|Home | About | Journals | Submit | Contact Us | Français|
The molecular basis for localization of the human immunodeficiency virus type 1 envelope glycoprotein (Env) in detergent-resistant membranes (DRMs), also called lipid rafts, still remains unclear. The C-terminal cytoplasmic tail of gp41 contains three membrane-interacting, amphipathic α-helical sequences, termed lentivirus lytic peptide 2 (LLP-2), LLP-3, and LLP-1, in that order. Here we identify determinants in the cytoplasmic tail which are crucial for Env's association with Triton X-100-resistant rafts. Truncations of LLP-1 greatly reduced Env localization in lipid rafts, and the property of Gag-independent gp41 localization in rafts was conserved among different strains. Analyses of mutants containing single deletions or substitutions in LLP-1 showed that the α-helical structure of the LLP-1 hydrophobic face has a more-critical role in Env-raft associations than that of the hydrophilic face. With the exception of a Pro substitution for Val-833, all Pro substitution and charge-inverting mutants showed wild-type virus-like one-cycle viral infectivity, replication kinetics, and Env incorporation into the virus. The intracellular localization and cell surface expression of mutants not localized in lipid rafts, such as the TM844, TM813, 829P, and 843P mutants, were apparently normal compared to those of wild-type Env. Cytoplasmic subdomain targeting analyses revealed that the sequence spanning LLP-3 and LLP-1 could target a cytoplasmic reporter protein to DRMs. Mutations of LLP-1 that affected Env association with lipid rafts also disrupted the DRM-targeting ability of the LLP-3/LLP-1 sequence. Our results clearly demonstrate that LLP motifs located in the C-terminal cytoplasmic tail of gp41 harbor Triton X-100-resistant raft association determinants.
Lentiviruses, including human immunodeficiency virus type 1 (HIV-1), are unusual in possessing a long cytoplasmic domain (~150 amino acids) in their envelope (Env) transmembrane (TM) glycoprotein compared to those of other retroviruses (20 to 50 amino acids). The cytoplasmic domain of HIV-1 TM protein gp41, which encompasses residues 706 to 856, has multiple functions during the virus life cycle, including viral replication, infectivity, transmission, and cytopathogenicity. Truncations of the HIV-1 cytoplasmic domains may modulate cell-cell fusion properties of the Env protein, presumably due to alterations in the levels of cell surface Env expression and conformation of the Env ectodomain (23, 81). The cytoplasmic domain is characterized by the presence of three structurally conserved, amphipathic α-helical segments, located at residues 828 to 856, 770 to 795, and 786 to 813 and referred to as lentivirus lytic peptide 1 (LLP-1), LLP-2, and LLP-3, respectively, at its C terminus (Fig. (Fig.1A).1A). The LLP-1 and LLP-2 sequences were shown to be inserted into viral membranes by a photoinduced chemical reaction (73). These LLP motifs have been implicated in a variety of functions, such as cell surface expression (12), Env fusogenicity (30), and Env incorporation into a virus (47, 56), as well as Env protein stability (33) and multimerization (34).
Gag and Env carry specific intracellular localization signals governing the site(s) of virus assembly/budding and release into the extracellular milieu. Env trafficking to the plasma membrane is regulated by the conserved C-terminal dileucine motif and the endocytic, membrane-proximal, tyrosine-based GY712SPL signal in the cytoplasmic tail of gp41 (Fig. (Fig.1A)1A) and by their respective interactions with the clathrin adaptor proteins, AP1 and AP2 (4, 9, 21, 49, 65, 77). A diaromatic motif, Y802W803, was shown to bind to TIP47, a protein required for the retrograde transport of mannose-6-phosphate receptors from late endosomes to the trans-Golgi network, and this interaction was involved in the retrograde transport of Env to the trans-Golgi network (8). Alterations of these intracellular localization signals may affect viral infectivity, Env assembly into the virus, and viral replication (8, 20). Likewise, Gag also contains important sequences required for its trafficking to and assembly at the plasma membrane. The matrix (MA) protein, p17, contains a myristoyl group and a cluster of basic amino acids, while p6 contains a late domain which interacts with the components of the endosomal sorting complex required for transport (ESCRT) pathway to mediate Gag trafficking to the virion assembly/budding site (for reviews, see references 25, 45, 57, and 59). It is well documented that the specific interaction between the cytoplasmic domain of gp41 and the trimeric MA protein in infected cells facilitates recruitment of the Env into virus assembly/budding sites on target membranes (for reviews, see references 18, 24, and 46). TIP47 was demonstrated to act as an adaptor to bridge the gp41 cytoplasmic domain and Gag, which allows the physical encounter between Gag and Env, resulting in efficient Env incorporation into the virus during the viral assembly/budding process (39).
Lipid rafts, also called detergent-resistant membranes (DRMs), are highly specialized membrane microdomains present in both the plasma and endosomal membranes of eukaryotic cells. These dynamic microdomains are characterized by their detergent insolubility, light density on a sucrose gradient, and enrichment of cholesterol, glycosphingolipids, and glycosylphosphatidylinositol (GPI)-linked proteins that are anchored in the membrane by their attached GPI moieties (1). HIV-1 utilizes lipid rafts to efficiently enter host cells (40, 74, 80) and selectively assembles and buds from lipid rafts on the surfaces of infected cells (27, 36, 48, 50, 54). Also, the HIV-1 Env protein was detected in lipid raft membranes (48, 54, 64). Lipid rafts are thought to facilitate Env-Gag interactions, to concentrate viral Env glycoproteins, and to promote multimerization of intracellular viral components (for a review, see reference 51). However, what governs Env transport to and localization in lipid rafts is a long-standing question.
Although the mechanisms by which proteins associate with lipid rafts are not fully understood, determinants for targeting of signal proteins to DRMs have been identified. These include a GPI anchor (2, 61) and an N-terminal Met-Gly-Cys in which Gly is myristylated and Cys is palmitoylated (43, 71). The latter includes certain dually acylated heterotrimeric guanine nucleotide-binding protein (G protein) α subunits (44). In addition, acylation by palmitoylation also serves as a signal to target signaling molecules to lipid rafts (for reviews, see references 11 and 60). Some Env proteins of membrane-enveloped viruses are known to be associated with lipid rafts (35, 41, 54, 69, 79), and acylation of viral Env proteins, in particular, palmitoylation, is important for targeting these Env proteins to lipid rafts (for reviews, see references 58 and 70).
