Comparative analyses of lysophosphatidic acid receptor-mediated signaling

Abstract

Lysophosphatidic acid (LPA) is a bioactive lipid mediator that activates G protein-coupled LPA receptors to exert fundamental cellular functions. Six LPA receptor genes have been identified in vertebrates and are classified into two subfamilies, the endothelial differentiation genes (edg) and the non-edg family. Studies using genetically engineered mice, frogs, and zebrafish have demonstrated that LPA receptor-mediated signaling has biological, developmental, and pathophysiological functions. Computational analyses have also identified several amino acids (aa) critical for LPA recognition by human LPA receptors. This review focuses on the evolutionary aspects of LPA receptor-mediated signaling by comparing the aa sequences of vertebrate LPA receptors and LPA-producing enzymes; it also summarizes the LPA receptor-dependent effects commonly observed in mouse, frog, and fish.

Introduction

Lysophosphatidic acid (LPA) is an intercellular lipid mediator that exerts a wide variety of cellular effects through G protein-coupled LPA receptors [1–5]. Thus far, six LPA receptor subtypes (Lpar1–Lpar6) have been identified in various vertebrates including human, mouse, rat, chick, frog, zebrafish, and medaka (Figs. 1, 2; Table 1). In other vertebrates, these LPA receptor genes have been predicted by in silico analysis. In addition, genes for LPA signal-related enzymes, such as the LPA-producing enzyme autotaxin (ATX) and phosphatidic acid-preferring phospholipase A1 (PA-PLA1), and LPA-degrading PA phosphatases (PAPs), have also been isolated or predicted in these vertebrates. PA-PLA1 has two isoforms, α and β, which are encoded by genes for lipase member H (Liph) and I (Lipi), respectively [6]. LPA-producing enzymes that show low amino acid (aa) identities (approximately 28 %) to vertebrate LIPH/LIPI have been identified in Drosophila melanogaster. However, no report has demonstrated the existence of LPA receptors in invertebrates. These observations suggest that extracellular LPA signaling was important in vertebrate evolution and plays common biological roles in a wide variety of vertebrate taxa.

Table 1.

The GenBank accession numbers or EMBL protein IDs for sequences used in the multiple sequence alignment and phylogenetic analysis

Numbers in black and red indicate validated and predicted sequences in the database, respectively

Table 2.

Tissue distribution of LPA receptors in mouse and medaka

Receptor subtype	Species	Tissue distribution	References
LPA1	Mouse Medaka	Brain, lung, heart, testes, small intestine, spleen, skeletal muscle Brain, heart, gut, testes, spleen, skeletal muscle	[27] [25]
LPA2 LPA2a LPA2b	Mouse Medaka Medaka	Brain, lung, kidney, testis Gut, testis, ovary, spleen Brain, heart, gut, testis, ovary, spleen, skeletal muscle	[27, 74] [25] [25]
LPA3	Mouse Medaka	Lung, kidney, testes, small intestine Gut, testis, ovary	[27] [25]
LPA4	Mouse Medaka	Brain, heart, lung, kidney, muscle, ovary, thymus Brain, heart, gut, testis, spleen, skeletal muscle	[22] [25]
LPA5 LPA5b	Mouse Medaka	Small intestine, spleen, thymus Heart, kidney, gut, spleen	[75] [25]
LPA6	Mouse Medaka	Ubiquitous including brain Brain, heart, liver, kidney, gut, testis, ovary, spleen, skeletal muscle	[55, 74] [25]

Since the first LPA receptor gene Lpar1 was identified in 1996, many studies have explored the expressions of LPA receptor genes and the biological functions of LPA receptor-mediated signaling in vitro and in vivo [1, 2, 7–9] (Table 2). Over the last decade, the biological and pathophysiological roles of each LPA receptor have been clarified using genetically engineered mice or zebrafish as well as in human genetic studies (Table 3). For example, LPA₁-mediated signaling is involved in neurogenesis, brain formation, neuropathic pain initiation, and pulmonary fibrosis formation [10–12]. In a zebrafish model, autotaxin-LPA₁ signaling is involved in the blood vessel formation [13]. LPA₂ affects synaptic transmission and tumorigenesis, and LPA₃ is important for the implantation of oocytes [14, 15]. LPA₄ mediates blood and lymphatic vessel formation [16]. Like LPA₁, LPA₅ is involved in neuropathic pain [17]. Lpar6 is a gene associated with autosomal recessive hypotrichosis [18]. Furthermore, transcriptome analyses for LPA receptors, ATX and LIPI in human normal and malignant tissues have demonstrated that different cancer entities display distinct expression profiles of LPA_1~6, ATX, and LIPI [19]. For instances, colon carcinomas show higher expression of LPA₃ and LPA₅, whereas other LPA receptor subtypes and LPA-generating enzymes are expressed at low levels. Ewing’s sarcomas exhibit higher expression of LPA₄ and LIPI. Another report demonstrates up-regulation of LIPH in lung cancer tissues [20]. Therefore, distinct LPA signaling might play a different role in cancer development, depending on the cancer types.

