Functional Complementation between a Novel Mammalian Polygenic Transport Complex and an Evolutionarily Ancient Organic Solute Transporter, OST (cid:1) - OST (cid:2) *

These studies identify an organic solute transporter (OST) that is generated when two novel gene products are co-expressed, namely human OST (cid:1) and OST (cid:2) or mouse OST (cid:1) and OST (cid:2) . The results also demonstrate that the mammalian proteins are functionally comple-mented by evolutionarily divergent Ost (cid:1) -Ost (cid:2) proteins recently identified in the little skate, Raja erinacea , even though the latter exhibit only 25–41% predicted amino acid identity with the mammalian proteins. Human, mouse, and skate OST (cid:1) proteins are predicted to contain seven transmembrane helices, whereas the OST (cid:2) sequences are predicted to have a single transmembrane helix. Human OST (cid:1) - OST (cid:2) and mouse Ost (cid:1) Ost (cid:2) cDNAs were cloned from liver mRNA, sequenced, expressed in Xenopus laevis oocytes, and tested for their ability to functionally complement the corresponding skate proteins by measuring transport of [ 3 H]estrone 3-sulfate. None of the proteins elicited a transport signal when expressed individually in oocytes; however, all nine OST (cid:1) -OST (cid:2) combinations ( i.e. then overnight. oocytes with the anti-FLAG M2 monoclonal antibody were incubated with an fluor 488 fragment of goat anti-mouse antibody in the 1 h (Molecular Probes; 2 mg/ml, diluted 1:200). Secondary antibody was removed by washing the oocytes with 1 (cid:3) PBS (cid:1) 1% Tween 20 three times for 5 min each, two times for 10 min each, and then overnight. Cells were imaged using a (cid:3) 10 objective on a Leica TCS-SP laser-scanning confocal microscope. Statistical Analysis— Kinetic data from experiments measuring uptake of radiolabeled substrate were fit to the Michaelis-Menten equa- tion by nonlinear least squares regression analysis. V max and K m values with standard errors were derived from these curves. Comparison of data measuring initial rates of uptake of radiolabeled substrates in the presence and absence of inhibitors was performed by unpaired Stu- dent’s t test and correlated to p (cid:4) 0.05.

must take up specific amounts of nutrients, metabolic precursors, inorganic ions, signaling molecules, and other macromolecules while also exporting signaling molecules, hormones, electrolytes, metabolic waste products, and xenobiotics. Recent studies have described some of the genes involved in these transport processes; however, it is clear that many other genes and gene products remain to be identified and characterized (1)(2)(3)(4)(5)(6)(7).
Using a comparative approach, a novel type of organic solute and steroid transporter was recently identified in the liver of an evolutionarily ancient vertebrate, the little skate Raja erinacea (8). In contrast to all other organic anion carriers, this skate transporter is generated by co-expression of two distinct and novel gene products, Ost␣ and Ost␤. 1 Substrates for this multispecific transporter include estrone 3-sulfate, taurocholate, digoxin, and prostaglandin E 2 . Interestingly, the overall predicted membrane topology of skate Ost␣-Ost␤ is similar to that of the heterodimeric sensory rhodopsins, suggesting that Ost␣-Ost␤ may have evolved from an ancestral rhodopsin-like molecule but has acquired the ability to transport steroids and eicosanoids, compounds that also function as ligands for some G-protein-coupled receptors.
