. phocoena, and K. breviceps. (all possess 18 repeats). Among noncetaceans, all ancestral nodes were assigned 15 repeat units; however, three species (rabbit, dog, and badger) were characterized by 14 units and, therefore, have subsequently lost an alanine. Molecular Evolution of Hoxd12 and Hoxd13 in Mammals Separate one-ratio models undertaken for all mammals, and for cetaceans only, revealed comparable trends for both genes. For Hoxd12, the x value was 0.105 for all mammals and 0.320 for cetaceans, whereas for Hoxd13, the x value was 0.024 for all mammals and 0.063 for cetaceans. Therefore, the ratio of nonsynonymous to synonymous substitutions, indicative of a change in selection pressure, was approximately three times higher in cetaceans than in all mammals. The results of the branch models, with log likelihood values, parameter estimates, and LRTs for hypothesis testing, are given in table 1. For Hoxd12, separate two-ratio branch models in which x was allowed to differ from the estimated background ratio in the ancestral branch of all 618 Wang et al. Table 1 Log Likelihood Values and Parameter Estimates for Hoxd12 (35 Sequences) under Different Branch Models, with Likelihood Ratio Tests for Testing Hypotheses (2Dl) Model npa l x0b xcb xtmb A. One ratio: x0 5 xc 5 xtm B. Two ratios: x0 5 xtm, xc C. Two ratios: x0 5 xc, xtm D. Two ratios: x0, xc 5 xtm 69 70 70 70 4,704.54 4,701.05 4,701.94 4,698.93 0.105 0.103 0.104 0.101 5 x0 0.785 5 x0 1.110 5 x0 5 x0 N 5 xc E. Three ratios: x0, xc, xtm F. Two ratios: x0 5 xc, xtm 5 1 G. Two ratios: x0, xc 5 xtm 5 1 H. Three ratios: x0, xc, xtm 5 1 71 69 69 70 4,698.41 4,702.43 4,698.94 4,698.90 0.101 0.104 0.101 0.101 0.785 5 x0 1 0.785 N 1 1 1 a b Models Compared 2Dl P Value B vs. A C vs. A D vs. A E vs. B E vs. C E vs. D C vs. F D vs. G E vs. H 6.98 5.20 11.22 5.28 7.06 1.04 0.98 0.02 0.98 0.0082 0.0226 0.0008 0.0216 0.0079 0.3078 0.3222 0.8875 0.3222 np, number of parameters. Parameters xc, xtm, and x0 are the x ratios for branches c, tm, and all other branches, respectively (see fig. 2). cetaceans (branch c in fig. 2), the ancestral branch of the tetradactyl mysticetes (branch tm in fig. 2), or both together (xc 5 xtm) showed significant model improvement over a one-ratio model, with higher x values in the foreground branches. The model fit of the former two tworatio models was improved further by the addition of a third ratio, suggesting that the x also differs between the branches c and tm. In this three-ratio model, the x values for branch c and the background were 0.785 and 0.101, respectively, whereas the x value for the branch labeled tm was ‘‘infinite’’ (i.e., dS 5 0 and dN . 0). A comparison of the two-ratio model (x0 5 0.101, xc 5 xtm 5 1.110) with the three-ratio model was not significant (LRT 5 1.04, P 5 0.3078, df 5 1), suggesting that the two-ratio model, of which x values of branch c and branch tm were similar and both higher than one, was not rejected. Two- and three-ratio models were not found to fit the data better than equivalent models in which the foreground x ratio was fixed at one. Therefore, accelerated or adaptive evolution has occurred in branches c and tm. The same branch model tests were also performed for Hoxd13; however, neither accelerated evolution nor positive selection was detected (data not shown). Additionally, separate tworatio models applied to ancestral branches of the perissodactyls and the cetartiodactyls also showed no significant improvement over a one-ratio model for either gene (data not shown). Site models (M0, M1a, M3, M7, and M8), in which x was allowed to vary among codons, did not reveal any positively selected sites in the protein-coding regions of Hoxd12 and Hoxd13 (data not shown). However, positively selected sites along branches c and tm were identified in branch-site model A for both genes, the log likelihood values and parameter estimates of which, along with LRT results, are given in supplementary table S1, Supplementary Material online. For Hoxd12, we detected seven sites under positive