An Inducible ESCRT-III Inhibition Tool to Control HIV-1 Budding

HIV-1 budding as well as many other cellular processes require the Endosomal Sorting Complex Required for Transport (ESCRT) machinery. Understanding the architecture of the native ESCRT-III complex at HIV-1 budding sites is limited due to spatial resolution and transient ESCRT-III recruitment. Here, we developed a drug-inducible transient HIV-1 budding inhibitory tool to enhance the ESCRT-III lifetime at budding sites. We generated autocleavable CHMP2A, CHMP3, and CHMP4B fusion proteins with the hepatitis C virus NS3 protease. We characterized the CHMP-NS3 fusion proteins in the absence and presence of protease inhibitor Glecaprevir with regard to expression, stability, localization, and HIV-1 Gag VLP budding. Immunoblotting experiments revealed rapid and stable accumulation of CHMP-NS3 fusion proteins. Notably, upon drug administration, CHMP2A-NS3 and CHMP4B-NS3 fusion proteins substantially decrease VLP release while CHMP3-NS3 exerted no effect but synergized with CHMP2A-NS3. Localization studies demonstrated the relocalization of CHMP-NS3 fusion proteins to the plasma membrane, endosomes, and Gag VLP budding sites. Through the combined use of transmission electron microscopy and video-microscopy, we unveiled drug-dependent accumulation of CHMP2A-NS3 and CHMP4B-NS3, causing a delay in HIV-1 Gag-VLP release. Our findings provide novel insight into the functional consequences of inhibiting ESCRT-III during HIV-1 budding and establish new tools to decipher the role of ESCRT-III at HIV-1 budding sites and other ESCRT-catalyzed cellular processes.