Initially, Ost␣ and Ost␤ orthologues were not identified in the human genome or in any other sequenced genomes, indicating that these genes might be specific to marine elasmobranchs. However, sequences for hypothetical human and mouse proteins have recently been entered into the data bases that exhibit 25-41% predicted amino acid sequence identity with skate Ost␣ and Ost␤ (see "Results"). The present study tested whether these mammalian genes are expressed, whether they encode for orthologues of the skate gene products, and if so, whether they functionally complement the transport activity of one another. Cloning Strategy-A multistep RT-PCR-based strategy was employed to obtain cDNAs for the putative open reading frames (ORFs) of hypothetical proteins OST␣ (GenBank TM /EBI accession number CAC51162), OST␤ (GenBank TM /EBI accession number XP_058693), and mOST␤ (GenBank TM /EBI accession number XP_134984/ AY279396). Oligonucleotide primers were designed to generate ORFs for each gene based on the cDNA sequences listed on GenBank TM (Table  I). RT-PCR products of predicted sizes were created for putative hOST␣, hOST␤, and mOst␤ using either human or mouse liver poly(A) ϩ RNA as a template (Clontech). The reaction products were isolated via agarose gel electrophoresis and ligated into the pCR-II TOPO vector (Invitrogen) utilizing Invitrogen's TOPO TA cloning kit. The resulting plasmids were used to transform DH5␣ Escherichia coli, from which positive clones were selected for ampicillin resistance and blue/white staining. Upon isolation and identification, a single clone expressing the new plasmid was used as a template in a PCR with primers designed to generate a final product flanked by two unique restriction sites (Table I). Following isolation and enzymatic digestion, the cDNAs with sticky ends were directionally ligated into the pSP64 poly(A) vector. These plasmids then were used to transform DH5␣ E. coli, and positive clones were selected for ampicillin resistance. Plasmid purifications were completed using Promega's DNA Wizard kit.

Materials and Animals-[G-
The cDNA for the hypothetical mOst␣ (GenBank TM /EBI accession numbers BC024441.1 and AAH25912) was obtained from the American Type Culture Collection (ATCC #6994476). This clone was used as a template in a PCR to generate the ORF for GenBank™ sequence BC024441.1. The primers were designed to add a HindIII site to the 5Ј end and an XbaI site to the 3Ј end of the PCR product (Table I). The PCR product was purified by agarose gel electrophoresis, and the band of expected size (1060 bp) was excised and enzymatically digested with HindIII and XbaI. The resulting cDNA sequence was then directionally cloned into the pSP64 poly(A) vector and used to transform DH5␣ E. coli. A positive clone was selected, and the plasmid was isolated with Promega's DNA Wizard miniprep kit.
Final clones for all four genes were sequenced in both directions with a series of specific oligonucleotide primers at the Mount Desert Island Biological Laboratory DNA Sequencing Core (Salsbury Cove, ME) to confirm sequence identity. The plasmids were then linearized with EcoRI and used to synthesize cRNA via Ambion's mMessage mMachine kit for injection into Xenopus oocytes.
Real Time Quantitative PCR to Determine Tissue Distribution and Expression Levels of hOST␣ and hOST␤ in 19 Human Tissue cDNAs-Human tissue cDNAs were purchased from Origene (human Sure-RACE TM panel). Synthetic oligonucleotide primers were designed to amplify portions of hOST␣, hOST␤, and human ␤-actin for use in quantitative real time PCR (Table I). Reactions were conducted and analyzed on a Roto-Gene 2000 real time light cycler from Corbett Research (Phenix Corporation, Hayward, CA). Qiagen's QuantiTect Sybr Green quantitative RT-PCR kit was used for PCR analysis. Expression levels are reported as a ratio to ␤-actin within each tissue examined.
Xenopus Oocyte Preparation, Microinjection, and Transport Assays-Isolation of Xenopus oocytes was performed as described by Goldin (9) and employed previously in our laboratory (10 -12). Stage V and VI defolliculated oocytes were selected and incubated at 18°C in modified Barth's solution (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO 3 , 0.82 mM MgSO 4 , 0.33 mM Ca(NO 3 ) 2 , 0.41 mM CaCl 2 , and 20 mM HEPES-Tris, pH 7.5) supplemented with penicillin (100 units/ml) and streptomycin (100 g/ml). The oocyte medium was changed daily, and healthy oocytes, defined as those with a clean brown animal pole and a distinct equator line, were selected for experiments. Transport studies were performed as described previously (10,11).
