The Nef-Like Effect of Murine Leukemia Virus Glycosylated Gag on HIV-1 Infectivity Is Mediated by Its Cytoplasmic Domain and Depends on the AP-2 Adaptor Complex

ABSTRACT

Human immunodeficiency virus type 1 (HIV-1) Nef enhances the infectivity of progeny virions. However, Nef is dispensable for the production of HIV-1 virions of optimal infectivity if the producer cells are superinfected with certain gammaretroviruses. In the case of the ecotropic Moloney murine leukemia virus (M-MLV), the Nef-like effect is mediated by the glycosylated Gag (glycoGag) protein. We now show that the N-terminal intracellular domain of the type II transmembrane protein glycoGag is responsible for its effect on HIV-1 infectivity. In the context of a fully active minimal M-MLV glycoGag construct, truncations of the cytoplasmic domain led to a near total loss of activity. Furthermore, the cytoplasmic domain of M-MLV glycoGag was fully sufficient to transfer the activity to an unrelated type II transmembrane protein. Although the intracellular region of glycoGag is relatively poorly conserved even among ecotropic and xenotropic MLVs, it was also fully sufficient for the rescue of nef-deficient HIV-1 when derived from a xenotropic virus. A mutagenic analysis showed that only a core region of the intracellular domain that exhibits at least some conservation between murine and feline leukemia viruses is crucial for activity. In particular, a conserved YXXL motif in the center of this core region was critical. In addition, expression of the μ2 subunit of the AP-2 adaptor complex in virus producer cells was essential for activity. We conclude that the ability to enhance HIV-1 infectivity is a conserved property of the MLV glycoGag cytoplasmic domain and involves AP-2-mediated endocytosis.

IMPORTANCE The Nef protein of HIV-1 and the entirely unrelated glycosylated Gag (glycoGag) protein of a murine leukemia virus (MLV) similarly enhance the infectiousness of HIV-1 particles by an unknown mechanism. MLV glycoGag is an alternative version of the structural viral Gag protein with an extra upstream region that provides a cytosolic domain and a plasma membrane anchor. We now show for the first time that the cytosolic domain of MLV glycoGag contains all the information needed to enhance HIV-1 infectivity and that this function of the cytosolic domain is conserved despite limited sequence conservation. Within the cytosolic domain, a motif that resembles a cellular sorting signal is critical for activity. Furthermore, the enhancement of HIV-1 infectivity depends on an endocytic cellular protein that is known to interact with such sorting signals. Together, our findings implicate the endocytic machinery in the enhancement of HIV-1 infectivity by MLV glycoGag.

INTRODUCTION

Human immunodeficiency virus type 1 (HIV-1) and all other primate lentiviruses encode a small myristylated protein called Nef. Although Nef is not required and often even dispensable for virus replication in cell culture, it is crucial for maintaining high virus loads in vivo, and consequently for pathogenicity (1–5). Nef engages clathrin-dependent endocytic machinery and induces the endocytosis and degradation of the viral receptor CD4 (6–9). Among several other cell surface proteins whose trafficking is affected by Nef are major histocompatibility class I molecules, which are downregulated by Nef to protect infected cells from recognition and lysis by cytotoxic T cells (10–14). Nef also modulates T cell antigen receptor signaling (15–19), interferes with T cell chemotaxis by inhibiting the remodeling of the actin cytoskeleton (20, 21), and triggers the release of extracellular vesicles (22–24). Furthermore, the Nef proteins of certain simian immunodeficiency viruses counteract the restriction factor BST2 (25, 26), which tethers budded virions to the cell surface (27).

In addition to its effects on the host cell environment, Nef enhances the specific infectivity of progeny virions (28–30). The enhancement of HIV-1 infectivity by Nef is impaired in cells depleted of clathrin or dynamin 2, indicating that the effect depends on the endocytosis or altered intracellular trafficking of a host factor (31). The effect of Nef on HIV-1 infectivity may be related to its effects on the morphology and function of the endocytic recycling compartment, because both depend on a dileucine-based sorting signal in Nef (32, 33). In support of this notion, the endocytic recycling regulator EHD4 has recently been implicated in the infectivity enhancement function of Nef (34). When CD4 is overexpressed, Nef can enhance HIV-1 infectivity by downregulating CD4, because high CD4 surface levels can sequester the HIV-1 envelope (Env) glycoproteins and prevent their incorporation into virions (35). However, the infectivity enhancement function depends on Nef residues that are dispensable for CD4 downregulation (36). Furthermore, Nef enhances the infectivity of HIV-1 virions even when these are produced in cells that lack CD4 or that express a CD4 that is resistant to downregulation by Nef (29, 37, 38).

HIV-1 virions contain only small amounts of Nef (39–41), and there is evidence indicating that the incorporation of Nef into virions is not relevant for its infectivity enhancement function (42). Thus, Nef appears to act on another component of the virion, although it has no evident effect on virus morphogenesis (30, 43). Several lines of evidence indicate that one virion component affected by Nef is HIV-1 Env. First, Nef does not enhance the infectivity of HIV-1 particles pseudotyped with the glycoprotein of vesicular stomatitis virus (44, 45). Second, although HIV-1 virus-cell fusion appears unaffected (46), Nef enhances the entry of HIV-1 capsids into the cytosol (47, 48). Third, the effect of Nef on the infectivity of virions pseudotyped with the Env proteins of different HIV-1 strains varies considerably (49, 50). Finally, Nef decreases the sensitivity of HIV-1 virions to neutralization by antibodies against the membrane-proximal external region of the transmembrane glycoprotein gp41 (49), which suggests that Nef affects the conformation of the gp41 ectodomain in the assembled Env trimer.

