DC-SIGN interactions with human immunodeficiency virus type 1 and 2 and simian immunodeficiency virus

Dendritic cells (DCs) efficiently bind and transmit human immunodeficiency virus (HIV) to cocultured T cells and so may play an important role in HIV transmission. DC-SIGN, a novel C-type lectin that is expressed in DCs, has recently been shown to bind R5 HIV type 1 (HIV-1) strains and a laboratory-adapted X4 strain. To characterize the interaction of DC-SIGN with primate lentiviruses, we investigated the structural determinants of DC-SIGN required for virus binding and transmission to permissive cells. We constructed a panel of DC-SIGN mutants and established conditions which allowed comparable cell surface expression of all mutants. We found that R5, X4, and R5X4 HIV-1 isolates as well as simian immunodeficiency and HIV-2 strains bound to DC-SIGN and could be transmitted to CD4/coreceptor-positive cell types. DC-SIGN contains a single N-linked carbohydrate chain that is important for efficient cell surface expression but is not required for DC-SIGN-mediated virus binding and transmission. In contrast, C-terminal deletions removing either the lectin binding domain or the repeat region abrogated DC-SIGN function. Trypsin-EDTA treatment inhibited DC-SIGN mediated infection, indicating that virus was maintained at the surface of the DC-SIGN-expressing cells used in this study. Finally, quantitative fluorescence-activated cell sorting analysis of AU1-tagged DC-SIGN revealed that the efficiency of virus transmission was strongly affected by variations in DC-SIGN expression levels. Thus, variations in DC-SIGN expression levels on DCs could greatly affect the susceptibility of human individuals to HIV infection.     

Identification of amino acid residues important for heparan sulfate proteoglycan interaction within variable region 3 of the feline immunodeficiency virus surface glycoprotein

Heparan sulfate proteoglycans (HSPGs) act as binding receptors or attachment factors for the viral envelope of many viruses, including strains of HIV and feline immunodeficiency virus (FIV). The FIV gp95 glycoprotein (SU) from laboratory-adapted strains (tissue culture adapted [TCA]) such as FIV-34TF10 can bind to HSPG, whereas SU from field strains (FS) such as FIV-PPR cannot. Previous studies indicate that SU-HSPG interactions occur within the V3 loop. We utilized a series of nested V3 peptides to further map the HSPG binding sites and found that both sides of the predicted V3 loop stem were critical for the binding but not the CXCR4 binding domain near the predicted tip of the V3 loop. Neutralization assays for TCA strain entry using the same set of V3 peptides showed that peptides targeting CXCR4 or HSPG binding sites can block infection, supporting the V3 loop as a critical neutralization target. Site-directed mutagenesis identified two highly conserved arginines, R379 and R389, on the N-terminal side of the V3 stem as critical for the contact between SU and HSPG. Residues K407, K409, K410, and K412 on the C-terminal side of the V3 stem form a second nonconserved domain necessary for HSPG binding, consistent with the observed specificity distinctions with FS FIV. Our findings discriminate structural determinants important for HSPG and CXCR4 binding by FIV SU and thus further define the importance of the V3 loop for virus entry and infection.     

DC-SIGN interactions with human immunodeficiency virus: virus binding and transfer are dissociable functions

The C-type lectins DC-SIGN and DC-SIGNR capture and transfer human immunodeficiency virus (HIV) to susceptible cells, although the underlying mechanism is unclear. Here we show that DC-SIGN/DC-SIGNR-mediated HIV transmission involves dissociable binding and transfer steps, indicating that efficient virus transmission is not simply due to tethering of virus to the cell surface.     

Enhancement of feline immunodeficiency virus infection after immunization with envelope glycoprotein subunit vaccines.

Cats were immunized three times with different recombinant feline immunodeficiency virus (FIV) candidate vaccines. Recombinant vaccinia virus (rVV)-expressed envelope glycoprotein with (vGR657) or without (vGR657 x 15) the cleavage site and an FIV envelope bacterial fusion protein (beta-Galactosidase-Env) were incorporated into immune-stimulating complexes or adjuvanted with Quil A. Although all immunized cats developed antibodies against the envelope protein, only the cats vaccinated with the rVV-expressed envelope glycoproteins developed antibodies which neutralized FIV infection of Crandell feline kidney cells. These antibodies failed to neutralize infection of thymocytes with a molecularly cloned homologous FIV. After the third immunization the cats were challenged with homologous FIV. Two weeks after challenge the cell-associated viral load proved to be significantly higher in the cats immunized with vGR657 and vGR657 x 15 than in the other cats. The cats immunized with vGR657 and vGR657 x 15 also developed antibodies against the Gag proteins more rapidly than the cats immunized with beta-Galactosidase-Env or the control cats. This suggested that immunization with rVV-expressed glycoprotein of FIV results in enhanced infectivity of FIV. It was shown that the observed enhancement could be transferred to naive cats with plasma collected at the day of challenge.     