DNA Sequence and Hydropathy Analysis-Double-stranded cDNA clones were sequenced in both directions at the Mount Desert Island Biological Laboratory DNA Sequencing Core. The full-length sequences for all clones were obtained using synthetic oligonucleotide primers. Sequence analysis was performed with the DNA and protein sequence analysis program Lasergene from DNAStar Inc. (Madison, WI). Membrane topology and putative membrane-spanning domains were determined by hidden Markov model analysis (www.enzim.hu/hmmtop) and by Kyte Doolittle hydropathy analysis (13). The presence of possible signal peptides was evaluated with the SignalP V1.1 program (www.cbs.dtu.dk/services/).
Synthesis of FLAG Epitope-tagged Skate and Human OST␣ and of c-MYC-tagged Skate and Human OST␤-The FLAG epitope (DYKD-DDDK) was added in-frame to the 3Ј end of skate and human OST␣ cDNA, and the c-MYC epitope (EQKLISEEDL) was added to the 3Ј end of skate and human OST␤ cDNA. Constructs were created via PCR using skate and human OST␣ and OST␤ cDNA as templates with oligonucleotide primers containing the FLAG or c-MYC sequences flanked by unique restriction sites. PCR products of predicted size were identified by agarose gel electrophoresis, excised, and purified with Qiagen's agarose gel purification kit. The isolated fragments were subcloned into the pSP64 poly(A) vector, and that vector was used to transform DH5␣ E. coli. Positive clones were grown and plasmid DNA isolated using Promega's DNA Wizard plasmid isolation kit. After restriction digestion and sequencing to establish clone identity, cRNA was prepared for each of the four constructs using Ambion's mMessage mMachine kit.
Immunofluorescence Labeling of Intact Oocytes-Intact Xenopus oocytes on day 3 after cRNA injection were fixed in methanol:acetone (1:1) for 10 min on ice, followed by four washes of 5 min each at room temperature and an overnight wash at 4°C in antibody dilution buffer (0.01 M PBS ϩ 0.05% Tween 20, 1% bovine serum albumin, 1% normal goat serum, and 0.01% sodium azide). Oocytes were washed at room temperature with 1ϫ PBS ϩ 1% Tween 20 for 1 h with 10-min buffer changes followed by a 15-min incubation in blocking solution (3% bovine serum albumin in antibody dilution buffer). Oocytes were then washed with 1ϫ PBS ϩ 1% Tween 20 for another 1 h with 10-min buffer changes followed by an overnight wash at 4°C. The oocytes were incubated with either anti-FLAG M2 monoclonal antibody (Sigma; 4.9 mg/ml, diluted 1:200 with antibody dilution buffer) or monoclonal antic-MYC Cy3-conjugated antibody (Sigma; 1.2 mg/ml, diluted 1:100 with antibody dilution buffer) for 1 h. To remove excess antibody, oocytes were washed in 1ϫ PBS ϩ 1% Tween 20 three times for 5 min each, two times for 10 min each, and then overnight. The oocytes labeled with the anti-FLAG M2 monoclonal antibody were incubated with an Alexa fluor 488 F(abЈ) 2 fragment of goat anti-mouse antibody in the dark for 1 h (Molecular Probes; 2 mg/ml, diluted 1:200). Secondary antibody was removed by washing the oocytes with 1ϫ PBS ϩ 1% Tween 20 three times for 5 min each, two times for 10 min each, and then overnight. Cells were imaged using a ϫ10 objective on a Leica TCS-SP laserscanning confocal microscope.
Statistical Analysis-Kinetic data from experiments measuring uptake of radiolabeled substrate were fit to the Michaelis-Menten equation by nonlinear least squares regression analysis. V max and K m values with standard errors were derived from these curves. Comparison of data measuring initial rates of uptake of radiolabeled substrates in the presence and absence of inhibitors was performed by unpaired Student's t test and correlated to p Ͻ 0.05.