Table 3.

LPA receptor-mediated functions revealed in genetically modified mouse, zebrafish, and frog

Receptor subtype	Species	Function	References
LPA1	Mouse Zebrafish	Brain development, psychiatric function, pain transmission, fibrosis, vascular formation Vascular formation	[1, 2, 39] [13, 43]
LPA2 LPA2a	Mouse Zebrafish	Synaptic transmission, vascular formation in concert with LPA1 Unknown	[14, 42]
LPA3	Mouse Zebrafish	Embryo implantation, spermatogenesis in concert with LPA1 and LPA2 Left–right asymmetry	[15, 47] [46]
LPA4	Mouse Zebrafish	Vasculogenesis, lymphogenesis, osteogenesis Vasculogenesis in concert with LPA1	[16] [13]
LPA5	Mouse Frog	Pain transmission Embryogenesis (?)	[17] [24]
LPA6	Mouse Frog	Hair development Forebrain development	[54] [56]

Although there are many excellent review articles describing the biological and pathophysiological roles of LPA signaling and LPA receptor-mediated signal transduction in mammals [1–3, 5, 11, 12, 19, 21–23], the effects of LPA signaling in non-mammals are not well known, and a comparative analysis of LPA receptor genes is necessary. Recent studies including ours have identified LPA receptor genes expressed in non-mammals, including frog, zebrafish, and medaka, and showed their expression profiles in medaka [13, 24, 25] (Table 2). Therefore, we have performed a comparative analysis of LPA receptors and discussed the evolution of LPA receptor-mediated signaling and its potential biological roles in non-mammals. To this end, we selected validated or predicted LPA receptor genes of human (Homo sapiens), mouse (Mus musculus), chicken (Gallus gallus), green sea turtle (Chelonia mydas), frog (Xenopus laevis and X. tropicalis), teleosts including medaka (Oryzias latipes), pufferfish (Fugu rubripes), and zebrafish (Danio rerio), and lamprey (Petromyzon marinus) (Figs. 1, 2, 3). Lamprey LPA_2a showed 65.4 and 63.9 % aa identities to human and zebrafish LPA₁, respectively, but 55.8 and 52.6 % identities to human and zebrafish LPA₂, respectively. Thus, we considered lamprey LPA_2a as a member of the LPA₁ family in this review. On the other hand, lamprey LPA₅ and LPA_5b had about 40 % aa identity to zebrafish LPA₄ or LPA₆, while they showed about 30 % aa identity to zebrafish LPA₅. However, as shown in the phylogenetic tree (Fig. 3b), lamprey LPA₅ and LPA_5b appeared to belong to the LPA₅ cluster, probably due to limited sequence information. Thus, we omitted lamprey LPA₅ and LPA_5b from the discussion.

Fig. 1.

Multiple sequence alignment for edg family LPA₁–LPA₃. The multiple sequence alignment was performed using the MAFFT web service (http://mafft.cbrc.jp/alignment/server/). The GenBank accession numbers for the LPA receptors used in this analysis are shown in Table 1. Red boxes indicate the seven putative transmembrane domains (TMD1–TMD7). Red arrowheads indicate the exon–intron boundary in the coding regions of the genes. Asterisks (*), colons (:), and periods (.) below each human LPA receptor sequence indicate no substitutions (identical), conserved substitutions, and semi-conserved substitutions across all animal species for each LPA receptor subtype, respectively. Asterisks (*), colons (:), and periods (.) in white on a black background below the LPA receptor sequences indicate no substitutions, conserved substitutions, and semi-conserved substitutions, respectively, across all animal species for LPA₁–LPA₃. Residues involved in LPA binding are shown in white on a black background. me medaka, pu pufferfish, ze zebrafish, xel Xenopus lavies, xet X. tropicalis, gr green sea turtle, ch chick, mo mouse, hu human

Fig. 2.

