Evolution of the protease-activated receptor family in vertebrates.

Belonging to the G protein-coupled receptor (GPcr) family, the protease-activated receptors (Pars) consist of 4 members, PAR1-4. PARs mediate the activation of cells via thrombin, serine and other proteases. Such protease-triggered signaling events are thought to be critical for hemostasis, thrombosis and other normal pathological processes. In the present study, we examined the evolution of PARs by analyzing phylogenetic trees, chromosome location, selective pressure and functional divergence based on the 169 functional gene alignment sequences from 57 vertebrate gene sequences. We found that the 4 Pars originated from 4 invertebrate ancestors by phylogenetic trees analysis. The selective pressure results revealed that only PAR1 appeared by positive selection during its evolution, while the other PAR members did not. In addition, we noticed that although these PARs evolved separately, the results of functional divergence indicated that their evolutional rates were similar and their functions did not significantly diverge. The findings of our study provide valuable insight into the evolutionary history of the vertebrate PAR family.

Introduction

Discovered in the 1990s, protease-activated receptors (PARs) are membrane-spanning proteins that belong to the G protein-coupled receptor (GPCR) family and exist on the surface of cells in a wide variety of tissues (1). The PARs contain 7 transmembrane (TM1-7) helices, an extracellular amino terminal domain, 3 intracellular loops (ICL1-3), 3 extracellular loops (ECL1-3) and an intracellular carboxyl terminal domain. As well as connecting TM4 and TM5, ECL2 also makes a disulfide bond with TM3 that is conserved amongst GPCRs and contributes to receptor or structural stability (2). The PAR family consists of 4 members, PAR1-4. PAR1, PAR3 and PAR4 are thrombin receptors, while PAR2 is a receptor for serine proteases (including trypsin, mast cell tryptase, as well as factors Xa and VIIa, etc.) (1,3,4).

Human PAR1 has been identified as a protein containing 425 residues, including an amino-terminal signal sequence of 20 residues and an extracellular amino-terminal domain of 75 residues (5). PAR1 is activated when cleaved by thrombin at a site between Arg41 and Ser42, exposing a new N-terminal tethered ligand domain with the sequence SFLLRN. The new N-terminus acts as a tethered ligand and binds intramolecularly to the body of the receptor to effect TM signaling (6,7).

Human PAR2 consists of 397 amino acids with a molecular weight of 44 kDa. It has an extracellular N-terminal domain of 75 amino acids tethered to a TM domain of 155 amino acids, assembled in 7 pseudo-parallel helical sequences connected by 3 extracellular and 3 intracellular loops of 117 amino acids, with a small eighth helix within the intracellular C-terminus of 50 amino acids. The extracellular N-terminus of PAR2 is cleaved at a site between Arg47 and Gly48 by protease (mainly serine protease), and the newly exposed N-terminus is GYPGQV. The exposed N-terminus, known as a tethered ligand, then folds back and self-activates PAR2 through binding to conserved regions of ECL2 and/or the PAR2 TM region (1,8–10).

PAR3 is a typical GPCR with a thrombin cleavage site between Lys38 and Thr39, which has 27% amino acid sequence similarity to PAR1 and 28% similarity to PAR2 (11). After being cleaved by thrombin, PAR3 exposes a new amino terminus (TFRGAP) which interacts with the receptor as a tethered ligand (12).

Human PAR4 is a 385-amino acid protein with a potential thrombin cleavage site in the extracellular amino-terminal domain between Arg47 and Gly48, and has 33% similarity to the other human PARs (13). PAR4 is one of the thrombin receptors on human platelets and a potential target for the management of thrombotic disorders. Briefly, when these receptors are activated by proteinases at the specific cleavage site, the extracellular N-terminus of the receptor, a new N-terminus, is exposed. The new N-terminus has been found to act as a tethered ligand and binds intramolecularly to initiate cellular signals (13).

