Cooperation of multiple CCR5 coreceptors is required for infections by human immunodeficiency virus type 1.

In addition to the primary cell surface receptor CD4, CCR5 or another coreceptor is necessary for infections by human immunodeficiency virus type 1 (HIV-1), yet the mechanisms of coreceptor function and their stoichiometries in the infection pathway remain substantially unknown. To address these issues, we studied the effects of CCR5 concentrations on HIV-1 infections using wild-type CCR5 and two attenuated mutant CCR5s, one with the mutation Y14N at a critical tyrosine sulfation site in the amino terminus and one with the mutation G163R in extracellular loop 2. The Y14N mutation converted a YYT sequence at positions 14 to 16 to an NYT consensus site for N-linked glycosylation, and the mutant protein was shown to be glycosylated at that position. The relationships between HIV-1 infectivity values and CCR5 concentrations took the form of sigmoidal (S-shaped) curves, which were dramatically altered in different ways by these mutations. Both mutations shifted the curves by factors of approximately 30- to 150-fold along the CCR5 concentration axis, consistent with evidence that they reduce affinities of virus for the coreceptor. In addition, the Y14N mutation specifically reduced the maximum efficiencies of infection that could be obtained at saturating CCR5 concentrations. The sigmoidal curves for all R5 HIV-1 isolates were quantitatively consistent with a simple mathematical model, implying that CCR5s reversibly associate with cell surface HIV-1 in a concentration-dependent manner, that approximately four to six CCR5s assemble around the virus to form a complex needed for infection, and that both mutations inhibit assembly of this complex but only the Y14N mutation also significantly reduces its ability to successfully mediate HIV-1 infections. Although several alternative models would be compatible with our data, a common feature of these alternatives is the cooperation of multiple CCR5s in the HIV-1 infection pathway. This cooperativity will need to be considered in future studies to address in detail the mechanism of CCR5-mediated HIV-1 membrane fusion.

Critical role of enhanced CD4 affinity in laboratory adaptation of human immunodeficiency virus type 1.

Strains of human immunodeficiency virus type 1 (HIV-1) that use the coreceptor CXCR4 (X4 strains) become laboratory adapted (LA) when selected for ability to replicate in leukemic T cell lines such as H9. Compared with patient X4 viruses, the gp120-gp41 complexes of LA viruses have a constellation of common properties including enhanced affinities for CD4, greater sensitivities to inactivations by diverse antibodies and by soluble CD4, increased shedding of gp120, and improved abilities to infect HeLa-CD4 cell clones that contain only trace quantities of CD4. These common characteristics, which may result from a concerted structural rearrangement of the gp120-gp41 complexes, have made it difficult to identify a specific feature that is critical for laboratory adaptation. To test the hypothesis that replication of patient X4 HIV-1 is limited by the low CD4 concentration in H9 cells (7.0 x 10(3) CD4/cell), we constructed H9 derivatives that express at least 10 times more of this receptor. Interestingly, most patient X4 isolates readily grew in these derivative cells, and the resulting virus preparations retained the characteristics of primary viruses throughout multiple passages. In contrast, selection of the same viruses in the parental H9 cells resulted in outgrowth of LA derivatives. We conclude that a weak interaction of patient X4 HIV-1 isolates with CD4 is the primary factor that limits their replication in leukemic T cell lines.

Polymorphisms of the cell surface receptor control mouse susceptibilities to xenotropic and polytropic leukemia viruses.

The differential susceptibilities of mouse strains to xenotropic and polytropic murine leukemia viruses (X-MLVs and P-MLVs, respectively) are poorly understood but may involve multiple mechanisms. Recent evidence has demonstrated that these viruses use a common cell surface receptor (the X-receptor) for infection of human cells. We describe the properties of X-receptor cDNAs with distinct sequences cloned from five laboratory and wild strains of mice and from hamsters and minks. Expression of these cDNAs in resistant cells conferred susceptibilities to the same viruses that naturally infect the animals from which the cDNAs were derived. Thus, a laboratory mouse (NIH Swiss) X-receptor conferred susceptibility to P-MLVs but not to X-MLVs, whereas those from humans, minks, and several wild mice (Mus dunni, SC-1 cells, and Mus spretus) mediated infections by both X-MLVs and P-MLVs. In contrast, X-receptors from the resistant mouse strain Mus castaneus and from hamsters were inactive as viral receptors. These results suggest that X-receptor polymorphisms are a primary cause of resistances of mice to members of the X-MLV/P-MLV family of retroviruses and are responsible for the xenotropism of X-MLVs in laboratory mice. By site-directed mutagenesis, we substituted sequences between the X-receptors of M. dunni and NIH Swiss mice. The NIH Swiss protein contains two key differences (K500E in presumptive extracellular loop 3 [ECL 3] and a T582 deletion in ECL 4) that are both required to block X-MLV infections. Accordingly, a single inverse mutation in the NIH Swiss protein conferred X-MLV susceptibility. Furthermore, expression of an X-MLV envelope glycoprotein in Chinese hamster ovary cells interfered efficiently with X-MLV and P-MLV infections mediated by X-receptors that contained K500 and/or T582 but had no effect on P-MLV infections mediated by X-receptors that lacked these amino acids. In contrast, moderate expression of a P-MLV (MCF247) envelope glycoprotein did not cause substantial interference, suggesting that X-MLV and P-MLV glycoproteins interfere nonreciprocally with X-receptor-mediated infections. We conclude that P-MLVs have become adapted to utilize X-receptors that lack K500 and T582. A penalty for this adaptation is a reduced ability to interfere with superinfection. Because failure of interference is a hallmark of several exceptionally pathogenic retroviruses, we propose that it contributes to P-MLV-induced diseases.

