HIV-1 Entry in SupT1-R5, CEM-ss, and Primary CD4⁺ T Cells Occurs at the Plasma Membrane and Does Not Require Endocytosis

Cell culture.

HEK293T cells were maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (Biochrom), 50 U/ml of penicillin, and 50 μg/ml of streptomycin (Gibco, Invitrogen). CEM-ss cells (40) were cultured in RPMI 1640 medium with GlutaMAX (Gibco, Invitrogen) with the same supplements. SupT1-R5 cells stably expressing exogenous CCR5 under puromycin selection (a kind gift from Robert Doms, University of Pennsylvania) were cultured in RPMI 1640 medium with GlutaMAX (Gibco, Invitrogen) with the same supplements indicated above and 0.3 μg/ml puromycin (Sigma-Aldrich). Peripheral blood mononuclear cells (PBMCs) were purified from the buffy coats of healthy blood donors by centrifugation through a Ficoll gradient (Amersham Biosciences) according to standard procedures, and CD4⁺ T cells were enriched by magnetic bead negative selection according to the manufacturer's protocol (CD4⁺ T cell isolation kit; Miltenyi Biotec). Alternatively, CD4⁺ T cells were directly enriched from the buffy coats of healthy blood donors using the RosetteSup human CD4⁺ T cell enrichment cocktail according to the manufacturer's protocol (Stemcell Technologies, Vancouver, Canada). CD4⁺ T cells were stimulated for 3 days using 100 IU/ml interleukin-2 (Biomol) and 2 μg/ml phytohemagglutinin (PHA; Sigma) and cultured in RPMI 1640 medium with GlutaMAX (Gibco, Invitrogen) with the supplements described above.

Plasmids.

The proviral HIV-1 plasmid pNL4-3 (41), pNL4-3.Env(−), and the Env expression vector pCAGGS.NL4-3-Xba (42), here designated pEnv.NL4-3, were described previously. The Env expression vector pCAGGS.NL4-3R5, encoding an R5-tropic Env glycoprotein that differs from Env_NL4-3 by seven point mutations within the V3 loop (42), was kindly provided by S. Pöhlmann (German Primate Center, Göttingen, Germany). An expression plasmid for the envelope glycoprotein of vesicular stomatitis virus (VSV-G) under the control of the cytomegalovirus promoter (pM3) was provided by D. von Laer (University of Innsbruck, Innsbruck, Austria). The empty expression vector pEnv.mock was constructed by deleting the Env-coding sequence from pEnv.NL4-3 using XhoI/EcoRI. Plasmid pEnvCD4(−) was constructed on the basis of pEnv.NL4-3 by exchanging an Acc65I/XhoI fragment for the corresponding fragment from the subviral plasmid pNL-A1(CD4⁻) carrying a deletion of the CD4 binding domain (43). Coreceptor binding-deficient pEnv.CoR(−) was constructed on the basis of pEnv.NL4-3 by exchanging an Acc65I/XhoI fragment for the corresponding fragment from pHenv319GE, an Env expression vector carrying a G-to-E mutation within the V3 loop at position 319 (44). Plasmid pEnv(Fus)V2E was constructed on the basis of pEnv.NL4-3 by exchanging an AleI/Acc65I fragment for the corresponding fragment from pcDNA-gp41V2E-TR752 (a kind gift from V. Bosch, DKFZ Heidelberg), carrying a V2E mutation in the gp41 fusion peptide (45). The expression vectors pEnv.5002 (plasma) and pEnv.4051 (plasma), carrying an X4-tropic or an R5-tropic env gene, respectively, directly isolated from patient plasma (46), were kindly provided by R. Swanstrom (University of North Carolina, Chapel Hill, NC, USA). Plasmids pNLC4-3.Env(−), pNLC4-3.Env(−).mCherry, and pNLC4-3.Env(−).eGFP were constructed on the basis of the previously described plasmids pNLC4-3, pNLC4-3.mCherry, and pNLC4-3.eGFP, respectively (30, 47), by exchange of an AgeI/XhoI fragment against the corresponding fragment from subviral plasmid PA3 (30), in which an NdeI site at nucleotide 6401 of the NL4-3 sequence was filled in with Klenow polymerase to generate a frameshift in the env open reading frame leading to a premature stop of translation.