selection on branch c based on branch-site model A, of which six had BEB values .0.90 (33S, 60N, 149S, 154E, 155T, and 174Q). The LRT test statistic (2Dl) of Test 1 was 8.84 (P 5 0.012, df 5 2), indicating that model A fit the data better than did model M1a, and the LRT test statistic of Test 2 was zero (nonsignificant, df 5 1). These results provide evidence of a change in selection pressure, and reject neutral evolution, on the branch leading to the cetaceans, though it is not possible to reject accelerated evolution due to relaxed selection. On the other hand, for lineage tm, though two sites were associated with high BEB posterior probabilities (57S with P 5 0.84, 60N with P 5 0.99), the LRT test statistics were not significant for either Test 1 (2Dl 5 3.98, P 5 0.1367, df 5 2) or Test 2 (2Dl 5 1.00, P 5 0.3173, df 5 1) and so we cannot reject neutral evolution. For Hoxd13, a moderately high BEB value was only found for a single site in each of branch c (74S with P 5 0.62) and tm (76S with P 5 0.75). The results of Tests 1 and 2 were not significant, either for branch c (2Dl 5 0.52, P 5 0.7711, df 5 2; 2Dl 5 0, P 5 1, df 5 1, respectively) or branch tm (2Dl 5 2.02, P 5 0.3642, df 5 2; 2Dl 5 0, P 5 1, df 5 1, respectively). Thus, on the basis of these models, neutral evolution cannot be rejected for this gene. A free-ratio model, which assumes independent x values for each branch, was generated for both genes in the cetaceans only (supplementary fig. S1, Supplementary Material online). Estimates of x for Hoxd12 ratios were infinite or greater than one in four branches: the ancestral branch of the three tetradactyl mysticetes, the ancestral branch of Megaptera novaeangliae and B. physalus, the branch leading to B. acutorostrata, and the branch leading to K. breviceps. For Hoxd13, x values were infinite in the branch leading to Stenella coeruleoalba and the branch leading to M. densirostris. However, for both genes, the free-ratio model was not better than the oneratio model (supplementary table S2, Supplementary Material online). The results of our sliding window analyses of dN/dS (x) across the protein-coding regions of Hoxd12 and Hoxd13 are shown in figure 4. Estimates of x of both genes are uniformly low, indicative of purifying selection, based on all mammals. In contrast, x estimates show dramatic variation along the genes in the cetaceans and exceeds one in two main sections of Hoxd12 and one section in Hoxd13. However, because these sections also include the sites identified in the branch-site model A (see above), this result is unlikely to be statistically significant. In both data sets, nonsynonymous substitutions were markedly more concentrated outside of the homeodomain, Adaptive Evolution of 5#HoxD Genes in Cetaceans 619 FIG. 4.—Sliding window analysis of dN/dS (x). Estimates are based on Nei–Gojobori method (Nei and Gojobori 1986) and given across the protein-coding regions of Hoxd12 (A) and Hoxd13 (B). In each case, the window was set at 60 bp and step at 12 bp. in areas that, to our knowledge, are not known or speculated to be key functional domains. Discussion The critical importance of the Hox genes for normal growth and patterning of the autopod skeleton has been established from screening individuals with morphological defects (Muragaki et al. 1996; Johnson et al. 1998; Zhao et al. 2007), as well as via both loss- (Herault et al. 1997; Tarchini et al. 2006) and gain-of-function (Kmita et al. 2000; Spitz et al. 2005) manipulations. Yet comparatively little consideration has been given to the role of Hox genes in limb structure and modification over an evolutionary time frame. Consequently, it is generally thought that these Hox genes are likely to be under strong selective constraint in tetrapods, in spite of several episodes of major limb reorganization. Our results suggest that Hoxd12 and Hoxd13 have undergone molecular adaptation in the cetaceans and that these changes correspond to major rearrangements in the number of digits. The cetacean fossil record provides