Inhibition of human immunodeficiency virus type 1 Env-mediated fusion by DC-SIGN

DC-SIGN, a lectin expressed on dendritic cell and macrophage subsets, binds to human immunodeficiency virus Env glycoproteins, allowing capture of viral particles. Captured virions either infect target cells or are efficiently transmitted to lymphocytes. Cellular mechanisms underlying the effects of DC-SIGN remain poorly understood. Here we have analyzed the effects of DC-SIGN on viral entry and on syncytium formation induced by Env glycoproteins. The lectin enhanced susceptibility to viral infection and dramatically increased virion internalization. Captured virions accumulated in the vesicular pathway, and their access to the cytosol was altered. Strikingly, the presence of DC-SIGN on target cells inhibited their ability to form syncytia with Env-expressing cells. However, increasing CD4 surface levels on target cells alleviated this inhibitory effect of DC-SIGN. Moreover, the potency of the viral fusion inhibitor T-20 was not affected in DC-SIGN-expressing cells. Altogether, our results indicate that DC-SIGN exerts subtle and complex effects during early steps of HIV type 1 replication. DC-SIGN facilitates capture and accumulation of viral particles in a vesicular compartment and inhibits viral fusion. Competition between CD4 and DC-SIGN for Env binding likely affects virus access to the cytosol and syncytium formation.     

Structural mapping of CD134 residues critical for interaction with feline immunodeficiency virus

CD134 is a primary binding receptor for feline immunodeficiency virus (FIV), and with CXCR4 facilitates infection of CD4(+) T cells. Human CD134 fails to support FIV infection. To delineate the regions important for defining virus specificity of CD134, we exchanged domains between human and feline CD134. The binding site for FIV surface glycoprotein (SU) is located in domain 1, in a region distinct from the natural ligand (CD134L)-binding site. Mutagenesis showed that Asp60 and Asp62 are required for interaction with FIV, and modeling studies localized these two residues to the outer edge of domain 1. Substitutions S60D and N62D, in conjunction with H45S, R59G and V64K, imparted both FIV SU binding and receptor function to human CD134. Finally, we demonstrated that soluble CD134 facilitates infection of CD134(-) CXCR4(+) target cells in a manner analogous to CD4 augmentation of HIV infection.     

Mapping of the CXCR4 binding site within variable region 3 of the feline immunodeficiency virus surface glycoprotein

Feline immunodeficiency virus (FIV) shares with T-cell tropic strains of human immunodeficiency virus type 1 (HIV-1) the use of the chemokine receptor CXCR4 for cellular entry. In order to map the interaction of the FIV envelope surface unit (SU) with CXCR4, full-length FIV SU-Fc as well as constructs with deletions of extended loop L2, V3, V4, or V5 were produced in stable CHO cell lines. Binding studies were performed using these proteins on 3201 cells (CXCR4(hi) CD134(-)), with or without the CXCR4 inhibitor AMD3100. The findings established that SU binding to CXCR4 specifically requires the V3 region of SU. Synthetic peptides spanning the V3 region as well as a panel of monoclonal antibodies (MAbs) to SU were used to further map the site of CXCR4 interaction. Both the SU V3-specific antibodies and the full-length V3 peptide potently blocked binding of SU to CXCR4 and virus entry. By using a set of nested peptides overlapping a region of SU specifically recognized by CD134-dependent neutralizing V3 MAbs, we showed that the neutralizing epitope and the region required for CXCR4 binding are within the same contiguous nine-amino-acid sequence of V3. Site-directed mutagenesis was used to reveal that serine 393 and tryptophan 394 at the predicted tip of V3 are required to facilitate entry into the target cell via CXCR4. Although the amino acid sequences are not identical between FIV and HIV, the ability of FIV to bind and utilize both feline and human CXCR4 makes the feline model an attractive venue for development of broad-based entry antagonists.     