It is generally believed that the association of HIV-1 Env with lipid rafts requires a palmitoylation signal(s) located in the cytoplasmic tail of gp41 (6, 64). Nevertheless, the two cytoplasmic palmitoylated Cys residues in the HXB2 strain Env protein are not conserved among HIV-1 isolates, and some isolates do not even contain cysteine residues in their cytoplasmic tail (32). In accordance with this notion, we previously demonstrated that the two cytoplasmic palmitoylated Cys residues in T-cell (T)- and macrophage (M)-tropic Env proteins do not play an obvious role in the virus life cycle, including Env's association with lipid rafts (13), suggesting that other factors may substitute for cytoplasmic palmitoylation to promote Env localization in lipid rafts. Clapham's group showed that mutations in MA or the cytoplasmic tail that prevent Env from incorporating into the virus and impair virus infectivity also interfere with Env's association with lipid rafts (7), indicating that the Gag-Env interaction drives efficient Env association with lipid rafts, which in turn modulates Env budding and assembly onto viral particles. In contrast to their findings, we previously also noted that the Env protein of the HXB2 strain expressed without Gag is still located in lipid rafts (13), providing compelling evidence for the proposal that the Env per se contains sufficient information for its sequestration into lipid rafts.
To further understand the nature of Env's association with lipid rafts, in the present study we show that sequestering Env in Triton X-100-resistant lipid rafts is an intrinsic property of Env and is independent of Gag-Env interactions. Additionally, the LLP motifs, in particular the α-helical structure of the hydrophobic face of LLP-1, play a crucial role in Env's localization in lipid rafts. Except for the 833P mutant of Env, which is unstable and degrades (33), all Pro-substituted mutants not located in lipid rafts exhibited wild-type (WT)-like phenotypes of intracellular localization, cell surface expression, incorporation into virions, and viral replication capacity. Importantly, the α-helix of the hydrophobic face of LLP-1 is also critical for the raft-targeting ability of the LLP-3/LLP-1 sequence. Our study depicts, for the first time, the critical role of the α-helix of the gp41 cytoplasmic domain in mediating Env's association with and targeting to Triton X-100-resistant lipid rafts.
The human embryonic kidney cell-derived 293T cell line that expresses the simian virus 40 large T antigen was maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS). The CD4+ T-cell lines CEM-SS and SupT1 and the hybridomas 902, Chessie 8, and 183 (clone H12-5C), which secrete mouse monoclonal antibodies (MAbs) specific for gp120, gp41, and capsid protein p24 of Gag, respectively, were maintained in RPMI 1640 supplemented with 10% FBS as previously described (16, 33). Rabbit anti-caveolin-1 antiserum and anti-flotillin-1 MAb were obtained commercially from BD Biosciences (San Jose, CA). Rabbit anti-placenta alkaline phosphatase (PLAP) was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Affinity-purified fluorescein isothiocyanate (FITC)- and rhodamine-conjugated secondary antibodies were purchased from Kirkegaard & Perry Laboratories (Gaithersburg, MD). An anti-β-galactosidase (anti-β-Gal) MAb was obtained from Promega (Madison, WI). Rabbit anti-calreticulin, β-actin MAb, methyl-β-cyclodextrin (β-MCD), and cholesterol were all purchased from Sigma (St. Louis, MO).
pHXB2R3-based cytoplasmic tail C-terminal truncation mutant proviral constructs (Fig. (Fig.1A)1A) were as previously described (78). Proviruses which encoded mutant Env proteins with single deletions at Thr-818, Ile-830, Val-833, Ala-836, Ile-840, Ile-843, Ile-847, and Ile-854 (Fig. (Fig.1B,1B, first group) and mutant proviral constructs encoding Ala, Ser, and Pro substitutions for Val-832 and Val-833 in Env (Fig. (Fig.1B,1B, second group) were as previously reported (33). The HIV-1 long terminal repeat (LTR)-driven Tat-expressing plasmid pIIIextat was as previously described (13). pcDNA3-based constructs that encoded cytoplasmic tail subdomains 706 to 752, 760 to 856, 760 to 795, and 816 to 856 were as previously described (16).
pHXB2R3-derived TM836, TM828, and TM820 mutant proviruses, which encode Env proteins truncated after residues Ala-836, Arg-828, and Ile-820, respectively, were created (Fig. (Fig.1A),1A), and mutant proviruses encoding Val829Pro, Ala839Pro, Ile843Pro, and Leu855Pro substitutions (Fig. (Fig.1B,1B, third group) and Glu831Arg, Arg838Glu, Arg841Glu, Arg845Glu, Arg846Glu, Arg848Glu, and Arg853Glu substitutions (Fig. (Fig.1B,1B, fourth group) were constructed by oligonucleotide-directed, site-specific mutagenesis using a PCR overlap extension, as previously indicated (33). The paired internal primers used in the PCR to encode each mutation in the gp41 C-terminal cytoplasmic tail are shown in Table Table1.1. Oligonucleotides 8423f (5′-GAAGAAGAAGGTGGAGAGAGA-3′) (sense; nucleotides 8423 to 8443 of the HXB2 strain of the provirus) and 8933r (5′-GCTACTTGTGATTGCTCC-3′) (antisense; nucleotides 8933 to 8916) were used as the outer paired primers. The BamHI- and XhoI-restricted PCR products were cloned into the pHXB2R3 provirus at the same sites to generate each mutant provirus. To generate HIV-1 LTR-controlled mutant Env expression plasmids, the KpnI-XhoI fragments isolated from mutant proviruses were substituted for the homologous sequence in a version of WT pSVE7puro in which the XhoI site located at the 5′-LTR was deleted. All mutant pHXB2R3 and pSVE7puro constructs were autosequenced with the 8423f oligonucleotide to confirm the mutations. The KpnI-XhoI fragments of the NL4-3 and ConB proviruses were also used to replace the corresponding sequence in pSVE7puro to obtain the NL4-3 and ConB env expression plasmids, respectively.