Remarkably, Nef is dispensable for optimal HIV-1 infectivity in cells infected with various gammaretroviruses (51). Many gammaretroviruses encode a glycosylated Gag (glycoGag) protein in addition to Gag, and in the case of Moloney murine leukemia virus (M-MLV), the rescue of HIV-1 infectivity was indeed shown to be mediated by glycoGag (51). M-MLV glycoGag is translated from a noncanonical initiation codon upstream of and in-frame with the gag gene (52). Its use results in the translation of a Gag molecule with an N-terminal extension that causes its membrane insertion with a type II orientation, where the N terminus remains in the cytosol and Gag forms a glycosylated extracellular domain (53). Like Nef, glycoGag is more important for virus replication in vivo than in cell culture (54–57). In murine cells, M-MLV glycoGag can enhance viral budding or release (58, 59). In contrast, no effect on virus release was observed for human T cell lines in which glycoGag potently enhanced HIV-1 infectivity (51). M-MLV glycoGag also counteracts the restriction factor APOBEC3 (60, 61). However, a recent study indicates that M-MLV glycoGag can also robustly enhance infectivity through an APOBEC-independent mechanism (62).

Although glycoGag does not downregulate CD4 (51), its effect on HIV-1 infectivity resembles that of Nef in several aspects. The effects of the two proteins are similarly determined by Env, even though neither Nef nor glycoGag affect the incorporation of Env into virions (49–51). Their effects on infectivity also exhibit a comparable dependence on the type of cell used for virus production and are particularly pronounced in T cell lines (51). Furthermore, both proteins exert their effects in producer cells, and in both cases, these effects become manifest in target cells at a very early stage of the replication cycle (51).

Most of the extracellular domain of M-MLV glycoGag appears dispensable for the rescue of nef-deficient HIV-1 (51), indicating that the region upstream of Gag accounts for the activity. In the present study, we provide evidence indicating that the Nef-like activity of glycoGag is exerted exclusively by its cytoplasmic domain. We also show that the ability of the glycoGag cytoplasmic domain to enhance HIV-1 infectivity is maintained among ecotropic and xenotropic MLVs in spite of relatively poor overall sequence conservation and is markedly dependent on a tyrosine residue within a conserved YXXL motif. Furthermore, the Nef-like activity of M-MLV glycoGag depends on the μ2 subunit of the AP-2 adaptor complex, which is known to bind to tyrosine-based sorting signals (63).

MATERIALS AND METHODS

Plasmids.

The env- and nef-deficient HIV-1 provirus HXB/Env⁻/Nef⁻ has been described previously (64). NL4-3/nef⁻ is a nef-deficient variant of HIV-1_NL4-3 (31). NL4-3/glycoMA has nucleotides (nt) 360 to 926 of M-MLV (GenBank accession number J02255) followed by a stop codon inserted between the Nef initiation codon and the unique XhoI site within the nef gene of pNL4-3. The HIV-1 Env expression vector pSVIIIenv, the xenotropic MLV Env expression vector pCMV-Xenogp85, and enhanced green fluorescent protein (EGFP)-Rab7A (Addgene plasmid 28047) have been described (65–67). To generate vectors expressing wild-type (WT) glycoMA (a C-terminally HA-tagged version of a fully active form of glycoGag) and the Δ1-16 (numbers indicate the range of truncated residues), Δ1-32, and Δ1-42 cytoplasmic domain truncation mutants, DNAs encoding residues 2 to 190, 17 to 190, 33 to 190, and 43 to 190 of M-MLV glycoGag preceded by a Kozak sequence and an ATG initiation codon and followed by a hemagglutinin (HA) tag and a stop codon were amplified from pNCA (68) and cloned into the mammalian expression vector pBJ5. The N25A mutation targets residue 25 of the matrix (MA) domain of WT glycoMA and was inserted by site-directed mutagenesis. For confocal imaging, the pBJ5-based vector encoding wild-type (WT) glycoMA was modified by inserting a sequence encoding a Thr-Gly-Ala-Gly linker followed by mCherry and a stop codon immediately 3′ of the glycoMA-HA coding sequence. DNA encoding the human asialoglycoprotein receptor 1 (AR) with a C-terminal HA tag was amplified from BC032130 (Open Biosystems) and cloned into pBJ5. DNAs encoding hybrid proteins with C-terminal HA tags were generated using an overlap extension PCR method (69) and also cloned into pBJ5. The templates used were pNCA (ecotropic MLV) and BC032130 (AR) or NZB-9-1 (xenotropic MLV) (70) and BC03210. The gCD-TM/AR and XgCD-TM/AR hybrid proteins have the intracellular and transmembrane regions of M-MLV or NZB-9-1 glycoGag (residues 2 to 85) fused to the AR extracellular region (residues 61 to 291). The gCD/AR and XgCD/AR hybrid proteins have the intracellular region of M-MLV or NZB-9-1 glycoGag (residues 2 to 66) fused to the AR transmembrane and extracellular regions (residues 41 to 291). The vector expressing full-length glycoGag is based on pBJ5 and has nt 360 to 2234 of the M-MLV genome (GenBank accession number J02255) inserted between an ATG initiation codon and a sequence encoding an HA tag. Deletions and/or point mutations were introduced into the gCD/AR, gCD-TM/AR, and glycoGag constructs by site-directed mutagenesis.