Roles of CD4 and coreceptors in binding, endocytosis, and proteolysis of gp120 envelope glycoproteins derived from human immunodeficiency virus type 1.

Infections by human immunodeficiency virus type 1 (HIV-1) involve interactions of the viral envelope glycoprotein gp120 with CD4 and then with a coreceptor. R5 isolates of HIV-1 use CCR5 as a coreceptor, whereas X4 isolates use CXCR4. It is not known whether coreceptors merely trigger fusion of the viral and cellular membranes or whether they also influence the energetics of virus adsorption, the placement of the membrane fusion reaction, and the metabolism of adsorbed gp120. Surprisingly, the pathway for metabolism of adsorbed gp120 has not been investigated thoroughly in any cells. To address these issues, we used purified (125)I-gp120s derived from the R5 isolate BaL and from the X4 isolate IIIB as ligands for binding onto human cells that expressed CD4 alone or CD4 with a coreceptor. The gp120 preparations were active in forming ternary complexes with CD4 and the appropriate coreceptor. Moreover, the cellular quantities of CD4 and coreceptors were sufficient for efficient infections by the corresponding HIV-1 isolates. In these conditions, the kinetics and affinities of (125)I-gp120 adsorptions and their subsequent metabolisms were strongly dependent on CD4 but were not significantly influenced by CCR5 or CXCR4. After binding to CD4, the (125)I-gp120s slowly became resistant to extraction from the cell monolayers by pH 3.0 buffer, suggesting that they were endocytosed with half-times of 1-2 h. Within 20-30 min of endocytosis, the (125)I-gp120s were proteolytically degraded to small products that were shed into the media. The weak base chloroquine strongly inhibited (125)I-gp120 proteolysis and caused its intracellular accumulation, suggesting involvement of a low pH organelle. Results supporting these methods and conclusions were obtained by confocal immunofluorescence microscopy. We conclude that the energetics, kinetics, and pathways of (125)I-gp120 binding, endocytosis, and proteolysis are determined principally by CD4 rather than by coreceptors in cells that contain sufficient coreceptors for efficient infections. Therefore, the role of coreceptors in HIV-1 infections probably does not include steerage or subcellular localization of adsorbed virus.

gp120 envelope glycoproteins of human immunodeficiency viruses competitively antagonize signaling by coreceptors CXCR4 and CCR5.

Signal transductions by the dual-function CXCR4 and CCR5 chemokine receptors/HIV type 1 (HIV-1) coreceptors were electrophysiologically monitored in Xenopus laevis oocytes that also coexpressed the viral receptor CD4 and a G protein-coupled inward-rectifying K+ channel (Kir 3.1). Large Kir 3.1-dependent currents generated in response to the corresponding chemokines (SDF-1alpha for CXCR4 and MIP-1alpha; MIP-1beta and RANTES for CCR5) were blocked by pertussis toxin, suggesting involvement of inhibitory guanine nucleotide-binding proteins. Prolonged exposures to chemokines caused substantial but incomplete desensitization of responses with time constants of 5-7 min and recovery time constants of 12-19 min. CXCR4 and CCR5 exhibited heterologous desensitization in this oocyte system, suggesting possible inhibition of a common downstream step in their signaling pathways. In contrast to chemokines, perfusion with monomeric or oligomeric preparations of the glycoprotein of Mr 120, 000 (gp120) derived from several isolates of HIV-1 did not activate signaling by CXCR4 or CCR5 regardless of CD4 coexpression. However, adsorption of the gp120 from a T-cell-tropic virus resulted in CD4-dependent antagonism of CXCR4 response to SDF-1alpha, whereas gp120 from macrophage-tropic viruses caused CD4-dependent antagonism of CCR5 response to MIP-1alpha. These antagonisms could be partially overcome by high concentrations of chemokines and were specific for coreceptors of the corresponding HIV-1 isolates, suggesting that they resulted from direct interactions of gp120-CD4 complexes with coreceptors and that they did not involve the desensitization pathway. These results indicate that monomeric or oligomeric gp120s specifically antagonize CXCR4 and CCR5 signaling in response to chemokines, but they do not exclude the possibility that gp120s might also function as weak agonists in some cells. The gp120-mediated disruption of CXCR4 and CCR5 signaling may contribute to AIDS pathogenesis.