It has been demonstrated that following activation, the PAR family is able to stimulate complex intracellular signaling networks through classical G protein-mediated pathways and β-arrestin signaling pathways (14). The signaling pathways induced by the activation of PARs play a key role in a number of physiological and pathological processes. For example, it has been demonstrated that PARs expressed on cells are involved in immune responses and inflammation, regulate endothelial-leukocyte interactions and modulate the secretion of inflammatory mediators or neuropeptides (4). PARs expressed on platelets and the vascular endothelium play important roles in normal blood vessel biology, which contribute to the pathogenesis of several cardiovascular diseases, including atherosclerosis, restenosis and thrombosis (15). Thus, it has been suggested that the PARs are potential therapeutic targets for the treatment of inflammation, hemostasis, thrombosis, vascular dysfunction and cancer (16–19). Apart from studies on the structure and function of PARs, few studies to date have focused on the genetic evolution of the family of PARs. Kahn et al compared the amino acid sequences, gene structures, locus organization and chromosomal locations in human and mouse PARs (20). Xu et al found that the function of PARs was conserved among vertebrates, after analyzing the expression patterns in zebrafish and their mammalian human and mouse counterparts (19). Thus far, however, to the best of our knowleged, there are no studies available on the global evolution of the PAR family in vertebrates. In this study, we describe the evolutionary genetic association between the members of the PAR family based on the analysis of phylogenetic trees, chromosome location, selective pressure and functional divergence.

Materials and methods

Sequence data collection

PAR gene sequences were obtained, based on their orthologous and paralogous relationship, by querying the Ensembl genome assemblies (http://www.ensembl.org/index.html) using the human PAR gene sequences. The obtained PAR sequences were then used as queries to obtain the best hit in BLAST at the NCBI database (http://www.ncbi.nlm.nih.gov/pubmed).

Multiple alignment and phylogenetic analysis

The protein coding sequences of the PAR gene family were aligned using the ClustalW program in MEGA 5.1 (21). The aligned sequences were used for the subsequent phylogenetic analysis. MEGA 5.1 was used to construct a maximum likelihood (ML) tree of the PAR family with the best fitting model of JTI+G+F that was selected by a model test in the same package with a bootstrap value based on 1,000 repetitions. MrBayes v3.1.2 was used to construct a Bayesian inference tree with the nucleotide substitution model (22). The Markov chain Monte Carlo (MCMC) technique was used for the PAR family. The repetitions run for PAR1, PAR2, PAR3 and PAR4 were 2,000,000, 3,000,000, 1,000,000 and 1,000,000 generations, respectively. PAR1, PAR2, PAR3 and PAR4 were sampled every 1,000 generations. The first 25% of the trees was discarded as burn-in. Convergence was assessed by determining the average standard deviation of split frequencies (<0.01). The posterior probabilities for internal nodes were calculated from the posterior density of trees.

Selective pressure analysis

In order to determine whether positive selection is involved in the evolution of the PAR family, CodeML in PAML 4.7 was used with the models M0 (one ratio), M1a (near neutral), M2a (positive selection), M3 (discrete), M7 (β) and M8 (β and ω) (23–25). M7 is a null model that does not allow for any codons with ω>1, whereas the M8 model allows for positive selective sites (ω>1). When the M8 model fit the data significantly better (P-value <0.05) than the null model (M7), the presence of positive selection sites was suggested. Conversely, results with a P-value >0.05 were regarded as lacking any positive selection sites. The non-synonymous/synonymous substitutions rate ratio (ω=dN/dS) was also used to indicate selective pressure;ω>1 indicates positive selection, ω<1 indicates negative selection and ω=1 indicates neutral evolution (26). Twice the difference in log likelihood between the M7 and M8 models (2Δl) was compared against χ2 with critical values of 0.01 significance levels. The number of non-synonymous substitutions (dN) per non-synonymous site and the number of synonymous substitutions (dS) per synonymous site were computed by MEGA 5.1 using the modified Nei-Gojobori method. The transition/transversion ratio computed by MEGA 5.1 with the ML method was 1.4, 1.58, 1.72 and 1.16 corresponding to PAR1, PAR2, PAR3 and PAR3, respectively.

