Oxazole-Benzenesulfonamide Derivatives Inhibit HIV-1 Reverse Transcriptase Interaction with Cellular eEF1A and Reduce Viral Replication

Antiretroviral drugs protect many HIV-positive people, but their success can be compromised by drug-resistant strains. To combat these strains, the development of new classes of HIV-1 inhibitors is essential and a priority in the field. In this study, we identified small molecules that bind directly to HIV-1 reverse transcriptase (RT) and inhibit its interaction with cellular eEF1A, an interaction which we have previously identified as crucial for HIV-1 replication. These compounds inhibit intracellular HIV-1 reverse transcription and replication of WT HIV-1, as well as HIV-1 mutants that are resistant to current RT inhibitors. A novel mechanism of action involving inhibition of the HIV-1 RT-eEF1A interaction is an important finding and a potential new way to combat drug-resistant HIV-1 strains in infected people.

ABSTRACT

HIV-1 replication requires direct interaction between HIV-1 reverse transcriptase (RT) and cellular eukaryotic translation elongation factor 1A (eEF1A). Our previous work showed that disrupting this interaction inhibited HIV-1 uncoating, reverse transcription, and replication, indicating its potential as an anti-HIV-1 target. In this study, we developed a sensitive, live-cell split-luciferase complementation assay (NanoBiT) to quantitatively measure inhibition of HIV-1 RT interaction with eEF1A. We used this to screen a small molecule library and discovered small-molecule oxazole-benzenesulfonamides (C7, C8, and C9), which dose dependently and specifically inhibited the HIV-1 RT interaction with eEF1A. These compounds directly bound to HIV-1 RT in a dose-dependent manner, as assessed by a biolayer interferometry (BLI) assay, but did not bind to eEF1A. These oxazole-benzenesulfonamides did not inhibit enzymatic activity of recombinant HIV-1 RT in a homopolymer assay but did inhibit reverse transcription and infection of both wild-type (WT) and nonnucleoside reverse transcriptase inhibitor (NNRTI)-resistant HIV-1 in a dose-dependent manner in HEK293T cells. Infection of HeLa cells was significantly inhibited by the oxazole-benzenesulfonamides, and the antiviral activity was most potent against replication stages before 8 h postinfection. In human primary activated CD4⁺ T cells, C7 inhibited HIV-1 infectivity and replication up to 6 days postinfection. The data suggest a novel mechanism of HIV-1 inhibition and further elucidate how the RT-eEF1A interaction is important for HIV-1 replication. These compounds provide potential to develop a new class of anti-HIV-1 drugs to treat WT and NNRTI-resistant strains in people infected with HIV.

IMPORTANCE Antiretroviral drugs protect many HIV-positive people, but their success can be compromised by drug-resistant strains. To combat these strains, the development of new classes of HIV-1 inhibitors is essential and a priority in the field. In this study, we identified small molecules that bind directly to HIV-1 reverse transcriptase (RT) and inhibit its interaction with cellular eEF1A, an interaction which we have previously identified as crucial for HIV-1 replication. These compounds inhibit intracellular HIV-1 reverse transcription and replication of WT HIV-1, as well as HIV-1 mutants that are resistant to current RT inhibitors. A novel mechanism of action involving inhibition of the HIV-1 RT-eEF1A interaction is an important finding and a potential new way to combat drug-resistant HIV-1 strains in infected people.

INTRODUCTION

Human immunodeficiency virus (HIV) is a member of the Retroviridae family and Lentivirus genus that infects and kills CD4⁺ T cells and can lead to AIDS. There are 37 million people living with HIV, and there were one million AIDS-related deaths in 2017 (1). There are over 20 approved antiretroviral (ART) drugs, and these are used in combination for maximal effectiveness and to minimize emergence of drug-resistant viral strains (2, 3). This treatment strategy is called combination antiretroviral therapy (cART) and has been highly successful for inhibiting HIV infection and preventing further transmission and progression to AIDS (4–6). However, HIV is able to rapidly mutate to become resistant to these antiretroviral drugs, and this resistance is a major cause of progression to AIDS (3, 7, 8). Nonnucleoside reverse transcriptase inhibitors (NNRTIs) are the most affordable and widely used first-line antiviral drug treatment, and these include the drugs nevirapine, efavirenez, and delavirdine. By the end of 2016, NNRTI resistance levels were between 4% and 28% in people with suppressed viral loads and 47% to 90% in people with unsuppressed viral loads (9). This highlights the need to further develop new classes of antiretroviral drugs with novel mechanisms of action to treat current and future drug-resistant HIV (10).