Polymorphisms in the CCR5 genes of African green monkeys and mice implicate specific amino acids in infections by simian and human immunodeficiency viruses.

CCR5, a receptor for the CC chemokines RANTES, Mip1alpha, and Mip1beta, has been identified as a coreceptor for infections by macrophage-tropic isolates of human immunodeficiency virus type 1 (HIV-1). To study its structure and function, we isolated cDNA clones of human, African green monkey (AGM), and NIH/Swiss mouse CCR5s, and we quantitatively analyzed infections by macrophage-tropic HIV-1 and SIVmac251 after transfecting human HeLa-CD4 cells with the CCR5 expression vectors. The AGM and NIH/Swiss mouse CCR5 proteins are 97.7 to 98.3% and 79.8% identical to the human protein, respectively. In addition, we analyzed site-directed mutants and chimeras of these CCR5s. Cell surface expression of CCR5 proteins was monitored by using a specific rabbit antiserum and by binding the chemokine [125I]Mip1beta. Our major results were as follows. (i) Two distinct AGM CCR5 sequences were reproducibly found in DNA from CV-1 cells. The AGM clone 1 CCR5 protein differs from that of clone 2 by two substitutions, Y14N in the amino-terminal extracellular region and L352F at the carboxyl terminus. Interestingly, AGM clone 1 CCR5 was inactive as a coreceptor for all tested macrophage-tropic isolates of HIV-1, whereas AGM clone 2 CCR5 was active. As shown by chimera studies and site-directed mutagenesis, the Y14N substitution in AGM clone 1 CCR5 was solely responsible for blocking HIV-1 infections. In contrast, both AGM CCR5 clones were active coreceptors for SIVmac251. Studies of DNA samples from other AGMs indicated frequent additional CCR5 polymorphisms, and we cloned an AGM clone 2 variant with a Q93R substitution in the extracellular loop 1 from one heterozygote. This variant CCR5 was active as a coreceptor for SIVmac251 but was only weakly active for macrophage-tropic isolates of HIV-1. In addition, SIVmac251 appeared to be dependent on the extracellular amino terminus and loop 2 regions of human CCR5 for maximal infection. Our results suggest major differences in the interactions of SIVmac251 and macrophage-tropic HIV-1 isolates with 19, N13, and Y14 in the amino terminus; with Q93 in extracellular loop 1; and with extracellular loop 2 of human CCR5. (ii) The NIH/Swiss mouse CCR5 protein differs at multiple positions from sequences recently reported for other inbred strains of mice. This CCR5 was inactive as a coreceptor for HIV-1 and SIVmac251. Studies of chimeras that contained different portions of NIH/Swiss mouse CCR5 substituted into human CCR5, as well as the reciprocal chimeras, indicated that the amino-terminal region and extracellular loops 1 and 2 of human CCR5 contribute to its coreceptor activity for macrophage-tropic isolates of HIV-1. Specific differences with previous CCR5 chimera results occurred because the NIH/Swiss mouse CCR5 contains a unique substitution corresponding to P183L in extracellular loop 2 that is nonpermissive for coreceptor activity. We conclude that diverse CCR5 sequences occur in AGMs and mice, that SIVmac251 and macrophage-tropic HIV-1 isolates interact differently with specific CCR5 amino acids, and that multiple regions of human CCR5 contribute to its coreceptor functions. In addition, we have identified naturally occurring amino acid polymorphisms in three extracellular regions of CCR5 (Y14N, Q93R, and P183L) that do not interfere with cell surface expression or Mip1beta binding but prevent infections by macrophage-tropic isolates of HIV-1. In contrast to previous evidence, these results suggest that CCR5 contains critical sites that are essential for HIV-1 infections.

CD4, CXCR-4, and CCR-5 dependencies for infections by primary patient and laboratory-adapted isolates of human immunodeficiency virus type 1.