Pseudovirion production and infectivity analysis.

HIV-1 pseudovirions capable of a single round of replication were produced by transfecting Jurkat T antigen (TAg) cells with Lipofectamine 2000 (Invitrogen). The cells were cotransfected with HXB/Env⁻/Nef⁻, a vector expressing Env_HXB2, and either a vector expressing WT or mutant glycoMA or glycoGag, a vector expressing AR or a glycoGag/AR hybrid protein, or empty pBJ5 at equivalent molar amounts. Supernatants were harvested 2 days posttransfection, clarified by low-speed centrifugation, filtered through 0.45-μm-pore-size filters, and then used immediately to infect TZM-bl indicator cells (71) (2 × 10⁵) in triplicate in T25 flasks. Aliquots of the supernatants were frozen for subsequent p24 antigen quantitation by a standard enzyme-linked immunosorbent assay (ELISA). Three days postinfection, the indicator cells were lysed in reporter lysis buffer supplied with the β-galactosidase (β-Gal) enzyme assay system (Promega), and β-Gal activity was measured according to the manufacturer's instructions. Values were normalized for the amount of p24 antigen present in the supernatants used for infection.

Conditional knockdown of AP2M1.

293T cells were cotransfected with pHDMHgpm2 encoding codon-optimized HIV-1 Gag-Pol, pTRIPZ-based doxycyline (Dox)-inducible lentiviral vectors encoding a nontargeting control shRNA (Thermo Scientific) or a short hairpin RNA (shRNA) targeting the AP-2 μ2 subunit (AP2M1; clone ID, V2THS_172835; Thermo Scientific), along with vectors expressing HIV-1 Rev, HIV-1 Tat, and the vesicular stomatitis virus G protein. Filtered supernatants were then used to transduce MOLT-4 clone 8 cells, followed by selection with 1 μg/ml puromycin. Dox was used at 2 μg/ml to induce shRNA expression in puromycin-resistant bulk cultures. The effect of Dox on AP2M1 expression was examined after 3 days of induction. To examine the role of AP2M1 in the infectivity of progeny virions, MOLT-4 clone 8 cells stably transduced with TRIPZ lentiviral vectors were kept in medium containing 2 μg/ml Dox after infection with NL4-3/nef⁻ or NL4-3/glycoMA. Progeny virus was harvested after 4 days of induction and used to infect TZM-bl cells after normalization for p24 antigen. To limit virus transmission to a single round, infections were done in the presence of the HIV-1 protease inhibitor saquinavir (2 μM). After a 12-h incubation period, the input virus was removed, and the infected cells were kept in the presence of saquinavir (2 μM) and AMD-3100 (5 μM) (72) to prevent both virus spreading and Env-mediated cell-cell fusion. Virus-induced β-Gal activity in the TZM-bl indicator cells was measured 3 days postinfection.

Western blot analysis.

The expression levels of AR-HA and of the glycoGag/AR-HA hybrid proteins, and the expression levels of WT and mutant glycoGag-HA, in the Jurkat TAg cells used for virus production were compared by Western blotting with anti-HA antibody HA.11 (Covance). To facilitate the detection of the glycoMA-HA constructs, appropriate expression vectors (1 μg) were transfected into 293T cells using calcium phosphate, and the cell lysates were examined by Western blotting with HA.11.

To examine the incorporation of the gCD/AR hybrid protein into HIV-1 virions, Jurkat TAg cells were transfected either with a full-length HIV-1 provirus harboring a disrupted gag gene (73) or with HXB/Env⁻/Nef⁻ (1 μg), along with pSVIIIenv (0.5 μg) and a vector expressing WT or mutant gCD/AR-HA (0.3 μg). Virions released by the transfected cells were pelleted through 20% sucrose cushions, and pelletable material was analyzed by Western blotting with the anti-HIV capsid (CA) antibody 183-H12-5C (74) and with the anti-HA antibody HA.11.

The presence of MLV CA in culture supernatants was detected with a goat anti-Rauscher MLV p30 antiserum (Quality Biotech; lot 78S221). AP2M1 in cell lysates was detected with anti-AP50 (BD Transduction Laboratories).

Confocal imaging.

U2-OS cells (2.5 × 10⁴) were seeded on 12-mm round glass coverslips (#1.5 thickness) in 24-well plates. The next day, the cells were transfected with pBJ5-glycoMA-mCherry (0.1 μg), either alone or together with a vector expressing EGFP-Rab7A (0.1 μg). One day posttransfection, the cells were fixed with 2% paraformaldehyde for 10 min at room temperature, washed in phosphate-buffered saline (PBS), and mounted in Vectashield antifade with 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories). Confocal microscopy was performed using a Solamere Technology Group (Salt Lake City, UT) CSU10B spinning disk confocal system attached to a Nikon TE2000-E2 motorized inverted fluorescence microscope (Nikon Instruments, Melville, NY). For confocal imaging, a ×100 oil objective was used, and cells were illuminated with lasers at 405 nm, 488 nm, or 561 nm.

RESULTS

Effects of cytoplasmic domain truncations on the Nef-like activity of a minimal glycoGag construct.