Reverse transcription of HIV-1 is the conversion of the positive-sense single-stranded RNA genome into double-stranded DNA, which is a precondition for integration into the host chromosomes for subsequent replication. This is primarily catalyzed by the enzyme HIV-1 reverse transcriptase (RT) as part of the reverse transcription complex (RTC), which contains several viral and host cell proteins (11). HIV-1 RT is a heterodimer composed of two related subunits, p66 and p51 (12–14). The p66 subunit contains the active sites for both DNA polymerase and RNase H activity, while p51 plays a structural role (12). Incoming deoxyribonucleotide triphosphates (dNTPs) bind at the polymerase active site and are polymerized to form double-stranded DNA by reverse transcription. HIV RT is an effective target of antiretroviral drugs, and NNRTIs bind HIV RT and inhibit enzymatic activity to prevent reverse transcription (15).

We were first to report that cellular factors were required for efficient reverse transcription (16–18). We subsequently identified that eukaryotic translation elongation factor 1A (eEF1A) was important for reverse transcription and may be the predominant RT-binding cellular protein in the RTC (19, 20). Mutations that reduce the RT-eEF1A interaction significantly impair HIV-1 replication in CD4⁺ T cells, highlighting the importance of this interaction in maintaining a stable RTC capable of completing reverse transcription (21). Using the eEF1A-binding compound didemnin B, proof of principle has been demonstrated that the RT-eEF1A complex is a druggable target for inhibiting HIV-1 replication (20). However, clinical trials with didemnin B in cancer patients showed significant toxicity and side effects (22, 23), likely because didemnin B binds eEF1A and inhibits translation and therefore cannot be pursued as an HIV-1 treatment. Therefore, we hypothesized that RT-binding compounds that inhibit the interaction of RT with eEF1A would inhibit reverse transcription and HIV-1 replication with lower toxicity. In this study, we evaluated a compound library and identified that oxazole-benzenesulfonamide derivatives bind RT (but not eEF1A) and inhibit both RT-eEF1A interaction and HIV-1 replication in cells.

RESULTS

Establishing a live-cell NanoBiT assay for quantitating inhibition of the RT-eEF1A interaction for small-molecule inhibitor screening.

Split-luciferase complementation (SLC) assays have been previously used as small-molecule screening assays to identify inhibitors of protein-protein interactions (24, 25). NanoBiT (Promega) is an SLC assay in which a NanoLuc luciferase signal is generated when the 17.6-kDa “Large BiT” (LgBiT) and 1.3-kDa “Small BiT” (SmBiT) subunits of the enzyme come together to form a functional luciferase, triggered by the interaction of their respective fusion proteins (26). In this study, the HIV-1 RTp66 coding region was inserted at the N terminus of LgBiT for expression of the RTp66-LgBiT fusion protein, and eEF1A was inserted at the N terminus of SmBiT for expression of the eEF1A-SmBiT fusion protein. As a control protein-protein interaction, the Zika virus (ZIKV) NS2B gene was inserted at the C terminus of LgBiT for expression of the LgBiT-NS2B fusion protein, and the protease domain of NS3 (NS3pro) was inserted at the N terminus of SmBiT for expression of the NS3pro-SmBiT fusion protein. HEK293T cells were lysed 24 h after transfection of these expression vectors, and lysates were analyzed by Western blotting. This showed successful expression of RTp66-LgBiT (Fig. 1A), eEF1A-SmBiT (Fig. 1B), ZIKV NS3pro-SmBiT (Fig. 1C), ZIKV LgBiT-NS2B (Fig. 1D), and the HaloTag-SmBiT negative control (Fig. 1B). LgBiT-NS2B and NS3pro-SmBiT were fused with FLAG and hemagglutinin (HA) tags, respectively, for detection by Western blotting, but the untagged vectors were used in NanoBiT. HEK293T cells expressing RTp66-LgBiT cotransfected with eEF1A-SmBiT produced a significantly higher NanoLuc luciferase signal than RTp66-LgBiT cotransfected with HaloTag-SmBiT (P < 0.0001) (Fig. 1E). Also, ZIKV LgBiT-NS2B cotransfected with NS3pro-SmBiT produced a significantly higher NanoLuc luciferase signal than LgBiT-NS2B cotransfected with HaloTag-SmBiT (P < 0.0001) (Fig. 1F). This indicates that a specific luciferase signal was produced as a result of the HIV-1 RT-eEF1A and the ZIKV NS2B-NS3pro interactions in cells.