We have used a focal infectivity method to quantitatively analyze the CD4, CXCR-4, and CCR-5 dependencies for infections by diverse primary patient (PR) and laboratory-adapted (LA) isolates of human immunodeficiency virus type 1 (HIV-1). Infectivities of T-cell-tropic viruses were analyzed in a panel of HeLa-CD4 cell clones that have distinct quantities of CD4 and in human astroglioma U87MG-CD4 cells that express a large quantity of CD4 and become highly susceptible to infection after transfection with a CXCR-4 expression vector. The latter analysis indicated that PR as well as LA T-cell-tropic viruses efficiently employ CXCR-4 as a coreceptor in an optimal human cell line that contains abundant CD4. Previous uncertainties regarding coreceptor usage by PR T-cell-tropic HIV-1 isolates may therefore have derived from the assay conditions. As reported previously, unrelated LA and PR T-cell-tropic HIV-1 isolates differ in infectivities for the HeLa-CD4 clonal panel, with LA viruses infecting all clones equally and PR viruses infecting the clones in proportion to cellular CD4 quantities (D. Kabat, S. L. Kozak, K. Wherly, and B. Chesebro, J. Virol. 68:2570-2577, 1994). To analyze the basis for this difference, we used the HeLa-CD4 panel to compare a molecularly cloned T-cell-tropic PR virus (ELI1) with six of its variants that grow to different extents in CD4-positive leukemic cell lines and that differ only at specific positions in their gp120 and gp41 envelope glycoproteins. All mutations in gp120 or gp41 that contributed to laboratory adaptation preferentially enhanced infectivity for cells that had little CD4 and thereby decreased the CD4 dependencies of the infections. There was a close correlation between abilities of T-cell-tropic ELI viruses to grow in an expanded repertoire of leukemic cell lines, the reduced CD4 dependencies of their infections of the HeLa-CD4 panel, and their sensitivities to inactivation by soluble CD4 (sCD4). Since all of the ELI viruses can efficiently use CXCR-4 as a coreceptor, we conclude that an increase in viral affinity for CD4 rather than a switch in coreceptor specificity is principally responsible for laboratory adaption of T-cell-tropic HIV-1. Syncytium-inducing activities of the ELI viruses, especially when analyzed on cells with low amounts of CD4, were also highly correlated with their laboratory-adapted properties. Results with macrophage-tropic HIV-1 were strikingly different in both coreceptor and CD4 dependencies. When assayed in HeLa-CD4 cells transfected with an expression vector for CCR-5, macrophage-tropic HIV-1 isolates that had been molecularly cloned shortly after removal from patients were equally infectious for cells that had low or high CD4 quantities. Moreover, despite their substantial infectivities for cells that had only a trace of CD4, macrophage-tropic isolates were relatively resistant to inactivation by sCD4. We conclude that T-cell-tropic PR viruses bind weakly to CD4 and preferentially infect cells that coexpress CXCR-4 and large amounts of CD4. Their laboratory adaptation involves corresponding increases in affinities for CD4 and in abilities to infect cells that have relatively little CD4. In contrast, macrophage-tropic HIV-1 appears to interact weakly with CD4 although it can infect cells that coexpress CCR-5 and small quantities of CD4. We propose that cooperative binding of macrophage-tropic HIV-1 onto CCR-5 and CD4 may enhance virus adsorption and infectivity for cells that have only a trace of CD4.

Infectious properties of human immunodeficiency virus type 1 mutants with distinct affinities for the CD4 receptor.

Recent evidence suggests that primary patient isolates of T-cell-tropic human immunodeficiency virus type 1 (HIV-1 ) have lower affinities for CD4 than their laboratory-adapted derivatives, that this may partly result from tighter gp120-gp41 bonds that constrain the CD4 binding sites of the primary viruses, and that selection for increased CD4 affinity may be the principal factor in laboratory adaptation of HIV-1 (S. L. Kozak, E. J. Platt, N. Madani, F. E. Ferro, Jr., K. Peden, and D. Kabat, J. Virol. 71:873-882, 1997). These conclusions were based on studies with a panel of HeLa-CD4 cell clones that differ in CD4 levels over a broad range, with laboratory-adapted viruses infecting all clones with equal efficiencies and primary T-cell-tropic viruses infecting the clones in proportion to cellular CD4 levels. Additionally, all of the primary and laboratory-adapted T-cell-tropic viruses efficiently used CXCR-4 (fusin) as a coreceptor. To test these conclusions by an independent approach, we studied mutations in the laboratory-adapted virus LAV/IIIB that alter the CD)4 binding region of gp120 and specifically reduce CD4 affinities of free gp 120 by 85 to 98% (U. Olshevsky et al., J. Virol. 64:5701-5707, 1990). These mutations reduced virus titers to widely varying extents that ranged from severalfold to several orders of magnitude and converted infectivities on the HeLa-CD4 panel from CD4 independency to a high degree of CD4 dependency that resembled the behavior of primary patient viruses. The relative infectivities of the mutants correlated closely with their sensitivities to inactivation by soluble CD4 but did not correlate with the relative CD4 affinities of their free gp120s. Most of the mutations did not substantially alter envelope glycoprotein synthesis, processing, expression on cell surfaces, incorporation into virions, or rates of gp120 shedding from virions. However, one mutation (D457R) caused a decrease in gp160 processing by approximately 80%. The fact that several mutations increased rates of spontaneous viral inactivation (especially D368P) suggests that HIV-1 life spans may be determined by structural stabilities of viral envelope glycoproteins. All of the wild-type and mutant viruses were only slowly and inefficiently adsorbed onto cultured CD4-positive cells at 37 degrees C, and the gradual declines in viral titers in the media were caused almost exclusively by spontaneous inactivation rather than by adsorption. The extreme inefficiency with which infectious HIV-1 is able to infect cultured susceptible CD4-positive cells in standard assay conditions casts doubt on previous inferences that the vast majority of retrovirions produced in cultures are noninfectious. Apparent infectivity of T-cell-tropic HIV-1 in culture is limited by productive associations with CD4 and is influenced in an interdependent manner by CD4 affinities of viral gp120-gp41 complexes and quantities of cell surface CD4.

