C-terminally biotinylated CAI peptide (ITFEDLLDYYPGGGSK-biotin) and unmodified CAI were synthesized by GL Biochem (Shanghai) Ltd. The Discovery Probe FDA-approved Drug Library was purchased from the National Compound Resource Center (Shanghai, China), whereas HTRF detection agents and white 384-well plates were purchased from PerkinElmer (Boston, MA, United States). Black 96-well plates were sourced from Greiner Bio-One (Darmstadt, Germany) and streptavidin-coated biosensors were purchased from Sartorius AG (Goettingen, Germany). Affinity chromatography resin for the purification of His-tagged and GST-tagged proteins was procured from Changzhou Smart-Lifesciences Biotechnology Co. Ltd. (Changzhou, China), while other reagents were obtained from Amresco (Solon, USA). Sennoside A (purity ≥99%) and sennoside B (purity ≥99) were commercially available from the supplier MedChem Express (Shanghai, China).

DNA encoding the CA CTD was cloned into BamHI-NotI sites of prokaryotic expression vector pGEX-4 T-1 with a N-terminal glutathione S-transferase (GST) tag. Recombinant CA CTD protein was expressed in E. coli BL21(DE3) and purified as previously described (Thenin-Houssier et al., 2016).

DNA encoding the wild type (WT) CA was cloned into Nde I-BamHI sites of prokaryotic expression vector pET11a. Expression and purification of WT CA proteins were performed using the same procedure as previously described (Lanman et al., 2002).

The plasmid expressing a quadruple mutant (A14C/E45C/W184A/M185A) CA (4Mu CA) was constructed by introducing the quadruple mutations to CA gene in plasmid for expression of prokaryotic WT CA. Expression of WT CA proteins were followed the same scheme as WT CA. The purification protocol for 4Mu CA followed the same scheme as WT CA, except that 200 mM β-mercaptoethanol was included in all buffers and ammonium sulfate fractionation was performed at 30% saturation.

After purification, CA proteins were dialyzed overnight against 20 mM Tris (pH 8.0) containing 5 mM β-ME. Subsequently, the purified proteins were concentrated to approximately 30 mg/mL, divided into aliquots, flash frozen in liquid nitrogen, and stored at −70°C until further analysis.

The Discovery Probe FDA-approved Drug Library comprised 1,971 compounds, which were available as stock solutions at 10 mM concentrations in DMSO. Primary screening of these compounds was conducted using a previously established HTRF assay (Zhang et al., 2019). Briefly, compound (1 μL), CAI-biotin peptide (2 μL) and GST-CA CTD (2 μL) were dispensed into 384-well microplates. After 30 min incubation at room temperature, 5 μL premixed HTRF agents were added. After an additional 1 h incubation at room temperature, the HTRF signal was acquired by reading the plates in a PerkinElmer Envision Multilabel Plate Reader. Data were analyzed and visualized using GraphPad Prism (version 5.0). To ensure the reproducibility of the results, duplicate assays were conducted for all drugs. Quality control measures were implemented, and only plates with a Z’-factor greater than 0.5 were included in the subsequent analysis. The quality of the assay was determined by calculating the Z’-factor using the following equation: Z’ = 1 - (3 × SD_max + 3 × SD_min) / |μ_max - μ_min|, where SD_max and SD_min represent the standard deviations of the positive and negative control measurements, respectively, and μ_max and μ_min represent the means of the respective positive and negative signal controls (Zhang et al., 1999).

To monitor the in vitro assembly of HIV-1 CA in the presence and absence of hit compounds, changes in sample absorbance over time at 350 nm were measured, as previously described (Lanman et al., 2002; Tang et al., 2003; Xu et al., 2018). To initiate the assay, 1 μL of 5 mM SA or SB was added to 74 μL of buffer composed of a mixture of 5 M NaCl and 200 mM NaH₂PO₄ (pH 8.0) in a ratio of 2:1 by volume. Subsequently, 25 μL of CA (120 μM) was added to each well and DMSO served as a vehicle control. After a 2-min equilibration period, sample absorbance values were measured at 350 nm every 60 s for 39 min using a Perkin Elmer Envision microplate reader and were corrected by subtracting the absorbance values of a blank sample without sodium chloride. The resulting data were analyzed using GraphPad Prism 5.0.

To obtain HIV-1 hexameric CA, the recombinant 4Mu CA was used, which have been shown to promote stable hexameric unit formation in a previous study (Pornillos et al., 2009, 2010). Briefly, Crosslinked CA hexamers were prepared by sequential dialysis of 30 mg/mL 4Mu CA into assembly buffer (NaH₂PO₄-Na₂H₂PO₄, pH8.0, 2.5 M NaCl) containing 200 mM β-ME, assembly buffer with 0.2 mM β-ME, and, finally, 20 mM Tris (pH 8.0). Each dialysis step was performed at 4°C, for at least 8 h. Biotinylation of both wild-type (WT) and mutant HIV-1 CA monomers as well as the 4Mu CA hexamer, was achieved using EZ-Link Sulfo-NHS-biotin reagents, according to the manufacturer’s instructions (Thermo Fisher).

