Conceived and designed the experiments: JDS RCM. Performed the experiments: JDS TJS CS. Analyzed the data: JDS TJS. Contributed reagents/materials/analysis tools: JDS TJS CS. Wrote the paper: JDS RCM.
The authors have declared that no competing interests exist.
The C-terminal tail (CTT) of the HIV-1 gp41 envelope (Env) protein is increasingly recognized as an important determinant of Env structure and functional properties, including fusogenicity and antigenicity. While the CTT has been commonly referred to as the “intracytoplasmic domain” based on the assumption of an exclusive localization inside the membrane lipid bilayer, early antigenicity studies and recent biochemical analyses have produced a credible case for surface exposure of specific CTT sequences, including the classical “Kennedy epitope” (KE) of gp41, leading to an alternative model of gp41 topology with multiple membrane-spanning domains. The current study was designed to test these conflicting models of CTT topology by characterizing the exposure of native CTT sequences and substituted VSV-G epitope tags in cell- and virion-associated Env to reference monoclonal antibodies (MAbs). Surface staining and FACS analysis of intact, Env-expressing cells demonstrated that the KE is accessible to binding by MAbs directed to both an inserted VSV-G epitope tag and the native KE sequence. Importantly, the VSV-G tag was only reactive when inserted into the KE; no reactivity was observed in cells expressing Env with the VSV-G tag inserted into the LLP2 domain. In contrast to cell-surface expressed Env, no binding of KE-directed MAbs was observed to Env on the surface of intact virions using either immune precipitation or surface plasmon resonance spectroscopy. These data indicate apparently distinct CTT topologies for virion- and cell-associated Env species and add to the case for a reconsideration of CTT topology that is more complex than currently envisioned.
Human immunodeficiency virus (HIV) infects humans predominantly through interaction of the viral envelope glycoprotein (Env) with the primary receptor CD4 and coreceptors CCR5 or CXCR4 on the surface of target cells. Env is produced as a 160 kDa polyprotein that is subsequently processed by extensive glycosylation, multimerization, and proteolytic cleavage to yield the virion-associated trimeric complexes of non-covalently associated gp120-gp41 dimers
Studies addressing the CTT have traditionally examined: (i) the role of the CTT in viral Env incorporation
A.) Traditional CTT model with one membrane-spanning α-helix and a completely intracytoplasmic localization of the remaining CTT sequence. LLP domains have been placed at their presumed membrane-localized position. B.) Alternative CTT model with multiple MSD segments as proposed by Hollier and Dimmock
Early evidence for an alternative topological model for the CTT was provided by Kennedy and colleagues
More recently, Dimmock and colleagues have attempted to address this apparent discrepancy between the traditional model of an exclusively intracytoplasmic CTT and an alternative model where the KE is exposed
This multiple MSD topology model is compatible with various studies that have identified intracellular partners proposed to interact with CTT sequences
Finally, more recent studies have examined the accessibility of the lentivirus lytic peptide (LLP) regions of the CTT (c.f.
In the current study, we sought to test the proposed topology models by experimentally determining the surface exposure of the KE sequences of gp41 for comparison to the external and internal locations predicted by the respective CTT models. Towards this goal, we have examined the accessibility of the KE to antibody binding on the surface of Env-expressing cells and on the surface of intact mature virions using a combination of flow cytometry, immunoprecipitation, and surface plasmon resonance (SPR) spectroscopy techniques that minimize the potential for disruption of the lipid membrane, the integrity of which is crucial to topology mapping studies (see
To experimentally determine the exposure of the KE on the surface of Env-expressing cells, the antibody reactivity of transfected cells expressing HIV-1 89.6 Env with the VSV-G tag substituted in the KE (VSV-G-KE) or LLP2 (VSV-G-LLP2) segments was analyzed. The selected epitope allowed the use of a high affinity commercial anti-VSV-G MAb to compare the accessibility of a common epitope in the context of the KE segment and in a putative internal LLP2 segment of the CTT
As shown in
Sequence alignment of the membrane-spanning domain (MSD) and CTT of the Env proteins used in this study. All Env proteins were full-length gp160, however, only the sequences from K665 to the C-terminus are shown for simplicity. Structural and sequence domains are indicated in boxes above the corresponding sequence. VSV-G epitope tag indicated in bold; SAR1 and 1577 epitope indicated in gray box. KE – Kennedy epitope; LLP – lentivirus lytic peptide, MSD – membrane spanning domain.