To construct pcDNA3-based β-Gal fusion chimeras that encoded the 786-813 and 786-856 regions attached at the C terminus of β-Gal, oligonucleotides 786fEcoRI (Table (Table1)1) and 856rXbaI were used as primers with the TM813 and WT pSVE7puro plasmids, respectively, in a PCR. To construct the 760-813 plasmid, 760fEcoRI and 856rXbaI were used as primers with TM813 pSVE7puro in a PCR. To construct β-Gal-786-856 fusion proteins with a deletion at Val-833 or a Pro substitution for Val-832 or Val-833, oligonucleotides 786fEcoRI and 856rXbaI were used as primers with their respective pHXB2R3 mutant proviruses in a PCR. To generate the β-Gal-786-856-based Ile-854 deletion, the DNA fragment was PCR amplified from the Δ854 mutant provirus, using 786fEcoRI as the forward primer and oligonucleotide 856r(Δ854)XbaI (Table (Table1)1) as the reverse primer. All constructs were cloned as previously described (16) and confirmed by sequencing.
Subconfluent 293T cells grown on 10-cm petri dishes were transfected with 10 μg of provirus, 10 μg of the pcDNA3-β-gal construct, or 10 μg of pSVE7puro together with 2 μg of pIIIextat, using a standard calcium phosphate coprecipitation method as previously described (53). To prepare vesicular stomatitis virus (VSV) glycoprotein G trans-complemented HIV-1 stocks, 293T cells were cotransfected with 7.5 μg each of the provirus and pHCMV-VSV G, a human cytomegalovirus (CMV) promoter-directed VSV G protein expression plasmid. At 6 hours posttransfection, excess DNA-calcium phosphate complexes were rinsed off with calcium-free phosphate-buffered saline (PBS), and fresh DMEM with 5% FBS was added to the culture. Cells were harvested for further analysis at 48 h posttransfection.
Culture supernatants obtained from 293T cells transfected with WT or mutant proviruses or cotransfected with pHCMV-VSV G and each of the WT and mutant proviruses were filtered through 0.45-μm membrane discs and normalized by reverse transcriptase (RT) activity as previously described (17). Cell-free viruses containing 2 × 104 cpm of RT activity were used to challenge 106 CEM-SS cells, and culture supernatants collected from infected cells were monitored at different times for RT activity. For VSV G trans-complemented viral infection, cell-free viruses containing 106 cpm of RT activity were used to challenge 107 CEM-SS cells for lipid raft membrane flotation assay or 1.25 × 106 SupT1 cells for analysis of viral protein expression.
LuSIV cells were seeded at 3 × 104 cells per well in a 96-well round-bottom plate, 100 μl of WT or mutant virus containing 1 × 105 cpm of RT activity was overlaid into each well, and the cultures were subjected to spin inoculation. The cultures were then incubated at 37°C overnight, washed twice with PBS, replenished with 100 μl of fresh medium per well, and allowed to recover for an additional 48 h. Cells were harvested and lysed, and firefly luciferase activity was assayed in triplicate. Results from three separate experiments were quantified.
Three-layer membrane flotation analysis to assess Env localization in lipid rafts was performed according to a previously described procedure (13, 17). To separate light and heavy DRMs from detergent-soluble membrane (DSM) fractions, a 14-fraction sucrose density gradient centrifugation method, as previously described (6, 7), was adopted for the Beckman SW-41 rotor. Cell pellets were lysed with 500 μl of 0.5% Triton X-100 in NTE buffer (25 mM Tris-HCl, pH 7.5, containing 0.15 M NaCl and 5 mM EDTA) supplemented with phenylmethylsulfonyl fluoride and a protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany), and 400 μl of clarified lysates was mixed with 1 ml of 85% sucrose prepared in NTE buffer. The mixture was then overlaid on a layer containing 250 μl of 80% sucrose in NTE buffer, which was placed in the bottom of an SW41 ultracentrifuge tube. Two milliliters each of 50%, 40%, 35%, 10%, and 5% sucrose, all prepared in NTE buffer, was then successively layered above the lysate-sucrose mix in the said order. Gradients were centrifuged at 100,000 × g and 4°C for 18 h. Upon completion, 850-μl fractions were collected from the top of the gradients by manual micropipetting. For CEM-SS cells expressing Env proteins, a scaled-down version of the sucrose density gradient centrifugation procedure, as previously described (17), was followed. For density gradient centrifugation of 293T cells expressing β-Gal-gp41 cytoplasmic tail fusion proteins, the three-layer sucrose density gradient centrifugation method described above was performed, and fractions were collected from the top of the gradients by manual micropipetting.
For proviral DNA transfection, the production, concentration, and analysis of cell- and virion-associated viral proteins were performed as previously described (15). Equal volumes of cell and virion lysates were then subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by Western blot analyses.
For sucrose gradient centrifugation, aliquots of samples in each fraction were diluted with an equal volume of NTE buffer, 20 μg of bovine serum albumin was added, and the mixture was concentrated by precipitation with 10% cold trichloroacetic acid prior to SDS-PAGE. For analysis of β-Gal-gp41 cytoplasmic tail fusion proteins, aliquots (200 μl) of each gradient fraction were added to 20 μg of bovine serum albumin, followed by the addition of 1 ml of ice-cold acetone. The acetone-precipitated samples were incubated at −20°C for 1 h and then centrifuged at 18,000 × g for 15 min at 4°C. After removing the upper, acetone layer, the lower, precipitated protein layer was mixed with 1 ml of ice-cold 80% (vol/vol) ethanol to remove excess sucrose. The protein pellets were then reprecipitated by centrifugation, as indicated above, prior to SDS-PAGE.
In the Western blot analysis, membrane blots were incubated with MAbs 902, Chessie 8, and 183, to detect, respectively, gp160 and gp120, gp41, and the Pr55 Gag precursor and its cleavage products (p41 and p24) or with the β-Gal MAb to detect β-Gal fusion proteins. After incubation with secondary antibodies, the immune complexes were visualized by an enhanced chemiluminescence (ECL) detection method. For quantitation, ECL images of the blots within the linear range of the film were scanned using a Microtek ScanMarker 8700 instrument (Carson, CA) and were quantitated using MetaMorph software (Universal Imaging, Downingtown, PA). The relative percentages of the levels of mutant proteins associated with DRMs to that of the WT protein are expressed, unless otherwise indicated.
Monolayers of transfected 293T cells were cultured in serum-free DMEM at 37°C for 1 h prior to treatment with β-MCD in serum-free DMEM for 30 min to extract cholesterol from plasma membranes. For cholesterol reconstitution after β-MCD treatment, the cholesterol-β-MCD complex (30 mg cholesterol to 1 g β-MCD), prepared as previously described (31), was diluted in serum-free DMEM to a final cholesterol concentration of 200 μg/ml. β-MCD-treated monolayers were then washed three times with PBS before being overlaid with the cholesterol-β-MCD solution for treatment at 37°C for 30 min. β-MCD-treated cells not replenished with cholesterol were overlaid with serum-free DMEM in parallel. Both cholesterol-depleted and -replenished cells were lysed with 1% Triton X-100 and subjected to sucrose density gradient centrifugation as described above.
