David M. IRWIN

?

Genomic organization and evolution of ruminant lysozymegenes

David M. IRWIN1,2,*

1Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Canada2Banting and Best Diabetes Centre, University of Toronto, Toronto, Canada

Ruminant stomach lysozyme is a long established model of adaptive gene evolution. Evolution of stomach lysozyme function required changes in the site of expression of the lysozymegene and changes in the enzymatic properties of the enzyme. In ruminant mammals, these changes were associated with a change in the size of the lysozymegene family. The recent release of near complete genome sequences from several ruminant species allows a more complete examination of the evolution and diversification of the lysozymegene family. Here we characterize the size of the lysozymegene family in extant ruminants and demonstrate that their pecoran ruminant ancestor had a family of at least 10 lysozymegenes, which included at least two pseudogenes. Evolutionary analysis of the ruminant lysozymegene sequences demonstrate that each of the four exons of the lysozymegene has a unique evolutionary history, indicating that they participated independently in concerted evolution. These analyses also show that episodic changes in the evolutionary constraints on the protein sequences occurred, with lysozymegenes expressed in the abomasum of the stomach of extant ruminant species showing the greatest levels of selective constraints.

Lysozyme; Ruminants; Gene family; Gene duplication; Concerted evolution; Mosaic evolution

INTRODUCTION

Ruminant mammals such as cow, sheep, and deer, rely on foregut fermentation to extract nutrients from their diet of plant material (Clauss et al, 2010; Janis, 1976; Mackie, 2002;Stevens & Hume, 1998). Foregut fermentation, by bacteria and microbes, produces short chain fatty acids that are absorbed through the stomach wall and provide energy for the ruminant animals; however, the microbial population responsible for this fermentation incorporates many of the other nutrients, such as nitrogen based compounds, into their own growing populations (Mackie, 2002; Stevens & Hume, 1998). To extract these essential nutrients from the microbial population, ruminant animals must break open these bacterial and microbial cells, to release their contents, to allow the stomach digestive enzymes in the abomasum to extract nutrients from their contents (Stevens & Hume, 1998). Since bacterial cells are typic-ally resistant to mammalian digestive enzymes, ruminant species have recruited the anti-bacterial enzyme, lysozy-me, to break open these cells (Callewaert & Michiels, 2010; Dobson et al, 1984; Irwin et al, 1992; Mackie, 2002;Prager & Jollès, 1996). Recruitment of lysozymeas a digestive enzyme has occurred at least twice within mammals, on the lineages leading to the ruminant artiodactyls and the leaf-eating monkeys (Dobson et al, 1984; Stewart et al, 1987; Ste-wart & Wilson, 1987), with a similar recruitment of a cal-c---ium-binding lysozyme occurring in the hoatzin, a leaf-eating bird (Kornegay et al, 1994; Kornegay, 1996).

Recruitment of lysozymeto become a digestive enzyme required changes both in the site of expression of the gene encoding this enzyme and in the amino acid sequence of the enzyme to allow function in the acidic stomach (Dobson et al, 1984; Irwin et al, 1992; Irwin, 1996; Prager, 1996). The major site of expression of lysozyme in mammals is macrophages, but it also secreted into some body fluids (such as tears), where it participates in host defense against bacterial infection (Callewaert & Michiels, 2010; Prager & Jollès, 1996; Short et al, 1996). The molecular basis for the recrui-tment of expression, at high levels, of lysozymein stomach cells is unknown. Typical mammalian lysozymeenzymes function in an environment at a neutral pH, and one that is free of digestive enzymes (Callewaert & Michiels, 2010; Prager & Jollès, 1996; Prager, 1996). Lysozymefunction in the abomasum of the stomach of ruminant animals, to digest bacterial cell walls, required adapting the lysozymeprotein sequence to function at an acidic pH and becoming resistant to the actions of stomach digestive enzymes and acids found in the abomasum (Dobson et al, 1984; Jollès et al, 1989; Prager, 1996). A number of convergent amino acid changes were seen between the lysozymesequences that have adapted for function in the stomachs of the langur, a leaf-eating monkey, and ruminants, have been identified and presumed to account for much of the functional adaptation (Stewart & Wilson, 1987; Stewart et al, 1987; Swanson et al, 1991; Prager, 1996). Some of these adaptive changes include replacement of lysine residues with arginine, which removes potential cleavage sites for digestive enzymes found in the stomach, and the loss of an aspartate-proline dipeptide, which is an acid-labile peptide bond (Jollès et al, 1989; Prager, 1996; Stewart & Wilson, 1987; Stewart et al, 1987; Swanson et al, 1991). These putative adaptive amino acid replacements are inferred to occur early in ruminant evolution, and thus may parallel the origin and evolution of the ruminant lifestyle (Irwin et al, 1992; Irwin, 1996).

Recruitment of lysozymeto a digestive role in ruminants is associated with an expansion of the size of the lysozymegene family (Jiang et al, 2014; Irwin & Wilson, 1989; Irwin et al, 1989, 1992). Most mammals have only one or a few lysozymegenes, while ruminant species have 10 or more (Callewaert & Michiels, 2010;Irwin & Wilson, 1989; Irwin et al, 1989, 1996; Prager & Jollès, 1996; Irwin et al, 2011; Jiang et al, 2014). The lysozymegene family of the cow has been better characterized than those of most other ruminants, where it was found that only some of the genes are expressed in the abomasum, while others retain more ancestral type of roles (Irwin & Wilson, 1989; Irwin et al, 1993; Irwin, 2004). Similar observations have been made for the lysozymegenes of sheep (Jiang et al, 2014). Several lysozymeproteins, and their cDNAs, have been characterized from the abomasums of the cow, sheep and deer (Dobson et al, 1984; Jollès et al, 1989; Irwin & Wilson, 1989, 1990). Intriguingly, phylogenetic analysis of the coding and 3' untranslated portions of the lysozymecDNA sequences yielded different trees, with the coding sequences implying duplications of the genes on each species lineage and the 3 untranslated region indicating more ancient duplications before the diverg-ence of these species (Irwin & Wilson, 1990). Selection at the protein level (e.g., lineage-specific adaptation of the protein sequences) does not explain the differences in the phylo-genies for the two regions, as synonymous difference (those that do not change the coding potential) also yield the same conclusions. It was concluded that the differences in the phylogenies was due to concerted evolution, mediated by gene conversion, acting on the coding sequences, while the 3' untranslated regions only experienced divergent evolution (Irwin & Wilson, 1990; Irwin et al, 1992; Irwin, 1996; Wen & Irwin, 1999; Yu & Irwin, 1996). Characterization of the genomic sequences of lysozyme genes expressed in the abomasum from the cow and sheep suggested that the concerted evolution was limited to only the coding exons, and did not involve the intronic sequences separating these exons (Irwin et al, 1993; Wen & Irwin, 1999). An analysis of larger number of lysozyme mRNA sequences, including genes that are not expressed in the stomach, suggested that some of the genes expressed in non-stomach tissues might have also experienced concerted evolution (Irwin, 1995, 2004; Takeuchi et al, 1993).