FIG 1.

Establishment of NanoBiT assay to screen inhibitors of the HIV-1 RT-eEF1A interaction. (A to D) Western blots of HEK293T cell lysates after 24 h of transfection with the indicated NanoBiT expression vectors using anti-RT (A), the HiBiT blotting system (B), anti-HA (C), or anti-FLAG (D). (E and F) Luminescent signal from NanoBiT assay in HEK293T cells expressing RTp66-LgBiT and eEF1A-SmBiT or RTp66-LgBiT and HaloTag-SmBiT (E) or ZIKV LgBiT-NS2B and NS3pro-SmBiT or LgBiT-NS2B and HaloTag-SmBiT (F). Data are the means from 3 independent experiments, and error bars represent SEMs.

The Open Collection Scaffolds library from Compounds Australia was screened using the NanoBiT system to identify compounds with inhibitory activity against the RT-eEF1A interaction. The library comprises 1,226 different scaffolds, and each scaffold has more than 30 members to enable screening of a small collection with ready access to compounds of related structure. One compound was selected at random from each of the 1,226 scaffolds. Compounds were plated in 96-well plates in assay-ready format by Compounds Australia. HEK293T cells cotransfected with the NanoBiT LgBiT and SmBiT plasmids were then seeded into the plates containing the compounds to a final concentration of 25 μM after 6 h of transfection. Cells were incubated with compounds overnight and NanoLuc luciferase signal was measured in live cells, immediately followed by a 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) cell proliferation assay to measure cell viability. The initial screen of 1,226 compounds identified 320 compounds that reduced the luciferase signal by at least 40% while maintaining more than 70% cell viability (Table 1). The 168 most potent compounds were selected for a repeat experiment with the same criteria as the primary (nonspecific) screen but with an added specificity control. Four compounds in this specific screen inhibited the HIV-1 RT-eEF1A interaction more than 1.5 times compared to the control, giving an overall hit rate of 0.33% for compounds with specific inhibition out of the initial library of 1,226 compounds. All compounds within these four scaffolds were then sourced from the Open Collection Scaffolds library (150 compounds) to establish structure-activity relationships (SAR) for each scaffold. A NanoBiT screen of the compound scaffold derivatives resulted in 16 hits that were evaluated for concentration response dependence. Nine compounds, featuring two different scaffolds, dose dependently and specifically inhibited the HIV-1 RT-eEF1A interaction and were evaluated in more detail.

TABLE 1.

Summary of compound library screens

Screen	No. of compounds	No. of hits	Hit rate (%)	No. of selected hits
Nonspecific screena	1,226	320	26.1	168
Specific screenb	168	4	2.38	4
Derivative screenc	150	16	10.66	16
Dose-response screend	16	9	56.25	9 + 1 negative control

^aNonspecific screen hit criteria: >40% RT:eEF1A NanoBit inhibition and >70% cell viability.

^bSpecific screen hit criteria: same as primary screen criteria plus >50% inhibition of RT:eEF1A NanoBiT signal compared to a control interaction NanoBiT signal.

^cDerivative screen hit criteria: same as secondary hit screen criteria.

