|Home | About | Journals | Submit | Contact Us | Français|
The generally accepted model for human immunodeficiency virus type 1 (HIV-1) envelope glycoprotein topology includes a single membrane-spanning domain. An alternate model has been proposed which features multiple membrane-spanning domains. Consistent with the alternate model, a high percentage of HIV-1-infected individuals produce unusually robust antibody responses to a region of envelope, the so-called “Kennedy epitope,” that in the conventional model should be in the cytoplasm. Here we show analogous, robust antibody responses in simian immunodeficiency virus SIVmac239-infected rhesus macaques to a region of SIVmac239 envelope located in the C-terminal domain, which in the conventional model should be inside the cell. Sera from SIV-infected rhesus macaques consistently reacted with overlapping oligopeptides corresponding to a region located within the cytoplasmic domain of gp41 by the generally accepted model, at intensities comparable to those observed for immunodominant areas of the surface component gp120. Rabbit serum raised against this highly immunogenic region (HIR) reacted with SIV envelope in cell surface-staining experiments, as did monoclonal anti-HIR antibodies isolated from an SIVmac239-infected rhesus macaque. However, control experiments demonstrated that this surface staining could be explained in whole or in part by the release of envelope protein from expressing cells into the supernatant and the subsequent attachment to the surfaces of cells in the culture. Serum and monoclonal antibodies directed against the HIR failed to neutralize even the highly neutralization-sensitive strain SIVmac316. Furthermore, a potential N-linked glycosylation site located close to the HIR and postulated to be outside the cell in the alternate model was not glycosylated. An artificially introduced glycosylation site within the HIR was also not utilized for glycosylation. Together, these data support the conventional model of SIV envelope as a type Ia transmembrane protein with a single membrane-spanning domain and without any extracellular loops.
The envelope glycoprotein (Env) of the human immunodeficiency virus (HIV) and of the simian immunodeficiency virus (SIV) is synthesized as a precursor protein, gp160, which is subsequently cleaved into surface (SU) and transmembrane (TM) subunits, also referred to as gp120 and gp41, respectively. The two subunits remain noncovalently associated after cleavage and are incorporated as trimers into virions during the budding process. In the mature virion, gp120 mediates the recognition of and binding to the host cell receptor, while gp41 anchors the envelope complex in the virion's plasma membrane and effects fusion with the host cell membrane.
The generally accepted model for Env describes it as a type Ia transmembrane protein, i.e., as having one extracellular domain including the amino terminus with a cleavable signal peptide, a single membrane-spanning domain, and one intracellular domain including the carboxy terminus. For the purposes of this report, we will refer to the sequences corresponding to the intracellular domain of the generally accepted model as gp41 C-terminal domain (gp41CTD). In contradiction to this classical model, several studies have described antibodies strongly reacting with a region situated C terminally to the membrane-spanning domain, thought to be located within the cell, in serum samples of HIV-infected patients (6, 10, 23, 30, 59). Furthermore, some groups have reported that antibodies against this region are able to modestly neutralize some strains of HIV type 1 (HIV-1) and HIV-2 under modified conditions in vitro (3, 9, 15, 19, 25, 35, 36).
Although not consistently supported by other studies (16, 34, 41, 45, 52), these observations have led to the proposal of an alternate model in which part of the HIV-1 gp41CTD forms an extracellular loop either constitutively or only during the fusion process, thereby exposing the immunogenic region outside of the cell (14, 17, 35). In such a conformation, however, the well-established membrane-proximal YXXΦ motif, shown unambiguously to effect clathrin-mediated endocytosis of Env, would be located outside the cell and therefore nonfunctional, in direct contradiction with several publications (1, 4, 5, 32, 43, 50, 53). Proponents of the alternate model have addressed this inconsistency by suggesting that only a minority of Env molecules assume the conformation with an extracellular loop or that the immunogenic region is only exposed during or after fusion. This alternate model remains controversial; while Steckbeck et al. (58) recently reported reactivity of antibodies against the immunogenic region on the surface of Env-expressing cells but not on intact virions, another recent study by Liu et al. (34) found no conclusive evidence supporting the formation of an extracellular loop on Env-expressing cells.
The envelope proteins of HIV-1 and SIV are structurally and functionally very similar, including their receptor usage and low spike number on the surface of infected cells and virions. However, they share only limited amino acid sequence identity, around 35%. The immunogenic region of the HIV-1 gp41CTD shares only ca. 11% amino acid sequence identity with the corresponding Env region of SIVmac isolates. Despite this lack of conservation, serum reactivity against the corresponding region of SIV Env was detected with samples from SIV-infected rhesus macaques used in previous studies (51, 64).