Expression vectors for wild-type (wt) Dyn-2 [Dyn-2(wt)] fused with enhanced green fluorescent protein (eGFP) [pDyn-2(aa)-GFP] and the Dyn-2 K44A variant fused with eGFP [pDyn-2(aa)K44A-GFP] have been described previously (48). pcDNA3.1(+)-Dyn-2-wild-type-mCherry and pcDNA3.1(+)-Dyn-2-K44A-mCherry (23) were kindly provided by G. Melikyan (Emory University, Atlanta, GA, USA). The Dyn-2 sequences from these plasmids were excised using HindIII and EcoRI and cloned into pmCherry-N1 (Clontech) in order to exchange the ampicillin resistance marker for a kanamycin (Kana) resistance marker, obtaining pDyn-2-wt-mCherry^Kana and pDyn-2-K44A-mCherry^Kana, respectively; this was done to avoid the background in the Vpr.β-lactamase (Vpr.BlaM) fusion assay. Plasmid pMM310 (49) was kindly provided by N. Landau (New York University, New York, NY, USA). Plasmid peGFP.Vpr (50) was a kind gift of T. Hope (Northwestern University, Chicago, IL, USA).

Transfection procedures and virus stocks.

HEK293T cells were transfected using polyethyleneimine (Sigma-Aldrich) according to standard procedures in 10-cm tissue culture dishes (Falcon; BD Bioscience). To produce fluorescently labeled virions, 7.5 μg of pNLC4-3.Env(−) was cotransfected with 7.5 μg of pNLC4-3.Env(−).mCherry and with 2.5 μg of the respective Env expression vector. For production of Vpr.β-lactamase-containing particles (HIV-1^BlaM), cells were transfected with 7.5 μg of pNLC4-3.Env(−), 1.5 μg of pMM310, and 7.5 μg of the respective Env expression vector. For production of eGFP.Vpr-containing particles, cells were transfected with 15 μg of either pNLC4-3 or pNL4-3 and 1.5 μg of peGFP.Vpr. For production of infectious HIV-1 pseudotyped with the respective Env variant, cells were transfected with 15 μg of pNL4-3.Env(−) and 1.5 μg of the respective Env expression vector. Media were harvested 48 h after transfection and purified through a 20% (wt/wt) sucrose cushion.

For electroporation of SupT1-R5 cells, 1.5 × 10⁷ cells were transferred to electroporation buffer (20 mM HEPES, 135 mM KCl, 2 mM MgCl₂, pH 7.6). Electroporation was performed using a Bio-Rad Gene Pulser Xcell system and 0.4-cm electroporation cuvettes (Invitrogen) in the presence of 15 to 40 μg of the plasmids encoding fluorescent fusion proteins of Lifeact, Dyn-2(wt), and dnDyn-2(K44A), respectively, using an exponential decay protocol at a voltage set between 250 to 300 V and a capacitance set to 1,000 μF. After electroporation, cells were resuspended in RPMI 1640 medium. For nucleofection, 5 × 10⁵ cells were washed in phosphate-buffered saline (PBS) and resuspended in 100 μl cell Nucleofection solution V (Amaxa) in the presence of 4 μg plasmid DNA. Program O-17 of the Nucleofector 2b device (Amaxa) was used. Imaging analyses were carried out at 18 h postelectroporation, unless indicated differently.

Virus binding and endocytosis.

Particle concentrations in purified preparations of fluorescently labeled HIV derivatives were estimated on the basis of CA quantification using quantitative immunoblotting (LI-COR Odyssey). For virus binding, 10⁶ SupT1-R5 cells were incubated with particles corresponding to ∼100 ng CA for 3 h at 16°C in CO₂-independent medium (Gibco, Invitrogen) and fixed with 3% paraformaldehyde (PFA) for 30 min at 16°C, and cell nuclei were counterstained with Hoechst 33258 (Sigma-Aldrich). Cells were imaged by wide-field microscopy (Zeiss Axiovert 200M), and the number of particles bound per cell was determined in maximum projections of z-sections comprising the complete cell volume using a custom-made semiautomated ImageJ plug-in.

To determine HIV-1 endocytosis, cells were distributed in 8-chamber glass-bottom slides (Lab-Tek, Nunc) at a density of 3 × 10⁵ cells per well and immediately incubated with viral particles (corresponding to ∼1 μg CA) for the times indicated below at 37°C or 22°C. Subsequently, cells were fixed for 20 min with 4% PFA at room temperature. Fixed cells were stained with phalloidin conjugated to either Atto488 or Atto633 (Atto-Tec), Dy415 (Dyomics), or tetramethylrhodamine-isothiocyanate (Molecular Probes, Invitrogen), washed with PBS, and analyzed using a spinning-disk confocal microscope (SDCM; ERS-6; PerkinElmer). Whole cells were imaged with a slice spacing of 150 nm for deconvolution (Huygens, Scientific Volume Imaging) or 400 nm and were reconstructed in three dimensions using Volocity analysis software (Improvision, PerkinElmer). The interior of the cell was defined by the extension of the stained actin cortex. Individual particles were scored as being internalized in the cell when they were found within the three-dimensional limits of the actin cortex.

