Quantitative imaging of endosome acidification and single retrovirus fusion with distinct pools of early endosomes

Abstract

Diverse enveloped viruses enter host cells through endocytosis and fuse with endosomal membranes upon encountering acidic pH. Currently, the pH dynamics in virus-carrying endosomes and the relationship between acidification and viral fusion are poorly characterized. Here, we examined the entry of avian retrovirus that requires two sequential stimuli—binding to a cognate receptor and low pH—to undergo fusion. A genetically encoded sensor incorporated into the viral membrane was used to measure the pH in virus-carrying endosomes. Acid-induced virus fusion was visualized as the release of a fluorescent viral content marker into the cytosol. The pH values in early acidic endosomes transporting the virus ranged from 5.6 to 6.5 but were relatively stable over time for a given vesicle. Analysis of viral motility and luminal pH showed that cells expressing the transmembrane isoform of the receptor (TVA950) preferentially sorted the virus into slowly trafficking, less acidic endosomes. In contrast, viruses internalized by cells expressing the GPI-anchored isoform (TVA800) were uniformly distributed between stationary and mobile compartments. We found that the lag times between acidification and fusion were significantly shorter and fusion pores were larger in dynamic endosomes than in more stationary compartments. Despite the same average pH within mobile compartments of cells expressing alternative receptor isoforms, TVA950 supported faster fusion than TVA800 receptor. Collectively, our results suggest that fusion steps downstream of the low-pH trigger are modulated by properties of intracellular compartments harboring the virus.

Many enveloped viruses use endocytosis and vesicular trafficking to enter host cells, where acidification of an endosomal lumen serves as a trigger for viral fusion (1, 2). Viral fusion proteins are fine-tuned to respond to different pH values. Those that are activated at less acidic pH (∼6.0) are thought to mediate fusion with early endosomes, whereas those with a lower pH threshold (∼5.0) appear to direct the virus entry from late endosomes (1, 3). However, recent evidence implies that factors other than low pH, such as specific endosome-resident lipids, can determine the intracellular compartments from which the viral capsid is released into the cytosol (2, 4, 5). Progress in understanding the complex regulation of virus–endosome fusion has been hindered by poor accessibility of intracellular compartments and lack of direct techniques for monitoring this process in situ.

Single-virus imaging is a powerful tool for gaining critical insights into virus entry (reviewed in ref. 6), but detection of virus–endosome fusion (here defined as the release of viral content into the cytoplasm) is technically challenging (7, 8). To understand the relationship between the pH trigger and virus–endosome fusion, it is essential to visualize these events in real time. A ratiometric method developed by the Zhuang group (9, 10) permits monitoring endosomal pH around single influenza viruses colabeled with pH-sensitive and pH-insensitive fluorescent markers. However, simultaneous measurements of the pH drop and the resulting virus fusion have not been reported so far.

We previously developed tools to visualize virus uptake (11) and to detect single retrovirus–endosome fusion (7, 8, 12) for differently labeled viruses in separate experiments. Here, we incorporated a FRET-based pH sensor into the membrane of the avian sarcoma and leukosis virus (ASLV) and loaded virions with a fluorescent content marker. This technique permitted simultaneous measurements of the pH in virus-carrying endosomes and the kinetics of postacidification steps of fusion. Analyses of acidification and movement patterns of virus-carrying endosomes indicated that ASLV entered two distinct pools of endosomes: slow moving and generally less acidic or dynamic and more acidic compartments. The rate of virus fusion depended on the pH and endosome motility and on the ASLV receptor isoform. For the same pH value, the kinetics of fusion and sizes of nascent fusion pores differed in cells expressing alternative isoforms of the ASLV receptor, which appear to target the virus entry through distinct endocytic routes (13, 14). These findings suggest that late steps of ASLV entry are modulated by properties of intracellular membranes.

Results

Genetically Engineered Sensor for Single Virus-Based pH Measurements.

