Research article

Visualization of retrovirus uptake and delivery into acidic endosomes

Kosuke Miyauchi, Mariana Marin, Gregory B. Melikyan


Diverse enveloped viruses enter cells by endocytosis and fusion with intracellular compartments. Recent evidence suggests that HIV also infects permissive cell lines by fusing with endosomes in a pH-independent manner. This finding highlights the importance of time-resolved monitoring of viral uptake. In the present study, we designed an imaging-based assay to measure endocytosis in real-time through probing the virus' accessibility to external solutions. Exposure of viruses bearing a pH-sensitive GFP (green fluorescent protein) variant on their surface to solutions of different acidity altered the fluorescence of surface-accessible particles, but not internalized viruses. By sequentially applying acidic and alkaline buffers with or without ammonium chloride, we were able to quantify the fractions of internalized and non-internalized virions, as well as the fraction of detached particles, over time. The exact time of single-virus internalization was assessed from the point when a particle ceased to respond to a perfusion with alternating acidic and alkaline buffers. We found that, surprisingly, HIV pseudoparticles entered acidic compartments shortly after internalization. These results suggest that the virus might be sorted to a quickly maturing pool of endocytic vesicles and thus be trafficked to fusion-permissive sites near the cell nucleus.

  • endocytosis
  • endosomal pH
  • pH-sensitive green fluorescent protein
  • single-particle imaging
  • virus detachment
  • virus fusion


Diverse viruses exploit endocytosis to enter host cells and initiate infection. The acidic environment in endosomes activates viral surface proteins resulting in virus–endosome fusion (enveloped viruses) or in disruption/permeabilization of an endosomal membrane (non-enveloped viruses) [1,2]. Viruses that do not rely on low pH to infect cells (many retro- and paramyxo-viruses) have long been thought to fuse directly with the plasma membrane. Accumulating evidence suggests, however, that these viruses can also enter and infect host cells via endocytosis [36]. We have recently reported that HIV-1 enters susceptible cell lines by receptor- and co-receptor-mediated endocytosis followed by pH-independent fusion with an endosomal membrane [7]. When HIV-1 did initiate fusion at the cell surface, the process was blocked at a step prior to the formation of a functional fusion pore, which is a prerequisite for the release of the nucleocapsid. The reliance of HIV-1 on endocytosis for entry is further supported by the demonstration that dynamin inhibitors suppress viral fusion and infection [7,8]. These findings highlight the importance of measuring HIV-1 uptake and assessing the relationship between endocytosis and productive fusion in different cell types.

Virus endocytosis has traditionally been measured in acute experiments involving the removal of surface-bound viruses by proteases and assessment of the remaining viral proteins (e.g. [911]). However, this approach is invasive and lacks the sensitivity to detect a small number of internalized virions. More recently, live-cell imaging has been implemented to directly monitor single-virus internalization and trafficking. Virus uptake has been deduced from the dispersal of fluorescently labelled clathrin-coat around the virus and/or from spikes in the virus-associated dynamin signal [1214]. An alternative approach to detect single-virus internalization has been introduced by Brandenburg et al. [15]. Polioviruses labelled with a pH-sensitive probe were judged to enter endosomal compartments when they became protected from changes in the extracellular pH. However, these measurements have not been implemented in a manner that would allow continuous monitoring of single-virus internalization, since extracellular solutions were manually replaced at a selected time, once per experiment.

In the present study, we designed an imaging-based protocol to measure endocytosis and delivery of HIV-1 pseudoviruses (designated HIVpp) into acidic compartments. To achieve this goal, the viral surface was labelled with a pH-sensitive GFP (green fluorescent protein) variant that was quenched at mildly acidic pH. Similar to the strategy used to detect the formation of endocytic vesicles [13], virus internalization was monitored by assessing the changes in mean fluorescence of multiple viral particles in buffers of varied acidity. These measurements showed that the virus uptake and detachment from cells occurred concomitantly and that the latter process was faster than endocytosis. The fluorescence-quenching assay also permitted time-resolved monitoring of single HIVpp endocytosis. We found that, shortly after being protected from changes in the external pH, HIVpp was delivered into acidic compartments. This result indicates that, similar to influenza virus [16], the internalized HIVpp is targeted to a quickly maturing population of endosomes. The methodology developed in the present work should be applicable to studies of uptake of both enveloped and non-enveloped viruses. Real-time imaging of single-virus endocytosis and subsequent fusion with intracellular compartments would help delineate the steps of productive entry.


Plasmids and virus production

To construct the vector expressing the ecliptic pHluorin–ICAM-1 (intercellular adhesion molecule 1) chimaera (designated EcpH-TM), the fragments of ecliptic pHluorin [17] and of the transmembrane domain of ICAM-1 were produced by PCR using the following primers: EcpH, 5′-TAAGCTTCTCGAGAGTAAAGGAGAAGAACTTTTCACTGG-3′ and 5′-TGAATTCTTGTATAGTTCATCCATGCCATGTG-3′; and ICAM-1 transmembrane domain, 5′-AGAATTCACGGTATGAGATTGTCATCATC-3′ and 5′-TGGATCCTCACCGCTGGCGGTTATAGAGGTA-3′. The pCDM8 ICAM-1 vector (Addgene) was used as a template for the transmembrane domain of ICAM-1, and the MLV (murine leukaemia virus) Gag-EcpH-expression vector [18], derived from the original ecliptic pHluorin construct, was used as a template for EcpH. PCR products of EcpH and ICAM-1 transmembrane fragment were cloned into the p3xFLAG CMV9 vector (Sigma) with HindIII and EcoRI, or EcoRI and BamHI sites respectively. To construct the HIV-1 Gag-mCherry expression vector, the Gag fragment of the p96ZM651gag-opt vector [National Institutes of Health ARRRP (AIDS Research and Reference Reagent Program)] was cloned into the pcDNA3.1zeo(+) plasmid (Invitrogen) using the HindIII and BamHI sites. The mCherry fragment was produced by PCR with pRSET-BmCherry (from Dr R. Tsien, University of California, San Diego, San Diego, CA, U.S.A.) as a template and the following primers: 5′-AGGATCCAAGGGCGAGGAGGATAACATGG-3′ and 5′-ACTCGAGTTACTTGTACAGCTCGTCCATGCCGCCGGTGGAGTGGC-3′. The fragment was cloned into the pcDNA3.1zeo(+) vector encoding HIV-1 Gag using the BamHI and XhoI sites.