Functional divergence analysis

The multiple alignment of amino acid sequences among clusters of PAR1 and PAR2, PAR1 and PAR3, PAR1 and PAR4, PAR2 and PAR3, PAR2 and PAR4, and PAR3 and PAR4, were input into a Clustal format and type I functional divergence analysis was performed using Diverge (v2.0) with the maximum-likelihood estimation (MLE) and model-free method (MFE) (27,28). The co-efficient of functional divergence θ (0<θ<1) was used to indicate the functional divergence that occurred in different clusters of the PAR family. A significant result, i.e., θ>0, suggests that it is desirable to identify the amino acid residues in the protein which have experienced a shift in their functional constraints. These sites may be relevant to the functional-structural differences between proteins.

Conservation of synteny analysis

To further study the evolution of the PAR gene family, we investigated the chromosomal distribution of PAR1, PAR2, PAR3 and PAR4 in eutheria, birds and teleosts. The location of chromosomes and the number of exons in the PAR family were searched in the Ensembl genome assemblies (http://www.ensembl.org/index.html).

Results

PAR gene repertoires in vertebrates

To examine the origin and genetic evolution of the PAR family in vertebrates, we collected 57 vertebrate gene sequences from Ensembl and tBLASTn (NCBI) using Homo sapiens genes as queries. Following the elimination of uncompleted sequences in Ensembl and NCBI, 169 functional gene sequences were applied to this study (Table I). The taxa comprised 18 non-mammals (9 teleost fish: coelacanth, platyfish, tilapia, zebrafish, cod, fugu, medaka, stickleback and tetraodon; 1 amphibian: xenopus; 3 reptiles: anole lizard, armadillo and Chinese softshell turtle; 5 birds: duck, chicken, flycatcher, turkey and zebra finch) and 38 mammals that include 10 primates (human, bushbaby, chimpanzee, gorilla, gibbon, orangutan, macaque, marmoset, mouse lemur and tarsier), 26 other mammals (alpaca, cow, cat, dog, dolphin, ferret, horse, megabat, microbat, pig, panda, shrew, elephant, hyrax, wallaby, lesser hedgehog tenrec, sloth, Tasmanian devil, guinea pig, mouse, rabbit, pika, rat, kangaroo rat, squirrel and tree shrew), 1 metatheria (opossum) and 1 prototheria (platypus). According to the completeness of genes, we divided the collected genes into 2 groups as follows: i) functional genes, which are sequences containing full-length open reading frames (ORFs) and ii) uncompleted genes, meaning those that lack start or stop codes or codes in the middle of sequences. Based on these criteria, we identified 169 functional genes and 69 uncompleted genes from a total of 238 PAR family genes. The number of PAR genes varied between the non-mammalian and mammalian vertebrates; 58 functional genes were identified in the non-mammalian species, ranging from 1 in the turkey to 11 in zebrafish; 49 uncompleted genes were found in the non-mammalian species, ranging from 1 in the duck, zebra finch, Chinese softshell turtle, platyfish and zebrafish to 9 in the coelacanth. A total of 111 functional genes were identified in mammals and these ranged from 1 in the mouse lemur, alpaca, hyrax, wallaby, and lesser hedgehog tenrec to 5 in the opossum.

Table I

Number of functional and uncompleted genes in the vertebrate PAR family.

