Sequences in Gibbon Ape Leukemia Virus Envelope That Confer Sensitivity to HIV-1 Accessory Protein Vpu^,

Abstract

HIV-1 efficiently forms pseudotyped particles with many gammaretrovirus glycoproteins, such as Friend murine leukemia virus (F-MLV) Env, but not with the related gibbon ape leukemia virus (GaLV) Env or with a chimeric F-MLV Env with a GaLV cytoplasmic tail domain (CTD). This incompatibility is modulated by the HIV-1 accessory protein Vpu. Because the GaLV Env CTD does not resemble tetherin or CD4, the well-studied targets of Vpu, we sought to characterize the modular sequence in the GaLV Env CTD required for this restriction in the presence of Vpu. Using a systematic mutagenesis scan, we determined that the motif that makes GaLV Env sensitive to Vpu is INxxIxxVKxxVxRxK. This region in the CTD of GaLV Env is predicted to form a helix. Mutations in the CTD that would break this helix abolish sensitivity to Vpu. Although many of these positions can be replaced with amino acids with similar biophysical properties without disrupting the Vpu sensitivity, the final lysine residue is required. This Vpu sensitivity sequence appears to be modular, as the unrelated Rous sarcoma virus (RSV) Env can be made Vpu sensitive by replacing its CTD with the GaLV Env CTD. In addition, F-MLV Env can be made Vpu sensitive by mutating two amino acids in its cytoplasmic tail to make it resemble more closely the Vpu sensitivity motif. Surprisingly, the core components of this Vpu sensitivity sequence are also present in the host surface protein CD4, which is also targeted by Vpu through its CTD.

INTRODUCTION

Human immunodeficiency virus type 1 (HIV-1), like many viruses, is capable of assembling infectious viral particles using the surface glycoproteins from foreign viruses by a process termed pseudotyping. However, not all virus/glycoprotein pairs are able to complement one another. HIV-1 is compatible with glycoproteins from many families of viruses, including rhabdoviruses, other retroviruses, and filoviruses, but the compatibility does not strictly follow family lines (14, 17, 29). For instance, HIV-1 is compatible with the Env glycoprotein from the gammaretrovirus Friend murine leukemia virus (F-MLV), but it is not compatible with the Env glycoprotein from gibbon ape leukemia virus (GaLV), even though F-MLV and GaLV belong to the same genus (6, 21, 24, 39). GaLV is a gammaretrovirus found in captive gibbon apes. It is closely related to a retrovirus found in wild koalas (koala retrovirus [KoRV]), but both viruses are believed to be fairly recent introductions that likely were derived from endogenous mouse retroviruses (reviewed in reference 42). The F-MLV and GaLV Env glycoproteins display 48% identity at the amino acid level. Both proteins have a native molecular weight of ∼85 kDa, and both are cleaved by a cellular protease into the 70-kDa surface (SU) and 15-kDa transmembrane (TM) domains, which remain associated after cleavage. Both the F-MLV and GaLV TM domains are additionally cleaved in their cytoplasmic tail domain (CTD) into a 12-kDa (p12E) and a 2-kDa peptide (p2, or R-peptide) by the virus-encoded protease during the viral assembly process (13, 32). This R-peptide cleavage is required for the viral glycoproteins to become fusogenically active (32). The component of GaLV Env that causes the incompatibility with HIV-1 has been mapped to its CTD (6, 38). Recently, we and others demonstrated that the incompatibility of HIV-1 with glycoproteins containing the CTD from GaLV Env is dictated by the HIV-1 accessory protein Vpu (7, 21). In the presence of Vpu, GaLV Env CTD containing glycoproteins are prevented from being incorporated into HIV-1 particles, whereas deletion of Vpu restores incorporation of these glycoproteins and infectivity of the resulting HIV-1 particles. The mechanism for this GaLV Env exclusion is not known, although it has been suggested that this may be affected by difference in trafficking of Env in the presence of Vpu (7).

Vpu is an 81-amino-acid HIV-1 protein that contains an N-terminal membrane-spanning domain followed by an ∼50-amino-acid cytoplasmic tail (40). Vpu is unique to HIV-1 and a few closely related lentiviruses. The first and most widely studied function of Vpu is to promote the degradation of the host surface protein CD4, the primary receptor for HIV-1. Since HIV-1 Env can bind to CD4 during transit through the endoplasmic reticulum (ER), binding can result in the proteins being sequestered in the ER. This can result in severe impairment to viral propagation and has been thought to be a major reason for Vpu's role in CD4 degradation (19, 41, 48). The C-terminal cytoplasmic domain of Vpu interacts with the CD4 cytoplasmic tail; consequently, the E3 ubiquitin ligase complex bearing β-TrCP is recruited to CD4 (23). CD4 is subsequently ubiquitinated and degraded by the proteasome (23, 34, 49). β-TrCP is critical for this function, and the phosphoserine residues in positions 52 and 56 of Vpu are required for β-TrCP recruitment and for CD4 degradation (36).