In a previous study, a minimal glycoGag molecule with a truncated extracellular Gag domain was as capable of enhancing HIV-1 infectivity as full-length M-MLV (51). The minimal glycoGag molecule retained the N-terminal cytoplasmic domain, the transmembrane region, and most of the MA domain. We used a C-terminally HA-tagged version of this fully active form, here called glycoMA, to examine the role of its cytoplasmic domain in the enhancement of HIV-1 infectivity. To determine the contribution of membrane-distal cytoplasmic domain regions that differ in their level of conservation (see Fig. 4A), we removed 16, 32, or 42 residues from the N terminus of the glycoMA molecule (Fig. 1A).

FIG 4.

A relatively conserved 30-amino-acid region within the M-MLV glycoGag cytoplasmic domain is crucial for its ability to enhance HIV-1 infectivity. (A) Sequence alignment of the cytoplasmic domains of the feline leukemia virus FeLV-61E (89), M-MLV, and NZB-9-1 glycoGag proteins and location of deletions introduced into the gCD/AR hybrid protein. Mutants whose infectivity enhancement activity was reduced less than 5-fold are indicated by solid lines. Mutants whose infectivity enhancement activity was reduced more than 30-fold are indicated by double lines. (B) Expression levels of AR and of WT and mutant versions of the gCD/AR protein in the virus-producing cells examined by Western blotting with anti-HA antibody. (C) Effects of the hybrid proteins on the infectivities of nef-deficient HIV-1 virions analyzed as in Fig. 1.

FIG 1.

Effects of cytoplasmic domain truncations on the Nef-like activity of a minimal glycoGag construct. (A) Schematic illustration of the glycoMA construct and of the cytoplasmic domain (CD) truncation mutants examined. TM, transmembrane region. (B) Expression levels and electrophoretic mobilities of the WT and mutant glycoMA proteins expressed in 293T cells and analyzed by Western blotting with anti-HA (α-HA) antibody. (C) Effects of the CD truncations on the ability of glycoMA to enhance the infectivity of nef-deficient HIV-1 virions. Pseudovirions produced in the presence of the indicated versions of glcyoMA were normalized for p24 antigen and analyzed on TZM-bl indicator target cells. Bars indicate standard deviations.

As shown in Fig. 1B, the parental glycoMA construct and the three truncation mutants were expressed at comparable levels. In each case, a major species and several additional bands were visible. Because glycoMA retains one of the glycosylation sites of glycoGag (75), it is expected to become N-glycosylated if it assumes a type II transmembrane topology. Indeed, the major species and a ladder of minor bands above it disappeared when Asn25 in the MA domain of glycoMA was replaced by Ala (Fig. 1B, lane 7). Unexpectedly, only the mobility of the major glycosylated species and of the unglycosylated form was affected by the truncations (Fig. 1B, lanes 2 to 5). We also note that the mobility of these two species appeared anomalous in the case of the Δ1-32 mutant. Similarly, we have previously observed that the effects of C-terminal truncations on the electrophoretic mobility of an HIV-1 Gag construct do not necessarily correlate with their effects on calculated molecular weight (76).

The mutants were then examined for their ability to rescue the infectivity of nef-deficient HIV-1. Viral stocks were produced by transfecting Jurkat TAg cells, because it has been shown that the effects of both Nef and glycoGag on HIV-1 infectivity are most pronounced in T cell lines (51). Pseudovirions capable of a single round of replication were produced by transfecting Jurkat TAg cells with an env- and nef-deficient HIV-1 provirus, a vector providing HIV-1_HXB2 Env in trans, and vectors expressing glycoMA or its mutants (1 μg each). The relative infectivities of the virus stocks, normalized for p24 content, were compared by measuring Tat-induced β-galactosidase activity in TZM-bl target cells. In agreement with previous results (51), glycoMA had a robust effect on the specific infectivity of nef-deficient HIV-1, which was enhanced almost 30-fold in the experiment shown in Fig. 1C. The Δ1-16 truncation mutant exhibited a comparable effect on HIV-1 infectivity, even though a slightly larger fraction of the mutant remained unglycosylated and thus appeared to lack a type II transmembrane topology. However, the Δ1-32 mutant, which lacked about half of the cytoplasmic domain, had a much smaller (2-fold) effect on the specific infectivity of nef-deficient HIV-1. Finally, the Δ1-42 mutant exhibited no activity in the same assay (Fig. 1C). Of note, the unglycosylated form of the poorly active Δ1-32 mutant was no more prominent than that of the highly active Δ1-16 mutant (Fig. 1B, lanes 3 and 4), indicating that the differences in activity were not due to different topologies. We conclude that the effect of glycoMA on HIV-1 infectivity is absolutely dependent on its N-terminal cytoplasmic domain, even though a significant portion of the cytoplasmic domain that shows no conservation among murine and feline leukemia viruses (see Fig. 4A) is dispensable.

The cytoplasmic domain of glycoGag is sufficient to confer the ability to enhance HIV-1 infectivity.