Modulation of ecotropic murine retroviruses by N-linked glycosylation of the cell surface receptor/amino acid transporter.

The cell surface receptor for ecotropic host-range (infection limited to mice or rats) murine leukemia viruses (MuLVs) is the widely expressed system y+ transporter for cationic amino acids (CAT-1). Like other retroviruses, ecotropic MuLV infection eliminates virus-binding sites from cell surfaces and results in complete interference to superinfection. Surprisingly, infection causes only partial (ca 40 to 60%) loss of mouse CAT-1 transporter activity. The NIH/Swiss mouse CAT-1 (mCAT-1) contains 622 amino acids with 14 hydrophobic potential membrane-spanning sequences, and it is known that the third extracellular loop from the amino terminus is required for virus binding. Although loop 3 is hypervariable in different species and mouse strains, consistent with its proposed role in virus-host coevolution, loop 3 sequences of both susceptible and resistant species contain consensus sites for N-linked glycosylation. Both of the consensus sites in loop 3 of mCAT-1 are known to be glycosylated and to contain oligosaccharides with diverse sizes (J. W. Kim and J. M. Cunningham, J. Biol. Chem. 268:16316-16320, 1993). We confirmed by several lines of evidence that N-linked glycosylation occludes a potentially functional virus-binding site in the CAT-1 protein of hamsters, thus contributing to resistance of that species. To study the role of receptor glycosylation in animals susceptible to infection, we eliminated loop 3 glycosylation sites by mutagenesis of an mCAT-1 cDNA clone, and we expressed wild-type and mutant receptors in mink fibroblasts and Xenopus oocytes. These receptors had indistinguishable transport properties, as determined by kinetic and voltage-jump electrophysiological studies of arginine uptake in oocytes and by analyses Of L-[3H]arginine uptake in mink cells. Bindings of ecotropic envelope glycoprotein gp7O to the accessible receptor sites on surfaces of mink cells expressing wild-type or mutant mCAT-1 were not significantly different in kinetics or in equilibrium affinities (i.e., K(D) approximately 3.7 X 10(-10) to 7.5 X 10(-10) M). However, when values were normalized to the same levels of mCAT-1 transporter expression, cells with wild-type glycosylated mCAT-1 had only approximately 50% as many sites for gp70 binding as cells with unglycosylated mCAT-1. Although infection with ecotropic MuLV had no effect on activity of the mink CAT-1 transporter that does not bind virus, it caused partial down-modulation of wild-type mCAT-1 and complete down-modulation of unglycosylated mutant mCAT-1. These results suggest that N-linked glycosylation causes wild-type mCAT-1 heterogeneity and that a significant proportion is inaccessible to virus. In part because only the interactive fraction of mCAT-1 can be down-modulated, infected murine cells conserve an amino acid transport capability that supports their viability.

Exceptional fusogenicity of Chinese hamster ovary cells with murine retroviruses suggests roles for cellular factor(s) and receptor clusters in the membrane fusion process.

Chinese hamster ovary (CHO) cells are naturally resistant to infection by amphotropic and ecotropic murine retroviruses, but they become susceptible after expressing corresponding receptors rRAM-1 and mCAT-1, respectively, and they then form abundant syncytia when exposed to these viruses. The fusogenic activities of CHO cell clones increase much more strongly with levels of receptor expression than do their susceptibilities to infection, suggesting that the assembly of receptor clusters may limit syncytium formation. However, other cell lines are not fusogenic, even if they express larger amounts of receptors. Our results suggest that a factor that is relatively abundant or active in CHO cells may functionally interact with rRAM-1 and mCAT-1 in a pathway that enables receptor-bearing membranes to fuse with membranes that contain viral envelope glycoproteins. In the case of CHO/rRAM-1 cells, syncytia form at foci of amphotropic 4070A virus infection by fusion-from-within of infected with uninfected cells. This fusogenic propensity is a sole property of the uninfected CHO/rRAM-1 cells, which fuse in cocultures with any cells infected with 4070A virus. With CHO/mCAT-1 cells, fusogenicity is even greater and involves fusion-from-without by ecotropic virion particles. In contrast to infection, which behaves as expected for a process limited by ecotropic virus attachment to single receptors, fusion-from-without increases dramatically for cells that express the highest levels of mCAT-1. We propose that infection and syncytium formation are limited at distinct steps of a common pathway that requires virus binding to a single receptor, assembly of multivalent virus-receptor complexes, structural changes in viral envelope glycoproteins, and membrane fusion. The limiting step in syncytium formation is a cellular process that depends on receptor clustering and is relatively active in CHO cells.

The envelope glycoprotein of an amphotropic murine retrovirus binds specifically to the cellular receptor/phosphate transporter of susceptible species.