The previous analyses were largely limited to lysozymegenes expressed in ruminant species. With the recent completion of draft genomic sequences from several ruminant species (including, cow, yak, zebu, goat, Tibetan antelope, and sheep: Canavez et al, 2012; Dong et al, 2013; Ge et al, 2013; Jiang et al, 2014; Qiu et al, 2012; Zimin et al, 2009) it now possible to more completely characterize the complete complement of lysozymegenes (including genenes that are not expressed) in the genomes of these species and examine the molecular evolution of these genes. Here we describe the lysozymegene complements of the cow and several other ruminant species. The lysozymegene cluster has largely been maintained within true ruminant (Infraorder Pecora) species. Analysis of these sequences shows that the ancestor of cow, sheep, and goats had 10 lysozymegenes, several of which were pseudogenes that were retained by diverse species. The exons of the lysozymegenes have differing evolutionary histories, suggesting that concerted evolution acted independently on each exon.

MATERIALS AND METHODS

Database searches

Previous searches of mammalian genomes indicated that the cow genome had about 12 lysozymegenes located in a cluster on cow chromosome 5, many of which were incompletely annotated in the Ensembl assembly (Irwin et al, 2011). To better characterize the lysozymegene cluster in the cow () genome we used the Blast algorithm (Altschul et al, 1990) to search the UMD 3.1 cow genome assembly (from Ensembl release 75 in June 2014; http://www.ensembl.org/index.html) with known and predicted cow lysozymecDNA and protein sequences. Lysozymegenes from the sheep (; Oar_v3.1), pig (; Sscrofa10.2), bottlenose dolphin (; Turtru1), dog (; CanFam3.1), panda (; AilMel1), horse (; EquCab2), and rhinoceros (; CerSimSim1 - preEnsembl) genomes from the Ensembl database were characterized by the approaches described above. A similar search strategy was used to identify lysozymegenes in the yak (), zebu (), water buffalo (), Tibetan antelope (chiru;), goat (), aplaca (), mi-nke whale (), kill-er whale (), Yangtze River dolphin (), and sperm whale () genomes from the NCBI Genomes (chromosome), Whole-genome shotgun contigs (wgs), and Nucleotide collection (nr/nt) databases (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

Genomic alignments and assignment of orthology

Genomic sequences encompassing lysozymegenes were downloaded from the Ensembl and NCBI databases. Intron-exon boundaries of the 4 exons and the 5' and 3' flanking sequences of the new lysozymegenes were annotated based on genomic alignments of genes using MultiPipMaker (Schwartz et al, 2000, 2003), using previously characterized artiodactyl lysozymegenes (Irwin et al, 1993, 1996; Irwin, 1995; Yu & Irwin, 1996; Wen & Irwin, 1999) as guides. Gene neighborhood organization was assessed as previously described for lysozymegenes (Irwin et al, 2011) with the flanking4 and6 genes identified using Blast. Ruminant lysozymegenes were named based on orthology (based on phylogeny, see below) and genomic location. Genes present in the common ancestor of sheep and cow were numbered (1-10), while lineage-specific duplic-ates have a letter (a-c) that follows the gene number. The alpaca lysozymegenes were numbered arbitrarily, thus their numbers do not indicate orthology with the rum-inant genes. All other species examined here have a single copy lysozymegene.

Phylogenetic analysis

Predicted protein coding sequences for lysozymecDNA sequences, extracted from the genomic alignments, were aligned with Muscle (Edgar, 2004) as implemented in Mega6.06 (Tamura et al, 2013). Alignments were edited manually to insert gaps to maintain open reading frames (due to the presence of frame shifting insertions in some pseudogenes). Phylogenetic trees were constr-ucted by the maximum likelihood, neighbor-joining and parsimony methods using Mega6.06 (Tamara et al, 2013). Alternative phylogenetic hypothesis, derived from the phylogenies of the different exons, were tested using Tree-puzzle (Strimmer & von Haeseler, 1996) as imple-m-ented on the Mobyle@Pasteur web site (http://mobyle.pasteur.fr/cgi-bin/portal.py?#welcome; Néron et al, 2009).

RESULTS AND DISCUSSION

Number and organization of lysozyme c genes in the cow genome

Analyses of genomic Southern blots had concluded that there were about 10 lysozymegenes in the cow genome (Irwin & Wilson, 1989; Irwin et al, 1989), with many of these genes clustered on chromosome 5 (Galla-gher et al, 1993). A recent (2011) search of the Btau4.0 (2ndrelease, assembled 2007) of the cow genome sequence assembly identified 12 lysozymegenes on chromosome 5 of the cow genome (Irwin et al, 2011). The genes identified in this search account for all of the previously characterized cow lysozymecDNA and protein sequences (Irwin et al, 2011). To better charac-terize the cow lysozymegene cluster, we searched the most current version (UMD3.1 – 3rdrelease) of the cow genome assembly (assembled 2009; Zimin et al, 2009) with Blast using the previously characterized lysozymecDNA and protein sequences. Our new searches identified a total of 14 lysosomegenes, 11 of which were annotated by Ensembl as genes (Table 1 and Figures 1, 2, and S1, supporting information at http://www.zoores.ac.cn/). The difference in the number of intact genes identified by the searches of the two different genome assemblies (12 in Btau4.0 and 11 in UMD3.1) is due to the earlier Btau4.0 assembly containing two copies of the tracheal lysozymegene (andin Irwin et al, 2011) while the most current assembly UMD 3.1 contains only single copy of this gene (here named2b).

In addition to the 11 annotated genes, each of which is composed of 4 exons consistent with the structure of a typical mammalian lysozymegene (Irwin et al, 1996; Callewaert & Michiels, 2010), Blast hits were found to map to additional locations that were distant from the annotated genes. Examination of these Blast hits suggested that they belong to three partial genes, which had not previously been annotated, with each being composed of only two, not four, exons (Table 1 and Figures 1, 2, and S1). The newly identified partial3a and3c genes contain exons 1 and 2 and exons 3 and 4, respectively, but are separated from each other by the3b gene that contains all 4 coding exons (Table 1 and Figures 1 and 2). No sequences similar to the missing exons were found near the3a and3c genes. The third partial gene,9, contains only exons 1 and 4,with the sequences between these exons showing no similarity to exons 2 or 3 of other lysozymegenes. Most of the lysozymegenes have the same orientation (annotated as the minus strand), but 4 of the 14 are on the opposite strand, indicating that the origin of this gene family is not just a simple series of tandem gene duplications. The cow genome, therefore, was found to contain 14 identifiable lysozymegenes (Table 1 and Figures 1,2, and S1).

Table 1 Locations of cow lysozyme c genes

a– Not annotated as a gene by Ensembl.

b– No ESTs with greater than 95% sequence identity identified in the NCBI EST database.