Pseudoviruses were produced by transfecting HEK-293T cells [HEK (human embryonic kidney)-293 cells expressing the large T-antigen of SV40 (simian virus 40); A.T.C.C.] with plasmids encoding HIV R8ΔEnv (from Dr D. Trono, University of Geneva, Geneva, Switzerland), Gag-mCherry, EcpH-TM, and either HXB2 Env or JRFL Env, using the calcium phosphate protocol, as described previously [7]. We found that HIV Gag-mCherry was not cleaved by viral proteases and was not released from the viral particles permeabilized by a brief treatment with saponin (results not shown). Therefore, even if a small fraction of virions do fuse with cells, these events should not result in the release in disappearance of the viral core marker.

Virus–cell fusion assay

Virus fusion with TZM-bl cells was quantified using the β-lactamase assay, as described previously [7]. Briefly, viruses were added to cells [MOI (multiplicity of infection) 0.7–1] and centrifuged at 2095 g at 4 °C for 30 min. Cells were washed to remove unbound viruses, and virus uptake and fusion were initiated by raising the temperature to 37 °C. Fusion was stopped at the times indicated by adding 1 μM C52L peptide derived from the C-terminal heptad repeat domain of gp41 [19]. At the end of incubation, cells were loaded with the fluorescent β-lactamase substrate CCF2-AM (CCF2-acetoxymethyl ester; Invitrogen). The intracellular β-lactamase activity (ratio of blue to green fluorescence) was measured after an overnight incubation at 14 °C.

Protease protection assay for virus endocytosis

Virus uptake by TZM-bl cells was determined essentially as described in [20]. Briefly, unlabelled pseudoviruses bearing HXB2 or JRFL Env were collected and concentrated by centrifugation (24000 rev./min for 2 h at 4 °C; rotor type SW41, Beckman) on to a 20% sucrose cushion. Virus binding to target TZM-bl cells (MOI 0.7) was aided by centrifugation at 2095 g (for 30 min at 4 °C). Following the shift to 37 °C, external viruses were stripped by treating cells with 2 mg/ml pronase (10 min on ice), cells were then washed with DMEM (Dulbecco's modified Eagle's medium) with 10% serum and lysed. The amount of p24 in the lysate was determined using a p24 ELISA Kit (PerkinElmer Life Sciences, catalogue number NEK050B001KT). The protease-resistant p24 signal at a given time point was normalized to the total amount of cell-bound p24 before raising the temperature.

HIV-1 immunostaining and immunoprecipitation

Double-labelled viruses (1×104 i.u.) containing EcpH-TM, Gag-mCherry and HXB2 Env were allowed to adhere to poly-L-lysine coverglass slides (Lab-Tek) for 30 min at 4 °C and were then blocked with DMEM containing 10% serum. All subsequent incubation steps were performed at 4 °C. Viruses were first incubated for 2 h with 0.4 μg/200 μl per well of the IgG1b12 anti-gp120 human monoclonal antibody (from ARRRP, donated by Dr Dennis Burton and Dr Carlos Barbas) or with an isotype-matched human IgG antibody (Sigma) diluted in DMEM/10% serum. Samples were washed and incubated for 1 h with 2 μg/ml Cy5 (indodicarbocyanine)-conjungated goat anti-human antibody (KPL), washed and imaged. Virus immunoprecipitation was carried out by incubating ~1×106 i.u. of double-labelled pseudoviruses overnight at 4 °C on a tube rotator with 10 μg of anti-FLAG M2 (Sigma) or mouse IgG isotype-control antibodies covalently attached to Protein G Plus/Protein A-Agarose beads (Calbiochem). Beads were centrifuged (4000 g for 5 min at 23 °C), and supernatants containing unbound particles were removed and titrated on TZM-bl cells, as described in [7].

Calculation of virus detachment and endocytosis

As described in the text, we used the following equations to evaluate the fraction of virions residing on the cell surface (Vs), in neutral endosomes (Vne), in acidic endosomes (Vae), as well as those detached from target cells (Vd). The initial number of viruses attached to cells at time (t)=0 prior to raising the temperature was calculated as (eqn 1): Embedded Image where F5.4(t), F8.8(t) and F8.8*(t) are the EcpH signals from cell-associated virions in respective buffers. The fraction of cell-surface-bound viruses was obtained from (eqn 2): Embedded Image

The fraction of virions residing in neutral and acidic compartments at any given time is (eqn 3 and eqn 4): Embedded Image Embedded Image

The fractions of detached virions and the total fraction of internalized particles (Ve=Vne+Vae) can be written as (eqn 5 and eqn 6): Embedded Image Embedded Image

Since we found that the number of viruses residing in non-acidic compartments (Vne) was negligibly small, eqn 5 and eqn 6 can be simplified (eqn 7 and eqn 8): Embedded Image Embedded Image

Single-virus imaging

Double-labelled pseudoviruses containing EcpH-TM and Gag-mCherry were pre-bound to TZM-bl cells grown on a coverslip by spinoculating at 2095 g for 40 min at 4 °C. Owing to the better adherence of HXB2 Env-pseudoviruses to cells compared with JRFL Env-pseudoviruses, coverslips were inoculated with 6×104 i.u. of HXB2 or with 4×105 i.u. of JRFL virus. This protocol allowed each cell to bind between 12 and 35 double-labelled particles. Coverslips with cells were then transferred to the imaging chamber mounted on the microscope stage. Virus endocytosis (and fusion) was triggered by quickly shifting cells to 37 °C. Solutions of different acidity were locally applied through a narrow tip positioned in close proximity to the imaged cells, using the image acquisition software-controlled four-channel miniature perfusion system (Bioscience Tools). The following three isotonic buffers were sequentially applied to cells for 30 s: pH 5.4 Mes, pH 8.8 Taps and pH 8.8 Taps supplemented with 30 mM ammonium chloride (referred to as the pH 8.8* buffer). To avoid applying excessive shear force, perfusion was carried out at a slow rate of ~0.2–0.3 ml/min.

