Making HIV its own worst enemy
Resistance to existing antivirals drives the need for identifying new ways to target HIV. Wang et al. screened a library of peptides from the HIV envelope protein to find ones that could inhibit the virus. One peptide, F9170, targeted HIV gp41 and was effective against free virions or HIV-infected cells. This peptide did not seem to be toxic or immunogenic in mice and is small enough to penetrate tissues that harbor HIV reservoirs. Simian-HIV–infected macaques treated with a short course of F9170 experienced a substantial drop in viral loads. Although further testing and development are needed, it could one day be part of a new treatment regimen for HIV.
Abstract
HIV-associated morbidity and mortality have markedly declined because of combinational antiretroviral therapy, but HIV readily mutates to develop drug resistance. Developing antivirals against previously undefined targets is essential to treat existing drug-resistant HIV strains. Some peptides derived from HIV-1 envelope glycoprotein (Env, gp120-gp41) have been shown to be effective in inhibiting HIV-1 infection. Therefore, we screened a peptide library from HIV-1 Env and identified a peptide from the cytoplasmic region, designated F9170, able to effectively inactivate HIV-1 virions and induce necrosis of HIV-1–infected cells, and reactivated latently infected cells. F9170 specifically targeted the conserved cytoplasmic tail of HIV-1 Env and effectively disrupted the integrity of the viral membrane. Short-term monoadministration of F9170 controlled viral loads to below the limit of detection in chronically SHIV-infected macaques. F9170 can enter the brain and lymph nodes, anatomic reservoirs for HIV latency. Therefore, F9170 shows promise as a drug candidate for HIV treatment.
INTRODUCTION
Throughout the course of the HIV epidemic, the virus has infected 74.9 million individuals and caused 37.9 million deaths globally (1). With the application of current antiretroviral therapy (ART), HIV infection is no longer a fatal disease. So far, four categories of antiretroviral drugs (ARDs) have been used in clinics, including nucleoside/nonnucleoside reverse transcriptase inhibitors (NRTIs/NNRTIs), protease inhibitors, integrase inhibitors (IIs), and entry inhibitors (EIs). RTIs, protease inhibitors, and IIs must enter the host cell to inhibit HIV replication, whereas EIs must act on the cell surface where the virus binds, making them ineffective against cell-free virions. Moreover, because of error-prone HIV replication, drug-resistant HIV isolates were identified for most ARDs within several weeks to several years after drug treatment (2–4). According to World Health Organization (WHO)’s HIV drug resistance report 2019, the proportion of pretreatment drug resistance to NNRTI among adults initiating first-line ART exceeded 10%, and for these patients, WHO recommends moving away from NNRTI-based ART (5). Besides, although ART can establish long-term control, it requires daily administration because it does not eradicate HIV type 1 (HIV-1) infection (6).
Recently, to overcome the limitation of classic ARDs, novel antivirals, such as antibodies, are being intensely researched for HIV. A wide array of broadly neutralizing antibodies (bNAbs) against HIV has been isolated, and such bNAbs are potential alternatives to ART. These bNAbs can effectively neutralize HIV virions by binding to epitopes on the HIV-1 envelope glycoprotein (Env), mainly including the CD4-binding site, N-linked glycans close to the third variable loop (V3), membrane-proximal external region (MPER) domain, trimer apex, and glycoprotein 120 (gp120)–gp41 interface. These anti-HIV bNAbs are more effective and better tolerated than some classic ARDs. However, treating viremic individuals with a single antibody still leads to resistance and viral escape (7, 8), whereas combining more than two types of bNAbs will definitely increase the cost. Given that HIV is more prevalent in developing countries (9), antibody-based drugs that will cause a heavy financial burden are unsuitable for widespread application. Therefore, affordable drugs that bind to still undiscovered targets will be desirable supplements to overcome the resistance to current ARDs. Moreover, to entirely solve drug resistance, we need to take further steps in identifying targets and drugs to completely eradicate HIV.
It is widely accepted that HIV reservoirs, including cell and tissue reservoirs, have remained the major barrier against an HIV cure (10). Latently HIV-infected cells are quiescent, but they can still undergo homeostatic proliferation, resulting in the long-term persistence of HIV. A “shock and kill” strategy has been proposed to eradicate HIV reservoirs by incorporating ARDs with a latency-reversing agent (LRA). It was reported that combining three bNAbs with LRAs reduced the reservoir size of HIV-1 in humanized mice (11). Part of the latent HIV reservoir could be eradicated because LRA-reactivated latent cells were able to produce new HIV-1 particles and express HIV-1 Env that could then be killed by Env-specific bNAbs via antibody-dependent cell-mediated cytotoxicity (ADCC), suggesting that HIV-1 Env is a feasible target for HIV cure. Although shock and kill strategy has been widely investigated in preclinical and clinical studies, no report has shown that this strategy could successfully eradicate HIV reservoir in a clinical study.
HIV-1 Env, which consists of receptor-binding subunit gp120 and transmembrane subunit gp41 (Fig. 1A), contains several classic and popular targets of ARDs and bNAbs, such as the CD4-binding site in gp120 (12), fusion peptide (FP) domain (13), six-helix bundle (6-HB) formation between the N- and C-terminal heptad repeats (NHR and CHR, respectively) (14, 15), and the MPER (16). We and others have reported that peptides derived from the HIV-1 gp41 CHR domain can interact with the NHR domain to block viral 6-HB formation, thus inhibiting HIV-1 fusion and entry into the target cell (14, 17–19). In this study, we screened a 15-mer peptide library derived from HIV-1 Env and identified a peptide that is effective in inhibiting HIV-1 infection.
(A) Schematic diagram of HIV-1 envelope glycoprotein (Env) and sequences of group M Env 15-mer peptides library. gp120, the surface subunit of HIV Env with a molecular weight of ~120 kDa; gp41, the transmembrane subunit of HIV Env with molecular weight of ~41 kDa; FP, fusion peptide; NHR, N-terminal heptad repeat; CHR, C-terminal heptad repeat; TM, transmembrane region; CT, cytoplasmic tail. (B) Inhibitory activities of peptides from group M Env 15-mer peptides library at 10 μM against HIV-1 IIIB infection in MT-2 cells that express CD4 and CXCR4. This test was repeated twice. (C) Schematic diagram of HIV-1 cytoplasmic region. LLP1, LLP2, and LLP3, lentiviral lytic peptide-1, -2, and -3, respectively. Each one can form amphipathic helix and bind to the inner leaflet of HIV-1 membrane or HIV-1–infected cell membrane. The sequences of LLP3, F9170, and F9-scr (the scrambled peptide of F9170) are shown. (D) Circular dichroism curve of F9170 and F9-scr. This test was repeated twice.