We previously visualized virus entry into acidic endosomes by incorporating a membrane-anchored ecliptic pHluorin (15) into the viral membrane (7, 11). The nearly complete quenching of the ecliptic pHluorin signal at pH below 6.2 marked the ASLV entry into mildly acidic compartments but did not allow quantitative measurements of endosomal pH. To enable simultaneous measurements of endosomal pH and ASLV fusion, we used a modified version of the FRET-based pH sensor, pHlameleon (16). The CFP (donor) module of pHlameleon was replaced with the monomeric teal fluorescent protein, mTFP1 (Fig. 1A) (17). The higher photostability and brightness of mTFP1 compared with CFP make this protein more suitable for single-virus imaging. The mTFP1-eYFP tandem was appended to the N terminus of the transmembrane domain of human ICAM-1 (Fig. 1A), which is known to incorporate selectively into retroviral particles (e.g., ref. 18). As we have demonstrated previously (7, 11), coexpression of ICAM-1–anchored fluorescent proteins with viral structural proteins results in efficient labeling of the viral membrane. Labeled viruses were immobilized on poly-l-lysine–coated coverslips and imaged using a Zeiss LSM780 confocal microscope in spectral mode (Fig. 1B). At basic pH, mTFP1 fluorescence is weak because of efficient FRET between donor and acceptor eYFP (Fig. S1 and ref. 19). Low pH markedly reduces sensitized emission of eYFP (pK_a 7.1) (20) by shifting its absorption peak (21) (Fig. 1 B and C) and thereby increases the signal from relatively pH-insensitive mTFP1 (pK_a 4.3) (17).

Fig. 1.

Construction of pH-sensing ASLV particles and pH measurements. (A) Drawings of triple-labeled ASLV viruses bearing a pH-sensor in their membranes at different pH. The sensor consisted of a CFP (mTFP1) attached to an eYFP by a short linker (shown respectively as colored cylinders or boxes on the enlarged diagram), which was in turn fused to the N terminus of the ICAM-1 transmembrane domain. The virus interior was labeled with the MLV Gag-mKate2 (red dots). (Left) At basic pH, mTFP1 is poorly fluorescent because of FRET, which is manifested in intense sensitized eYFP emission. (Right) In an acidic environment, eYFP fluorescence is quenched (gray cylinders or boxes), and FRET is disrupted, leading to an increase in the mTFP1 signal. (B) The pH-dependent spectral shift in the tandem protein emission. ASLV pseudoviruses colabeled with mTFP1-eYFP-ICAM and Gag-mKate2 were allowed to adhere to poly-l-lysine–coated coverslips and were imaged at room temperature in citrate/phosphate buffers adjusted to different pH values. Representative micrographs taken at neutral pH (Left) and acidic pH (Right) are shown along with the coloring chart. The mKate2 fluorescence is not included for visual clarity. (C) The mTFP1-eYFP-ICAM emission spectra as a function of pH. (D) The calibration curve obtained by plotting the average ratio of the mTFP1 and eYFP signals from all viruses in the image field at different pH.

To determine the relationship between the pH and emission spectrum of the mTFP1-eYFP sensor, viruses were imaged in solutions of different acidity (Fig. 1 B and C). To ensure that the measured signal was from viruses, particles were colabeled with the content marker, murine leukemia virus (MLV) Gag-mKate2 (red emission). Only puncta positive for all three labels, mTFP1, eYFP, and mKate2, were analyzed, yielding the calibration curve that related the mTFP1/eYFP emission ratio to the external pH (Fig. 1D). The observed spectral shift in the tandem protein fluorescence enabled sensitive pH measurements in individual ASLV-carrying endosomes within a physiologically relevant range. Normally, FRET measurements using sensitized emission require corrections for direct excitation of the acceptor by the 458-nm laser line and spectral bleed-through (donor photons detected in the acceptor channel). However, for intramolecular FRET between the tandem mTFP1-eYFP proteins, the donor/acceptor stoichiometry and their emission ratio are fixed, thus providing a reliable direct measure of pH based on the emission ratio (17).

Imaging the pH Drop in a Single-Virus–Carrying Endosome and Viral Content Release into the Cytosol.

