Slc22 family organic anion and cation transporters (OATs, OCTs, and OCTNs) are transmembrane proteins expressed predominantly in kidney and liver. These proteins mediate the uptake or excretion of numerous physiologically (and pharmacologically) important compounds, and accordingly have been the focus of intensive study. Here we investigate the molecular phylogeny of the slc22 transporters, identifying homologs in Drosophila and C. elegans, several of which are developmentally regulated, as well as reporting the cloning of a novel human family member, UST6, expressed exclusively in liver in both embryo and adult. The latter helps define a subfamily within the OATs, which appears to have human- and rodent-specific members, raising potential issues with respect to the use of rodents as models for the transport of organic anions (which include many pharmaceuticals) in humans. Although this phylogenetic inference could not be made on the basis of sequence alignment, analysis of intron phasing suggests that the OAT, OCT, and OCTN lineages of the slc22 family formed after the divergence of vertebrates and invertebrates. Subsequently, these lineages expanded through independent tandem duplications to produce multiple gene pairs. After analyzing over 200 other transporter genes, we find such pairing to be relatively specific to vertebrate organic anion and cation transporters, suggesting selection for gene pairing operating within this family in particular. This might reflect a requirement for redundancy or broader substrate specificity in vertebrates (compared to invertebrates), due to their greater physiological complexity and thus potentially broader exposure to organic ions.
- Caenorhabditis elegans
- SLC gene pairs
the proximal tubule of the kidney contains highly avid systems for the secretion of organic anions and cations. Organic anion uptake at the basolateral surface (from blood) is driven by exchange for dicarboxylates, which are maintained at high intracellular concentrations by cellular metabolic processes and by the action of a Na+-dicarboxylate cotransporter (in turn driven by the Na+ gradient generated by the Na+-K+-ATPase). Apical efflux (into urine) occurs through mechanisms that are less well understood but that may involve exchange for luminal anions (in particular urate). In the case of organic cations, uptake appears to be membrane-potential driven and efflux mediated, in part, by exchange for luminal protons (Fig. 1; reviewed in Refs. 6, 7).
Although there is a long history of physiological investigation of this system, it is only relatively recently, with the identification of multiple genes encoding organic anion and cation transmembrane transporters (OATs, OCTs, and OCTNs), that its molecular underpinnings have begun to be studied (3, 10, 22). The organic anion transporters include OAT1–4 (2, 4, 23, 24, 35, 37, 40) and RST/URAT1 (8, 26); the organic cation transporters include OCT1–3 (15, 20, 28), OCTN1–3 (42–44, 48), and Flipt2/CT2 (9, 13). In addition, several homologs of as yet unknown function have been identified: ORCTL3 and ORCTL4 (27), Flipt1 (13), UST1 (34), UST3 (Takanaga H, Ohtsuki S, and Terasaki T, unpublished observations), and OAT5 (13, 38); the latter three are postulated to serve as organic anion transporters on the basis of sequence homology. Notwithstanding their nomenclature, OATs, OCTs, and OCTNs are encoded by related genes and indeed share some substrates and inhibitors (7, 19). They are accordingly considered members of the same family, termed slc22 (3, 22). Among the OATs, OAT1 and OAT3 have been localized to the basolateral surface of the proximal tubule and OAT4 and RST/URAT1 to its apical surface, while among the OCTs and OCTNs, OCT1 and OCT2 are basolateral and OCTN1 and OCTN 2 are apical (Fig. 1; reviewed in Refs. 10, 22). Interestingly, each of these pairs of basolateral or apical, anion or cation transporters is also paired in the genome (i.e., the encoding genes are immediate chromosomal neighbors; Ref. 12).