one of the most spectacular examples of evolutionary transition of all mammals, from the small semiaquatic tetrapodal forms of the lower Eocene (Thewissen et al. 2007) to the large fully aquatic species of the late Eocene, in which the hind limbs were already reduced in size and would later disappear altogether (Thewissen et al. 2006). Fossil evidence and ancestral state reconstruction (Sober 1991; Shimamura et al. 1997; Nikaido et al. 1999; Price et al. 2005) both suggest that the ancestors of the cetaceans possessed five digits (Gingerich et al. 2001; Berta et al. 2006) and, therefore, might have regained a digit after their earlier divergence from the hippopotomids (Nikaido et al. 1999), whose forelimbs have four digits (Boisserie et al. 2005). An alternative scenario, which seems more plausible in terms of morphology, but which is less concordant with the result of our selection tests, is that digit number in the early cetaceans is a primitive state and that independent reductions in digit number have occurred in four ancestral artiodactyl lineages, leading to the camels, the pigs, the common ancestor of cows and goats, and the hippopotomids. The broad pentadactyl flipper is thought to be well adapted for the slow maneuverable swimming necessary for turning and prey capture in shallow lagoons (Benke 1993). Similarly, the subsequent loss of a digit (I) in some mysticetes (Cooper et al. 2007) in the late Oligocene (Fitzgerald 2006; Demere et al. 2008) conferred a more narrow, elongate flipper that is considered well adapted for swimming at high speeds, which is necessary for filter feeding (Goldbogen et al. 2006; Woodward et al.2006). The results of our analyses of the molecular evolution of Hoxd12 are thus consistent with these adaptive explanations for transitions in digit number associated with changes in feeding behavior, prey and habitat use, with significantly higher omega ratios found on the same two branches (ancestral to the cetaceans and to the tetradactyl mysticetes) than on the background (see table 1). Although branch model comparisons were unable to reject models in which the omega value of these branches was fixed at one, models of positive selection along branches c and tm were also never rejected. The role of Hoxd12 in the modulation of digit number and development of the flipper was also suggested by the results of branch-site models, in which several sites under positive selection were detected on the branch leading to the cetaceans (see supplementary table S1, Supplementary Material online). Further research is needed to elucidate the functional significance of the sites in Hoxd12 gene function. Analyses of the Hoxd13 gene showed no signature of adaptive change; however, PAML-based analyses did not take into account the observed length variation in the polyalanine tract at the 5#-end coding region among the cetaceans surveyed. Our ancestral reconstructions revealed that the addition of two alanine residues occurred in the ancestral branch of the cetaceans and thus coincided with a gain in digit number and web forming. Abnormally high numbers of alanine residues (.15) in this stretch in humans and mice have been linked to morphological deformities characterized by extra and webbed digits (Muragaki et al. 1996; Johnson et al. 1998) and, more generally, repeat length mutations are known to correlate with a range of human conditions or developmental syndromes in humans (Lavoie et al. 2003; Amiel et al. 2004; Anan et al. 2007), reviewed in Amiel et al. (2004) and Trochet et al. (2007). Moreover, several of these disease-linked polyalanine domains have also been reported from Hox genes (Innis et al. 2004). Recently, Fondon and Garner (2004) proposed that length polymorphisms in amino acid repeat domains might underpin rapid morphological evolution, 620 Wang et al. following their discovery that repeat variation in the coding regions of Alx-4 and Runx-2 were associated with differences in dog limb and skull morphology. Our findings from Hoxd13 in cetaceans appear to