SiRNA depletion experiments suggested that HIV-1 budding requires, in principle, only one CHMP4 and one CHMP2 isoform to facilitate egress, whereas CHMP4 filament assembly provides a platform for downstream CHMP2 recruitment [52].However, CHMP3 cooperates with CHMP2A to increase budding efficiency substantially, while CHMP2B acts independent of CHMP3 [53].
Live cell imaging of budding sites showed that Gag recruitment to the plasma membrane and VLP formation is, on average, completed within ~10 min [54].Alix and Tsg101 (ESCRT-I) progressively accumulate with Gag [55], and once Gag recruitment is terminated, ESCRT-III proteins and VPS4 transiently appear at the budding site, followed by virus release [55,56].Gag colocalizing ESCRT-III clusters showed closed, circular structures with an average size of 45 to 60 nm [57].Recruitment is sequential, with CHMP4B arriving before CHMP2A followed by VPS4.This was suggested to constrict the neck, and in case scission does not occur within minutes after ESCRT-III remodeling and disassembly, ESCRT-III and VPS4 are recruited again to the same budding site [58].The residence time of ESCRTs at the budding site is very short, a few minutes or less [59], and only 1 to 3% of total budding sites per cell exhibit ESCRT-III and Gag colocalization [57].
In order to facilitate imaging, which is challenged by the transient nature of ESCRT-III function, particularly in a context where cell physiology is not excessively compromised, we developed a drug-inducible transient inhibitory ESCRT-III system.To this end, CHMP2A, CHMP3, and CHMP4B (referred to collectively as CHMP) proteins were fused to the hepatitis C virus protease NS3 (NS3) via a short linker containing the NS3 cleavage site, which permits autocleavage (see Table S1).Treatment of cells with the NS3 inhibitor Glecaprevir would induce the temporal accumulation of full-length CHMP-NS3-FP proteins.This accumulation potentially exerts a dominant-negative effect on ESCRT-III function, as previously reported when a heterologous protein is fused to CHMP proteins (Figure 1) [17,18].We find that inducible CHMP3-NS3-FP had no effect on VLP release in contrast to CHMP3-YFP fusions, but increased the inhibitory effect in combination with CHMP2A-NS3-FP.Furthermore, we demonstrate that inducible CHMP4B-NS3-FP fusions and CHMP2A-NS3-FP fusion proteins can be fine-tuned to reduce Gag-VLP release and temporally increase the number of budding sites, revealing Gag-ESCRT colocalization.
For immunoblotting analysis, cells were seeded at 60% confluence into 100 mm dishes and transfected 24 h later using the Jetprime technique (Ozyme, Saint Cyr lʹEcole, France).The cells were cotransfected with 0.5 µg of Gag, the specified amount of CHMP-NS3-FP, and a total transfected DNA amount of 8 µg.Glecaprevir (CliniSciences, Nanterre, France) was used at a concentration of 25 µM.
Whole cells and VLPs protein extracts were prepared as previously described [64].Western blots were conducted using the following antibodies: Anti-HIV-1 p24 (ARP-3537, NIH AIDS Reagent Program), Anti-HIV-1 Vpu (ARP-969, NIH AIDS Reagent Program), Anti-Bst2 (ARP-11721, NIH AIDS Reagent Program), anti-GFP (#A11122, Life Technologies, Courtaboeuf, France), and Anti-Flag (#F7625, Sigma Aldrich, St. Quentin Fallavier , France).Quantification was based on densitometry comparing Gag detection in the VLP fraction to total Gag within cells.To additionally correct for Gag intensities, densitometry values were normalized for Gag to 1 (by dividing the raw mean values by those measured in Gag).
For immunoblotting analysis, cells were seeded at 60% confluence into 100 mm dishes and transfected 24 h later using the Jetprime technique (Ozyme, Saint Cyr l'Ecole, France).The cells were cotransfected with 0.5 µg of Gag, the specified amount of CHMP-NS3-FP, and a total transfected DNA amount of 8 µg.Glecaprevir (CliniSciences, Nanterre, France) was used at a concentration of 25 µM.
Whole cells and VLPs protein extracts were prepared as previously described [64].Western blots were conducted using the following antibodies: Anti-HIV-1 p24 (ARP-3537, NIH AIDS Reagent Program), Anti-HIV-1 Vpu (ARP-969, NIH AIDS Reagent Program), Anti-Bst2 (ARP-11721, NIH AIDS Reagent Program), anti-GFP (#A11122, Life Technologies, Courtaboeuf, France), and Anti-Flag (#F7625, Sigma Aldrich, St. Quentin Fallavier, France).Quantification was based on densitometry comparing Gag detection in the VLP fraction to total Gag within cells.To additionally correct for Gag intensities, densitometry values were normalized for Gag to 1 (by dividing the raw mean values by those measured in Gag).
For live cell imaging, cells were seeded in fibronectin-coated glass-bottomed µ-dishes (Biovalley, Nanterre, France).HeLa CCL2 cells beyond passage p15 and HeLa Kyoto BST2-cells were transfected with an excess of an untagged version of Gag to prevent the morphological defects associated with particles assembled solely from fluorescently tagged Gag [67].Cells were cotransfected with 0.4 µg of pGag, 0.1 µg of pGag-mCherry, and either 0.5 µg of GFP-VPS4A_E228Q, 0.5 µg of GFP-VPS4B_R253A, or 0.5 µg of pCHMP-NS3-FP.Fluorescent imaging of live cells was conducted 24 h after transfection, following a 4 h treatment with Glecaprevir (25 µM, CliniSciences, Nanterre, France) or an equivalent volume of DMSO.An exception was made in the case of the HeLa CCL2 TIRF videomicroscopy experiment, where Glecaprevir incubation was limited to 2 h.
For electron microscopy imaging, FreeStyle293F cells were seeded in a 6-well plate (Life Technologies, Courtaboeuf, France) at a density of 2 × 105 cells per well.After 24 h, the cultures were cotransfected with 2 µg of pCHMP-NS3-Flag and 2 µg of pGag using 4 µL of 293Fectin Reagent (Life Technologies, Courtaboeuf, France).Cells were pelleted and high-pressure frozen as described in [68] using an HPM100 system (Leica Microsystems, Vienna, Austria).After freezing, samples were cryo-substituted in an AFS2 machine (Leica Microsystems, Vienna, Austria), dehydrated, and embedded in anhydrous Araldite resin.
Electron microscopy: Ultrathin sections were observed using an FEI G2 Tecnai transmission electron microscope (TEM) equipped with an Orius SC1000 CCD camera.