For Env intracellular localization, transfected HeLa cells were fixed with 4% paraformaldehyde and permeabilized with 0.25% Triton X-100. After being blocked, the slides were successively incubated with Chessie 8, FITC-conjugated anti-mouse immunoglobulin G (IgG), rabbit anti-calreticulin, and rhodamine-conjugated anti-rabbit IgG, and immunostained cells were analyzed by confocal microscopy as previously described (14). To quantify Env expression, a previously described procedure (14) was followed. For total Env expression, proviral DNA-transfected 293T cells were fixed with 4% paraformaldehyde, permeabilized with 0.25% Triton X-100, and then successively incubated with MAb 902 and FITC-conjugated anti-mouse IgG. For cell surface Env expression, transfected cells, after being blocked, were incubated with MAb 902, fixed, and then incubated with FITC-conjugated anti-mouse IgG. Immunostained cells were quantitated by fluorescence-activated cell sorter (FACS) analysis using a FACSCalibur flow cytometer (Becton Dickinson Immunocytometry Systems, San Jose, CA).
We previously showed that gp41 was associated with lipid rafts when Env was expressed from the HIV-1 LTR-driven env expression plasmid pSVE7puro without other viral protein coexpression (13), suggesting that gp41 itself contains information sufficient for Env association with lipid rafts. We thus further identified and characterized determinants in Env that are crucial for Env-lipid raft association. Since the cytoplasmic domain of gp41 is characterized by the three structurally conserved, membrane-interacting, amphipathic α-helical LLP motifs located at its C terminus (Fig. (Fig.1A),1A), we thereby examined the involvement of these LLP sequences in Env localization in lipid rafts.
We first studied the effects of progressive truncations of the C terminus of the cytoplasmic tail on Env's association with lipid rafts in the context of proviral DNA transfection. A series of pHXB2R3-based cytoplasmic tail truncation mutants, i.e., TM844, TM836, TM828, TM820, and TM813 (Fig. (Fig.1A),1A), along with the WT provirus, were used to transfect 293T cells. Cells were extracted with 1% cold Triton X-100, and lysates were analyzed by sucrose gradient centrifugation as previously described (13, 17). To assess the reliability of this raft fractionation technique in separating raft proteins from soluble proteins, the raft association properties of three known raft markers, PLAP, flotillin-1, and caveolin-1, were analyzed. A significant fraction of PLAP was fractionated into DRMs, i.e., fractions 7 and 8, which were positioned in the interface between the 30% and 5% sucrose layers (Fig. (Fig.2A,2A, top panel). Flotillin-1 and caveolin-1 were distributed predominantly in lipid raft membranes (Fig. (Fig.2A,2A, middle and bottom panels, respectively). DRM-associated WT Env also floated to fractions 7 and 8, while detergent-soluble WT Env was located in the bottom cytosolic fractions 1 to 3 (Fig. (Fig.2B,2B, first panel). The TM844 mutant, which has a truncation of the last 12 amino acids of the C terminus of gp41, showed greatly reduced Env localization in lipid rafts compared to WT Env (Fig. (Fig.2B,2B, panel 2). Further truncations toward the N terminus of LLP-1 strikingly abolished Env distribution into lipid raft fractions (Fig. (Fig.2B,2B, panels 3 to 6). The quantitative results from three separate studies are shown in the bottom graph of Fig. Fig.2B2B.
Next, a set of pSVE7puro Env plasmids that encoded a series of truncations from the C terminus of the cytoplasmic tail was examined. The TM844 mutant showed greatly impaired Env association with lipid rafts when Env was expressed alone (Fig. (Fig.2C,2C, panel 2). Additional truncations toward the N terminus of LLP-1 also strikingly reduced Env-lipid raft association (Fig. (Fig.2C,2C, panels 3 to 6). The results of quantification from three independent analyses are shown in the bottom graph of Fig. Fig.2C.2C. Moreover, quantitative results showed that similar values for the absolute raft association ability of Env, i.e., the percentage of Env distributed into raft fractions relative to the total Env level, were obtained for expression from proviral DNA transfection and subgenomic expression (Fig. (Fig.2D),2D), confirming that Env is indeed efficiently associated with lipid rafts even in the absence of other viral protein expression.
We also examined the role of the gp41 cytoplasmic tail in Env's association with lipid rafts in HIV-1 natural target CD4+ T cells, using a previously reported high-level, transient HIV-1 expression system based on trans-complementation with the VSV G protein (3, 38). The TM844 mutant strikingly reduced detection of gp41 in lipid raft fractions (Fig. (Fig.2E).2E). Similarly, deletion of the entire LLP-1, e.g., in mutants TM828 and TM813, abrogated Env-lipid raft association (Fig. (Fig.2E).2E). These results together indicate that the cytoplasmic tail of gp41, in particular, LLP-1, plays a crucial role in Env localization in lipid rafts in epithelial cells as well as in CD4+ T cells.
Bhattacharya et al. (7) showed that HIV-1 LTR-driven Env proteins of the HXB2 and NL4-3 strains were excluded from DRMs when coexpressed with an env-deficient NL4-3 mutant 12LE provirus in which the Env-Gag interaction is disrupted, indicating that Env localization in lipid rafts is driven by the Env-Gag interaction. To clarify the discrepancy between their findings and ours, we reexamined whether Env proteins encoded by WT, 12LE, and 30LE NL4-3 proviruses are located in lipid rafts. First, we determined whether incorporation of Env into the virus in CD4+ T cells was impaired, based on trans-complementation with the VSV G protein. As previously reported (26), the two Gag mutants, 12LE and 30LE, which contain Glu substitutions for Leu at residues 12 and 30 in MA, respectively, inhibited Env incorporation into the virus (Fig. (Fig.3A,3A, compare lanes 5 and 6 to lane 4), presumably due to the lack of interaction between gp41 and mutant MA proteins. Env's incorporation into the 12LE and 30LE mutant virions was also severely impaired when examined in 293T cells (data not shown). Moreover, the Env proteins expressed by WT, 12LE, and 30LE mutant NL4-3 proviruses as well as the Env encoded by the HXB2 provirus were found to be located in DRMs when subjected to conventional three-layer sucrose density gradient centrifugation (Fig. (Fig.3B3B).