The image acquisition, local perfusion and autofocus function to compensate for the focal drift were implemented using the MultiTime macros available through the Zeiss LSM510 software. Cells and viruses were imaged using the C-Apo 40×/1.2 NA (numerical aperture) water-immersion objective. Three Z-sections (bottom, middle and top) separated by 2.5 μm were acquired per time point with the pinhole set to 340 μm in order to collect some out-of-focus light and thereby reduce the number of required Z-stacks. Image analysis was performed using the Volocity software (Improvision/PerkinElmer). For every image frame, three-dimensional objects positive for both EcpH and mCherry markers were identified and their total fluorescence was calculated as a function of time. Single-particle tracking was performed using the red (mCherry) channel, as described previously [7].


Production and characterization of pH-reporter virions

In order to continuously monitor virus uptake, the pH-sensitive GFP variant {referred to as ecliptic pHluorin (EcpH) [17]} was incorporated into the viral membrane by appending EcpH to the N-terminus of the transmembrane domain of ICAM-1 (Figure 1A, EcpH-TM), an intercellular adhesion molecule known to selectively incorporate into HIV particles [21,22]. Double-labelled pseudoviruses were obtained by co-transfecting HEK-293T cells with plasmids encoding EcpH-TM, HIV Gag-mCherry and HIV Env (along with the HIVΔEnv backbone vector). Approx. 30–40% of Gag-mCherry-labelled virions obtained by the above protocol contained detectable amounts of EcpH-TM in their membranes (Figure 1B). To eliminate the contribution to the fluorescence signal from the EcpH-TM-labelled particles (probably membrane vesicles) lacking the Gag-mCherry, we analysed only virions positive for both markers (see below). The choice of mCherry as a marker for the viral core was dictated by the relatively low sensitivity to a drop in pH (pKa<4.5 [23]), which should permit visualization of virions in acidic compartments where the EcpH fluorescence is fully quenched (Figure 1B). As expected, the EcpH- and mCherry-based markers attached to distant proteins did not exhibit FRET (fluorescence resonance energy transfer) (Supplementary Figure S1 at

Figure 1 Construction and characterization of pH-reporter pseudoviruses

(A) A diagram of EcpH attached to the transmembrane domain of ICAM-1 (designated EcpH-TM). (B) Diagrams and images of pseudoviruses labelled with EcpH-TM and the HIV-1 Gag-mCherry at alkaline and acidic pH. (C) Double-labelled HXB2 Env-pseudotyped viruses can be quantitatively pulled down with the anti-FLAG M2 antibody, but not with an isotype control antibody. (D) A representative image of HXB2 Env-carrying pseudoviruses co-labelled with Gag-mCherry (red), EcpH-TM (green) and immunostained for Env, using the IgG1b12 antibody (blue). Particles positive for all three markers appear white. (E) A contour plot of Gag-mCherry, EcpH-TM and Env co-localization obtained by measuring the total intensities of red, green and blue signals respectively, for 1027 individual particles. A ‘hot spot’ of high Env-expressing particles is observed for viruses with larger amounts of Gag-mCherry and moderate amounts of EcpH-TM. (F) Env co-localization with Gag-mCherry+ particles (left-hand bar) and with Gag-mCherry+/EcpH-TM+ particles (right-hand bar). (G) Incorporation of EcpH-TM does not compete with incorporation of Env, as shown by a weak positive correlation between the EcpH and Env signals from Gag-mCherry+ particles.

In order to assess the presence of Env in double-labelled virions, coverslip-immobilized particles were immunostained with the IgG1b12 anti-gp120 antibody and analysed by fluorescence microscopy (Figures 1D and 1E). Nearly 55% of Gag-mCherry+/EcpH-TM+ particles exhibited detectable Env staining (Figure 1F). This relatively low co-localization of fluorescent markers and Env was more apparent than real due to the detection limit of fluorescence microscopy. In reality, EcpH-TM was incorporated into nearly all particles containing both Gag and Env, since >99% of infectious viruses could be precipitated by the M2 anti-FLAG antibody (Figure 1C). Because EcpH and Env signals tended to correlate (Figure 1G), incorporation of this membrane marker into virions did not appear to displace Env. These findings imply that monitoring delivery of double-labelled particles into acidic endosomes should faithfully reflect virus internalization.