^dDose-response screen hit criteria: dose-dependent inhibition of the RT:eEF1A NanoBiT signal and less inhibition of the ZIKV NS2B:NS3pro NanoBiT signal.

Benzimidazole-pyrrolones and oxazole-benzenesulfonamides inhibit the RT-eEF1A interaction and HIV-1 infectivity.

The nine compounds identified from dose-response screening plus a negative control were designated C1 to C10 and are described in Table 2 and Materials and Methods. C1, C2, C3, C4, and C5 are benzimidazole-pyrrolones, sharing a common substructure of 5-amino-4-(1H-benzo[d]imidazol-2-yl)-1-phenyl-1,2-dihydro-3H-pyrrol-3-one. C6, C7, C8, C9, and C10 are oxazole-benzenesulfonamides, sharing the common substructure of 4-(oxazol-5-yl)benzenesulfonamide. These compounds were serially diluted 2-fold from 25 μM to 0.195 μM and incubated with HEK293T cells expressing either HIV-1 RTp66-LgBiT and eEF1A-SmBiT or ZIKV LgBiT-NS2B and NS3pro-SmBiT. NanoLuc luciferase and cell viability were measured the following day, and the percentage compared to those for dimethyl sulfoxide (DMSO)-treated cells was calculated. The benzimidazole-pyrrolones C1 to C5 inhibited the RT-eEF1A interaction more strongly than the ZIKV NS2B-NS3pro interaction (Fig. 2). While compound C3 inhibited the RT-eEF1A interaction by 50% at 1 μM (IC₅₀), it did not elicit dose-dependent inhibition at higher concentrations (Fig. 2C), and this was similar for the structurally related compound C2 (Fig. 2B). C1, C4, and C5 showed dose-dependent inhibition of the RT-eEF1A interaction from 25 μM to 0.78 μM (Fig. 2A, D, and E). C4 (IC₅₀, 2 μM) and C5 (IC₅₀, 3 μM) had comparable inhibitor potency, but cell viability was slightly greater for C5 (Fig. 2E). The 50% cytotoxic concentration (CC₅₀) was not reached for any of the compounds; however, the reduction in cell viability at each concentration must be taken into consideration when analyzing the inhibition of RT-eEF1A interaction at each concentration. The results indicate that the benzimidazole-pyrrolone compounds identified can specifically inhibit the HIV-1 RT-eEF1A interaction with various potencies.

TABLE 2.

Details of hit compounds for further investigation

Compound code	ChemDiv no.	Compounds Australia no.	Scaffold series	Mol wt	Log P
C1	D086-0046	SN00785766	Benzimidazole-pyrrolone	320.35	3.1
C2	D086-0114	SN00785768	Benzimidazole-pyrrolone	338.80	4.0
C3	D086-0148	SN00785769	Benzimidazole-pyrrolone	304.35	3.4
C4	D487-0468	SN00786708	Benzimidazole-pyrrolone	322.34	3.1
C5	D487-0490	SN00786711	Benzimidazole-pyrrolone	334.38	2.9
C6	G408-0087	SN00791985	Oxazole-benzenesulfonamide	354.43	4.9
C7	G408-2173	SN00792040	Oxazole-benzenesulfonamide	380.47	5.5
C8	G408-2184F	SN00792041	Oxazole-benzenesulfonamide	368.46	5.4
C9	G408-2081	SN00792038	Oxazole-benzenesulfonamide	354.4	5.5
C10	G408-0073	SN00791982	Oxazole-benzenesulfonamide	383.47	4.2

FIG 2.

NanoBiT assay for benzimidazole-pyrrolone series compounds. Shown are compound structures and NanoBiT assay results for HEK293T cells incubated in C1 (A), C2 (B), C3 (C), C4 (D), and C5 (E). NanoBiT assay was performed with HIV-1 RTp66-LgBiT cotransfected with eEF1A-SmBiT (red line) alongside ZIKV LgBiT-NS2B cotransfected with NS3pro-SmBiT (blue line). After luciferase was measured in live cells, MTS assay was performed to measure cell viability (black line). Luciferase units from the NanoBiT assay and cell viability from the MTS assay measured at each compound concentration were normalized to cells incubated in DMSO only. Data are the means from 3 independent experiments performed in duplicate, and error bars represent SEMs.