Vpu also enhances viral release by modulating the host defense protein tetherin (also known as BST-2, CD317, or HM1.24) (27, 45). Tetherin is an alpha interferon-induced antiviral protein that contains an N-terminal membrane-spanning domain and a C-terminal glycophosphatidylinositol anchor that physically tethers enveloped viruses to the infected cell's surface after release. Human tetherin expression at the cell surface is efficiently modulated by Vpu, resulting in enhanced virus release (27, 45). Unlike with CD4, recognition of tetherin by Vpu appears to be facilitated through the membrane-spanning domains of the two proteins (10, 31). Vpu with a scrambled membrane-spanning domain cannot enhance viral release or modulate tetherin activity (30, 35, 44). Tetherin antagonism by Vpu has been reported to be β-TrCP dependent (8, 25) and has been reported to involve the sequestration of tetherin in the trans-Golgi network or the endo-lysosomal compartments (9). In addition to modulating tetherin and CD4, Vpu has also been reported to degrade or modulate other host cellular proteins, including TASK-1, major histocompatibility complex class I (MHC-I), MHC-II, CD1d, and NTB-A (11–12, 16, 26, 37).

GaLV and HIV-1 infect different primates and belong to different genera of retroviruses. The reason for the antagonism of GaLV Env by HIV-1 Vpu is not clear. We hypothesize that the proteins containing the GaLV Env CTD resemble human proteins that are normally targeted by Vpu and are mistakenly recognized by Vpu and excluded from viral particles. The possibility that the resemblance between GaLV Env and the natural target of Vpu has a functional significance cannot be excluded. An alanine-scanning mutagenesis strategy combined with a single-round infectivity assay in the presence or absence of Vpu was performed to identify the precise amino acid sequence in the GaLV Env CTD that confers Vpu sensitivity. The primary protein sequence conferring sensitivity to Vpu was further validated by showing that additional viral glycoproteins could be made sensitive to Vpu by mutating select amino acid residues to match the identified Vpu-sensitivity sequence.

MATERIALS AND METHODS

Plasmids.

The ecotropic F-MLV Env (isolate 57) expression construct was kindly provided by Walther Mothes (Yale University). The F-MLV/GaLV chimeric Env construct containing the full-length form of GaLV Env CTD was constructed using oligonucleotide linkers coding for the cytoplasmic tail of GaLV. These oligonucleotides were inserted between the ClaI site (encoded within the NRL amino acid coding sequence, 30 amino acids upstream of the C terminus as shown in Fig. 1) and the EcoRI site that occurs after the stop codon in the cytoplasmic tail region of the F-MLV Env expression construct. Plasmids expressing the truncated forms of the F-MLV/GaLV Env (i.e., lacking 4, 8, 9, 10, 11, or 12 amino acids at the C terminus) were constructed by introducing a stop codon at the appropriate position on the linkers used to replace the fragment between ClaI and EcoRI. Mutant chimeric F-MLV/GaLV Env-expressing constructs were made in the context of the Δ8 F-MLV/GaLV chimeric Env (lacking 8 amino acids at the C terminus) and were constructed by site-directed mutagenesis of the fragment between ClaI and EcoRI. Two silent mutations were made, from CTCATT to CTGATC, to encode LII amino acids 48 residues upstream of the C terminus in the transmembrane region to introduce a BclI site into the region encoding the transmembrane domain of F-MLV Env. For mutations in the transmembrane region or the membrane-proximal cytoplasmic domain of the chimeric F-MLV/GaLV Env, the fragment between BclI and EcoRI was replaced by PCR, and the respective mutations were introduced through the primers. The order of the amino acid mutations in this study is indicated by numbering from the C terminus inward. Rous sarcoma virus (RSV) Schmidt-Ruppin A Env with GaLV CTD was created by a two-step PCR with oligonucleotides amplifying the sequences from the RSV Env transmembrane domain and ectodomain and the GaLV Env CTD. The fragment of DNA between EcoRI (encoding GIP) and SacII (after the stop codon) on the RSV Env expression plasmid was replaced by the PCR product. The RSV/GaLV Env fusion sequence was CLPC/ILNR. The C-terminal 27 amino acids in the vesicular stomatitis virus G protein (VSV-G) were deleted, and the amino acids LRV were silently mutated to introduce a unique MluI restriction site. The GaLV Env CTD, amplified by PCR, was introduced into the VSV-G-encoding plasmid between MluI (encoding LRV) and BamHI (after the stop codon) sites to construct a VSV-G/GaLV chimeric Env. The VSV-G/GaLV Env fusion sequence was IHLCI/LNRLV. The influenza hemagglutinin (HA) and neuraminidase (NA) expression constructs pEWSN-HA and pCAGGS-WNA15, respectively, were kindly provided by Yoshihiro Kawaoka (University of Wisconsin—Madison). Influenza HA with GaLV Env CTD was created by a two-step PCR and replacement of the fragment between BstXI (encoding ASSLV) and XhoI (after the stop codon). The HA/GaLV Env fusion sequence was LGAIS/QFIND. The sequences of the inserted region in all the constructs were confirmed by sequencing. NL4-3-derived HIV-cytomegalovirus (CMV)-green fluorescent protein (GFP) was kindly provided by Vineet KewalRamani (National Cancer Institute [NCI]-Frederick). This proviral vector lacks the genes encoding Vif, Vpr, Vpu, Nef, and Env and has a CMV immediate-early promoter-driven GFP in the place of Nef. The Vpu⁺ HIV-CMV-GFP was created by replacing the fragment between BamHI and SalI sites in HIV-CMV-GFP from the equivalent BamHI-SalI fragment encoding Vpu from the plasmid ΔR8.2 (51).