Clade	Species	Name	Functional genes	Uncompleted genes	Total
Eutheria	Homo sapiens	Human	4		4
	Otolemur garnettii	Bushbaby	4		4
	Pan troglodytes	Chimpanzee	4		4
	Nomascus leucogenys	Gibbon	3	1	4
	Gorilla gorilla	Gorilla	4		4
	Pongo abelii	Orangutan	4		4
	Macaca mulatta	Macaque	4		4
	Callithrix jacchus	Marmoset	4		4
	Microcebus murinus	Mouse lemur	1	2	3
	Tarsius syrichta	Tarsier		1	1
	Vicugna pacos	Alpaca	1	2	3
	Bos taurus	Cow	4		4
	Felis catus	Cat	4		4
	Canis lupus familiaris	Dog	4		4
	Tursiops truncatus	Dolphin	3	1	4
	Mustela putorius furo	Ferret	4		4
	Equus caballus	Horse	3	1	4
	Pteropus vampyrus	Megabat	3	1	4
	Myotis lucifugus	Microbat	2	1	3
	Sus scrofa	Pig	4		4
	Ailuropoda melanoleuca	Panda	3	1	4
	Sorex araneus	Shrew	2		2
	Loxodonta africana	Elephant	3	1	4
	Procavia capensis	Hyrax	1	3	4
	Macropus eugenii	Wallaby	1	1	2
	Echinops telfairi	Lesser hedgehog tenrec	1	1	2
	Choloepus hoffmanni	Sloth	2	1	3
	Sarcophilus harrisii	Tasmanian devil	4		4
	Cacia porcellus	Guinea pig	2	1	3
	Mus musculus	Mouse	4		4
	Oryctolagus cuniculus	Rabbit	2	1	3
	Ochotona princeps	Pika	2		2
	Dipodomys ordii	Kangaroo rat		3	3
	Rattus norvegicus	Rat	4		4
	Ictidomys tridecemlineatus	Squirrel	2		2
	Tupaia belangeri	Tree shrew		2	2
Metatheria	Monodelphis domestica	Opossum	5		5
Prototheria	Ornithorhynchus anatinus	Platypus	3		3
Birds	Anas platyrhynchos	Duck	4	1	5
	Gallus gallus	Chicken	5		5
	Ficedula albicollis	Flycatcher	5		5
	Meleagris gallopavo	Turkey	1	4	5
	Taeniopygia guttata	Zebra finch	3	1	4
Reptiles	Anolis carolinensis	Anole lizard	2		2
	Dasypus novemcinctus	Armadillo	3		3
	Pelodiscus sinensis	Chinese softshell turtle	2	1	3
Amphibia	Xenopus tropicalis	Xenopus	5		5
Teleost	Latimeria chalumnae	Coelacanth	7	9	16
	Xiphophorus maculatus	Platyfish	6	1	7
	Oreochromis niloticus	Tilapia	6	4	10
	Danio rerio	Zebrafish	11	1	12
	Gadus morhua	Cod		6	6
	Takifugu rubripes	Fugu	2	3	5
	Oryzias latipes	Medaka	2	3	5
	Gasterosteus aculeatus	Stickleback		6	6
	Tetraodon nigroviridis	Tetraodon		4	4
Total		57	169	69	238

PAR, protease-activated receptor.

Phylogenetic analysis of the PAR genes

To examine the evolutionary relationship of the PAR gene family, the 169 functional genes that were identified from all 238 genes were analyzed by the ML method and Bayesian inference. These yielded a similar result. The phylogenetic tree constructed by ML is shown in Fig. 1. Based on this model, the PAR gene family can be divided into 4 clades, PAR1, PAR2, PAR3 and PAR4, suggesting that PAR gene sequences have major differences. To further examine the phylogenetic relationship of each clade, we constructed 4 sub-trees for PAR1, PAR2, PAR3 and PAR4 (Figs. 2–5) using Bayesian inference. In the PAR1 clade, the posterior probability is >58. Teleost PAR1 genes form a group separate from eutherian and bird genes. We found that there is only 1 AR1 copy in eutheria, metatheria, bird, amphibian and reptile, whereas there are 6 copies in the zebrafish, 3 copies in the tilapia and 2 copies in the platyfish. In addition, the posterior probability of the teleost genes is higher than that of eutherian and bird genes. In the PAR2 clade, the posterior probability is >54. Primates, birds and teleost PAR2s form a group separate from other vertebrate genes. There is only 1 PAR2 copy in most vertebrates apart from teleosts (2 copies in zebrafish, 2 copies in tilapia, 2 copies in platyfish and 2 copies in coelacanth). In the PAR3 clade, the posterior probability is >54, with birds and teleosts each forming a group separate from other species. There is only 1 PAR3 copy in most vertebrates apart from zebrafish (2 copies). The posterior probability of primates is 100 and the posterior probabilities between megabat and microbat, cow and pig, zebra finch and flycatcher, tilapia and zebrafish, opossum and Tasmanian devil, rat and mouse are also 100. In the PAR4 clade, the posterior probability is >57, with birds and teleosts again forming a group separate from other species. Teleosts, amphibians and birds have multiple copies in the PAR4 clade (3 copies in coelacanth, 2 copies in xenopus, 2 copies in flycatcher and 2 copies in the chicken).