Cloning Putative Human OST␣ and OST␤ and Mouse
Ost␣ and Ost␤-Hypothetical human and mouse proteins (Gen-Bank TM /EBI accession numbers CAC51162 and AAH25912, respectively) that were recently added to the data bases exhibit 41% predicted amino acid identity with skate Ost␣ and share 83% amino acid identity with each other (Fig. 1). Because of the many conserved amino acid substitutions, the extent of amino acid similarity is ϳ70% between skate Ost␣ and these putative mammalian orthologues and 89% between the hypothetical mouse and human proteins (Fig. 1). Interestingly, all three deduced amino acid sequences share a highly unusual cluster of cysteine residues in a predicted hydrophilic cytosolic loop between transmembrane (TM) domains 3 and 4 ( Fig. 1). This relatively high overall amino acid identity along with the conserved cysteine cluster in the human, mouse, and skate gene products suggests that they are functional orthologues.
Likewise, possible human and mouse orthologues of skate Ost␤ were recently entered into the data bases, but these sequences exhibit only 25-29% predicted amino acid identity with skate Ost␤ (Fig. 2). The hypothetical human protein with GenBank TM sequence XP_058693 and the hypothetical mouse protein with GenBank TM sequence XP_134984/AY279396 exhibit 25 and 29% predicted amino acid identity with skate Ost␤, respectively (Fig. 2). These hypothetical human and mouse proteins exhibit 62.5% amino acid identity with each other, both encoding for proteins containing 128 amino acids with a putative single TM domain (Fig. 2).
An RT-PCR-based strategy was employed to obtain cDNAs for the predicted ORFs of putative human OST␣ and OST␤ using human liver mRNA and of putative mouse Ost␤ using mouse liver mRNA. The cDNA for the putative mouse Ost␣ was obtained from the American Type Culture Collection (ATCC #6994476). Sequence analysis of the synthesized cDNAs for the four genes indicated that both mouse clones were identical to GenBank TM sequences, whereas the human sequences varied by a single nucleotide in the reading frame of each gene. Both differences in the human gene sequences result in amino acid substitutions (Figs. 1 and 2; GenBank TM /EBI accession numbers AY194243 and AY194242). It is unclear whether the observed differences are due to polymorphisms or whether they result from mutations introduced during PCR. They are unlikely to be sequence artifacts as both occur in regions of strong sequence data. When compared with the published human genomic DNA data base (www.ncbi.nlm.hih.gov:80/BLAST/), our human OST␣ sequence matches perfectly, whereas the GenBank TM cDNA sequence for the protein with GenBank TM / EBI accession number CAC51162 contains an adenine at the base in question rather than a guanine (position 604 of the ORF). This difference predicts a valine at position 202 of human OST␣ (circled in Fig. 1), whereas the GenBank TM sequence CAC51162 predicts an isoleucine. In contrast, our human OST␤ sequence differs from the genomic DNA sequence at the base in question, containing a cytosine at nucleotide 197 of the ORF instead of an adenine. As a result, the sequence for human OST␤ obtained in the present study predicts a threonine at position 66 (circled in Fig. 2), whereas the GenBank TM sequence XP_058693 predicts a lysine. Because the human liver poly(A) ϩ RNA used as a template for RT-PCR in the present study was pooled from four different people, the reaction product containing the inserts used for cloning human OST␤ was sequenced. Upon analysis it was determined that an equal distribution of adenine and cytosine occurred at this nucleotide position (data not shown), indicating that this may be a naturally existing polymorphism.
A search of the human genomic data base (www.ncbi.nlm. hih.gov:80/BLAST/) revealed that human OST␣ is located on chromosome 3 and is coded by nine exons in the 3q29 region, whereas human OST␤ is on chromosome 15 and is coded by four exons located in the 15q21 region. Mouse Ost␣ is located on chromosome 16 and is coded by nine exons at 16B2, whereas mouse Ost␤ is found on chromosome 9 and is coded by four exons located at 9C.