Single Particle Tracking
The open-source Icy software 2.0.3 (https://icy.bioimageanalysis.org/,accessed on 16 October 2020) was used to semiautomatically track a large number of individual Gag-mCherry puncta [69].The procedure included defining the region of interest at the cell's leading edge, where Gag-mCherry spots were not densely distributed.Gag-mCherry particles were detected with the spot detector plug-in, and their tracks trajectories were subsequently determined using the spot tracking plug-in [69][70][71].Manually, tracks at the edge of the region of interest and short tracks (≤2) were excluded.Subsequently, the track manager tool was employed to calculate velocities and tracking durations for each particle.Statistical analysis was conducted using GraphPad Prism 9 software.

Expression and Autocleavage of CHMP-NS3-FP
Immunoblotting experiments were conducted to assess protein expression of CHMP-NS3-FP constructs in HEK-293 cells.Control cells treated with DMSO exhibited a timedependent accumulation of cleaved NS3-FP proteins after transfection with CHMP2A-NS3-FP (Figure 2a), CHMP3-NS3-FP (Figure 2b, left panels), and CHMP4B-NS3-FP (Figure 2c).The raw dataset for Western blots is available in Figure S1.Glecaprevir treatment led to the temporal accumulation of noncleaved full-length CHMP-NS3-FP proteins as early as one hour (Figure 2a-c, right panels), and longer incubation times led to an increased presence of uncleaved CHMP-NS3-FP proteins.Notably, CHMP4B-NS3-FP showed the highest accumulation of uncleaved fusion proteins upon Glecaprevir treatment, while CHMP2A-NS3-FP reached its maximum after 4 h, and CHMP3-NS3-FP was least efficient in the inhibition of autocleavage (Figure 2a-c).
showed the highest accumulation of uncleaved fusion proteins upon Glecaprevir treatment, while CHMP2A-NS3-FP reached its maximum after 4 h, and CHMP3-NS3-FP was least efficient in the inhibition of autocleavage (Figure 2a-c).
In order to assess the stability of the CHMP-NS3-FP proteins over time, a pulse-chase experiment was conducted.Notably, an increase in the amount of NS3-FP protein was observed, suggesting that newly expressed CHMP2A-NS3-FP, CHMP3-NS3-FP, and CHMP4B-NS3-FP proteins are autocleaved (Figure 2d-f).Remarkably, the quantity of full-length CHMP2A-NS3-FP, CHMP3-NS3-FP, and CHMP4B-NS3-FP was not equal and corresponded to the efficiency of fusion protein generation upon Glecaprevir treatment.
We conclude that the addition of Glecaprevir leads to a rapid and stable accumulation of full-length CHMP-NS3-FP proteins over time.In order to assess the stability of the CHMP-NS3-FP proteins over time, a pulsechase experiment was conducted.Notably, an increase in the amount of NS3-FP protein was observed, suggesting that newly expressed CHMP2A-NS3-FP, CHMP3-NS3-FP, and CHMP4B-NS3-FP proteins are autocleaved (Figure 2d-f).Remarkably, the quantity of full-length CHMP2A-NS3-FP, CHMP3-NS3-FP, and CHMP4B-NS3-FP was not equal and corresponded to the efficiency of fusion protein generation upon Glecaprevir treatment.
We conclude that the addition of Glecaprevir leads to a rapid and stable accumulation of full-length CHMP-NS3-FP proteins over time.
We conclude that Glecaprevir administration permits imaging of Gag colocalization with CHMP2A-NS3-green, CHMP3-NS3-green, and CHMP4B-NS3-green at the plasma membrane, indicative of a prolonged half-life of uncleaved CHMP-NS3-FP proteins at HIV-1 Gag budding sites.

Partial Inhibition of Gag VLP Release by CHMP-NS3-FP Proteins
We evaluated Gag in released VLPs and in whole cell extract (WCE) using immunoblotting in HEK-293 cells (raw dataset for Western blots is available in Figure S1).As controls, we included uncleavable CHMP-mutNS3-FP proteins without NS3 protease cleavage site and catalytic inactive dominant negative GFP-VPS4A E228Q [14] (Table S1).As expected, Gag cotransfection with GFP-VPS4A E228Q, CHMP2A-mutNS3-FP, or CHMP4B-mutNS3-FP resulted in a significant impairment of HIV-1 Gag VLP discharge, to the extent that emitted VLPs were undetectable compared to wild-type Gag VLP release (Figure 4a-c).We also tested the VPS4B R253A mutant that revealed slower kinetics in disassembling ESCRT-III CHMP2A-CHMP3 helical polymers in vitro [64] and demonstrated a reduction in VLP release (Figure 4f).In DMSO-treated cells, VLP emission was minimally affected when CHMP3-NS3-FP or CHMP4B-NS3-FP were expressed while CHMP2A-NS3-FP expression led to a 20% decrease in VLP release (Figure 4a-c).