Subtype A ASLV envelope glycoprotein (EnvA) becomes competent to mediate low-pH–dependent membrane fusion only after binding to the avian tumor virus receptor A (TVA receptor) (22, 23). Both isoforms of this receptor, the transmembrane (TVA950) and GPI-anchored (TVA800) proteins (24), support ASLV fusion but appear to direct the virus entry through different endocytic routes (13, 14). To determine the timing and the strength of the pH trigger that activates the receptor-primed ASLV entering through different routes, we imaged fusion of triple-labeled pseudoviruses into cells expressing either TVA800 or TVA950. Viruses were prebound to cells under cold conditions, and their entry was initiated by raising the temperature. ASLV internalization and delivery into acidic endosomes resulted in diminution of the eYFP signal and concomitant increase of the mTFP1 fluorescence (Fig. 2). These changes could be seen readily as a white-to-purple shift in the virus color (Fig. 2, a–b transitions). For most virions, endosomal pH dropped fairly quickly (<20 s) and usually stabilized around pH 6.0 (Fig. 2 B and C and Fig. S2).

Fig. 2.

Measuring the pH drop in ASLV-carrying endosomes and virus–endosome fusion. (A) Diagram of ASLV colabeled with mTFP1-eYFP-ICAM and Gag-mKate2 (a) and of the color changes associated with virus entry into acidic endosomes (b) and fusion (c). (B and C) Triple-labeled ASLV pseudoviruses (appearing as white puncta) were prebound to CV-1 cells expressing either TVA950 (B) or TVA800 (C) in the cold, and fusion was initiated by shifting to 37 °C. The mTFP1 fluorescence is shown in blue, the sensitized eYFP emission in green, and the mKate2 signal in red. Image panels represent consecutive snapshots of viruses before internalization (marked “a”), immediately after entry into acidic endosomes (purple puncta, “b”), and after fusion, which results in the release of mKate2 (blue puncta, “c”). White dashed lines in images B and C outline the cell nucleus. The lower left panels in B and C show the particle’s trajectory pseudocolored according to the time from the onset of the experiment (see corresponding Movies S1 and S2). The graphs show the fluorescence intensities of viral markers as a function of time obtained by single-particle tracking using the mTFP1 channel. Endosomal pH (gray dots) and smoothed pH data (black lines) are shown also. Vertical gray dashed lines illustrate the lag time between acidification and fusion.

Low-pH–dependent ASLV fusion was detected based on the loss of the viral content marker, nucleocapsid-mKate2, a product of the MLV Gag-mKate2 cleavage occurring upon virus maturation (Fig. 2) (25). We found that ∼20% of triple-labeled particles fused with acidic endosomes, as evidenced by the quick loss of the mKate2 signal following the pH drop (Fig. 2, b–c transitions). In contrast to eYFP, the more acid-resistant mTFP1 signal persisted under acidic conditions and thus continued to mark the viral membrane after fusion took place. Similar acidification and fusion patterns were observed in cells expressing TVA950 and TVA800 (Fig. 2 B and C). Fusion was not detected in control experiments performed in the presence of a fully inhibitory dose (0.1 mg/mL) of the ASLV EnvA-derived R99 peptide (26). Thus we were able to measure the pH dynamics in virus-carrying endosomes and detect ASLV fusion in the same experiment.

Endosome Acidity and pH-Dependence of ASLV Fusion Are Not Affected by TVA Isoforms.

Analysis of the changes in fluorescence of the virus-borne pH-sensor showed that acidity of individual endosomes was fairly stable over time (Fig. S2). We did not detect further acidification of lumen for as long as we could reliably measure the pH (usually up to 20 min, at which point most viruses entered the autofluorescent perinuclear area). Steady luminal pH aided more precise determination of this parameter for each vesicle by averaging the mTFP1/eYFP ratio signal (Fig. 2 B and C). The time-averaged values of pH in virus-carrying endosomes were broadly distributed (Fig. 3A, circles), and its mean value before fusion was independent of the receptor isoform: 5.98 ± 0.16 (n = 39) for TVA800 and 5.99 ± 0.14 (n = 57) for TVA950 cells (Fig. 3A, diamonds). The fraction of fusion events as a function of pH was distributed normally, with most events occurring between pH 5.8 and 6.1 (Fig. 3B, black and open bars).

Fig. 3.