Multiple sequence alignment for non-edg family LPA₄–LPA₆. A multiple sequence alignment was performed using the MAFFT web service (http://mafft.cbrc.jp/alignment/server/). The GenBank accession numbers for the LPA receptors used in this analysis are shown in Table 1. Red boxes indicate the seven putative transmembrane domains (TMD1–TMD7). Red arrowheads indicate the exon–intron boundary in the coding region of the genes. Asterisks (*), colons (:), and periods (.) below each human LPA receptor sequence indicate no substitutions, conserved substitutions, and semi-conserved substitutions, respectively, across all animal species for each LPA receptor subtypes. Asterisks (*), colons (:), and periods (.) in white on a black background below the LPA receptor sequences indicate no substitutions, conserved substitutions, and semi-conserved substitutions, respectively, across all animal species for LPA₄–LPA₆. Residues involved in LPA binding are shown in white on a black background. me medaka, pu pufferfish, ze zebrafish, xel Xenopus lavies, xet Xenopus tropicalis, gr green sea turtle, ch chick, mo mouse, hu human

Fig. 3.

Phylogenetic trees and gene structures of the edg (a) and non-edg LPA receptors (b). For each family, closest outgroups [S1P receptors, GPR55, and platelet-activating factor receptors (PAFR)] were selected based on a preceding analysis by Yanagida and Ishii [69] using human LPA receptor genes. The amino acid sequences of LPA receptors of various vertebrates were aligned by MAFFT [70] with the auto option. The resulting alignments were subjected to PhyML [71] to infer phylogenetic trees based on the maximum likelihood (ML) method assuming the WAG [72] plus discrete Gamma model with four rate categories [73]. The GenBank accession numbers and EMBL protein ID numbers for the LPA receptors used in this analysis are shown in Table 1. The GenBank accession numbers for S1P receptors, GPR55, and PAFRs are shown in the parenthesis on the right of the receptor names. Bootstrap values are indicated at each node of the tree. On the right, each exon corresponding to the open reading frame is illustrated as a box. Dashed lines indicate no information for gene structures existing in the database. la lamprey, me medaka, pu pufferfish, ze zebrafish, xel Xenopus lavies, xet X. tropicalis, gr green sea turtle, mo mouse, hu human

LPA receptor genes

The first LPA receptor gene Lpar1 encoding LPA₁ was identified in the developing cerebral cortex of mice in 1996 [8]. The aa sequence showed 40.5 % identity to a product of the endothelial differentiation gene (edg)-1, which was later identified as a receptor for sphingosine 1-phosphate (S1P) [26]. Subsequent studies have further isolated Lpar2 and Lpar3, which encode LPA₂ and LPA₃, respectively, from mouse tissues and demonstrate that pairwise comparisons between LPA₁, LPA₂, and LPA₃ show over 50 % aa identity [27].

Although they mediate some biological effects of LPA, several cellular activities induced by LPA could not be explained by the activation of these LPA receptors. These findings implied the existence of additional LPA receptor(s), and in 2003, human Lpar4 was identified. It has been classified as a member of the purinergic receptor family and is distantly related to Lpar1–Lpar3 [28]. Subsequently, two additional LPA receptor genes, Lpar5 and Lpar6, have been discovered in mammals, and these three receptors (Lpar4, Lpar5, and Lpar6) have been termed the non-edg family LPA receptors [22].

The LPA receptor family is divided into the edg and non-edg family LPA receptors (Fig. 3), which together can account for most of the biological functions of LPA. Some other G protein-coupled receptor (GPCR) genes in mammals are thought to belong to the LPA receptor family, including GPR87 and GPR35 [29, 30]. However, further research is needed to confirm that these GPCRs function as LPA receptors. Thus, we will not focus on these yet unconfirmed genes in this review.

Edg family LPA receptor genes

Lpar1

A multiple protein sequence alignment of the LPA receptors of various vertebrates revealed that Lpar1 is highly conserved (Figs. 1, 3a). Human LPA₁ shared more than 88 % aa identity with medaka LPA₁ and 64.5 % aa identity even with lamprey LPA₁, indicating that ligand specificity and intracellular signal transduction pathways as well as LPA₁-dependent biological functions are common in vertebrates.