In addition to kidney, slc22 transporters are expressed largely in liver and mediate renal and hepatic excretion of many pharmaceuticals (reviewed in Refs. 3, 7, 10). However, it is not necessarily clear that the principal function of organic anion and cation transporters is detoxification/excretion, as they also transport numerous physiologically important substrates, including amino and fatty acids, neurotransmitters, nucleotides, prostaglandins, and miscellaneous metabolites, such as dicarboxylates and carnitine. Elucidation of the phylogeny (and thus evolutionary history) of the slc22 family will help delineate any additional role(s) of these transporters in organismal physiology. Accordingly, we have identified homologs in fly and worm, and we have used analysis of intron phasing to determine phylogenetic relationships that were not revealed by sequence alignments. We have also identified a novel human OAT-homolog, termed UST6, that helps define a subfamily within the OATs with apparently human and rodent-specific members. Finally, we find that gene pairing is rare among the invertebrate homologs, as well as among over 200 other solute transporters, suggesting selection for pairing specifically among vertebrate slc22 genes, possibly to subserve functional redundancy or broader substrate specificity. These results clarify the phylogeny of organic anion and cation transporters and will contribute to an understanding of the evolutionary physiology of organic ion transport.
MATERIALS AND METHODS
Sequence searches and analyses.
Fly and worm homologs of slc22 genes were identified through searches of the Drosophila melanogaster and Caenorhabditis elegans genomes (http://www.ncbi.nlm.nih.gov) using peptide sequences of various divergent slc22 family members (derived from GenBank; http://www.ncbi.nlm.nih.gov) as “queries.” Sequence alignments were performed with ClustalX (available at http://www-igbmc.u-strasbg.fr/BioInfo/ClustalX/Top.html) with the gap opening penalty set to 10 and the gap extension penalty set to either 1 or 0.1; in either case, the identical neighbor-joining dendrograms were obtained as visualized using TreeView (available from http://taxonomy.zoology.gla.ac.uk/rod/treeview.html). Percent-identity and similarity scores were calculated from sequence alignments using GeneDoc software (available from http://www.psc.edu/biomed/genedoc/). Exon-intron structures of the various slc22 transporter genes and their homologs were each individually determined in the identical manner, using the coding sequences as “queries” in BLAT searches (http://genome.ucsc.edu) of the most current versions of the corresponding genomes (July 2003 for human, October 2003 for mouse, June 2003 for rat, January 2003 for Drosophila, and May 2003 for C. elegans). Intron phasings were determined by comparison of the resulting sequence alignments to the translations of the coding sequences. Of over 300 intron boundaries examined, each conformed to the GT/AG rule except the 12th intron of the C. elegans gene K05F1.6, which has the following donor and acceptor sites, respectively (exon sequence capitalized, intron sequence in lower case): TTCAAGAAATgctagttgaa…ttattttcagGCGAGTCGTA.
To identify novel human members of the slc22 family, peptide sequences of various divergent family members (as above) were used as “queries” in searches of the draft human genome sequence (http://www.ensembl.org). The identical set of high-scoring “hits” was obtained in each case; these included all the known OATs, OCTs, and OCTNs, as well as additional sequences potentially encoding novel organic ion transporters. However, all but one of these “hits” appeared to represent pseudogenes: they mapped to loci that either lacked evidence of transcription (such as overlapping mRNAs or spliced ESTs) or that did not encode full-length transporters. Potential exons in the remaining “hit,” UST6, which had greatest homology to the putative OAT, UST3, were identified through inspection of the corresponding locus in the human genome browser (http://genome.cse.ucsc.edu) for ESTs, gene predictions, and mouse-human homology (i.e., evolutionarily conserved) regions, which could, together, encode a full-length OAT-like gene. Transmembrane topology was determined by TopPred (http://bioweb.pasteur.fr/seqanal/interfaces/toppred.html). Chromosomal locations of the mammalian OAT clusters were determined by BLAT searches of genomic sequences from the most current versions of the various genomic browsers. Human genes formally classified as solute carriers, i.e., carrying the “slc” designation, were identified by searches of the draft human genome sequence (http://genome.ucsc.edu).