Figure 1

Maximum likelihood (ML) tree of the protease-activated receptor (PAR) gene family in vertebrates.

Figure 2

The sub-trees of protease-activated receptor (PAR1) were constructed using Bayesian inference.

Figure 3

The sub-trees of protease-activated receptor (PAR2) were constructed using Bayesian inference.

Figure 4

The sub-trees of protease-activated receptor (PAR3) were constructed using Bayesian inference.

Figure 5

The sub-trees of protease-activated receptor (PAR4) were constructed using Bayesian inference.

The 4 PAR clades were analyzed to determine whether there is functional divergence between them (PAR1 vs. PAR2, PAR1 vs. PAR3, PAR1 vs. PAR4, PAR2 vs. PAR3, PAR2 vs. PAR4 and PAR3 vs. PAR4). We estimated type I divergence using Diverge v2.0 (Table II). The results revealed that all θ values were significant (>0), indicating that a site-specific rate shift during gene duplication was a common phenomenon in the evolution of the vertebrate PAR family. We also noticed that the 6 θ values of PAR1 vs. PAR2, PAR1 vs. PAR3, PAR1 vs. PAR4, PAR2 vs. PAR3, PAR2 vs. PAR4 and PAR3 vs. PAR4 were all <0.5, suggesting that they did not diverge markedly.

Table II

Functional divergence of the PAR family genes.

	θ value		Standard error
	MFE method	MLE method	MFE method
PAR1 vs. PAR2	0.397	0.364	0.056
PAR1 vs. PAR3	0.370	0.395	0.060
PAR1 vs. PAR4	0.360	0.382	0.057
PAR2 vs. PAR3	0.433	0.494	0.061
PAR2 vs. PAR4	0.313	0.372	0.053
PAR3 vs. PAR4	0.323	0.372	0.058

PAR, protease-activated receptor; θ, coefficient of functional divergence; MFE, model-free method; MLE, maximum-likelihood estimation.

To determine whether positive selection drove the evolution of the PAR gene family, we calculated the dN and dS distances between each pair of the sequences from the 4 clades (Fig. 6). The distance between dN and dS is large in the pairwise comparison of the PAR1 sequences, suggesting that positive selection is involved in PAR1 evolution. The value of dN is not significantly higher than the value of dS in the pairwise comparisons of the PAR2, PAR3 and PAR4 sequences. Most values of dN/dS in these sequences were distributed below the diagonal, suggesting that positive selection is not involved in the evolution of these 3 clades. In addition, site-specific tests for positive selection were performed on the vertebrate PAR gene family using PAML4.7 (Table III) and some positive selection sites for PAR1 were found. The amino acids, 65G** and 76K**, were identified as positive selection sites (referring to Homo sapiens) in the PAR1 gene, whereas there were no positive selection sites found in the PAR2, PAR3 and PAR4 genes. The positive selection sites identified by PAML 4.7 were the same as those identified by dN/dS analysis using MEGA 5.1.

Figure 6

Pairwise comparisons of the number of non-synonymous substitutions (dN) per non-synonymous site and the number of synonymous substitutions (dS) per synonymous site for the protease-activated receptor (PAR) gene family. Pairwise comparisons of dN and dS for (A) PAR1, (B) PAR2, (C) PAR3 and (D) PAR4.

Table III

Site-specific tests for positive selection sites in the PAR gene family.

	Models	InL	Estimates of parameters	2ΔI	Positive selection sites
PAR1	M7 (β)	−18520.05	p=0.42638, q=1.45124	22.02 (P<0.01)	NA
	M8 (β and ω)	−18498.03	p0=0.95469, p=0.54737, q=2.73176 (p1=0.04531) w=1.26332	22.02 (P<0.01)	65G** 76K**
PAR2	M7 (β)	−19806.32	p=0.49569, q=2.76102	0 (P=1.00)	NA
	M8 (β and ω)	−19806.32	p0=0.99999, p=0.49569, q=2.76102 (p1=0.00001) w=2.89180	0 (P=1.00)	None
PAR3	M7 (β)	−13970.25	p=0.70619, q=3.15181	0 (P=1.00)	NA
	M8 (β and ω)	−13970.25	p0=0.99999, p=0.70622 q=3.15206 (p1=0.00001), w=1.00000	0 (P=1.00)	None
PAR4	M7 (β)	−16496.19	p=0.70253, q=3.17231	0 (P=1.00)	NA
	M8 (β and ω)	−16496.19	p0=0.99999, p=0.70253, q=3.17230, (p1=0.00001), w=5.42283	0 (P=1.00)	None

PAR, protease-activated receptor; lnL, the log-likelihood difference between the two models; 2ΔI, twice the log-likelihood difference between the 2 models; NA, not allowed; None, no positive selection sites identified.