A rat cDNA (rRam-1), which was cloned on the basis that it enables Chinese hamster ovary (CHO) cells to be infected by amphotropic host range murine retroviruses, was recently found to encode a widely expressed Na(+)-phosphate symporter (M. P. Kavanaugh, D. G. Miller, W. Zhang, W. Law, S. L. Kozak, D. Kabat, and A. D. Miller, Proc. Natl. Acad. Sci. USA 91:7071-7075, 1994). CHO cells express the hamster homolog of Ram-1 but are resistant to amphotropic retroviruses. Although the amphotropic envelope glycoprotein gp70 bound weakly onto control CHO cells, CHO/rRam-1 cells had novel high-affinity binding sites, and the resulting strongly adsorbed gp70 was only slowly removed from cell surfaces, with a half-life of greater than 6 h. CHO/rRam-1 cells were also specifically and efficiently killed by exposure to amphotropic gp70 followed by antiserum to gp70 in the presence of complement. Infection with an appropriately pseudotyped form of amphotropic retrovirus 4070A did not perturb control CHO cells or inhibit their phosphate transport. In contrast, 4070A infection of CHO/rRam-1 cells caused major alterations including cell-cell fusions, a specific 40% down-modulation of the rRam-1 component of phosphate transport, and complete interference to super-infection by amphotropic viruses. The 4070A virus-infected CHO/rRam-1 cells retained a substantial cell surface pool of rRam-1 that functioned as a phosphate transporter but not as a viral receptor. We conclude that amphotropic gp70 binds more strongly to rRam-1 than to the homologous hamster protein and that this stable attachment is necessary for infection, interference, membrane fusion, and pathogenesis.

Amplified and tissue-directed expression of retroviral vectors using ping-pong techniques.

Ping-pong amplification is an efficient process by which helper-free retrovirions replicate in cocultures of cell lines that package retroviruses into distinct host-range envelopes [11]. Transfection of a retroviral vector DNA into these cocultures results in massive virus production, with potentially endless cross-infection between different types of packaging cells. Because the helper-free virus spreads efficiently throughout the coculture, it is unnecessary to use dominant selectable marker genes, and the retroviral vectors can be simplified and optimized for expressing a single gene of interest. The most efficient ping-pong vector, pSFF, derived from the Friend erythroleukemia virus, has been used for high-level expression of several genes that could not be expressed with commonly employed two-gene retroviral vectors. Contrary to previous claims, problems of vector recombination are not inherent to ping-pong methods. Indeed, the pSFF vector has not formed replication-competent recombinants as shown by stringent assays. Here we review these methods, characterize the ping-pong process using the human erythropoietin gene as a model, and describe a new vector (pSFY) designed for enhanced expression in T lymphocytes. Factors that limit tissue-specific expression are reviewed.

Deletions in one domain of the Friend virus-encoded membrane glycoprotein overcome host range restrictions for erythroleukemia.

Although the Friend virus-encoded membrane glycoprotein (gp55) activates erythropoietin receptors (EpoR) to cause erythroblastosis only in certain inbred strains of mice but not in other species, mutant viruses can overcome aspects of mouse resistance. Thus, mice homozygous for the resistance allele of the Fv-2 gene are unaffected by gp55 but are susceptible to mutant glycoproteins that have partial deletions in their ecotropic domains. These and other results have suggested that proteins coded for by polymorphic Fv-2 alleles might directly or indirectly interact with EpoR and that changes in gp55 can overcome this defense. A new viral mutant with an exceptionally large deletion in its ecotropic domain is now also shown to overcome Fv-2rr resistance. In all cases, the glycoproteins that activate EpoR are processed to cell surfaces as disulfide-bonded dimers. To initiate analysis of nonmurine resistances, we expressed human EpoR and mouse EpoR in the interleukin 3-dependent mouse cell line BaF3 and compared the abilities of Friend virus-encoded glycoproteins to convert these cells to growth factor independence. Human EpoR was activated in these cells by erythropoietin but was resistant to gp55. However, human EpoR was efficiently activated in these cells by the same viral mutants that overcome Fv-2rr resistance in mice. By construction and analysis of human-mouse EpoR chimeras, we obtained evidence that the cytosolic domain of human EpoR contributes to its resistance to gp55 and that this resistance is mediated by accessory cellular factors. Aspects of host resistance in both murine and nonmurine species are targeted specifically against the ecotropic domain of gp55.

Cell-surface receptors for gibbon ape leukemia virus and amphotropic murine retrovirus are inducible sodium-dependent phosphate symporters.