Here we characterize the highly immunogenic region (HIR) in the gp41CTD of SIVmac239 in detail and address the possibility of an extracellular loop experimentally. Using cell surface staining, glycosylation scanning, and neutralization assays, we found no evidence to support the existence of such an extracellular loop. Instead, we present evidence that the detection of the HIR on the surface of Env-expressing cells is an artifact of the standard methods commonly used.
All serum samples used in this study were obtained from rhesus macaques (Macaca mulatta) infected with SIVmac239 or its mutants SIVmac239 g23 and SIVmac239 g123, as detailed in reference 64 in the context of other studies. All animal procedures were performed in accordance with the Guide for the Care and Use of Laboratory Animals and approved by the Harvard Medical School Animal Care and Use Committee (40). Infected animals were housed at the New England Primate Research Center in a biosafety-level-3 containment facility in compliance with the standards established by the Association for Assessment and Accreditation of Laboratory Animal Care and the Harvard Medical School Animal Care and Use Committee.
A collection of oligopeptides, each 15 amino acids (aa) in length, covering the entire sequence of SIVmac239 Env was obtained from the NIH AIDS Research and Reference Reagent Program (NIH ARRRP). The sequence of each oligopeptide overlapped the sequence of its adjacent oligopeptides by 11 aa. Assays were conducted as described previously (51, 64). Briefly, single wells of 96-well plates were each coated with one of the peptides, incubated overnight at 4°C, blocked with 5% nonfat dry milk (NFDM) in phosphate-buffered saline (PBS), and incubated for 1 h at 37°C. Subsequently, wells were incubated with serum diluted 1:20 in 5% NFDM in PBS for 2 h at 37°C. Six washes were performed with PBS plus 0.05% Tween 20, and samples were incubated with goat anti-human immunoglobulin G antibody conjugated to horseradish peroxidase (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) in 5% NFDM in PBS for 1 h at 37°C. After 10 more washes with PBS plus 0.05% Tween 20, tetramethylbenzidine reagent (Calbiochem, Gibbstown, NJ) was added to each sample, and plates were incubated at room temperature for 30 min. Finally, the reaction was stopped by addition of 250 mM hydrochloric acid, and the optical density of each sample was gauged at a wavelength of 450 nm by use of a Wallac Victor V multilabel counter (PerkinElmer, Waltham, MA).
Hydrophilicity plots were performed using MacVector software version 9.5.2. (MacVector, Inc., Cary, NC). Plots are based on the Kyte-Doolittle scale and have a window size of 7 aa.
Serum from rabbits inoculated with synthetic oligopeptide corresponding to the highly immunogenic region (HIR) in the C-terminal domain of gp41 was obtained from Sigma-Genosys (Sigma-Genosys, The Woodlands, TX). A synthetic oligopeptide of the sequence CFSSPPSYFQQTHIQQDPALPTREGKER, coupled via the N-terminal cysteine to keyhole limpet hemocyanin (KLH), was used to inoculate New Zealand White rabbits (Oryctolagus cuniculus). Animals received three booster inoculations after the initial inoculation, 14 days apart each. Serum was drawn 84 days after the initial inoculation.
Total immunoglobulin G (IgG) was purified from this anti-HIR serum with protein A/G agarose beads (Thermo Fisher Scientific, Inc., Rockford, IL), following the manufacturer's instructions. Dialysis cassettes with a molecular mass cutoff of 10,000 Da (Thermo Fisher Scientific, Inc.) were used to replace the elution buffer with PBS, pH 7.2. The final volume was adjusted to equal the original volume of serum used for purification. The purity of the isolated IgG was verified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) with subsequent Coomassie staining.
Serum samples from SIVmac239-infected rhesus macaques were tested by standard enzyme-linked immunosorbent assay (ELISA) for their ability to react with the same oligopeptide used to inoculate rabbits to generate anti-HIR serum (see above). A lymph node biopsy was performed on the animal whose serum displayed the highest reactivity. After extensive disaggregation, the cell suspension was repeatedly washed, followed by mononuclear cell separation by standard Ficoll density gradient. IgG-expressing cells were then isolated with anti-IgG magnetic beads (Miltenyi Biotec Inc., Auburn, CA). Cells were subsequently incubated with supernatant containing macacine herpesvirus 4 (rhesus Epstein-Barr virus[EBV]-like herpesvirus) in the presence of type B stimulatory oligonucleotides (1 μM CpG; Invivogen, San Diego, CA) and then cultured over irradiated feeder cells on 96-well plates at 37°C in a humid atmosphere of 5% CO2 until B-cell transformation occurred. Approximately 4.5 weeks later, supernatants were screened by standard ELISA for reactivity against the oligopeptide corresponding to the HIR. Two positive wells were selected, and dilution cloning was performed until the monoclonality of the population of interest was ensured. Both cell lines were then expanded and transferred to serum-free medium. Next, IgG was affinity purified with protein A/G Sepharose beads (GE Healthcare, Piscataway, NJ) followed by quantification with a Nanodrop UV spectrometer (Thermo Fisher Scientific, Inc.). The specificity of both monoclonal antibodies, designated as 4B2 and 6E6, was confirmed by testing their reactivity with the oligopeptide corresponding to the HIR by ELISA. Neither antibody reacted with recombinant SIVmac239 proteins gp140 and gp120 (Immune Technology Corp., New York, NY), which do not include the HIR. The monoclonal B-cell lines secreting antibodies 4B2 and 6E6, respectively, will be made available through the NIH ARRRP.
