Lenacapavir binding to immature Gag induces giant virions and causes protease-dependent inhibition of viral release

Significance

Lenacapavir (LEN) is a long-acting therapy that targets the capsid of HIV, yet the mechanisms underlying its effects during the late stages of viral replication remain poorly understood. Here, we showed that LEN binds to precursor Gag during particle formation, promoting the formation of enlarged virus-like particles. We demonstrated that LEN aberrantly enhanced intracellular Gag processing, which may contribute to impaired viral release. In addition, we found that the addition of LEN interfered with p24 detection by ELISA, potentially leading to an underestimation of its antiviral activity. LEN-induced giant particles retained their membrane fusion capability but exhibited defects at postentry stages. These findings provide insights into the mechanisms of capsid-targeting antivirals and may inform the design of next-generation antiviral therapeutics.

Abstract

Lenacapavir (LEN), a potent capsid inhibitor, suppresses reverse transcription and nuclear import by disrupting capsid core formation in HIV type 1 (HIV-1). However, its effects on late-stage viral assembly remain unclear. Although p24 ELISA suggested that LEN substantially inhibited HIV-1, both RT-dPCR and Vpr-HiBiT assays indicated that suppression of viral release required relatively high LEN concentrations. This discrepancy was attributed to LEN-induced reduction in p24 solubility, leading to an overestimation of its inhibitory effect on viral release. In addition, excessive intracellular Gag processing by the viral protease was observed in the presence of LEN, which contributed to impaired viral release. Furthermore, LEN induced numerous abnormal, discrete clusters of Gag at the plasma membrane. Using mass photometry, which was applied to analyze HIV-1 particle sizes, and transmission electron microscopy, we identified heterogeneous LEN-induced viral-like particles (LENiVLPs; diameter >200 nm) containing both Gag and Env proteins. Although LENiVLPs retained membrane fusion capability, they were noninfectious owing to postentry defects that prevented viral replication. Thus, LEN binds to the precursor Gag and promotes aberrant particle formation, offering insights that could guide the development of next-generation LEN-based therapeutics targeting the late stages of the viral life cycle.

In 2022, lenacapavir (LEN; SUNLENCA; formerly GS-6207; Gilead Sciences, Inc.) gained clinical approval as the first-in-class, small-molecule, long-acting HIV type 1 (HIV-1) capsid (CA) inhibitor (1, 2). PF-3450074 (PF74), with a binding pocket similar to LEN (3), was developed previously (4). However, compared with LEN, PF74 exhibits lower antiviral potency and a narrower selectivity window, limiting its therapeutic potential (4). Similarly, GS-CA1, another compound targeting the same site, exhibits picomolar potency and long-acting activity; however, further optimization of this scaffold led to LEN, which demonstrates improved pharmacological properties (5). Bevirimat (BVM, also known as PA-457) is a maturation inhibitor that acts on the six-helix bundle at the CA C-terminal domain (CTD) and spacer peptide 1 (SP1) of the HIV-1 Gag polyprotein (6). Despite its potential, it has not been approved because naturally occurring polymorphisms in the Gag CA-SP1 region, particularly within the SP1 domain, reduce susceptibility to the compound and result in variable clinical responses among patients (7). These limitations have positioned LEN as the first approved antiretroviral drug (ARV) that directly targets the HIV-1 Gag, particularly its CA domain (1, 2). LEN offers the advantage of twice-yearly administration compared with the more frequent dosing schedules required for long-acting cabotegravir and rilpivirine (8), exhibits picomolar-level antiviral activity, does not exhibit cross-resistance with existing ARVs, and the ability to inhibit multiple stages of the HIV-1 life cycle (1).

The HIV-1 CA protein critically influences the entire viral life cycle, making it a promising therapeutic target (9–11). During the late replication phase, Gag and Gag-Pol proteins, which contain the CA domain essential for protein interactions, are expressed (12). HIV-1 protease (PR) cleaves the precursor Gag, thereby releasing CA, which then self-assembles into a mature fullerene-cone-shaped capsid composed of approximately 250 hexamers and 12 pentamers (13–15). This structure is essential for postentry steps (16, 17). Upon fusion with the host membrane, the viral core enters the cytoplasm and undergoes uncoating. This process is regulated by host factors, such as CPSF6 and CypA, through their interaction with CA. These interactions contribute to efficient reverse transcription and subsequent integration of the viral genome (18–20).

LEN binds to the N-terminal domain (NTD)-CTD interface of CA monomers, located within a hydrophobic phenylalanine-glycine (FG)-binding pocket (2). The conserved FG-binding pocket plays an important role in capsid stability and uncoating during cytoplasmic transport (21, 22). LEN reportedly accelerates capsid assembly, leading to aberrant capsid formation and inhibition of infection (3). Additionally, it destabilizes Gag and Gag-Pol, thereby interfering with the precursor Gag assembly and release (1, 2). Although the FG-binding pocket is absent in immature capsids, it remains unclear whether LEN inhibits viral release at this stage of infection.