RESULTS
Identification of a short peptide with anti–HIV-1 activity from a 15-mer HIV-1 Env peptide library
To find more drug targets on Env, we screened a 15-mer peptide library from the National Institutes of Health (NIH) AIDS Reagent Program (Fig. 1A). These serial peptides comprise the whole consensus sequence from the Env region of the HIV-1 M group, which is the most common group (20), having 11 amino acids overlap between sequential peptides. An inhibition test against HIV-1 showed that most of these peptides were ineffective in inhibiting strain HIV-1 IIIB infection under the concentration of 10 μM. Peptide F9170 (residues 789 to 803) exhibited potent anti–HIV-1 activity, whereas the four adjacent peptides, F9168 (residues 781 to 795), F9169 (residues 785 to 799), F9171 (residues 793 to 807), and F9172 (residues 797 to 811), had no HIV-1 inhibitory activity (Fig. 1B).
All five of these peptides fully or partly overlap the gp41 lentiviral lytic peptide-3 (LLP3) sequence (Fig. 1C). HIV-1 gp41 LLP1, LLP2, and LLP3 domains, which are all located inside the viral particle (Fig. 1C), can form an amphipathic helix with their hydrophobic side immersed into the HIV-1 membrane (21, 22). To investigate the biophysical properties of F9170, we analyzed its secondary structure by circular dichroism (CD) spectrum. Both F9170 and its scrambled peptide (F9-scr) manifested double minima at 208 and 222 nm, which is a curve of α-helical conformation, but the helical content of F9-scr (33%) was much lower than that of F9170 (65%) (Fig. 1D).
F9170 inhibits HIV-1 infection and inactivates cell-free HIV-1 virions
To investigate whether peptide F9170 has broad anti–HIV-1 activity, we tested a total of 25 HIV-1 strains, including 2 HIV-1 laboratory-adapted strains and 20 clinical isolates, with different subtypes and tropisms, as well as 3 fusion inhibitor-resistant strains. As shown in Fig. 2A and table S1, F9170 inhibited HIV-1 infection in a dose-dependent manner, whereas F9-scr was not inhibitory at 10 μM. The half-maximal inhibitory concentration (IC50) of F9170 against HIV-1 ranged from 0.11 to 1.78 μM, with an average of 0.83 μM. Although F9170 is not as potent as T20 (enfuvirtide, IC50 ranged from 0.01 to 0.07 μM), the only U.S. Food and Drug Administration–approved gp41-derived HIV EI, F9170 could strongly inhibit infection of T20-resistant strains with IC50 from 0.11 to 0.68 μM, suggesting that the target of F9170 is different from that of T20.
(A) Inhibitory activity of F9170, T20, or F9-scr against HIV-1 IIIB (left) and BaL (right) infection. F9170 was mixed with 100× 50% tissue culture infective doses (TCID50) of HIV-1 live virus, and then MT-2 (for X4 virus) or CEMx 174 5.25 M7 cells (for R5 virus) were added. The magnitude of HIV replication was represented by quantification of p24 in the culture medium at day 3. (B) Inactivation of cell-free HIV-1 IIIB (left) and BaL (right) virions. F9170 was mixed with 200× TCID50 of HIV-1 live virus, and the target cells were added after the removal of peptide. Similarly, p24 quantification in the culture medium at day 3 was measured to represent the magnitude of HIV replication. Both experiments were repeated twice.
Then, we tested whether F9170 could inactivate HIV-1 viral particles. In this inactivation assay, before being added to the target cells, HIV-1 particles were preincubated with serially diluted peptide. After the removal of peptide, the peptide-pretreated HIV-1 particles were then added to the target cells. Peptide F9170 could inactivate HIV-1 in a dose-dependent manner, whereas neither T20 nor F9-scr showed inactivation of the virions (Fig. 2B). Moreover, F9170 could inactivate all 21 tested clinical isolates of different subtypes and tropisms, as well as HIV-1 mutants with resistance to gp41 CHR-derived peptide-based HIV-1 EIs (T20 and T2635), with EC50 (half-maximal effective concentration) ranging from 0.09 to 0.95 μM (table S2). These results further suggest that gp41 LLP3-derived peptide F9170 has a mechanism of action different from that of the gp41 CHR-derived peptides.
F9170 targets the gp41 cytoplasmic LLP1 domain
The inactivation activity of F9170 against cell-free virions suggests that this peptide may take effect before HIV-1 attaches to the target cell. To investigate whether it targets HIV-1 Env, we tested the inhibitory activity of F9170 against infection of the pseudo-typed HIV-1 and Middle East respiratory syndrome–related coronavirus (MERS-CoV), which have the same HIV NL4-3 backbone proteins, but different Envs, i.e., HIV-1 Env and MERS-CoV Env, respectively. As shown in Fig. 3 (A and B), F9170 could inhibit infection of pseudo–HIV-1 in a dose-dependent manner, but not that of pseudo–MERS-CoV, whereas MERS-CoV EI HR2PM2 (23) worked in the opposite way. This result suggests that F9170 can specifically target the HIV-1 Env protein gp120 or gp41.
(A and B) Inhibition of F9170 upon entry of the HIV-1 pseudo-virus into the target cell, TZM-bl. Pseudo-virus NL4-3 consists of HIV-1 NL4-3 backbone and HIV-1 NL4-3 Env and MERS-CoV consists of HIV-1 NL4-3 backbone and MERS-CoV Env. HR2PM2 is a MERS-CoV entry inhibitor. (C) Sequences of the peptides used in this study for identification of F9170’s target. (D) Effect of F9170 at different concentrations and gp120-specific antibody N6 on binding between gp120 and soluble CD4 (sCD4). OD450, optical density at 450 nm. (E) Inhibition of F9170 and the FP-binding peptide VIR-164 on FP-mediated hemolysis. (F) Inhibition of F9170 and the 6-HB inhibitory peptide T2635 on 6-HB formation between N36 and C34 peptides. (G) Binding of F9170 to the gp41-derived peptides N63, C34, MPER, LLP1, LLP2, and LLP3, respectively. (H) Binding of F9170 to the LLP1-derived peptides. The above tests were repeated twice.