A computational analysis of edg family LPA receptors indicated several aa residues critical for LPA recognition and binding within the 3rd and 7th transmembrane domains (TMDs) [31]. Three residues identified by mutagenesis of human LPA₁ are important for the activation of LPA₁, specifically R3.28, Q3.29, and K7.36. For example, mutations of R3.28 or K7.36 to alanine result in changes in the efficacy and potency of LPA₁. A Q3.29 mutation to alanine also decreases ligand interaction. These three residues are conserved across all vertebrates examined, suggesting that they are commonly critical for the activation of LPA₁ in animal species (Fig. 1). Indeed, human, mouse, frog, zebrafish, and medaka LPA₁ can couple to G proteins and activate intracellular signaling in response to LPA exposure with a comparable degree of efficacy and potency in a heterologous expression system [13, 25, 32–34].

Another unique feature of LPA₁ sequences is the class I PDZ (PSD95/Dlg1/ZO-1)-binding motif consisting of SVV at the C-terminal end [35–37]. This motif sequence generally consists of X-(S/T)-X-(V/I/L)-COOH; it interacts with PDZ proteins and is involved in the regulation of receptor-mediated intracellular signaling. Mammalian LPA₁ interacts with RhoGEF or GIPC (GAIP-interacting protein, C-terminus), leading to the activation of the Rho pathway or intracellular trafficking of LPA₁. The finding that LPA₁ of all species except for lamprey LPA₁ and LPA_2a contains the SVV sequence suggests the existence of common signaling pathways and biological roles (Fig. 1). Lamprey Lpar2a contains the initiation and stop codons, whereas no stop codon is found in the predicted lamprey Lpar1. Thus, the precise function of the lamprey LPA₁ sequence remains unclear.

A recent report identified three amino acid residues, Y85, L87, and I325, which are essential for the proper folding and signal transduction of human LPA₁ in the endoplasmic reticulum [38]. These resides are also conserved in all vertebrates except for lamprey LPA_2a. Therefore, critical amino acid residues for ligand binding, receptor activation and folding, and the PDZ-binding motif are mostly conserved across animal species, strongly suggesting that the functions of LPA₁ are largely common in these animals.

Although Lpar1 is involved in brain development, psychiatric and sensory functions, and fibrosis in mice [2, 10, 39, 40] (Table 3), it is not known whether similar LPA₁-mediated effects are common in vertebrates. One biological function of Lpar1 commonly observed in mice and zebrafish is blood vessel formation. In Lpar1-deficient mice, a small fraction of infants showed frontal hematoma caused by abnormal blood vessel formation [41]. The incidence of hematoma appearance is increased in Lpar1/Lpar2-double knockout mice [42]. On the other hand, zebrafish embryos co-injected with morpholino antisense oligonucleotides for LPA₁ and/or LPA₄ have defects in vascular formation [13, 43]. These findings suggest that LPA₁-mediated signaling cooperates with other LPA receptor subtypes, leading to normal blood vessel formation in vertebrates. This idea is also supported by the findings that ATX plays a role in vascular formation in both mice and zebrafish [13, 44].

Lpar2

In the phylogenetic analysis, the Lpar2 cluster was clearly divided into two subclusters, one consisting of loci in teleosts and the other consisting of loci in other vertebrates (Fig. 3a). Teleost Lpar2 was further duplicated, generating Lpar2a and Lpar2b. Computational analyses and mutagenesis studies have identified several specific residues, R3.28, Q3.29, R5.38, and K7.36, which are critical for human LPA₂ activation [31]. The R3.28, R5.38, and K7.36 residues are conserved in the LPA₂ sequences of all species. Q3.29 is conserved in non-teleost LPA₂ and teleost LPA_2b, but not teleost LPA_2a, in which the corresponding aa is replaced by glycine (Fig. 1). Because mutations of Q3.29 to alanine or glutamic acid decrease human LPA₂ activation [31], teleost LPA_2a may show some unique interactions with LPA compared with teleost LPA_2b or non-teleost LPA₂. However, zebrafish LPA_2a responds to LPA exposure by secreting transforming growth factor-α (TGF-α) in a TGF-α shedding assay, with a similar degree of efficacy and potency as zebrafish LPA_2b [13]. Thus, the replacement of glutamine with glycine is unlikely to affect the recognition of LPA by LPA_2a.

Similar to LPA₁, human and mouse LPA₂ contain a PDZ-binding motif consisting of STL at the C-terminal region. This motif of human LPA₂ can interact with PDZ proteins, such as Na⁺–H⁺ exchanger regulatory factor 2 or RhoGEF, resulting in a change in the activation of intracellular signaling pathways [37, 45]. Although chicken and turtle LPA₂ also contain a similar PDZ-binding motif consisting of SSI or SSL, frog and teleost LPA_2a and LPA_2b have no such motif (Fig. 1). Thus, C-terminal region-mediated signaling of LPA₂ varies depending on animal species.