The Human and Mouse Proteins Function as Organic Solute Transporters-To assess whether the human and mouse proteins function as organic solute transporters, uptake of [ 3 H]estrone 3-sulfate was measured in X. laevis oocytes injected with cRNA synthesized from the human, mouse, or skate genes. As expected, co-expression of skate Ost␣ and Ost␤ was required to generate transport activity (Fig. 3A) (8). When the putative human OST␣ and mouse OST␣ were expressed individually in oocytes, no transport activity was detected (Fig. 3A); however, when these proteins were co-expressed with skate Ost␤, a strong transport signal was obtained (Fig. 3A), indicating that human OST␣ and mouse OST␣ can functionally complement the corresponding skate protein. Likewise, human OST␤ and mouse OST␤ did not induce transport activity when expressed individually in oocytes but generated a functional transporter  2. OST␤ amino acid alignments. The deduced amino acid sequences for human, mouse, and skate OST␤ were aligned using DNAStar's MegAlign computer program running the Jotun Hein algorithm. Amino acid identity is displayed with black shading, and the predicted TM domain is boxed. The cDNA sequence obtained in the present study for human OST␤ predicts a threonine at position 66 (circled) (GenBank TM /EBI accession number AY194242), whereas the GenBank TM sequence XP_058693 predicts a lysine at this position. This change results from a single nucleotide difference at base 197 of the ORF, where GenBank TM sequence AY194242 contains a cytosine and GenBank TM sequence XP_058693 contains an adenine. The RXR motif is located at amino acid position 61 in the mouse and 92 in the skate. when co-expressed with skate Ost␣ (Fig. 3B). Thus, the human and mouse proteins not only are functional orthologues of the skate proteins but also are able to complement each other across species. Moreover, co-expression of the two human (OST␣-OST␤) or the two mouse (mOST␣-mOST␤) proteins generated a very strong transport signal, as did the human-mouse ␣-␤ pairs (Fig. 3C).
Characteristics of Human OST␣-OST␤-and Mouse OST␣-OST␤-mediated Transport-Oocytes injected with human OST␣ and OST␤ cRNA (1 ng each) or with mouse Ost␣ and Ost␤ cRNA (1 ng each) were able to transport taurocholate, estrone 3-sulfate, digoxin, and prostaglandin E 2 but not estradiol 17␤-D-glucuronide or p-aminohippurate (Fig. 4), indicating that this transport system is multispecific and that it may participate in cellular uptake of conjugated steroids and eicosanoids. This substrate profile is similar to that of skate Ost␣-Ost␤ (8).
The skate, mouse, and human transporters shared a number of other features as well. Transport was sodium-independent (Fig. 5), saturable (Fig. 6), and inhibited by bile salts, steroids, and other organic anions (Table II). Replacement of the NaCl in the oocyte incubation medium with either choline chloride or lithium chloride had no effect on estrone 3-sulfate uptake (Fig.  5), indicating that transport is not coupled to the sodium electrochemical gradient. Initial rates of estrone 3-sulfate uptake into human OST␣-OST␤-or mouse OST␣-OST␤-expressing oocytes were saturable, although the apparent Michaelis con-stants (K m ) were relatively high (320 Ϯ 30 M and 290 Ϯ 24 M, respectively; Fig. 6). The K m for estrone 3-sulfate uptake by skate Ost␣-Ost␤ is lower (85 M) (8). Uptake of [ 3 H]estrone 3-sulfate in hOST␣-hOST␤-and mOST␣-mOST␤-expressing oocytes was inhibited by a variety of bile salts, steroids, and other organic anions (Table II). As reported previously for skate Ost␣-Ost␤ (8), [ 3 H]estrone 3-sulfate uptake was inhibited by sulfated steroids, including lithocholic acid sulfate and taurolithocholic acid sulfate (Table II).