The plasmid encoding expression-optimized SIVmac239 Env, designated as 64S, was a generous gift from George Pavlakis (NCI Frederick) (49). Mutations were introduced into this plasmid using the QuikChange II site-directed mutagenesis kit (Agilent Technologies, Inc., Santa Clara, CA) according to the manufacturer's instructions. Primers were synthesized by Sigma-Aldrich (St. Louis, MO). The specificity and exclusivity of mutations were verified by sequencing of the entire env reading frame (Retrogen, Inc., San Diego, CA).
HEK293T/17 cells (ATCC, Manassas, VA) were maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum, 25 mM HEPES, 2 mM l-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin (all purchased from Invitrogen Corp., Carlsbad, CA).
For flow cytometry and immunoblot analysis, cells were transiently transfected with FuGENE 6 (Roche Applied Science, Indianapolis, IN) according to the manufacturer's instructions. A ratio of 3 μl FuGENE 6 per 1 μg DNA and per 97 μl untreated DMEM was used for transfection mixtures.
SIVmac316 virions were prepared and quantified as described previously (64).
Thirty-eight hours after transfection as described above, HEK293T cells were detached by brief (<15 s) incubation with 0.05% trypsin in Dulbecco's phosphate-buffered saline (DPBS) (pH 7.4, without Ca2+ or Mg2+; Invitrogen Corp.), washed off the tissue culture flask with ice-cold DPBS, and stored on ice while being counted. Washing cells off the flask without the use of trypsin led to essentially equivalent results but increased nonspecific staining. Per sample, 5 × 105 cells were transferred to polystyrene tubes and washed twice with ice-cold 2.5% goat serum in DPBS. Samples were then incubated at 4°C with primary antibody diluted in 2.5% goat serum in DPBS for 30 min. A final concentration of 5 ng/μl was used for monoclonal rhesus antibodies 1.9C (27), 4B2, and 6E6, and of 2 ng/μl for monoclonal mouse anti-FLAG antibody M2 (Sigma-Aldrich). Rabbit anti-HIR serum and purified IgG from this serum were used at a final dilution of 1:1,000. Subsequently, cells were washed twice with ice-cold 2.5% goat serum in DPBS and then incubated for 30 min at 4°C in the dark with the appropriate secondary antibody coupled to a fluorophore, diluted in 2.5% goat serum in DPBS. Samples were then washed three times with ice-cold DPBS, fixed in ice-cold 2% formaldehyde in DPBS, and stored at 4°C until analysis by flow cytometry, which was performed within 12 to 18 h after staining. Data were analyzed using FlowJo software version 6.4.1. (Tree Star, Inc., Ashland, OR).
For supernatant transfer experiments, the medium containing transfection reagent and plasmid DNA was removed 12 h after transfection, and cells were washed once with fresh medium and then incubated with fresh medium for another 24 h. Thirty-six hours after transfection, medium from cells transfected to express Env or controls was removed and centrifuged for 5 min at room temperature and 1,000 rpm to pellet cellular debris and floating cells. The supernatant was used to replace the medium of cells transfected with unrelated empty vector DNA. These samples were then incubated for another 2.5 h at 37°C prior to processing for flow cytometry staining as described above.
Thirty-eight hours after transfection as described above, HEK293T cells were washed off the tissue culture plate with ice-cold DPBS, washed once with ice-cold DPBS, and then lysed in NP-40 lysis buffer (50 mM HEPES, 150 mM NaCl, 1% NP-40 in H2O) with Complete protease inhibitor cocktail (Roche Applied Science). Samples were vortexed vigorously for 1 min and then centrifuged for 10 min at 1.6 × 104 relative centrifugal force (rcf) and 4°C. Supernatants were subsequently transferred to fresh vials and mixed with Laemmli buffer (Sigma-Aldrich). Proteins were then separated by SDS-PAGE on a 12% gel and transferred onto a polyvinylidene fluoride membrane. gp160 and gp41 were detected with hybridoma supernatant containing the monoclonal mouse anti-gp41 antibody KK41 (NIH ARRRP) (31); endogenous β-tubulin was detected with a rabbit anti-β-tubulin antibody from Cell Signaling Technology (Danvers, MA).