For colocalization studies, SupT1-R5 cells were incubated with eGFP.Vpr containing viral particles for 3 h at 37°C prior to fixation in 4% PFA for 20 min and subsequent permeabilization with 0.05% digitonin (Sigma-Aldrich). Staining was performed using Fab fragments of human anti-Env 2G12 or a polyclonal rabbit anti-CD4 serum (a kind gift from V. Bosch), Atto565-labeled mouse anti-human Fab, and Alexa 647-labeled goat anti-rabbit IgG.

To quantify clathrin-dependent endocytosis, 10⁶ SupT1-R5 cells were incubated for 7 min with Alexa 488- or Alexa 647-transferrin (Tfn; Invitrogen) at 37°C. Subsequently, surface-bound Tfn was removed by an acid wash (a wash for 1 min at pH 2.5 with 0.5 M NaCl, 0.2 N acetic acid), cells were fixed, and fluorescence signals were measured by flow cytometry.

For inhibition experiments with AMD-3100 (Sigma-Aldrich) or maraviroc (MVC; ViiV Healthcare), cells were preincubated for 30 min at 37°C with the respective compound at a concentration of 10 μM prior to addition of virus. For experiments with the fusion inhibitor T-1249 (76) (kindly provided by S. Urban, University of Heidelberg, Heidelberg, Germany), cells were preincubated for 30 min at 37°C at a concentration of 250 nM. T-20 (enfuvirtide; Roche) was used at a concentration of 50 μM and was added at the time points indicated below for the different experiments. For analysis of peptide uptake, C-terminally Atto488 labeled T-20 and Atto488-labeled Tat peptide (as a control) were produced and kindly supplied by C. Kaufmann and W. Mier (University Hospital Heidelberg, Heidelberg, Germany) and were used at a concentration of 50 μM. For experiments using soluble CD4 (sCD4; Gibco, Invitrogen), virus was preincubated with sCD4 at a concentration of 10 μg/ml for 15 min at 37°C prior to addition to cells.

Vpr.β-lactamase entry assay.

Virus cell entry was measured using a previously described β-lactamase HIV fusion assay (51). SupT1-R5, CEM-ss, or activated primary human CD4⁺ T cells (2 × 10⁵ to 3 × 10⁵) were incubated with the amounts of particles carrying the Vpr.β-lactamase fusion protein indicated below for the times indicated below at 22°C or 37°C in the presence or absence of T-20 using a V-bottom 96-well plate (Costar; Corning). Cells were washed with PBS and incubated in the dark at room temperature with 2 μM CCF2-AM (Invitrogen), prepared according to the manufacturer's instructions, in CO₂-independent medium (Gibco, Invitrogen) containing 20 μl/ml probenecid (Invitrogen). Cells were washed, fixed, and analyzed by flow cytometry, using a LSR-II or FACSVerse flow cytometer (Becton Dickinson). CCF2 fluorescence was excited at 405 nm, and emission maxima at 447 nm (CCF2blue) and 520 nm (CCF2green) were recorded using appropriate filter settings. Data were analyzed using FlowJo software (TreeStar).

Infectivity assays.

To determine HIV-1 infectivity, 2 × 10⁵ to 10⁶ SupT1-R5, CEM-ss, or activated primary human CD4⁺ T cells per well were distributed into V-bottom 96-well plates (Costar; Corning). Subsequently, cells were incubated with different amounts of HIV-1 variants for different periods of time at 37°C or 22°C. T-20 (50 μM) (52), AMD-3100 (2 μM) (53), or MVC (2 μM) (54) was added at the time points indicated below. If compounds were added prior to infection, preincubation with the compound was performed for 1 h. For temperature shift experiments, medium was removed after the times indicated below, fresh medium containing 50 μM T-20 was added, and incubation was continued for 48 h at 37°C. Subsequently, cells were fixed (4% PFA for 60 min at room temperature) and stained for intracellular CA using fluorescein isothiocyanate-conjugated anti HIV-1 CA antibody KC57 (Beckman Coulter) and Perm/Wash buffer (BD Bioscience). Analysis was performed by flow cytometry. For temperature shift experiments in the presence of inhibitors of endosomal acidification, cells were preincubated with 100 nM bafilomycin A1 (Sigma-Aldrich) or 10 mM ammonium chloride (Sigma-Aldrich) for 1 h prior to addition of virus, and further incubation was performed in the presence of the same drugs.

Ethics statement.

Buffy coats which otherwise would have been discarded were obtained from routine whole-blood preparations from healthy donors who signed an informed consent.

HIV-1 endocytosis is Env and CD4 dependent in SupT1-R5 cells.