Neutralization of feline immunodeficiency virus by polyclonal feline antibody: simultaneous involvement of hypervariable regions 4 and 5 of the surface glycoprotein.

Sites involved in antibody-mediated neutralization of feline immunodeficiency virus were mapped by reciprocal exchange of envelope fragments or amino acids between molecular clones of feline immunodeficiency virus with different susceptibilities to neutralization by a polyclonal cat serum. Combinations of mutations within HV-4 or within HV-4 and HV-5 changed the susceptibility of the viruses to neutralizing antibody.     

CD4 binding site antibodies inhibit human immunodeficiency virus gp120 envelope glycoprotein interaction with CCR5

The human immunodeficiency virus type 1 (HIV-1) gp120 exterior glycoprotein is conformationally flexible. Upon binding the host cell receptor, CD4, gp120 assumes a conformation that is able to bind the chemokine receptors CCR5 or CXCR4, which act as coreceptors for the virus. CD4-binding-site (CD4BS) antibodies are neutralizing antibodies elicited during natural infection that are directed against gp120 epitopes that overlap the binding site for CD4. Recent studies (S. H. Xiang et al., J. Virol. 76:9888-9899, 2002) suggest that CD4BS antibodies recognize conformations of gp120 distinct from the CD4-bound conformation. This predicts that the binding of CD4BS antibodies will inhibit chemokine receptor binding. Here, we show that Fab fragments and complete immunoglobulin molecules of CD4BS antibodies inhibit CD4-independent gp120 binding to CCR5 and cell-cell fusion mediated by CD4-independent HIV-1 envelope glycoproteins. These results are consistent with a model in which the binding of CD4BS antibodies limits the ability of gp120 to assume a conformation required for coreceptor binding.     

Functional evaluation of DC-SIGN monoclonal antibodies reveals DC-SIGN interactions with ICAM-3 do not promote human immunodeficiency virus type 1 transmission

DC-SIGN, a type II membrane-spanning C-type lectin that is expressed on the surface of dendritic cells (DC), captures and promotes human and simian immunodeficiency virus (HIV and SIV) infection of CD4(+) T cells in trans. To better understand the mechanism of DC-SIGN-mediated virus transmission, we generated and functionally evaluated a panel of seven monoclonal antibodies (MAbs) against DC-SIGN family molecules. Six of the MAbs reacted with myeloid-lineage DC, whereas one MAb preferentially bound DC-SIGNR/L-SIGN, a homolog of DC-SIGN. Characterization of hematopoietic cells also revealed that stimulation of monocytes with interleukin-4 (IL-4) or IL-13 was sufficient to induce expression of DC-SIGN. All DC-SIGN-reactive MAbs competed with intercellular adhesion molecule 3 (ICAM-3) for adhesion to DC-SIGN and blocked HIV-1 transmission to T cells that was mediated by THP-1 cells expressing DC-SIGN. Similar but less efficient MAb blocking of DC-mediated HIV-1 transmission was observed, indicating that HIV-1 transmission to target cells via DC may not be dependent solely on DC-SIGN. Attempts to neutralize DC-SIGN capture and transmission of HIV-1 with soluble ICAM-3 prophylaxis were limited in success, with a maximal inhibition of 60%. In addition, disrupting DC-SIGN/ICAM-3 interactions between cells with MAbs did not impair DC-SIGN-mediated HIV-1 transmission. Finally, forced expression of ICAM-3 on target cells did not increase their susceptibility to HIV-1 transmission mediated by DC-SIGN. While these findings do not discount the role of intercellular contact in facilitating HIV-1 transmission, our in vitro data indicate that DC-SIGN interactions with ICAM-3 do not promote DC-SIGN-mediated virus transmission.     

Analysis of the interaction of the human immunodeficiency virus type 1 gp120 envelope glycoprotein with the gp41 transmembrane glycoprotein.