Rationale for measuring virus uptake and dissociation

The high pKa value (~7.0 [17]) of EcpH makes it useful for assessing early stages of virus uptake and endosomal acidification since its fluorescence is virtually fully quenched at pH 6.2 (Figure 1B). Thus the drop in the EcpH fluorescence should, in principle, reflect the time-course of virus delivery into mildly acidic compartments. These measurements would not directly report the time of HIV envelopment by an endocytic vesicle if there is a significant lag before the interior of virus-carrying endosomes becomes acidic. Another confounding factor is that, in agreement with [7,24], a large fraction of virions detach from cells during an imaging experiment. Virus detachment was manifested in the significant drop of the relatively pH-independent Gag-mCherry signal over time at 37 °C (see Figure 3A), which was not due to sample photobleaching (Supplementary Figure 2A at

Our strategy for monitoring HIVpp detachment, uptake and delivery into acidic compartments was as follows. The initial number of virions adhered to cells at the beginning of incubation (Vini) is redistributed into four populations of virions upon incubation at physiological temperature: surface-bound (Vs), detached (Vd), viruses in neutral endosomes (Vne) and those in acidic endosomes (Vae). Detached viruses no longer contribute to the fluorescence signal. To determine the fractions of virions in those populations, cells were exposed to the following three buffers: pH 5.4 Mes, pH 8.8 Taps, and pH 8.8 Taps supplemented with 30 mM ammonium chloride (referred to as the pH 8.8* buffer). The diagram in Figure 2(A) illustrates that only particles residing in neutral endosomes (Vne) should contribute to the EcpH signal in a membrane-impermeant acidic buffer. Thus virus accumulation in neutral endosomes would increase the EcpH fluorescence at external pH 5.4 (eqn 3 in the Experimental section). Raising the external pH to 8.8 is expected to maximize the EcpH fluorescence of surface-bound viruses, but not those residing in acidic compartments, thus yielding Vs+Vne. The inclusion of ammonium chloride in the pH 8.8* buffer would raise endosomal pH and reveal the signals from all cell-associated virions, Vs+Vne+Vae. Therefore the signal difference in a pH 8.8 buffer with and without ammonium chloride is proportional to the number of particles in acidic compartments.

Figure 2 Rationale for HIVpp endocytosis measurements and snapshots of viruses on cells subjected to the triple-perfusion protocol

(A) Expected changes in the EcpH-TM fluorescence of virion entering target cells upon exposure to solutions of different acidity. For simplicity, the Gag-mCherry fluorescence of double-labelled viruses is not shown. Viruses residing on the cell surface (Vs), in neutral and acidic endosomes (Vne and Vae), as well as detached viruses (Vd), are illustrated. Viruses in acidic endosomes are coloured grey, whereas those in a neutral environment are green. An alkaline buffer supplemented with ammonium chloride raises the endosomal pH and reveals the EcpH signal from all three populations of virions (right-hand panel). (B) Representative images of TZM-bl cells with bound EcpH-TM/Gag-mCherry labelled viruses were acquired during successive perfusion with pH 5.4, pH 8.8 and pH 8.8* (pH 8.8+NH4Cl) buffers prior to (upper panel) and 43 min after (lower panel) shifting to 37 °C. Arrows mark surface-accessible virions that respond to changes in extracellular pH, whereas arrowheads indicate internalized particles that respond only to perfusion with the pH 8.8* buffer that raises the endosomal pH. Scale bar=10 μm.

Prior to shifting cells to 37 °C, all viruses are accessible to external solutions, hence, the initial number of cell-bound particles is proportional to the difference between the EcpH fluorescence in alkaline and acidic buffers at time t=0 (ΔF0, see eqn 1). By normalizing the difference in the EcpH signal in buffers of different acidity at any given time to ΔF0, one can obtain the fraction of cell-surface-bound, detached and internalized virions residing in different compartments (eqns 26). Although we did not directly measure endosomal pH in the pH 8.8* buffer, this buffer raised the cytosolic pH (Supplementary Figure S3 at and increased the EcpH fluorescence of internalized viruses (see Figures 2, 3 and 6). These results support the notion that the pH 8.8* buffer dissipates the pH gradient across both the plasma and endosomal membranes, thus raising the endosomal pH above neutrality. In our calculations, the differences in the Vne fluorescence in membrane-impermeant buffers and in the pH 8.8* buffer are ignored. As we will show in the next section, this omission is inconsequential.

Figure 3 Analysis of pH-dependent changes in fluorescence of cell-associated HIVpp

(A) EcpH-TM/Gag-mCherry labelled virions on TZM-bl cells were sequentially perfused with pH 5.4, pH 8.8 and pH 8.8* buffers. The first perfusion cycle was carried out at room temperature, after which time the cells were shifted to 37 °C. The duration of each perfusion session is marked by grey boxes. The pH 5.4 perfusion was initiated ~15 s before the image acquisition. The ○ and ● symbols show the changes in the EcpH and mCherry signals respectively. Each datum point is the mean fluorescence of a three-dimensional region of interest after subtracting the background signal. The grey broken lines trace the changes in fluorescence signals in pH 8.8 and pH 8.8* buffers. (B) The magnified view of one perfusion session. (C) Control imaging experiment performed at 26 °C. Fluorescence measurements were carried out as described in (A). a.u., arbitrary units.

HIVpp endocytosis is restricted by the kinetically dominant detachment process

The amount of HIVpp input was adjusted so that a few hundred double-labelled particles were initially attached to TZM-bl cells in the image field. Virus uptake was triggered by quickly raising the temperature to 37 °C. The virus dynamic on the cell surface was monitored by successive applications of pH 5.4, pH 8.8 and pH 8.8* buffers to cells in the image field. The local perfusion setup was synchronized with the image acquisition and was carried out every 8.5 min in brief sessions lasting approx. 1.5 min (30 s for each buffer). Images were acquired during, but not between, the perfusion sessions, and only three Z-sections were obtained for every time point. This protocol was designed to reduce adverse effects of shear force and non-physiological pH and to minimize sample photobleaching. A brief exposure to low pH minimizes the cytosol acidification ([13] and Supplementary Figure S3C), which could inhibit endocytosis. We also found that HIVpp–cell fusion was not affected by pre-treatment with acidic buffers (Supplementary Figure S2B).