Fig. 1.

(A) Schematic of the F-MLV and GaLV Env proteins. The amino acid sequences in the CTD of F-MLV Env, GaLV Env, and CD4 are depicted. The arrow indicates the point at which the F-MLV Env CTD was exchanged for the equivalent sequence from GaLV Env CTD in the F-MLV/GaLV chimeric Env construct. The vertical lines indicate the alignment of the amino acid residues between the CTD of F-MLV Env and GaLV Env or between GaLV Env and CD4. (B) Schematic of the single-round infectivity assay. HIV-CMV-GFP (±Vpu) is cotransfected with different Env constructs into 293FT cells. Supernatant is applied to fresh 293T mCAT-1 cells, and infected cells are quantitated by flow cytometry. Projected on the plot is the side scatter (y axis) against the GFP fluorescence (x axis). Vpu sensitivity is quantified as a ratio of infectivity (percent cells infected) from provirus without Vpu divided by the infectivity from a parallel assay using a provirus containing Vpu. Relative infectivity is quantified as a ratio of infectivity in the absence of Vpu to infectivity with wild-type F-MLV Env in the same experiment.

Cell culture.

The 293FT cell line was obtained from Invitrogen. The cell line expressing the ecotropic F-MLV Env receptor, 293T mCAT-1, was kindly provided by Walther Mothes. The 293T TVA cell line expressing the receptor for Rous sarcoma virus (RSV) Env was provided by John Young (Scripps Research Institute) (20). All three cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 2 mM glutamine, 1 mM sodium pyruvate, 10 mM nonessential amino acids, and 0.5 mg/ml G418. In phases of cell culture involving transfection or transduction, G418 was not added to the medium.

Infectivity/Vpu sensitivity assays.

Infectivity assays using HIV-CMV-GFP and its derivative were performed by transfection of 293FT cells with 500 ng of the HIV proviral DNA and 500 ng of the Env or the mutant Env expression construct using 3 μl of FuGene 6 (Roche) or 3 μl of 1 mg/ml polyethylenimine (PEI) (3) in six-well plates. When influenza HA or its derivatives were used, 293FT cells were transfected with 500 ng of the provirus, 250 ng of the HA expression vector or its derivative, and 250 ng of the NA expression vector. The media were replaced 16 to 24 h posttransfection in the case of FuGene 6 and 4 to 6 h posttransfection in the case of PEI to remove any residual transfection reagent. Supernatant was collected 24 h after the media were exchanged and then frozen at −80°C for at least 2 h to lyse any cells in the supernatant. The supernatant was thawed in a 37°C water bath and spun at 2,500 × g for 10 min to pellet any cells or cell debris. For viruses pseudotyped with F-MLV Env or influenza HA, 1 ml of the supernatant was applied to fresh 293T mCAT-1 cells. For the transduction of RSV Env pseudotyped viruses, 1 ml of the supernatant was applied to fresh 293T TVA cells. Cells were collected 48 h later, fixed with 4% paraformaldehyde, and analyzed by flow cytometry using a FACScan or Accuri C6 flow cytometer system. Data were analyzed using FlowJo software (version 7.5.5; Tree Star). Vpu sensitivity is expressed as a ratio of the percentage of cells infected with the HIV-1 from provirus lacking Vpu to the percentage of cells infected with the HIV-1 from provirus containing Vpu (Fig. 1B). Relative infectivity is expressed as a ratio of percentage of cells infected with Vpu⁻ HIV-1 pseudotyped with a mutant Env to the percentage of infectivity with Vpu⁻ HIV-1 pseudotyped with F-MLV Env in the same experiment (Fig. 1B).

Western blotting.