Cell surface receptors for gibbon ape leukemia virus (Glvr-1) and murine amphotropic retrovirus (Ram-1) are distinct but related proteins having multiple membrane-spanning regions. Distant homology with a putative phosphate permease of Neurospora crassa suggested that these receptors might serve transport functions. By expression in Xenopus laevis oocytes and in mammalian cells, we have identified Glvr-1 and Ram-1 as sodium-dependent phosphate symporters. Two-electrode voltage-clamp analysis indicates net cation influx, suggesting that phosphate is transported with excess sodium ions. Phosphate uptake was reduced by > 50% in mouse fibroblasts expressing amphotropic envelope glycoprotein, which binds to Ram-1, indicating that Ram-1 is a major phosphate transporter in these cells. RNA analysis shows wide but distinct tissue distributions, with Glvr-1 expression being highest in bone marrow and Ram-1 in heart. Overexpression of Ram-1 severely repressed Glvr-1 synthesis in fibroblasts, suggesting that transporter expression may be controlled by net phosphate accumulation. Accordingly, depletion of extracellular phosphate increased Ram-1 and Glvr-1 expression 3- to 5-fold. These results suggest simple methods to modulate retroviral receptor expression, with possible applications to human gene therapy.

A Friend virus mutant encodes a small glycoprotein that causes erythroleukemia.

The Pvu delta mutant of Friend spleen focus-forming virus encodes the smallest env glycoprotein (apparent M(r), 41,000) known to activate erythropoietin receptors. In vivo, Pvu delta causes erythroblastosis and the development of erythroleukemia. We isolated two leukemic cell lines that contain Pvu delta; both synthesize hemoglobin in response to dimethyl sulfoxide. The Pvu delta env gene contains a 204-base deletion in the ecotropic-specific region, suggesting that this domain of the glycoprotein is not essential for viral pathogenesis.

Differences in CD4 dependence for infectivity of laboratory-adapted and primary patient isolates of human immunodeficiency virus type 1.

CD4 is known to be an important receptor for human immunodeficiency virus type 1 (HIV-1) infection of T lymphocytes and macrophages. However, the limiting steps in CD4-dependent HIV-1 infections in vivo and in vitro are poorly understood. To address this issue, we produced a panel of HeLa-CD4 cell clones that express widely different amounts of CD4 and quantitatively analyzed their infection by laboratory-adapted and primary patient HIV-1 isolates. For all HIV-1 isolates, adsorption from the medium onto HeLa-CD4 cells was inefficient and appeared to be limiting for infection in the conditions of our assays. Adsorption of HIV-1 onto CD4-positive peripheral blood mononuclear cells was also inefficient. Moreover, there was a striking difference between laboratory-adapted and primary T-cell-tropic HIV-1 isolates in the infectivity titers detected on different HeLa-CD4 cells. Laboratory-adapted HIV-1 isolates infected all HeLa-CD4 cell clones with equal efficiencies regardless of the levels of CD4, whereas primary HIV-1 isolates infected these clones in direct proportion to cellular CD4 expression. Our interpretation is that for laboratory-adapted isolates, a barrier step that preceeds CD4 encounter was limiting and the subsequent CD4-dependent virus capture process was highly efficient, even at very low cell surface concentrations. In contrast, for primary HIV-1 isolates, the CD4-dependent steps appeared to be much less efficient. We conclude that primary isolates of HIV-1 infect inefficiently following contact with surfaces of CD4-positive cells, and we propose that this confers a selective disadvantage during passage in rapidly dividing leukemia cell lines. Conversely, in vivo selective pressure appears to favor HIV-1 strains that require large amounts of CD4 for infection.

A Friend virus mutant that overcomes Fv-2rr host resistance encodes a small glycoprotein that dimerizes, is processed to cell surfaces, and specifically activates erythropoietin receptors.

The env gene of Friend spleen focus-forming virus (SFFV) encodes a membrane glycoprotein (gp55) that is inefficiently (3 to 5%) processed from the rough endoplasmic reticulum to form a larger dimeric plasma membrane derivative (gp55p). Moreover, the SFFV env glycoprotein associates with erythropoietin receptors (EpoR) to cause proliferation of infected erythroblasts [J.-P. Li, A. D. D'Andrea, H. F. Lodish, and D. Baltimore, Nature (London) 343:762-764, 1990]. Interestingly, the mitogenic effect of SFFV is blocked in mice homozygous for the Fv-2r resistance gene, but mutant SFFVs can overcome this resistance. Recent evidence suggested that these mutants contain partial env deletions that truncate the membrane-proximal extracellular domain of the encoded glycoproteins (M. H. Majumdar, C.-L. Cho, M. T. Fox, K. L. Eckner, S. Kozak, D. Kabat, and R. W. Geib, J. Virol. 66:3652-3660, 1992). Mutant BB6, which encodes a gp42 glycoprotein that has a large deletion in this domain, causes erythroblastosis in DBA/2 (Fv-2s) as well as in congenic D2.R (Fv-2r) mice. Analogous to gp55, gp42 is processed inefficiently as a disulfide-bonded dimer to form cell surface gp42p. Retroviral vectors with SFFV and BB6 env genes have no effect on interleukin 3-dependent BaF3 hematopoietic cells, but they cause growth factor independency of BaF3/EpoR cells, a derivative that contains recombinant EpoR. After binding 125I-Epo to surface EpoR on these factor-independent cells and adding the covalent cross-linking reagent disuccinimidyl suberate, complexes that had immunological properties and sizes demonstrating that they consisted of 125I-Epo-gp55p and 125I-Epo-gp42p were isolated from cell lysates. Contrary to a previous report, SFFV or BB6 env glycoproteins did not promiscuously activate other members of the EpoR superfamily. Although the related env glycoproteins encoded by dualtropic murine leukemia viruses formed detectable complexes with EpoR, strong mitogenic signalling did not ensue. Our results indicate that the SFFV and BB6 env glycoproteins specifically activate EpoR; they help to define the glycoprotein properties important for its functions; and they strongly suggest that the Fv-2 leukemia control gene encodes an EpoR-associated regulatory factor.