To identify F9170’s binding site(s) in HIV-1 Env, we compared its inhibitory activity with that of a series of inhibitors with known target sites in HIV-1 Env (Fig. 3C). First, we tested whether F9170 could block the interaction between gp120 and its receptor CD4, using the bNAb N6 that specifically targets the CD4-binding site in gp120 (24) as a positive control. Whereas N6 at 7 nM could block gp120 binding to CD4, F9170 had no inhibition at the concentration as high as 10 μM (Fig. 3D), suggesting that the CD4-binding site in gp120 was not the target of F9170.
To determine whether F9170 interacts with the FP domain in HIV-1 Env, we performed an FP-mediated hemolysis assay, as previously described (25), using the peptide VIR-164, which inhibits HIV-1 entry by binding to HIV-1 FP domain (13), as a control. Consistently, VIR-164 could effectively inhibit FP-mediated hemolysis in a dose-dependent manner, whereas, again, F9170 had no inhibition at the concentration as high as 100 μM (Fig. 3E), suggesting that FP domain in HIV-1 gp41 may not be the target of F9170.
We then investigated whether F9170 could block the formation of 6-HB between the gp41 NHR and CHR using a CHR peptide, T2635, which inhibits HIV-1 fusion and entry by blocking 6-HB formation (26), as a control. As shown in Fig. 3F, F9170 did not affect 6-HB formation at the concentration as high as 10 μM, whereas T2635 at 0.6 μM could inhibit 80% of 6-HB formation, suggesting that F9170 may not target the gp41 NHR or CHR domain.
Subsequently, we synthesized biotin-labeled F9170 to detect whether F9170 could bind to several HIV-1 Env-derived proteins or peptides using the enzyme-linked immunosorbent assay (ELISA), as previously described (27). We found that F9170 neither bound to protein gp120 or gp140 (the full-length HIV-1 Env without cytoplasmic region) (fig. S1) nor peptides derived from the gp41 NHR, CHR, and MPER (Fig. 3G).
We then investigated whether F9170 targeted the gp41 cytoplasmic region. F9170 bound to the LLP1 (residues 828 to 856) domain (Fig. 3G), especially peptide LLP1-GQ, which consists of residues 835 to 849 (aligned to sequence of HIV-1 HXB2) (Fig. 3H). To further elucidate the interaction model between F9170 and LLP1-GQ, we introduced single-site mutations to peptide LLP1-GQ and F9170, respectively. First, we detected the binding ability of LLP1-GQ mutants to biotin-labeled F9170. Binding activity of LLP1-GQ-A4, -A8, -A9, -A10, and -A11 to F9170 decreased significantly (P < 0.01; Fig. 4A). Then, we detected the binding activity of F9170 mutants to biotin-labeled LLP1-GQ. As shown in Fig. 4B, F9170-A3, -A5, -A7, -A10, -A11, -A12, -A14, and -A15 exhibited reduced binding activity to LLP1-GQ. These results suggest that R838, H842, I843, P844, R845, and I847 in LLP1-GQ and E791, L793, Y795, N798, L799, L800, Q802, and Y803 in F9170 play important roles in the interaction between F9170 and LLP1-GQ.
(A) Sequences of mutants of LLP1-GQ (top) and the percentage of change in binding of F9170 to LLP1-GQ mutants, compared with that to LLP1-GQ (bottom). (B) Sequences of mutants of F9170 (top) and the percentage of change in binding of LLP1-GQ to F9170 mutants, compared with that to F9170 (bottom). (C) Percent change of in IC50 of F9170 mutant peptides for inhibiting HIV-1 IIIB infection, compared with that of F9170 peptide. (D) Simulated model of interaction between LLP1-GQ and F9170. (E) Hypothetical schematic of F9170 for inactivating HIV-1 virions. (F) Inactivation of pseudo-typed HIV-1 infection. 3 mut, three mutations. (G) Consensus analysis of LLP1-GQ sequence. The above tests were repeated twice. Data were analyzed by ANOVA test, the P value between F9170 and each group was adjusted by Bonferroni correction. *P < 0.05, **P < 0.01, and ***P < 0.001.
To further investigate whether these important-for-binding sites were also vital for F9170’s inhibitory activity, we tested the potency of F9170 mutants against HIV-1 infection. Compared with the result from the binding test, more residues with mutations were responsible for the decreased inhibitory activity of F9170 (Fig. 4C and fig. S2). By using the Iterative Threading ASSEmbly Refinement (I-TASSER) server (28), we simulated the F9170/LLP1-GQ interaction model and found that one of the simulated interaction models had a good fit with our experimental results. As shown in Fig. 4D, R838, H842, and R845 in LLP1 and Q791, Y795, D798, and Y802 in F9170 are involved in the interaction, thus mutations in these loci resulting in a significant decrease in binding between LLP1-GQ and F9170. A hydrophobic zone formed by L799, L800, and W803 is vital for the construction of amphipathic helix, which also affected the interaction between LLP1-GQ and F9170. This model may also explain why F9170-A2, -A4, -A8, -A9, and -A13 exhibited weak anti–HIV-1 activity, but no significant decrease in binding to LLP1-GQ, because the mutations of these residues that are located on the hydrophobic surface of F9170 may also disturb the amphipathic helical formation of F9170, thus suppressing the interaction of F9170 with the lipid membrane. Therefore, we speculate that F9170, through its hydrophobic surface, interacts first with the outer leaflet of lipid bilayers of viral membranes, which is consistent with the report that LLP3 peptide can insert into phospholipid membranes (21), and then with LLP1, the hydrophobic surface of which is located in the inner leaflet. The association between the hydrophilic surfaces of F9170 and LLP1 results in the formation of a hydrophilic channel (Fig. 4E) in the membrane, which may affect the function of HIV-1 virion.