Primers for amplification of UST6 cDNA fragments were designed from predicted exons using Primer3 (http://www-genome.wi.mit.edu) and consisted of the following combinations of forward and reverse primers: CTCCTCTTGGGGATCACTTG and AGGGCTCTGTATCTGGCTCA; CATACTTGATCATCGCTGCTGG and GTGCAGACATCACCAACTGG; ATGGTCTTACCTCCAGCTCG and GCAAAGTAAGGCCCCAAAA-AGG; CTCCAGTTGGTGATGTCTGC and AGGGCTCCCCCAATATTAGC; CTCCTGGCCAATTGTGTTGC and TTCCTGCCCTGCTTTTCTGG. Amplifications were performed from “Marathon” human liver cDNA (Clontech, Palo Alto, CA), and used HotStart Taq (Qiagen, Hilden, Germany) under reaction conditions recommended by the manufacturer. Cycle parameters were: denaturing at 95°C for 15 min; followed by 40 cycles of 94°C denaturing for 20 s, 60°C annealing for 20 s, and 72°C extension for 30 s. PCR products were visualized on 1.5% agarose gels stained with ethidium bromide.
The tissue distribution of UST6 was determined by RT-PCR. A panel of 16 human adult tissue cDNAs (source RNAs were pooled from multiple individuals in each case) and 8 human fetal tissue cDNAs (derived from pooled RNAs from multiple fetuses at 20–25 wk of gestation) was purchased from Clontech. Primers employed were CTCCTCTTGGGGATCACTTG and AGGGCTCTGTATCTGGCTCA. Control human G3PDH primers were provided by Clontech. Amplifications were performed with the identical conditions and cycle parameters as described above, with the exception that amplifications of G3PDH cDNA required only 30 cycles.
Similarly, expression patterns of fly slc22 homologs were examined by RT-PCR on Drosophila cDNAs derived from various adult and embryonic tissues (purchased from OriGene, Rockville, MD). Primers used were as follows: GCTTATGTCGGTGGGTAAAGTGG and GACTGGATGCTGCACACAATAGC (CG5592), GGACAA-TCTTAAACTGCCCTTGG and GCGAAAACAAAGAATAGCCTTGC (CG9317), TGATCCTCACACCATACCTGTCG and CACTACGTGCTACGCAAACAAGG (CG6006), TTTGGCTATATTC-ACCCCATTCG and GTGGCCAATAAGCTGAACACTCC (CG6600), TCACATTTTGATGGGCAGATACG and ATCCGGTAGTTTTCGGTTGAAGG (CG4630), GTTTGCCAACACCAGTGTCTACG and GGAAGTACGAGCATTTGCTGACC (CG3790), GCTGCTCCTCTTCACTTTGAACC and GTGATTAACTGCTCGTCGTCTCG (Orct), CGTGTGGGATCTCTTCTGCTACC and GACATCTGCCTTGGTCTTCTTGC (CG13610), and GGTCGAATAGGTTCCATTGTTGC and AAGGCAGTTGGAGAACAGCTACG (CG8654).
Control RP49 primers were provided by OriGene. PCR and electrophoresis conditions were as described above. Data on the developmental expression patterns of the various C. elegans genes were obtained from databases maintained at http://www.wormbase.org.
Cloning and Sequencing of UST6 cDNA.
Amplifications of overlapping UST6 cDNA fragments were electrophoresed on a 1% preparative agarose gel. The appropriate bands were excised, and the DNA within them was extracted using QIAquick columns (Qiagen). Approximately 100 ng of each PCR product was ligated to the pGEM-T vector (Promega, Madison, WI), and ligations were used to transform XL1-Blue competent cells (Stratagene, La Jolla, CA). Plasmids were purified from individual transformant colonies using QIAprep columns (Qiagen). Recombinants containing inserts of the expected size were identified by restriction digests with SacI and SphI (from New England Biolabs, Beverly, MA) and were sequenced at the University of California, San Diego, core sequencing facility, with an automated sequencer using the PRISM BigDye Terminator Cycle Sequencing Ready Reaction Kit version 3. Multiple independent clones were sequenced so as to control for misincorporation by Taq polymerase.