As mentioned above, mice lacking both Lpar1 and Lpar2, but not those lacking just Lpar2, show more severe frontal hematoma and loss of LPA responses that affect cortical development, suggesting a non-essential role for LPA₂ in mouse development [42] (Table 3). LPA₂ also plays a role in excitatory synaptic transmission [14] (Table 3). However, there is no direct evidence that Lpar2 has the same biological functions in other vertebrates.

Lpar3

Mutagenesis studies have identified several residues involved in human LPA₃ activation, including R3.28, Q3.29, W4.64, R5.38, and K7.35 [31]. LPA₃ of all vertebrates examined contained R3.28, Q3.29, and W4.64. R5.38, in which arginine was replaced by threonine, was also present in LPA₃ of most species, with the exception of medaka (Fig. 1). In the 7th TMD, both R7.36 and K7.35, which are important for LPA recognition by LPA₃, were observed in all animal species considered (Fig. 1). Therefore, LPA₃-mediated signaling is likely conserved across animal species. Indeed, stimulation of mouse and medaka LPA₃ results in the same cytoskeletal responses in a heterologous expression system [25].

Mouse LPA₃ is important for embryo implantation in the uterus [15] (Table 3). However, considering that most non-mammals are oviparous and never develop eggs within the mother, LPA₃ has distinct biological roles in non-mammals. In zebrafish, it regulates left–right asymmetry. This is based on the finding that knockdown of Lpar3 impaired calcium influxes in dorsal forerunner cells, leading to the inhibition of Kupffer’s vesicle formation and the disruption of organ left–right asymmetries [46] (Table 3). However, abnormal organ formation, which might be caused by disrupted left–right asymmetries, has not been demonstrated in mice lacking Lpar3 or Lpar1–Lpar3 [47]. Another potential biological role for Lpar3 is the regulation of male reproduction. Lpar3 is highly expressed in the testis of human, mouse, and medaka [25, 48, 49]. Indeed, a triple deletion of Lpar1–Lpar3 results in defects in germ cell survival and an increased prevalence of azoospermia in aging mice, although single deletions of Lpar3 have no effect [47] (Table 3). Thus, LPA₃ might co-operate with LPA₁ and LPA₂ to regulate the maintenance of reproductive organs. More detailed analyses of LPA₃-mediated effects in the reproduction systems of fish and other animal species are needed.

Non-edg family LPA receptor genes

Lpar4

Lpar4 was originally identified as the orphan GPCR GPR23/p2y9, which belongs to the purinergic receptor family, whose ligands are adenosine, ADP, or ATP. Ligand screening studies identified LPA as an agonist for human GPR23 [28]. In our analysis, the aa identity of human LPA₄ to the edg family LPA receptors was about 25 %. There were no critical residues for receptor activation corresponding to R3.28, Q3.29, or K7.36, which each exist in LPA₁ and LPA₂, and in LPA₄ of most animal species (Fig. 2). Computational analyses show that the polar head group of LPA interacts with LPA₄ at distinct regions compared with LPA₁–LPA₃ [50]. S3.28, a threonine in the second extracellular loop (EL2), and Y6.51 are critical residues for human LPA₄ activation. S3.28 and the threonine in EL2 were conserved across LPA₄ of all animal species, while Y6.51 was present in animal species except for frog, in which tyrosine was replaced by phenylalanine (Fig. 2). This observation indicates that LPA₄ has maintained LPA binding activity during evolution. Indeed, mammalian and teleost LPA₄ respond to LPA with cellular activation, such as cytoskeletal changes and TGF-α release in a heterologous expression system [13, 25, 51].

In mice, Lpar4 is involved in vasculogenesis, lymphogenesis, and osteogenesis [16] (Table 3). In zebrafish, Lpar4 is involved in vasculogenesis in concert with Lpar1 [13] (Table 3). Further detailed analyses are needed to clarify the biological roles of LPA₄-mediated signaling.