Transfections for Western blots were performed as described for infectivity assays. Viral supernatants were spun at 13,200 × g for 2 h. The pellets were resuspended in 2× SDS-PAGE loading buffer, and the equivalent of 1 ml of viral supernatant was analyzed by 10% discontinuous SDS-PAGE. Cells were pelleted and resuspended in 1× SDS-PAGE loading buffer, and approximately 2% of the total amount of cells was analyzed in parallel with the viral supernatants. Proteins were transferred manually onto a 0.22-μm polyvinylidene difluoride (PVDF) membrane. The membranes were blocked with 5% nonfat dry milk and probed with goat anti-MLV Env gp70 (kindly provided by Alan Rein, NCI-Frederick) diluted 1:10,000 and anti-HIV p24 hybridoma medium diluted 1:500 (HIV-1 p24 hybridoma [183-H12-5C]; obtained through the AIDS Research and Reference Reagent program, Division of AIDS, NIAID, NIH) from Bruce Chesebro (5) and anti-HIV MA hybridoma medium diluted 1:5 (HB-8975; MH-SVM33C9; obtained through the ATCC) (33). Blots were then probed with horseradish peroxidase (HRP)-conjugated anti-goat antibody diluted 1:10,000 or anti-mouse antibody diluted 1:10,000, both from Sigma. Luminata Crescendo Western HRP substrate from Millipore was used for visualization of the membranes with a chemiluminescence image analyzer, LAS3000 from Fujifilm. Blots were also probed with infrared (IR) dye 700DX-labeled anti-mouse IgG and visualized using an Odyssey infrared imaging system from LI-COR biosciences.

RESULTS

F-MLV and GaLV are both gammaretroviruses, and the Env proteins of these viruses are closely related. The amino acids sequences of the TM domains of these two proteins are 65% identical, and both proteins contain a conserved viral protease cleavage site that removes the C-terminal R-peptide. The R-peptide itself is less conserved than the rest of Env and differs in length by one amino acid (Fig. 1A). In spite of these similarities, an F-MLV Env that contains the GaLV Env CTD is incompatible with HIV-1 proviruses that contain Vpu, while wild-type F-MLV Env is compatible. This incompatibility typically causes a 50- to 100-fold reduction in the number of infectious HIV-1 particles produced (21). Vpu's action on GaLV Env appears mechanistically similar to its action on the host protein CD4. GaLV Env and CD4 contain short CTDs that dictate Vpu sensitivity. The modulation of CD4 and GaLV Env by Vpu is abolished by mutation of the serine residues at positions 52 and 56 in Vpu (21, 36), but it is not affected by mutations in the membrane-spanning domain of Vpu (21, 49). However, the CTD of GaLV Env is similar to the CTD of F-MLV but very dissimilar from the CTD of CD4 (Fig. 1A), and yet GaLV Env but not F-MLV Env is sensitive to Vpu. In addition, while Vpu has typically been thought to target CD4 for degradation, the action on GaLV appears distinct. Vpu prevents GaLV Env CTD-containing F-MLV Env from forming infectious particles with HIV-1 but not with F-MLV (6, 21), which suggests that the mechanism of restriction is not based simply on protein degradation. These data suggest that the recognition sequence and modulation mechanism of GaLV Env by Vpu is distinct from that of CD4.

The boundary for Vpu sensitivity lies around the tenth amino acid from the C terminus.

To identify the precise sequence required for modulation of GaLV Env by Vpu, we first defined the C-terminal boundary that confers Vpu sensitivity. An F-MLV/GaLV Env chimera containing the C-terminal 29 amino acid CTD from GaLV Env was engineered to lack its C-terminal 4, 8, or 12 amino acids (denoted Δ4, Δ8, and Δ12) (Fig. 2). The F-MLV/GaLV chimera was used so that the only differences between the Vpu-insensitive F-MLV Env and the Vpu-sensitive F-MLV/GaLV Env were in the CTD. Infectivity experiments were performed in parallel with Vpu⁺ and Vpu⁻ HIV proviruses using a 1:1 (wt/wt) ratio of proviral DNA to Env. This ratio of transfection yielded high infectivity and at least a 10-fold reduction in infectivity in the presence of Vpu with Env proteins that contained the GaLV CTD. A Vpu insensitive Env, F-MLV Env, was used as a control to ensure that observed differences were due to Vpu. In all cases, infectivity with F-MLV Env was essentially identical with the Vpu⁺ and Vpu⁻ proviruses. The F-MLV/GaLV Env Δ4 and Δ8 proteins alone remained sensitive to Vpu; however, this sensitivity was abolished with the Δ12 truncation. Curiously, F-MLV/GaLV Env chimeras missing 8 to 12 amino acids at the C terminus produced severalfold more infectious particles than full-length F-MLV/GaLV Env (Fig. 2B). Full-length F-MLV/GaLV Env was incorporated into viral particles less efficiently than wild-type F-MLV Env, F-MLV/GaLV Env Δ8, or F-MLV/GaLV Env Δ12 (see Fig. S1A in the supplemental material). Reduced incorporation of full-length GaLV Env into HIV particles, even in the absence of Vpu, has been noted previously (7). The reduction in incorporation also correlated with loss in infectivity with these mutant Envs but did not affect Vpu sensitivity (Fig. 2B) (see Fig. S1B in the supplemental material). To further define the Vpu sensitivity boundary, F-MLV/GaLV chimeric Env proteins lacking 9, 10, or 11 amino acids from the C terminus were created. The sensitivity toward Vpu was reduced stepwise with 9 or 10 amino acids being removed. Upon removal of 11 amino acids, the sensitivity toward Vpu was abolished (Fig. 2A and B). Hence, the boundary of the region sensitive to Vpu is the cytoplasmic tail extending to the Tyr at the tenth position from the C terminus.