The human immunodeficiency virus type 1 (HIV-1) gp120 exterior envelope glycoprotein interacts with the viral receptor (CD4) and with the gp41 transmembrane envelope glycoprotein. To study the interaction of the gp120 and gp41 envelope glycoproteins, we compared the abilities of anti-gp120 monoclonal antibodies to bind soluble gp120 and a soluble glycoprotein, sgp140, that contains gp120 and gp41 exterior domains. The occlusion or alteration of a subset of gp120 epitopes on the latter molecule allowed the definition of a gp41 "footprint" on the gp120 antibody competition map. The occlusion of these epitopes on the sgp140 glycoprotein was decreased by the binding of soluble CD4. The gp120 epitopes implicated in the interaction with the gp41 ectodomain were disrupted by deletions of the first (C1) and fifth (C5) conserved gp120 regions. These deletions did not affect the integrity of the discontinuous binding sites for CD4 and neutralizing monoclonal antibodies. Thus, the gp41 interface on the HIV-1 gp120 glycoprotein, which elicits nonneutralizing antibodies, can be removed while retaining immunologically desirable gp120 structures.     

Caprine arthritis-encephalitis virus envelope surface glycoprotein regions interacting with the transmembrane glycoprotein: structural and functional parallels with human immunodeficiency virus type 1 gp120

A sequence similarity between surface envelope glycoprotein (SU) gp135 of the lentiviruses maedi-visna virus and caprine arthritis-encephalitis virus (CAEV) and human immunodeficiency virus type 1 (HIV-1) gp120 has been described. The regions of sequence similarity are in the second and fifth conserved regions of gp120, and the similarity is highest in sequences coinciding with beta-strands 4 to 8 and 25, which are located in the most virion-proximal region of the gp120 inner domain. A subset of this structure, formed by gp120 beta-strands 4, 5, and 25, is conserved in most or all lentiviruses. Because of the orientation of gp120 on the virion, this highly conserved virion-proximal region of the gp120 core may interact with the transmembrane glycoprotein (TM) together with the amino and carboxy termini of full-length gp120. Therefore, interactions between SU and TM of lentiviruses may be structurally related. Here we tested whether the amino acid residues in the putative virion-proximal region of CAEV gp135 comprising putative beta-strands 4, 5, and 25, as well as its amino and carboxy termini, are important for stable interactions with TM. An amino acid change at gp135 position 119 or 521, located in the turn between putative beta-strands 4 and 5 and near beta-strand 25, respectively, specifically disrupted the epitope recognized by monoclonal antibody 29A. Thus, similar to the corresponding gp120 regions, these gp135 residues are located in close proximity to each other in the folded protein, supporting the hypothesis of a structural similarity between the gp120 virion-proximal inner domain and gp135. Amino acid changes in the amino- and carboxy-terminal and putative virion-proximal regions of gp135 increased gp135 shedding from the cell surface, indicating that these gp135 regions are involved in interactions with TM. Our results indicate structural and functional parallels between CAEV gp135 and HIV-1 gp120 that may be more broadly applicable to the SU of other lentiviruses.     

Differential N-linked glycosylation of human immunodeficiency virus and Ebola virus envelope glycoproteins modulates interactions with DC-SIGN and DC-SIGNR

The C-type lectins DC-SIGN and DC-SIGNR [collectively referred to as DC-SIGN(R)] bind and transmit human immunodeficiency virus (HIV) and simian immunodeficiency virus to T cells via the viral envelope glycoprotein (Env). Other viruses containing heavily glycosylated glycoproteins (GPs) fail to interact with DC-SIGN(R), suggesting some degree of specificity in this interaction. We show here that DC-SIGN(R) selectively interact with HIV Env and Ebola virus GPs containing more high-mannose than complex carbohydrate structures. Modulation of N-glycans on Env or GP through production of viruses in different primary cells or in the presence of the mannosidase I inhibitor deoxymannojirimycin dramatically affected DC-SIGN(R) infectivity enhancement. Further, murine leukemia virus, which typically does not interact efficiently with DC-SIGN(R), could do so when produced in the presence of deoxymannojirimycin. We predict that other viruses containing GPs with a large proportion of high-mannose N-glycans will efficiently interact with DC-SIGN(R), whereas those with solely complex N-glycans will not. Thus, the virus-producing cell type is an important factor in dictating both N-glycan status and virus interactions with DC-SIGN(R), which may impact virus tropism and transmissibility in vivo.     

Sulfation of the human immunodeficiency virus envelope glycoprotein.