Although previous studies on the inhibitory effects of LEN have primarily focused on viral maturation and postentry processes (1, 2, 21–24), its effects on viral release remain poorly understood. Moreover, assessments of viral release have largely relied on p24 ELISA assays, limiting the depth of the analysis. In this study, we applied mass photometry, a recently developed label-free single-particle analysis quantitative analysis of the mass profiles of individual particles in solution, together with velocity gradient centrifugation and electron microscopy, to investigate the effects of LEN on the late stage of HIV-1 replication. Through a comprehensive analysis of the released viral particles, we elucidated the mechanism by which LEN inhibits HIV-1 during the late phase of viral replication.

Results

LEN Inhibits HIV-1 Release at High-Concentrations.

We isolated LEN-resistant HIV-1 from patients treated with LEN and identified three critical amino acid substitutions in the CA domain (Q67H, K70R, and T107N). A recombinant virus carrying these substitutions, referred to as the LEN-resistant mutant (LRM), confers resistance to LEN (25). To evaluate the inhibitory effect of LEN on viral release, we measured the Vpr-HiBiT levels in virions from 293T/Vpr-HiBiT cells. During Gag assembly, Vpr-HiBiT is incorporated into progeny virions. Nanoluciferase activity measured via HiBiT and LgBiT complementation indirectly reflects the number of released viral particles (26). Because postentry inhibition by LEN is not expected when 293T/Vpr-HiBiT cells are transfected with full-length HIV-1, LEN was added immediately after transfection. Under the tested conditions, LEN had minimal effect on nanoluciferase activity in either X4-tropic HIV-1 (NL4-3) or R5-tropic HIV-1 (JR-FL), indicating that substantial inhibition occurs only at higher concentrations (Fig. 1 A and B and SI Appendix, Table S1). Similarly, LEN showed limited inhibitory effects on the nanoluciferase activity of transmitted/founder HIV-1 clones in transfected 293T/Vpr-HiBiT cells, with less than a twofold reduction observed at 60 nM (SI Appendix, Table S2).

Fig. 1.

LEN inhibits HIV-1 release under high-concentration conditions. (A–E) Measurement of Vpr-HiBiT amount to assess HIV-1 release from pNL4-3-transfected 293T/Vpr-HiBiT cells (A), pJR-FL-transfected 293T/Vpr-HiBiT cells (B), VSV-G-pseudotyped NL4-3-infected 293T/Vpr-HiBiT cells (C), VSV-G-pseudotyped NL4-3-infected Jurkat/Vpr-HiBiT cells (D), and NL4-3-infected Jurkat/Vpr-HiBiT cells (E). LEN was added immediately after transfection (A and B) or 15 h after infection (C–E). (F) Quantification of vRNA levels in virions from LEN-treated, pNL4-3-transfected 293T cells. LEN was added immediately after transfection. (G–I) Quantification of virion-associated p24 Gag. Virions were harvested from LEN-treated, pNL4-3-transfected 293T/Vpr-HiBiT cells (G and I) or infected primary CD4+ T cells (H). LEN was added immediately after transfection (G and I) or 15 h after infection (H). p24 levels were measured using the p24 ELISA kit (Rimco). (I) Virions were subjected to acidic treatment prior to p24 treatment, and a LEN concentration of 60 nM was used. (J) Quantification of reverse transcriptase (RT) activity in virions from NL4-3-transfected 293T/Vpr-HiBiT cells. LEN was added immediately after transfection. (K) Representative western blot showing Pr55 Gag, p24 Gag, and GAPDH in pNL4-3-transfected 293T/Vpr-HiBiT cells treated with 60 nM LEN. (L–N) Quantification of band intensities in (K) using ImageJ software. The processing efficiency in cells was calculated as the ratio of intracellular p24 to intracellular Gag (Pr55 + p41 + p24). The processing efficiency in the virus was calculated as the ratio of particle-associated p24 to particle-associated Gag (Pr55 + p41 + p24) (M). Virus release efficiency was calculated as the ratio of extracellular Gag (Pr55 + p24) to total Gag (intracellular Gag + extracellular Gag) (N). Data are presented as mean ± SD (A–J). Data in (A–J) are from three independent experiments. (L) shows band intensities quantified from four independent experiments. *IC₅₀ values are shown as mean ± SD with CV (%) and 95% CI are provided in SI Appendix, Table S1.

Furthermore, VSV-G-pseudotyped NL4-3 was used to infect 293T/Vpr-HiBiT or Jurkat/Vpr-HiBiT cells, which exhibited low susceptibility to LEN when treated 15 h postinfection, with IC₅₀ values of 3.4 µM and 2.5 µM, respectively (Fig. 1 C and D and SI Appendix, Table S1). In contrast, VSV-G-pseudotyped NL4-3/LRM exhibited high resistance to LEN. As reported previously (1, 5), LEN inhibits postentry steps; therefore, it was not administered during the 15 h required for viral genome integration. Replication-competent NL4-3 and NL4-3/LRM also exhibited no substantial inhibition at 15 h postinfection, with an IC₅₀ value of 1.3 µM for the former (Fig. 1E and SI Appendix, Table S1). These values were notably higher than the previously reported picomolar IC₅₀ values (1, 5). However, LEN strongly inhibited infection when added at the time of infection, consistent with previously reported postentry antiviral effects (SI Appendix, Fig. S1 A–D and Table S3) (1, 5). LEN also showed no substantial reduction in vRNA copy numbers at lower concentrations, with an IC₅₀ value > 250 nM (Fig. 1F and SI Appendix, Table S1). PF74 and BVM did not block nanoluciferase activity, consistent with previous studies (4, 6) (SI Appendix, Fig. S1 E and F).