Tissue Distribution of Human OST␣ and OST␤ mRNA-OST␣ and OST␤ mRNA levels were measured in 19 human tissue cDNAs and were expressed relative to ␤-actin mRNA levels using quantitative real time PCR analysis. The results revealed that OST␣ and OST␤ are widely expressed in human tissues (Fig. 7). Tissues that had high levels of OST␣ mRNA generally also had high levels of OST␤ mRNA, indicating coexpression of these genes. Relatively high levels of both mRNAs were found in testis, colon, liver, small intestine, kidney, ovary, and adrenal gland (Fig. 7A); and lower levels were measured in heart, lung, brain, pituitary, thyroid gland, uterus, prostate, mammary gland, and fat (Fig. 7B). The mRNA for OST␣ and OST␤ was below our limit of detection in skeletal muscle and peripheral blood leukocytes (data not shown).
Trafficking of OST␣ and OST␤ to the Plasma Membrane-To gain insight into the mechanism by which OST␣ and OST␤ interact to generate a functional transporter, epitope-tagged FIG. 3. The human and mouse OST proteins functionally complement the corresponding skate Ost proteins. A, human OST␣ (hOST␣) and mouse OST␣ (mOST␣) functionally complement skate Ost␣ (sOst␣). B, human OST␤ (hOST␤) and mouse OST␤ (mOST␤) functionally complement skate Ost␤ (sOst␤). C, human OST␣-␤ (hOST␣-hOST␤) and mouse OST␣-␤ (mOST␣-mOST␤) are transport-competent. Oocytes were microinjected with 2 ng of cRNA when individual genes were expressed (sOst␣, sOst␤, hOST␣, hOST␤, mOst␣, or mOst␤) and with 1 ng of each cRNA when pairs of genes were expressed (sOst␣-sOst␤, hOST␣-sOst␤, mOst␣-sOst␤, sOst␣-hOST␤, sOst␣-mOst␤, hOST␣-hOST␤, mOst␣-mOst␤, hOST␣-mOst␤, or mOst␣-hOST␤). Oocytes were cultured for 3 days, and uptake of 50 nM [ 3 H]estrone 3-sulfate was measured for 1 h at 25°C. The values are means Ϯ S.E., n ϭ 3; each of the three separate experiments was performed in triplicate. constructs were synthesized, and these constructs were tested for functional activity (Fig. 8) and cellular localization (Fig. 9) in Xenopus oocytes. The FLAG epitope was added in-frame to the 3Ј end of skate and human OST␣ cDNA, and the c-MYC epitope was added to the 3Ј end of skate and human OST␤ cDNA. As expected, there was no transport activity when the individual epitope-tagged proteins were expressed in oocytes; however, when human or skate ␣-FLAG and ␤-c-MYC were co-expressed, there was strong estrone 3-sulfate transport activity (Fig. 8). Interestingly, immunofluorescence analysis revealed that OST␣-FLAG and OST␤-c-MYC were able to reach the plasma membrane when expressed either individually or in pairs (Fig. 9). Thus, the inability of the individual proteins to generate transport activity does not appear to be due to impaired trafficking to the cell surface. DISCUSSION The present results identify a novel mammalian organic solute transporter for which the mRNA is widely distributed and highly expressed in human tissues. This transporter is unique among mammalian organic anion transporters in that it is generated when two distinct gene products are co-expressed, namely a putative 7-TM domain membrane protein, OST␣, and a smaller, single-TM domain polypeptide, OST␤. Interestingly, the human and mouse proteins were able to complement each other as well as those from an evolutionarily ancient vertebrate, the little skate, indicating a high degree of conservation throughout evolution. Although the physiological functions of this transporter are not known, its broad tissue expression and its ability to transport steroids and prostaglandin E 2 suggest an important role in cellular functions.
Identification of OST␣ and OST␤ was made possible by the recent cloning of orthologous genes from the liver of an evolutionarily ancient marine vertebrate, the little skate (8). Surprisingly, when skate Ost␣ and Ost␤ genes were identified in 2001, comparable genes were not present in any of the sequenced genomes, including the human genome, suggesting that they may be unique to elasmobranchs (8). However, because the human genome remains in draft form, many genes have not yet been discovered.