Figure 1 Organization of lysozymegenes in the cow genome

Schematic of the arrangement of lysozymegenes. And their neighbors, in the cow genome. Vertical lines represent exons, with splicing indicated by the lines joining the exons. Gene names are indicated above (plus strand) or below (minus strand) indicating strand with coding potential. Sizes of genes and distances are proportional. The genes are located between 44.3 and 44.8 Mb on chromosome 5.

Figure 2 Amino acid sequences predicted by cow lysozymegenes

Sequences of predicted lysozymeproteins from the cow genome are shown in single letter code, with differences from the Cow Lyz1 sequence shown and identities indicated by dots (.). Sequence is numbered avove the sequences from the N-terminus of the Lyz1 sequence, with the signal peptide numbered backwards and in italics. Dashes (-) indicate gaps introduced to maximize alignment. Question marks (?) indicate incomplete codons due to missing sequence. Residues involved in disulfide bridging ($) and active site residues (#) are marked below the sequences. Residues marked in red are likely damaging pseudogenes and disrupt initiation, disrupt disulfide bridging or introduce stop codons. Asterisks (*) indicate inframe stop codons. Xs refer to codons that have less than 3 bases and thus cause frame shifts. The initiation codon of Lyz8 is not methionine (M).

Pairwise sequence comparisons revealed that the DNA sequence identities of the coding sequences among the 14 genes ranged from 74.8% to 97.5%, with most pairs showing 80%-90% identity (Table 2). The similarity between3a and3c could not be measured, as there is no overlap between these two genes (3a has exons 3 and 4, while3c has exons 1 and 2, see Table 1). These two partial genes were most similar to the3b gene, showing greater than 96% identify in the coding sequence (Table 2), raising the possibility that they are recent gene duplicates. The3a,3b, and3c genes are also adjacent to each other in the genome, suggesting that the3a and3b gene were generated by partial tandem duplications of different parts of the3b gene (Figure 1). The partial gene9 did not show particularly strong similarity to any other specific cow lysozymegene (Table 2), suggesting that it is not a product of a very recent segmental duplication event (Liu et al, 2009; Seo et al, 2013). Among the intact lysozyme genes, the coding sequence of the genes encoding the lysozymes expressed in the abomasum (Irwin & Wilson, 1989; Irwin et al, 1993)5/6/7 share about 97% identity and the2a/2b/2c genes, which includes the tracheal lysozyme gene (Takeuchi et al, 1993), share about 96% identity (Table 2). The high level of identity shared by these sets of genes suggests that these triplets are products of recent segmental duplic-ation / gene duplication events (Liu et al, 2009; Seo et al, 2013).5/6/7 are adjacent to each other and are in the same orientation (Table 1; Figure 1), thus could be generated by a simple series of tandem gene duplication events. A more complicated duplication history is needed to explain the diversification of the2a/2b/2c genes. While2a and2b are in tandem, several other lysozymegenes (3a,3b, and3c) are located between the2a/2b gene pair and the2c gene (Table 1 and Figure 1).

Table 2 Pairwise percent DNA sequence identity between cow lysozyme c coding sequences

a– These two genes do not overlap.

Since the three partial genes3a,3b, and9 do not contain all four coding exons; they cannot predict intact open reading frames. In addition to the missing exon sequence, all three of these genes also contain in frame stop codons or frameshifts that also would prevent translation (Figure 2). Among the lysozymegenes possessing all four exons, two,3b and8, fail to predict intact open reading frames (Figure 2).3b, which was previously called lysozymeyNS4 (Irwin, 1995) contains a frameshift, which is shared with3a, which prevents translation of the reading frame, while8 contains both in frame stop codons and a replacement at the initiating codon (Figure 2). Thus of the 14 lysozymegenes found in the cow genome, only 9 potentially encode functional lysozymeproteins. To further investigate the functional potential of these lysozymegenes we searched for evidence of expre-ssion for all 14 cow lysozymegenes in the NCBI expre-ssed sequence tag (EST) database. ESTs were found for only 8 of the 9 intact genes, and for none of the 5 pseudogenes (Table 1). While2a has an intact open reading frame (Figure 2), no ESTs highly similar (>98% identity) to it were found in the NCBI database (Table 1), raising the possibility that this gene is not expressed.

Many of the cow lysozymegene annotations in the Ensembl database do not include 5' and/or 3' untranslated sequences. Since previous work had shown that the 5' and 3' untranslated sequences of known lysozymegenes from diverse mammalian species have considerable sequence similarity (Irwin & Wilson, 1989; Irwin, 1995, 2004), we used this similarity to predict the extent of these regions for each gene (see Figure S1) from alignments generated by MultiPipMaker (Schwartz et al, 2000, 2003). Complete 5' untranslated regions could be predicted for all of the lysozymegenes that had exon 1, however the full 3' untranslated regions could not be predicted for all exon 4 sequences, as the 3' end of the 3' untranslated region could not be found for the cow3a and3b genes (see Figure S1). This observation is consistent with an earlier failure to identify homologous sequences for the entire 3' untra-nslated region of the cowyNS4 (3b) gene (Irwin, 1995, 2004).

Lysozyme c genes in other ruminant genomes

To better characterize the evolutionary history of the ruminant lysozymegenes, we identified lysozymegenes in the genomes of other ruminant species and their close relatives (Table 3 and Figures 3 and S1). As expected, from previous work (Callewaert & Michiels, 2010; Irwin et al, 1989, 1996, 2011; Prager & Jollès, 1996), only a single lysozymegene was found in the genomes of carnivores (dog,; and panda,) and perrisodactyls (horse,; and rhinoceros,) (Table 3 and Figures 3 and S1). The single lysozymegene in the outgroup species is located between the4 and6 genes (Figure 3), as found in most other mammalian species (Irwin et al, 2011). This ancestral mammalian genomic arrangement has been retained in the cow, with the amplification of the lysozymegenes occurring between the4 and6 genes (Irwin et al, 2011) (Figures 1 and 3). The tylopod lineage (e.g., camels and alpacas) represents one branch of the earliest divergence within artiodactyls (Morgan et al, 2013; Romiguier et al, 2013), with these species being pseudoruminants with a simpler multi-chambered stomach than the true ruminants (Clauss et al, 2010; Janis, 1976; Mackie, 2002). Searches of the alpaca () genome in the Ensembl database identified three genomic sequences encoding partial lysozymegene sequences, indicating that multiple lysozymegenes exist in this genome (results not shown). Searches of the NCBI Genomes (chromosomes) database identified an updated larger genomic contig that predicted 4 complete lysozymegenes (and included all of the gene sequences found in the Ensembl alpaca genome assembly) at one end of a contig sequence (Table 3 and Figure 3). The4 gene was found to be adjacent to one side of the lysozymegene cluster, however no genes were found on the other side of the lysozymegene cluster in this genomic contig (Figure 3). The presence of the4 gene adjacent to the alpaca lysozymegenes suggests that a similar genomic neighborhood exists in alpaca, but since the lysozymegenes were at one end of the genomic contig it is possible that additional unsequenced lysozymegenes may exist in the alpaca genome. The pig () is a representative of the family Suidea, which is the next diverging lineage within artiodactyls (Morgan et al, 2013; Romiguier et al, 2013). As expected, and previously reported (Irwin et al, 1989; Yu & Irwin, 1996), only a single lysozyme gene is found in this species (Table 3, Figures 3 and S1). As previously reported (Irwin et al, 2011), the genomic neighborhood surrounding the pig lysozyme gene differs from that of other mammals, raising the possibility that this genomic area has experienced recombination (Figure 3). Cetaceans (e.g., whales and dolphins) fall within artiodactyls, thus yielding cetartiodactyla (Morgan et al, 2013; Romiguier et al, 2013). A single lysozymegene was identified in all five cetacean (bottlenose dolphin,; minke whale,; killer whale,; Yangtze river dolphin,; and sperm whale,) genomes (Table 3 and Figure S1), which is found in genomic location cons-istent with the ancestral genomic organiza-tion (Figure 3).