A typical imaging experiment is illustrated in Figures 2 and 3. At the beginning of incubation at 37 °C, the EcpH on all particles responded to the pH changes by switching between fluorescent and non-fluorescent states. Towards the end of the experiment, however, fewer red puncta were observed, and only a fraction of virions exhibited green fluorescence upon raising the external pH (Figure 2B, arrows compared with arrowheads). The dynamics of HIVpp detachment and entry were quantified by plotting the changes in mean EcpH and mCherry signals from double-labelled particles in solutions of different acidity (Figure 3). The application of the pH 5.4 buffer throughout the experiment abrogated the EcpH fluorescence, whereas the pH 8.8* buffer maximized the cell-associated signal (Figures 2 and 3A). Both EcpH and mCherry signals dropped over time, even in the pH 8.8* buffer (Figure 3A), demonstrating that a large fraction of viruses detached from target cells. The lack of detectable EcpH fluorescence at acidic pH implies that, at any given time, the fraction of viruses residing in neutral endosomes (Vne) was negligible. Indeed, if viruses were to accumulate in neutral endosomes to a significant extent, the EcpH fluorescence at an external acidic pH would have increased relative to the initial level (Figure 2A). We therefore used simplified eqns (7) and (8) to calculate Vd and Ve.

As expected, EcpH fluorescence was identical in pH 8.8 and pH 8.8* buffers before and immediately after raising the temperature. Over time, the difference between the EcpH signals in pH 8.8 and pH 8.8* buffers became more apparent (Figures 3A and 3B), revealing the increased fraction of endosome-resident virions (Ve). To verify that the differences in the EcpH fluorescence with and without ammonium chloride were due to the virus uptake into acidic compartments, imaging experiments were performed at room temperature (25–27 °C). As expected for conditions that slow down endocytosis, the pH 8.8* signal was only slightly greater than the pH 8.8 signal by the end of the experiment (Figure 3C). However, judging by the drop in the EcpH and mCherry signals over time at alkaline pH (Figure 3C and results not shown), viruses still effectively detached from cells at reduced temperature.

To obtain the plots of surface-associated (Vs), endocytosed (Ve) and detached (Vd) viruses against time of incubation, the EcpH-quenching data shown in Figure 3 were analysed using eqns (2), (7) and (8). This analysis showed that the majority of virions detached from cells, whereas a smaller fraction was internalized (Figures 4A and 4B). In agreement with [7,24], HIV-1 dissociation was virtually completed within ~40 min at 37 °C. By that time, most of the cell-associated viruses were internalized with only a minor fraction remaining at the cell surface. The kinetics of virus detachment based on the decay in the overall mCherry signal was similar to that obtained from the EcpH-quenching data (results not shown). We found that the virus uptake was marginal at room temperature, whereas its detachment was comparable with that at 37 °C (Figure 4A, stars).

Figure 4 Calculated fractions of cell surface-adhered (Vs), endocytosed (Ve=Vne+Vae) and dissociated (Vd) viruses as a function of time at 37 °C

(A) Changes in the fractions of cell-adhered (△), internalized (●) and dissociated (○) HXB2 Env-bearing virions were calculated as the percentage of the initially pre-bound particles from the representative experiment shown in Figure 3, using eqns (2), (7) and (8). Virus detachment and endocytosis at room temperature (RT) are shown by open and closed stars respectively. (B) Same as in (A), but for JRFL Env-bearing virions. (C and D) Kinetics of HXB2 (C) and JRFL (D) dissociation (○) and internalization (●) obtained by averaging the data from three to five independent experiments. Solid lines are single exponential fits to data points. The rate of endocytosis based on single-particle tracking is shown by Δ in (C) after normalizing the final extent of virus uptake to that determined by multi-particle analysis. Also shown are the rate of EcpH quenching upon HXB2 endocytosis in the absence of perfusion (grey noisy line) and the sum of endocytosis and detachment (Ve+Vd, ■). (E) JRFL and HXB2 pseudovirus binding to TZM-bl cells expressed as a fraction of HIV-1 p24 input that remained associated with cells after virus spinoculation at 4 °C and washing (n≥6). There was a statistically significant difference in pseudovirus binding (P<0.043). Error bars are S.E.M.

Interestingly, JRFL Env-pseudotyped particles appeared more prone to dissociate from cells and less capable of entering the cells than HXB2 Env pseudotypes. To verify whether this trend was real, we assessed the average values of virus shedding and endocytosis from several independent experiments (Figures 4C and 4D). These results supported the initial observation that JRFL was more readily lost from the cell surface than HXB2. As determined by fitting the data with an exponential function (Figures 4C and 4D, solid lines), JRFL and HXB2 uptake was, respectively, 4-fold and 2.5-fold slower than their dissociation from cells. As a result, only 13% of JRFL and 21% of HXB2 virions were endocytosed by the end of the experiment. HXB2 uptake tended to exceed that of JRFL, but these differences were not statistically significant (P>0.29) due to a limited sample size. Inverse correlation between HIVpp uptake and dissociation suggests that the kinetic competition between these processes determines the efficacy of virus entry and fusion with endosomes.

At 4 °C, HIV-1 associates with HeLa-derived target cells in a CD4-independent manner [9,20,25]. We therefore asked whether the virus' propensity to associate with cells in a CD4-independent manner could affect its detachment at elevated temperature. Both single-virus imaging (see the Experimental section) and the p24 assay (Figure 4E) confirmed that HXB2 Env-pseudotyped particles attached to TZM-bl cells in the cold more efficiently than JRFL virions. This suggests that, in agreement with the inverse correlation between HIVpp dissociation and endocytosis (Figures 4C and 4D), a tighter CD4-independent attachment to cells in the cold can slow down virus dissociation and thus favour its uptake.