To explore the possibility that the different sucrose gradient centrifugation methods used by Bhattacharya et al. and us may have contributed to the differential properties of Env's association with lipid rafts, we also employed multilayer sucrose gradient centrifugation as described by Clapham's group to address the raft association property of Env. As controls, significant fractions of PLAP, flotillin-1, and caveolin-1 were fractionated into light-density DRM (DRM-L) as well as heavy-density DRM (DRM-H) (Fig. (Fig.3C).3C). All of the Env proteins encoded by WT, 12LE, and 30LE mutant NL4-3 proviruses and by the HXB2 provirus were found to be localized in DRM-L as well as in DRM-H, although a significant portion of these proteins was still located in the DSM fraction (Fig. (Fig.3D).3D). Furthermore, Env proteins derived from HXB2 and NL4-3 proviruses and from 12LE and 30LE mutant pNL4-3 proviruses were also detected in lipid raft fractions in CEM-SS cells in a three-layer membrane flotation analysis (Fig. (Fig.3E).3E). These results together indicate that the Env-Gag interaction is not required for Env localization in DRMs in monolayer CD4− 293T cells or in suspended CD4+ T cells.
We then confirmed that the HXB2 strain Env expressed without Gag is also localized in lipid rafts by employing the multilayer, stepwise sucrose gradient centrifugation method. Just like the Env encoded by the WT pHXB2R3 provirus, which was found to be associated with DRM-L and DRM-H (Fig. (Fig.4A,4A, top panel), the Env proteins encoded by a pCAGGS promoter-driven Env plasmid (17), marked as pCX-Env, and by WT pSVE7puro were also found to be localized in DRM-L and DRM-H (Fig. (Fig.4A,4A, middle and bottom panels, respectively). When the raft association property of Env was assessed, the percentages of Env distributed into DRM-L, DRM-H, and DSM were 16.9%, 23.3%, and 59.8%, respectively, for the HXB2 proviral Env; 18.8%, 22.1%, and 59.1%, respectively, for the pCAGGS Env; and 17.6%, 26.7%, and 55.7%, respectively, for the pSVE7puro Env. These observations further indicate that Env without Gag coexpression is still effectively associated with raft membranes.
To understand whether localization in lipid rafts is a general property of HIV-1 Env, the Env proteins of T-tropic NL4-3 and M-tropic ConB (13) clones, encoded by the pSVE7puro vector, were examined. As observed with the Env derived from the HXB2 strain, the NL4-3 and ConB Env proteins were also localized to lipid rafts in the absence of Gag expression (Fig. (Fig.4B).4B). The percentage of Env located in lipid rafts relative to the total Env level was quantified to be 19.2%, 12.7%, and 24.9%, respectively, for Env proteins of HXB2, NL4-3, and ConB strains without Gag coexpression. This observation further reveals that association of gp41 with lipid rafts does not require Gag coexpression and is an intrinsic property of Env proteins of different subtypes.
When 293T cells transfected with the WT provirus were treated with different concentrations of β-MCD, which depletes cholesterol from the cell surface, at 37°C for 30 min before extraction with cold Triton X-100, β-MCD decreased Env localization in lipid rafts in a dose-dependent manner (Fig. (Fig.5A).5A). In the context of Env expression alone, β-MCD treatment also greatly abolished the association of Env with lipid rafts (Fig. (Fig.5B,5B, compare the top and middle panels). However, replenishment of β-MCD-treated cells with cholesterol restored the ability of Env to associate with lipid rafts (Fig. (Fig.5B,5B, bottom panel). These results together indicate that HIV-1 Env is specifically localized in lipid raft membranes and that its binding to lipid raft membranes is cholesterol dependent.
The LLP-1 motif is characterized by its amphipathic α-helical feature. To examine the involvement of the α-helical structure of this motif in the Env-lipid raft association, a series of mutant proviruses encoding single point deletions in LLP-1 (Fig. (Fig.1B,1B, first group) was examined. Deletions of single amino acids in the N-terminal and central regions, such as in the Δ830, Δ833, Δ836, Δ840, Δ843, and Δ847 mutants, with deletions of Ile-830, Val-833, Ala-836, Ile-840, Ile-843, and Ile-847, respectively, abrogated or greatly reduced Env-lipid raft interactions (Fig. (Fig.6A).6A). However, Env with a deletion of Ile-854, which is positioned at the extreme C terminus of LLP-1, was still located in lipid rafts (Fig. (Fig.6A).6A). Interestingly, deletion of Thr-818, which is situated between the LLP-3 and LLP-1 motifs, also greatly reduced Env localization in lipid rafts (Fig. (Fig.6A6A).
To study the specificity of displacement of the amphipathic α-helix of LLP-1 in Env-lipid raft interactions, the effects of Pro substitutions for residues in this motif were examined. Pro was expected to disrupt the structure of the α-helix more severely than were other amino acids. Mutant proteins with Pro substitutions for Val-829, Val-832, Val-833, Ala-839, Ile-843, and Leu-855 (Fig. (Fig.1B,1B, second group) in Env were examined. Pro substitutions for residues in the hydrophobic face of LLP-1, such as Val-829, Val-833, and Ile-843, all diminished or greatly decreased gp41-raft associations compared to those with WT Env (Fig. (Fig.6B).6B). However, substitutions of Pro for residues in the hydrophilic face, such as Val-832, Ala-839, and Leu-855, did not greatly alter Env-lipid raft associations (Fig. (Fig.6B6B).
To further study the specificity of Pro substitution for residues in the hydrophilic and hydrophobic faces of LLP-1 in Env localization in lipid rafts, the effects of Ala, Ser, and Pro substitutions for Val-832 and Val-833 (Fig. (Fig.1B,1B, third group) were compared. Unlike the Pro substitution for Val-833, substitutions of Ala and Ser, respectively, for Val-833 did not greatly affect Env-raft associations (Fig. (Fig.6C),6C), while Ala, Ser, and Pro substitutions for Val-832 had no apparent effects on Env-lipid raft associations (Fig. (Fig.6C6C).