The sensitivity of infectious SIVmac316 virions to neutralization by IgG purified from anti-HIR serum and monoclonal anti-HIR antibodies 4B2 and 6E6 was measured with a secreted alkaline phosphatase (SEAP) reporter assay as described previously (37, 64). Virus equivalent to 5 ng p27 was used to infect the C8166-45 SIV-SEAP reporter cell line (5,000 cells/well on 96-well plates), which expresses SEAP under the control of the SIV long terminal repeat (LTR) region. Before infection, virions were incubated with a series of 2- or 4-fold dilutions of purified IgG from anti-HIR serum, heat-inactivated control serum from rhesus macaques, monoclonal anti-HIR antibodies 4B2 or 6E6, or hybridoma supernatant containing the monoclonal mouse anti-gp41 antibody KK41 (31) for 1 h at 37°C. Supernatants were harvested at day 3 or 5 after infection, and SEAP activity was determined by means of the chemiluminescent Phospha-Light SEAP assay system (Applied Biosystems, Foster City, CA), following the manufacturer's instructions with the previously described modifications (37). Luminescence was measured using a Wallac Victor V multilabel counter (PerkinElmer). All samples were measured in triplicate and are reported as a percentage of the SEAP activity detected in the absence of immunoglobulin.
In two previous unrelated studies (51, 64), sera from rhesus macaques (Macaca mulatta) infected with SIVmac239 or mutants thereof reacted strongly, as measured by ELOISAs, with overlapping oligopeptides representing a region of gp41 generally believed to be located within the cytoplasm. In order to verify and quantify this observation, we pooled and reanalyzed the raw data of these experiments and combined them with three additional, independent sample sets (Fig. 1A). The reactivity of the highly immunogenic region (HIR) in the C-terminal domain of gp41 (gp41CTD) with sera from SIV-infected rhesus macaques was comparable in intensity and standard deviation with the reactivity of the well-established immunodominant regions of Env, i.e., the V1/V2 loop region, the C-terminal end of gp120, and a region in the extracellular domain of gp41. Separate analysis of each of the sample subsets yielded qualitatively identical results (data not shown). Further analysis identified the HIR as an area of 27 amino acids (aa), namely, F725 to R751 (Fig. 1B), encompassing peptides 6709, 6710, 6711, and 6712 of the set of 15-mer oligopeptides, each overlapping by 11 aa. All of the 10 serum samples analyzed reacted strongly with at least one of these peptides. With the exception of two samples which also reacted with peptide 6744, representing the C-terminal 15 aa (A865 to L879), serum reactivity within gp41CTD was confined to peptides 6706 to 6712, representing aa M713 to R751.
The HIR within gp41CTD of SIVmac239 is highly similar in location, if not in sequence, to the immunogenic region of the HIV-1 gp41CTD, the so-called “Kennedy epitope” (13, 30, 58). It was this high immunogenicity in HIV-1 which led to the proposal of an alternative topology featuring an extracellular loop (14, 17, 26). In order to assess whether any part of the gp41CTD of SIVmac239 fulfills the biophysical requirements for such an extracellular loop, we performed an in silico analysis of its hydrophilicity distribution (Fig. 2). Amino acids L684 to A715 form a consistently hydrophobic region, corresponding well to the membrane-spanning domain of the generally accepted model. This is followed by a highly hydrophilic area (aa K716 to E767), which includes the entire HIR. Importantly, this hydrophilic region leads into a stretch of moderate but distinct hydrophobicity (aa Y768 to L802). These results mirror the outcome of a similar analysis on HIV-1 Env reported by Hollier and Dimmock (26), again despite a lack of high sequence similarity. The second hydrophobic stretch (aa Y768 to L802) is long enough to form an additional membrane-spanning domain; however, it includes five charged residues and a palmitoylated cysteine (C787) (2, 62), making it an unlikely candidate for a membrane crossing. Theoretically, a very short membrane-spanning domain could also be formed by aa Y768 to F784, reentering the cytoplasm before the palmitoylated cysteine; such a conformation would still have to accommodate three charged residues inside the plasma membrane. Nonetheless, these properties are consistent with the hypothetical possibility of an alternative structure in which the classical membrane-spanning domain would traverse the lipid bilayer, make a turn near the interface with the cytoplasm, and cross back toward the extracellular surface, thus allowing the gp41CTD to form an extracellular loop (including the HIR) and to eventually enter the cytoplasm via a second membrane-spanning domain (Fig. 2C).