Hypothetical human and mouse genes that recently were entered into the data bases are predicted to encode proteins exhibiting 25-41% amino acid identity with skate Ost␣ and Ost␤ ( Figs. 1 and 2). Although this level of amino acid identity is low, it is not insignificant given an evolutionary distance of 200 million years between skates and humans. Thus, despite the modest level of amino acid identity, these skate and mammalian proteins may be carrying out the same biological functions. To test this possibility, the present study assessed whether these hypothetical human and mouse genes are expressed in human and mouse liver, and if so, whether they function as organic solute transporters. Our results demonstrate not only that human OST␣-OST␤ and mouse Ost␣-Ost␤ are orthologues of skate Ost␣-Ost␤ but also that these proteins are able to functionally complement one another across species. That is, the 7-TM domain OST␣ proteins from humans, mice, or skates are able to partner with any of the OST␤ proteins from these three species to generate a functional transporter. Cross-species complementation indicates a high degree of functional conservation throughout evolution.
The mechanism by which OST␣ and OST␤ interact to generate transport activity is unknown, although the present results indicate that co-expression is not required for trafficking to the cell surface (Fig. 9). OST␣ and OST␤ were able to reach the plasma membrane when expressed individually in oocytes; however, they were not functional (Fig. 8). These findings argue against a chaperone function and suggest that either one protein may play a regulatory role or the two proteins may be forming a heterodimer or hetero-oligomers. The present results also indicate that comparable levels of transport activity are Oocytes were injected with water, 1 ng of hOST␣ cRNA plus 1 ng of hOST␤ cRNA, or 1 ng of mOst␣ cRNA plus 1 ng of mOst␤ cRNA. After 3 days, uptake of 50 nM [ 3 H]estrone 3-sulfate in hOST␣-hOST␤-or mOST␣-mOST␤Ϫexpressing oocytes was measured in the presence of NaCl-containing medium (regular modified Barth's solution) or medium in which NaCl (88 mM) was substituted isosmotically with choline chloride or lithium chloride. Uptake was measured at 25°C for 1 h, and values are presented as means Ϯ S.E., n ϭ 3. fmol, femtomoles. generated by proteins that exhibit only 25% amino acid identity, suggesting that only a few conserved amino acids may be required for this interaction. For example, human OST␣ was activated equally well by human OST␤ or skate Ost␤ (Fig. 3) despite only a 25% amino acid identity for the latter two proteins (Fig. 2). Alternatively, the interaction between OST␣ and OST␤ may be determined more by their three-dimensional structures or post-translational modifications than by primary  amino acid sequences or may require the participation of a third, as yet unidentified protein or cofactor. Interestingly, the predicted transmembrane domain architectures of human, mouse, and skate OST␣ are similar, as are those of the three OST␤ proteins (Figs. 1 and 2), providing additional evidence that these proteins carry out the same biological functions. Each of the predicted TM domains and hydrophilic loops in OST␣ and OST␤ from human, mouse, and skate are similar in length and relative position within the polypeptides (Figs. 1 and 2). The only significant exception is skate Ost␤, which has a longer amino terminus region; however, the first 27 amino acids of skate Ost␤ are predicted to be a signal peptide (8) such that the mature protein may be comparable in length to the human and mouse proteins. The conserved membrane architecture between these evolutionarily divergent proteins indicates that this membrane structure is essential for function.