There are multiple positively charged Arg and negatively charged Glu residues clustered in LLP-1 (Fig. (Fig.1A).1A). To determine the involvement of these charged residues in Env localization in lipid rafts, mutant proteins with Glu substitutions for Arg-838, Arg-841, Arg-845, Arg-846, Arg-848, and Arg-853 and an Arg substitution for Glu-831 (Fig. (Fig.1B,1B, fourth group) were analyzed. None of these mutations greatly altered the raft association property of Env (Fig. (Fig.6D6D).
To understand whether Env localization in lipid rafts may have a role in virus infection, the replication kinetics of WT and Pro-substituted mutants were assessed. As previously shown (33), the 833P mutant virus was severely impaired in its replication, and no RT activity was detected even up to 32 days after virus challenge (Fig. (Fig.7A).7A). Despite a slight delay in viral replication kinetics, the 829P mutant still replicated productively, whereas all other mutants replicated with kinetics similar to those of the WT virus (Fig. (Fig.7A).7A). Moreover, all charge-switching mutants replicated as productively as the WT virus, although the R853E mutant showed a slight delay in its replication kinetics compared to the WT virus (Fig. (Fig.7B7B).
We then examined the effects of these mutations on viral infectivity, using LuSIV cells, a firefly luciferase gene-harboring reporter cell line (62). This cell line allows assessment of nearly one-cycle viral entry based on the ability of Tat, upon viral infection, to transactivate HIV-1 LTR-linked luciferase gene expression in CEMx174 cells. All Pro substitution mutants examined here, including the 833P mutant, exhibited viral infectivities similar to or even greater than that of the WT virus (Fig. (Fig.7C).7C). To provide evidence for the observation that the 833P mutant possesses WT-like one-cycle viral entry in this reporter gene activation assay, the infectivity of the Δ833 and ΔLWYIK mutants was also assessed. Due to its instable nature, the Δ833 mutant, just like the 833P mutant, showed severely impaired replication kinetics compared to the WT virus (33). The ΔLWYIK mutant, in which the LWYIK motif located immediately proximal to the TM region of gp41 is deleted, had an impaired fusion ability, resulting in inhibited one-cycle viral infectivity and delayed replication kinetics compared to those of the WT virus (17). The ΔLWYIK and Δ833 mutants showed inhibited one-cycle viral infectivities compared to the WT virus (Fig. (Fig.7C).7C). These results together indicate that despite the phenotype of reduced Env assembly in the virions, the level of Env present on the 833P mutant may still exceed the threshold Env level required to initiate successful CD4 and coreceptor binding events, resulting in reporter gene expression following viral entry. Nevertheless, the effect of this mutation on Env destabilization may be augmented further in later rounds of viral replication, resulting in detrimental consequences in viral replication kinetic assays (Fig. (Fig.7A).7A). As to charge-reversing mutants, the E831R mutant still possessed 75% of the one-round viral infectivity of the WT virus, while other mutants did not affect one-cycle viral infectivity compared to that of the WT virus (Fig. (Fig.7D).7D). These results collectively indicate that all mutants, but not the 833P mutant, exhibit WT virus-like replication capacities.
To determine whether Pro-substituted mutants can impair the Env maturation process and/or incorporation of Env into the virus, viral protein expression of these mutants was assessed in 293T cells, which supported one-round viral assembly/budding. All of the WT and mutant viruses produced comparable amounts of intracellular and virion-associated Gag Pr55 precursors and its cleaved products, p41 and p24 (Fig. (Fig.7E).7E). As shown previously (33), the 833P mutant produced smaller amounts of intracellular gp160, gp120, and gp41 and virion-associated gp120 and gp41 than those produced by WT virus expression (Fig. (Fig.7E,7E, compare lanes 4 and 11 to lanes 1 and 8). None of the other mutants had a notable effect on gp160 synthesis, precursor processing, or incorporation of gp120 and gp41 into the virus (Fig. (Fig.7E).7E). When charge-switching mutants were assessed, synthesis, precursor processing, and assembly/budding of Gag and Env and their maturation and incorporation into mutant viruses were apparently normal compared to those observed with the WT virus (Fig. (Fig.7F7F).
Next, viral protein expression in SupT1 cells, which are natural host cells permissive for HIV-1 infection, was assessed by the HIV-1 expression system based on trans-complementation with the VSV G protein. Again, none of the Pro substitutions in LLP-1, except for that in the 833P mutant, affected Env synthesis, precursor processing, or Env incorporation into the virus (Fig. (Fig.7G).7G). Similarly, none of the charge-switching mutants showed obviously altered Env phenotypes compared to those of WT Env (Fig. (Fig.7H).7H). These results together indicate that whether the Env is localized in Triton X-100-resistant DRMs does not necessarily have a consequential effect on the viral replication capacity or on Env assembly in the virus.
To examine whether exclusion of mutants from Triton X-100-resistant rafts may possibly be due to their aberrant intracellular transport to the cell surface, the intracellular localization of the 829P, 833P, 843P, TM844, and TM813 mutants was examined by confocal microscopy, along with the raft-associated WT and 832P mutant proteins. Since Env is known to be retained largely in the endoplasmic reticulum (ER) or in a cis-Golgi compartment (5, 19, 76), colocalization with calreticulin, an ER marker, was examined. A significant fraction of the WT Env was colocalized with calreticulin in the perinuclear area, and it was also colocalized as peripheric dots in the cytoplasm and as speckles near or on the surface (Fig. (Fig.8A).8A). Also, a fraction of WT Env located in the cytoplasm was not colocalized with calreticulin (Fig. (Fig.8A).8A). All mutant proteins also exhibited intracellular distribution patterns similar to that of WT Env (Fig. (Fig.8A8A).
Next, the total and cell surface expression levels of WT and mutant proteins were examined by FACS analysis (Fig. (Fig.8B).8B). Consist with its destabilized nature, the 833P mutant exhibited reduced levels of total and cell surface expression compared to the WT and other mutants (Fig. (Fig.8B).8B). Although total levels of the TM813 and 843P mutants were slightly lower than those of the WT and the other mutants, which could have been due to variations in transfection efficiencies of these two mutants in this particular analysis, these two mutants still showed similar levels of Env on the cell surface compared to the WT and other mutants (Fig. (Fig.8B).8B). Quantitative results from three independent experiments showed that with the exception of the 833P mutant, all mutants showed similar levels of total and cell surface expression to those of WT Env (Fig. (Fig.8C).8C). These studies together indicate that these mutants are synthesized and transported to and expressed on the plasma membrane as effectively as WT Env.