To address the question of whether part of the gp41CTD might be exposed outside of the cell or virion, we obtained serum from rabbits inoculated with an oligopeptide corresponding to the HIR sequence (anti-HIR serum) and used it for surface staining of HEK293T cells transiently transfected with expression-optimized SIVmac239 env by flow cytometry. Both anti-HIR serum and IgG purified from it stained the surface of Env-expressing cells, but not the surface of cells not expressing Env (Fig. 3A). Serum drawn from the same animal prior to immunization did not stain Env-expressing cells (Fig. 3A). Anti-HIR serum also stained the surface of cells expressing Env of SIVmac316 (Env316), which differs in its HIR sequence from Env of SIVmac239 (Env239) only in the last amino acid (R751G) (Fig. 3B).
To ensure the specificity of binding, we performed cell surface staining with progressively truncated Env239 (Fig. 3C). Anti-HIR serum reacted with full-length Env and with Env that was truncated C terminally to the HIR (E767stop) but not with Env truncated within the N-terminal first quarter (Y731stop) or at the N-terminal end (F725stop) of the HIR. Truncation of gp41CTD drastically increased the cell surface expression of Env compared to the full length protein, as shown by staining with the monoclonal anti-gp120 antibody 1.9C. This observation is consistent with previous reports describing that such truncations decrease the rate of Env endocytosis and increase the levels of Env surface expression (5, 65). However, the reactivity of anti-HIR serum with cells expressing Env E767stop did not reflect this increase compared to full-length Env, in contrast to what would be predicted if the HIR were indeed exposed on the cell surface.
To obtain further evidence for the specificity of cell surface staining of the HIR, we isolated two monoclonal anti-HIR antibodies from lymph node biopsies of a rhesus macaque infected with SIVmac239. These antibodies, 4B2 and 6E6, reacted with an oligopeptide corresponding to the complete HIR sequence (Fig. 4A). They did not react with gp120 or with Env protein truncated prior to the classical membrane-spanning domain (gp140) (Fig. 4A). In addition to its reactivity with the complete HIR epitope, 4B2 also reacted with the core of the HIR, represented by the oligopeptide centered around the sequence common to all four HIR oligopeptides (Fig. 1B) (peptide 6711; aa sequence QQTHIQQDPALPTRE). Similarly, in an ELOISA with overlapping oligopeptides representing the entirety of Env239, 4B2 reacted only with peptide 6711 and the adjacent 6712 (IQQDPALPTREGKER), confirming that 4B2 is indeed specific for the HIR and recognizes the linear epitope IQQDPALPTRE (Fig. 4B). Conversely, 6E6 reacted only with the complete HIR epitope but did not react with the HIR core epitope (Fig. 4A) or any 15-mer oligopeptide in an Env-specific ELOISA (Fig. 4B), indicating that 6E6 may recognize a nonlinear epitope formed by the complete HIR sequence.
Having established the specificity of 4B2 and 6E6 for the HIR, we next tested their ability to detect Env239 and Env316 on the surface of transiently transfected cells by flow cytometry. Both 4B2 and 6E6 efficiently stained cells expressing Env239 and Env316, but neither displayed significant reactivity with the Env239 mutant truncated at the N-terminal end of the HIR (F725stop) (Fig. 4C).
As mentioned previously, the palmitoylated cysteine 787 is located within the hydrophobic region of gp41CTD, C terminally to the HIR. If gp41CTD did indeed form an extracellular loop, a second membrane-spanning domain might traverse the lipid bilayer and reach the cytoplasm shortly before C787, which could then be palmitoylated and thus anchor the structure. Therefore, preventing this palmitoylation might be predicted to change the conformation of a hypothetical extracellular loop and lead to a change in cell surface exposure of the HIR. Hence, we tested reactivity of anti-HIR serum with an Env mutant deficient in the palmitoylated cysteine (C787A) on the surface of transiently transfected HEK293T cells (Fig. 5A). However, the C787A mutation had no effect on the reactivity of Env with anti-HIR serum or with 1.9C.
The classical membrane-spanning domain contains a well-conserved arginine, R705, of unknown function. For the alternative structure model to be viable, the first membrane-spanning domain would have to enter the lipid bilayer, almost traverse it, form a turn, and then cross back toward the extracellular side. Such a turn could potentially be accomplished by the charged side chain of R705 interacting with the polar interface of phospholipids and the cytosol, thereby “pulling” the center of the membrane-spanning domain toward the cytosol—a mechanism referred to as “snorkeling” (8, 20, 61, 63). A disruption of this snorkeling by mutating R705 to a hydrophobic amino acid might thus be expected to interfere with the formation of a hypothetical extracellular loop. Therefore, we introduced the mutation R705L into full-length Env and Env E767stop and tested cell surface reactivity of anti-HIR serum with HEK293T cells transiently expressing these constructs (Fig. 5B). However, the mutation did not change the reactivity of Env with anti-HIR serum in the context of full-length or truncated Env.