Homologs of slc22 transporters in fly and worm.
We searched the fly, worm, and yeast genomes for homologs of OATs, OCTs, and OCTNs, to extend our understanding of the evolutionary history of this gene family. We set a threshold of similarity of 30% for selection of possible homologs, since the most divergent of the mammalian slc22 transporters have percent amino acid similarities of ∼35%. For example, rat UST1 and OCT3 are 36% similar (and 25% identical); likewise, rat UST1 and OCTN2 are 36% similar (27% identical). From among “hits” meeting this criterion, we retained only sequences with conservation of the lineage-specific motifs STIVTEWN/DLVC and ELYPTVIR (34). Conservation was defined as having similar or identical amino acids at 7 of 11 positions (STIVTEWN/DLVC) or at 6 of 8 positions (ELYPTVIR); these thresholds were chosen based on the degree of conservation among the known slc22 transporters. Significantly, none of the hits with less than 30% similarity had conservation of these motifs.
Although yeast lacks homologs of the mammalian slc22 transporters (by the above criteria), we identified a total of four homologs in C. elegans and nine in D. melanogaster. Among these were the two previously characterized worm organic ion transporters, CeOAT1 (14) and CeOCT1 (47), which have been demonstrated to transport organic anions and cations respectively, and the fly gene, Orct, which had previously been noted to be homologous to mammalian slc22 transporters, but not functionally characterized (45). The remaining genes were ZK455.8a and K05F1.6 (worm), and CG13610, CG8654, CG3790, CG4630, CG6600, CG6006, CG9317, and CG5592 (fly) (Fig. 2). As noted in the introduction, most mammalian slc22 genes occur in the genome as pairs (12). By contrast, only Orct and CG13610 (which have adjacent locations on chromosome 3R: 20.1 Mb) are paired among the 13 invertebrate genes noted above (see Fig. 4; discussed below).
We obtained data on the developmental expression patterns of the putative worm transporters from Wormbase [which contains such data (based on microarray expression profiles) on virtually all of the genes in the C. elegans genome; http://www.wormbase.org]. These data suggest that three of the four worm genes, CeOAT1, CeOCT1, and K05F1.6, have lower expression in embryo and adult and higher expression in intermediate larval stages (Fig. 3A). Comprehensive surveys of gene expression are lacking for Drosophila; accordingly, we investigated the ontogeny of the putative fly transporters by performing RT-PCR on a panel of cDNAs derived from flies of different developmental stages (Fig. 3B). We were able to detect robust signals for all the fly genes except CG6006, for which specific PCR products could not be detected, possibly owing to low expression levels (not shown). Of the remaining fly genes, six (CG8654, CG3790, CG4630, CG6600, CG9317, and CG5592) were expressed largely in adult (in contrast to C. elegans) with onset of expression generally late in development (3rd instar or pupa). Among these, expression of CG3790 and CG9317 appeared to be restricted to the head in the adult, and that of CG5592 was present only in males. Interestingly, the two genes expressed at earlier developmental stages were the paired Orct and CG13610, both of which had onset of expression as early as the 4 h embryo. Expression of Orct persisted throughout the remainder of development, whereas CG13610 appeared to be temporarily downregulated between the 12 h embryo and 1st instar stages.
Phylogeny of the vertebrate and invertebrate organic ion transporters.
We aligned the peptide sequences of the fly, worm, and mammalian (human) organic ion transporters to generate a dendrogram delineating their phylogeny (Fig. 4). Although the resulting “family tree” of transporters suggests groupings of OAT-like, OCT-like, OCTN-like, and invertebrate genes, the earliest (most evolutionarily remote) branch points in the dendrogram are not well-supported by bootstrap values; i.e., there is uncertainty regarding their precise phylogenetic relationships (which is visually conveyed by the fact that the corresponding nodes are very close together on the tree). Specifically, it is not clear-cut whether the common ancestor of OATs, OCTs, and OCTNs existed before or after the divergence of vertebrates and invertebrates, which could have implications for any functional commonalities among these various genes.