Bulk uptake and productive endocytosis of HIVpp occur at the same rate

Our perfusion protocol was designed to minimize possible adverse effects of non-physiological pH [26,27] and of the sheer force produced by a local perfusion. To verify that these factors did not significantly affect virus endocytosis, we compared the above results with the kinetics of EcpH quenching at neutral pH in the absence of perfusion. Under these conditions, the reduction in EcpH signals occurred as a result of both virus detachment and delivery into acidic compartments. Since HIVpp does not appear to accumulate in neutral compartments (Figure 3), the loss of EcpH signal should be proportional to Vd+Ve. The measured kinetics of the EcpH fluorescence decay in the absence of perfusion (Figure 4C, noisy grey line) and the combined fraction of endocytosed and detached viruses obtained from the perfusion experiments (black squares) were statistically indistinguishable (P>0.43). These results imply that the perfusion protocol itself did not significantly perturb HIVpp uptake.

Next, we compared the rate of virus uptake obtained by the perfusion assay with the rate of the HIV-1 p24 accumulation in TZM-bl cells. The extent of virus internalization was determined from the fraction of protease-resistant p24, as described in [20]. For a given virus (JRFL or HXB2), the rates of EcpH protection from low pH and from a protease were close (Figure 5, P>0.40). However, the fraction of the HIV-1 p24 resistant to proteolysis at any given time was greater than the fraction protected from an acidic environment. These differences are likely due to the limited access of proteases to partially internalized virions (e.g. those in clathrin-coated pits), which are still accessible to small molecules and ions [28]. To conclude, our imaging assay is suitable for the time-resolved measurements of the virus dynamics on the cell surface. This assay appears to more accurately report the point of virus sequestration within endosomal compartments compared with traditional biochemical assays employed to monitor the virus uptake.

Figure 5 The rates of virus endocytosis measured by different techniques

(A) Pseudovirus uptake by TZM-bl cells determined based on the EcpH-quenching assay and based on the p24 assay. The HIV-1 p24 internalization assay measured the fraction of viruses that remain associated with cells following the protease treatment. Endocytosis of JRFL and HXB2 cells is shown by open and filled symbols respectively. Virus uptake based on the EcpH-quenching data (re-plotted from Figures 4C and 4D) is shown by circles, whereas the time course of p24 internalization is shown by triangles. (B) Comparison of the HXB2 uptake (EcpH-quenching data, ●) and the virus escape from the inhibitory peptide C52L (◇), as measured by the β-lactamase assay. Error bars are S.E.M. for at least three independent experiments.

The EcpH-quenching assay reports the bulk virus uptake that may or may not lead to productive entry. We therefore sought to compare the above results with the functional measurements of productive HIVpp entry. The latter assay is based on the measurement of the time course of the virus escape from the C52L peptide [19], which competitively inhibits fusogenic conformational changes of Env. A fully inhibitory concentration of C52L was added at varied times of virus–cell incubation at 37 °C, and the resulting HIV–cell fusion was measured using the β-lactamase assay [29]. Because HIVpp enters TZM-bl cells by endocytosis followed by fusion with endosomes [7], protection from the membrane-impermeant peptide should occur as a result of productive endocytosis, which later culminates in virus fusion. The kinetics of HIVpp uptake (EcpH quenching) and of escape from C52L were indistinguishable (Figure 5B, P>0.90). The finding that bulk virus uptake and its productive entry occur at similar rates supports the validity of the EcpH-quenching assay.

Internalized HIVpp quickly enters mildly acidic compartments

Our initial analysis was concerned with the average fraction of internalized HIVpp obtained by integrating the pH-dependent EcpH signal from multiple particles. Next, we examined the exact time of single-virus uptake and delivery into acidic endosomes. To improve the time resolution for detecting the virus uptake, extracellular pH was shifted between 5.4 and 8.8 every 24 s (Figure 6A). Also, images (three Z-stacks) were acquired every 7–8 s instead of every 8.5 min (compare Figures 3 and 6). These experiments were carried out with HXB2 pseudoviruses that were more likely to become internalized than JRFL particles (Figure 4).

Figure 6 Imaging single HIVpp endocytosis

(A) Perfusion protocol and particle internalization at alkaline and acidic pH. AE, acidic endosomes; CCP, clathrin-coated pits; NE, neutral endosomes; PM, plasma membrane. EcpH-labelled viruses residing in alkaline and acidic compartments are shown in green and grey respectively. (B and C) Representative images and single-particle tracking for an HXB2 Env-pseudotyped virion internalized by a TZM-bl cell at alkaline pH (see also Supplementary Movie S1 at Internalization of pre-bound viruses was monitored by continuously perfusing cells with alternating pH 5.4 and pH 8.8 buffers, each applied for 24 s (grey and white areas on the graphs respectively). The external pH and the time after raising the temperature are shown at the top of images. The graph illustrates the mCherry (dark red) and EcpH (dark green triangles) signals from the selected particle (arrow) as a function of time. The pH 5.4 perfusion session applied after ~14 min (second image panel) did not initially quench the EcpH fluorescence, which dropped to the background level shortly thereafter (red block arrow) and did not recover upon subsequent exposure to pH 8.8 (third image panel). A reference particle (arrowhead), which remained on the cell surface throughout the experiment, went through repetitive EcpH quenching–dequenching cycles (light green trace), whereas its mCherry fluorescence remained steady (light red). (D and E) Single-particle tracking and representative images of a virus (arrow) endocytosed during the low pH cycle (see also Supplementary Movie S3 at The EcpH signal that was lost at low pH (second image panel) did not reappear after shifting to a basic solution (third image panel). The arrowhead marks a surface-accessible virus. The blue numbers above the traces in (C) and (D) correspond to sequential images in (B) and (E) respectively. Perfusion with the pH 8.8* buffer recovered the EcpH fluorescence of the endocytosed particle (last image panels in B and E). The velocity of the particle is shown by a broken line.