In both the classical and the alternative models of gp41CTD topology, the C terminus is located in the cytoplasm. We added a C-terminal FLAG epitope to Env (Env239 C-FLAG) and tested for cell surface exposure on transiently transfected HEK293T cells (Fig. 5C). Unexpectedly, anti-FLAG antibody M2 reacted very efficiently with Env239 C-FLAG in the absence of cell permeabilization, indicating the exposure of the C terminus of Env on the surface of cells. These data provided the first evidence that cell surface reactivity of anti-HIR serum and monoclonal anti-HIR antibodies with Env may be an experimental artifact and may not reflect the formation of an extracellular loop by gp41CTD.
Based on this observation, we hypothesized that the supernatant of Env-expressing cells contains released, free Env molecules which can bind to the surface of cells. To test this experimentally, medium from Env-expressing cells was removed, cleared of cells and cellular debris by centrifugation, and the supernatant was transferred onto mock-transfected cells. After incubation for 2.5 h, these mock-transfected cells were washed and processed for flow cytometry using the same protocol as in the previous experiments. Importantly, the monoclonal anti-HIR antibody 4B2 efficiently stained mock-transfected cells that had been incubated with the supernatant of Env239- or Env239 C-FLAG-expressing cells but not mock-transfected cells that had been incubated with the supernatant of mock-transfected cells (Fig. 6, top). Similarly, the highly specific, monoclonal anti-FLAG antibody M2 stained mock-transfected cells that had been incubated with the supernatant from cells expressing Env239 C-FLAG but not mock-transfected cells that had been incubated with the supernatant from Env239-expressing or mock-transfected cells (Fig. 6, bottom). The same results were obtained when untransfected cells were used as supernatant recipients rather than mock-transfected cells (data not shown). These data unambiguously demonstrate that detection of the HIR on Env-expressing cells is due, in whole or in part, to Env release into the supernatant and subsequent binding of this free Env protein to the surface of cells in the culture.
As surface staining of Env-expressing cells was shown to be unsuitable for determining whether the HIR is naturally exposed on the surface of Env-expressing cells, we used a completely different approach to address this question. Potential N-linked glycosylation motifs, which are characterized by the sequence NXS/T (where X can be any amino acid except proline; acidic residues are suboptimal), are usually glycosylated when exposed on the extracellular side, whereas the same motif remains unmodified when localized to the cytosol (11). gp41CTD features exactly one potential N-linked glycosylation motif, 759NSS761, located only 8 aa C terminally of the HIR and N terminally of the hypothetical second membrane-spanning domain. If gp41CTD did indeed form an extracellular loop, this motif may be expected to be glycosylated, leading to a size shift detectable by Western blotting. Therefore, we introduced the mutation N759Q into Env, thus eliminating the motif. However, the acceptor asparagine of an N-linked glycosylation motif generally needs to be at least 14 aa removed from the following membrane-spanning domain in order to be efficiently glycosylated (11, 42, 46, 47). N759 is located only 8 aa N terminally of the earliest possible start site of a hypothetical second membrane-spanning domain and may thus not be accessible to glycosylation, depending on the exact site of gp41CTD reinsertion into the membrane. Therefore, we created another Env mutant into which we introduced an artificial N-linked glycosylation motif within the HIR (Q733N), well removed from any putative membrane-spanning domain. As controls, we also used two Env variants missing a single N-linked glycosylation site each in the N-terminal extracellular part of gp41 (N634Q and N650Q) (28, 64). These mutant Envs were expressed in HEK293T cells by transient transfection, and the cell lysates were subjected to SDS-PAGE with subsequent immunoblotting (Fig. 7). Neither the removal of the preexisting glycosylation motif in gp41CTD (N759Q) nor the introduction of an artificial motif (Q733N) had any effect on gp41 mobility, whereas removal of either one of the N-terminal extracellular sites effected a clear size shift. Thus, the region of gp41CTD predicted by the alternative structure to be exposed extracellularly is not glycosylated, even when an optimal glycosylation site is introduced into it.