Cells transfected with HIV-1 89.6 WT, VSV-G-KE, or VSV-G-LLP2 were analyzed by FACS to determine VSV-G epitope accessibility to anti-VSV-G MAb. A.) Intact, non-permeabilized Env-expressing cells were stained with α-VSV-G (AlexaFluor (AF) 700) and α-actin (AF 488). B.) Intact cells expressing Env were stained for surface exposure of the KE using a native KE antibody, SAR1. C.) Same as (B) except cells were permeabilized prior to staining.
Finally, as the CTT has been shown to contain signals important for association with detergent-resistant membranes (DRMs)
Env association with detergent resistant membranes was determined by sucrose gradient floatation followed by western blot analysis. Lane 1 represents the top of the gradient and lane 12 the bottom of the gradient. Blots were stained with anti-gp41 MAb Chessie 8 to determine the localization of gp41 in the bands shown.
Next, immunoprecipitation studies were performed to assess the ability of MAbs directed at the KE to pull down intact viral particles using protein G-coated paramagnetic beads. Both a gp120-specific antibody (termed C11) and a gp41 ectodomain specific antibody (7B2) were used to provide positive controls for Env binding and virion pull down. A p24-specific MAb (183-H12-5C) was used to detect disrupted virions in the pull down assays as a control for virion integrity. Of particular importance, several reference MAbs (2F5, 4E10, Z13e1) directed to the membrane proximal external region (MPER) of gp41 were used as controls, as this epitope is known to be closely associated with the membrane surface
To ensure that all the reference MAbs were able to bind their target protein, immunoprecipitations were initially performed with MAbs incubated with purified virions solubilized with Triton X-100. The product immunoprecipitates were then characterized by SDS-PAGE and western blot for the respective target protein (e.g. C11 immunoblotted for gp120, 7B2 immunoblotted for gp41, etc.). The open bars in
The indicated proteins and viral particles were immunoprecipitated using reference MAbs coupled to protein G-coated paramagnetic beads. (Top) Open bars represent % of target antigen precipitated when incubated with solubilized virus, while closed bars represent the % of input p24 precipitated by the corresponding MAb under native (intact virus) conditions. # = p<0.05 for MAbs compared to IgG control with solubilized virus; * = p<0.05 for MAbs with intact virus compared to IgG control. (Bottom) Representative p24 bands immunoprecipitated using the MAbs indicated in top panel with intact virus (Intact) or the bands of the target antigen from each MAb in detergent-disrupted virus (Solublized).
Having established that the MAbs were able to bind their target antigen in this assay format, we next tested the ability of the reference MAbs to precipitate intact virions under conditions similar to those used above without detergent solubilization. Results presented in
To complement these immunoprecipitation studies, binding analyses of gp41-specific MAbs to purified virions in solution were performed using SPR spectroscopy. This technique allowed the assessment of qualitative differences in relative antibody affinity in real time during the course of the interaction. Specifically, SPR assays were used to address the possibility that both KE and MPER-specific MAbs bound viral particles equally well, but that KE MAb binding to virions was unstable during the washing procedures used in the immunoprecipitation assay.
As shown in
MAbs 2F5 (MPER) and SAR1 (KE) were compared for relative binding rates and affinity using SPR spectroscopy to monitor antibody binding to purified intact HIV-1 virions. RU – resonance units.
The current studies were designed to distinguish between the two distinct topological models currently proposed for HIV-1 gp41 by determining the accessibility of key CTT sequences in the context of virions and Env-expressing cells to binding by reference MAbs. The results of these studies for the first time reveal in parallel assays a marked difference in the accessibility to antibody binding of the KE in a primary-isolate Env expressed on the surface of cells compared to intact virions. Thus, these observations indicate that the CTT may exist in different topologies depending on the membrane environment, suggesting that the CTT may actually be dynamic in its association with different lipid membranes or other cofactors.