To quantify the p24 antigen levels in the supernatants from transfected 293T/Vpr-HiBiT cells and infected primary CD4⁺ T cells, we used two commercially available p24 ELISA kits (27). LEN reduced p24 detection by over 100-fold, with an IC₅₀ of 12 nM and 11 nM in 293T/Vpr-HiBiT and CD4⁺ T cells, respectively (Fig. 1 G and H and SI Appendix, Fig. S2A and Table S1). These values are consistent with previous reports (1) and contradict the Vpr-HiBiT and vRNA measurements (Fig. 1 A–F and SI Appendix, Table S1). LEN binds within the FG-binding pockets of the CA domain, becoming deeply embedded and stabilized by hydrogen bonds. We hypothesized that LEN-induced alterations in Gag structure or aggregation may apparently reduce the detection of p24 to the monoclonal antibody. To validate this, LEN was added directly to the virion lysate following virion collection. This resulted in a dose-dependent decrease in p24 detection from 1 nM to 2 μM LEN treatment (SI Appendix, Fig. S2B), supporting the possibility that LEN-induced Gag conformational changes or aggregation are maintained under mild lysis conditions, resulting in reduced accessibility of the epitope recognized by the commercial ELISA antibody. Subsequently, we explored whether physical treatments could reverse the masking effect. Boiling in SDS buffer did not restore p24 levels, likely due to a 10-fold reduction in anti-p24 antibody reactivity in the absence of LEN (SI Appendix, Fig. S2 C and D). Notably, freezing samples in PBS enhanced detection 10-fold in the presence of LEN (SI Appendix, Fig. S2 C and D). As PBS becomes mildly acidic when frozen (28), the increased acidity may induce conformational changes that render epitopes more accessible to the antibodies. Accordingly, incubating virions under acidic conditions (pH 3.0) restored p24 detection by up to 10-fold in the presence of 60 nM LEN (Fig. 1I and SI Appendix, Fig. S2E). These results indicate that LEN structurally masks the p24 epitope, and this effect can be reversed under acidic conditions, facilitating antibody binding. As masking likely depends on the specific antibody epitopes, previous ELISA-based evaluations of LEN’s effect on viral release may have overestimated its inhibitory activity because of epitope interference. Of note, since the p24 levels did not fully recover under acidic conditions, we could not exclude a potential contribution of LEN to impaired virus release.

To assess whether LEN interferes with Gag-Pol incorporation in virions, the RT activity in virions produced from transfected cells was measured. LEN treatment modestly inhibited RT activity at higher concentrations (Fig. 1J and SI Appendix, Table S1). Consistent with this, LEN reduced the protein levels of RT and Vpr in the cell supernatant at 60 nM (SI Appendix, Fig. S3 A and B), although the reduction was less pronounced than that of p24, in agreement with the Vpr-HiBiT and RT activity results (Fig. 1 A and J). Next, Gag proteins within viral particles were fully denatured using SDS and then quantified by immunoblotting. To avoid epitope masking, a polyclonal antibody (HIV-Ig), an anti-p24 antibody which binds to the N terminus of CA domain and an anti-p17 monoclonal antibody which binds to the HIV-1 Gag matrix (MA) domain were used. In the absence of PR, Pr55 Gag levels in immature particles were not significantly affected by LEN treatment (Fig. 1 K and L and SI Appendix, Fig. S4A). By contrast, in the presence of PR, p24 Gag and p17 MA levels in the virus lysates were significantly reduced by LEN at concentrations of 60 nM and 200 nM, but not at 4 nM, even under acidic conditions (Fig. 1 K–M and SI Appendix, Fig. S4 B–H). In addition, intracellular p24 levels were increased following LEN treatment (Fig. 1 K–M and SI Appendix, Fig. S4 B, C, E, and H). The viral release efficiency, calculated as viral Gag over total Gag, was more than 50% reduced even under 60 nM LEN condition (Fig. 1N). These results indicate that active viral protease is required for both the LEN-induced aberrant Gag processing during virion formation and the LEN-induced reduction in virus release efficiency.

Importantly, the reduced extracellular p24 was not attributable to epitope masking or incomplete solubilization of LEN-induced Gag aggregates, as similar results were obtained using higher SDS concentrations and commercially available high-stringency lysis buffers (SI Appendix, Fig. S5A). Moreover, varying plasmid DNA input did not alter the phenotype at the LEN concentrations primarily used (SI Appendix, Fig. S5B). Consistent with this, intracellular p24 remained detectable after trypsin removal of cell surface-associated virions, suggesting that LEN may lead to abnormal intracellular Gag processing (SI Appendix, Fig. S5C). We next considered the possibility that mature particles, in which CA is detached from the membrane, may be mechanically more fragile than immature particles, in which the membrane is supported by an underlying Gag lattice. We hypothesized that LEN-containing mature particles might collapse during centrifugation, resulting in the release of soluble p24, which would not pellet during centrifugation and thus lead to a reduced level of pellet-associated p24. To evaluate this hypothesis, we used a 10-kDa molecular weight cutoff filter to recover and concentrate soluble p24 from the supernatant after centrifugation. Under these conditions, LEN-treated samples did not show elevated soluble p24 relative to controls (SI Appendix, Fig. S5D). These results indicate that the LEN-associated reduction in p24 observed in the presence of protease is not attributable to physical disruption during centrifugation. Importantly, we observed slight degradation of intracellular Pr55 Gag in 293T/Vpr-HiBiT cells and more pronounced degradation in T cell lines (Jurkat, Jurkat/Vpr-HiBiT, and MT-2 cells) at higher LEN concentrations, consistent with previous reports (23) (SI Appendix, Fig. S5 E–J). In contrast, LEN treatment increased intracellular p24 Gag levels in 293T cells and did not reduce them in T cell lines. Together, these findings suggest that LEN differentially modulates viral protease-mediated Gag processing, promoting intracellular Gag processing in producer cells while reducing the extent of Gag processing within viral particles.