To further validate the important binding site(s) of F9170 in LLP1, we constructed pseudo-typed HIV-1 virus with single mutation of R838A, H842A, or R845A or a combination of three replacements (three mutations). The infectivity of each mutant was slightly or markedly decreased (fig. S3). Therefore, the ability of F9170 to inactivate these pseudo-viruses was tested after their baseline infectivity was normalized to each other. As shown in Fig. 4F, The EC50 values of F9170 for inactivating pseudo-typed HIV-1 IIIB with Env carrying mutation of R838A, H842A, R845A, or a combination of three mutations were 5.3-, 6.7-, 14.7-, and 20.7-fold higher than that in wild-type HIV-1 IIIB, respectively. The loss of F9170’s activity on these pseudo-viruses further proved that R838, H842, and R845A were important for the binding of F9170 to the cytoplasmic tail of the HIV-1 Env. We further analyzed whether the binding sites of F9170 were conserved among the diverse HIV-1 strains. After using LLP1-GQ in Blast tool of NIH website and collecting all the 19,998 aligned sequences, we analyzed the ratio of each amino acid in every site with AnalyzeAlign (29). As shown in Fig. 4G, the predicted binding sites of F9170 (838R, 842H, and 845R) are highly conserved among the 19,998 HIV-1 strains analyzed, indicating that F9170 has broad anti–HIV-1 activity against divergent HIV-1 strains, including those resistant to the gp41-derived HIV-1 fusion inhibitors, T20 and T2635.
F9170 inactivates cell-free HIV-1 virions by disrupting the integrity of HIV-1 particles
To verify our hypothesis that F9170 inactivates HIV-1 virions by destabilizing the viral membrane through its interaction with LLP1 domain, we investigated whether F9170 could disrupt the integrity of HIV viral membrane. We incubated the pseudo-typed HIV-1 and MERS-CoV with F9170 and tested the release of HIV-1 RNA genome from viral particles using reverse transcription quantitative polymerase chain reaction (RT-qPCR). As shown in Fig. 5A, 100% of the HIV-1 RNA genome was released from 10 μM F9170-treated HIV pseudo-virus particles in both tests with different primers, similar to that released from 1% Triton X-100–treated virions. However, 10 μM of F9170 did not cause RNA release from MERS-CoV pseudo-virus particles, indicating that F9170 could only disrupt membrane of viral particles with HIV Env.
(A) The amplifiable viral genome released from the HIV-1 (left) or MERS-CoV (right) pseudo-typed virions pretreated with F9170 or 1% Triton X-100 and digested by ribonuclease was detected by RT-qPCR. (B) HIV-1 genome in each sucrose gradient fraction was detected after HIV-1 particles were pretreated with F9170, 1% Dulbecco’s modified Eagle’s medium, or 1% Triton X-100 and centrifuged in sucrose by RT-qPCR (left) and by SDS–polyacrylamide gel electrophoresis and Western blot, using anti–HIV-1 p24 antibody (right). (C) Cytotoxicity of F9170 to CHO-Env cells (left), Jurkat-HXBc2 cells (middle), and H9/IIIB cells (right). CHO-Env, CHO cells that stably express HIV-1 Env on the cell surface; Jurkat-HXBc2, Jurkat cells that express full-length HIV-1 HXBc2 Env on the cell surface; Jurkat-713 STOP, Jurkat cells that express HIV-1 HXBc2 Env ectodomain (without cytoplasmic domain); H9/IIIB, chronically HIV-1 IIIB-infected H9 cells that express HIV-1 IIIB Env. Cell viability was assessed by CCK8 kit. Inhibition rate = (positive value-tested value)/(positive value-negative value). Cells without treatment and cells treated with 1% Triton X-100 served as positive and negative control, respectively. (D) PI fluorescence and anti–annexin V–FITC antibody stain of untreated (left), F9170-treated (middle), and F9-scr–treated (right) H9/IIIB cells. (E) Killing of chidamide-reactivated and nonreactivated HIV-1latently infected cells by F9170 (right) and F9-scr (left). The above tests were repeated twice. Data were analyzed by Student’s unpaired two-tailed t test. *P < 0.05 and ***P < 0.001. N.S., no significance.
To further confirm the release of viral genome from the F9170-treated HIV-1 particles, we performed a sucrose density gradient assay as previously described (30). After centrifugation of HIV-1 particles pretreated with 10 μM of F9170, 1% Triton X-100 (positive control), or 1% dimethyl sulfoxide (DMSO) (negative control), respectively, in gradient sucrose, we analyzed the fractions from different layers and demonstrated that HIV-1 genomic RNA and core protein p24 were concentrated in fraction nos. 4 and 5 in the DMSO group (Fig. 5B), implying that most HIV-1 particles remained intact after treatment with 1% DMSO. However, HIV-1 genomic RNA was concentrated in fraction no. 7, and most p24 was detected in fraction nos. 1 and 2 in the F9170-treated group, similar to the result from the 1% Triton X-100 group. These results suggest that F9170 can disrupt the integrity of HIV-1 particles, resulting in the release of HIV-1 genome.
F9170 induced necrosis of HIV-1 Env-expressing cells and HIV-1–infected cells
The above results suggested that F9170 could disrupt membrane by interacting with HIV-1 Env. To investigate whether it could also affect cells expressing HIV-1 Env, we incubated Chinese hamster ovary (CHO)–Env cells, which constitutively express HIV-1 gp160 on their membrane (31), with F9170, using CHO cells expressing no HIV-1 Env as control cells. After a 72-hour incubation with 10 μM of F9170, CCK8 was added to evaluate cell viability. We found that about 80% of CHO-Env cells were dead and that the cytotoxicity of F9170 on CHO-Env cells was dose-dependent. However, no detectable cytotoxicity of F9170 was observed on CHO cells at the concentration as high as 10 μM (Fig. 5C). Consistently, about 87% of Jurkat-HXBc2 cells expressing HIV-1 Env and 5% of Jurkat-713 STOP cells expressing HIV-1 Env without gp41 cytoplasmic tail (32) were dead after they were incubated with 10 μM of F9170 for 72 hours (Fig. 5C). These results suggested that F9170 might interact with the gp41 cytoplasmic tail to exert its cytotoxic function.
Because HIV-1–infected cells also express HIV-1 Env on the cell membrane (33), we reasoned that F9170 might also kill HIV-1–infected cells. Thus, we incubated chronically HIV-1 IIIB–infected H9 (H9/IIIB) cells (using uninfected normal H9 cells as control) with F9170 and found that about 84% of the H9/IIIB and 12% of H9 cells died after incubation with 10 μM of F9170 for 72 hours (Fig. 5C), confirming that F9170 was effective in killing HIV-1–infected cells.