Fig. 2.

HIV-1 Vpu modulates infectivity with F-MLV/GaLV Env. (A) Infectivity plots are as described for Fig. 1, and the percentage of infected cells is indicated in the dot plots. (B) Scheme of truncated mutants. “X” represents a deleted amino acid, and “-” represents an unchanged amino acid. Vpu sensitivity and relative infectivity have been calculated as in Fig. 1 and are the averages from two experiments. Mean relative infectivity ± standard deviation (SD) from the experiments is shown.

Vpu sensitivity motif.

To identify the amino acids required for sensitivity toward Vpu, an alanine-scanning mutagenesis was performed, starting from the minimal sensitive protein, F-MLV/GaLV Env Δ8. Initially, five amino acids at a time were mutated to alanine in the CTD of the chimeric Env. F-MLV/GaLV Env Δ8 containing mutations to alanine at positions (from the C terminus) 9 to 13 (designated 9-13A), 14-18A, and 24-28A formed infectious particles but were resistant to Vpu (see Fig. S2 in the supplemental material). The construct with mutations 19-23A, however, did not form infectious particles in the presence or absence of Vpu. We fine-tuned the mutagenesis by mutating only two amino acids at a time to alanine and observed that 11-12A, 13-14A, 15-16A, 17-18A, 19-20A, 21-22A, and 25-26A formed infectious particles but were not sensitive to Vpu. The exceptions were 23-24A and 27-28A, which did not produce infectious particles in the presence or absence of Vpu, and 9-10A, which produced infectious particles and remained sensitive to Vpu (see Fig. S3 in the supplemental material).

Finally, the amino acids in the CTD of F-MLV/GaLV Env Δ8 were mutated individually. Of the 30 amino acids mutated, eight independently abolished sensitivity to Vpu. Mutants were scored as “Vpu insensitive” if the presence of Vpu reduced the infectivity by <4-fold. The Vpu sensitivity-conveying residues were spread over a 16-residue segment beginning 26 amino acids from the C terminus and conveyed Vpu sensitivity through the motif INxxIxxVKxLxxRxK (Fig. 3). Mutation of amino acids V29, R31, N32, and L33 to alanine abolished the production of infectious particles with HIV-1 in the presence and absence of Vpu and therefore cannot be excluded as being required for Vpu sensitivity. Several other point mutations altered the infectivity of particles produced, but there was no obvious correlation between changes that affect Vpu sensitivity and changes that affect infectivity (Fig. 3). The “sensitivity motif” is composed largely of hydrophobic amino acids and a few positive charges but is devoid of proline and glycine residues, which is consistent with an alpha-helical structure.

Fig. 3.

Scanning mutagenesis of the GaLV CTD. Mutants are depicted as in Fig. 2. A letter at a particular position represents mutation of the amino acid in that position. Vpu sensitivity and relative infectivity have been calculated as for Fig. 1 and are the averages from two experiments for mutations in amino acid residues 32 through 35, from three experiments for mutations of amino acids 11, 12, 13, 15, 16, 23, 27, 29, 30, 31, and 38, from four experiments for mutations of residues 18, 19, 20, 22, and 26, from five experiments for mutations of residues 10, 17, 21, 24, 25, 28, 36, 37, and 39, from 7 experiments for mutation of amino acid residue 14, and from 31 experiments for the Δ8 construct. NA, not applicable. Mean relative infectivity ± SD from the experiments is shown.

Modularity of the Vpu sensitivity motif-containing GaLV CTD.

The GaLV Env CTD is required, but not necessarily sufficient, to confer Vpu sensitivity. Additional requirements for Vpu sensitivity may be present in F-MLV and GaLV Envs. To determine if this sensitivity is necessary and sufficient for mediating Vpu sensitivity, the CTD of RSV Env, HA, and VSV-G were replaced with the Vpu sensitivity motif from GaLV Env. Replacement of the CTDs of VSV-G and HA resulted in chimeric proteins unable to form infectious particles with HIV-1 in the presence or absence of Vpu (data not shown). However, replacement of the RSV CTD with the Vpu sensitivity motif from GaLV resulted in a chimera that produced infectious particles with HIV-1 and was sensitive to Vpu (Fig. 4). Wild-type RSV Env could pseudotype HIV-1 particles efficiently and remained Vpu insensitive. Hence, sensitivity toward Vpu can be conferred on other proteins by exchanging the CTD with the GaLV Env CTD, but determination of whether the protein can form infectious particles depends on additional factors.

Fig. 4.