To resolve these ambiguities, we determined the exon-intron structure, and thus the intron phasing, of each of the above genes as an independent measure of phylogenetic relatedness (Fig. 5). The rationale for this approach is that the phase (phase 0 = intron falls between codons, phase 1 = after the 1st base of a codon, phase 2 = after the 2nd) of a coding region intron (as opposed to one located in the 5′ or 3′ untranslated regions) tends to be evolutionarily conserved. This is because an intron phase change normally requires an insertion or deletion in the upstream flanking exon, which would result in disruption of the reading frame, unless there were a precisely compensating insertion or deletion in the downstream flanking exon. In other words, for a phase change to occur and not be deleterious, there would have to be simultaneous mutations at distinct genomic sites. Thus phase changes should be rare events, and the degree of similarity of phasing in homologous genes is likely to reflect the degree of their phylogenetic relatedness. Indeed, analysis of intron phasing has been used to support phylogenetic inferences (1, 32, 46). Although the theoretical possibility exists that introns having the identical position and phase in homologous genes need not have been derived from a common ancestor, but might have been independently acquired due to preferential insertion into “hot spots,” this is believed to be an infrequent occurrence (reviewed in Ref. 31).
Our results suggest that OATs, OCTs, and OCTNs, as well as the ORCTLs, are more closely related to one other than they are to any of the fly or worm slc22 homologs, or to the recently described human slc22 family members, the “fly-like putative transporters,” Flipt1 and Flipt2/CT2 (9, 13): the former all have related intron phasings beginning with 0212021 [with 12 appended in the case of the OCTNs to yield 9 coding region introns with the phases 021202112, 212 appended in the case of the OCTs, yielding 0212021212, and 22 in the case of the OATs (except OAT5, discussed below) and ORCTLs, yielding 021202122], while the latter have divergent intron phasings. Therefore, OATs, OCTs, OCTNs, and ORCTLs appear to be phylogenetically distinct from both the invertebrate homologs and the Flipts. Importantly, analysis of exon-intron structure led to unambiguous phylogenetic inferences where none could be obtained using standard methods of phylogenetic reconstruction based on sequence alignments. Among the invertebrate genes, however, intron phasing was not helpful in assigning phylogenetic relationships: with the exception of Orct and CG13610 (which both lack introns altogether), no indisputably similar intron phasings (identically phased introns occupying homologous sequence positions) could be identified, including in the functionally characterized worm transporters CeOAT1 and CeOCT1.
Molecular cloning of UST6, a putative novel OAT.
To further clarify the molecular phylogeny of the slc22 family we searched for novel family members in the draft human genome sequence. Using various known OATs, OCTs, and OCTNs as “queries” in BLAST searches, we identified a genomic region on chromosome 11q12.3 that appeared to encode a novel organic anion transporter, based on its high level of homology to OAT “queries,” in particular UST3 (see materials and methods). Numerous ESTs were found to map to this location, all of which (where specified) were derived from either fetal or adult liver. We subsequently identified potential exons through a combination of comparison to ESTs, gene predictions, and conceptual translations of subregions conserved between human and mouse. Primers were designed from these potential exons and used in RT-PCR reactions to amplify overlapping fragments from human liver cDNA. The resulting amplicons were cloned and their sequences assembled to generate the full-length coding sequence of the putative novel OAT-like gene (presented in Figs. 6 and 7), termed UST6 (GenBank Accession No. AY437532).