To directly study HIV-1 binding and uptake by imaging, it is crucial to discriminate between specific, receptor-mediated events and nonspecific uptake. The T-cell line SupT1-R5 was chosen for use in this study because it is efficiently infected by both R5- and X4-tropic HIV-1 strains and displays low surface expression of HSPGs (55). Initial experiments established that uptake of labeled HIV-1 particles can be detected in SupT1-R5 cells by fluorescence microscopy. Cells were incubated with fluorescently labeled HIV-1 particles carrying MA.mCherry, MA.eGFP, or eGFP.Vpr. After fixation, the F-actin cytoskeleton was stained using fluorescent phalloidin conjugates. This allowed identification of cell boundaries as the dense layer of cortical F-actin in all three dimensions using a spinning disk confocal microscope (SDCM) (Fig. 1A to C). Viral particles appeared as diffraction-limited dots and could be visually classified as being bound to the cell surface or being internalized, i.e., being located within the boundaries of the actin cortex.

Next, we wanted to determine whether Env and the viral receptor and coreceptor are functionally required for HIV-1 attachment and endocytosis in SupT1-R5 cells. For this, we performed experiments using fluorescent HIV-1 particles carrying Env proteins with different coreceptor tropisms or carrying mutations in the receptor or coreceptor binding site, as well as particles lacking Env. Particle attachment to the cell surface was quantified following virus binding at 16°C to prevent endocytosis (56). Cells were imaged in a wide-field microscope, and the number of particles bound was determined in maximum projections covering the complete cell volume. Cell binding efficiency was significantly lower for viruses lacking HIV-1 Env than isogenic viruses carrying Env_NL4-3 (Fig. 2A). Impaired attachment was also observed for HIV-1 particles carrying Env with a deletion in the CD4-binding domain, while mutation of the coreceptor binding V3 loop in gp120 had no significant influence on virus cell binding (Fig. 2A). Thus, stable attachment of HIV-1 to SupT1-R5 cells requires the viral receptor CD4 with some contribution of nonspecific binding, while the coreceptor appears to have no role in this process.

To assess particle endocytosis, SupT1-R5 cells were incubated with fluorescent HIV-1 carrying different Env proteins, and fluorescent particles within the F-actin boundary were counted using a series of confocal optical sections (z-stacks) covering the whole cell volume. Most cells (∼75%) incubated with HIV-1 carrying either X4-tropic or R5-tropic Env displayed internalized particles (Fig. 2B and D), with an average of ∼2.5 particles per cell and no significant difference between the two Env variants being found. In contrast, incubation with particles lacking HIV-1 Env yielded internalized particles in only ∼10% of cells (Fig. 2B), and a single particle per cell was observed in positive cells. A significant difference between viruses carrying or lacking Env proteins was also observed for primary CD4⁺ T cells (Fig. 2B). The overall particle uptake was lower in this case, however, and the differences were less pronounced.

The potential role of receptor and coreceptor binding of HIV-1 Env for viral endocytosis was analyzed using the binding mutants described above as well as specific inhibitors. For compound treatment, virus or cells were preincubated with the respective substance (soluble CD4 [sCD4] for virus, AMD-3100 or MVC for cells), followed by infection for 3 h and SDCM to detect intracellular particles. Either preincubation with sCD4 or a deletion in the CD4 binding site of gp120 abolished specific HIV-1 endocytosis, reducing uptake to a level close to the background for an Env-negative [Env(−)] control (Fig. 2C). Since these results indicated CD4-dependent HIV-1 endocytosis in this cell line, we analyzed whether particles in endosomes were associated with CD4. Cells were incubated with fluorescent HIV-1 as described above and were subsequently fixed, permeabilized, and stained for Env and CD4. Immunofluorescence analysis clearly revealed fluorescent intracellular particles that stained positive for the Env glycoprotein and were associated with a CD4 signal (Fig. 1D to F). Thus, endocytosis of HIV-1 in SupT1-R5 cells depends on the CD4 receptor, and endocytosed virions appear to remain engaged with CD4.

The role of coreceptor binding was analyzed using fluorescent HIV-1 carrying an X4-tropic Env variant with a point mutation in the coreceptor binding site. Alternatively, coreceptor binding was inhibited by preincubation of cells with antagonists. Neither mutation of the coreceptor binding site nor treatment with the coreceptor antagonists caused a decrease in particle endocytosis (Fig. 2D). A slight increase of internalization was observed upon AMD-3100 treatment. Effective inhibition of coreceptor binding was validated using a β-lactamase reporter-based HIV-1 fusion assay (51), revealing a complete block of fusion for viruses carrying a mutation in the coreceptor binding site or with pharmacological inhibition of coreceptor binding (data not shown). Thus, coreceptor binding is dispensable for HIV-1 endocytosis but not for cytoplasmic entry in SupT1-R5 cells. As expected, blocking of membrane fusion by mutation of the fusion peptide or by treatment with the fusion inhibitor T-1249 also did not decrease HIV-1 endocytosis in these cells but led to a slight increase of endocytosed particles in the case of T-1249 treatment (Fig. 2E).