LEN Alters the HIV-1 Gag Distribution.

Unlike the mature CA hexamer, which contains multiple FG-binding pockets essential for LEN binding, the Pr55 Gag lacks these defined pockets (14, 29). Our data demonstrated that LEN did not reduce Pr55 Gag levels in cell lysates, but did inhibit particle release. Notably, p24 levels were higher in cell lysates from LEN-treated cells (Fig. 1N), suggesting the intracellular retention of Gag. Consequently, we hypothesized that LEN can bind to Pr55 Gag in the absence of the canonical FG-binding pockets, as previously reported (29, 30). To investigate this hypothesis, we examined the intracellular Gag distribution in the presence of LEN using the anti-p24 antibody AG3.0 (31). In both transfected and infected cells, LEN treatment led to a distinct punctate pattern of Pr55 Gag localization, in contrast to the diffuse signals observed under control conditions (Fig. 2 A and B and SI Appendix, Fig. S6A). Because LEN treatment could potentially alter antibody recognition of Gag, we evaluated the reactivity of the anti-p24 antibody AG3.0 by flow cytometry. Under these conditions, AG3.0 showed no reduction in antibody reactivity following LEN treatment (SI Appendix, Fig. S6B). This observation suggests that LEN-associated changes in p24 detection may depend on the antibody used and may be affected by differences in epitope accessibility. Notably, LEN had no observable effect on the localization of GagVenus, a Gag protein tagged with yellow fluorescent protein (YFP) at the C-terminus (SI Appendix, Fig. S7A). These results suggest that the C-terminal YFP tag may have interfered with the detection of LEN-induced changes in Gag localization. Furthermore, supporting superresolution imaging using direct stochastic optical reconstruction microscopy (dSTORM) revealed significant Gag–Gag aggregation in LEN-treated cells, suggesting enhanced oligomerization of Gag at the plasma membrane (Fig. 2 C and D). This was confirmed via a Gag multimerization assay using a multilayered sucrose gradient to separate Gag monomers from oligomers and higher-order multimers. In cells transfected with the pNL4-3/GagFlag, control samples showed a peak in fraction 4. However, in LEN-treated samples, the signal in fraction 5 increased, resulting in a broadening of the peak around fraction 4, indicating the accumulation of heavier oligomeric forms. This shift was not observed in the NL4-3/LRM, which exhibited comparable behavior under both the control and LEN-treated conditions. These results suggest that LEN promotes abnormal Gag oligomerization at the plasma membrane (Fig. 2 E and F) by binding to a putative FG-binding pocket-like structure within the Pr55 Gag hexamer, consistent with previous cryoelectron microscopy (cryo-EM) structural studies (29, 30). The molecular mechanism underlying this abnormal Gag oligomerization was elucidated through molecular simulations of a CA-SP1 hexamer with or without LEN, which revealed that LEN binds to immature CA or Pr55 Gag (SI Appendix, Fig. S8A). Moreover, LEN was predicted to cause steric clashes between residues E79 and R82 at the interface of immature CA monomers, potentially disrupting the hexameric CA-SP1 lattice (SI Appendix, Fig. S8B), although the predicted structural change could be overestimated because of a CA-SP1 hexamer but not Gag lattice as simulation models. This disruption was capable of destabilizing the sixfold symmetry of the lattice (Movies S1–S3), altering the particle density and Gag oligomerization. Overall, LEN interacts with immature Gag, destabilizing the CA hexamer structure and promoting abnormal oligomerization at the plasma membrane.

Fig. 2.

LEN induces the abnormal multimerization of Gag at the plasma membrane. (A and B) Representative confocal images of pNL4-3-transfected 293T/Vpr-HiBiT cells (A) or VSV-G/NL4-3-infected Jurkat/Vpr-HiBiT cells (B) treated with LEN. The cells were stained for Gag (green) and nuclei (blue). [Scale bars, 20 μm (A) and 30 μm (B).] (C) Representative dSTORM images of pNL4-3-transfected 293T/Vpr-HiBiT cells treated with LEN. [Scale bars, 5 μm (Left panels) and 1 μm (enlarged panels).] (D) Gag signal intensities were quantified using ImageJ software from 50 regions of interest across 10 cells. Data are presented as mean ± SD from three independent experiments. (E) Western blotting analysis of Gag multimerization in cells treated with LEN using sucrose gradient analysis. (F) Gag band intensities in each fraction were quantified using the ImageJ software. Data are presented as mean ± SD from three independent experiments.