Lpar5

Computational analyses have identified four ligand recognition residues, R2.60, H4.64, R6.62, and R7.32, in human LPA₅, all of which are predicted to bind to the polar head group of LPA [52]. R2.60 also existed in LPA₅ of other vertebrate species examined (Fig. 2). However, H4.64 was replaced by phenylalanine and isoleucine in medaka and zebrafish LPA_5b, respectively (Fig. 2). R6.62 was replaced by lysine in chicken and turtle LPA₅ and zebrafish LPA_5a, and absent from medaka LPA_5b and zebrafish LPA_5b; these LPA_5b sequences were very short (Fig. 2). R7.32 was replaced by lysine in frog LPA_5X-1, asparagine in medaka LPA_5b, and phenylalanine in zebrafish LPA₅ (Fig. 2). Thus, such variation at critical aa residues across various animal species suggests the differential ability of LPA₅ to recognize and bind LPA, resulting in variation in LPA₅ activation. Indeed, zebrafish LPA_5a and LPA_5b, and medaka LPA_5b fail to show cellular responses after treatment with LPA, such as TGF-α release or actin rearrangement in a heterologous expression system, whereas mammalian LPA₅ is activated by LPA treatment [13, 25]. Alternatively, the action of LPA_5b may not be coupled to G_12/13 because the C-terminal regions are shorter than those of non-teleost LPA₅ sequences (Fig. 2).

LPA₅-dependent signaling plays an important role in pain transmission as evidenced by experiments using Lpar5-deficient mice [17] (Table 3). However, there are no reports of LPA₅-mediated biological functions in other vertebrates, although maternal Lpar5 transcripts are present in the animal pole of frog embryos [24]. The exploration of the biological roles of LPA₅ has just commenced.

Lpar6

A recent in silico study on human LPA₆ has identified several missense mutations leading to a shift in orientation of LPA at the binding site, which may affect LPA₆ activation [53]. These include S3T, D63V, G146R, I188F, E189K, D248Y, and L277P, which have been severely associated with abnormal hair development. LPA₆ of other vertebrates contained glutamine and glutamic acid corresponding to D63 and E189 of human LPA₆, respectively (Fig. 2). Isoleucine and leucine corresponding to I188 and L277 of human LPA₆, respectively, were also conserved in all LPA₆ except for zebrafish LPA_6b (Fig. 2). However, other aa residues varied among animal species. These analyses employ the docking assay using human LPA₆ and 1-acyl-LPA. Because LPA₆ preferentially binds 2-acyl LPA [5], other residues that interact with this type of LPA might exist. Nonetheless, medaka LPA₆ activation results in cytoskeletal changes similar to those observed for mammalian LPA₆, suggesting that LPA₆-mediated signaling is highly conserved during evolution [25].

Although Lpar6 plays an important role in mammalian hair development in concert with Liph [18, 54] (Table 3), most tissues seem to ubiquitously express Lpar6 in mammals and fish [25, 55]. Thus, LPA₆-mediated signaling might have additional fundamental and ubiquitous biological functions; this is partially supported by a recent finding that Lpar6 is involved in forebrain formation of X. lavies [56] (Table 3).

Phylogenetic analysis of LPA receptor genes

Phylogenetic analysis of non-edg LPA receptor genes

A unique feature of edg family LPA receptor genes is the insertion of an intron in the middle of the coding region for the 6th TMD [27]. The intron insertion within the coding region for the 6th TMD was found not only in mammals but also in all vertebrates examined (Figs. 1, 3a). This observation suggests that the intron insertion occurred in the ancestral edg sequence and has likely been evolutionarily preserved. However, another edg family consisting of the S1P receptor genes (S1pr1–S1pr5) does not contain an intron insertion within the open reading frame [57]. Thus, it is likely that the ancestral edg diverged into two classes, the edg family LPA receptor genes and the edg family S1P receptor genes, and the intron insertion appeared only in the edg family LPA receptor lineage during evolution.

In addition to the conserved intron insertion, human, mouse, chicken, and turtle Lpar1 genes contained an additional intron insertion within the coding region at the N-terminus (Figs. 1, 3a). In pufferfish Lpar1, two intron insertions are likely to be present within the coding region at the N-terminus. By contrast, lamprey Lpar1 and Lpar2a genes contained an intron insertion within the coding region at the C-terminus, although common features are not observed for the insertions in these two genes. However, because the sequences of turtle, pufferfish, and lamprey LPA receptor genes are only predicted by in silico analysis, the precise gene structures need to be determined.