Distribution of endosomal pH and pH-dependence of ASLV fusion. (A) Distributions of pH in virus-carrying endosomes in TVA800 and TVA950 cells from 10 independent experiments performed for each cell line. Mean pH values (diamonds) and scatterplots of individual pH values (circles) are shown. Also shown are the mean pH (squares) and scatterplots for mobile (“fast”) and less mobile (“slow”) subpopulations of endosomes. Error bars indicate SDs. (B) pH histograms for nonfusing (hatched and gray bars) and fusing particles (black and open bars). Solid and dashed lines are Gaussian fits to data for TVA800 and TVA950 cells, respectively; curve fits for fusion events are shown by black lines and for nonfusing particles by gray lines. (C) The pH-dependence of ASLV fusion with endosomes is shown as the ratio of fused vs. inactive particles for TVA800 (black circles) and TVA950 (open circles) cells. (D) Velocities as a function of time are shown for 20 representative fusion-competent virions in TVA950 cells before fusion. Velocities below 0.5 μm/s (marked by a thick dashed line) define slow endosomes. (E) Relative frequencies of ASLV fusion with fast (black segments) and slow (open segments) endosomes of TVA800 and TVA950 cells. (F) Decay times for the viral content (mKate2) signal upon ASLV fusion. Twenty individual fluorescence traces for fast (open triangles) and slow endosomes (open circles) in TVA950 cells and for fast (black triangles) and slow (black circles) endosomes in TVA800 cells were averaged and plotted as a function of time. Error bars indicate SEM. The corresponding single exponential fits are shown by solid lines. Pores were significantly larger in fast TVA950 endosomes than in fast TVA800 endosomes (P = 0.001) or in slow TVA950 endosomes (P = 0.032).

The reduced number of fusion events at more acidic pH (Fig. 3B) apparently was caused by the lower abundance of virus carriers with luminal pH below 5.8. This notion was verified by measuring the pH distribution around triple-labeled particles that entered acidic compartments but failed to fuse (Fig. 3B): Only a minor fraction of fusion-incompetent viruses were exposed to pH <5.8. On the other hand, the low occurrence of fusion events relative to the virus delivery into less acidic compartments (pH >6.2) likely reflects suboptimal activation of receptor-primed EnvA. This result is in agreement with the pH threshold of ∼6.2 obtained from ensemble measurements of ASLV fusion (27) and from ASLV EnvA-mediated cell–cell fusion (28).

The mean pH in endosomes carrying nonfusogenic viruses was 6.10 ± 0.19 (n = 46) and 6.10 ± 0.22 (n = 43) in TVA800 and TVA950 cells, respectively. These values were significantly higher (P < 0.022) than the average pH in endosomes transporting fusion-competent viruses. Assuming that all endocytosed virions are fusion competent, one can estimate the pH-dependence of fusion by normalizing the number of viral content release events to the number of internalized particles (from Fig. 3B) across the pH range. ASLV fusion with TVA800 and TVA950 cells exhibited similar pH-dependence (Fig. 3C), indicating that both receptor isoforms efficiently primed EnvA for low-pH–induced fusion. The obtained pH-dependence of ASLV–endosome fusion appeared less steep than for EnvA-mediated cell–cell fusion, which reached the maximal level below pH 5.7 (28; see also ref. 27).

ASLV Enters and Fuses with Distinct Populations of Endosomes.

Visual inspection of fusion events showed that a fraction of viruses entered acidic endosomes and released their content without undergoing considerable displacement from their initial position on the cell surface (Fig. 2B). In contrast, other particles moved quickly before being exposed to low pH and undergoing fusion. Because endosome acidification is coupled to their retrograde trafficking (29), we explored the relationship between endosome motility and luminal pH. Viruses/endosomes that moved faster than an arbitrary threshold (set at 0.5 μm/s) for at least two consecutive frames before or at the time of fusion were regarded as “fast” (Fig. 3D and Fig. S3). This relatively fast and directional movement is typical for trafficking along microtubules (10). Particles that did not reach the threshold velocity were designated “slow” In both cell lines, rapid movement correlated with acidification and usually occurred within 1 min from the pH drop (Fig. S3C). Interestingly, the average pH was significantly higher in slow endosomes than in fast compartments. Fast and slow endosomes in TVA800 cells exhibited average pH values of 6.04 ± 0.16 (n = 20) and 5.92 ± 0.15 (n = 19), respectively (P = 0.022) (Fig. 3A, squares). Likewise, slow endosomes in TVA950 cells were less acidic than fast endosomes: 6.02 ± 0.15 (n = 35) vs. 5.93 ± 0.13 (n = 14) (P = 0.048). The differences in average pH and different motion patterns of fast and slow endosomes suggest these endosomes are distinct, perhaps corresponding to quickly and slowly maturing pools of vesicles, respectively (9).