Table 3 Locations of lysozyme c genes in diverse artiodactyls and relatives

Continued

GeneChromosome / scaffoldStrandBasesEnsembl gene ID / NCBI accessionMissing exons Lyz4NW_005100667Plus6 670 972-6 67 2304NAa1, 2b Lyz5NW_005100667Minus6 689 705-6 694 702XM_005680192 Lyz6NW_005100667Minus6 715 083-6 719 993NM_001287566 Lyz7NW_005100667Minus6 741 973-6 747 565NAa Lyz8NW_005100667Minus6 761 023-6 765 243XM_005680235 Lyz9NW_005100667Minus6 783 196-6 785 752NAa2, 3 Lyz10NW_005100667Minus6 834 500-6 839 640NM_001285711 Sheep (Ovis aries) Lyz13Minus150 165 176-150 170 352ENSOARG00000020393 Lyz23Plus150 225 205-150 229 630ENSOARG00000020417 Lyz3JH921983.1Minus4 380-1 032ENSOARG00000000543 Lyz43Plus150 266 228-150 270 937ENSOARG00000020429 Lyz53Minus150 288 480-150 293 529ENSOARG00000020393 Lyz63Minus150 313 875-150 318 810ENSOARG00000020439 Lyz73Minus150 342 914-150 348 498NAa Lyz83Minus150 362 122-150 366 351ENSOARG00000020476 Lyz93Minus150 385 062-150 387 578NAa2, 3 Lyz103Minus150 434 372-150 439 510ENSOARG00000020515 Pig (Sus scrofa) Lyz5Minus36 179 949-36 185 488ENSSSCG000000004925 Alpaca (Vicugna pacos) Lyz1NT_167289.2Minus1 670 090-1 675 333NAa Lyz2NT_167289.2Minus1 707 719-1 713 452NAa Lyz3NT_167289.2Plus1 722 393-1 728 285NAa Lyz4NT_167289.2Plus1 760 703-1 766 608NAa Bottlenose dolphin (Tursiops truncates) Lyzscaffold_114746Plus182 136-187 936ENSTTRG00000013948 Minke whale (Balaenopteraacutorostrata scammoni) LyzNW_006733011Minus27 704 269-27 709 850XM_007195043 Killer whale (Orcinus orca) LyzNW_004438568Plus1 319 177-1 324 490XM_004281877 Yangtze River dolphin (Lipotes vexillifer) LyzNW_006790307Minus1 455 813-1 461 420XM_007463554 Sperm whale (Physeter catodon) LyzNW_006716048Minus6 880-11 985XM_007118874 Dog (Canis lupus familiaris) Lyz10Plus11 346 500-11 350 639ENSCAFG00000000426 Panda (Ailuropoda melanoleuca) LyzGL192893.1Plus308 956-313 372ENSAMEG00000011820 Horse (Equus caballus) Lyz6Plus84 276 158-84 280 173ENSECAG00000018113 Rhinoceros (Ceratotherium simum simum) LyzJH767750.1Plus25 463 742-25 467 903ENSP00000261267_1

a– Not annotated as a gene in Ensembl or NCBI.

b– Possibly missing due to incomplete gene.

Figure 3 Organization of lysozymegenes in diverse Artiodactyls and relatives

Schematic of the genomic arrangement of lysozymegenes, and their neighbors, derived from genomic sequences in the Ensembl and NCBI databases. Sizes of genes, and distances between genes are not to scale. Genes shown above the lines are encoded by the plus strand, while those below are on the minus strand. Genomic sequences are listed in Table 3.

Pecoran artiodactyls (cow, sheep, deer, and relatives) are true ruminants with a stomach composed of four chambers (Clauss et al, 2010; Janis, 1976; Mackie, 2002). In addition to the sheep () genome (Jiang et al, 2014), which is available from Ensembl, genome sequences of 5 other pecoran ruminant species (yak,(Qiu et al, 2012); zebu,(Canavez et al, 2012); water buffalo,; Tibetan antelope (chiru) (Ge et al, 2013),; and goat,(Dong et al, 2013)) are available in the NCBI database. The genomes of all pecoran ruminant species contained multiple lysozymegenes (Table 3 and Figures 3 and S1), in accord with previous results (Irwin & Wilson, 1989; Irwin et al, 1989, 2011). For most pecoran species, lysozymegenes could be mapped to large genomic contigs, or chromosomes, that show organizations similar to that seen in the cow (Table 3 and Figure 3). In the sheep, one gene (3) was not mapped to chromosome 3, but instead to an unm-apped contig (Table 3). Since the goat genes all map to one contig (Table 3) it is possible that the sheep3 gene has been misplaced (Figure 3), although move-ment to a new location through recombination cannot be excluded. The yak lysozymegenes map to two contigs, with one containing a large gap that corresponds to the location where one or more missing lysozymegenes might exist (Table 3 and Figure 3). The lysozymegenes in both the Tibetan antelope and water buffalo map to two genomic contigs that might be adjacent in their genomes (Table 3 and Figure 3). Lysozymegenes in the zebu are each on separate contigs, but could be arranged as seen in the cow and other pecoran species (Table 3 and Figure 3).