We then used a double staining system with annexin V–fluorescein isothiocyanate (FITC) and propidium iodide (PI) (34) to detect whether these H9/IIIB cells died from apoptosis or necrosis. As shown in Fig. 5D, about 86% of F9170-treated H9/IIIB cells were PI+ and FITC−, indicating the necrosis of cells (34). In contrast, F9-scr–treated H9/IIIB cells exhibited no marked change in the proportion of cells of each quadrant. Meanwhile, the numbers of FITC+ cells (both PI+ and PI−) in F9170-treatment group were not obviously higher than those in the untreated and F9-scr treatment groups, indicating that F9170 does not induce apparent apoptosis.
Latent HIV-infected cells neither produce new virions, nor do they express viral Env. However, once they are reactivated by an LRA, they can produce HIV particles and express HIV Env on the cell surface (35). Histone deacetylase inhibitor (HDACi), chidamide and romidepsin (RMD), were reported to be capable of effectively reactivating HIV latency cells (36). Here, we tested the potential cytotoxic effect of F9170 and F9-scr on latent HIV-infected cells, ACH-2 cells, before and after reactivation with chidamide or RMD. As shown in Fig. 5E and fig. S4A, F9170 could kill the reactivated ACH-2 cells, but not the nonreactivated ACH-2 cells, whereas F9-scr could kill neither reactivated nor nonreactivated ACH-2 cells, suggesting that F9170 can be used in combination with an LRA to eradicate latent HIV reservoir. The result was similar in another in vitro latency model, U1 cells (fig. S4B).
F9170 is safe in vitro and in vivo
A previous study revealed that the LLP1 domain can trigger pore formation in lipid membranes (37). Because F9170 is derived from the adjacent LLP3 domain, we tested the potential cytotoxicity of F9170 on 3T3, U87-CD4+-CCR5+, and TZM-bl cells, all of which usually serve as HIV-1 target cells in in vitro assays. As shown in Fig. 6A, F9170 did not attenuate the viability of these three cell lines with concentration ranging from 0.39 to 12 μM. We also tested the in vivo toxicity of F9170 in an ICR (strain named under Institute of Cancer Research) mouse model. Mice treated with F9170, whether under low (20 mg/kg) or high (100 mg/kg) concentration, had weight change ratio similar to that of the phosphate-buffered saline (PBS) group (negative) (Fig. 6B and fig. S5A). In addition, the serum alanine aminotransferase (ALT) concentration, which indicates hepatotoxicity, and the serum creatinine concentration, which indicates nephrotoxicity, of F9170-treated mice did not present a substantial difference compared to the PBS group (Fig. 6C). Histology further confirmed that injection of F9170 did not cause any pathological change to mouse liver and kidney (fig. S4B).
(A) Viability of cells expressing no HIV-1 Env when incubated with F9170. (B) Percent initial starting weight of male mice at different intervals after injection of PBS and F9170 at 20 and 100 mg/kg, respectively. Group size = 5. (C) The amounts of serum glutamic-pyruvic transaminase (ALT) and creatinine in F9170- and PBS-treated mice. Group size = 5. (D and E) Detection of F9170-specific antibody in mice (D) and rhesus macaques (E) at different intervals after injection of PBS or saline and F9170 at 100 mg/kg for mice and 3 mg/kg for rhesus macaques. Group size (mouse) = 6; group size (macaque) = 4. (F) Detection of preexisting antibody specific for F9170 in HIV-1–infected patients. Data were analyzed by Student’s unpaired two-tailed t test.
Drug-specific antibodies will substantially attenuate drug efficacy, making immunogenicity a vital characteristic to evaluate. We tested the immunogenicity of F9170 by measuring F9170-specific antibodies in F9170-treated mouse serum (Fig. 6D). The mice did not develop specific antibodies toward F9170 at any time point monitored within 6 weeks after one-time immunization.
Rhesus monkey models chronically infected with simian-HIV (SHIVSF162P3) were also adopted for the immunogenicity test. Four macaques were intramuscularly injected with F9170 in saline at 3 mg/kg of body weight, and another four macaques were injected with the same volume of saline only twice daily for 1 week and once daily for another week. We detected no F9170-specific antibodies in serum samples from these four F9170-treated macaques at any time point monitored within 6 weeks after first-time injection (Fig. 6E), indicating low immunogenicity of F9170, which is an ideal trait for drug development.
The HIV-1 virus has been reported to induce antibodies against HIV-1 Env-derived drugs in infected patients. These preexisting antibodies could counteract the inhibitory activity of Env-derived drugs, such as T20 (38). Here, we tested 10 serum samples from HIV-1–infected patients. As shown in Fig. 6F, binding of F9170 to patients’ serum proteins was similar to that of F9-scr or PBS, indicating that there are no preexisting anti-F9170 antibodies. To further study the druggability of F9170, we compared the inhibitory activity of F9170 stored at room temperature with F9170 stored at −20°C. As shown in fig. S6, the inhibition curves of F9170 stored at room temperature and −20°C are overlapping, indicating that F9170 peptide could possibly be stored and transported at ambient temperature, which could reduce the cost for drug transportation.
No neural toxicity was detected in F9170-treated mice
Before evaluating the efficacy of F9170 against HIV-1 infection in an animal model, we used an in vivo fluorescence imaging system to examine the distribution of F9170 in mice. After administration of the fluorescence-labeled peptide F9170-Cy5 via tail vein, we have detected the fluorescent signal in the area around the liver, bladder, and brain (fig. S7, A and B). We further measured the fluorescence intensity of F9170-Cy5 and F9-scr–Cy5 (as control) in the dissected organs. The fluorescence signal of F9170-Cy5 could be detected in the liver, brain, uterus, kidney, lymph nodes (LNs), spleen, and heart. The fluorescence signal of F9-scr–Cy5 was detected in the liver, kidney, spleen, and heart, whereas moderate signal of F9-scr-Cy5 was detected in the LN (fig. S7, C to K), suggesting that F9170-Cy5 was distributed evenly through the blood circulation into mouse organs. F9170, but not F9-scr-Cy5, could cross the blood-brain barrier (BBB) and the blood-perilymph barrier to penetrate the central nervous system (CNS) and peri-LN. It is well known that the CNS (39) and LN (40) are both sanctuary sites for HIV-1, preventing the entry of many drugs. Therefore, F9170 may gain entry to these anatomic reservoirs to inactivate HIV-1 virions, kill HIV-1–infected cells, and even eradicate latent HIV-1.