Velocity analysis for all triple-labeled particles, irrespective of whether they underwent fusion, revealed that the number of internalized ASLV was divided nearly evenly between fast and slow endosomes in both TVA800 and TVA950 cells. However, although the number of fusion events in TVA800 cells also was equally distributed between the two populations of endosomes, ASLV fused less frequently with fast endosomes in TVA950 cells (Fig. 3E). If fusing and nonfusing viruses are sorted similarly by TVA950 cells, the less frequent fusion with fast endosomes would indicate that these compartments are less permissive of fusion than slow endosomes.

To compare the fusion-permissiveness of fast and slow compartments, we assessed the effective sizes of fusion pores formed by ASLV EnvA. Larger initial fusion pores usually correlate with a more robust fusion process, consistent with the higher probability of ASLV fusion and faster release of the viral content (a measure of the pore diameter) in TVA950 as compared with TVA800 cells (7). We found that the decay of the mKate2 signal as a result of fusion with TVA950 cells was faster than for TVA800 cells (Fig. 3F). Thus, pores formed through the GPI-anchored receptor were smaller, because they restricted the efflux of mKate2 from virions. The fact that larger, more robust fusion pores were observed in mobile endosomes of TVA950 cells (Fig. 3F) argues against the possibility that these endosomes are less permissive to fusion than stationary compartments.

Quickly Moving Endosomes Support Faster ASLV Fusion than Less Mobile Compartments.

Measurements of the lag times from shifting to 37 °C to individual acidification events showed that ASLV entered acidic compartments in TVA950 cells earlier than in TVA800 cells (Fig. S3D). This result was expected, based on our previous findings that TVA800 internalized ASLV more slowly than the transmembrane receptor (7, 14). In TVA800 cells, endosome acidification occurred with the same time course, irrespective of their motion pattern, whereas in TVA950 cells the lumen of fast endosomes became acidic at earlier times than in slow compartments (P = 0.023).

We next determined the true kinetics of acid-mediated ASLV fusion with different endosomes by measuring the interval between the pH drop below pH 6.3 (Fig. 3C) and the onset of mKate2 release, as illustrated in Fig. 2 B and C (vertical dashed lines). The lag times to fusion with fast endosomes in both TVA800 and TVA950 cells were considerably shorter than with their respective slow endosomes (P = 0.012 and 0.001, respectively) (Fig. 4A). In other words, the virus’ ability to undergo quick fusion correlated with accelerated/directional movement of carrier endosomes. In addition, fusion with mobile endosomes was faster in TVA950 than in TVA800 cells (P = 0.012), whereas the fusion rates with slow endosomes in the two cell lines were not significantly different (P = 0.151). These shorter lags to ASLV fusion in fast endosomes in TVA950 cells are inconsistent with the possibility that these compartments are less permissive to fusion than slow endosomes.

Fig. 4.

Kinetics of low-pH–triggered ASLV fusion with fast and slow endosomes. (A) The time intervals between the pH drop below 6.3 and the onset of mKate2 release were determined for fusion events in TVA800 (filled symbols) and TVA950 (open symbols) cells and were plotted separately for fast and slow endosomes (triangles and circles, respectively). (B) Correlation between endosomal pH and the delay between acidification and fusion for TVA800 (filled circles) and TVA950 (open circles) cells. Solid and dashed lines are linear regressions for TVA800 and TVA950 data, respectively.

Accelerated fusion with fast endosomes could be driven, at least in part, by their lower luminal pH compared with slow endosomes (Fig. 3A). However, the differences in the mean pH in dynamic and stationary endosomes in TVA950 cells were of only borderline significance (Fig. 3A). More importantly, the waiting times to fusion did not correlate strictly with luminal pH (Fig. 4B). For both TVA950 and TVA800 cells, lag times to fusion exhibited relatively weak dependences on endosomal pH with a greater scatter observed at pH above 5.95. This pattern suggests that, although low pH triggers fusion, other factors, such as heterogeneity of viruses and of endosomal membranes, can modulate postacidification steps of this process. It appears that endosomal compartments in TVA950 cells, but not in TVA800 cells, support the formation of a relatively long-lived (∼1 min) hemifusion intermediate (7). The existence of such an intermediate is consistent with delays between acidification and content release (full fusion) observed in this work (Fig. 4). Together, our findings are in line with the idea that endosomes exhibiting distinct motion patterns vary in their propensity to support ASLV fusion.