Two Lpar2 genes, Lpar2a and Lpar2b, have been identified or predicted in medaka, zebrafish, and pufferfish, but not in mammals [13, 25, 40], indicating that they may be specific to teleosts. The structures of these teleost Lpar2a and Lpar2b differ from each other (Fig. 3a). In addition to the conserved intron insertion within the coding region for the 6th TMD, medaka and pufferfish Lpar2b had an additional intron insertion within the coding region for the 2nd TMD, while zebrafish Lpar2b contained an intron within the coding region for the 3rd TMD (Figs. 1, 3a). Pufferfish Lpar2b also contained an intron in the coding region for the C-terminus, while zebrafish Lpar2b contained two introns in the C-terminal coding region (Figs. 1, 3a). Likewise, the medaka and pufferfish Lpar2a genes included an intron within the coding region at the C-terminus (Figs. 1, 3a). However, shared features were not observed for the intron insertions in the two teleost Lpar2 genes, probably due to their high aa sequence diversities. Taken together, the two Lpar2 genes Lpar2a and Lpar2b might have undergone a duplication event in the ancestral teleost lineage, and exon usage might have evolved independently.

Phylogenetic analysis of non-edg LPA receptor genes

The medaka and pufferfish Lpar4 genes contained a unique insertion of a short intron consisting of about 80 and 270 bp within the coding region for the 5th TMD, respectively [25] (Gene ID: LOC101078478). This insertion did not exist in non-teleost or zebrafish Lpar4, or in other LPA receptor genes, indicating that it arose in the medaka and pufferfish lineage.

The Lpar5 clusters in the phylogenetic tree were more characteristic than other LPA receptor genes (Fig. 3b). During evolution, an ancestral Lpar5 gene diverged from the Lpar4 and Lpar6 clusters. Furthermore, the Lpar5 cluster was divided into two subclusters, one consisting of teleost loci and the other consisting of other vertebrate loci. Teleost Lpar5 was further duplicated, generating Lpar5a and Lpar5b. Identification of endogenous ligand or signaling pathways for teleost LPA_5a and LPA_5b will provide a better understanding of the evolutionary significance of Lpar5.

Evolutionary rate

In phylogenetic trees of LPA receptor genes (Fig. 3), the evolutionary rate greatly differs between different subtypes. This tree reveals a possible relationship between evolutionary rate of a gene and its tissue distribution. In LPA₁, the evolutionary distance between mammals and teleosts is 0.255 on this tree. On the other hand, the corresponding distance is 1.06 in LPA₃ (Fig. 3a). The former is more than four times slower than the latter in evolutionary rate. LPA₁ expression in brain is commonly observed in various vertebrates including human, mouse, rat, frog, zebrafish, and medaka (Table 2) [34, 43, 58]. In mouse, LPA₁ is known to be involved in brain development and functions, while LPA₃ is involved in reproductive functions (Table 3). The observation that evolutionary rate of brain-expressed genes is relatively slow is consistent with the finding by Kuma et al. [59] on protein kinase and immunoglobulin genes. They argued that many proteins are involved in complex biochemical networks in the brain or in central nervous system [60], and thus these proteins interact with a larger number of molecules than in other tissues. Accordingly, brain-expressed genes are under stronger functional constraints to maintain such interactions [61]. As a result, their evolutionary rates at the amino acid level remain low.

A similar tendency is observed in the non-edg LPA receptor family. LPA₄ and LPA₆, expressed in brain, are more slowly evolving than LPA₅ (Fig. 3b). As the functions and tissue distributions of LPA receptors in different species are determined more clearly in the future, it may become possible to interpret the changes in evolutionary rate or specific amino acid substitutions according to functional changes.

Phylogenetic analysis of genes for LPA-generating enzymes

Signaling LPA is extracellularly generated via the catalytic action of two structurally distinct lipases, ATX and PA-PLA1 [6, 62–65]. ATX was firstly identified as ectonucleotide pyrophosphatase/phosphodiesterase encoded by the Enpp2 gene, and later shown to exhibit higher lysophospholipase D activity, which mainly involves cleaving lysophospholipids to produce LPA. Interestingly, among Enpp family members, consisting of Enpp1–7, only ATX/Enpp2 shows lipase activity to produce LPA [62, 65]. Human ATX shows high aa identities to other vertebrate ATXs, and higher aa identities when catalytic domains located in the central region of ATX protein are compared (Fig. 4a). The central catalytic domain is responsible for lipase activity, although the N-terminal somatomedin B-like domain and C-terminal nuclease domain are cooperative for the activity [65]. Thus, it is suggested that lysophospholipase D activity of ATX is highly conserved across animal species. We found two possible ATX-like proteins in Caenorhabditis elegans in the database (accession numbers NP_001041085 and NP_001041087). However, these show about 15 % aa identity to a catalytic domain of human ATX, and are likely to be closer to ENPP1 and ENPP3, respectively.