RSV/GaLV Env is Vpu sensitive. CTD amino acid sequences of the indicated Envs are depicted. Vpu sensitivity with F-MLV Env, Δ8, and Δ12 Envs are the averages from four experiments, those from for RSV/GaLV Env are the averages from three experiments, and those for for RSV Env are the averages from two experiments. Mean relative infectivity ± SD from the experiments is shown.

Sensitivity toward Vpu requires a putative alpha helix.

The frequency and positioning of hydrophobic amino acids in the GaLV Env CTD is consistent with an alpha-helical domain. Secondary structure predictors indicate that the CTDs of F-MLV Env and GaLV Env are alpha-helical with an unstructured chain of nine amino acids at the C terminus (4, 15, 43). The putative alpha helix was disrupted to determine if this structure is important for Vpu sensitivity. Glycine and proline have a low propensity to fall into an alpha helix (28). Several point mutations were made throughout the Vpu sensitivity sequence by changing individual amino acid residues to proline. Mutations of six different amino acids to proline in a stretch of 16 amino acids each abolished sensitivity toward Vpu (Fig. 5A). In each case, HIV-1 particles pseudotyped with these mutant Envs displayed viral infectivity equivalent to that of the parent protein, suggesting the mutations did not alter normal protein function. These data are consistent with an alpha helix in the CTD of GaLV Env being required for Vpu sensitivity but not for regular infectivity function of Env in cell culture.

Fig. 5.

Vpu sensitivity requires a predicted alpha helix. (A) Amino acid sequences are indicated as in Fig. 2. Vpu sensitivity and relative infectivity are calculated as for Fig. 1 and are the averages from three experiments for mutations of residues 11, 14, and 17, from five experiments for mutation of residue 23, from eight experiments for mutation of residues 20 and 26, and from 11 experiments for F-MLV and Δ8 Envs. Mean relative infectivity ± SD from the experiments is shown. (B) Helix representation of GaLV Env CTD is depicted, starting from L33 at position d and ending at Y10 at position f. Amino acids required for Vpu sensitivity are underlined.

If the GaLV Env CTD is represented as an alpha helix, then all amino acids in the sensitivity motif, with the exception of R13, align on one side of the helix. Positions a and d in the helix contain largely hydrophobic amino acids and hence may be the interaction face of a coiled-coil structure (Fig. 5B). In fact, F-MLV Env is known to possess a trimeric structure in the ectodomain, and it is conceivable that the cytoplasmic domain also trimerizes with these hydrophobic amino acids, providing the interaction for trimerization (43). The protease cleavage site for R-peptide removal is located at position g of the helix and is free from the interaction face to be accessed by the viral protease to cleave the R-peptide and activate the Env (43). At position e of the helix is a positively charged face in the predicted alpha helix.

F-MLV Env can be made sensitive to Vpu.

To validate the Vpu sensitivity motif, we tested whether Vpu sensitivity could be conferred to F-MLV Env with minimal changes. Although the F-MLV Env and GaLV Env CTDs differ at 12 residues out of 29 positions, only four of these positions are part of the predicted Vpu sensitivity motif (Fig. 6A). Of these four, we focused on K11, K18, and N25, which are in the position e of the putative alpha helix (Fig. 6B). The tail lengths of F-MLV and GaLV Env differ by one amino acid. N25, K18, and K11 in the GaLV Env from the C terminus correspond to K26, Q19, and Q12 in the F-MLV Env from the C terminus. These amino acids fall on the same face of the putative alpha helix. The Q12K mutation alone conferred a slight sensitivity to Vpu upon F-MLV Env. However, mutations Q12K and Q19K were sufficient to make F-MLV Env acutely sensitive to Vpu. Q12K and K26N mutations, in contrast, were not sufficient to confer Vpu sensitivity (Fig. 6C).

Fig. 6.

F-MLV Env can be made sensitive to Vpu. (A) F-MLV and GaLV Env CTD along with the Vpu sensitivity motif are aligned, and the differences between F-MLV and GaLV Env CTD within the motif are boxed. (B) The alpha-helical representation is of the F-MLV Env CTD starting from L34 at position d and ending at Y11 at position f. F-MLV Env CTD differs from GaLV Env CTD within the Vpu sensitivity motif at the underlined positions (C) F-MLV Env CTD has been mutated to more closely resemble the Vpu sensitivity motif. Amino acid sequences are indicated as in Fig. 2. Vpu sensitivity and relative infectivity are calculated as shown in Fig. 1 and are averages from six experiments for F-MLV Env mutants and from nine experiments for the wild-type F-MLV Env. (D) The lysines that can render F-MLV Env sensitive to Vpu have been mutated in the GaLV Env CTD. Data for K18Q Env from Fig. 3 have been included for comparison. Vpu sensitivity shown is the average from four experiments for the Δ8 and F-MLV/GaLV K11R K18R Envs and from three experiments for K11R and K18R single-mutation Envs. Mean relative infectivity ± SD from the experiments is shown.