Cell-Line-Specific and Dose-Dependent Effects of CHMP-NS3-FP Proteins
Our assay was designed as an imaging tool for studying ESCRT-III proteins, prompting us to assess its efficacy in the HeLa CCL2 cell line.Notably, unlike Hek293 cells, HeLa CCL2 cells have been documented to express BST2 tetherin proteins, known for their ability to restrict the release of HIV-1 virions [72,73].We thus include a HeLa Kyoto cell line which has been BST2-deleted and stably expresses MKLP1-GFP [66].As previously reported, HeLa Kyoto BST2+ cells, when transfected with Gag, did not liberate VLPs, while HeLa Kyoto BST2-cells did (Figure S2a) [73].Surprisingly, our HeLa CCL2 cells displayed a significant accumulation of free VLPs, indicating a moderate expression of BST2 in this particular HeLa CCL2 cell line.In support of this, immunoblotting revealed reduced BST2 expression in HeLa CCL2 compared to HeLa Kyoto BST2+ cells, with no detectable expression of BST2 in either Hek293 or HeLa Kyoto BST2-cells (Figure S2b).Furthermore, cotransfection with Vpu, the BST2 inhibiting factor, poorly enhanced the accumulation of free VLPs in HeLa CCL2 cells (Figure S2a).In conclusion, our results suggest that the weak expression of BST2 in the HeLa CCL2 cell line does not efficiently impede the liberation of VLPs.
To assess assay efficiency in HeLa Kyoto BST2− and Hek293 cell lines, we cotransfected these cells with Gag and varying amounts of CHMP-mutNS3-green or Glecaprevir-treated CHMP-NS3-green constructs, as outlined in Figure 5 (raw dataset for Western blots is available in Figure S1).It is worth noting that in HeLa Kyoto BST2− cells, exposure to Glecaprevir led to a modest reduction in VLP release when using the CHMP4B-NS3mutgreen and CHMP2A-NS3mut-green constructs.Nonetheless, the results unequivocally demonstrate a dose-dependent decrease in HIV-1 VLP release (Figure 5).We conclude that budding inhibition efficiency by Glecaprevir-treated CHMP4B-NS3-green and CHMP2A-NS3-green proteins is cell-line-dependent and correlates with expression.HIV-1 Gag proteins were detected using an anti-p24 antibody, while CHMPs-NS3-green were detected using an Anti-Flag antibody.Note that the exposure time for Western blot revelation was typically 30 s for Hek293 cells and 5 min for HeLa CCL2 and HeLa Kyoto cells.