Dyn-2 is required for HIV-1 endocytosis but dispensable for cytoplasmic entry and infection.

Having established that endocytotic uptake of HIV-1 into SupT1-R5 cells is Env and CD4 dependent, we wanted to investigate its functional relevance. The most straightforward approach, previously employed to address related questions, would be to test the effect of pharmacological inhibition of clathrin-mediated endocytosis on HIV-1 cell entry and infection. Dynasore (57), pitstop-2 (22), and dyngo-4a (58) have all been described to be specific inhibitors of clathrin-mediated endocytosis. Control experiments measuring inhibition of Tfn endocytosis and cytotoxic effects in parallel indicated, however, that efficient inhibition of clathrin-mediated endocytosis occurred only at cytotoxic concentrations of these compounds in SupT1-R5 and primary CD4⁺ T cells (data not shown).

We therefore made use of a dominant negative (dn) mutant of Dyn-2 [dnDyn-2(K44A)] to block clathrin-mediated endocytosis. SupT1-R5 cells were transfected with plasmids carrying eGFP-tagged wild-type Dyn-2 [Dyn-2(wt)] or mutant Dyn-2, followed by incubation with fluorescent HIV-1 carrying wt Env_NL4-3. Tfn-Alexa 647 was added for the last 5 min of incubation to control for inhibition of endocytosis. Cells were fixed and stained for actin, and four-color z-stacks of whole cells were recorded by SDCM. Figure 3A displays individual z-slices for eGFP–Dyn-2(wt)-expressing cells (top) and eGFP–dnDyn-2(K44A)-expressing cells (bottom). The eGFP signal served to identify cells expressing Dyn-2(wt) or dnDyn-2(K44A). Cells displaying a disrupted F-actin cytoskeleton were excluded from the analysis to minimize confounding nonspecific cytotoxic effects. The intracellular Tfn-Alexa 647 signal allowed validation of the dnDyn-2(K44A) effect and stratification of dnDyn-2(K44A)-expressing cells into a population that showed complete inhibition of Tfn uptake [Tfn(−)] and a population that exhibited residual Tfn uptake [Tfn(+)]. Internalized virus particles (Fig. 3A, white arrowheads) were detected in both eGFP-positive cells and untransfected cells (asterisks) but were strongly reduced in dnDyn-2(K44A)-expressing cells. Quantitative analysis revealed that ∼75% of all Dyn-2(wt)-expressing cells carried internalized HIV-1 particles (Fig. 3B), similar to the situation for untransfected SupT1-R5 cells (Fig. 2B). In contrast, endocytosed HIV-1 was detected in only ∼25% of all dnDyn-2(K44A)-expressing cells. The degree of inhibition of HIV-1 endocytosis correlated with the efficiency of inhibition of transferrin uptake: ∼35% of dnDyn-2(K44A)-expressing cells with residual Tfn uptake exhibited intracellular HIV-1 particles, while only 12% of those classified to be negative for Tfn uptake showed intracellular HIV-1 (Fig. 3B). The latter proportion was similar to the unspecific endocytosis observed for Env(−) particles (represented by a dashed line in Fig. 3B; compare Fig. 3B and 2B).

To determine the functional relevance of the observed Env-, CD4-, and Dyn-2-dependent endocytosis for cytosolic entry of HIV-1 in SupT1-R5 cells, we employed an established entry assay detecting the activity of β-lactamase (BlaM) incorporated into HIV-1 particles as a Vpr.BlaM fusion protein (51). Cytosolic delivery of active BlaM upon virus-cell fusion can be detected by a shift in the emission maximum of a fluorogenic BlaM substrate following cleavage. Since eGFP fluorescence would interfere with this readout, mCherry-tagged fusions of Dyn-2(wt) and dnDyn-2(K44A) were used for these experiments. Inhibition of endocytosis was verified by incubation of mCherry–Dyn-2-expressing cells with Tfn-Alexa 488. Expression of mCherry–dnDyn-2(K44A) significantly reduced transferrin uptake to ∼8% of the control level, while expression of mCherry–Dyn-2(wt) had no effect (Fig. 4A). Cells expressing either mCherry–Dyn-2(wt) or mCherry–dnDyn-2(K44A) were incubated for 3 h with different amounts of HIV-1^BlaM reporter particles, followed by CCF2 staining and multicolor flow cytometry. Quantitative evaluation of three independent experiments performed with two different reporter virus preparations showed no significant effect of dnDyn-2(K44A)-mediated inhibition of endocytosis on the proportion of entry-positive cells (Fig. 4B). Similar entry rates were observed for the Dyn-2(wt)- and dnDyn-2(K44A)-positive populations, independent of the virus dose (from 2.5% to 80% BlaM-positive cells; Fig. 4B and data not shown). Thus, Dyn-2-dependent HIV-1 endocytosis is dispensable for HIV-1 fusion and cytosolic entry in SupT1-R5 cells.