LEN Induces the Formation of Enlarged HIV-1 Particles.

During HIV-1 Gag assembly, approximately 1,500 to 2,000 Gag molecules are assembled at virus-forming regions to produce each progeny virion (32). We hypothesized that LEN disrupts the orderly assembly of Gag at the plasma membrane, thereby affecting the formation and physical properties of viral particles. To directly interrogate the physical properties of individual HIV-1 particles, we employed mass photometry, an emerging single-particle analytical technique based on interferometric scattering (iSCAT), which enables label-free detection of individual particles and quantitative measurement of their scattering contrast (33, 34). Because particles larger than approximately 200 nm are difficult to analyze by mass photometry, this approach primarily captures the population of conventionally sized HIV-1 particles. Under normal conditions, HIV-1 particles typically measure 100 to 130 nm in diameter and contain approximately 1,500 to 2,000 Gag molecules (32, 35). Consistent with these characteristics, mass photometric analysis revealed a homogeneous population of particles with contrast values of approximately 0.3. In contrast, analysis of progeny virions harvested from LEN-treated producer cells revealed a marked reduction in regularly sized viral particles and a leftward shift in the particle density distribution of NL4-3 virions (Fig. 3 A and B). LEN also reduced the number and density of immature virions (SI Appendix, Fig. S9A). This reduction in normal-sized particles was especially pronounced at higher LEN concentrations in both NL4-3 and NL4-3/LRM but was not evident at lower LEN concentrations (SI Appendix, Fig. S9B). Consistent with the lack of effect of LEN on GagVenus localization (SI Appendix, Fig. S7A), LEN also had no significant effect on the abundance of GagVenus particles (SI Appendix, Fig. S9C). Notably, GagVenus particles exhibited a broader distribution of particle densities than normal particles (SI Appendix, Fig. S9C), suggesting that their localization pattern may not accurately reflect that of intact Gag. Consistent with the effects on immature virions, similar changes were observed in VSV-G-pseudotyped NL4-3 virions derived from Jurkat/Vpr-HiBiT cells exposed to increasing concentrations of LEN (SI Appendix, Fig. S10A), as well as in PF74-treated transfected cells (SI Appendix, Fig. S10B). In contrast, BVM treatment did not noticeably alter the particle size or density (SI Appendix, Fig. S10C). Notably, LEN treatment did not reduce the size or density of GagVenus particles at 60 nM (SI Appendix, Fig. S7B). Although the reduction in particle number within Gate A did not fully align with the vRNA copy number and nanoluciferase measurements (Fig. 1 A–F), these results suggest that LEN disrupts the formation of regular-sized viral particles by altering the particle density and reducing the proportion of mature virions. To address the discrepancies between the mass photometry results and Vpr-HiBiT quantification, we analyzed the density and size of particles using a sucrose velocity gradient (5 to 30%) based on Stokes’ principle, which separates particles based on size and density following ultracentrifugation. Under control conditions, both mature (NL4-3) and immature virions [NL4-3/PR(−)] peaked in fraction #7 (Fig. 3C and SI Appendix, Fig. S11A). In contrast, treatment with LEN caused a shift in particle distribution, with a peak observed in fraction #11 (Fig. 3C and SI Appendix, Fig. S11A). Additionally, a higher concentration (200 nM) of LEN significantly reduced the number of particles in fraction #7 by two- to threefold (SI Appendix, Fig. S11B). In contrast, a lower concentration (4 nM) had minimal effects (SI Appendix, Fig. S11C). Similarly, 293T/Vpr-HiBiT cells infected with VSV-G-pseudotyped NL4-3 and NL4-3/LRM exhibited identical results (SI Appendix, Fig. S11D). Predictably, PF74, a compound with similar capsid-targeting properties, also induced a comparable shift (SI Appendix, Fig. S11E), whereas BVM did not (SI Appendix, Fig. S11F). This finding is consistent with the report by Keller et al., in which BVM treatment did not cause major changes in virion morphology or particle size (36). The infectivity of the particles separated by the sucrose gradient was evaluated by examining the virions in each fraction. Among the untreated samples, fractions #6 and #7 contained infectious viruses. However, in the presence of LEN, these fractions lost their infectivity. Notably, fraction #8 from LEN-treated NL4-3/LRM (60 nM) did not retain infectivity (Fig. 3D), suggesting that NL4-3/LRM is resistant at the late stage of the viral life cycle but remains susceptible to loss of infectivity at higher doses of LEN. These findings indicate that LEN significantly alters the physical characteristics of HIV-1 particles, specifically their density and size, likely attributable to increased Gag oligomerization during assembly.

Fig. 3.