Mosaic evolutionary histories for exons of cow lysozyme c genes

To examine the evolutionary history of the cow lysozymegenes, a phylogeny of the sequences was established. Phylogenetic trees were constructed for each exon of the lysozymegenes (Figure 4) as previous analyses suggested that they might have experienced different histories (Irwin, 2004; Irwin & Wilson, 1990; Irwin et al, 1993, 1996; Wen & Irwin, 1999). As shown in Figure 4, different phylogenies were identified for each exons, with similar trees found if different outgroup species were used or if phylogenies were constructed using distance or parsimony methods or if only synony-mous substitutions were used (results not shown). Some consistent phylogenetic patterns were observed across all exons, such as the clustering of the2a,2b, and2c genes and3a or3c being closest to3b (Figure 4). In contrast, the placement of some genes differed greatly between exons, such as the placement of1 or4 (Figure 4). To test whether there were statistically significant differences between the tree topologies estimated by each exon, we used Tree-puzzle (Strimmer & von Haeseler, 1996) to compare the four separate exon tree topologies with data for each exon. Despite the short lengths of some exons, at least two of the three alternative topologies could be excluded by all three of the KH, SH, and ELW statistical tests used by Tree-puzzle, with all three being excluded by at least one of the tests (Table 4). We cannot exclude the possibility that exons 2 and 3 share an identical evolutionary history, as these trees were not excluded by all three of the statistical tests, but exons 1 and 4 have evolutionary histories that are incompatible with each other and with exons 2 and 3 indicating that at least three different histories are represented by these four exons (Table 4). The differences in the topologies are unlikely to be due to convergent evolution acting on the lysozymeprotein sequences as the differences in the topologies were also seen when only synonymous differences were examined (results not shown).

Table 4 Phylogenies predicted from different cow lysozyme c gene exons are significantly different

a– Probability of observing the tree, given the data, from the statistical one sided KH test based on pairwise SH tests (KH), the Shimodaira-Hasegawa test (SH), and the expected likelihood weight test (ELW) from Tree-puzzle (Strimmer & von Haesler, 1996).

b– Probability that the data is compatible with the tree is less than 0.05.

Figure 4 Phylogeny of cow lysozymegenes derive from sequences of (A) exon 1, (B) exon 2, (C) exon 3, and (D) exon 4

Phylogenies for each of the 4 exons of the lysozymegenes were estimated using maximum likelihood, as implemented in Mega6.06 (Tamura et al, 2013), using the Kimura 2-paramater model with a gamma distribution, which was the best fitting model for the sequence data. Similar results were obtained with the neighbor-joining method or parsimony, or the use of different outgroups. Phylogenies were generated from 152, 156, 74, and 306 aligned bases present in all sequences for exons 1, 2, 3, and 4, respectively. The presented phylogenies were bootstrapped 500 times.

These results are in agreement with previous conclusions that lysozymegenes expressed in the abomasum of ruminants have experienced mosaic evolution due to gene conversion occurring between the coding exons (Irwin, 2004; Irwin & Wilson, 1990; Irwin et al, 1993, 1996; Wen & Irwin, 1999), and suggested that the 3' untranslated (exon 4) sequences likely best reflect the evolutionary history of the divergent genes, as this sequence appears to have experienced the fewest number of concerted evolution events.

Origin and evolutionary history of lysozyme c genes in ruminant genomes

To better examine the evolution of the duplicated lysozymegenes in ruminant species, a phylogenetic tree was established for the lysozymesequences from the diverse ruminants (e.g., cow, sheep, and Tibetan antelope) and their close relatives (e.g., pig, cetaceans, and carnivores) (Figure 5). Exon 4 sequences were chos-en to construct this phylogeny as they likely best reflect the divergence of the genes, and have experienced lower levels of concerted evolution (see above). The phylogeny shown in Figure 5 was derived by maximum likelihood, and similar phylogenies were generated when neighbor-joining or parsimony was used (results not shown). The exon 4 phylogeny shown in Figure 5 of the lysozymegenes yield strong evidence for the orthology of 8 of 10 types of lysozymegenes found in ruminants (Figure 5).3,4,5,6,7,8,9, and10 orthology groups each have high (88%-100%) bootstrap support, with the species relationships within each group in general accord with the accepted species relationships (Figure 5). This observation implies that these 8 genes existed in the common ancestor of pecoran ruminants. The phylogenetic analysis did not resolve1 or2 genes as monophyletic groups, but instead suggested some intermixing of these genes (Figure 5).1 and2 sequences from species of tribe Bovine (cow, yak, zebu, and water buffalo) formed a moderately supported monophyletic group that had a primary divergence between the1 and2 sequences. The tribe Bovini1 and2 sequences were then grouped with1 sequences from the other pecoran ruminants (Tibetan antelope, goat and sheep), with the2 sequences from these same species being the outgoup to all of the1 and2 sequences. While it is possible that this distribution could be explained by an ancestor having four genes, and pairs of genes being lost in each species, an alternative explanation is that the pecoran ancestor possessed two genes, and that a concerted evolution event transferred sequences from the tribe Bovini1 exon 4 sequence to the tribe Bovini2 gene, resulting in the grouping of these sequences. Support for the monophyly of the1 and2 genes was found from phylogenies of exon 2 and exon 3 sequences (results not shown). These results suggest that the ancestor of pecoran ruminants possessed 10 lysozymegenes.

While the ancestor of modern pecoran ruminants may have had 10 lysozymegenes, several extant species have a higher number of genes, such as cow with 14 genes and the zebu with 12 genes (Tables 1 and 3). The increased numbers of lysozymegenes in some ruminant species appear to be due to lineage-specific gene duplications. The phylogeny presented in Figure 5 implies lineage-specific duplications in three genes,2,3 and7, all of which occurred in species (cow and zebu) of the genus. Both cow and zebu have three2 genes (2a,2b, and2c) (Tables 1 and 3). Only a single2 gene was found in the yak, however a gap in the genome assembly was found at this location (Table 3), thus it is possible that additional2 genes exist in this genome. Better assembly of thegenome sequences are needed to determine whether the triplic-ated genes have a single origin, or represent parallel duplication, a conclusion that does have some support from the phylogenetic analysis (Figure 5). Duplicated3 genes were only found in the cow, although the lack of this gene in the yak, potentially due to a gap in the assembly, and the poor assembly of the zebu genome do not rule out the possibility that multiple3 genes exist in these species (Tables 1 and 3). The duplications of the2 and3 genes inlikely represent products of segmental duplications (Liu et al, 2009; Seo et al, 2013). It is possible that segmental duplications may also exist in other pecoran ruminant species, but were collapsed as single genes during the genome sequence assembly process, and thus the increased numbers seen in the genussimple reflect the better cow genome assembly.

The distribution of the numbers of lysozymegenes in the genomes of ruminant species and their close relatives is consistent with an amplification of the lysozymegene on the lineage leading to true ruminants as previously proposed (Irwin & Wilson, 1989; Irwin et al, 1989, 1992, 2011; Yu & Irwin, 1996). In contrast, our current phylogenetic analysis of the 3' untranslated regions of lysozymegenes suggests that the amplif-ication of these genes was initiated very early in the artiodactyl lineage, before the divergence of the rumi-nants and tylopod (e.g., alpaca) lineages, and implying that the pig and cetaceans have lost genes, however these early divergences are very poorly supported (Figure 5). Indeed, phylogenetic analysis of exon 1, exon 2, or exon 3 sequences by themselves yielded differing conclusions concerning these earliest duplications, although again, none of these analysis yielded strong conclusions (results not shown). Analysis of larger amounts of genomic sequences (e.g., intronic and flanking sequence) poten-tially could resolve the order of the earliest divergences of the paralogous lysozymegenes and cetartiodactyl species. While the alpaca has multiple lysozymegenes (Table 3), our phylogenetic analysis suggests that they originated through a parallel series of lineage-specific independent duplications.