provide support for this assertion. As well as changes in digit number, the cetacean manus also shows dramatic interspecific polymorphisms in the number of carpals and phalanges, and even intraspecific polymorphisms in the number of phalanges (Fedak and Hall 2004; Berta et al. 2006; Cooper et al. 2007). It is therefore intriguing that omega estimates (based on free-ratio models) for both genes were seen to vary markedly among the branches of a tree that included cetaceans only, and, on average, were much higher in cetaceans than in other mammals. Sliding windows of omega ratios also showed considerable fluctuations outside of the homeodomain region of both genes. These adjacent domains have not been studied individually; however, because both genes are transcription factors that are typically composed of a DNA-binding domain and an activator domain, the area of positively selected sites detected by branch-site model A and the sliding windows might correspond to the latter domain and thus influence the efficiency of DNA binding (Agalioti et al. 2000; Berg et al. 2002). We thus conclude that the homeodomain is under strong functional constraint in cetaceans and is not involved in the observed variation and adaptive change detected in Hoxd12 and Hoxd13. In contrast to the cetaceans, no evidence of changes in selection pressure in 5#HoxD genes was recorded in the evolution of perissodactyls and artiodactyls, both of which have also undergone reductions in digit number (Vaughan et al. 2000). Yet given that diversification of the limb skeleton can occur via multiple routes (Tanaka et al. 2005; Weatherbee et al. 2006), this lack of parallel signature is perhaps not surprising. More interesting, however, is how these Hox genes, which are so highly conserved across vertebrates, including the ungulates, have been able to undergo such relatively large numbers of changes in the cetaceans. One clue to this might come from the recent finding that forelimbs and hind limbs share a common genetic control path for digit development (Cobb and Duboule 2005). It is thus plausible that the early loss of the hind limbs in cetaceans reduced the developmental constraint placed on 5#HoxD genes. Therefore, because they no longer controlled appendicular developments in both limbs (Thewissen et al. 2006), these genes were able to adapt to meet the morphological demands of different cetacean species during their adaptive radiation. Previous evidence of the role of Hox genes in evolution has nearly all come from studies of expression (Zakany et al. 1997). Patterns of expression have been used to assign digit identity in bird wings (Vargas and Fallon 2005), and are linked to pelvis loss in the pufferfish Takifugu rubripes (Tanaka et al. 2005). Expression studies have also suggested that HoxD and Tbx genes were coopted in the evolution of paired fins (Freitas et al. 2006) and that Hox genes such as those on the HoxA and HoxD clusters were involved in the subsequent evolution of an ancestral fin-like state into tetrapod appendages (Zakany and Duboule 2007). Some studies have examined the molecular evolution of Hox genes involved in limb organization (Kappen 2000). Lynch et al. (2006) found that paralog groups of Hox genes showed signatures of adaptive evolution following duplication of Hox cluster in the early evolution of vertebrates, whereas work by Amores et al. (2004) suggests that gene diversification following Hox gene duplication has contributed to the radiation of the teleosts. By comparison, examples of rapid Hox gene evolution in mammals are uncommon (for an exception, see Wang and Zhang 2007). Our findings, suggesting that limb modification in the mammals has involved the molecular evolution of the Hox genes Hoxd12 and Hoxd13, represent an important addition to this growing body of work and highlights the potential role that Hox genes are likely to have had in more recent evolutionary radiations. Supplementary Material Supplementary tables S1 and S2 and supplementary figure S1 are available at Molecular