LEN alters the size and density of HIV-1 particles. (A and B) Mass photometry analysis of virion size (nm) and density (contrast; arbitrary units) in the presence (orange) or absence of LEN (blue). The number of particles and density (contrast; arbitrary units) within gate A, as shown in (A), were quantified in (B). (C) Sucrose velocity gradient analysis of HIV-1 particles released from pNL4-3-transfected 293T/Vpr-HiBiT cells treated with LEN, quantified using Vpr-HiBiT activity. (D) Infectivity of virions purified from the sucrose fractions shown in (C), assessed by firefly luciferase activity in TZM-bl cells after normalization to Vpr-HiBiT activity. Data are presented as the mean ± SD from three independent experiments (A–D). (E and F). Representative transmission electron microscopy (TEM) images of virions produced from pNL4-3-transfected 293T/Vpr-HiBiT cells in the presence or absence of LEN are shown. The colored dashed boxes indicate the enlarged regions. [Scale bars, 0.5 μm (main panels) and 0.2 μm (Insets).] (G) Quantification of virion size and classification (mature, black; immature, dark gray; aberrant, light gray) from the TEM images shown in (E and F), was performed using ImageJ. n indicates the number of virions analyzed. Statistical analysis was performed using one-way ANOVA with Sidak’s multiple-comparisons test. (H) Representative TEM images of NL4-3/PR(−) particles produced from pNL4-3/PR(−)-transfected 293T/Vpr–HiBiT cells in the presence or absence of LEN. The colored dashed boxes indicate the enlarged regions. [Scale bars, 0.5 μm (main panels) and 0.2 μm (Insets).] (I–K) Quantification of radial density profiles (I), particle size (J), and the outer leaflet (OL)–inner leaflet (IL) distance (K) from the TEM images shown in (H), performed using ImageJ. Statistical analysis was performed using one-way ANOVA with Sidak’s multiple-comparisons test.

To clarify the presence of particles containing Vpr-HiBiT and vRNA in the heavy fraction, we used TEM to visualize the morphology of HIV-1 particles released from 293T and T cells. Mature HIV-1 particles are characterized by a dark central density, corresponding to the mature capsid core (37). Owing to sample sectioning during TEM preparation, capsids may not consistently appear conical or fullerene-shaped; particles can still be reliably classified as mature, immature, or abnormal (SI Appendix, Fig. S12A and Table S4). In the presence of LEN, NL4-3 particles exhibited high variability in morphology and size, with many particles exceeding 200 nm in diameter (Fig. 3 E and G). The increase in the number of aberrant particles induced by LEN is consistent with previous reports (23). This effect was also observed in NL4-3/PR(−) particles (Fig. 3 H–J), making them undetectable by mass photometry, as it exceeds its upper size detection limit. A similar particle size heterogeneity was observed in 293T/Vpr-HiBiT and Jurkat/Vpr-HiBiT cells infected with VSV-G-pseudotyped NL4-3 and NL4-3/LRM (SI Appendix, Fig. S12 B–D). Immunogold electron microscopy revealed specific Gag labeling within these aberrantly enlarged particles (SI Appendix, Fig. S12E). Scanning electron microscopy (SEM) revealed heterogeneity among virions attached to the cell surface (SI Appendix, Fig. S12F). These abnormally large, irregular particles are referred to as LEN-induced virus-like particles (LENiVLPs). Notably, in NL4-3/PR(−) particles treated with LEN, LENiVLPs appeared more spherical with a decreased radial distance between the Gag and membranes of the viral envelope (Fig. 3 H, I, and K). Overall, these results suggest that LEN interacts with immature Gag and promotes the formation of abnormally large LENiVLPs, likely by altering the structural organization and radial spacing of the viral membrane layers.

LEN Inhibits Viral Infectivity.

LEN inhibits HIV-1 infectivity (1, 2). To assess the infectivity of LENiVLPs, virions were harvested 48 h posttransfection from 293T/Vpr-HiBiT cells and normalized based on Vpr-HiBiT levels. They were subsequently used to infect TZM-bl, CEM-GFP, and Jurkat cells. Viral infectivity in TZM-bl cells was assessed by measuring HIV-1 Tat-driven firefly luciferase activity at 48 h postinfection. LEN inhibited the infectivity of NL4-3 with an IC₅₀ that was 2.5 times lower than that of NL4-3/LRM, indicating increased sensitivity of NL4-3 to LEN (Fig. 4A and SI Appendix, Table S5). PF74 demonstrated similar infectivity inhibition at 5 μM, consistent with previous reports (4) (SI Appendix, Fig. S13A). BVM exhibited similar effects (SI Appendix, Fig. S13B). Additionally, when NL4-3 produced from LEN-treated 293T/Vpr-HiBiT cells was used to infect CEM-GFP cells, the IC₅₀ was 3.1-fold lower than that of NL4-3/LRM (Fig. 4B and SI Appendix, Table S5). Similarly, infection of Jurkat cells resulted in a 2.4-fold reduction in IC₅₀ (Fig. 4C and SI Appendix, Fig. S13 C and D and Table S5). Moreover, virions produced from LEN-treated infected-T cells exhibited a 55-fold reduction in infectivity compared to NL4-3/LRM when tested in TZM-bl cells (Fig. 4D and SI Appendix, Table S5). A similar pattern was observed in the virions harvested from primary CD4⁺ T cells (Fig. 4E and SI Appendix, Table S5), confirming the broad inhibitory effects of LEN on various T cell types. Despite this inhibition, LENiVLPs successfully incorporated envelope proteins during the late-stage assembly in the presence of LEN (Fig. 4 F and G and SI Appendix, Fig. S13E). These particles maintained an effective membrane fusion capacity (Fig. 4H), suggesting that LEN inhibits infectivity at the postentry step, specifically by blocking viral uncoating and/or the nuclear import of reverse transcription complexes.

Fig. 4.