We further examined whether F9170 could cause neural damage in vivo in normal mice without viral infection using the same protocol for the hepatotoxicity test as described above. Mouse brain sections were analyzed for quantification of neuronal nuclear protein–positive (NeuN+) cells (neuron) and glial fibrillary acidic protein–positive (GFAP+) cells (astrocytes). As shown in fig. S8, no apparent loss of neurons and significant increase in astrocytes (i.e., astrogliosis) were observed, suggesting that F9170 does not cause apparent neural damage in mice.
F9170 is effective in controlling SHIV infection in nonhuman primates
We then evaluated the in vivo therapeutic efficacy of F9170 in a nonhuman primate (NHP) model. Eight rhesus macaques chronically infected with SHIVSF162P3 were randomly assigned to two groups and intramuscularly injected with F9170 in saline at 3 mg/kg of body weight (F1, F2, F3, and F4) or saline only (C1, C2, C3, and C4) twice daily for 1 week and once daily for another week. The dose choice of F9170 was based on a preliminary study on mice where F1970 (20 mg/kg) resulted in a maximum concentration (Cmax) of 22 μM (fig. S9). Under conversion based on body surface area, F9170 (3 mg/kg) is supposed to result in a Cmax of ~10 μM, a safe and effective concentration in in vitro studies, in rhesus macaques. As shown in Fig. 7 (A to C), plasma viral loads were maintained at around 4 to 5 log10 RNA copies/ml and declined sharply after injection of F9170, whereas plasma viral loads exhibited no substantial change after injection of saline. In the F9170-treated group, the plasma viral loads in monkeys F1, F3, and F4 dropped below the limit of detection (<100 copies/ml) on day 3 after injection. As expected, the viral loads rebounded in all treated monkeys after stopping F9170 administration. This result suggests that intramuscular administration of F9170 as a monotherapy exhibits in vivo efficacy.
Eight female rhesus macaques chronically infected with SHIVSF162P3 were intramuscularly injected with saline or F9170 twice daily for 1 week and once daily for another week. Viral loads in plasma were determined by a quantitative real-time reverse transcription PCR (qRT-PCR) assay at the indicated time after first injection. (A) Plasma viral loads of F9170-treated rhesus macaques. Group size = 4. (B) Plasma viral loads of saline-treated rhesus macaques. Group size = 4. (C) Statistical analysis of results of (A) and (B). The comparison was performed between viral loads of F9170- and saline-treated group at the same day. The significant difference was analyzed by Student’s unpaired two-tailed t test. *P < 0.05 and **P < 0.01. (D) F9170 serum concentration in F9170-treated rhesus macaques at different time point after intravenous (iv; left) or intramuscular (im; right) injection. 1 μM = 1.983 μg/ml. Group size = 2.
We then evaluated the in vivo half-life of F9170 in rhesus macaques. Briefly, rhesus macaques were intravenously or intramuscularly injected with F9170 in saline at 3 mg/kg of body weight, and the concentration of F9170 at different time points after injection was determined with liquid chromatography–tandem mass spectrometry (LC-MS/MS). F9170 could be detected until 12 hours after injection with an in vivo half-life of 8.94 hours (intravenously) and 11 hours (intramuscularly) (Fig. 7D), which is much longer than that of most short peptide–based drugs (41). In addition, the concentration of F9170 in macaques’ sera at 12 hours after injection is above IC90 (1.32 μM) of F9170 against HIV-1 IIIB in vitro.
Last, we assessed whether F9170 could cause brain damage in the F9170-treated rhesus macaques and found no apparent loss of neurons in the F9170-treated rhesus macaques (fig. S10, A to D). Consistent with previous reports on simian immunodeficiency virus–infected macaques (42, 43), SHIV infection also resulted in astrogliosis and up-regulated GFAP expression in astrocytes (fig. S10, E and F). Although F9170 treatment did not attenuate the astrogliosis in the SHIV-infected macaques, no further upgraded expression of GFAP in the brains of F9170-treated macaques was detected, compared with the untreated SHIV-infected macaques (fig. S10, G and H). These results suggest that administration of F9170 to rhesus macaques does not cause neural damage, consistent with results from the study on F9170-treated mice.
DISCUSSION
The high mutation rate of HIV leads to a correspondingly rapid emergence of drug-resistant viruses during treatment with current ARDs. Therefore, identification of previously undefined drug targets and discovery of ARDs that can act on these targets still remain urgent. In this study, we identified a short 15-mer amphipathic peptide, designated F9170, which exhibited potent and broad inhibitory activity against pseudo-typed and live HIV-1 strains with different subtypes and tropisms. Most current ARDs must enter and wait inside the host cell to interact with intracellular proteins, such as reverse transcriptase and protease, to inhibit HIV-1 replication (44, 45). In contrast, F9170 can attack free virions in blood circulation or tissues before virions attach to, or enter into, the cells. Some nonionic surfactants, such as N9, and surface active agents, including C31G, an alkyl sulfate, and SDS, can also inactivate HIV-1 virions in vitro (46), but their high cytotoxicity and low specificity prevent them from systemic in vivo application. F9170 has very low in vitro cytotoxicity and in vivo toxic effect, and it is highly specific for virions and cells expressing HIV-1 Env. Furthermore, F9170 can kill HIV-1–infected cells and cells expressing full-length HIV-1 Env. Because HIV-1 can spread directly from productively infected cells to neighboring cells via virological synapses, a more efficient way than infection with cell-free virions (47), F9170 is expected to be effective in eliminating these HIV-1–infected cells, thus blocking cell-cell transmission of HIV-1. The suppressive effect of F9170 against SHIV infection of rhesus macaques may be attributed to its ability to inactivate HIV-1 virions and kill SHIV-infected cells.