The UST6 cDNA sequence exactly matches the draft human genome with the exception of the transition A to G at position 1456 of the cDNA sequence, resulting in the substitution of valine for methionine at position 486 of the peptide sequence (boxed in Fig. 6). The polypeptides encoded by either variant of UST6 have the same predicted topology, 12 transmembrane domains (concordant with that of the other slc22 transporters; Refs. 33, 39), and the peptide sequences of OAT1, OAT3, and URAT1 also have a valine at this position (Fig. 7). These findings suggest that this sequence variation represents an authentic polymorphism rather than a sequencing or cloning artifact. Importantly, the single nucleotide polymorphism database, dbSNP, has an entry, rs6591771, submitted by the Human Genome Sequencing Center, Baylor College of Medicine, which corresponds to the variant that we have identified (note that the sequence of the complementary strand is presented in the database entry). However, whether this site is truly polymorphic can only be determined following the sequencing of UST6 from multiple humans.
UST6 comprises 9 exons and 8 introns (Fig. 6), with donor and acceptor sites conforming to the GT/AG rule in every instance (not shown). The phasing of the introns is 02120212; i.e., identical to that of OAT5 and differing from that of the other OATs only in lacking their terminal phase 2 introns. This intron phasing suggests multiple theoretical possibilities for alternatively spliced transcripts (joining exons bordered by introns of the same phase). However, as there are no symmetric exons (exons flanked by introns of the same phase), each such alternatively spliced transcript would require the deletion of at least 2 exons. UST3 and OAT5 appear to be the closest homologs of UST6, with peptide sequence similarities of 81% and 75%, respectively (compared with ∼50% for most of the other OATs) (Fig. 7). Moreover, just as UST3 and OAT5 appear to be liver specific (12, 13), UST6 expression is also apparently restricted to liver in both adult and fetus (Fig. 8).
Phylogeny of the mammalian OATs.
We noted earlier that OATs, OCTs, and OCTNs occur as pairs in the human genome (12). Among the OATs, the pairs are OAT1 and OAT3, OAT4 and URAT1, and OAT5 and UST3, all of which are located on adjacent segments of chromosome 11 at 11q12.3–11q13.1. UST6 is located between the OAT1/OAT3 and OAT5/UST3 gene pairs, with no other genes interposed among these five OATs (Fig. 9A). Thus this locus can now be seen to comprise a cluster, although the distance between OAT1 and OAT3 (8 kb) is much smaller than the distance to UST6 (148 kb), the next gene in the cluster. The OAT4/URAT1 gene pair is located ∼1.2 Mb further downstream, whereas the sole unpaired OAT family member, OAT2, is on a different chromosome. Of note, the recently published chimp genome (http://genome.cse.ucsc.edu) has very high-scoring matches (97–99% identity at the peptide level) to each of the above human sequences (not shown), although there are as yet no mRNA or EST sequences validating transcription from these loci.
We inspected the mouse and rat draft genome sequences for similar clustering of OAT genes. Rodent orthologs have been demonstrated for only some of the above genes: mouse and rat OAT1–3 (2, 21, 23, 24, 35, 37, 40) and the murine ortholog of URAT1, termed RST (26). The relative locations of these genes are indeed consistent with those of their human counterparts, with OAT1 and OAT3 paired and with RST occupying a nearby (although not adjoining) location (Fig. 9A); mouse or rat OAT2, as with human, is situated on a different chromosome than the other OATs. However, inspection of the rat (but not mouse) genomic locus syntenic to the human OAT cluster (http://genome.cse.ucsc.edu) also reveals the presence of two novel putative OATs, UST4r and UST5r (cloned from liver; Refs. 17, 18), as well as the previously described UST1r (34) (Fig. 9A).
