The cellular receptor for gibbon ape leukemia virus is a novel high affinity sodium-dependent phosphate transporter.

The primate type C retrovirus gibbon ape leukemia virus (GaLV) has been shown to use a widely expressed, multiple membrane-spanning protein of unknown function as its cell surface receptor on human cells (GLVR1) (Johann, S. V., Gibbons, J. J., and O'Hara, B. (1992) J. Virol. 66, 1635-1640; O'Hara, B., Johann, S. V., Klinger, H. P., Blair, D. G., Rubinson, H., Dunni, K.J., Sass, P., Vitek, S. M., and Robins, T. (1990) Cell Growth Diff. 1, 119-127). Here we present evidence that the receptor for GaLV (GLVR1) functions as a sodium-dependent transporter of inorganic phosphate. GLVR1 is shown to have approximately 3-4-fold higher affinity for phosphate than other mammalian phosphate transporters described to date. Productive infection of GLVR1-expressing cells by GaLV, but not other retroviruses, results in the complete blockade of GLVR1-specific uptake of inorganic phosphate. Since productive infection of cells with GaLV is generally not cytotoxic, it is likely that more than one phosphate transporter exists on the cell surface. Our data suggest that GLVR1 represents a sodium-dependent phosphate transporter that differs from other mammalian phosphate transporters in structure, affinity for phosphate, and function.

The primate type C retrovirus gibbon ape leukemia virus (GaLV) has been shown to use a widely expressed, multiple membrane-spanning protein of unknown function as its cell surface receptor on human cells (GLVR1) ( Here we present evidence that the receptor for GaLV (GLVR1) functions as a sodium-dependent transporter of inorganic phosphate. GLVRl is shown to have approximately "fold higher affinity for phosphate than other mammalian phosphate transporters described to date. Productive infection of GLVR1-expressing cells by GaLV, but not other retroviruses, results in the complete blockade of GLVR1-specific uptake of inorganic phosphate. Since productive infection of cells with GaLV is generally not cytotoxic, it is likely that more than one phosphate transporter exists on the cell surface. Our data suggest that GLVRl represents a sodium-dependent phosphate transporter that differs from other mammalian phosphate transporters in structure, affinity for phosphate, and function.
Virus receptors play a critical role in viral infection. At present, the cellular functions for only two retroviral receptors have been identified. The receptor for the lentivirus, human immunodeficiency virus (HIV),' has been identified as the CD4 molecule (1)(2)(3), and the receptor for the type C ecotropic murine leukemia virus (E-MuLV) has been demonstrated to function as a sodium-independent cell surface transporter of basic amino acids (4)(5)(6). The discovery that the CD4 molecule serves as the receptor for HIV has increased the understanding of the role of CD4 in viral-mediated pathology and enabled the development of CD4-based antiviral strategies (7, 8).
As a transporter of basic amino acids, the E-MuLV receptor serves a critical cell function. Paradoxically, E-MuLV infection is not cytopathic to most cell types, even though productive infection of cells by E-MuLV renders the receptor inaccessible to incoming virus (9). The puzzle of how cells maintain productive infection with E-MuLV in the absence of a cytopathic effect was solved when it was shown that mink cells that express the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. l l To whom reprint requests should be addressed. A receptor for a second type C retrovirus, gibbon ape leukemia virus (GaLV), has been cloned and sequenced (11). This receptor (designated GLVR1) is ubiquitously expressed; the mRNAis present at very high levels in brain and thymus but is detectable in all tissues examined (12). In addition to the broad tissue distribution of this receptor, the GLVRl protein is present in a wide variety of species. Functional GaLV receptors are present on cells derived from cat, dog, cow, some birds, bat, mink, rabbit, monkey, primate, and most rodents with the notable exception of mice (13). The murine GaLV receptor homologue shares 90% amino acid identity with the functional human form of the receptor (12). It has been shown previously that substitution of as few as 2 amino acids in the murine form of GLVRl allows for GaLV receptor function (14). These subtle differences in the primary structure of the form of the protein expressed on murine cells account for the resistance of murine cells t o GaLV infection.
The cellular function of this widespread receptor has not been determined. It has been reported that GLVRl shares 31% amino acid identity with a phosphate permease from Neurospora crussa, Pho-4' (12,15). Furthermore, we noted that GLVRl also shares 17% amino acid identity with the sodiumdependent inorganic phosphate transporter identified from human kidney cortex (16). Based on these observations, we sought to determine whether GLVRl may function as a transporter of inorganic phosphate and how cells are able to maintain productive infection with GaLV in the absence of cytopathic effect.
We report here that GLVRl is a ubiquitously expressed, high affinity phosphate transporter, which presumably functions in maintaining cellular phosphate levels.