We previously reported that a peptide derived from the stem region of Zika virus (ZIKV) E protein could inactivate ZIKV by disrupting the integrity of viral membranes (30). Holthausen et al. (48) also reported that an amphibian host defense peptide was virucidal to human H1 hemagglutinin-bearing influenza viruses. Here, the results from mechanistic studies also suggest that F9170 inactivates virions and induces necrosis of HIV-1–infected cells by disrupting the integrity of the membranes of cell-free virions and cells expressing HIV-1 Env through its interaction with the LLP1 domain in the cytoplasmic tail of gp41. The mechanism of action of F9170 is different from that of classic antiviral drugs. In general, after binding of a drug to its target protein, the biological function of the viral protein is attenuated, resulting in the inhibition of viral infection. However, binding of F9170 to the LLP1 domain of gp41 does not affect the function of gp41 but rather results in the disruption of the integrity of the viral and infected cell membranes. Therefore, this concept may also be used to identify more antiviral molecules against other enveloped viruses.
Recently, Murphy et al. (22) have solved the solution structure of the gp41 cytoplasmic tail in a micellar environment and its membrane-associated structure, providing valuable insights into the functions and mechanisms of the HIV Env cytoplasmic tail. They have shown that a large hydrophobic surface of the gp41 cytoplasmic tail is buried in the viral membrane and the residues R838, H842, and R845 are on the hydrophilic surface, which is consistent with the result from our simulated interaction study. Furthermore, they proposed that the conserved arginine in LLP sequence might contribute to membrane bilayer destabilization by keeping the charged side chain in the hydrophobic interior. F9170’s interaction with LLP1 domain may strengthen this process. Meanwhile, the structure revealed that the linker between LLP1 and LLP3 was short (22). In our simulated interaction model, N terminus of F9170 peptide derived from LLP3 interacts with the C terminus of LLP1-GQ peptide. The short linker limited the interaction between the homologous LLP3 and LLP1.
The current ARDs can treat, but not cure, HIV-1 infection because they cannot eradicate the latent HIV reservoir (35). These HIV reservoirs can survive for a long time with a half-life time of 43.9 months in patients (49). Several drugs can serve as LRAs to reactivate the integrated cells, including HDACis (50, 51), protein kinase C agonist (52), and Toll-like receptor agonist (53). The reactivated cells proceed to produce HIV particles and express HIV Env on the cell surface. Although ARDs can suppress replication of the new viral particles, the drugs cannot kill these reactivated HIV-producing cells. It was reported that these reactivated HIV-infected cells can be killed by specific cytotoxic T lymphocytes (CTLs) (54). However, the in vivo efficacy of CTLs is limited because many latent viruses carry CTL escape mutations, rendering the reactivated HIV-infected cells insensitive to CTLs (55). Unlike the F9-scr peptide, we show that F9170 is effective in killing latent HIV-infected cells, ACH-2 or U1 cells, after reactivation with HDACi-based LRA but that it is ineffective in killing the nonreactivated cells. This result suggests that F9170 may be promising in combination with an LRA to eradicate the latent HIV reservoirs.
Another hindrance to HIV eradication is the limited accessibility of anatomic reservoirs. The CNS is an optimal sanctuary for HIV where HIV-1 can replicate in astrocytes (56) and macrophages (57), and the replication can be persistent and independent during treatment, resulting in the development of resistant strains different from those in plasma of the patients (58). One reason is the limited penetration of ARDs into the CNS. It was reported that the concentration of the small-molecule compound-based ARDs in the cerebrospinal fluid of ARD-treated patients with HIV is much lower than that in their plasma. Meanwhile, the peptide-based anti-HIV drug, enfuvirtide, cannot easily cross the BBB (39). In contrast, a number of amphipathic peptides have good BBB-penetrating properties (59). Consistently, the short amphipathic peptide F9170 can get into the mouse brain just 1 hour after intravenous injection of F9170, suggesting that F9170 may inactivate HIV-1 virions and kill HIV-1–infected cells in the CNS. Estes et al. (40) reported that HIV replication persisted in lymphoid tissue (LT) and drug concentrations were relatively lower in LT than in peripheral blood. Here, we found that the fluorescence signal of F9170-Cy5 could also be detected in mouse inguinal LNs, suggesting that F9170 may also enter LT and take effect there. It has been reported that transcription profile of HIV is different in the gut than that in the blood, suggesting that the gut may also be a sanctuary for HIV latency (60). The F9170’s presence in the gut has not been assessed in this study. Therefore, further investigation of distribution and concentration of F9170 in the gut is needed.
Recombinant multifunctional proteins, such as 2DLT and 4Dm2m (61, 62), and anti–HIV-1 bNAbs can also neutralize free virions in blood circulation and kill HIV-1–infected cells through ADCC (63). Compared with these antibodies, F9170 has several strengths. First, F9170 can be easily synthesized, thus having much lower production cost than bNAbs. Second, F9170 can penetrate the tissues to the viral sanctuary sites, such as CNS, because of its amphipathic trait, whereas a bNAb cannot easily access these sites because of its large molecular size. Third, F9170 can directly induce necrosis of HIV-infected cells, whereas bNAbs may indirectly kill the cells through ADCC. Fourth, F9170 peptide could possibly be stored and transported at room temperature, whereas the antibody drugs generally needed to be stored and transported at low temperature.
F9170 has an in vivo half-life of 8.9 hours in rhesus macaques, about fourfold longer than most peptide drugs. However, F9170 may still require once or twice daily drug administration, making it less ideal for HIV treatment, which currently needs life-long adherence. Many approaches have been proved successful in extending the in vivo half-life of a peptide, such as nanosuspension (64), implantation devices (65), and conjugation with polyethylene glycol, albumin-binding motif (66), or a CD4-binding adnectin (67, 68). In particular, we recently reported that a short anti–HIV-1 peptide CP24 conjugated with the human immunoglobulin G Fc-binding peptide (IBP) exhibited a prolonged half-life about 26-fold longer than that of CP24 without IBP (69). Therefore, half-life of F9170 can be further improved with these strategies to reduce the injection frequency. Compared with other small molecular ARDs that can be taken orally, F9170 administration would need to be parenteral (intravenously or intramuscularly), which would limit its clinical use. However, the trend in HIV treatment is to develop long-acting drugs, most of which need parenteral use or even implantation. Therefore, if F9170’s half-life can be further extended, then administration of F9170 once a month or every 2 weeks would be more acceptable. Another barrier for F9170 to enter the next step to the clinical use is its high-cost potential because at least 300 mg/day (3 mg/kg*50 kg*twice a day) will be used; this could partially be overcome by sequence optimization to improve its efficacy or extend its half-life as described above, to reduce its dosage and thus its cost. Although studies on relationship between F9170 potential target mutations and F9170 specific resistance have been carried out, there is still a lack of systematic data on induction of F9170-resistant strains, especially in animal models. In the NHP model, the effectiveness of F9170 has only been observed for a short period of time. Therefore, it is still unknown whether long-term use will be constantly effective and whether a F9170-resistant strain will be induced. The analysis of F9170-resistant strains will help to further clarify the mechanism and target of F9170, as well as its potential feasibility in clinical application. It is still cannot be ruled out that F9170 has other targets or mechanisms for inactivating HIV-1.