To analyze whether expression of Dyn-2(wt) or dnDyn-2(K44A) affects productive HIV-1 infection, we infected SupT1-R5 cells previously transfected with the respective expression vector with HIV-1_NL4-3 (Fig. 4C). Infectivity was scored by flow cytometry, determining the proportion of CA-positive cells in the Dyn-2-mCherry-positive (and thus transfected) population. Cell viability was monitored in parallel by 7-amino-actinomycin D staining, which showed no significant difference between Dyn-2(wt)- and dnDyn-2(K44A)-expressing cells (data not shown). Quantitative analysis of three independent infection experiments, each performed with several different virus preparations, revealed no significant difference in HIV-1 infection between cells expressing dominant negative Dyn-2 and control cells (Fig. 4C). Similar results were observed for HIV-1 pseudotyped with the G glycoprotein of vesicular stomatitis virus (VSV-G) (Fig. 4D). The latter result appears to be counterintuitive, since pseudotyping with VSV-G is suggested to reroute viral entry to the endocytotic pathway. No effect of dnDyn-2(K44A) on infection by VSV-G-pseudotyped HIV-1 was also observed in our previous work (21), and other studies reported a marginal 2- to 3-fold reduction of infectivity of VSV-G pseudotypes by dnDyn-2(K44A) (59, 60). Accordingly, Dyn-2 appears to be largely dispensable for the entry of VSV-G-pseudotyped HIV-1.

Productive HIV-1 entry occurs primarily at the plasma membrane in SupT1-R5, CEM-ss, and primary CD4⁺ T cells.

The fact that Dyn-2-dependent endosomal uptake of HIV-1 could be blocked without any effect on cytoplasmic entry or infection suggested that endocytosis is dispensable for productive HIV-1 entry in SupT1-R5 cells. However, entry via Dyn-2-independent endocytosis was not excluded by our experiments, and the results observed for VSV-G pseudotypes also suggested alternative pathways. To directly determine the contribution of the plasma membrane versus endosomal fusion to HIV-1 infection in different cells, we made use of the differential temperature requirements of endocytosis and HIV-1 Env-mediated fusion. Previous studies have shown that HIV-1 fusion is virtually abolished at temperatures below 23°C (61, 62), while endocytosis is retained at temperatures above 16°C (63, 64). Incubation of HIV-1 particles with target cells for several hours at temperatures below 23°C leads to a so-called temperature-arrested state (TAS), where virions remain susceptible to inhibition by coreceptor antagonists and do not undergo membrane fusion (61, 62).

First, we investigated whether HIV-1 endocytosis occurred at 22°C. Incubating SupT1-R5 cells with fluorescent HIV-1 for 5 h at 22°C or 37°C revealed intracellular particles in both cases, but endocytosis was reduced 3-fold at the lower temperature (Fig. 5A and B). Similar results were obtained for Tfn, which showed a 2-fold reduction in uptake after incubation for 1 h at 22°C compared to the level of uptake after incubation at 37°C (data not shown). Analysis of the temperature dependence of HIV-1 fusion performed in parallel confirmed that incubation for 5 h at 22°C did not permit fusion, while efficient cell entry was observed at 37°C (Fig. 5C). Based on these observations, we developed an experimental approach (Fig. 5D) where SupT1-R5 cells were incubated with HIV-1 for 5 h at 22°C to allow endocytosis without fusion. During the last hour of incubation at 22°C, a high concentration of the non-cell-permeant (65) peptidic fusion inhibitor T-20 was added. Subsequently, the temperature was switched to 37°C. T-20 does not reach previously established endosomal compartments (66, 67) and can thus inhibit only the fusion of virus particles remaining surface accessible up to the time of T-20 addition. A lack of cell entry of T-20 was confirmed microscopically for a T-20 peptide labeled with Atto488 at its C terminus using an equally labeled HIV-1 Tat peptide as a control (data not shown). The proportion of HIV-1 entry and infection resistant to T-20 addition after prior incubation of cells and virus without T-20 at 22°C should thus reveal the contribution of endosomal uptake to HIV-1 infection if the T-20 concentration is sufficient to block all cell surface entry. In contrast, the small-molecule drugs AMD-3100 and MVC should also reach previously endocytosed viruses and should thus inhibit all HIV-1 entry events.