Biology and Evolution online (http://www.mbe.oxfordjournals.org/). Acknowledgments We are very grateful to Rob Deaville for samples of cetaceans. We also thank Yang Liu for assistance in the laboratory and Annalisa Berta and John Gatesy for helpful comments on an earlier version of the manuscript. This work was funded by a grant awarded to S. Zhang under the Key Construction Program of the National ‘‘985’’ Project and ‘‘211’’ Project, a Royal Society Research Fellowship awarded to S.J.R., and a Darwin grant (14/008) awarded to G.J. and S.Z. Literature Cited Agalioti T, Lomvardas S, Parekh B, Yie J, Maniatis T, Thanos D. 2000. Ordered recruitment of chromatin modifying and general transcription factors to the IFN-beta promoter. Cell. 103:667–678. Amiel J, Trochet D, Clement-Ziza M, Munnich A, Lyonnet S. 2004. Polyalanine expansions in human. Hum Mol Genet. 2:R235–R243. Amores A, Suzuki T, Yan YL, Pomeroy J, Singer A, Amemiya C, Postlethwait JH. 2004. Developmental roles of pufferfish Hox clusters and genome evolution in ray-fin fish. Genome Res. 14:1–10. Anan K, Yoshida N, Kataoka Y, Sato M, Ichise H, Nasu M, Ueda S. 2007. Morphological change caused by loss of the taxon-specific polyalanine tract in Hoxd-13. Mol Biol Evol. 24:281–287. Benke H. 1993. Investigations on the osteology and the functional morphology of the flipper of whales and dolphins (Cetacea). Invest Cetacea. 24:9–252. Berg JM, Tymoczko JL, Stryer L. 2002. The control of gene expression. In: Moran S, Hadler GL, Zimmerman P, editors. Biochemistry, 5th ed. New York: W.H. Freeman and Company. p. 1280–1319. Berta A, Sumich JL, Kovacs KM. 2006. Marine mammals: evolutionary biology. San Diego: Academic Press. Adaptive Evolution of 5#HoxD Genes in Cetaceans 621 Boisserie JR, Lihoreau F, Brunet M. 2005. The position of Hippopotamidae within Cetartiodactyla. Proc Natl Acad Sci USA. 102:1537–1541. Chen CH, Cretekos CJ, Rasweiler JJ, Behringer RR. 2005. Hoxd13 expression in the developing limbs of the short-tailed fruit bat, Carollia perspicillata. Evol Dev. 7:130–141. Cobb J, Duboule D. 2005. Comparative analysis of genes downstream of the Hoxd cluster in developing digits and external genitalia. Development. 132:3055–3067. Cooper LN, Berta A, Dawson SD, Reidenberg JS. 2007. Evolution of hyperphalangy and digit reduction in the cetacean manus. Anat Rec. 290:654–672. Davis AP, Capecchi MR. 1996. A mutational analysis of the 5# HoxD genes: dissection of genetic interactions during limb development in the mouse. Development. 122:1175–1185. Davis MC, Dahn RD, Shubin NH. 2007. An autopodial-like pattern of Hox expression in the fins of a basal actinopterygian fish. Nature. 447:473–476. Demere TA, Mcgowen MR, Berta A, Gatesy J. 2008. Morphological and molecular evidence for a stepwise evolutionary transition from teeth to baleen in mysticete whales. Syst Biol. 57:15–37. Fedak TJ, Hall BK. 2004. Perspectives on hyperphalangy: patterns and processes. J Anat. 204:151–163. Fitzgerald EM. 2006. A bizarre new toothed mysticete (Cetacea) from Australia and the early evolution of baleen whales. Proc Biol Sci. 273:2955–2963. Fondon JW 3rd, Garner HR. 2004. Molecular origins of rapid and continuous morphological evolution. Proc Natl Acad Sci USA. 101:18058–18063. Freitas R, Zhang G, Cohn MJ. 2006. Evidence that mechanisms of fin development evolved in the midline of early vertebrates. Nature. 442:1033–1037. Freitas R, Zhang G, Cohn MJ. 2007. Biphasic Hoxd gene expression in shark paired fins reveals an ancient origin of the distal limb domain. PLoS ONE. 2:e754. Gingerich PD, Haq M, Zalmout IS, Khan IH, Malkani MS. 2001. Origin of whales from early artiodactyls: hands and feet of Eocene Protocetidae from Pakistan. Science. 293:2239–2242. Goldbogen JA, Calambokidis J, Shadwick RE, Oleson EM, McDonald MA, Hildebrand JA. 2006. Kinematics of foraging dives and lunge-feeding in fin whales. J Exp Biol. 209:1231–1244. Golley FB, Whitmoyer TF. 1959. Zygodactyly in the meadow mouse. J Mammal. 