In a previous study, a minimal M-MLV glycoGag molecule that retained only 10 residues of the extracellular domain maintained a significant ability to rescue the infectivity of nef-deficient HIV-1 (51). To determine whether activity in this assay depends on any extracellular glycoGag residues, we entirely replaced the extracellular domain of M-MLV glycoGag by the corresponding domain of an unrelated type II transmembrane protein. As the fusion partner, we used the asialoglycoprotein receptor 1 (AR), for which there is direct evidence that its N terminus faces the cytoplasm whereas its C terminus is extracellular (77). There is also direct evidence that the transmembrane orientation of AR is determined primarily by charged residues flanking the hydrophobic transmembrane region (78). This result was predicted by the charge difference method (79), which also predicts that the gCD-TM/AR fusion protein illustrated in Fig. 2A maintains the type II orientation assumed by glycoGag itself.

FIG 2.

The cytoplasmic domain of M-MLV glycoGag potently enhances the infectivity of nef-deficient HIV-1 when fused to an unrelated protein. (A) Schematic illustration of the AR-based hybrid proteins examined. (B) Expression levels of AR and of the indicated hybrid proteins in virus-producing cells (after transfection of 1 μg of each expression vector), and their effects on the infectivities of nef-deficient HIV-1 virions. LU, light units. (C) Effects of the gCD/AR hybrid protein at lower expression levels. For comparison, 1 μg of the vector expressing the gCD-TM/AR construct was transfected.

The gCD-TM/AR fusion protein harbors the cytoplasmic domain and the transmembrane region of M-MLV glycoGag, whereas the ectodomain is entirely derived from AR. To determine its effect on HIV-1 infectivity, 1 μg of a vector encoding the gCD-TM/AR fusion protein with a C-terminal HA tag was transfected into Jurkat TAg cells together with env- and nef-deficient HIV-1 and a vector providing HIV-1_HXB2 Env. In parallel transfections, the vector encoding the fusion protein was replaced by a vector encoding AR with a C-terminal HA tag or by the empty vector. Immunoblotting of the cell lysates with anti-HA antibody revealed that the steady-state levels of the gCD-TM/AR fusion protein were dramatically lower than those of unmodified AR cloned into the same expression vector. In spite of its very low expression levels, the gCD-TM/AR fusion protein caused a 31-fold increase in the specific infectivity of virions produced by nef-deficient HIV-1 (Fig. 2B). In contrast, unmodified AR had no effect in this assay (Fig. 2B). We thus conclude that the remarkably high specific activity of the gCD-TM/AR fusion protein is entirely conferred by the cytosolic and transmembrane regions of M-MLV glycoGag.

To determine whether the transmembrane domain of M-MLV glycoGag is required for activity, we generated the gCD/AR fusion protein illustrated in Fig. 2A. In contrast to the original gCD-TM/AR construct, the gCD/AR version retains only the cytoplasmic domain of M-MLV glycoGag, and the transmembrane and extracellular domains are both from HA-tagged AR. As shown in Fig. 2B, the cellular steady-state levels obtained with the gCD-TM/AR construct were much higher than those observed for the gCD-TM/AR version. Since the two fusion proteins differed only in the transmembrane region, these data imply that the cysteine-rich transmembrane region of M-MLV glycoGag was responsible for the very poor expression of the gCD-TM/AR fusion protein. Importantly, the gCD/AR fusion protein enhanced the specific infectivity of nef-deficient HIV-1 34-fold in the experiment shown in Fig. 2B. Although this effect was dose dependent, as little as 30 ng of the gCD/AR construct still led to a 15-fold enhancement in infectivity (Fig. 2C). Thus, the transfer of the cytoplasmic domain of M-MLV glycoGag was sufficient to transfer the ability to rescue the infectivity of nef-deficient HIV-1.

The ability to enhance HIV-1 infectivity is conserved among the divergent cytoplasmic domains of ecotropic and xenotropic glycoGags.

A previous study indicates that several gammaretroviruses can rescue the infectivity of nef-deficient HIV-1 when present in producer cells, including xenotropic MLV and gibbon ape leukemia virus (51). However, except in the case of M-MLV, it is not known whether glycoGag is responsible for the rescue. Notably, the cytoplasmic domain of glycoGag is rather poorly conserved even among relatively closely related species of gammaretroviruses (see Fig. 4A). For instance, the cytoplasmic domain of glycoGag is only about 65% identical between the ecotropic M-MLV and the xenotropic MLV isolate NZB-9-1. Given the relatively poor sequence conservation of the non-Gag portion of glycoGag, we asked whether the ability to rescue nef-deficient HIV-1 is a unique property of the cytoplasmic domain of M-MLV glycoGag or is more widely conserved.

To examine whether the non-Gag portion of a xenotropic glycoGag can affect HIV-1 infectivity, we generated vectors encoding the XgCD-TM/AR and XgCD/AR hybrid proteins illustrated in Fig. 3A. The XgCD-TM/AR hybrid protein harbors the cytoplasmic domain and the transmembrane region of a xenotropic glycoGag, whereas the XgCD/AR hybrid protein harbors only the cytoplasmic domain. Transfection of these vectors (1 μg each) into Jurkat TAg cells showed that the XgCD-TM/AR and XgCD/AR hybrid proteins were expressed at comparable levels (Fig. 3B). Of note, the XgCD-TM/AR hybrid protein was expressed at much higher levels than the equivalent gCD-TM/AR hybrid protein (Fig. 3B), which is identical except that the glycoGag portion is from the ecotropic M-MLV. The XgCD-TM/AR and XgCD/AR hybrid proteins enhanced the infectivity of nef-deficient HIV-1 48- and 69-fold, respectively (Fig. 3B). In the same experiment, the very poorly expressed gCD-TM/AR hybrid protein still had a 38-fold effect on HIV-1 infectivity (Fig. 3B). Similarly, an experiment in which the vectors expressing the XgCD-TM/AR and XgCD/AR chimeras were titrated down revealed that only minute quantities of these hybrid proteins were required to robustly enhance HIV-1 infectivity (Fig. 3C). We conclude that the cytoplasmic domains of ecotropic and xenotropic glycoGags share the ability to dramatically enhance HIV-1 infectivity.