We showed that the C-terminal two-thirds, but not the N-terminal one-third, of the cytoplasmic tail contains membrane-associated sequences (16). To further understand the molecular basis of Env localization in lipid rafts, an Escherichia coli cytosolic reporter protein, β-Gal, was individually tagged with various subdomains of the gp41 cytoplasmic tail, and the ability of these cytoplasmic segments to target β-Gal to DRMs was examined. As a control, β-Gal by itself was distributed in the soluble fraction (Fig. (Fig.9A,9A, panel 1). The 760-856 segment, but not the 706-752 segment, was able to target β-Gal to lipid rafts (Fig. (Fig.9A,9A, panels 3 and 2, respectively). Next, the ability of each LLP sequence, spanning residues 816 to 856, 760 to 795, or 786 to 813, was assessed. Intriguingly, this motif by itself did not possess lipid raft-targeting ability (Fig. (Fig.9A,9A, panels 4 to 6). We then determined whether two contiguous LLP sequences possessed raft-targeting ability. Unlike the 760-856 segment, the 760-813 segment, which contains LLP-2 and LLP-3, did not confer raft-targeting ability (Fig. (Fig.9A,9A, panel 7). Interestingly, the region spanning both the LLP-3 and LLP-1 motifs, i.e., the 786-856 sequence, was able to target β-Gal to lipid rafts despite its lower raft-targeting efficiency, i.e., 38% of that of the 760-856 segment (Fig. (Fig.9A,9A, compare panel 8 to panel 3 and also see the graph). Nevertheless, deletion of the last 12 amino acid residues from LLP-1, i.e., the 786-844 sequence, greatly reduced the raft-targeting ability of the 786-856 segment (Fig. (Fig.9A,9A, panel 9).
We showed that the α-helix of the hydrophobic face of LLP-1 is critical for Env association with lipid rafts. We then determined whether the hydrophobic face of the LLP-1 α-helix containing the 786-856 sequence also plays a pivotal role in targeting β-Gal to lipid rafts. Deletion of Val-833 but not Ile-854 disrupted the lipid raft-targeting ability of the 786-856 segment (Fig. (Fig.9B,9B, compare panels 3 and 4, respectively, to panel 2). Moreover, a mutant with Pro substituted for Val-832 still possessed 72% of the raft-targeting ability of the 786-856 segment, whereas a mutant with Pro substituted for Val-833 greatly abolished the raft-targeting ability (Fig. (Fig.9B,9B, compare panels 6 and 5, respectively, to panel 2 and also see the graph).
In the present study, we extend our previous efforts (17) to unravel the molecular basis of HIV-1 Env localization in lipid rafts. We show that LLP motifs located in the C-terminal cytoplasmic tail of gp41 play a crucial role in Env localization in Triton X-100-resistant lipid rafts and that Env is capable of localizing in lipid rafts without functional Gag-Env interaction.
A deletion in the α-helix is expected to skew the structure in a way that residues originally aligned on either the hydrophobic or hydrophilic face of the helix become displaced. Point deletions in the N-terminal and central regions of the LLP-1 motif, e.g., in the Δ830 and Δ843 mutants, resulted in dispersion of the positively and negatively charged residues into the hydrophobic face (Fig. 10A). However, deletion at the extreme C terminus of LLP-1, such as in the Δ854 mutant, did not affect the amphipathic α-helical feature of LLP-1 (Fig. 10A). Thus, the observation that displacement of the amphipathic α-helix of LLP-1 by a single deletion alters Env association with DRMs (Fig. (Fig.6A)6A) suggests that the α-helical feature of LLP-1 is important for Env localization in these raft microdomains. Pro is expected to disrupt the local α-helical structure more severely than do other amino acids. Pro substitutions for residues in the hydrophobic face, such as Val-829, Val-833, and Ile-843, had a greater impact on gp41-raft association than did substitutions of those in the hydrophilic face, such as Val-832, Ala-839, and Leu-855 (Fig. (Fig.6B6B and 10B). Moreover, Pro, Ser, and Ala are expected to alter the local or neighboring α-helical conformation in an increasingly severe order of Pro > Ser > Ala, as shown by previous peptide modeling studies on the leucine zipper-like motif of HIV-1 gp41 (75). The differential effects of these substitutions on Val-832 and Val-833 (Fig. (Fig.6C)6C) further confirm that the α-helical structure of the hydrophobic face of LLP-1 has a more critical role in Env-raft associations than does the hydrophilic face. This notion is in accordance with the results showing that charge-switching mutations in the hydrophilic face did not greatly alter gp41 association with lipid rafts (Fig. (Fig.6D6D and 10C).
The critical involvement of LLP sequences in Env-raft association was further supported by the ability of the C-terminal two-thirds, but not the N-terminal one-third, of the cytoplasmic domain to target a cytoplasmic reporter protein, e.g., β-Gal, to DRMs (Fig. (Fig.9A).9A). Nevertheless, each of the LLP motifs by itself was insufficient to target β-Gal to DRMs. Remarkably, the intact LLP-3 and LLP-1 motifs, but not the LLP-2 or LLP-3 motif, acted in tandem as a raft-targeting signal. The observation that the 786-856 segment possessed only about 38% of the lipid raft-targeting ability of the 760-856 segment (Fig. (Fig.9A)9A) implies that LLP-2, although insufficient for lipid raft targeting, is still required for the maximal raft-targeting capacity of the LLP-3 and LLP-1 motifs.
We previously showed that LLP motifs possess self-assembly and membrane-interacting abilities and that these motifs can target a cytosolic protein to the ER (16, 34). This observation together with the present study reveals that a long, contiguous, membrane-interacting sequence encompassing intact LLP-3 and LLP-1 motifs is necessary to direct a cytosolic reporter protein to Triton X-100-resistant raft-like microdomains in the ER membrane (10, 55, 67). The mutagenesis and subdomain targeting studies pointed out that the long stretch of LLP-3 and LLP-1, particularly the α-helical structure of the hydrophobic face in LLP-1, not only is involved in but also mediates Env binding to Triton X-100-resistant lipid rafts (Fig. (Fig.2,2, ,6,6, and and9).9). Also, the presence of LLP-2 may increase the interface of LLP-3 and LLP-1 with lipid rafts or stabilize the association of LLP-3 and LLP-1 sequences with lipid rafts. Alternatively, it is likely that the hydrophobic residues of LLP-3 and LLP-1, by virtue of being buried within the hydrophobic core of lipid raft-associated cellular proteins, may directly interact with the lipid portions of Triton X-100-resistant raft membranes. Disruption of the lipid raft integrity or structure by cholesterol depletion thereby hinders Env's association with lipid raft membranes (Fig. (Fig.55).