SupT1-R5 cells were incubated at 22°C with the X4-tropic HIV-1_NL4-3 or with an R5-tropic derivative, followed by addition of inhibitor and a temperature shift as depicted in the schematic in Fig. 5D. Medium was changed 5 h after the temperature shift, and T-20 was added to all samples at this time to prevent multiple rounds of infection. Cells were fixed and stained for CA at 48 h after the temperature shift. Approximately 8% of cells were infected by either the X4- or the R5-tropic HIV-1 variant under these conditions, and infection by both variants was completely abrogated when T-20 was added from the beginning (Fig. 5E and F). Importantly, the same negative result was obtained when T-20 was added 4 h after incubation at 22°C. No residual infectivity was detected, although significant endocytosis of fluorescent HIV-1 occurred at this temperature (Fig. 5B, E, and F). Incomplete inhibition was observed for the coreceptor antagonists when infection was with a virus targeting the respective coreceptor (Fig. 5E and F). This could be due to insufficient drug concentrations but is also consistent with the observation that a subset of viral particles in the TAS becomes resistant to coreceptor antagonists over time (62).

The full sensitivity of HIV-1 infection to T-20 inhibition after prior incubation of cells and virus at 22°C for 4 h in the absence of T-20 suggested that endocytosis may not contribute to productive fusion in this cell line. Alternatively, endocytosed HIV-1 particles may have been inactivated by the low endosomal pH over time, thus obscuring their potential for productive entry. To determine whether a lack of infection was due to exposure to a low pH, the experiment described above was repeated in the presence or absence of compounds preventing endosomal acidification. SupT1-R5 cells were infected with HIV-1 in the presence or absence of ammonium chloride or bafilomycin A1. While bafilomycin A1 reduced infection by a factor of ∼2, ammonium chloride did not affect productive infection at 37°C (Fig. 5G, left), consistent with the fact that HIV-1 entry is pH insensitive (13, 14). Incubation with HIV-1 at 22°C in the presence or absence of these compounds, followed by addition of T-20 and a temperature shift, also revealed no difference (Fig. 5F), indicating that acid exposure cannot explain the lack of infection of SupT1-R5 cells by previously endocytosed HIV-1.

To investigate whether the HIV-1 entry pathway depends on the T cell line used, we performed similar experiments in CEM-ss cells. This T cell line had been used in a previous study that suggested exclusive HIV-1 entry via endocytosis, with plasma membrane entry apparently blocked at the stage of hemifusion (23). Incubation of CEM-ss cells with HIV-1_NL4-3 for 4 h at 22°C, followed by addition of T-20 and a temperature shift (as depicted schematically in Fig. 5D), again led to an almost complete loss of infection (Fig. 5G, right), indicating that endocytosis is also dispensable for productive HIV-1 entry in CEM-ss cells. As was observed for SupT1-R5 cells, addition of ammonium chloride had no effect on HIV-1 infectivity and did not restore infection in CEM-ss cells treated with T-20 after preincubation with virus at 22°C (Fig. 5G, right).

The observation that HIV-1 infection was completely lost, while fluorescent particle endocytosis was reduced only 3-fold, indicated that endocytosis may not contribute significantly to HIV-1 infection. These results do not rule out a threshold effect, however, where a certain amount of endocytosis could be needed to score positive in the infection assay and the reduction of particle uptake would not linearly correlate with infection. To address this point, we extended the total period of incubation at 22°C from 5 h to 16 h. Endocytosis of fluorescent HIV-1 in SupT1-R5 cells was similar at 37°C and 22°C for this longer incubation time (Fig. 6A and B). Incubation for 16 h at 22°C led to a significant increase in HIV-1 fusion compared to that with 5 h of incubation (compare Fig. 5C and 6C), indicating that the TAS is not stable over time and may thus be better called a temperature-delayed state. Importantly, the amount of fusion at 22°C indicated in Fig. 6C constitutes the sum of all fusion events of particles that escaped TAS during the 16-h period before the temperature shift, irrespective of their subcellular localization. This number of fusion events thus provides a baseline for the events that were induced by the temperature shift in the following infection experiment.