FIG 3.

The ability of the glycoGag cytoplasmic domain to enhance HIV-1 infectivity is conserved among ecotropic and xenotropic MLVs. (A) Schematic illustration of the AR-based hybrid proteins examined. The XgCD-TM/AR and gCD-TM/AR hybrid proteins harbor the intracellular and transmembrane domains of xenotropic and ecotropic glycoGag, respectively. The XgCD/AR hybrid protein harbors only the intracellular domain of xenotropic glycoGag. (B) Expression levels of AR and of the hybrid proteins in virus-producing cells (after transfection of 1 μg of each expression vector), and their effects on the infectivities of nef-deficient HIV-1 virions. (C) Effects of the XgCD-TM/AR and XgCD/AR hybrid proteins at lower expression levels.

GlycoGag cytoplasmic domain regions required for the enhancement of HIV-1 infectivity.

The gCD/AR construct, which harbors only the cytoplasmic domain of M-MLV glycoGag, was used to map the cytoplasmic domain residues responsible for the enhancement of HIV-1 infectivity. Consecutive deletions were generated throughout the glycoGag cytoplasmic domain (Fig. 4A). Except for the Δ57-64 mutation, the deletions only moderately affected the expression levels of the gCD/AR hybrid protein (Fig. 4B). The low steady-state levels of the Δ57-64 mutant may have been a consequence of aberrant membrane insertion, because the Δ57-64 mutation removes two basic residues flanking the transmembrane region that are expected to be crucial for the topology of the hybrid protein (79).

The Δ1-16 and Δ17-24 mutations removed membrane-distal cytoplasmic domain sequences that show no evident conservation among murine and feline leukemia viruses (Fig. 4A). As expected, the Δ1-16 mutation, which did not significantly affect the activity of the glycoMA construct (Fig. 1), also had no significant effect in the context of the gCD/AR hybrid protein. Similarly, the Δ17-24 mutation reduced activity in the rescue assay only about 2-fold. In marked contrast, the consecutive Δ25-32, Δ33-40, Δ41-48, and Δ49-54 mutations, which targeted regions that are at least partially conserved among murine and feline leukemia viruses (Fig. 4A), each largely abolished the activity of gCD/AR (Fig. 4C). The Δ55-56 mutation, which specifically removed a membrane-proximal dileucine motif that is only conserved among MLVs (Fig. 4A), caused only a modest (less than 2-fold) reduction in activity (Fig. 4C). Finally, the Δ57-64 mutant, which lacked the conserved membrane-proximal region of the glycoGag cytoplasmic domain (Fig. 4A), also retained significant activity (Fig. 4C). Indeed, when the differences in expression levels are taken into account, the activity of the Δ57-64 mutant appeared to be comparable to that of WT gCD/AR. Together, these results identify a relatively well-conserved core region encompassing about 30 residues that is crucial for the ability to rescue nef-deficient HIV-1.

Role of a conserved YXXL motif.

The critical 30-amino-acid region of the M-MLV glycoGag cytoplasmic domain contains a central Y₃₆XXL₃₉ motif that is conserved among murine and feline leukemia viruses (Fig. 4A) and is located in a region identified as required for infectivity enhancement by the Δ33-40 mutation (Fig. 4C). Since conserved YXXL motifs in retroviral proteins can serve as docking sites for cellular proteins (76, 80–82), we examined whether Tyr36 and Leu39 in the M-MLV glycoGag cytoplasmic domain are important for the ability of the gCD/AR hybrid protein to enhance HIV-1 infectivity. In parallel, we examined the role of vicinal proline residues that, while only conserved among MLVs (Fig. 4A), are also in a region identified as absolutely essential for infectivity enhancement (Fig. 4C). As shown in Fig. 5A and B, the Y₃₆A, L₃₉A, and PP_50,51AA mutations did not affect the overall expression levels of the gCD/AR hybrid protein. However, the Y₃₆A mutation nearly abolished its ability to enhance HIV-1 infectivity (Fig. 5A), and the L₃₉A mutation reduced this activity between 5- and 10-fold (Fig. 5A and B). In contrast, the PP_50,51AA mutation had only a moderate effect (Fig. 5A).

FIG 5.