Rates of evolution in ruminant lysozyme c genes

Duplication of the lysozymegene on the ruminant lineage has allowed the specialization of gene expression in distinct tissues, such as different chambers of the stomach, and thus evolution of novel gene function (Callewaert & Michiels, 2010; Jiang et al, 2014; Irwin et al, 1992; Irwin, 1995, 2004; Prager & Jollès, 1996). Changes in the function of lysozymelikely leads to changes in the evolutionary constraints acting upon these genes. To examine this issue we calculated the diverge-nce at nonsynonymous and synonymous sites among lysozymegenes, with the results from three divergent representatives of pecoran ruminants (cow, goat, and Tibetan antelope), and between genes in these three species and the single copy lysozymegene sequences found in pig and horse shown in Table 5 (similar results were seen with the other pecoran rumin-ant species). The relative rates of nonsynonymous to synonymous substit-utions (dn/ds) varied between genes when comp-ared among ruminants, from low values for the5 and6 genes, which imply that they are strongly constra-ined, to high values for the3 and9 genes, suggesting that there is little constraint on their protein sequences (Table 5). The cow3 and9 genes fail to predict intact open reading frames, suggesting that they are pseudogenes (Table 1 and Figure 2) and thus should have no evolutionary constraints on their protein sequences.

Figure 5 Phylogeny of ruminant lysozymegenes derived from exon 4 sequences of predicted lysozymegenes

The phylogeny of the lysozymegenes was estimated from aligned exon 4 sequences (192 aligned bases in all sequences) using maximum likelihood, as implemented in Mega6.06 (Tamura et al, 2013), using the Kimura 2-paramater model with a gamma distribution, which was the best fitting model for the sequence data. Similar results were obtained with the neighbor-joining method or parsimony. The phylogeny was bootstrapped 500 times. Outgroups used to root the phylogeny are shown at the bottom. The ten types of lysozymegenes are indicated on the right, with the bootstrap values that support 8 of these clades (all except the1 and2 clade) shown in bold.

Table 5 Rates of evolution of ruminant lysozyme c genes

a– Sheep4 used to replace the incomplete goat4 for comparisons.

The cow genome contains three3-like genes, with only one being a full-length gene sequence (3b), and a single copy of this gene was found in most of the other pecoran ruminant species (Tables 1 and 3). Cow3b gene was previously identified as the cow lyso-zymepseudogene??NS4 (Irwin, 1995, 2004). While the3 gene sequences from tribe Bovini (cow, zebu andwater buffalo) all share a frame shift mutation in exon 3 (amino acid residue 100 in Figure 2), which would prevent translation of a functional product, and additio-nal mutations that potentially disrupt functions found in some sequences at other locations, the sequences from sheep, goat, and Tibetan antelope all predict a full-length open reading frame (Figure S1). This observation might suggest that the3 gene became a pseudogene, due to a frame-shifting mutation, on the lineage leading to tribe Bovini, after divergence from the other pecoran ruminant lineages. However, a high rate of nonsynonymous substitutions is also observed between the goat and Tibetan antelope3 gene sequences (Table 5) and between the sheep and both the goat and Tibetan antelope sequences (results not shown) sugge-sting that few evolutionary constraints were acting on this sequence and that this gene may have been non-functional in the common ancestor of all pecoran rumin-ants. It is possible that a mutation that prevented expres-sion, or an amino acid substation that that preve-nted function, rather than a mutation that prevents translation of an intact product, was the initial mutation that created this pseudogene.

The second gene with a very high dn/ds ratio is the9 gene, which is composed of only 2 exons, exons 1 and 4, in the cow, due to the loss of exons 2 and 3 (Table 1 and Figure 2). Orthologs of the9 genes in other pecoran ruminant species also have similar gene structures (Tables 1 and 3 and Figure S1), suggesting that this structure exited in the Lyz9 gene in the ancestor of all pecoran ruminants. The loss of exon 2 and 3 sequences from9 prevents the translation of a functional lysozyme, thus it can be concluded that this pseudogene originated before the radiation of the pecoran ruminants. Consistent with this conclusion, a high rate of divergence at nonsynonymous sites is observed in the9 gene sequence among all pecoran ruminant species (Table 5 and results not shown).

The3 and9 genes account for 4 of the five predicted lysozymepseudogenes in the cow genome (Table 1 and Figure 2). In addition to a pair of inframe stop codons (located between amino acid residues 24 and 25, and residue 26, Figure 2), the initiation codon for the cow8 gene is valine rather than methionine (amino acid -18 in Figure 2 and Figure S1). Orthologs of the8 gene from members of the tribe Bovini (yak, zebu, and water buffalo) share the inframe stop codons, as well other mutations such as a 9 base deletion in exon 2 (3 codons – residues 66-68 in Figure 2), while8 sequences from other pecoran species (sheep, goat and Tibetan antelope) do not possess any obvious harmful amino acid substitution, other than the valine substitution at the initiation codon (Figure S1). In contrast to the3 and9 pseudogenes, a much lower dn/ds ratio was obser-ved in the pairwise comparisons among cow, goat, and Tibetan antelope (Table 5), which would be consistent with functional constraints acting on some, but not necessarily all, of the Lyz8 protein sequences. These observations appear to suggest, that despite the replacement of the initiator methionine with valine, the Lyz8 protein sequences in the sheep, goat and Tibetan antelope is functional, while a mutation occurred on the lineage leading to tribe Bovini to producing the8 pseudogene. How a functional protein can be translated from the8 gene, or evolutionary constraints that mirror protein function, is unclear. A downstream ATG, at codon 85 of the mature protein sequence (Figure 2), would be predicted to yield a protein of only 45 amino acid residues, far shorter than the typical 145 amino acid long protein lysozymeprecursor, with most of the sequence not being translated and thus not under evolutionary constraint for protein function.