All cell-bound particles initially exhibited repetitive EcpH quenching–dequenching cycles (Figure 6 and Supplementary Movies S1–S3 at At later times, double-labelled virions permanently lost the EcpH fluorescence irrespective of the external pH. These events were often associated with particle displacement towards the cell nucleus and thus marked the virus uptake and delivery into acidic compartments. Since the EcpH signal vanishes at pH 6.2 (Figure 1B), virions internalized during the alkaline perfusion cycle must experience a pH drop of more than 2 units (from 8.8 to ~6.2) in order to lose their green signal. We deduced the time of full virus engulfment by an endosomal membrane from the point when the EcpH fluorescence ceased to respond to pH changes (Figure 6A). Ranking the internalization times of 26 randomly chosen particles yielded the rate of endocytosis, which was not significantly different from that of the bulk virus uptake determined by ensemble analysis of EcpH quenching (P>0.05, Figure 4C, open triangles compared with black circles). The fact that raising the endosomal pH with the pH 8.8* buffer at the end of the experiment fully recovered the EcpH fluorescence (Figures 6B and 6E, right-hand panels) confirmed that the loss of this signal was due to the virus delivery into acidic endosomes.

Early endosomes undergo maturation whereupon their interior becomes increasingly acidic [3032]. Viruses internalized during the alkaline perfusion cycle allowed the measurement of the time between uptake and delivery into mildly acidic compartments. If these virions spend considerable time in early non-acidic endosomes, they should appear as EcpH-positive particles during the next acid perfusion cycle(s) (Figure 6A). Surprisingly, the external pH-independent green fluorescence of virions that entered at alkaline pH was short-lived. The representative virus marked by an arrow in Figure 6B (see also Supplementary Movie S1) was sealed off from the external milieu at alkaline pH after ~14 min of incubation at 37 °C, as shown by the brief retention of green fluorescence during the next alkaline pH cycle (Figure 6C). The loss of the EcpH signal during the next alkaline pH cycle (red block arrow) demonstrates that the virus-carrying endosome became acidic shortly after budding from the plasma membrane. Even assuming that the particle was internalized at the onset of the preceding alkaline cycle, the pH around the virus must have dropped in less than 48 s (the duration of one full pH 8.8/5.4 perfusion cycle). In contrast, surface-associated virions (marked by arrowheads in Figures 6B and 6E) continued to change their fluorescence in solutions of different acidity, thereby reporting the exact times of the local pH shift (Figure 6C, light green trace).

We found that 36 out of the 80 analysed virions entered endosomes during the alkaline perfusion cycle. These particles were rendered non-fluorescent within one or two perfusion cycles (for another example of a relatively long-lived endosomal EcpH signal, see Supplementary Figure S4 at and Supplementary Movie S2). Seven of these virions quickly lost their EcpH fluorescence while still at alkaline pH (results not shown) and were thus engulfed by endosomes and entered acidic compartments in a time shorter than a single perfusion cycle (21–24 s). The early acidification of an endosomal lumen did not correlate with significant particle movement (Figures 6C and 6D, broken lines). The 44 virions that entered during acid perfusion did not exhibit intracellular EcpH fluorescence (Figure 6A and [13]) and therefore did not permit the determination of the time required to deliver the virus into acidic compartments. An example of a virus entering at acidic external pH is shown in Figures 6(D) and 6(E) (see also Supplementary Movie S3).

In order to determine the virus' residence time in neutral endosomes, we measured the interval from the time of retention of EcpH fluorescence at acidic pH and the subsequent loss of green signal (designated TEcpH; Figure 7A). The exact time of the pH drop around the particle of interest was determined based on the EcpH fluorescence quenching of a neighbouring surface-associated particle (Figure 7A, broken line). Note that the above time interval is the lower limit of the actual time the virus spends in non-acidic endosomes, since its uptake occurred at some point during the preceding alkaline perfusion cycle. To account for this uncertainty, half the time of a perfusion cycle was added to the total time EcpH signal was observed after low pH application (TEcpH), yielding an approximate residence time in neutral compartments. As shown in Figure 7(B), nearly 80% of virions entered acidic endosomes within 25 s after being internalized.

Figure 7 HIVpp residence time in non-acidic endosomes

(A) TZM-bl cells with bound HXB2 Env-pseudotyped viruses were perfused with alternating pH 8.8 Taps and pH 5.4 Mes (grey-shaded areas) buffers every 21 s. The EcpH fluorescence of surface-bound particles (crosses and broken line) marks the time of pH drop near the virion of interest (black triangles and solid line). The virus residence time in neutral endosomes was obtained from the interval between the first time point when the EcpH signal failed to drop at acidic pH and the time when this signal disappeared due to entry into acidic compartments (TEcpH). To account for the uncertainty arising from the virus internalization at some time during the preceding alkaline cycle, the cycle half-time (T1/2=10.5 s) was added to the measured interval, thus yielding the residence time in non-acidic endosomes (Tne=TEcpH+T1/2). (B) Distribution of virus residence times in non-acidic endosomes measured as described above.

The practical implication of quick virus entry into acidic compartments is that the laborious perfusion assay could be potentially replaced by a less invasive protocol which simply monitors the time course of EcpH quenching in the absence of perfusion. Indeed, the rate of EcpH quenching and the combined rate of HIV-1 detachment and internalization determined by our perfusion protocol were close (Figure 4C). After correcting the EcpH-quenching data for the loss of signal due to virus detachment (measured independently), one can obtain the rate of virus uptake.


The imaging-based method described in the present study enables continuous monitoring of virus uptake and dissociation from cells. This assay revealed that HIVpp dissociated from HeLa cells expressing CD4 and co-receptors faster than they were internalized. Therefore, in agreement with [24], the kinetically dominant virus-dissociation process limits its ability to infect these target cells. Interestingly, the JRFL Env-pseudotyped virions bound less efficiently to TZM-bl cells in the cold and detached somewhat more readily than isogenic particles bearing HXB2 Env. Hence, tighter receptor-independent HIVpp binding, probably occurring via interactions with heparans and other cellular factors [25,33,34], could slow down virus dissociation and favour endocytosis. Platt et al. [24] have convincingly demonstrated that HIV–cell binding mediated by polycations enhances the infectivity mainly by preventing virus detachment from cells.