Role of a conserved YXXL motif within the M-MLV glycoGag cytoplasmic domain. (A) Expression levels of WT and mutant versions of the gCD/AR hybrid protein in virus-producing Jurkat TAg cells (after transfection of 1 μg of each expression vector) and their effects on the infectivities of nef-deficient HIV-1 virions. (B) Effects of smaller amounts of WT gCD/AR and of the L₃₉A mutant on HIV-1 infectivity. (C) Association of WT gCD/AR and of the Y₃₆A mutant with HIV-1 virions. Jurkat TAg cells were transfected with1 μg of a gag-deficient full-length HIV-1 provirus (lane 1) or HXB/Env⁻/Nef⁻ (lanes 2 and 3), along with pSVIIIenv (0.5 μg) and vectors expressing WT or mutant gCD/AR (0.3 μg). Purified viral particles and the cell lysates were analyzed by Western blotting with anti-HA and anti-HIV CA as indicated. (D) Effect of WT gCD/AR and of the Y₃₆A mutant on MLV infectivity. Jurkat TAg cells were cotransfected with a vector expressing MLV Gag-Pol (1 μg), a MLV LTR-GFP reporter (1 μg), pCMV-Xenogp85 (0.5 μg), and vectors expressing AR, WT gCD/AR, or Y₃₆A gCD/AR (0.5 μg). Aliquots of the culture supernatants were examined by Western blotting with anti-MLV p30 to compare the amounts of mature MLV CA released by the transfected cells. The infectivities of the released MLV pseudovirions for HT1080 cells were compared by counting GFP-expressing cells. (E) Effect of the Y₃₆A mutation in the context of full-length M-MLV glycoGag. Jurkat TAg cells were cotransfected with HXB/Env⁻/Nef⁻ (1 μg), pSVIIIenv (0.5 μg), and vectors expressing WT M-MLV glycoGag-HA or a version with the Y₃₆A mutation (0.03 μg). The glycoGag-HA expression levels in the transfected cells and the infectivities of the released pseudovirions for TZM-bl cells are shown.

Western blotting of purified virion preparations indicated that only a trace amount of the gCD/AR hybrid protein was incorporated into the Jurkat TAg-derived HIV-1 pseudovirions (Fig. 5C, lane 2). However, the Y₃₆A mutation selectively increased the incorporation of two minor forms of gCD/AR that migrated as diffuse bands (Fig. 5C, lane 3). These minor species were not detected in cells expressing WT gCD/AR (lane 2) but were visible in cells expressing the Y₃₆A mutant (lane 3), indicating that they were stabilized by the Y₃₆A mutation.

Since it has been demonstrated that glycoGag enhances the infectivity of Jurkat TAg-derived MLV virions pseudotyped with a xenotropic Env for HT1080 cells (51), we also examined the effects of the gCD/AR hybrid protein and of the Y₃₆A mutant in this experimental system. We found that the parental gCD/AR construct enhanced the specific infectivity of MLV virions bearing a xenotropic Env for HT1080 cells about 14-fold, whereas the Y₃₆A mutant was nearly inactive (Fig. 5D).

Finally, we examined the effect of the Y₃₆A mutation on the ability of full-length M-MLV glycoGag to enhance the infectivity of nef-deficient HIV-1 virions. In the experiment shown in Fig. 5E, WT M-MLV glycoGag-HA had a 68-fold effect on HIV-1 infectivity. The Y₃₆A mutant version was expressed at comparable levels but enhanced HIV-1 infectivity less than 8-fold. We thus conclude that Y₃₆ makes a critical contribution to the infectivity enhancement function of M-MLV glycoGag.

The effect of glycoGag on HIV-1 infectivity depends on AP-2.

YXXL motifs within the cytoplasmic domains of transmembrane proteins can serve as endocytosis signals that physically interact with the μ2 subunit of the AP-2 adapter complex (63, 83). Consistent with the notion that glycoGag is endocytosed, confocal microscopy showed that glycoMA with mCherry attached to its C-terminal ectodomain exhibits a punctate perinuclear distribution indicative of entrapment in an endosomal compartment (Fig. 6A). Furthermore, a subpopulation of glycoMA-mCherry puncta clearly coincided with the lumen of vesicles that were decorated with the late endosomal marker EGFP-Rab7A (Fig. 6B). However, the Y₃₆A mutation in its cytoplasmic domain did not significantly alter the overall subcellular localization of glycoMA-mCherry (data not shown).

FIG 6.

Confocal microscopy of U2-OS cells expressing (A) glycoMA-mCherry (red) or (B) glycoMA-mCherry (red) together with EGFP-Rab7A (green) and stained with the nuclear dye DAPI (blue). Arrows in the insets mark vesicular structures decorated with EGFP-Rab7A that surround puncta formed by glycoMA-mCherry.

Because our virion incorporation data suggested that the Y₃₆A mutation may selectively affect the trafficking of a minor species, we determined directly whether AP-2 is involved in the activity of glycoMA. To this end, we transduced MOLT-4 clone 8 T lymphoid cells with lentiviral vectors encoding a Dox-inducible nontargeting shRNA (sh_Ctrl-i cells) or an shRNA targeting the μ2 subunit of AP-2 (sh_AP2M1-i cells). Whereas Dox had no effect on AP2M1 expression levels in sh_Ctrl-i cells, as expected (data not shown), AP2M1 could be silenced in sh_AP2M1-i cells in a Dox-dependent manner (Fig. 7A). Both cell lines were infected with NL4-3/nef⁻ or NL4-3/glycoMA and kept in the presence of Dox to induce shRNA expression. Progeny virus was then harvested at a time when AP2M1 was efficiently silenced in the sh_AP2M1-i cells. As shown in Fig. 7B, glycoMA enhanced the specific infectivity of virions derived from sh_Ctrl-i cells about 100-fold but had essentially no effect on the infectivity of virions derived from AP2M1-depleted sh_AP2M1-i cells. We conclude that AP-2 is essential for the effect of M-MLV glycoGag on virus infectivity.

FIG 7.