LEN inhibits HIV-1 infectivity at a postentry step. (A) Infectivity of HIV-1 virions produced from pNL4-3-transfected 293T/Vpr-HiBiT cells treated with LEN, assessed by Tat-driven firefly luciferase activity in TZM-bl cells. (B) Infectivity of virions from pNL4-3-transfected 293T/Vpr-HiBiT cells, assessed in CEM-GFP cells by flow cytometry based on the percentage of GFP-positive cells. (C) Infectivity of virions from pNL4-3-transfected 293T/Vpr-HiBiT cells, assessed in Jurkat cells by flow cytometry based on intracellular p24-staining. (D and E) Infectivity of HIV-1 virions produced from pNL4-3-infected Jurkat/Vpr-HiBiT cells (D) and primary CD4+ T cells (E) treated with LEN, assessed by Tat-driven firefly luciferase activity in TZM-bl cells. Data are presented as mean ± SD from three independent experiments (biological triplicates where applicable) (A–E). (F) Representative western blot analysis of gp120 incorporation into HIV-1 virions produced in the presence or absence of LEN. (G) Quantification of gp120 incorporation normalized to p24 levels, based on ImageJ analysis. Data are presented as mean ± SD from four independent experiments. (H) HIV-1 fusion assessed using the BlaM-Vpr reporter assay by flow cytometric detection of CCF4 cleavage in the presence of LEN. Data are presented as mean ± SD from three independent experiments. *IC₅₀ values are shown as mean ± SD with CV (%) and 95% CI are provided in SI Appendix, Table S5.

Discussion

LEN, a potent capsid inhibitor, binds to the FG-binding pockets located at the mature interfaces between p24 capsid subunits (2), thereby inhibiting nuclear import, virion assembly, and core formation (1, 2). However, its role in the late-stage of HIV-1 life cycle remains poorly understood. Our findings provide important insights into the effects of LEN on viral assembly, release, and infection (Fig. 5). High concentrations of LEN significantly inhibited viral release in both 293T and T cells, reducing HIV-1 vRNA copy number, RT activity, and Vpr-HiBiT release, while p24 levels were significantly decreased even at lower concentrations. This is thought to result from LEN altering p24 antibody recognition in the presence of nonionic detergents, thereby inhibiting antibody binding (SI Appendix, Fig. S2B). This interpretation is consistent with the partial recovery of p24 antigen detection after freezing and the possibility that protonation of histidine residues under acidic conditions induces transitions in capsid structure (38). However, although altered antibody recognition contributes to the reduced extracellular p24 signal, it cannot fully account for the marked reduction in extracellular p24 levels. Indeed, p17 and p24 signals remained reduced even under strongly denaturing SDS conditions (SI Appendix, Fig. S5A), and intracellular p24 levels were not reduced by LEN treatment. Furthermore, high concentrations of LEN reduced the release of viral RNA, RT activity, Vpr-HiBiT, and Gag, indicating that LEN not only affects p24 detection but also impairs particle assembly and release. These findings support a model in which LEN-induced aberrant Gag assembly promotes aberrant viral protease-mediated Gag cleavage within producer cells, thereby impairing particle assembly and reducing virion release. Notably, p17 and p24 were selectively reduced in LENiVLPs, whereas Vpr-HiBiT and viral RNA were relatively preserved. These observations suggest that LENiVLPs differ in molecular composition from normal virions, consistent with altered Gag processing during particle assembly. Such compositional changes may contribute to defective virion maturation after release.

Fig. 5.

LEN induces abnormal Gag multimerization, leading to the formation of LENiVLPs. Gag is abnormally assembled in the presence of LEN, leading to the formation of LENiVLPs. The epitopes of these LENiVLPs are masked by LEN hence p24 amount are underestimated. The infectivity of LENiVLPs is lost due to postentry defects, even though envelope glycoproteins are incorporated.

Previous studies reported that GS-CA1 (5) and LEN (1) reduced intracellular levels of Pr55 and p24, suggesting that binding of these compounds to Gag and Gag-Pol may compromise the stability, trafficking, and/or assembly of CA precursor proteins. In contrast, we detected no reduction in intracellular p24 over the concentration range tested (2 nM to 2 µM) in either 293T or T cell lines (SI Appendix, Fig. S5 E–J), whereas intracellular Pr55 levels were reduced only at high LEN concentrations in a cell type–dependent manner (SI Appendix, Fig. S5 E–J), which is consistent with a previous report (23). A recent study by Ricaña et al. (30) demonstrated that the apparent effect of LEN on intracellular Gag abundance is highly dependent on the protein extraction method. Specifically, they showed that extraction with SDS resulted in an apparent reduction in intracellular Pr55 at high LEN concentrations, whereas extraction with 8 M urea largely restored Pr55 recovery. In our study, cells were first lysed with 1% Triton X-100, followed by solubilization of the remaining insoluble fraction with Laemmli SDS sample buffer. These differences in protein extraction procedures may partly account for the discrepancies in intracellular Gag levels among different studies. In addition, we found that different anti-Gag antibodies showed different sensitivities for detecting Pr55 and p24 (SI Appendix, Fig. S5 E–G). Notably, HIV-Ig appeared to be less affected by the reduced detection sensitivity caused by aberrant Gag aggregation induced by high concentrations of LEN. Despite using the same protein solubilization conditions and detecting Gag with HIV-Ig, T cell lines showed a reduction in Pr55 levels and only a minimal increase in p24 under 200 nM LEN treatment (SI Appendix, Fig. S5 E, H, and J). These cell type–dependent differences may reflect differences in the extent of intracellular Gag cleavage and/or in the LEN concentration required to induce abnormal Gag aggregation. Taken together, accurate evaluation of LEN-induced changes in intracellular Gag abundance will require further studies that consider the conditions used for protein solubilization, the anti-Gag antibodies used for detection, and the cellular context.