Next, we tested whether HIR-specific immunoglobulins were capable of neutralizing infectious SIVmac virions. As SIVmac239 is quite resistant to neutralization, and as anti-HIR serum (Fig. 3B), 4B2, and 6E6 (Fig. 4C) bind equally well to the HIR of Env239 and Env316, we performed neutralization assays with the closely related but very neutralization-sensitive strain SIVmac316. As shown in Fig. 8A, IgG purified from anti-HIR serum displayed no neutralizing activity against SIVmac316. Conversely, SIVmac316 was effectively neutralized by sera from SIV-infected rhesus macaques and by the monoclonal anti-gp41 antibody KK41 (31). Serum from a specific-pathogen-free (SPF) rhesus macaque had no neutralizing ability. Similarly, neither 4B2 nor 6E6 were capable of neutralizing SIVmac316 (Fig. 8B). 4B2 and 6E6 also did not neutralize SIVmac239 (data not shown). The inability of three different sources of IgG, all proven to react efficiently with the HIR of complete Env under nonreducing and nondenaturing conditions, to neutralize even the highly neutralization-sensitive strain SIVmac316 is consistent with the absence of exposure of the HIR on the surface of SIV virions.
The topology of HIV-1 gp41 continues to be the subject of considerable controversy. Reports have appeared even recently in support of the alternate model (58) and of the classical model (34). In our current study, we have expanded this debate for the first time to include gp41 of SIVmac. We have characterized a highly immunogenic region in gp41CTD which elicits strong serum reactivity in SIV-infected rhesus macaques. An analogous reactivity against the corresponding region of HIV-1 gp41CTD has been documented for sera from HIV-infected individuals (6, 10, 23, 30, 59). The existence of such highly immunogenic regions in the gp41CTD of both HIV-1 and SIVmac, which in the generally accepted model should not be exposed on the surface, underscores the functional conservation between the envelope glycoproteins of these two viruses in the absence of consistently high sequence conservation. This is further emphasized by the correspondence in location of these two highly immunogenic regions within the linear amino acid sequence and the conservation of a hydrophobic stretch with the theoretical potential to span the plasma membrane just downstream of the highly immunogenic region in both cases.
In the serum of an infected individual, the presence of immunoglobulins specific for cytoplasmic or nuclear viral antigens is a common phenomenon. For example, serum of patients infected with human herpesvirus 8 (HHV-8) characteristically reacts with latency-associated nuclear antigen 1 (LANA-1), which is localized to the nucleus (21, 22, 39). Similarly, HIV-1-infected individuals routinely develop antibodies against Gag, which is either located in the cytoplasm or encapsulated in virions during its normal life cycle (24, 29, 54). However, the reactivity of patient sera against such intracellular antigens is typically much lower than against extracellular antigens that are naturally and directly exposed to the humoral immune system. The observation that sera from SIV-infected rhesus macaques display reactivity against a specific area of the gp41CTD at levels comparable to the reactivity against the extracellular immunodominant regions of Env suggests that this region of the CTD either is naturally exposed on the cell or virion surface or is unusually immunogenic. The data presented here argue strongly that the natural topology of gp41 does not feature an extracellular loop of gp41CTD and that it is its inherent immunogenicity that must be responsible for the consistently strong antibody responses.
While anti-HIR serum and the monoclonal anti-HIR antibodies 4B2 and 6E6 did indeed bind to the surface of Env-expressing cells (Fig. 3 and and4C,4C, respectively), we show here that this reactivity is due in whole or in part to free Env released from Env-expressing cells binding to the surface of cells in the culture with remarkable efficiency (Fig. 6). This observation may have important implications for other studies as well, as it indicates that epitopes which appear to be accessible by cell surface staining may actually not be accessible on functional Env trimers as expressed naturally on the surface of cells. For the envelope of HIV, several groups have reported cell surface staining of HIV-infected cells with serum or antibodies directed against the HIR of gp41CTD (12, 15, 48), whereas others have described the absence of such staining (45, 52). One recent study reported surface staining of cells expressing HIV-1 Env only under conditions which triggered fusion and at cooled temperature and concluded that gp41CTD was exposed only during the fusion event itself (35). The data presented here differ from these previous observations, as Env was expressed in HEK293T cells. These cells do not express CD4 and are therefore not able to trigger the fusion mechanism of Env; thus, the observed cell surface presentation of the HIR described here is independent of the fusion event.