While the goal of this study was to distinguish between the alternative models of CTT topology, the current data is only partially compatible with both models. In the case of virion-associated Env, the lack of KE exposure to antibody binding is consistent with the classical model of Env topology (
A dynamic model for CTT topology is consistent with the published studies by the Dimmock lab
Recent structural studies of diverse membrane proteins have resulted in an increased appreciation of the complexity and dynamics of membrane protein topology
A second aspect of current membrane protein structure studies that appears to be relevant to the case for a reconsideration of HIV-1 CTT membrane topology is the unique role of arginine (Arg) in membrane-associated proteins. While the early paradigm predicted Arg residues to be excluded from hydrophobic environments like a membrane lipid bilayer, more recent structural data indicate unique biochemical properties of Arg that produce unexpected functional properties for membrane proteins. For example, Arg residues can be incorporated as critical components of the membrane spanning domains of proteins, as observed with highly conserved Arg residue in the MSD of HIV-1 (c.f.
A second unique property of Arg is its ability in a repeat sequence (or homopeptide) to mediate transport of proteins and DNA or RNA across intact biological membranes, in contrast to homopeptides of other charged amino acids such as lysine (Lys) and histidine (His) that fail to efficiently cross lipid bilayers
A unique characteristic of the gp41 CTT of HIV-1 is that it contains two cationic amphipathic helical segments (LLP1 and LLP2) that are rich in Arg. In addition, the CTT contains a number of other highly conserved Arg residues. We and others have previously shown that synthetic LLP1 and LLP2 peptides assume a random coil conformation in aqueous environments, but rapidly convert to a helical conformation in a hydrophobic or membrane mimetic environment
Finally, we and others have previously reported that modifications in CTT length or amino acid composition can have marked affects on global Env structure and function, including virion incorporation, fusogenicity, and antibody reactivity. For example, we found that the mutation of two Arg residues to glutamate (Glu) residues in LLP2 of an HIV-1 provirus substantially reduced Env fusogenicity and altered the antibody reactivity and neutralization sensitivity of epitopes located in both the gp120 protein and the ectodomain of gp41
As often happens in HIV-1 research, our efforts to test two relatively simple models for CTT topology have indicated the need to consider more complicated and dynamic models of CTT membrane topology. This is perhaps a timely pursuit in light of the new paradigms of membrane structure and the importance of understanding the mechanisms by which the CTT serves as a determinant of HIV-1 Env structure and function on the surface of virions and cells.
HEK293T cells (ATCC) were maintained in DMEM (Invitrogen) supplemented with 10% (v/v) FBS. Cells were used at passage numbers less than 30. Gradient purified aldrithiol-2-inactivated HIV-1 NL4-3 virus (generously provided by Dr. Jeff Lifson, SAIC Frederick, Inc) used in this study has been described previously
As described for topology mapping of other membrane proteins
HEK293T cells were plated into six-well plates 24 hours prior to transfection. 80% confluent cells were transfected with 2.5 µg of the selected HIV-1 Env mutant DNA using LipofectamineTM LTX reagent and PLUS™ reagent, as recommended by the manufacturer (Invitrogen).
Cells were harvested 24 hours post-transfection by treatment with 2 mM EDTA. Cells were resuspended and washed twice with FACS wash buffer (1X PBS with 5% FBS) at 4°C prior to antibody staining. Reference MAbs to VSV-G (Roche Diagnostics), gp41 (SAR1), and actin (Sigma-Aldrich) were labeled immediately prior to staining using Zenon labeling kits (Invitrogen) following manufacturer's instructions. 106 cells were dual-stained by incubating with 5 µg fluorophore-labeled antibody (VSV-G or gp41) and 7.5 µg fluorophore-labeled actin antibody for 30 minutes on ice. Following staining, cells were washed thrice with FACS wash buffer at 4°C. Washed cells were resuspended and stained with 7-amino-actinomycin D (7-AAD). Cells were not fixed prior to analysis. Fluorescently-labeled cells were analyzed on a FACSAria (BD Biosciences). Live, intact cells (7-AAD negative) were selected for scatter characteristics, including selection of single-cell populations by doublet-discrimination analysis. PMT settings were adjusted on identically-stained mock transfected cells prior to analysis of Env-transfected cells. Data was collected for 5×104 7-AAD negative cells and analyzed for reactivity with fluorescently-labeled VSV-G or gp41 and actin antibodies.