AP-2 is required for the effect of glycoGag on HIV-1 infectivity. (A) Effect of a Dox-inducible shRNA on the expression levels of AP2M1 in MOLT-4 clone 8 cells. (B) Dox-induced silencing of AP2M1 in MOLT-4 clone 8 cells infected with NL4-3/nef⁻ or NL4-3/glycoMA dramatically inhibits the effect of glycoMA on the single-cycle infectivity of progeny virions.

DISCUSSION

A recent study demonstrated that several gammaretroviruses rescue the infectivity of nef-deficient HIV-1 when both viruses are coexpressed in producer cells (51). Specifically, Nef was dispensable in producer cells infected with an ecotropic MLV, a xenotropic MLV, or a gibbon ape leukemia virus (51). The rescue of nef-deficient HIV-1 by the ecotropic M-MLV was shown to be dependent on the capacity to express glycoGag, but whether glycoGag also accounted for the Nef-like effect of the other gammaretroviruses was not examined (51).

In the case of M-MLV glycoGag, most of the extracellular domain was shown not to be strictly required to rescue the infectivity of nef-deficient HIV-1 (51), indicating that this activity is mediated by the cytoplasmic domain and/or the transmembrane domain of ecotropic glycoGag. In the present study, we show that the cytoplasmic domain is both essential and sufficient to confer a Nef-like activity. In the context of a fully active minimal glycoGag molecule, truncations that removed 50% or more of its cytoplasmic domain essentially abolished the ability to enhance the infectivity of nef-deficient HIV-1. It is unlikely that the loss of activity was due to aberrant membrane insertion, because a truncation mutant that retained robust activity exhibited exactly the same glycosylation pattern as an inactive mutant with a more extensive truncation of the cytoplasmic domain.

In a hybrid protein context, the presence of the cytoplasmic domain of ecotropic glycoGag was sufficient for the ability to robustly enhance HIV-1 infectivity. The effect on HIV-1 infectivity was not further enhanced when the transmembrane region of ecotropic glycoGag was also included in the hybrid protein context, but the latter construct was very poorly expressed and thus appeared to possess a very high specific activity. However, our plasmid titration experiments show that the presence of the cytoplasmic domain of ecotropic glycoGag alone also conferred a remarkably high specific activity.

Even between ecotropic and xenotropic MLVs, the non-Gag portion of glycoGag is only poorly conserved. Nevertheless, we find that the cytoplasmic domains of ecotropic and xenotropic glycoGag proteins can both mediate a robust enhancement of HIV-1 infectivity in a hybrid protein context. Expression levels and activity in the infectivity rescue assay remained comparable when the transmembrane region of xenotropic glycoGag was included together with its cytoplasmic domain. These observations support the notion that the transmembrane region of glycoGag does not play a significant role in the rescue of HIV-1 infectivity. We also conclude that the dramatic negative effect of the transmembrane region of ecotropic glycoGag on hybrid protein steady-state levels did not reflect a property that is conserved among ecotropic and xenotropic MLVs.

Our deletion analysis indicates that only a core region of the glycoGag cytoplasmic domain that shows at least some conservation between murine and feline leukemia viruses is critical for the infectivity enhancement activity. Conserved membrane-proximal residues were expected to be important for topology based on the charge-difference rule (79). Indeed, a mutant that lacked these residues was expressed at exceptionally low levels, consistent with the possibility that proper membrane insertion was required for stability. However, the mutant retained considerable activity in the infectivity rescue assay, indicating that only a trace amount of the glycoGag cytoplasmic domain needs to be in the correct orientation to significantly affect HIV-1 infectivity.

The infectivity enhancement function of Nef depends on clathrin and dynamin 2 (31), and on a dileuine-based sorting signal that connects Nef to the clathrin-dependent endocytic machinery and is required for its localization to coated pits (32, 84–86). It has also been reported that some Nef proteins interact with clathrin-associated adaptor complexes through tyrosine-based endocytosis signals of the YXXØ type (87). We have therefore examined the roles of potential sorting motifs within regions of the glycoGag cytoplasmic domain that are conserved between ecotropic and xenotropic MLVs. We found that a tyrosine residue (Y₃₆) within a conserved region that fits the YXXØ motif was essential for the ability of the gCD/AR hybrid protein to enhance the infectivities of HIV-1 and MLV pseudovirions. Furthermore, Y₃₆ was critical for the infectivity enhancement activity of full-length M-MLV glycoGag.

Y₃₆ was not required for the punctate perinuclear localization of the glycoMA construct. However, Y₃₆ may primarily affect the trafficking of a specific pool of glycoGag molecules, because its absence led to the selective accumulation of two minor forms of the gCD/AR hybrid construct in HIV-1 virions. Indeed, our results show that the μ2 subunit of AP-2, which interacts with YXXØ-type sorting signals (63), is essential for the activity of a minimal glycoGag molecule.

Publicly available sequences indicate that the cytoplasmic domain of glycoGag exhibits virtually no primary sequence similarity between more distantly related gammaretroviruses. In particular, the YXXØ motif is not universally conserved. However, we note that the YXXØ motif is present in the glycoGag proteins of feline leukemia viruses (88), whose intracellular regions otherwise exhibit very limited sequence similarity to those of MLV glycoGag proteins. Together, our observations strongly suggest that the infectivity enhancement activity exhibited by at least some of the highly divergent glycoGag proteins, like the enhancement of HIV-1 infectivity by Nef itself, depends on the engagement of cellular sorting machinery.