Sulfation is a posttranslational modification of proteins which occurs on either the tyrosine residues or the carbohydrate moieties of some glycoproteins. In the case of secretory proteins, sulfation has been hypothesized to act as a signal for export from the cell. We have shown that the human immunodeficiency virus type 1 (HIV-1) envelope glycoprotein precursor (gp160) as well as the surface (gp120) and transmembrane (gp41) subunits can be specifically labelled with 35SO42-. Sulfated HIV-1 envelope glycoproteins were identified in H9 cells infected with the IIIB isolate of HIV-1 and in the cell lysates and culture media of cells infected with vaccinia virus recombinants expressing a full-length or truncated, secreted form of the HIV-1 gp160 gene. N-glycosidase F digestion of 35SO4(2-)-labelled envelope proteins removed virtually all radiolabel from gp160, gp120, and gp41, indicating that sulfate was linked to the carbohydrate chains of the glycoprotein. The 35SO42-label was at least partially resistant to endoglycosidase H digestion, indicating that some sulfate was linked to complex carbohydrates. Brefeldin A, a compound that inhibits the endoplasmic reticulum to Golgi transport of glycoproteins, was found to inhibit the sulfation of the envelope glycoproteins. Envelope glycoproteins synthesized in cells treated with chlorate failed to incorporate 35SO42-. However, HIV glycoproteins were still secreted from cells in the presence of chlorate, indicating that sulfation is not a requirement for secretion of envelope glycoproteins. Sulfation of HIV-2 and simian immunodeficiency virus envelope glycoproteins has also been demonstrated by using vaccinia virus-based expression systems. Sulfation is a major determinant of negative charge and could play a role in biological functions and antigenic properties of HIV glycoproteins.     

Interaction of short modified peptides deriving from glycoprotein gp36 of feline immunodeficiency virus with phospholipid membranes

A tryptophan-rich octapeptide, C8 (Ac-Trp-Glu-Asp-Trp-Val-Gly-Trp-Ile-NH₂), modelled on the membrane-proximal external region of the feline immunodeficiency virus (FIV) gp36 glycoprotein ectodomain, exhibits potent antiviral activity against FIV. A mechanism has been proposed by which the peptide, being positioned on the surface of the cell membrane, inhibits its fusion with the virus. In the present work, peptide–lipid interactions of C8 with dimyristoyl phosphatidylcholine liposomes are investigated using electron spin resonance spectroscopy of spin-labelled lipids. Three other peptides, obtained from modifications of C8, have also been investigated, in an attempt to clarify the essential molecular features of the interactions involving the tryptophan residues. The results show that C8 adsorbs strongly on the bilayer surface. Membrane binding requires not only the presence of the Trp residues in the sequence, but also their common orientation on one side of the peptide that is engendered by the WX₂ WX₂ W motif. Membrane interaction correlates closely with peptide antiviral activity, indicating that the membrane is essential in stabilizing the peptide conformation that will be able to inhibit viral infection.     

DC-SIGN and CLEC-2 mediate human immunodeficiency virus type 1 capture by platelets

Platelets can engulf human immunodeficiency virus type 1 (HIV-1), and a significant amount of HIV-1 in the blood of infected individuals is associated with these cells. However, it is unclear how platelets capture HIV-1 and whether platelet-associated virus remains infectious. DC-SIGN and other lectins contribute to capture of HIV-1 by dendritic cells (DCs) and facilitate HIV-1 spread in DC/T-cell cocultures. Here, we show that platelets express both the C-type lectin-like receptor 2 (CLEC-2) and low levels of DC-SIGN. CLEC-2 bound to HIV-1, irrespective of the presence of the viral envelope protein, and facilitated HIV-1 capture by platelets. However, a substantial fraction of the HIV-1 binding activity of platelets was dependent on DC-SIGN. A combination of DC-SIGN and CLEC-2 inhibitors strongly reduced HIV-1 association with platelets, indicating that these lectins are required for efficient HIV-1 binding to platelets. Captured HIV-1 was maintained in an infectious state over several days, suggesting that HIV-1 can escape degradation by platelets and might use these cells to promote its spread. Our results identify CLEC-2 as a novel HIV-1 attachment factor and provide evidence that platelets capture and transfer infectious HIV-1 via DC-SIGN and CLEC-2, thereby possibly facilitating HIV-1 dissemination in infected patients.     