We aligned the peptide sequences of these various OATs (along with those of representative OCTs and OCTNs) to construct a dendrogram representing their phylogeny (Fig. 9B). [In this instance, comparison of intron phasing was of limited utility, since the phasing of each of the human, mouse, and rat OATs (excepting UST6 and OAT5 as noted above) is identical (021202122).] This dendrogram reveals that the rat genes, UST1r, UST4r, and UST5r, and the human genes, UST6, UST3, and OAT5, together appear to constitute a distinct subfamily (which may be termed the USTs after the first identified member of this group; Ref. 34). (We attempted to identify orthologs for any of these genes in the mouse genome, but were unsuccessful, possibly because of as yet incomplete sequence information.) Since the rat and human USTs have homologous chromosomal locations (Fig. 9A), one might have expected that the former represent the rat orthologs of the latter. However, this expectation is not supported by the phylogeny depicted by the slc22 dendrogram (Fig. 9B), which suggests, rather, that the rat USTs are more closely related to one another than to any of the human USTs, and vice versa. (Note that, by contrast, the dendrogram does clearly reveal the orthologous relationships between the rodent and human versions of OAT1, 2, and 3.) The human and rat UST genes may have originated in independent gene duplications that occurred after the divergence of the rodent and human lineages. Alternatively, gene conversion may have occurred among these neighboring genes, resulting in paralogs seeming more similar than orthologs. At present there are no reports on the function of any of the six USTs, and it is not known what organic ions they may transport.
Gene pairing among solute transporters.
We sought to place our finding of slc22 gene pairing in context by inspecting the human genome (http://genome.cse.ucsc.edu) to determine the prevalence of pairing or clustering (defined as the occurrence of related transporters on adjoining chromosomal locations without any other genes interposed between them) among other solute carrier genes. Thirteen of 16 (81%) human OAT, OCT, and OCTN genes are paired or clustered (12), including 11 of the 13 (85%) formally classified as solute carriers (carrying the designation slc); OCTN1 and 2 (slc22a4 and 5), OAT1 and 3 (slc22a6 and 8), OAT4 and URAT1 (slc22a11 and 12), and ORCTL3 and 4 (slc22a13 and 14) are paired, and OCT1–3 (slc22a1–3) are clustered (Fig. 10). By contrast, fewer than 10% of the ∼200 other human slc genes are organized in this manner (Fig. 10).
These other paired/clustered transporters occur in five families. Interestingly, two of these families are involved in organic anion transport (although they are phylogenetically and mechanistically distinct from the OATs): slc21 (OATPs; 16), of which 4 of 9 (44%) occur in a cluster, and slc17 (although originally designated as phosphate transporters, family members have recently been demonstrated to function primarily in the transport of organic anions; 30), of which 4 of 7 (57%) occur in a cluster. The other three families containing pairs or clusters are slc14 (urea transporters; 36), the 2 members of which are paired, slc6 (amino acid and neurotransmitter transporters; Ref. 5), 4 of 13 (31%) of which occur as two pairs, and slc38 (sodium-coupled neutral amino acid transporters; Ref. 25), 3 of 6 (50%) of which occur in a cluster.
The role of slc22 organic anion and cation transporters (OATs, OCTs, and OCTNs) in the renal and hepatic excretion of xenobiotics [including such clinically important pharmaceuticals as β-lactam antibiotics, loop and thiazide diuretics, nonsteroidal anti-inflammatory drugs (NSAIDs), and various antivirals, anti-arrhythmics, and antihypertensives; 3, 7, 10] has been well-characterized. By contrast, the physiological role of these transporters (at the level of the whole organism) is less clear, although they are known to interact (at the molecular level) with numerous endogenous compounds, including amino and fatty acids, neurotransmitters, nucleotides, prostaglandins, dicarboxylates, and carnitine, among others. Elucidation of the evolutionary history of the slc22 family will help define the role of these transporters in organismal physiology. Therefore, we have identified slc22 homologs in Drosophila and C. elegans, as well as a novel human family member, UST6. We have used sequence alignment and analysis of intron phasing of these genes to determine their likely phylogenetic relationship to the known OATs, OCTs, and OCTNs. We have also investigated the prevalence of gene pairing among transporters, finding it to be relatively specific to vertebrate slc22 genes. These results clarify the phylogeny of the slc22 family and thus contribute to our understanding of the evolution of organic ion transport.