Quantitative Analysis of HIV-1 Gag VLP Dynamics Reveals Drug-Dependent Prolongation of Gag VLP Retention
We hypothesized that CHMP-NS3-FP's inhibition of VLP liberation could potentially lead to a delay in VLP release.To precisely measure the timing of VLP release at the cellular level, we conducted thorough TIRF video-microscopy imaging on HeLa CCL2 and HeLa Kyoto BST2-cells.We recorded movies spanning 10 min with a temporal resolution of 0.45 s per frame.It is important to mention that HeLa Kyoto BST2-cells cotransfected with Gag/Gag-mCherry/CHMP2A-NS3-FP or CHMP4B-NS3-FP and treated with Glecaprevir exhibited a rounded morphology in over 25% of the cells, indicating a potential toxic effect associated with the transfected constructs.Interestingly, this phenomenon was not observed in HeLa CCL2 cells.We tracked Gag-mCherry spots and conducted quantitative analyses of their maximum velocity and duration (details in Materials and Methods).Notably, the distribution of these values appeared consistent across different cells within each experimental condition (Figure S3).Gag-mCherry spots might correspond to (i) VLPs in the process of assembly or assembled but not yet released, (ii) VLPs release from the plasma membrane but tethered by the glycocalix and/or proteins or adhering to the glass surface, and (iii) free VLPs moving in the narrow space between the cell and the coverslip.Freemoving virions have been described to exhibit a high maximum velocity [56,74].Consistent with previous findings [56], we observed in both HeLa CCL2 and HeLa Kyoto BST2-cells that the coexpression of VPS4A_E228Q significantly reduced the population of spots with high maximum velocity (>0.625 µm/s) (Figure S4a,b).Furthermore, the expression of both CHMP2A-NS3-FP and CHMP4B-NS3-FP led to a reduction in the number of spots with high maximum velocity (Figure S4c,d).We conclude that in HeLa CCL2 cells, both CHMP2A-NS3-FP and CHMP4B-NS3-FP proteins markedly reduce the number of VLPs released at the individual cell level.
In addition, these findings affirm the accuracy of our tracking analysis.Next, we analyzed the individual tracking time duration of Gag-mCherry dots.As expected, cells transfected with a mixture of Gag-mCherry/Gag and Vps4A_E228Q displayed a strong increase in spots tracked for the whole movie duration (19.5 ± 5.8% compared to 1.5 ± 0.9% in control, p < 0.0001, unpaired t-test; Figure 6a).These spots most probably correspond to VLPs assembled but blocked in the process of membrane fission.For CHMP-NS3-FP constructs, Glecaprevir addition enhanced the duration time in a specific way.CHMP2A-NS3-FP triggered a 3.4-fold enhancement of Gag-mCherry dots that lasted for the entire 10 min recording time from 1.4 ± 0.7% in DMSO to 4.1 ± 1.3% in the Glecaprevir group (p = 0.0072, unpaired t-test; Figure 6c).In contrast, CHMP4B-NS3-FP enhanced the proportion of tracks that had a duration greater than one minute from 7.7 ± 0.5% in DMSO to 18.7 ± 2.8% in the Glecaprevir group (p < 0.0001, unpaired t-test; Figure 6d).Notably, the observed phenotype of CHMP2A-NS3-FP and CHMP4B-NS3-FP fusion enhancing tracks lasting time was significantly improved when examining colocalized spots (Figure 6c,d).
fusion enhancing tracks lasting time was significantly improved when examining colocalized spots (Figure 6c,d).
Our findings underscore a drug-dependent prolongation of the lifetime of HIV-1 Gag VLPs at a cell's surface.This effect was particularly pronounced at sites where CHMP4B-NS3-FP and CHMP2A-NS3-FP accumulate.Our findings underscore a drug-dependent prolongation of the lifetime of HIV-1 Gag VLPs at a cell's surface.This effect was particularly pronounced at sites where CHMP4B-NS3-FP and CHMP2A-NS3-FP accumulate.

Electron Microscopy Reveals Drug-Induced Impairment of Late-Stage HIV-1 VLP Budding
To gain insight into the structure of VLPs within cells, both in the presence and absence of the drug, we conducted electron microscopy imaging.Specifically, we examined 293FS cells that had been cotransfected with CHMP2A-NS3 or CHMP4B-NS3 along with Gag.These cells were subjected to high-pressure freezing, resin embedding, and subsequent imaging.
In cells treated with DMSO, we observed only a minimal presence of cell-associated VLPs (as depicted in Figure 7a,g).In stark contrast, cells treated with Boceprevir, an NS3 inhibitor, for a duration of 2 h exhibited numerous particles that remained tethered to the parental cell via membrane stalks (Figure 7).This observation suggests that VLPs had assembled but were unable to undergo release.
In conclusion, our findings suggest that fusion proteins CHMP2A-NS3 or CHMP4B-NS3 impair the late stage of budding, specifically impeding plasma membrane fission.

Electron Microscopy Reveals Drug-Induced Impairment of Late-Stage HIV-1 VLP Budding
To gain insight into the structure of VLPs within cells, both in the presence and absence of the drug, we conducted electron microscopy imaging.Specifically, we examined 293FS cells that had been cotransfected with CHMP2A-NS3 or CHMP4B-NS3 along with Gag.These cells were subjected to high-pressure freezing, resin embedding, and subsequent imaging.
In cells treated with DMSO, we observed only a minimal presence of cell-associated VLPs (as depicted in Figure 7a,g).In stark contrast, cells treated with Boceprevir, an NS3 inhibitor, for a duration of 2 h exhibited numerous particles that remained tethered to the parental cell via membrane stalks (Figure 7).This observation suggests that VLPs had assembled but were unable to undergo release.
In conclusion, our findings suggest that fusion proteins CHMP2A-NS3 or CHMP4B-NS3 impair the late stage of budding, specifically impeding plasma membrane fission.