MDTF cells expressing the human cDNA for the GaLV receptor were infected with wild type GaLV, strain SEATO, as described previously (17). Infection was monitored by measuring the reverse transcriptase activity from the cell media of the infected cells (18). By 2-3 weeks post-infection, the reverse transcriptase activity had reached a peak level, at which time productive GaLV infection was assessed by a superinfection interference assay. GLVR1-expressing MDTF cells were also infected with two different molecular clones of ecotropic murine leukemia virus (E-MuLV), Friend murine leukemia virus (57A) (19), or Moloney murine leukemia virus (MLV-K) (20). Productive infection by E-MuLVs was assessed by reverse transcriptase assays, followed by superinfection interference assays.
Superinfection interference assays were performed as described previously (17). Briefly, uninfected or productively infected cells were exposed to retroviral vectors bearing either GaLV envelope glycoproteins (PGl3/GlBgSvN) (21,22)  glycoproteins (CREBAG) (23,24). Both the GlBgSvN and BAG genomes encode two bacterial proteins, P-galactosidase and neomycin phosphotransferase, allowing for detection of infected cells by two different assays. In this study, infection was measured by counting the number of foci of P-galactosidase-positive cells detected histochemically 48-72 h after exposure of uninfected and infected cells to the retroviral vectors as described previously (17). The extent of superinfection interference was then determined by comparing the apparent titers of the vectors on infected cells to those on uninfected cells. A greater than 100-fold inhibition of superinfection was used to confirm productive infection of the cells. The productively infected cells were then used in phosphate uptake assays, as described below.
Anion Uptake Assays-All [32Plphosphate uptake experiments were performed essentially as described (26). Cells were cultivated in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum until they reached confluence in 24-well culture trays (Costar, Cambridge, MA). The attached cells were washed once with uptake medium (10 mM HEPES-KOH buffer, pH 6.5, containing 0.1 mM KH,PO,, 137 mM NaCl, 5.4 mM KCl, 2.8 mM CaCl,, and 1.2 mM MgSO,). For sulfate uptake the [MgSO,] was lowered to 0.1 mM in the uptake medium. Uptake medium (200 pl/well) supplemented with carrier free KH,3'P04 or Na,3%O, (5 pCi/well) was added to the cells to initiate anion transport. The rate of sodium-dependent phosphate uptake was found to be linear through 3 min under all conditions reported. Incubations for 2 min at 37 "C were routinely employed to insure linearity and to allow determination of the kinetic parameters of sodium-dependent phosphate uptake as indicated for MDTFIGLVRl cells. Following incubation at 37 "C for 2 min, transport was terminated by rapidly removing the uptake medium and washing the cells four times with ice-cold phosphate-buffered saline. The washed cells were solubilized in a lysis buffer (50 mM Tris-HC1, pH 7.2, containing 150 mM NaCl, 1% Triton X-100, 0.5% deoxycholate, 0.1% SDS, 5 mM EDTA), and aliquots of the solubilized cell extracts were counted in a liquid scintillation counter. The protein concentrations of the lysis buffer-solubilized cell extracts were determined with the use of protein assay reagent (BCA, Pierce) according to the procedure provided by the manufacturer, using bovine serum albumin as protein standard.
The cation specificity of Pi uptake was tested in the standard uptake medium a s described above, except that Na' was replaced with 137 mM Li', K', NH;, or choline, as described previously (26).

RESULTS
Functional Properties of the GLVRl Protein-As described above, murine cells are resistant to GaLV infection because the murine form of the GLVRl protein does not function as a GaLV receptor (12,14). This property of murine cells provides a means of selecting murine cells that express the human form of the GLVRl cDNA based on their acquired susceptibility to GaLV infection. To determine whether GLVRl can function as a phosphate transporter, two murine fibroblast cell lines, NIH 3T3 and MDTF (M. dunni tail fibroblasts), were infected by a  (Fig. 1). Asimilar increase in anion transport was not observed when sulfate was substituted for phosphate under identical experimental conditions, suggesting transport was selective for Pi anions.
Several functional parameters of GLVR1-specific Pi transport were assessed and compared to the previously reported parameters established for the endogenous Pi transport function in NIH 3T3 cells (26) and for the Na/P, transporter in human kidney cortex (16). The effects of different monovalent cations on GLVR1-dependent phosphate uptake were compared by substituting potassium, lithium, NH,, or choline chloride for sodium chloride in the uptake medium of MDTF and MDTFI GLVRl cells. As shown in Fig. 2, replacement of sodium chloride with each of these cations reduced phosphate transport dramatically. These results indicate that both the human GLVRl transporter and the endogenous murine phosphate transporter(s1 are sodium-dependent. The phosphate transporter identified in human kidney cortex has a pH optimum of 8.0 for phosphate uptake (16). Phosphate uptake was measured in uptake media with pH values ranging from 5.5 to 8.5. The endogenous murine transporter and the human GLVRl transporter appear to exhibit two peaks of phosphate uptake: one occurring at pH 6.5 and another at pH 7.5 (Fig. 3).