Vpu sensitivity requires lysine at the eleventh position from the C terminus.

Vpu modulation of the cellular targets CD4 and tetherin have both been proposed to involve ubiquitination. Ubiquitin, a 76-amino-acid and 19-kDa moiety, may be conjugated to proteins to modulate various functions, such as altered trafficking, or to target proteins for proteasome-mediated degradation. Ubiquitin is most often conjugated to lysine residues in the target protein. The CTD of the F-MLV/GaLV chimeric Env contains two lysine residues, both of which are part of the Vpu sensitivity motif (Fig. 3). In the light of the ability to make F-MLV Env sensitive to Vpu, each of the lysines was tested for necessity toward Vpu sensitivity. Each of the lysines was mutated to arginine, which is biophysically similar to lysine but not prone to ubiquitination. When both lysines were mutated to arginine, Vpu sensitivity was abolished (Fig. 6D). Individually, the K18R was slightly less sensitive to Vpu, but the K11R abolished Vpu sensitivity. Hence, the lysine at the eleventh position from the C terminus is required for Vpu sensitivity. When K18 was mutated to Q, to better resemble the F-MLV protease cleavage site (6), sensitivity was abolished (Fig. 3). These data suggest that a positively charged amino acid is required at the eighteenth position for sensitivity toward Vpu and a lysine is specifically required at the eleventh position. Collectively, these data suggest that the Vpu sensitivity motif consists of a C-terminal alpha helix with a hydrophobic face adjacent to a positively charged face, including at least one lysine.

Loss of the processed form of a Vpu-sensitive Env in the presence of Vpu.

Our lab has previously reported that Vpu specifically blocks the incorporation of the F-MLV/GaLV Env into HIV-1 particles (21). It has been reported previously that Vpu expression results in a loss of cleaved TM protein in cells (7). To ensure that the addition of Vpu does not enforce reduction in infectivity by means other than preventing incorporation of Env into HIV-1 particles, cells were transfected with 500 ng of HIV-CMV-GFP with or without Vpu and 500 ng of different Env constructs as indicated in Fig. 7A. Env is produced as a gp85 precursor and then processed by a cellular protease into gp70 (SU) and p15 (TM). In the case of the Vpu sensitive Envs, in the presence of Vpu, the processed or the gp70 form of the Env is lost, and this correlates with the loss of Env incorporation into viral particles (Fig. 7A). The infectivity of the pseudotyped viral particles in these experiments was also determined, and Vpu sensitivity was calculated (Fig. 7B).

Fig. 7.

Loss of the processed form of a Vpu-sensitive Env in the presence of Vpu. (A) 293FT cells were transfected with HIV-CMV-GFP ± Vpu and different Envs as indicated. Western blot analysis was performed for the transfected cells and pelleted viral supernatants. A representative image from the experiments performed is shown. (B) Vpu sensitivity of the various envelopes used in these experiments is shown and is an average from 19 independent experiments. Error bars represent standard deviations from the mean from 19 independent experiments.

To revisit the question of whether GaLV Env and CD4 contain the same Vpu target, we re-examined the sequence similarity in light of the identification of the precise Vpu sensitivity motif between the GaLV Env CTD, other gamma retroviral Env CTDs, and the human CD4 CTD (Fig. 8). Among the gammaretroviruses very closely related to GaLV, namely, KoRV and wooly monkey sarcoma virus (WMSV), the amino acids contributing to Vpu sensitivity are conserved. The KoRV Env contains a K18R variation, but this CTD would be predicted to remain Vpu sensitive (Fig. 6). All gammaretroviral Env CTDs maintained the predicted hydrophobic alpha helix, but no other Envs contained lysines at the critical 11th and 18th positions. In fact, the vast majority of gammaretroviral Envs contain highly conserved glutamines residues at these two positions. In contrast, the human CD4 CTD is predicted to contain an alpha helix with hydrophobic residues at the appropriate locations, and it contains lysines in the positions equivalent to the 11th and 18th positions (Fig. 8). Collectively, these data suggest that GaLV Env and CD4 unexpectedly contain sequences with equivalent Vpu recognition sequences.

Fig. 8.

Alignment the CTDs from GaLV Env, WMSV Env, KoRV Env, Moloney murine leukemia virus (MoMLV) Env, F-MLV Env, MLV AKR, Feline leukemia virus (FLV) Env, Kirsten murine leukemia virus (KMLV) Env, xenotropic murine leukemia related virus (XMRV) Env, spleen necrosis virus (SNV) Env, endogenous retroviral (ER) Envs, and human CD4, along with the Vpu sensitivity motif. Dark boxes highlight the critical lysine positions; light boxes highlight the other residues critical for Vpu sensitivity in GaLV Env.