Discussion
A large number of cellular membrane remodeling processes are catalyzed by the ESCRT machinery, which can be efficiently blocked by dominant negative forms of ESCRT-III or VPS4 [22,[75][76][77].However, strong dominant negative inhibition affects multiple cellular processes, and the observed effect on a given process may be influenced by blocking many essential cellular functions, not only the targeted pathway.To address this issue, we developed an inducible and transient ESCRT-III inhibition system based on the NS3 hepatitis C protease, described by the Tsien lab [78,79].In brief, a cis-acting NS3 protease was fused to the C-terminus of ESCRT-III via a short linker containing the NS3 cleavage site.To track the fusion protein, a fluorescent protein (FP) was further genetically fused Cterminally of NS3.The CHMP-NS3-FP fusion proteins underwent autocleavage, producing transiently wild-type-like CHMP, consistent with previous reports employing cis-acting NS3 [78].NS3-catalyzed autocleavage likely occurs immediately upon expression since no fusion protein can be detected upon one hour of expression.Likewise, inhibition of autocleavage with Glecaprevir led to the transient accumulation of full-length CHMP-NS3-FP fusion proteins detectable within just 30 min, and maximum levels after 4 h of treatment for CHMP2A-NS3-green and CHMP3-NS3-green, or 24 h for CHMP4B-NS3-Green.
C-terminal fusions of CHMP1 have been shown to exert a dominant negative effect on vesicle trafficking [80], and HIV-1 budding can be inhibited with CHMP3-YFP, CHMP3-RFP, and CHMP4A/C-RFP fusion proteins [17,18].Furthermore, N-terminal ESCRT-III fusion proteins were also reported to be dominant negative [18,81].We chose to target the C-terminus because N-terminal fusions of CHMP proteins act differently from C-terminal ESCRT-III fusions.N-terminal fusions likely interfere with membrane interaction due to the N-terminal membrane insertion domains present in most ESCRT-III proteins, which may in turn also affect polymerization on membranes [34,43,82].In contrast, the effect of C-terminal fusions is likely linked to the function of VPS4, which is required for ESCRT-III remodeling and disassembly [25].Here, we showed that uncleavable CHMP-NS3-FP fusion proteins containing CHMP2A and CHMP4 effectively impede the release of HIV-1 Gag VLPs.Notably, CHMP3 demonstrates a limited impact on inhibiting VLP discharge, in contrast to the strong inhibitory effects observed with CHMP3-YFP and CHMP3-RFP [17,18].We hypothesize that the observed dominant negative effect of C-terminal fusions is due to the overall conformation of the fusion protein, which can sterically interfere with the activity of VPS4.This is consistent with CHMP2B, CHMP3, and CHMP4B tagged with GFP via a long flexible linker (LAP-tag) showing no detectable perturbation of abscission in HeLa cells [37].We therefore propose that CHMP4B-NS3-FP and CHMP2A-NS3-FP perturb VPS4 activity, which in turn reduces VLP release.In contrast, CHMP3-NS3-FP does not affect VPS4 activity, suggesting that its structure is compatible with VPS4 function.Blocking autocleavage of CHMP4B and CHMP2A by NS3 with Glecaprevir does not fully inhibit VLP release, and, notably, blocking autocleavage of CHMP3-NS3-FP has no effect on VLP release, but enhances the effect of CHMP2A-NS3-FP, in agreement with the reported synergy of CHMP2A-CHMP3 on HIV-1 budding [53] and CHM2A-CHMP3 heteropolymer formation in vitro [34].Interestingly, a VPS4B mutant that was previously shown to slow down the kinetics of ESCRT-III disassembly in vitro [64] showed a similar reduction in VLP release as inhibition of CHMP4B and CHMP2A autocleavage.We propose that the inhibited fusion proteins slow down the ESCRT-III/VPS4 machinery, leading to a reduced detection of VLPs.
The effect of Glecaprevir, the accumulation of full-length CHMP-NS3-FP proteins, appears to be counterbalanced by protein synthesis, resulting in the continuous presence of cleavage products, NS3-FP and CHMP proteins.The latter are supposed to cooperate with native CHMPs in regular ESCRT-III function.Strikingly, the full-length CHMP-NS3-FP protein accumulation correlated with their dominant negative activity, suggesting that the dosage of full-length CHMP-NS3-FP proteins determines the slow-down effect of ESCRT-III function.
We demonstrated that the reduction in VLP release is cell-line-dependent.Hek293 cells exhibited a stronger dominant negative activity of CHMP-NS3-FP, whereas HeLa cells displayed lower inhibition of VLP release.Notably, HeLa cells have been reported to consistently express BST2, which traps assembled HIV-1 particles at the cell surface and is countered by the HIV-1 accessory protein Vpu [72,73].In this study, we observed that BST2 expression is higher in HeLa Kyoto cells compared to HeLa CCL2 cells.Interestingly, in HeLa CCL2 cells, HIV-1 VLP particles were efficiently liberated even in the absence of Vpu, although its presence did enhance the release slightly.These findings are consistent with a certain BST2 threshold concentration being necessary for sequestering HIV-1 particles.
We further showed that the NS3-FP fusions affect ESCRT-III localization.Full-length CHMP3-NS3-green was mainly distributed throughout the cytosol, while CHMP2A-NS3-FP and CHMP4B-NS3-FP revealed some punctuate staining at the plasma membrane and at endosomes and/or multivesicular endosomes.HIV-1 Gag has been shown to accumulate at the plasma membrane until Gag VLP bud formation, which triggers the recruitment of ESCRT-III followed by Vps4A [54,55,59].This recruitment process is of short duration, typically lasting less than two minutes, resulting in a relatively low fraction of HIV-1 budding sites displaying colocalization with ESCRT complexes, estimated to be between 1.5% and 3.4% [57,58].
In our current study, we provided compelling evidence that noncleaved CHMP4B-NS3-FP and CHMP2A-NS3-FP proteins exhibit an extended half-life at HIV-1 budding sites, concurrently demonstrating a substantial colocalization of these proteins with Gag-mCherry.In addition, expression of CHMP4B-NS3-FP or CHMP2A-NS3-FP proteins extends the period during which VLP particles could be tracked at the cell surface.We also directly observed an increased number of particles connected to the parental cell by a plasma membrane neck.From these observations, we infer that the accumulation of CHMP4B-NS3-FP and CHMP2A-NS3-FP proteins acts to decelerate the process of plasma membrane fission at HIV-1 budding sites.
In summary, we set up a novel inducible ESCRT-III system that will be helpful for imaging ESCRT-III at HIV-1 budding sites to obtain more insight into its structure and function.Finally, the system will be useful to study the large plethora of ESCRT-dependent membrane remodeling processes.