HEK293T cells were transfected with 89.6 WT, VSV-G-KE, or VSV-G-LLP2 plasmids. 48 hours post-transfection, cells were subjected to sucrose gradient floatation as previously described
Protein G Dynabeads (Invitrogen) were prepared according to the manufacturer's directions. Briefly, anti-Env or anti-Gag antibodies (4 µg) were incubated with 20 µl protein G Dynabeads in 35 µl citrate-phosphate buffer, pH 5.0, with gentle shaking for 45 minutes at room temperature. Isotype-matched IgG controls were used for each species (murine, human, etc.) from which a MAb was derived. Beads were washed thrice with 0.5 ml citrate-phosphate buffer followed by resuspension in either 26 µl PBS (for intact virus) or 26 ul PBS with 1% Triton X-100 (for lysed virus) and 4 µl NL4-3 virus (equivalent to 1.1 µg p24). Virus-bead suspensions were incubated at 4°C for one hour with gentle shaking and subsequently washed thrice with 1X PBS. Following the final wash, beads were resuspended in NuPAGE SDS-PAGE buffer, heated at 70°C for 10 minutes, and the supernatant loaded onto 4–12% Bis-Tris NuPAGE gels. Gels were electrophoresed followed by transfer to polyvinylidene fluoride (PVDF) membranes using the Invitrogen iBlot system. Blots were blocked for one hour in 5% blotto (1X PBS with 5% dry milk). After blocking, blots were cut to allow separate staining of gp120 (>60 kDa), gp41 (30–60 kDa), and p24 (<30 kDa). gp120 was stained with rabbit anti-gp120 (Advanced Biotechnologies, Inc.), gp41 stained with Chessie 8, and p24 stained with Ag3.0 for 1.5 hours at room temperature. Blots were washed thrice with 1X PBS and 0.025% Tween 20 (PBS-T), followed by incubation with appropriate secondary antibody (anti-rabbit IgG or anti-mouse IgG conjugated to horseradish peroxidase) for one hour at room temperature. Blots were washed thrice in PBS-T with the gp120 blot receiving an additional wash in 1X PBS with 0.1% Triton X-100. Finally, blots were incubated with PicoWest substrate (Pierce) for one minute and reassembled for visualization on X-ray film.
Antibody IPs were quantified by densitometry analysis. For each IP, three independent X-ray exposures were scanned and analyzed using ImageJ (NIH). For intact virus, p24 bands, and for solublized virus bands from each MAb's target antigen, were selected for each protein, and the integrated area under the densitometry curve was compared to that of the viral input band to yield percent of the input protein in the immunoprecipitate. Percent inputs for each protein for each of the three exposures were averaged to yield the overall percent input immunoprecipitated per experiment. This was repeated for three independent experiments, and the results were averaged to yield the final percent input immunoprecipitated per antibody. One-way ANOVA analysis was performed in Prism 5.0b with the Dunnett Multiple Comparison test and IgG as the control column. The P value of the test was <0.0001. Individual P values for antibodies compared to the IgG control are indicated in each data set.
Antibody-virus interactions were analyzed using SPR spectroscopy using a Biacore 3000 instrument. Protein A was immobilized on CM5 sensor chips as described
The authors thank Dr. Jeff Lifson and the AIDS and Cancer Virus Program, SAIC Frederick, Inc., National Cancer Institute, Frederick, MD for providing the purified, inactivated HIV-1 NL4-3 used in these studies, Dr. Nigel Dimmock for providing the SAR1 hybridoma cell line, and Dr. Jodi Craigo for the creation of