The kinetic parameters (V,,,,, and K,) of phosphate uptake were determined for the endogenous transporter(s) expressed in MDTF cells and for the human form of the GLVRl transporter overexpressed in MDTF cells (Fig. 4). The V, , determined for both the endogenous MDTF transporter and the GLVRl transporter (2.3 and 1.9 nmol of P/min/mg of protein, respectively) is approximately 10 times higher than that reported previously for phosphate uptake in NIH 3T3 fibroblasts (V,, = 0.25 nmol of P,/min/mg of protein) (26). The affinity for Pi of the sodium-dependent transporter present in MDTF cells (K, = 133 PM) is in the same range as that reported for the human kidney cortex N O i transporter (K, = 170 p~) (16), but higher than that observed with NIH 3T3 cells ( K , = 220 PM) (26). The K, for phosphate (53 p~) determined for the GLVRl transporter is significantly lower than that for any of these transporters, suggesting that the GaLV receptor has a higher affinity for phosphate than other mammalian phosphate transporters described to date.

Viral Receptor Is Novel NaIP, Dansporter
GaLV Infection Blocks GLVRl specific Pi Transport-When cells expressing a functional GaLV receptor are infected with GaLV, their viral receptors are no longer accessible (27). To determine what effect GaLV infection and the concomitant loss of functional GaLV receptors might have on GLVR1-specific Pi transport, MDTF/GLVRl cells were exposed to GaLV strain SEATO. At 2-3 weeks post-exposure, viral infection assays were performed (as described under "Materials and Methods") to insure virus had spread throughout the culture. The absence of functional human GLVRl encoded viral receptors on the surface of infected MDTF/GLVRl cells was demonstrated by the more than 100-fold diminution in viral-induced foci observed when infected MDTF/GLVRl cells were exposed to homologous challenge virus (e.g. E-MuLV or GaLV) (data not shown). Infection of MDTF/GLVRl cells with GaLV resulted in a reduction of Pi uptake to the level of the control MDTF cells (1240 and 1212.5 pmol of P,/min/mg of protein, respectively), compared to a nearly 2-fold increase observed in uninfected MDTF/GLVRl cells (2129 pmol of Pjminlmg of protein) (Fig. 5).
As a control for the possibility that productive retroviral infection contributes to a reduction of Pi uptake by a nonspecific mechanism, we productively infected MDTF/GLVRl cells with retroviruses that use a receptor other than GLVRl to infect cells. MDTF/GLVRl cells were infected with two different strains of E-MuLV, Friend murine leukemia virus, and Moloney murine leukemia virus as described under "Materials and Methods." E-MuLVs have been shown to use a cell surface transporter of cationic amino acids as a viral receptor (5,6) and therefore provide a useful control for the general effects of retroviral infection on Pi transport. In contrast to the results obtained following GaLV infection, infection of MDTF/GLVRl cells with either Friend or Moloney murine leukemia virus did not result in a marked reduction in the level of Pi uptake relative to that of uninfected MDTF/GLVRl cells (Fig. 5 ) . This result shows that transport of phosphate via the GLVRl receptor is completely and specifically blocked in GaLV-infected cells. The observation that cells productively infected by GaLV do not demonstrate any cytopathic effect, despite the fact that phosphate uptake is presumably critical to the viability of the cells, argues for the presence of more than one phosphate transporter in these cell types. DISCUSSION The receptor for the gibbon ape leukemia virus (GaLV) is an integral membrane protein found on most types of cells. We have now determined that the GaLV receptor functions as a high affinity sodium-dependent phosphate transporter.