Conservation of synteny in the PAR gene family

Based on the results shown in Fig. 7, we found that PAR1, PAR2, PAR3 are located on the same chromosome (F2R encoding PAR1, F2RL1 encoding PAR2 and F2RL2 encoding PAR3) and PAR4 is located on a different chromosome (F2RL3 encoding PAR4). For example, PAR1, PAR2 and PAR3 are located on chromosome 5 and PAR4 on chromosome 19 in humans, chimpanzees and orangutans. However, in teleosts, the chromosomal distribution of PARs differs from that observed in primates and birds. In this case, PAR1, PAR2 and PAR3 are not located on just 1 chromosome, but are distributed among different chromosomes. For example, in zebrafish, PAR1 is on chromosome 5, PAR2 is on chromosome 21 and PAR4 is on chromosome 14, while PAR3 is found on chromosomes 5 and 21. In the medaka, PAR1 is found on chromosome 9, PAR2 and PAR4 are found on chromosome 12 and PAR3 is found on chromosomes 9 and 12. Most of the PAR genes have 2 orthologous exons. However, there are exceptions in the PAR genes of many species (Table IV). The dog, dolphin, horse and macaque have 3 exons, while the marmoset and shrew have 4 exons in their PAR1 genes. The squirrel, ferret, megabat, medaka and lesser hedgehog have 3 exons, while the cat, wallaby and xenopus have 4 exons in their PAR2 genes. Megabat has three exons in its PAR3 gene. The bushbaby and gibbon have 1 exon in their PAR4 genes, while the orangutan has 4 exons and the ferret has 3 exons in their PAR4 genes.

Figure 7

Chromosomal distribution of the protease-activated receptor (PAR) family.

Table IV

The number of exons in the vertebrate PAR gene family.

Species	No. of exons
	in PAR1	in PAR2	in PAR3	in PAR4
Human	2	2	2	2
Bushbaby	2	2	2	1
Chimpanzee	2	2	2	2
Gibbon	2	2		1
Gorilla	2	2	2	2
Orangutan		2	2	4
Macaque	3	2	2	2
Marmoset	4	2	2	2
Mouse lemur			2
Alpaca	2	2
Cow	2	2	2	2
Cat	2	4	2
Dog	3	2		2
Dolphin	3	2		2
Ferret	2	3	2	3
Horse	3	1	2
Megabat	2	3	3
Pig	2	2	2	2
Panda		3		2
Shrew	4
Elephant		2
Hyrax		2
Wallaby		4
Lesser hedgehog		3
tenrec
Sloth	2	2
Tasmanian devil	2	2	2	2
Mouse	2	2	2	2
Rabbit		2
Rat		2	2	2
Squirrel	2	3
Opossum	2	2	2	2
Platypus		2	1
Chicken	2	2	2
Flycatcher	2		2
Chinese softshell	2	1		2
turtle
Armadillo	1
Xenopus		4	2
Coelacanth	3	2		2
Platyfish		2		2
Tilapia	2	2	2
Zebrafish 1	2	2	2	2
Zebrafish 2	4
Zebrafish 3	3
Zebrafish 4	4
Zebrafish 5	3
Medaka		3

PAR, protease-activated receptor.

Discussion

The PAR family (PAR1-4) members belong to the GPCR family and are found in vertebrates. Based on an analysis of a number of different species, in the present study, we found that specific-specific gene duplication did not occur in the vertebrate PAR genes and that the lineage-specific gene expansion of PARs was observed only in teleosts, indicating that gene duplication occurred prior to the separation of these vertebrate species and the teleost cluster evolved following a birth-and-death model (4,14,29).