Confocal and dSTORM imaging revealed enlarged Gag puncta with LEN, indicating that LEN alters Gag distribution and promotes its aggregation at the plasma membrane, consistent with the previous reports (30). This supports the ability of LEN to interact with immature Gag and induce oligomerization even in the absence of FG-binding pockets. On the other hand, Gag structures surrounding the FG-binding pocket may influence the architecture of the Gag hexamer during particle assembly, thereby promoting aberrant accumulation of Gag. Thus, mutations in the FG-binding pocket could potentially affect particle morphology or the efficiency of viral release. In other words, it is conceivable that resistance mutations are selected that either prevent LEN binding or allow LEN binding but do not perturb particle assembly. With regard to GagVenus, large puncta were present at the plasma membrane even without LEN, and their particle number remained unchanged upon LEN treatment (SI Appendix, Fig. S7B). The reason why LEN does not induce oversized particles in GagVenus is unclear. One possible explanation is that the addition of Venus to the CTD increases the size of Gag by 1.5-fold, which may offset the effects of LEN on CA structural changes. It is also possible that LEN binds less efficiently to GagVenus, warranting further investigation.

We identified giant viral particles, termed LENiVLPs, with aberrant capsid structures in both the transfected and infected cells. LENiVLPs appear to incorporate Env proteins during late-stage assembly. The incorporated Envs remained functional and mediated efficient membrane fusion (Fig. 4 F–H). This suggests that Gag within LENiVLPs is likely cleaved by viral proteases and forms an aberrant core, consistent with a previous report (30). Despite their enlarged size and aberrant cores, LENiVLPs efficiently attach to and fuse with target cells, indicating that particle size and core morphology do not substantially affect the early steps. However, the formation of an aberrant core may impair nuclear import of the core and hinder reverse transcription. These findings further support the inference that LEN inhibits the postentry stage of the viral replication cycle.

We propose that LEN destabilizes the CA-SP1 hexameric structure (SI Appendix, Figs. S8B and S9), resulting in abnormal Gag accumulation, the formation of LENiVLPs, and consequent changes in particle density (Figs. 3C and 5). We observed a reduced radial distance between the Gag layer (appearing as the IL) and the plasma membrane (appearing as the OL) in LEN-treated samples (Fig. 3K). This could be attributed to the structural collapse or modification of the MA domain, which anchors Gag to the plasma membrane. Alternatively, Gag multimers may adopt a flatter conformation, leading to decreased spacing between the OLs and ILs. Additional structural studies, such as cryo-EM, are necessary to conclusively determine this. While performing cryo-EM, mature LENiVLPs should not be dismissed as a rare species, given their diverse and abnormal morphologies.

Overall, these results expand our current understanding of LEN’s antiviral mechanism by demonstrating its ability to target both mature and immature Gag assemblies, as revealed by advanced virus analysis techniques that incorporate cutting-edge technologies. This dual-phase inhibition, through the disruption of precursor Gag assembly and proper uncoating, presents opportunities for refining capsid-targeting therapeutics.

Materials and Methods

Experimental details and methods can be found in SI Appendix, including sources of cell lines and reagents, generation of 293T/Vpr-HiBiT and Jurkat/Vpr-HibiT cell lines, and procedures for virus production and infection. Details of Viral release and infectivity, western blotting analysis, sucrose velocity gradients, Gag multimerization assay, confocal microscopy, dSTORM, TEM/SEM, Immunogold electron microscopy, FACS, viral fusion assay, mass photometry analysis, and computational molecular dynamics simulation are also described in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

CA-SP1 hexamerization in the absence of LEN (wo LEN). All-atom molecular dynamics simulation of the HIV-1 CA-SP1 hexamer in the absence of LEN. The structural model was generated by homology modeling using PDB 7ASL, 6V2F, and 6BHR in MODELLER (version 9v8). Simulations were performed using the FF14SB and GAFF force fields.

Movie S2.

LEN binding before CA-SP1 hexamerization destabilizes the assembly. All-atom molecular dynamics simulation of the HIV-1 CA-SP1 hexamer with LEN introduced prior to complete hexamerization. The structural model is shown above the RMSD trajectory. The model was generated by homology modeling using PDB 7ASL, 6V2F, and 6BHR in MODELLER (version 9v8). Simulations were performed using the FF14SB and GAFF force fields.

Movie S3.

LEN binding after CA-SP1 hexamerization alters capsid stability. All-atom molecular dynamics simulation of the HIV-1 CA-SP1 hexamer with LEN bound after CA-SP1 hexamer formation. The structural model was generated by homology modeling using PDB 7ASL, 6V2F, and 6BHR in MODELLER (version 9v8). Simulations were performed using the FF14SB and GAFF force fields.

Data, Materials, and Software Availability

Study data are included in the article and/or supporting information.