The contribution of endocytosis to productive HIV-1 infection was determined by incubating SupT1-R5 cells (Fig. 6D, left) or CEM-ss cells (Fig. 6D, right) with different concentrations of HIV-1 for a total of 16 h at 22°C. T-20 was added after 15 h, and the temperature was shifted to 37°C after one additional hour at 22°C (see the schematic in Fig. 5D). Addition of T-20 from the start of the experiment completely abrogated infectivity in all cases, while addition of T-20 after 15 h of incubation at 22°C reduced infectivity to a residual level of <10% of that for the respective control (Fig. 6D and E). This phenotype was independent of the infectious dose. Infectivity was reduced only 2-fold or less upon addition of AMD-3100 after 15 h at 22°C, indicating that >50% of HIV-1 particles had already proceeded from TAS to a coreceptor-engaged state during this period (Fig. 6E). This result also suggested that HIV-1 infectivity was not severely affected by the extended preincubation. The infection events observed following addition of T-20 after a 15-h preincubation at 22°C and a subsequent temperature shift represent the sum of particle fusions that occurred during the low-temperature incubation period (fusion both from the plasma membrane and from the endosome) and of fusion events from previously endocytosed HIV-1 particles after the temperature shift. Infectivity after preincubation at the lower temperature, T-20 addition, and the subsequent temperature shift was reduced >10-fold (Fig. 6D and E), whereas fusion during the 16 h of incubation at 22°C was reduced only ∼5-fold (Fig. 6C). Thus, cytoplasmic entry during the period of preincubation could completely explain the residual infectivity.

While T-cell lines recapitulate many features of HIV-1 target cells, it is important to validate the findings in natural target cells of HIV-1. We thus performed similar experiments in primary human CD4⁺ T cells. Cells from three different donors were either infected with HIV-1_NL4-3, incubated with HIV-1^BlaM at 37°C in the presence or absence of T-20, or incubated with the virus for 4 h or 15 h at 22°C, followed by addition of T-20 and a temperature shift after one more hour at 22°C. Viral entry by membrane fusion was almost completely abolished when T-20 was added at the time of virus addition or 4 h after addition of virus and was reduced to <20% of the level for the control when particles were allowed to enter at 22°C for 15 h, followed by addition of T-20 (Fig. 7A and C). Similar observations were made for productive HIV-1 infection. Addition of T-20 at the time of infection abolished HIV-1 infectivity, and addition of T-20 after 4 h or 15 h of incubation at 22°C reduced HIV-1 infection by a factor of 30 (Fig. 7B and D). Notably, the reduction of infectivity was even more pronounced than the reduction of membrane fusion in this case. No difference in infectivity was observed when cells were infected in the presence or absence of ammonium chloride, excluding a detrimental effect of prolonged exposure to low pH of endosomally entering viruses (Fig. 7E).

Besides cell-type dependent differences, HIV-1 entry may also be influenced by the viral Env proteins, and we thus expanded our analyses beyond the commonly used laboratory-adapted strain HIV-1_NL43. To this end, we produced HIV-1 pseudotypes by cotransfection of an env-defective proviral plasmid with expression vectors carrying env genes directly amplified from the plasma of well-characterized HIV-1-infected patients. These primary env sequences give rise to highly T-cell-tropic viruses (requiring high CD4 surface levels on the target cell), circulate in patients at high levels, and are thought to be the predominant species mediating interindividual HIV transmission. Hence, they should better reflect the contribution of endocytosis to HIV infection that occurs under physiological rather than laboratory-adapted conditions. To determine a potential difference dependent on coreceptor usage, we performed these experiments with two well-characterized Env proteins: the X4-tropic Env₅₀₀₂ and the R5-tropic Env₄₀₅₁ (46). HIV-1^BlaM particles pseudotyped with either Env₅₀₀₂ or Env₄₀₅₁ were incubated with primary CD4⁺ T cells from two different donors for different periods of time in the presence or absence of T-20. Addition of T-20 from the beginning abrogated membrane fusion and infection in all cases (Fig. 7F to I). Membrane fusion and viral infectivity were also reduced to background levels for both Env pseudotypes when T-20 was added after virus and cells had been incubated for 4 h at 22°C (Fig. 7F and G). These results indicate that endocytosis is also dispensable for infection of primary CD4⁺ T cells with HIV-1 particles carrying primary T-cell-tropic Env proteins using either CXCR4 or CCR5 as the coreceptor. Membrane fusion was reduced only 3-fold compared to the level of fusion in the control incubation at 37°C when virus and cells were incubated for 15 h at 22°C (Fig. 7H), indicating that the TAS is much less tight for these primary envelopes. Importantly, however, addition of T-20 after incubation for 15 h at 22°C of primary CD4⁺ T cells with HIV-1 pseudotyped with either Env₅₀₀₂ or Env₄₀₅₁ almost completely abrogated infectivity (Fig. 7I), indicating that entry from endosomal compartments is dispensable for productive infection of HIV-1 carrying primary envelope proteins.