To confirm the affinity of the compound for HIV-1 CA, biolayer interferometry (BLI) experiments were conducted using an Octet96 system at 25°C. Streptavidin biosensor (SA) sensors were immersed in phosphate-buffered saline (pH 7.5) with Tween 20 (0.02%) and DMSO (1%) for at least 10 min prior to the run. Biotinylated proteins were loaded onto the tips of SA sensors. The double reference subtracted method was conducted, as previously described, to determine three affinity constants: k_on, k_off, and K_D (Zhang et al., 2023).

Four crystal structures, including CAI-bound CA CTD (PDB ID:2BUO), native CA hexamer (4XFX), PF74 bound native CA hexamer (4XFZ), and GS-6207 bound CA_{A14C/E45C/W184A/M185A} hexamer (6VKV) were used for molecular docking. All molecular simulations were performed using the SwissDock on-line server (Grosdidier et al., 2011a,b), and the results were visualized using PyMOL. ¹

Anti-HIV-1 activity was determined by an assay based on virus-induced cytopathic effects (CPE) (Zhang et al., 2020). Briefly, C8166 cells were infected with HIV-1 at a multiplicity of infection (M.O.I) of 0.03, with DMSO or diluted SA or SB in 96-well plates. After 72 h, the CPE was determined by light microscopy and the 50% effective concentrations (EC₅₀) were calculated. The cytotoxicity of the compounds was determined by a tetrazolium dye (MTT) based assay. C8166 cells were co-incubated with diluted SA or SB in 96-well plates. After 3 days, cell viability was determined using MTT, and the 50% cytotoxicity concentration (CC₅₀) was calculated. The selectivity index (SI) was calculated using the formula: SI = CC₅₀/EC₅₀, which indicates the magnitude between cytotoxic and effective concentrations of the compound. Tenofovir disoproxil fumarate (TDF) was used as a positive control.

Data are presented as means ± SEM. Paired-test was used for statistical significance determination using the GraphPad Prism 5.0. Significance was set at p < 0.05.

Our study does not include any human research participants and animals.

HTRF is a highly effective platform for conducting high-throughput screening (HTS) (Rectenwald et al., 2019). We employed this method to create an assay to identify small-molecule drugs that could disrupt the interaction between peptide CAI and CA CTD (Zhang et al., 2019). The screening approach and workflow are shown in Figure 2A. We screened a library of 1,971 small molecules, all of which were FDA-approved. Each compound was tested in duplicate at a concentration of 50 μM. To evaluate the quality of the screening process, we used the Z′-factor based on the signals between the 1.0% DMSO (negative control) and 50 μM CAI-treated groups. The average Z-factor for all screenings was 0.84, indicating that the results were reliable (Figure 2B). Screening of 1,971 compounds also demonstrated high reproducibility, with a correlation coefficient (R) of 0.9992 (Figure 2C). Eight hits with over 70% inhibition (Figure 2D) were identified including biotin, an assay interference compound, which was excluded from further investigation. The remaining compounds were further validated through dose–response analysis using the HTRF assay.

Figure 3A and 1.62 μM (Figure 3G), respectively. These results indicate that both compounds bind directly to CA and compete with CAI for binding to the CA CTD. Therefore, based on these findings, SA and SB were chosen as the preferred candidates for subsequent analyses in further experiments.

Seven hit compounds were repurchased and subjected to a 12-point two-fold serial dilution, from 50 to 0.024 μM. Of these seven compounds, five were found to be effective in dose–response assays (Figure 2D). Specifically, two of the compounds, namely sennoside A (SA) and sennoside B (SB), exhibited strong inhibitory activity with IC₅₀ values of 0.18 μM (95% confidence interval [CI], 0.15 μM–0.23 μM) (Figure 2E) and 0.08 μM (95% confidence interval [CI], 0.06 μM–0.12 μM) (Figure 2F), respectively. Their chemical structures are shown in Figures 2G,H. To further confirm their affinity for CA, both SA and SB were analyzed using the BLI assay. This assay was conducted using three different concentrations of the compounds to determine their binding to monomeric CA. The results showed that both compounds interacted directly with the CA monomer with a K_D value of 2.34 μM (Figure 3A) and 1.62 μM (Figure 3G), respectively. These results indicate that both compounds bind directly to CA and compete with CAI for binding to the CA CTD. Therefore, based on these findings, SA and SB were chosen as the preferred candidates for subsequent analyses in further experiments.