The PAR genes were divided into 4 clades (PAR1, PAR2, PAR3 and PAR4) in the ML tree, suggesting that the PAR genes originated from 4 ancestors. In the ML tree, PAR1 and PAR2 are clustered into 1 subfamily and PAR3 and PAR4 are clustered into another subfamily, which suggested that the genes encoding PAR1 and PAR2 and the genes encoding PAR3 and PAR4 arose from a gene duplication event that was relatively recent in the history of the PAR family. In each of the 4 PAR clades, teleosts, primates and other mammalian species are separated from the other vertebrates, forming independent clusters. In the process of evolution, only PAR1 was found to have arisen from positive selection, while no evidence for positive selection was found with PAR2, PAR3 and PAR4. The identified positive selection sites of PAR1 are 65G and 76K (a signal peptide of approximately 20 amino acids in length is cleaved). These two amino acids are located on the N-terminus of PAR1. The thrombin cleavage site on the N-terminus of PAR1 is located between Arg41 and Ser42, thus, the positive selection sites are still present on the new N-terminus created following the activation of PAR1. These positive selection sites may affect the cleavage of PAR1 or the subsequent binding of the novel N-terminus with the body of the receptor. The results of positive selection analysis revealed that the PAR2, PAR3 and PAR4 lineages are highly conserved in vertebrates, while PAR1 is associated with environmental adaptation. The way in which PAR1 helps vertebrates adapt to their environment warrants further investigation. Based on the conservation of synteny analysis of the PAR gene family, we found that the PAR4 genes are located on a chromosome distinct from those of the genes encoding PAR1-3 in all vertebrates apart fromteleosts, in which the PARs genes are distributed on at least two chromosomes and PAR4 is not on a chromosome alone. This finding suggested that chromosomal fusion may have occurred during the evolution of teleosts. There are 2 exons in most of the vertebrate PARs, implying that they may be derived from the same ancestor and that exon rearrangement in PARs may not have occurred during the evolution of the PAR gene family in vertebrates. Functional divergences were found to have occurred between the PAR1, PAR2, PAR3 and PAR4 genes following analysis by the software, Diverge v2.0, implying that the PARs evolved individually in vertebrates. Different PARs have various functions.

Studies have demonstrated that PAR1 is involved in many functions, including the alteration of vascular tone and permeability, angiogenesis, and smooth muscle cell proliferation (30–32). As a receptor for serine proteases, following cleavage at a special site on the N-terminus of PAR2, activated PAR2 has been found to play a critical role in inflammation, immunity and angiogenesis (33). It has been demonstrated that PAR3 reduces the platelet response to thrombin, thereby providing protection from thrombosis, and mediates activated protein C anti-apoptotic signaling (34). PAR4 is involved in a number of functions, such as reducing platelet response to thrombin, providing protection from thrombosis in the cardiovascular system, regulating colonic nociception and inhibiting hypersensitivity in the nervous system (35,36). Interaction also exists among the members of the PAR family and may affect the individual functions of the PARs. PAR3 can affect the signaling of PAR4 and strengthen the activation of PAR4. In a study on mouse platelets, PAR3 was activated at low thrombin concentrations, while PAR4 was activated at high thrombin concentrations (37). It has been demonstrated that PAR3 serves as a co-factor for the activation of PAR4 at low thrombin concentrations (38). In contrast to PAR3 co-factoring with PAR4, PAR1 modulates the activity of PAR2 through a different mechanism. Following cleavage by thrombin, the N-terminus of PAR1 was found to unmask a tethered ligand domain, which binds in trans to activate PAR2 through an intermolecular liganding mechanism that elicits a special signaling response (39).

In conclusion, in this study, we examined the evolution of the PAR family by constructing phylogenetic trees, positive selection analysis, functional divergence analysis and conservation of synteny analysis. We concluded that the 4 members of the vertebrate PAR family originated from 4 ancestors and only PAR1 evolved by positive selection. In addition, in their separate evolution, there was no significant divergence of the functions of individual PARs. The evolutionary rates of the 4 PAR members were consistent. The findings of our study provide a theoretical background for the evolution of the PAR gene family in vertebrates.