Cell surface site for mitogenic interaction of erythropoietin receptors with the membrane glycoprotein encoded by Friend erythroleukemia virus.

The membrane glycoprotein (gp55) encoded by the env gene of Friend spleen focus-forming virus (SFFV) causes mitogenesis of infected erythroblasts and is inefficiently (3-5%) processed from the rough endoplasmic reticulum (RER) to plasma membranes. Recent evidence suggested that gp55 binds to erythropoietin receptors (EpoR) in the RER, and it was proposed that this intracellular interaction causes mitogenesis (Li, J.-P., D'Andrea, A. D., Lodish, H. F., and Baltimore, D. (1990) Nature 343, 762-764). Other evidence has indicated that the plasma membrane component (gp55P) can also complex with EpoR. To identify the site of functional complexes and to study the factors that control their formation, we analyzed eight SFFV env mutants that contain in-frame deletions or linker insertions. The three nonpathogenic mutants encode gp55s that fail to leave the RER, whereas the five pathogenic mutants encode glycoproteins that occur on cell surfaces. Although BaF3 hematopoietic cells are interleukin 3 (IL-3)-dependent, a cellular derivative BaF3/EpoR that contains EpoR survives and grows in either IL-3 or erythropoietin (Epo). The five pathogenic SFFV env mutants converted BaF3/EpoR but not BaF3 cells to growth factor independence, whereas the nonpathogenic mutants did not eliminate growth factor requirements of any cells. Studies using 125I-Epo and the covalent cross-linking reagent disuccinimidyl suberate provided unambiguous evidence for ternary complexes of 125I-Epo.EpoR.gp55P on surfaces of all factor-independent cells. Size of the cell surface complex was correspondingly reduced in the case of a newly isolated SFFV mutant that encodes a severely truncated (M(r) approximately 41,000) but mitogenically active env glycoprotein. Our results support the hypothesis that activation of EpoR by the SFFV env glycoprotein occurs exclusively on cell surfaces.

Disulfide bonding controls the processing of retroviral envelope glycoproteins.

The mitogenic membrane glycoprotein (gp55) encoded by Friend erythroleukemia virus is inefficiently processed from the rough endoplasmic reticulum (RER) and only 3-5% reaches plasma membranes. Because this processed component (gp55P) contains larger and more complex oligosaccharides, it can be separated from RER gp55. In nonreducing conditions, gp55P is a unique disulfide-bonded dimer, whereas RER gp55 consists of monomers and dimers with diverse intrachain and interchain disulfide bonds. This suggests that gp55 folds heterogeneously and that only one homodimer is competent for export from the RER. Pulse-chase analyses of gp55 components labeled with radioactive amino acids indicated that formation of diverse disulfide-bonded components occurred within minutes of polypeptide synthesis and that malfolded components did not later isomerize to generate dimers competent for export from the RER. Chemical studies suggested that all 12 cysteines of gp55 were oxidized within 5 min after synthesis of the protein. In contrast, the envelope glycoprotein precursor (gPr90) encoded by a replication-competent murine leukemia virus folds more homogeneously, and it is then processed and cleaved to form an extracellular glycoprotein gp70 plus a transmembrane protein p15E. The fully processed glycoprotein contains an unoxidized cysteine sulfhydryl that isomerizes reversibly with a disulfide bond that links gp70 to p15E. Consequently, only a proportion of gp70 and p15E is disulfide-bonded, and dissociation occurs when the environment becomes even slightly reducing. The gp55 glycoprotein appears to be an extreme example of protein malfolding associated with imprecise and irreversible disulfide bonding. We discuss evidence that folding inefficiencies are common for retroviral proteins that have newly evolving pathogenic functions.

The Sfpi-1 proviral integration site of Friend erythroleukemia encodes the ets-related transcription factor Pu.1.

Previous studies identified a common site (Sfpi-1) for proviral integration in immortalized Friend erythroleukemias. cDNAs corresponding to a 1.5-kb Sfpi-1 mRNA were isolated and sequenced. These were larger than an independently isolated Sfpi-1 cDNA described by researchers from another laboratory, and they contained common differences from that sequence, including in the coding region four extra nucleotides that altered the reading frame. The properly translated protein is identical to Pu.1, a transcription activation factor that is related to the ets oncogene family. Genetic methods were used to map Sfpi-1 with respect to other loci on mouse chromosome 2. Our results suggest that Pu.1 blocks erythroblast differentiation and thereby causes immortalization.