A recent publication by Steckbeck et al. (58) employing a similar method of transfection and detection found that an epitope tag inserted into HIV-1 gp41CTD reacted with its antibody in cell surface staining when located in the immunogenic region but not when inserted closer to the C terminus. This observation is in direct contradiction to our results obtained with a C-terminal FLAG epitope (Fig. 5C). The results of Steckbeck et al. might reflect an inaccessibility of the C-terminal epitope tag at the location of insertion, a possibility that was not experimentally excluded by the authors and that is supported by the fact that the tag was inserted into the lentivirus lytic peptide 2 (LLP2) region, which is believed to form a tight secondary structure and possibly even a direct interaction with the plasma membrane (18, 57, 60). Interestingly, an Env239 expression construct with a FLAG epitope substitution located only 21 to 29 aa before the C terminus, close to the equivalent of the LLP1 region in HIV-1 gp41, exhibited little or no reactivity with the anti-FLAG antibody in our own surface-staining experiments (data not shown). The same Env mutant reacted with the monoclonal anti-gp120 antibody 1.9C at the same level as wild-type Env. Our data demonstrate unequivocally that surface staining of the HIR can be fully explained by the release of Env protein into the cell-free supernatant and the subsequent attachment of this Env to the surface of cells.
The results of our control experiments for the detection of cell surface staining raised serious concerns about the suitability of this assay for examining the potential extracellular exposure of the HIR. Independent evidence against such a natural HIR exposure on the surface of Env-expressing cells was obtained by other means. We have shown that the HIR is not glycosylated, neither at a naturally occurring nor at an artificially introduced glycosylation motif (Fig. 7). N-linked glycosylation of extracellularly exposed NXS/T motifs is so reliable that so-called “glycosylation scanning,” where artificial glycosylation motifs are introduced at various points within a protein of interest, has been commonly used to determine membrane topology (for examples, see references 11, 33, 38, 55, and 56). The naturally occurring 759NSS761 sequence within gp41CTD may or may not be too close to the unknown starting location of a hypothetical second membrane-spanning domain to be glycosylated; the artificial site introduced by the Q733N mutation, however, is far removed from any potential steric interference. Neither motif was glycosylated, providing strong evidence against a natural exposure of the HIR outside of the cell.
While the previously discussed methods were aimed at investigating the possibility of an extracellular loop formed by gp41CTD on the surface of Env-expressing cells, we also examined the possibility of HIR exposure on infectious virions by means of neutralization assays. It has been suggested that the ability of an antibody to neutralize viral infectivity relates directly to its ability to bind to the surfaces of target virion (7, 44). The SIVmac316 strain is closely related to SIVmac239 but is exquisitely sensitive to neutralization by a wide range of antibodies with a wide range of specificities (27). If the HIR were indeed exposed on the surfaces of virions, anti-HIR antibodies would be expected to neutralize SIVmac316 to one extent or another. However, neither IgG purified from anti-HIR serum (Fig. 8A) nor the monoclonal anti-HIR antibodies 4B2 and 6E6 (Fig. 8B) displayed any ability to neutralize SIVmac316. This observation is consistent with the classical model of gp41CTD conformation without an external loop. It is also consistent with the report by Steckbeck et al. (58), who did not find any evidence for exposure of the HIV-1 gp41CTD immunogenic region on intact virions.
Taken together, the data presented in this study argue strongly against a natural extracellular exposure of the highly immunogenic region located within the C-terminal domain of SIVmac gp41. We have shown that recognition of the HIR on the surface of Env-expressing cells by anti-HIR serum and monoclonal anti-HIR antibodies does not increase with the level of cell surface expression of Env (Fig. 3C) and is independent of structural elements that would be expected to play a role in the formation of an extracellular loop (Fig. 5A and B). Instead, this surface reactivity can be fully explained by the release of free Env protein from Env-expressing cells into the supernatant and the subsequent binding of this free Env to the surface of cells in the culture (Fig. 6). Further, the region of gp41CTD that is predicted to be exposed extracellularly in the alternative model is not glycosylated to any detectable extent (Fig. 7), providing evidence that this region of gp41 is not presented to the lumen of the endoplasmic reticulum and therefore not located outside of the cell after transport to the cell membrane. Finally, anti-HIR serum and anti-HIR antibodies are not able to neutralize even the highly neutralization-sensitive strain SIVmac316 (Fig. 8), despite efficient binding to the Env protein under nondenaturing and nonreducing conditions.
This work was supported by grant 5RO1AI025328 from the National Institutes of Health (NIH) to R.C.D., by funding from the IAVI Neutralizing Antibody Consortium to R.C.D., by the graduate college program GRK1071 of the Deutsche Forschungsgemeinschaft, and by a scholar stipend from GlaxoSmithKline to T.S.P. This work was also supported by base grant RR00168 from the NIH to the New England Primate Research Center.
We are grateful to George Pavlakis (NCI Frederick) for kindly sharing his expression plasmids and to Jacqueline Bixby for excellent technical assistance. Furthermore, we thank Jens Kuhn (Integrated Research Facility at Fort Detrick, NIAID) and Sheli Radoshitzky (USAMRIID) for critically reading the manuscript.
Published ahead of print 9 November 2011