MATERIALS AND METHODS
Study design
The objective of this study is to identify a novel anti-HIV drug candidate for HIV treatment. HIV-1 infection inhibition assay was used to screen a peptide library consisting of 211 peptides derived from HIV-1 Env. Pseudo-typed virus inhibition assay, ELISA, viral genome release and sucrose density gradient assays, and specific cytotoxicity assays were applied to determine the drug target and mechanism of action of the anti–HIV-1 peptide identified. A series of animal experiments on mouse and rhesus macaque models were performed to assess the in vivo safety and efficacy. All the mice and rhesus macaques were randomly assigned to F9170 and control groups. Group sizes were determined on the basis of preliminary experiments to give a statistical power of 0.8. Except for the efficacy assay in rhesus macaques, experiments and analyses of this study were performed unblinded. Each in vitro experiment was repeated at least twice, and each in vivo experiment was performed at least once with at least two animals in one group. Individual-level primary data are reported in data file S1.
Ethics statement
All animal experiments were conducted under ethical guidelines and approved by the Institutional Laboratory Animal Care and Use Committee at Fudan University (20160927-2). ICR mice were used and bred at the Department of Laboratory Animal Science of Fudan University. All rhesus macaques were housed in an Association for Assessment and Accreditation of Laboratory Animal Care–accredited facility. Protocols for the use of animals were approved by the Institutional Animal Care and Use Committee of the Institute of Laboratory Animal Science, Chinese Academy of Medical Sciences (no. XJ17005). All animals were anesthetized with ketamine hydrochloride (10 mg/kg) before the procedures.
Blood samples were obtained from healthy and HIV-1–infected individuals with the consent of the individuals. All the related experiments were conducted under ethical guidelines and approved by the Ethics Committee at Shanghai Medical College, Fudan University (02, 2017.10.26). The detailed information of these individuals with HIV-1 is provided in table S3.
In vivo therapeutic efficacies and neural toxicity of F9170 in SHIV-infected rhesus macaques
The in vivo therapeutic efficacy of inhibitors was evaluated in SHIV-infected rhesus macaques as described previously (70). Eight female rhesus macaques (8 to 10 years old) chronically infected with SHIVSF162P3 (NIH AIDS Reagent Program) were randomly assigned to two treatment groups for administration of saline and F9170, respectively. One milliliter of 0.9% saline solution or F9170 in saline solution (3 mg/kg) was intramuscularly injected twice daily for the first week and once daily for the second week. Viral loads in plasma were determined by a quantitative real-time RT-PCR (qRT-PCR) assay. To evaluate the potential neural toxicity of F9170 in the four SHIV-infected macaques treated with F9170, four SHIV-infected macaques without F9170 treatment and one untreated and uninfected rhesus macaque were used as controls. All these animals were anesthetized with ketamine hydrochloride, and their brains were collected and fixed in formalin for immunohistochemical assay. The tissue slices were prepared by technicians at Servicebio. Anti-NeuN antibody (ab177487, Abcam) and anti-GFAP antibody (GB11096, Servicebio) were used for quantification of neurons and astrocytes, respectively. The NeuN+ and GFAP+ areas were measured with ImageJ (https://imagej.nih.gov/ij/). The blood samples of F9170- or saline-treated macaques at indicated day were collected to test the potential F9170-specific antibodies in blood induced by F9170 after its intravenous or intramuscular administration of F9170.
To determine the optimal dose of F9170 that would be used in macaques, we performed a preliminary study on mice. Four mice were divided into two groups and intravenously injected with F9170 (20 and 100 mg/kg), respectively. The blood samples of mice were collected at 0, 15, and 30 min, and 1.5, 3, 6, 9, 12, and 24 hours after injection and serum concentration of F9170 was assessed by LC-MS/MS.
Statistical analysis
Unless specially indicated, data are shown as means ± SD. Statistics were performed with Prism 7 (GraphPad Software Inc.). Unpaired Student’s t test was used to analyze the significance of difference between two groups. Analysis of variance (ANOVA) test was used to analyze the significance of difference among three or more groups, and the P value between each group was adjusted by Bonferroni correction. P < 0.05 was considered significant.
SUPPLEMENTARY MATERIALS
stm.sciencemag.org/cgi/content/full/12/546/eaaz2254/DC1
Materials and Methods
Fig. S1. Binding of F9170 to HIV-1 gp120 and gp140.
Fig. S2. Percent change of in IC50 of F9170 mutant peptides for inhibiting HIV-1 BaL infection, compared with that of F9170 peptide.
Fig. S3. Infectivity of mutated and wild-type pseudo-virus to TZM-bl target cells.
Fig. S4. Cytotoxicity of F9170 to reactivated latently infected cells.
Fig. S5. Safety of F9170 in mice.
Fig. S6. Inhibitory activity of F9170 against HIV-1 IIIB infection in MT-2 cells.
Fig. S7. Distribution of F9170 in mice.
Fig. S8. Safety of F9170 in the CNS of mice.
Fig. S9. F9170 serum concentration in F9170-treated mice at different time point after intravenous injection.
Fig. S10. Safety of F9170 in the CNS of SHIV-infected rhesus macaques.
Table S1. Inhibitory activity of F9170 against HIV-1 laboratory-adapted strains, primary isolates, and T20-resistant strains.
Table S2. Inactivation of cell-free HIV-1 R5 and X4 strains by F9170.
Table S3. Clinical information of 10 patients with HIV whose blood were tested for preexisting antibodies to F9170.
Data file S1. Primary data.
This is an article distributed under the terms of the Science Journals Default License.
REFERENCES AND NOTES
- Copyright © 2020 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works