Given the known Env phenotypes of the mutants described here and those which were previously characterized, we cannot simply attribute the property of Env localization in Triton X-100-resistant lipid rafts to viral replication or Env assembly in the virus. For instance, truncations of the last 12 and 43 amino acids from the C terminus of the cytoplasmic tail, i.e., the TM844 and TM813 mutants, respectively, greatly disrupted Env localization in lipid rafts (Fig. (Fig.2),2), but the incorporation of gp120 into the virus for these two mutants was normal despite the drastically reduced or lacking infectivity of these two mutants compared to that of the WT virus (78). All single-deletion mutants in LLP-1, but not the Δ854 mutant, abrogated the lipid raft association property of Env (Fig. (Fig.6A);6A); nonetheless, only N-terminal LLP-1 deletion mutants, such as the Δ830 and Δ833 mutants, exhibited severely impaired viral replication, which was attributable to the instability of their Env proteins. The Δ843 and Δ847 mutants still displayed WT-like viral replication kinetics and Env incorporation into the virus, whereas the Δ836 and Δ840 mutants were also replication competent (33). Furthermore, Pro substitutions for residues located in the hydrophobic face, e.g., in the 829P and 843P mutants, showed greatly decreased Env localization to lipid rafts (Fig. (Fig.6B);6B); however, these mutations did not have significant impacts on one-cycle viral infectivity, virus replication kinetics, or Env incorporation (Fig. (Fig.7).7). On the other hand, Env localization in lipid rafts does not ensure viral infectivity. We recently showed that the highly conserved putative cholesterol-binding LWYIK motif, located at residues 679 to 683, does not have an apparent role in Env's association with lipid rafts, despite its role in the gp41-mediated fusion process (17).
Of note, exclusion of mutant proteins, such as those of the 829P, 843P, and TM mutants, from Triton X-100-resistant lipid rafts cannot be attributed to their abnormal intracellular trafficking to the plasma membrane. The observation that all Pro-substituted mutants, with the exception of the 833P mutant, were efficiently processed to yield gp120 and gp41 (Fig. 7E and G) implies that they are normally transported to the trans-Golgi network, where the Env precursor is cleaved to produce gp120 and gp41. Protein folding and oligomerization play an essential role in targeting viral Env glycoproteins to the plasma membrane (for reviews, see references 29 and 63). The fact that those Pro-substituted mutants (but not the 833P mutant) that were not localized in Triton X-100-resistant lipid rafts were still incorporated effectively into the virus (Fig. 7E and G) suggests that they are normally assembled and folded into an oligomeric structure, which is subsequently transported to the plasma membrane. This proposition was further supported by their similar intracellular localization and cell surface expression patterns compared to the WT Env (Fig. (Fig.8).8). Moreover, the normal Env incorporation phenotype of the TM844 and TM813 mutants (78) is in accordance with their normal expression on the cell surface (Fig. (Fig.8).8). Similarly, the observation that all LLP-1 deletion mutants, but not the Δ830 and Δ833 mutants, exhibited WT-like phenotypes of viral replication kinetics and Env incorporation into the virus (33) also implies that they are normally transported to and expressed on the cell surface. The findings that comparable amounts of gp41 and gp120 were associated with the 829P, 843P, TM844, and TM813 mutant virions also imply that these mutations neither affected the effectiveness of the TM region in anchoring the mutant Env into membranes nor altered gp120-gp41 interactions, despite these mutants being sequestered in non-Triton X-100-resistant membranes.
The recruitment of Env into lipid rafts without participation of Gag should have biological implications. Gag targeting to raft-like domains on the plasma membrane requires MA myristylation and protein oligomerization (22, 36). Nevertheless, in a model system, myristylated green fluorescent proteins containing palmitoylated or polybasic sequences colocalize with cholesterol and ganglioside GM1-enriched membrane domains, but not in lipid rafts/caveolae (42). This raises the possibility that the membrane targeting signal, i.e., myristylation and the basic residues, in MA might not be sufficient to target Gag to lipid rafts, and other factors may also contribute to HIV-1's localization to lipid rafts. HIV-1 MA targeting to lipid rafts may occur via its binding to phosphatidylinositol diphosphate [PI(4,5)P2] (66, 72), which is enriched in lipid rafts (28). Our results raise the interesting possibility that under certain circumstances, Gag targeting to lipid rafts may proceed via its interaction with Env, which is integrated into lipid raft membranes regardless of whether or not Gag is expressed. In support of the hypothesis that MA interacting with Env plays an active role during viral assembly at lipid rafts, coexpression with Env can redirect Gag assembly and budding in polarized epithelial cells, and mutations in MA and truncations in the gp41 cytoplasmic tail can abrogate this polarized budding (37, 52). Moreover, it is likely that accumulation of Env at lipid rafts may increase the affinity of Gag to bind to lipid rafts, thus promoting constant recruitment of Gag into these specialized membrane microdomains (68).
In sum, our results show that association with Triton X-100-resistant lipid rafts is an intrinsic property of HIV-1 Env and that gp41 harbors sequences and/or structural determinants mediating Env's association with and targeting to Triton X-100-resistant lipid rafts. To the best of our knowledge, this is the first report demonstrating that the amphipathic α-helix located in the cytoplasmic tail of a TM protein confers the ability to target the protein to lipid rafts.
We are grateful to Tun-Hou Lee for providing the pHXB2R3-based gp41 cytoplasmic tail truncation mutants and to Eric O. Freed for providing the pNL4-3-derived 12LE and 30LE Gag mutants. pNL4-3 was obtained from Malcolm Martin, pHXBn-PLAP-IRES-N+ from Benjamin K. Chen and David Baltimore, and LuSIV from Jason W. Roos and Janice E. Clements, through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH.
This work was supported by grants from the National Health Research Institute (NHRI-EX95-9431SI and NHRI-EX96-9431SI), Miaoli, Taiwan, and the National Science Council (NSC98-2320-B-001-012-MY3) and by Theme Program Project grants (5202401023-23-4m) from Academia Sinica, Taipei, Taiwan, Republic of China.
Published ahead of print on 30 September 2009.