It is worth pointing out that the present study examined HIV-1 uptake using pseudoviruses carrying a non-viral marker (EcpH-TM) and target cells engineered to express proper receptors. Thus caution should be exercised when extrapolating these data to detachment and internalization of infectious HIV-1 grown in primary target cells. PBMC (peripheral blood mononuclear cell)-grown HIV-1 contains intercellular adhesion molecules, such as ICAM-1 [21,22,35]. Interactions between the viral ICAM-1 and LFA-1 (lymphocyte function-associated antigen 1) expressed on CD4+ T-cells and macrophages enhance infection through stabilizing the HIV-1 adhesion to natural target cells [3638]. It is therefore expected that infectious HIV-1 would be less prone to detach from T-cells compared with pseudoviruses bound to HeLa-derived target cells. On the other hand, the HIVpp model has been extensively validated and employed for delineating the virus entry and nucleocapsid trafficking. Our control experiments (Figure 1) imply that EcpH-TM is picked up by nearly all entry-competent pseudoviruses without noticeably affecting the incorporation or function of Env. These considerations and the necessity to monitor the virus uptake in real-time appear to justify the usage of labelled pseudoviruses and adherent cell lines to gain insight into this process. The imaging-based method introduced in the present study establishes proof-of-concept for future studies of host cell-grown HIV-1 detachment and internalization by primary CD4+ T-cells and macrophages.

We found that, once internalized, HIVpp are quickly sorted into acidic compartments of HeLa-derived target cells. First, the lack of detectable EcpH signal from multiple virions at low external pH (Figure 3) implies that, at any given time, virtually all cell-associated viruses resided either at the cell surface or in acidic endosomes. Secondly, single-virus tracking combined with continuous cycling between acidic and alkaline pH revealed that the residence time in neutral endosomes was usually less than 30 s. By comparison, HeLa cells sorted a fluid-phase marker into mildly acidic endosomes within 10 min after internalization [39]. In different cell lines, Semliki Forest Virus and influenza virus were trafficked to acidic endosomes within 2–5 min following their uptake [16,40,41]. These times are considerably longer than the brief HIVpp residence time in neutral compartments observed in the present study. On the other hand, the interior of single endocytic vesicles carrying the EcpH-tagged transferrin receptor was acidified within ~30 s, as determined by monitoring the EcpH fluorescence while changing the external pH [13]. This quick transferrin delivery into mildly acidic compartments agrees with the rate of HIVpp entry into these compartments obtained in our experiments. Thus although measurements of internalization of a conventional endocytic marker have not been performed in our experimental system, based on the work of others [13], we are confident that our results are valid. In addition, the similar kinetics of HIVpp entry into HeLa-derived cells in the absence of pH cycling and upon applying our perfusion protocol (Figure 4C) imply that transient exposure to acidic and alkaline buffers did not affect the virus uptake. We thus propose that, like influenza virus [16], HIVpp is preferentially sorted into a quickly maturing population of endosomes.

Note, however, that the quick maturation of influenza-carrying endosomes was defined based on the accumulation of the late endosomal marker Rab7, whereas the lumen of Rab7-positive endosomes became acidic after a considerable delay [16,40]. The exceptionally quick acidification of HIVpp-carrying endosomes indicates that this virus enters through a distinct endocytic pathway, which may ensure quick transport to a perinuclear space where the virus appears to fuse [7]. Sorting into quickly maturing endosomes may therefore be beneficial for enhancing the efficiency of nuclear entry and integration. It is not clear, however, what prevents HIVpp from fusing with the plasma membrane and/or what cellular factors facilitate its fusion with endosomes. Future studies of the HIVpp trafficking and dynamic co-localization with endosomal markers at the time of fusion should further our understanding of its productive entry pathway.

The experimental approach described in the present paper can be readily modified for monitoring the uptake of viruses that fuse at low pH. To avoid prematurely activating viral-fusion proteins, an acidic buffer used in the triple-perfusion protocol should be substituted by a neutral buffer. Even though EcpH would not be fully quenched at neutral pH, the increase in its signal upon shifting to alkaline pH should still be sufficient to monitor virus endocytosis and detachment. The usage of amine-reactive dyes such as CypHer5 [15], which fluoresce brighter at acidic pH, would provide a better and perhaps more universal labelling strategy for studies of virus uptake. The ability to detect both virus delivery into acidic compartments and its subsequent fusion in the same experiment is essential for elucidating the key steps of viral entry. This could be accomplished by incorporating EcpH-TM into viruses containing a lipophilic dye and a diffusible content marker [7].


Kosuke Miyauchi and Mariana Marin performed the experiments and co-wrote the manuscript. Gregory Melikyan conceived and performed the experiments, and co-wrote the manuscript.


This work was supported by the National Institutes of Health [grant number R01 GM054787] (to G.B.M.).


We thank Dr D. Kabat, Dr L. Chernomordik and Dr S. Padilla-Parra for critical reading of the manuscript and helpful discussions prior to submissions, and Dr G. Miesenbock (University of Oxford, Oxford, U.K.) for providing the pHluorin construct.

Abbreviations: ARRRP, AIDS Research and Reference Reagent Program; DMEM, Dulbecco's modified Eagle's medium; GFP, green fluorescent protein; HEK-293T cells, human embryonic kidney-293 cells expressing the large T-antigen of SV40 (simian virus 40); HIVpp, HIV pseudovirus; ICAM-1, intercellular adhesion molecule 1; MOI, multiplicity of infection


View Abstract