Fig. 4.

Phylogenetic trees of ATX (a), and LIPH/LIPI (b). For each family, closest outgroups [ENPP, phosphatidylserine-specific PLA1 (PLA1A), hepatic triacylglycerol lipase (LIPC), endothelial lipase (LIPG), and lipoprotein lipase (LPL)] were selected based on preceding analyses [6, 62, 65]. The amino acid sequences of ATX, or LIPH and LIPI of various vertebrates were used to infer phylogenetic trees, as described in Fig. 3 legend. The GenBank accession numbers and EMBL protein ID numbers for ATX or LIPH and LIPI used in this analysis are shown in the parenthesis. Designation of LIPH and LIPI is based on the gene deposited in the database. Bootstrap values are indicated above each branch of the tree. The percentage aa identities shown on the right were determined by using the homology search tool of the software GENETYX-MAC^®. For ATX, full length (FL) and a catalytic domain (CD) corresponding to the region of aa 161–540 of human ATX were used for the analysis. fl fly, la lamprey, me medaka, pu pufferfish, ze zebrafish, xel Xenopus lavies, gr green sea turtle, mo mouse, hu human

Liph and Lipi belong to the lipase gene family consisting of phosphatidylserine-specific PLA1, hepatic lipase, lipoprotein lipase, endothelial lipase, pancreatic lipase, and pancreatic lipase-related protein [6]. However, there is no established theory on evolutionary relationship between Liph and Lipi. Holmes and Cox [66] discussed that the mammalian Lipi arose from a recent gene duplication event of an ancestral Liph-like gene during mammalian evolution, but the phylogenetic tree (Fig. 5 in Ref. [66]) does not support this theory in a straightforward way. On the other hand, Hesse et al. [67] performed a careful analysis using syntenic information and showed that the Lipi locus already existed in the common ancestor of mammals and teleosts; however, the syntenic conservation of teleost “Liph” is unclear. Phylogenetic reconstruction of Liph and Lipi is difficult due to a great difference in evolutionary rate between both isoforms. Generally, lineages with high evolutionary rate are often misestimated to have an ancient origin, and it is known that the maximum likelihood (ML) method is relatively robust in such a situation [68].

A phylogenetic tree based on the ML method (Fig. 4b) is easier to interpret than trees in the two above-mentioned papers. Assuming this tree topology and the syntenic information reported by Hesse et al. [67], a possible evolutionary scenario is: (1) Lipi is the ancestral locus in early bony fishes, (2) Liph was duplicated from Lipi in the lineage of ancestral amniote, (3) only in the mammalian lineage, the evolutionary rate of Lipi was greatly increased, (4) in the teleost lineage, an independent gene duplication occurred. The increase in evolutionary rate of mammalian Lipi may be due to the change in the expression pattern; mammalian Lipi is expressed only in testis, but avian Lipi is more broadly expressed [67], although there is a possibility of many other factors. Since the statistical support is not high in our tree, the possibility of ancient duplication between Liph and Lipi is not rejected.

All vertebrate lipases in the LIPH/LIPI cluster contain a catalytic triad consisting of three amino acids, serine, aspartic acid, and histidine, essential for catalytic activity common to the lipase gene family [6]. Unlike LPA receptors or ATX, two LIPH/LIPI-like proteins in D. melanogaster were found in the database (NP_573259 and NP_648652). NP_573259 possess a catalytic triad, whereas NP_648652 lacks serine. However, whether NP_573259 catalyzes PA to produce LPA remains unknown. Even if it is the case, it is unclear whether LPA produced in fly acts as an extracellular signaling mediator. Further comparative studies on LIPH/LIPI structures and functions would help us to understand the biological role of LPA or emergence of signaling LPA during evolution.

Conclusion

In the present study, we performed comparative analyses of LPA receptors across fish, frog, turtle, chicken, and mammals, and discussed the potentially conserved biological roles of LPA receptor-mediated signaling, although limited in vivo studies are available to assess its biological importance in non-mammals. We believe that further analyses that consider non-mammals will reveal more conserved biological roles of LPA signaling, particularly with respect to the development of innate behaviors common to many vertebrates. Recent genome editing technologies, widely applied to many animals including mouse and fish will facilitate the generation of LPA receptor gene deletion animals to explore the functions of LPA signaling.