Binding of recombinant feline immunodeficiency virus surface glycoprotein to feline cells: role of CXCR4, cell-surface heparans, and an unidentified non-CXCR4 receptor

To address the role of CXCR4 in the cell-surface attachment of the feline immunodeficency virus (FIV), a soluble fusion protein, gp95-Fc, consisting of the surface glycoprotein (SU, gp95) of either a primary (PPR) or cell line-adapted (34TF10) FIV strain was fused in frame with the Fc domain of human immunoglobulin G1. The recombinant SU-immunoadhesins were used as probes to investigate the cellular binding of FIV SU. In agreement with the host cell range properties of both viruses, binding of 34TF10 gp95-Fc was observed for all cell lines tested, whereas PPR gp95-Fc bound only to primary feline T cells. 34TF10 gp95-Fc also bound to Jurkat and HeLa cells, consistent with the ability of FIV to use human CXCR4 as a fusion receptor. As expected, 34TF10 gp95-Fc binding to Jurkat cells was blocked by addition of stromal cell-derived factor 1alpha (SDF-1alpha), as was binding to the 3201 feline lymphoma cell line. However, SDF-1alpha, RANTES, macrophage inflammatory protein 1beta, and heparin all failed to inhibit the binding of either gp95-Fc to primary T cells, suggesting that a non-CXCR4 receptor is involved in the binding of FIV SU. In this regard, an unidentified 40-kDa protein species from the surface of primary T cells but not Jurkat and 3201 cells specifically coprecipitated with both gp95-Fc. Yet another type of binding of 34TF10 gp95-Fc to adherent kidney cells was noted. SDF-1alpha failed to block the binding of 34TF10 gp95-Fc to either HeLa, Crandel feline leukemia, or G355-5 cells. However, binding was severely impaired in the presence of soluble heparin, as well as after enzymatic removal of surface heparans or on cells deficient in heparan expression. These overall findings suggest that in addition to CXCR4, a non-CXCR4 receptor and cell-surface heparans also play an important role in FIV gp95 cell surface interactions on specific target cells.     

Structural analysis of the principal immunodominant domain of the feline immunodeficiency virus transmembrane glycoprotein.

In the transmembrane envelope glycoprotein (TM) of lentiviruses, including human immunodeficiency virus type 1 (HIV-1) and feline immunodeficiency virus (FIV), two cysteine residues, conserved in most retroviruses, are thought to form a loop containing five to seven amino acids. These elements make up a B-cell epitope recognized by nearly 100% of sera from infected patients or animals, designated the principal immunodominant domain (PID). The PID amino acid sequences are highly conserved between isolates of the same lentivirus but are unrelated, except for the two cysteines, when divergent lentiviruses are compared. The aim of this study was to analyze the relationship between amino acid sequence in the PID and envelope function. We introduced two kinds of mutations in the PID of FIV: mutations which impeded the formation of a loop and mutations which substituted the sequence of FIV with the corresponding sequences from other lentiviruses, HIV-1, visna virus, and equine infectious anemia virus. We analyzed antibody recognition, processing, and fusogenic properties of the modified envelopes, using two methods of Env expression: a cell-free expression system and transfection of a feline fibroblast cell line with gag-pol-deleted FIV proviruses. Most mutations in the PID of FIV severely affected envelope processing and abolished syncytium formation. Only the chimeric envelope containing the HIV-1 PID sequence was correctly processed and maintained the capacity to induce syncytium formation, although less efficiently than the wild-type envelope. We computed three-dimensional structural models of the PID, which were consistent with mutagenesis data and confirmed the similarity of FIV and HIV-1 PID structures, despite their divergence in amino acid sequence. Considering these results, we discussed the respective importance of selection exerted by functional requirements or host antibodies to explain the observed variations of the PIDs in lentiviruses.     

Cell surface effects of human immunodeficiency virus

Cell killing by human immunodeficiency virus (HIV) is thought to contribute to many of the defects of the acquired immunodeficiency syndrome (AIDS). Two types of cytopathology are observed in HIV-infected cultured cells: cell-cell fusion and killing of single cells. Both killing processes appear to involve cell surface effects of HIV. A model is proposed for the HIV-mediated cell surface processes which could result in cell-cell fusion and single cell killing. The purpose of this model is to define the potential roles of individual viral envelope and cell surface molecules in cell killing processes and to identify alternative routes to the establishment of persistently-infected cells. Elucidation of HIV-induced cell surface effects may provide the basis for a rational approach to the design of antiviral agents which are selective for HIV-infected cells.