DISCUSSION

In this study, the sequence in GaLV Env CTD that modulates Vpu sensitivity has been identified comprehensively to be INxxIxxVKxxVxRxK. Previous reports had identified the KRLLSEKKT sequence in CD4 to be the minimal sequence required for restriction by Vpu; this was later narrowed down to EKKT in the CD4 CTD (18, 46, 50). This study, however, shows that the lysines, separated by 7 amino acids or two turns of the helix, within KRLLSEKKT are required. The predicted motif required for restriction is essentially an alpha helix with a positively charged face that includes at least one lysine and a hydrophobic face. The alpha helix structural requirement in the CD4 CTD for Vpu-mediated downregulation has been previously reported (50).

Vpu is thought to modulate CD4 expression by inducing its polyubiquitination and subsequent degradation from the ER (47–48). The enhanced turnover and ubiquitination of CD4 in the presence of Vpu in the CD4 CTD is well documented (23, 34). Ubiquitination of proteins occurs predominantly at lysine residues. However, mutational studies of the CD4 CTD have indicated a role for amino acids other than lysine in Vpu-mediated degradation (1, 22). Vpu-dependent CD4 ubiquitination is abolished only when lysine, serine, and threonine in the CD4 CTD are mutated (22). In our study to identify amino acids involved in Vpu-dependent modulation of GaLV Env, we found that serines in the GaLV Env CTD are not necessary. The only amino acids likely to be ubiquitinated in the motif are the two lysine residues (Fig. 3, 6, and 8). In fact, F-MLV Env could be made sensitive to Vpu by introduction of these two lysines in the appropriate locations of the F-MLV Env CTD (Fig. 6).

Our finding that the membrane-proximal lysine (K18) can be replaced by arginine, but not glutamine, without affecting Vpu sensitivity (Fig. 5) appears to differ from the findings of a previous report (7). Christodoulopoulos et al. reported that mutating this lysine residue (denoted K618 in the previous report) to arginine abolished Vpu sensitivity and to glutamine partially alleviated sensitivity. The reason for this discrepancy is not known, but there are two key differences between this study and the previous study. First, the present study analyzed the GaLV Env CTD in the context of an F-MLV Env protein; the previous study was performed with a full-length GaLV Env. Second, the output of the current study was loss of infectious particle production; the output of the previous study was loss of mature GaLV Env expression within the cell. In the present study, the K18R mutation did reduce sensitivity to Vpu, but it did not abolish it. Thus, the difference between these results and previous results could simply reflect differences in sensitivity of the two assays.

In addition to identifying the motif, this current study found that when the sequences of GaLV Env CTD and CD4 CTD are aligned in the context of the Vpu sensitivity motif, the CD4 CTD also contains a sequence consistent with the GaLV Env sensitivity motif, namely, two lysines separated by two helicals of a predicted alpha helix (Fig. 8). TASK-1, another protein reported to be targeted by Vpu (11), also contains a sequence consistent with this sensitivity motif. Tetherin is reported to be targeted by Vpu through its transmembrane region rather than its CTD; not surprising, no sequence consistent with the GaLV Env Vpu sensitivity motif is found in tetherin's cytoplasmic domain. It is likely that Vpu-based restriction of proteins by cytoplasmic tail recognition is different from transmembrane domain recognition (35).

The infectivity of the F-MLV/GaLV Δ8 Env is greater than that of the full-length F-MLV/GaLV Env. These observations indicate the presence of an inhibitory/regulatory sequence in the C-terminal eight amino acids in the GaLV Env CTD. Because these amino acids are all removed by R-peptide cleavage, their inhibitory effect presumably occurs prior to cleavage. Indeed, F-MLV/GaLV Env is not incorporated into HIV-1 particles at the same levels as wild-type F-MLV Env (see Fig. S1 in the supplemental material). A similar phenomenon has been observed with the gammaretrovirus Env protein from the feline endogenous retrovirus RD114. This Env is poorly incorporated into lentiviral particles, and the inhibitory sequence has been mapped to an acidic motif in the last six amino acids of the protein (2). The loss of incorporation is attributed to alterations in protein trafficking due to the acid motif. Altering two of the negatively charged amino acids alleviated the retroviral restriction (2). Although the C-terminal tail of GaLV Env is not as negatively charged as RD114 Env, the last eight amino acids of GaLV Env do include two negatively charged residues.

Gammaretroviral Env proteins have been important components of several gene therapy applications. Because different applications require different virus/glycoprotein combinations, it is important to understand the molecular details that lead to compatibility among these combinations. This study provides an understanding of what causes incompatibility between a gammaretroviral Env and a Vpu-containing lentivirus. We can now predict other glycoproteins that are likely to be incompatible with such a virus, and the information provided here could be used to eliminate this incompatibility.

In conclusion, the Vpu sensitivity motif appears to be an alpha helix on the cytoplasmic side of a membrane protein with a positively charged face, including at least one lysine and a hydrophobic face. This information can potentially be used to identify additional Vpu target proteins by in silico screening for proteins with these properties and to develop new lentivirus-based gene transfer vectors.