Murine cells express a form of the GLVRl protein that, in contrast to the human GLVRl protein, does not function as a GaLV receptor (12,14). Expression of the human form of this GLVR1-specific phosphate uptake.  GLVRl has several structural features that are found in other transporters/phosphate permeases. A hydropathy plot of the deduced amino acid sequence of GLVRl indicates 10 membrane-spanning a-helical segments (11). According to determinations of the mean hydrophobicity and hydrophobic moments of each of the membrane-spanning domains, four of the segments (segments 1, 6, 9, and 10) are highly hydrophobic, while six of the segments (2-5, 7, and 8) are more amphipathic in nature ( Table I). Since each of these a-helical structures can take on a cylinder-like shape, a model was prepared to indicate the location of the hydrophilic and hydrophobic amino acid residues that make up each of the a-helical segments. As depicted in Fig. 6, the amino acid residues within the six amphipathic transmembrane segments (2-5,7, and 8) are arrayed such that the hydrophilic residues would be on one side of the cylinder and the hydrophobic residues would be on the other. In this model, with the amphipathic transmembrane segments oriented so that their hydrophobic sides face toward the hydrophobic lipid environment of the membrane, the hydrophilic sides of these six transmembrane segments form a channel or pore to facilitate phosphate uptake. A similar pentagonal pore structure has been hypothesized for members of the glucose transporter family (29).  Another feature shared by GLVRl and other Na+-dependent transporters is the presence of a proposed Na' binding domain identified by Deguchi et al. (30) This sequence, G1y-X-X-X-X-Leu-X-X-X-Gly-Arg, is present in the kidney-specific Na'l phosphate transporters, the rabbit and human Na+/glucose transporters in the intestine, and the Na'iglutamate and the Na+/proline transporters ofEscherichia coli (31). We have identified a sequence spanning amino acid residues 571-581 that conforms to these consensus sequence requirements. Two possibilities that might account for the ability of GaLV to block GLVR1-mediated Pi transport are as follows: 1) GaLV binding could directly block the ability of Pi to interact with GLVRl due to the physical proximity of the Pi and GaLV binding sites, or 2) GaLV binding could inhibit the ability of Na' to interact with GLVR1. Since residues 550-558 in the fourth extracellular do-  (Table I). The spatial location of hydrophilic (lightly shaded) and hydrophobic (darkly shaded) amino acid residues within each of the transmembrane segments is presented in the model. When the amphipathic transmembrane domains (segments 2-5, 7, and 8 ) are oriented so that their hydrophobic sides face the membrane lipid environment and their hydrophilic sides face inward, they can form a hexagonal pore consisting of these six a-helical segments of the GaLV receptor/phosphate transporter. main have been shown to be critical for GaLV infection (14,32), the proximity of the putative Na' binding site and the virus binding site may account for the reduction in GLVR1-specific sodium-dependent phosphate transport that accompanies GaLV infection.

7-
Although GLVRl shares a high degree of homology and a similar topology with Pho-4+, the phosphate permease of N.
crassa, it is structurally and functionally distinct and exhibits a different distribution pattern than the Nap,-1, -2, and -3 kidney-specific phosphate transporters. The NaP,-l, -2, and -3 transporters are tissue-restricted in their expression (16,28), whereas GLVRl is expressed in all tissues examined (12). Functional parameters also distinguish GLVR1-specific Pi transport from Pi transport by the human kidney Nap,-3 transporter. First, Nap,-3 transport responds dramatically to increasing pH, doubling the level of Pi transport as the pH is changed from 7.0 to 8.0. This phenomenon is thought to be due to allosteric regulation by protons of the N e , cotransport system (31). Both the endogenous NaR, transporter in MDTF cells and the GLVR1-specific Pi transporter exhibit peaks of maximal phosphate uptake a t pH 6.5 and 7.5, with a dramatic decrease in Pi uptake at pH 8.0 and above. This suggests a different effect of pH on the regulation of GLVR1-specific Pi transport compared to the Nap,-3 transporter. Second, the affinity for phosphate of the GLVRl sodium-dependent phosphate transporter is approximately 3-4-fold higher than that of other mammalian phosphate transporters (16,26,28), suggesting that the human form of GLVRl transporter is a high aGnity transporter for phosphate. Finally, unlike the kidney transporters, GLVRl is highly evolutionary conserved and is expressed in a wide variety of tissues and cell lines (11,12).
The process by which mammalian retroviruses choose their receptors is an opportunistic one. Since productive infection by retroviruses results in loss of available receptor for infection, there are essentially three types of cell surface molecules that are candidates as retroviral receptors, namely (a) molecules whose functional or substrate binding site is different from that of the retrovirus, so that cellular function is not abrogated by infection (e.g. the receptor for ecotropic murine leukemia viruses (10); ( h ) molecules whose function is a luxury one, so that loss of cell function does not result in cell death and loss of viral host function (for example, CD4, the receptor for HIV), and (c) molecules whose function in the cell is redundant, so that loss of cellular function is not synonymous with cell death. The GaLV receptor appears to be the first example of the third type; our data suggest that cells susceptible to GaLV contain at least one phosphate transporter in addition to the GaLV receptor so that infection with GaLV does not inhibit all phosphate transport. Finally, given the role of inorganic phosphate as an important regulator of cellular metabolism, it will be of interest to better understand how this novel ubiquitous transporter functions in maintaining cellular phosphate homeostasis.