Figure 1 .
Figure 1.Schematic illustrating our method for transient expression of ESCRT-III fused to NS3 protease and a fluorescent protein using a drug.ESCRT-III proteins are fused to the NS3 cleavage site, NS3 protease, and fluorescent protein.Cells transfected with this construct express wild-type ESCRT-III proteins.Upon addition of the drug, a full-length fusion protein accumulates.For imaging and immunoblotting detection, fluorescent proteins (mostly mNeonGreen) and a Flag tag were further added to the NS3 C-terminus, creating CHMP4B/2A/3-NS3-FP-Flag constructs (hereafter collectively referred to as CHMP-NS3-FP).

Figure 1 .
Figure 1.Schematic illustrating our method for transient expression of ESCRT-III fused to NS3 protease and a fluorescent protein using a drug.ESCRT-III proteins are fused to the NS3 cleavage site, NS3 protease, and fluorescent protein.Cells transfected with this construct express wild-type ESCRT-III proteins.Upon addition of the drug, a full-length fusion protein accumulates.For imaging and immunoblotting detection, fluorescent proteins (mostly mNeonGreen) and a Flag tag were further added to the NS3 C-terminus, creating CHMP4B/2A/3-NS3-FP-Flag constructs (hereafter collectively referred to as CHMP-NS3-FP).

Figure 7 .
Figure 7. Electron microscopy images of 293FS cells showing VLP arrested in budding at the plasma membrane.(a-e) 293FS cells transfected with Gag and CHMP2A-NS3 and treated for 2 h with either