40:595–599. Herault Y, Fraudeau N, Zakany J, Duboule D. 1997. Ulnaless (Ul), a regulatory mutation inducing both loss-of-function and gain-of-function of posterior Hoxd genes. Development. 124:3493–3500. Innis JW, Mortlock D, Chen Z, et al. (13 co-authors). 2004. Polyalanine expansion in HOXA13: three new affected families and the molecular consequences in a mouse model. Hum Mol Genet. 13:2841–2851. Johnson KR, Sweet HO, Donahue LR, Ward-Bailey P, Bronson RT, Davisson MT. 1998. A new spontaneous mouse mutation of Hoxd13 with a polyalanine expansion and phenotype similar to human synpolydactyly. Hum Mol Genet. 7:1033–1038. Kappen C. 2000. Analysis of a complete homeobox gene repertoire: implications for the evolution of diversity. Proc Natl Acad Sci USA. 97:4481–4486. Kmita M, Kondo T, Duboule D. 2000. Targeted inversion of a polar silencer within the HoxD complex re-allocates domains of enhancer sharing. Nat Genet. 26:451–454. Knezevic V, De Santo R, Schughart K, Huffstadt U, Chiang C, Mahon KA, Mackem S. 1997. Hoxd-12 differentially affects preaxial and postaxial chondrogenic branches in the limb and regulates Sonic hedgehog in a positive feedback loop. Development. 124:4523–4536. Lavoie H, Debeane F, Trinh QD, Turcotte JF, Corbeil-Girard LP, Dicaire MJ, Saint-Denis A, Page M, Rouleau GA, Brais B. 2003. Polymorphism, shared functions and convergent evolution of genes with sequences coding for polyalanine domains. Hum Mol Genet. 12:2967–2979. Lee AP, Koh EGL, Tay A, Brenner S, Venkatesht B. 2006. Highly conserved syntenic blocks at the vertebrate Hox loci and conserved regulatory elements within and outside Hox gene clusters. Proc Natl Acad Sci USA. 103: 6994–6999. Lemons D, McGinnis W. 2006. Genomic evolution of Hox gene clusters. Science. 313:1918–1922. Litingtung Y, Dahn RD, Li Y, Fallon JF, Chiang C. 2002. Shh and Gli3 are dispensable for limb skeleton formation but regulate digit number and identity. Nature. 418:979–983. Lull RS. 1904. Fossil footprints of the Jura-Trias of north America. Mem Boston Soc Nut Hist. 5:461–557. Lynch VJ, Roth JJ, Wagner GP. 2006. Adaptive evolution of Hox-gene homeodomains after cluster duplications. BMC Evol Biol. 6:86. Maddison WP, Maddison DR. 2007. Mesquite: a modular system for evolutionary analysis. Version 2.01. http://mesquiteproject. org. Muragaki Y, Mundlos S, Upton J, Olsen BR. 1996. Altered growth and branching patterns in synpolydactyly caused by mutations in HOXD13. Science. 272:548–551. Murphy WJ, Eizirik E, O’Brien SJ, et al. (11 co-authors). 2001. Resolution of the early placental mammal radiation using Bayesian phylogenetics. Science. 294:2348–2351. Nei M, Gojobori T. 1986. Simple methods for estimating the numbers of synonymous and nonsynonymous nucleotide substitutions. Mol Biol Evol. 3:418–426. Nei M, Kumar S. 2000. Molecular evolution and phylogenetics. Oxford: Oxford University Press. Nikaido M, Rooney AP, Okada N. 1999. Phylogenetic relationships among cetartiodactyls based on insertions of short and long interpersed elements: hippopotamuses are the closest extant relatives of whales. Proc Natl Acad Sci USA. 96:10261–10266. O#Leary MA, Gatesy J. 2008. Impact of increased character sampling on the phylogeny of Cetartiodactyla (mammalia): combined analysis including fossils. Cladistics. 24: 397–442. Papageorgiou S. 2007. HOX gene expression. New York: Springer Press. Price SA, Bininda-Emonds OR, Gittleman JL. 2005. A complete phylogeny of the whales, dolphins and even-toed hoofed mammals (Cetartiodactyla). Biol Rev Camb Philos Soc. 80:445–473. Pride DT. 2000. SWAAP: a tool for analyzing substitutions and similarity in multiple alignments. Version 1.0.2. http:// www.bacteriamuseum.org/SWAAP/SwaapPage.htm. Rose KD. 1982. Skeleton of Diacodexis, oldest known Artiodactyl. Science. 216:621–623. Sheth R, Bastida MF, Ros M. 2007. Hoxd and Gli3 interactions modulate digit number in the amniote limb. Dev Biol. 310:430–441. Shimamura M, Yasue H, Ohshima K, Abe H, Kato H, Kishiro T, Goto M, Munechika I, Okada N. 1997. Molecular evidence from retroposons that whales form a clade within even-toed ungulates. Nature. 