As indicated above, the amino acid identity between the human, mouse, and skate proteins is not restricted to the TM helices but is also seen in putative intracellular and extracellular loops (Figs. 1 and 2). For OST␣, several amino acid regions appear highly conserved in the hydrophilic loops, including an unusual stretch of 6 -7 cysteine residues that reside in a predicted cytosolic loop between TM domains 3 and 4 (TGPCCCCCPC(C/L)P; Fig. 1). The significance of this cysteine motif in OST␣ is not known, although it may function either as a ligand or substrate binding site, a site of interaction with OST␤, or as a site of membrane association. In general, cysteine residues play important roles in protein secondary structure, metal coordination, oligomerization, and post-translational modifications. Proteins that contain comparable cysteine-rich motifs include the human t-SNARE protein syntaxin 11 (GenBank TM /EBI accession number NP_003755), a chicken protocadherin isoform (GenBank TM /EBI accession number AAK57196), a candidate gene for human Cat-Eye syndrome (GenBank TM /EBI accession number AAK30049), the cysteine-string proteins, two proteins of unknown function (Chic1 and CHIC2) AF121796). Although each of these proteins contains a short polycysteine motif, they share no additional sequence identity with OST␣. It is interesting to note, however, that many of these proteins are associated with the cell membrane and are either directly or indirectly involved in membrane transport (14 -21).
It is also interesting to note that human, mouse, and skate OST␣ and OST␤ proteins all appear to have membrane-targeting sequences in their carboxyl-terminal, putative cytosolic domains. Skate Ost␣ and Ost␤ and mouse OST␤ have an Arg-X-Arg (RXR) motif, whereas human OST␣ and mouse OST␣ have an RRK sequence at the corresponding location in the sequence (Figs. 1 and 2). RXR sequences in hetero-oligomeric proteins function as retention or retrieval signals that must be masked before the corresponding protein complexes can be transported from the endoplasmic reticulum (22)(23)(24).
Although there are currently no known primary structural homologues for OST␣ and OST␤ in the mouse or human genomes, the OST␤ sequences from mouse and skate exhibit a low level of predicted amino acid identity (18 -22%) with the carboxyl-terminal 200 amino acids (i.e. the single-TM helix and the carboxyl-terminal domain) of protocadherin-␥, a cell surface glycoprotein that belongs to the cadherin superfamily (8).
Cadherins are involved in cell recognition, signaling, morphogenesis, and angiogenesis, and one study has identified a role for these proteins in organic solute transport (25). Dantzig and co-workers (25) discovered a cadherin homologue that was associated with the acquisition of peptide transport activity by transport-deficient cells (human peptide transporter-1, HPT-1). Because HPT-1 has only one putative TM domain, these authors speculated that this cadherin homologue may be one component of a heteromultimeric complex or may self-associate to form a homomultimeric complex. Although these hypotheses have not yet been tested, these observations by Dantzig and colleagues suggest that some members of the cadherin superfamily of proteins or perhaps structurally related molecules such as OST␤ may contribute to membrane transport activity. However, because current methods of identifying homologues rely on comparing primary sequence data, these methods cannot identify structural or functional homologues that exhibit only a low level of primary sequence identity. As better modeling programs and algorithms arise, OST structural homologues may be identified that are not currently apparent from primary sequence data.
The absence of primary structural homologues for OST␣ and OST␤ and the fact that this transporter has survived evolutionary selection provide support for the hypothesis that these genes play a necessary and perhaps unique physiological role in humans. The present results indicate that OST␣-OST␤ can function as a transporter for steroids such as estrone 3-sulfate, taurocholate, and digoxin, as well as the eicosanoid prostaglandin E 2 (Fig. 4). Because steroids and eicosanoids are involved in many cellular functions, this transporter may play a central role in regulating these activities. Thus, one possible role of OST␣-OST␤ is to regulate cellular entry and/or exit of signaling molecules.
Additional evidence for an essential physiological role of these genes is provided by the broad tissue distribution of OST␣ and OST␤ mRNA and by the high levels of mRNA found in several human tissues (Fig. 7). mRNA expression was detected in 17 of 19 tissues examined with relatively high levels found in testis, colon, liver, small intestine, kidney, ovary, and adrenal gland. In these tissues, OST mRNA was present at 5-40% of the levels of ␤-actin mRNA. Because ␤-actin is a relatively abundant transcript, this indicates that OST␣ and OST␤ are expressed at high levels. However, because transcript abundance does not always correlate with high protein expression, additional studies are needed to evaluate relative protein abundance as well as the cellular and subcellular localization of these proteins.