Episodic evolution of ruminant lysozyme c genes

The cow lysozymegenes displaying the lowest dn/ds ratios among ruminant species, and thus implying the strongest evolutionary constraints, are the5 and6 genes (Table 5), which are expressed in the abomasum (Table 1). Lysozymegenes expressed predominantly in non-stomach tissues (Irwin, 2004), such as1, in milk,2, in the trachea, and10, in macrophages, have intermediate dn/ds ratios, but ratios that lower than those seen for the3,8, and9 pseudogenes (Table 5). However, when the dn/ds ratios are calculated between the ruminant genes (cow, goat, and Tibetan antelope) and an outgroup sequence (pig or horse), the stomach expressed5 and6 genes are seen to have dn/ds ratios that are either similar (when pig is the outgroup) or higher (horse being the outgroup) than those seen for the non-stomach (1,2, and10) genes (Table 5). To obtain this pattern of results, these observations suggest that the dn/ds ratio on the common ancestral lineage leading to the ruminants, after divergence from pig or horse, but before radiation of the pecoran ruminants, was higher for the lysozymegenes expressed in the abomasum (5 and6) than for those expressed in non-stomach tissues (1,2, and10). This suggests that the rates of evolution of lysozymegenes expressed in the abomasum display an episodic pattern, with more rapid evolution on the early ruminant lineage, and a slower rate within the pecoran ruminants. These results are consistent with previous findings of accelerated evolution of lysozymeprotein sequences obtained from the abomasum of ruminant species (Jollès et al, 1989; Irwin & Wilson, 1990; Irwin et al, 1992, 1993).

CONCLUSIONS

Genome sequences have advanced our understan-ding of the evolution of the lysozymegene family in ruminant species. Genomic sequences from seven diver-g-ent pecoran ruminant species allowed us to demonstrate that the genome of the pecoran ruminant common ancestor possessed at least 10 lysozymegenes, and that these genes have largely been retained by extant rumin-ant species. More recent gene duplication, likely via segmental duplications (Liu et al, 2009; Seo et al, 2013), have resulted in increases in the number of lysozymegenes on some lineages, with 14 genes found in the cow, but we can not exclude the possibility that some duplications may have been missed during assembly of some genomes. Lysozymegenes have not evolved in a simple divergent manner, but rather by concerted evolution acting independently on each exon, yielding differing phylogenetic relationships for the ten types of lysozymegenes. Some lysozymegenes have become pseudogenes, either due to mutations in their coding sequence (e.g.,3 and8) or by deletion of exon sequences (e.g.,9). Some pseudogenes may have been generated by incomplete duplication of genes, such as3a and3c in the cow. Despite being presumably non-functional, at least two pseudogenes that exited in the ancestral pecoran ruminant (3 and9) have been retained in diverse ruminant species. A third lysozymegene has its initiation codon mutated to valine (from methionine), yet shows evidence that its coding sequence is evolutionary constrained on some ruminant lineages. This suggests that some lysozymepseudogenes may retain biological functions, however, how protein function in this sequence is maintained is unclear. Changes in the rates of nonsynonymous substit-utions suggest that changes have occurred in the functi-onal constraints acting on lysozymeprotein sequences, and these changes have occurred in an episodic fashion.

ACKNOWLEDGEMENTS

I thank the editor and two reviewers for helpful comments that have improved the manuscript.

REFERENCES

Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. 1990. Basic local alignment search tool.215(3): 403-410.

Callewaert L, Michiels CW. 2010. Lysozymes in the animal kingdom.35(1): 127-160.

Canavez FC, Luche DD, Stothard P, Leite KR, Sousa-Canavez JM, Plastow G, Meidanis J, Souza MA, Feijao P, Moore SS, Camara-Lopes LH. 2012. Genome sequence and assembly of.103(3): 342-348.

Clauss M, Hume ID, Hummel J. 2010. Evolutionary adaptations of ruminants and their potential relevance for modern production systems.4(7): 979-992.

Dobson DE, Prager EM, Wilson AC. 1984. Stomach lysozymes of ruminants. I. Distribution and catalytic properties.259(18): 11607-11616.

Dong Y, Xie M, Jiang Y, Xiao NQ, Du XY, Zhang WG, Tosser-Klopp G, Wang JH, Yang S, Liang J, Chen WB, Chen J, Zeng P, Hou Y, Bian C, Pan SK, Li YX, Liu X, Wang WL, Servin B, Sayre B, Zhu B, Sweeney D, Moore R, Nie W, Shen Y, Zhao R, Zhang G, Li J, Faraut T, Womack J, Zhang Y, Kijas J, Cockett N, Xu X, Zhao S, Wang J, Wang W. 2013. Sequencing and automated whole-genome optical mapping of the genome of a domestic goat ().31(2): 135-141.

Edgar RC. 2004. MUSCLE: multiple sequence alignment with high accuracy and high throughput.32(5): 1792-1797.

Gallagher DS Jr, Threadgill DW, Ryan AM, Womack JE, Irwin DM. 1993. Physical mapping of the lysozyme gene family in cattle.4(7): 368-373.

Ge RL, Cai Q, Shen YY, San A, Ma L, Zhang Y, Yi X, Chen Y, Yang L, Huang Y, He R, Hui Y, Hao M, Li Y, Wang B, Ou X, Xu J, Zhang Y, Wu K, Geng C, Zhou W, Zhou T, Irwin DM, Yang Y, Ying L, Bao H, Kim J, Larkin DM, Ma J, Lewin HA, Xing J, Platt RN 2nd, Ray DA, Auvil L, Capitanu B, Zhang X, Zhang G, Murphy RW, Wang J, Zhang YP, Wang J. 2013. Draft genome sequence of the Tibetan antelope.4: 1858.

Irwin DM. 1995. Evolution of the bovine lysozyme gene family: changes in gene expression and reversion of function.41(3): 299-312.

Irwin DM. 1996. Molecular evolution of ruminant lysozymes.: Jollès P. Lysozymes: Model Enzymes in Biochemistry and Molecular Biology. Basel: Birkh?user Verlag, 347-361.

Irwin DM. 2004. Evolution of cow nonstomach lysozyme genes.47(6): 1082-1090.

Irwin DM, Wilson AC. 1989. Multiple cDNA sequences and the evolution of bovine stomach lysozyme.264(19): 11387-11393.

Irwin DM, Wilson AC. 1990. Concerted evolution of ruminant stomach lysozymes. Characterization of lysozyme cDNA clones from sheep and deer.265(9): 4944-4952.

Irwin DM, Sidow A, White RT, Wilson AC. 1989. Multiple genes for ruminant lysozymes.: Smith-Gill SJ, Sercarz EE. The Immune Response to Structurally Defined Proteins. NY: Adenine Press, 73-85.

Irwin DM, Prager EM, Wilson AC. 1992. Evolutionary genetics of ruminant lysozymes.23(3): 193-202.

Irwin DM, White RT, Wilson AC. 1993. Characterization of the cow stomach lysozyme genes: repetitive DNA and concerted evolution.37(4): 355-366.

Irwin DM, Yu M, Wen Y. 1996. Isolation and characterization of vertebrate lysozyme genes.: Jollès P. Lysozymes: Model Enzymes in Biochemistry and Molecular Biology. Basel: Birkh?user Verlag, 225-241.

Irwin DM, Biegel JM, Stewart CB. 2011. Evolution of the mammalian lysozyme gene family.11: 166.

Janis C. 1976. The evolutionary strategy of Equidae and the origins of rumen and cecal digestion.30(4): 757-774.