Because dynamic endosomes support faster ASLV and faster release of the viral content in TVA950 than in TVA800 cells, it is unlikely that dynamic compartments disfavor ASLV fusion. We therefore reinterpreted the apparent probability of fusion obtained by comparing the frequencies of fusing and nonfusing particles across the pH range (Fig. 3E). It appears that, although nonfusing particles are distributed evenly between the two populations of endosomes, fusion-competent virions are sorted preferentially to slow endosomes in TVA950 cells. Although fewer virions enter fast endosomes, these compartments appear to support ASLV fusion optimally.

Discussion

In this study, retroviruses tagged with a genetically encoded pH-sensor and a fluorescent content marker enabled simultaneous measurements of the pH drop within virus-carrying vesicles and the resulting virus–endosome fusion. From several FRET-based sensors designed to measure endosomal pH (16, 30–32), we used a derivative of pHlameleon (16) incorporated into the viral membrane. Unlike other approaches, this strategy avoided chemical modifications of the virus, which could compromise its ability to undergo fusion. The ICAM-1–anchored pH sensors can be incorporated into retroviral particles pseudotyped with other viral fusion proteins (11) and thus may be used for studies of entry of other enveloped viruses.

The main limitation of single virus-based pH measurements is the relatively weak signal from the tandem fluorescent protein. Although this sensor allowed pH measurements around single viruses down to ∼5.2 in vitro (Fig. 1D), the ability to determine the pH reliably in late endosomes was limited by cell autofluorescence. This problem can be alleviated by incorporating a larger number of pHlameleon-based sensors into the viral membrane and/or by implementing alternative labeling strategies (32). Further improvements in sensitivity and temporal resolution of pH measurements could shed light on biogenesis of acidic compartments. The initial pH drop to ∼6.0 followed by a stationary phase (Fig. 2 and Fig. S2) appears surprising, considering that endosome maturation is thought to be coupled to gradual acidification. However, to our knowledge, pH measurements in a single maturing endosome have not been reported (perhaps with the exception of refs. 9 and 10). Ensemble measurements of endosomal pH are not likely to resolve the pH profile observed here for single endosomes.

Simultaneous measurements of endosomal pH and the release of viral content into the cytosol afford a unique opportunity to define the relationship between the strength of the fusion trigger and the rate and efficiency of this process. A weak correlation between endosomal pH and the lag before fusion (Fig. 4B) implies that additional factors, such as the property of endosomal membranes, can modulate the fusion reaction. Indeed, the ASLV fusion pores were larger (Fig. 3F) and the lag times to fusion were shorter (Fig. 4A) in dynamic endosomes than in slow endosomes of TVA950 cells. In contrast, nascent fusion pores formed in either fast or slow endosomes of TVA800 cells were not statistically different. A possible explanation is that, because the GPI-anchored receptor internalizes the virus through a raft-dependent pathway (14), particles trafficked to distinct compartments may remain associated with raft-like subdomains that are equally permissive to fusion.

Paradoxically, quick and robust ASLV fusion with fast endosomes in TVA950 cells (Figs. 3F and 4A) was observed relatively infrequently as compared with respective compartments in TVA800 cells (Fig. 3E). This difference could result from differential sorting of fusion-competent and -incompetent particles in TVA950 cells. If fusogenic particles incorporate more Env glycoproteins than fusion-incompetent ones, they could induce stronger signaling and be sorted into a different endosome population than inactive particles. The lack of selective sorting (and perhaps signaling) of ASLV in TVA800 cells is consistent with its slow uptake occurring at the rate of receptor-independent endocytosis in parental CV-1 cells (7).

The possibility of selective sorting of fusion-competent, but not inactive, ASLV into a subset of endosomes makes it difficult to evaluate the fusion efficiency simply by normalizing the number of fusion events to the total number of internalized particles. Thus it is possible that the real pH-dependence of ASLV–endosome fusion is steeper than shown in Fig. 3C and is similar to EnvA-mediated cell–cell fusion (28). The apparently selective ASLV entry into slow endosomes in TVA950 cells is in contrast to sorting of the influenza virus to mobile, quickly maturing endosomes (9). These differences likely reflect the markedly lower pH optimum for influenza fusion (pH 4.8–5.2) (33) compared with ASLV fusion (pH ∼5.5) (27, 28). Entry into quickly maturing endosomes likely ensures faster delivery of influenza viruses into sufficiently acidic compartments.