388:666–670. Shubin NH, Marshall CR. 2000. Fossils, genes, and the origin of novelty. Paleobiology. 26:324–340. Sober E. 1991. Reconstructing the past: parsimony, evolution, and inference. Cambridge: MIT Press. 622 Wang et al. Spitz F, Herkenne C, Morris MA, Duboule D. 2005. Inversioninduced disruption of the Hoxd cluster leads to the partition of regulatory landscapes. Nat Genet. 37:889–893. Tabin CJ. 1992. Why we have (only) five fingers per hand: hox genes and the evolution of paired limbs. Development. 116:289–296. Tamura K, Dudley J, Nei M, Kumar S. 2007. MEGA4: Molecular Evolutionary Genetics Analysis (MEGA) software version 4.0. Mol Biol Evol. 24:1596–1599. Tanaka M, Hale LA, Amores A, Yan YL, Cresko WA, Suzuki T, Postlethwait JH. 2005. Developmental genetic basis for the evolution of pelvic fin loss in the pufferfish Takifugu rubripes. Dev Biol. 281:227–239. Tarchini B, Duboule D, Kmita M. 2006. Regulatory constraints in the evolution of the tetrapod limb anterior-posterior polarity. Nature. 443:985–988. Teeling EC, Springer MS, Madsen O, Bates P, O#Brien SJ, Murphy WJ. 2005. A molecular phylogeny for bats illuminates biogeography and the fossil record. Science. 307:580–584. Thewissen JG, Cohn MJ, Stevens LS, Bajpai S, Heyning J, Horton WE Jr. 2006. Developmental basis for hind-limb loss in dolphins and origin of the cetacean bodyplan. Proc Natl Acad Sci USA. 103:8414–8418. Thewissen JG, Cooper LN, Clementz MT, Bajpai S, Tiwari BN. 2007. Whales originated from aquatic artiodactyls in the Eocene epoch of India. Nature. 450:1190–1194. Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG. 1997. The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25:4876–4882. Tickle C. 2006. Making digit patterns in the vertebrate limb. Nat Rev Mol Cell Biol. 7:45–53. Trochet D, de Pontual L, Keren B, Munnich A, Vekemans M, Lyonnet S, Amiel J. 2007. Polyalanine expansions might not result from unequal crossing-over. Hum Mutat. 28: 1043–1044. Vargas AO, Fallon JF. 2005. Birds have dinosaur wings: the molecular evidence. J Exp Zool Part B. 304B:86–90. Vaughan TA, Ryan JM, Czaplewski NJ. 2000. Mammalogy. Philadelphia: Saunders College Publishing. Wang X, Zhang J. 2007. Rapid evolution of primate ESX1, an X-linked placenta- and testis-expressed homeobox gene. Hum Mol Genet. 16:2053–2060. Weatherbee SD, Behringer RR, Rasweiler JJt Niswander LA. 2006. Interdigital webbing retention in bat wings illustrates genetic changes underlying amniote limb diversification. Proc Natl Acad Sci USA. 103:15103–15107. Woodward BL, Winn JP, Fish FE. 2006. Morphological specializations of baleen whales associated with hydrodynamic performance and ecological niche. J Morphol. 267:1284–1294. Yang Z. 1998. Likelihood ratio tests for detecting positive selection and application to primate lysozyme evolution. Mol Biol Evol. 15:568–573. Yang Z. 2006. Computational molecular evolution. London: Oxford University Press. Yang Z. 2007. PAML 4: phylogenetic analysis by maximum likelihood. Mol Biol Evol. 24:1586–1591. Yang Z, Wong WS, Nielsen R. 2005. Bayes empirical bayes inference of amino acid sites under positive selection. Mol Biol Evol. 22:1107–1118. Zakany J, Duboule D. 2007. The role of Hox genes during vertebrate limb development. Curr Opin Genet Dev. 17:359–366. Zakany J, Fromental-Ramain C, Warot X, Duboule D. 1997. Regulation of number and size of digits by posterior Hox genes: a dose-dependent mechanism with potential evolutionary implications. Proc Natl Acad Sci USA. 94:13695–13700. Zakany J, Kmita M, Duboule D. 2004. A dual role for Hox genes in limb anterior–posterior asymmetry. Science. 304: 1669–1672. Zhang J, Nielsen R, Yang Z. 2005. Evaluation of an improved branch-site likelihood method for detecting positive selection at the molecular level. Mol Biol Evol. 22:2472–2479. Zhao X, Sun M, Zhao J, et al. (15 co-authors). 2007. Mutations in HOXD13 underlie syndactyly type V and a novel brachydactyly–syndactyly syndrome. Am J Hum Genet. 80:361–371. Claudia Kappen, Associate Editor Accepted December 6, 2008