We previously noted that several mammalian slc22 transporters manifest transient embryonic expression in a manner distinct from their expression patterns in the adult, suggesting a role in development or differentiation (29). It may be salient in this regard that there are no clear slc22 homologs in the unicellular yeast and that the worm has less than half as many putative organic ion transporters as the morphologically more complex fly, despite the greater total number of genes in the former (http://genome.cse.ucsc.edu). The apparently developmentally regulated expression of several worm and fly slc22 homologs (Fig. 3) is also suggestive of a role for organic anion and cation transporters in development. Last, it is noteworthy that, although of unknown function, Orct was cloned from the lemming (lmg) locus of Drosophila, mutations in which result in apoptosis in imaginal tissues (45).
Our phylogenetic analysis suggests that OAT, OCT, and OCTN genes (although not necessarily Flipt1 and Flipt2/CT2) emerged (i.e., shared their most recent common ancestor) after the divergence of vertebrates and invertebrates (an important conclusion that was not apparent from the standard analysis based on alignment of coding sequences). This in turn may imply that functions shared among OATs, OCTs, and OCTNs (i.e., transport of organic ions) might not necessarily be conserved in the fly and worm homologs (although it should be noted here that the polypeptides encoded by the worm genes CeOAT1 and CeOCT1 have in fact been demonstrated to transport organic anions and cations, respectively; Refs. 14, 47). Conversely, functions shared between the Flipts (of which, Flipt2/CT2 has been shown to function as a carnitine transporter; Ref. 9) on the one hand, and OATs, OCTs, and OCTNs on the other might be particularly informative regarding “primitive” activities of these transporters that are conserved across a broad phylogenetic spectrum. This type of analysis may ultimately provide insights into what certain members of this large class of genes might be doing apart from handling the disposal of common drugs.
As noted earlier, UST6 helps define a subfamily within the OATs, the USTs, with apparently human- and rodent-specific members (Fig. 9B). Regardless of whether these genes originated in independent gene duplications or are the products of gene conversion, their species-specificity might reflect differing selective constraints on humans and rodents. Conceivably, for example, USTs might interact preferentially with xenobiotics (while the other OATs interact preferentially with endobiotics). As humans and rodents occupy different ecological niches, their xenobiotic exposure might be expected to be different, accounting for the species-specific evolution of the USTs. Functional characterization of the USTs is needed to address this hypothesis. In this regard, it is important to note, from a practical standpoint, that any physiological differences between humans and rodents in the transport of organic anions could have significant implications for the use of rats or mice as experimental surrogates for humans, in particular as models for human drug-handling capacity.
Finally, we show that the phenomenon of transporter gene pairing is relatively specific to the vertebrate slc22 family, as it is rare among the invertebrate homologs as well as among over 200 other solute carriers. In conjunction with the observation that pairing appears to be the result of recurrent tandem duplication (11, 12), this specificity suggests the existence of selective pressure for pair formation. Since pair members are functionally alike (Fig. 1), pairing may confer the benefits of redundancy or broader substrate specificity. After all, the greater physiological and behavioral complexity of the vertebrates (relative to the invertebrates) might be expected to lead, respectively, to exposure to a broader array of endogenous and exogenous organic ions, ultimately imposing a requirement for a more robust system for the transport of these substrates.
S. A. Eraly was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant K08-DK-064839, and S. K. Nigam was supported by National Institute of Child Health and Human Development Grant R01-HD-40011.
We acknowledge the excellent technical assistance of Ashwin Mudaliar and helpful discussions with Stan Mendoza.
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: S. K. Nigam, Dept. of Medicine, Univ. of California, San Diego, 9500 Gilman Drive, La Jolla, California 92093-0693 (E-mail:).
- Copyright © 2004 the American Physiological Society