Accumulating evidence supports the notion that TVA950 and TVA800 direct the ASLV entry through distinct routes (13, 14). It currently is not known whether ASLV fuses with early endosomes in both cell lines; such fusion would suggest quick convergence of different endocytic routes to early endosomes where fusion takes place shortly after acidification. Future studies with cells expressing specific markers of early and late endosomes, Rab5 and Rab7, should provide further insights into ASLV-trafficking pathways and reveal the identity of endosomes supporting the fusion reaction.

Methods

Cell Lines, Plasmids, and Virus Preparation.

Cell lines and reagents used in this study, as well as pseudovirus production protocols, have been described before (7, 8, 12). For details, see SI Methods.

Calibration of a Nano-pH-Meter.

MLV-based pseudoparticles carrying ASLV EnvA were colabeled with Gag-mKate2 (content marker) and mTFP1-eYFP-ICAM (membrane marker). Viruses were adhered to poly-l-lysine–coated eight-well coverslips (Nunc) and visualized using a Zeiss LSM780 confocal microscope (Carl Zeiss Microsystems). To relate changes in the mTFP1/YFP-sensitized emission to extraviral pH, viruses were exposed to a set of citrate/phosphate buffers between pH 4.3 and 7.3 and were imaged using a 63×/1.4 NA oil immersion objective. Fluorophores were excited with the 458-nm line of an argon laser, and emission spectra between 468 and 600 nm were acquired with a 32-channel GaAsP detector in the spectral mode. To minimize the effect of photobleaching, the calibration curve was obtained from nine separate experiments, each using a buffer of different acidity. The fluorescence ratios used for the calibration curve were obtained by summing the intensity of spectral images between 468–503 nm for the mTFP1 emission and 512–547 nm for the sensitized eYFP emission. This configuration was preserved when carrying out measurements in live cells using the same GaAsP detector. The fluorescence signal from all viruses in the image field (usually around 100 particles) was integrated and plotted as a function of wavelength. Background subtraction and a low-pass filter were applied on all spectral images using ImageJ (National Institutes of Health). The experimental points were fitted, as described in ref. 16, yielding the apparent pK_a 5.95.

Imaging Virus Entry into Acidic Compartments and Fusion.

CV-1 cells expressing either TVA950 or TVA800 were grown to near confluency on glass-bottomed 35-mm Petri dishes (MatTek) in phenol red-free medium. Dishes were placed on ice, washed with cold HBSS, and centrifuged with ∼1.5 × 10⁴ IU of pseudoviruses colabeled with mTFP1-eYFP-ICAM and Gag-mKate2 at 2,100 × g (4 °C) for 20 min. Unbound viruses were removed by washing, and cells were mounted onto a microscope stage maintained at 37 °C. Virus internalization and fusion were imaged using a Zeiss LSM 780 confocal microscope with a 63×/1.4 NA oil immersion objective. mTFP1 and mKate2 were excited with the 458- and 561-nm laser lines, respectively. Fluorescence emission was detected with the 32-channel spectral detector. The first 16 channels recorded the mTFP1 signal, and the remaining 16 channels acquired the sensitized eYFP emission signal. The mKate2 fluorescence was detected using a cooled photomultiplier tube. Images were acquired for 35–40 min approximately every 8 s.

Single-virus tracking was performed with Imaris (BitPlane) or Volocity (Perkin-Elmer). Changes in the average fluorescence ratio of mTFP1 fluorescence and of sensitized eYFP emission from internalized viruses were converted to pH units using the calibration curve. Virus-carrying endosomes were considered quickly moving (“fast”) if (i) the particle’s velocity exceeded 0.5 μm/s for at least three consecutive frames before virus fusion and (ii) increased velocity was associated with directional movement toward the nucleus. The endosomal pH distributions were compared using the two-tailed Student t test, and distributions of waiting times for fusion were compared using the Mann–Whitney rank sum test.