To identify the binding site of SA and SB on the CA protein, several key residues within the CA protein, including Y169, N183, E187, and L211, which are critical for the binding of CAI to the CA CTD, were selected for mutagenesis experiments. We introduced the mutations Y169A, N183A, E187A, and L211A/S into each plasmid and purified the mutant proteins from E. coli (Bartonova et al., 2008). Subsequently, we used BLI to determine the equilibrium dissociation (K_D), association rate (K_on), and dissociation rate (K_off) constants for the binding of SA and SB to the mutant CA monomers. For SA, the Y169A, E187A, and L211S substitutions affected both the K_on and K_off values of SA, resulting in an approximately 3- to 3.7-fold decrease in the binding affinity of the two natural products to the CA monomer (Figures 3A–C,M). The L211A substitution resulted in a 1.35-fold reduction in the K_D value, primarily because of a similar reduction in K_off, whereas the K_on values of SA association for L211A and WT CA showed a small difference (Figures 3A,D,M). In contrast, the N183A mutation led to a slight decrease in the K_D value with reduced levels of K_off, although the K_on values of SA association for N183A and WT CA were similar. In the case of SB, all five substitutions affected both the K_on and K_off constants of the compound, leading to a 1.72- to 4.17-fold decrease in the binding affinity to the CA monomer (Figures 3G–M). The E187A substitution had the most pronounced effect on the SB equilibrium dissociation (K_D), compared to other substitutions, such as Y169A, L211A, L211S, and N183A substitutions had the most pronounced effect on SB equilibrium dissociation (K_D). Taken together, these results provide compelling evidence that SA and SB bind directly to the CA protein as direct competitors of the CAI-CA CTD interaction.

Under high ionic strength conditions, recombinant CA can spontaneously oligomerize and form open-ended helical tubes consisting of repeating CA hexamers. To investigate the impact of SA and SB on CA tube assembly, we monitored in vitro CA assembly in the presence and absence of SA and SB by measuring changes in light absorbance over time in 2.5 M NaCl. CA (25 μM) for CA. In contrast to the CAI peptide (positive control), which inhibited CA assembly in vitro, both SA and SB increased in vitro CA assembly in a dose-dependent manner (Figures 4A,B). We used BLI to determine the K_D values of SA and SB in the HIV-1 CA hexamer. Figures 4C,D presented the results, which indicated that they bound firmly to the hexameric CA, with K_D values of 1.75 μM and 2.25 μM, respectively. These findings suggest that SA or SB increases capsid in vitro assembly by binding to the CA hexamer.

To better understand how SA and SB interact with the CA hexamer, we employed induced-fit docking using Swiss-Dock. Figure 5 shows the docked poses of SA or SB within the crystal structure of the CA_{A14C/E45C/W184A/M185A} hexamer. Our results showed that SA (Figure 5A) and SB (Figure 5C) were bound to the pocket formed by two neighboring CA subunits. SA established a hydrogen-bonding network with residues N57, V59, Q63, K70, and N74 of CA1-NTD and Q179 of CA2-CTD (Figure 5B). SB forms hydrogen bonds with the N53, N70, and N74 residues of CA1-NTD, along with the A177 and Q179 residues of CA2-CTD (Figure 5D).

As mentioned previously, this pocket is the binding site for two host cell proteins, CPSF6 and NUP153 (Price et al., 2014). As such, if SA/SB interacts within this pocket it should inhibit the interaction of peptides derived from CPSF6 and NUP153 with hexameric CA. Therefore, we synthesized two peptides CPSF6_313–327 (PVLFPGQPFGQPPLG) and NUP153_1407–1423 (TNNSPSGVFTFGANSST), both biotinylated at the N-terminus to allow capture on a streptavidin coated sensor chip. Both CPSF6 and NUP153 peptides have been previously demonstrated to interact with hexameric CA. We then performed a BLI-based competition assay. CA hexamer at a concentration of 5 μM, either alone or in combination with high concentrations of SA or SB or PF-74, were used to bind to the SA biosensor loaded with biotinylated CPSF6_313–327 or NUP153_1407–1423. As Figure 6 showed, PF-74 inhibited, to a degree, the CA hexamer/CPSF6 peptide and the CA hexamer/NUP153 peptide interaction. Our newly discovered compound SA or SB, also inhibited the CA hexamer/CPSF6 peptide and the CA hexamer/NUP153 peptide interaction indicating that it is binding in the pocket formed by two neighboring CA subunits.

To further evaluate the antiviral effects of SA and SB, we tested them for protecting C8166 cells from HIV-1-induced CPE. The cytotoxic effect of these compounds on C8166 cells was also assessed. We observed that, SA and SB showed weak inhibitory activity against HIV-1 induced CPE with IC₅₀ values of 2.27 μM and 4.73 μM respectively, while the CC₅₀ of SA and SB are 22.63 μM and 33.41 μM. The selectivity index of both compounds with antiviral activity was less than 10.