Jiang Y, Xie M, Chen W, Talbot R, Maddox JF, Faraut T, Wu C, Muzny DM, Li Y, Zhang W, Stanton JA, Brauning R, Barris WC, Hourlier T, Aken BL, Searle SM, Adelson DL, Bian C, Cam GR, Chen Y, Cheng S, DeSilva U, Dixen K, Dong Y, Fan G, Franklin IR, Fu S, Fuentes-Utrilla P, Guan R, Highland MA, Holder ME, Huang G, Ingham AB, Jhangiani SN, Kalra D, Kovar CL, Lee SL, Liu W, Liu X, Lu C, Lü T, Mathew T, McWilliam S, Menzies M, Pan S, Robelin D, Servin B, Townley D, Wang W, Wei B, White SN, Yang X, Ye C, Yue Y, Zeng P, Zhou Q, Hansen JB, Kristiansen K, Gibbs RA, Flicek P, Warkup CC, Jones HE, Oddy VH, Nicholas FW, McEwan JC, Kijas JW, Wang J, Worley KC, Archibald AL, Cockett N, Xu X, Wang W, Dalrymple BP. 2014. The sheep genome illuminates biology of the rumen and lipid metabolism.344(6188): 1168-1173.

Jollès J, Jollès P, Bowman BH, Prager EM, Stewart CB, Wilson AC. 1989. Episodic evolution in the stomach lysozymes of ruminants.28(6): 528-535.

Kornegay JR, Schilling JW, Wilson AC. 1994. Molecular adaptation of a leaf-eating bird: stomach lysozyme of the hoatzin.11(6): 921-928.

Kornegay JR. 1996. Molecular genetics and evolution of stomach and nonstomach lysozymes in the hoatzin.42(6): 676-684.

Liu GE, Ventura M, Cellamare A, Chen L, Cheng Z, Zhu B, Li C, Song J, Eichler EE. 2009. Analysis of recent segmental duplications in the bovine genome.10: 571.

Mackie RI. 2002. Mutualistic fermentative digestion in the gastrointestinal tract: diversity and evolution.42(2): 319-326.

Morgan CC, Foster PG, Webb AE, Pisani D, McInerney JO, O'Connell MJ. 2013. Heterogeneous models place the root of the placental mammal phylogeny.30(9): 2145-2156.

Néron B, Ménager H, Maufrais C, Joly N, Maupetit J, Letort S, Carrere S, Tuffery P, Letondal C. 2009. Mobyle: a new full web bioinformatics framework.25(22): 3005-3011.

Prager EM. 1996. Adaptive evolution of lysozyme: changes in amino acid sequence, regulation of expression and gene number.: Jollès P. Lysozymes: Model Enzymes in Biochemistry and Molecular Biology. Basel: Birkh?user Verlag, 323-345.

Prager EM, Jollès P. 1996. Animal lysozymesand: an overview.: Jollès P. Lysozymes: Model Enzymes in Biochemistry and Molecular Biology. Basel: Birkh?user Verlag, 9-31.

Qiu Q, Zhang GJ, Ma T, Qian WB, Wang JY, Ye ZQ, Cao CC, Hu QJ, Kim J, Larkin DM, Auvil L, Capitanu B, Ma J, Lewin HA, Qian XJ, Lang YS, Zhou R, Wang LZ, Wang K, Xia JQ, Liao SG, Pan SK, Lu X, Hou HL, Wang Y, Zang XT, Yin Y, Ma H, Zhang J, Wang ZF, Zhang YM, Zhang DW, Yonezawa T, Hasegawa M, Zhong Y, Liu WB, Zhang Y, Huang ZY, Zhang SX, Long RJ, Yang HM, Wang J, Lenstra JA, Cooper DN, Wu Y, Wang J, Shi P, Wang J, Liu JQ. 2012. The yak genome and adaptation to life at high altitude.44(8): 946-949.

Romiguier J, Ranwez V, Delsuc F, Galtier N, Douzery EJ. 2013. Less is more in mammalian phylogenomics: AT-rich genes minimize tree conflicts and unravel the root of placental mammals.30(9): 2134-2144.

Schwartz S, Zhang Z, Frazer KA, Smit A, Riemer C, Bouck J, Gibbs R, Hardison R, Miller W. 2000. PipMaker--a web server for aligning two genomic DNA sequences.10(4): 577-586.

Schwartz S, Elnitski L, Li M, Weirauch M, Riemer C, Smit A; NISC Comparative Sequencing Program, Green ED, Hardison RC, Miller W. 2003. MultiPipMaker and supporting tools: Alignments and analysis of multiple genomic DNA sequences.31(13): 3518-3524.

Seo S, Larkin DM, Loor JJ. 2013. Cattle genomics and its implications for future nutritional strategies for dairy cattle.7(S1): 172-183.

Short ML, Nickel J, Schmitz A, Renkawitz R. 1996. Lysozyme gene expression and regulation.: Jollès P. Lysozymes: Model Enzymes in Biochemistry and Molecular Biology. Basel: Birkh?user Verlag, 243-257.

Stevens CE, Hume ID. 1998. Contributions of microbes in vertebrate gastrointestinal tract to production and conservation of nutrients.78(2): 393-427.

Stewart CB, Wilson AC. 1987. Sequence convergence and functional adaptation of stomach lysozymes from foregut fermenters.52: 891-899.

Stewart CB, Schilling JW, Wilson AC. 1987. Adaptive evolution in the stomach lysozymes of foregut fermenters.330(6146): 401-404.

Strimmer K, von Haeseler A. 1996. Quartet puzzling: A quartet maximum likelihood method for reconstructing tree topologies.13(7): 964-969.

Swanson KW, Irwin DM, Wilson AC. 1991. Stomach lysozyme gene of the langur monkey: tests for convergence and positive selection.33(5): 418-425.

Takeuchi K, Irwin DM, Gallup M, Shinbrot E, Kai H, Stewart CB, Basbaum C. 1993. Multiple cDNA sequences of bovine tracheal lysozyme., 268(36): 27440-27446.

Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. 2013. MEGA6: Molecular evolutionary genetics analysis version 6.0.30(12): 2725-2729.

Wen Y, Irwin DM. 1999. Mosaic evolution of ruminant stomach lysozyme genes.13(3): 474-482.

Yu M, Irwin DM. 1996. Evolution of stomach lysozyme: the pig lysozyme gene.5(2): 298-308.

Zimin AV, Delcher AL, Florea L, Kelley DR, Schatz MC, Puiu D, Hanrahan F, Pertea G, Van Tassell CP, Sonstegard TS, Mar?ais G, Roberts M, Subramanian P, Yorke JA, Salzberg SL. 2009. A whole-genome assembly of the domestic cow,.10(4): R42.

08 October 2014; Accepted: 02 December 2014

This study was supported by grants from the Natural Sciences and Engineering Research Council (RGPIN 183701)

, E-mail: david.irwin@utoronto.ca

10.13918/j.issn.2095-8137.2015.1.1