The oxazole-benzenesulfonamides C6, C7, C8, and C9 inhibited the HIV-1 RT-eEF1A interaction substantially more than the ZIKV NS2B-NS3pro control interaction, with cells still being viable (Fig. 3A to D). C6 had higher cytotoxicity and inhibition of the control ZIKV NS2B-NS3pro interaction than C7, C8, and C9 and was not used further. C10 did not inhibit the RT-eEF1A interaction more than the ZIKV NS2B-NS3pro interaction and was therefore used as a negative control (Fig. 3E). C7 inhibited the RT-eEF1A interaction more effectively (IC₅₀, 0.35 μM) than the ZIKV NS2B-NS3pro interaction (IC₅₀, 17.5 μM) and the CC₅₀ was not reached with the concentrations tested; however, the cell viability was reduced by approximately 20% at the IC₅₀ (Fig. 3B). C8 had a potency against RT-eEF1A (IC₅₀ 0.5 μM) comparable to that of C7, and the cell viability was also reduced by approximately 20% at the IC₅₀ (Fig. 3C). C9 was more cytotoxic than C7 and C8 at 0.195 μM and had an IC₅₀s of 0.2 μM for the RT-eEF1A interaction and 12.5 μM for the ZIKV NS2B-NS3pro interaction. A nonlinear inhibition curve was not fit with the data because the inhibition by C7, C8, and C9 did not represent a simple inhibition model, where inhibition was stronger at 0.39 μM than 0.78 μM and 1.56 μM for C7 (Fig. 3B) and C9 (Fig. 3D) and inhibition was stronger at 0.78 μM than 1.56 μM and 3.125 μM for C8 (Fig. 3C). The reason for this is unknown. However, inhibition was dose dependent at a majority of concentrations. For all compounds except C6, the inhibition of the NanoBiT signal for the ZIKV NS2B-NS3pro interaction mostly matched the cell viability, indicating that inhibition of this interaction was due to cytotoxicity, while the increased inhibition of the HIV-1 RT-eEF1A was specific. Therefore, the oxazole-benzenesulfonamides C7 to C9 are identifiable inhibitors of the HIV-1 RT-eEF1A interaction.

FIG 3.

NanoBiT assay for oxazole-benzenesulfonamide series compounds. Shown are compound structures and NanoBiT assay results for HEK293T cells incubated in C6 (A), C7 (B), C8 (C), C9 (D), and C10 (E). NanoBiT assay was performed with HIV-1 RTp66-LgBiT cotransfected with eEF1A-SmBiT (red line) alongside ZIKV LgBiT-NS2B cotransfected with NS3pro-SmBiT (blue line). After luciferase was measured in live cells, MTS assay was performed to measure cell viability (black line). Luciferase units from NanoBiT assay and cell viability from MTS assay measured at each compound concentration were normalized to cells incubated in DMSO only. NanoBiT data are the means from 4 independent experiments performed in duplicate, and error bars represent SEMs. HEK293T cells were infected with HIV-1_{pNL4-3.Luc.R-E-} and incubated with 0.1 to 10 μM C7 (B), C8 (C), C9 (D), or C10 (E). Cells were lysed after 29 h, and luciferase units (RLU) at each concentration were normalized to the DMSO-only treatment control (green line). Infection data are the means of 2 independent experiments performed in duplicate, and error bars represent SEMs.

Next we investigated the effects of C7 to C10 in a single cycle of HIV-1 replication in the presence of the oxazole-benzenesulfonamides using a firefly luciferase reporter HIV-1. HEK293T cells were incubated with vesicular stomatitis virus G protein (VSV-G)-pseudotyped HIV-1 in the presence of various concentrations of C7 to C10, the cells were lysed at 29 h postinfection (hpi), and luciferase expression in cell lysates was measured. C7, C8, and C9 dose-dependently inhibited luciferase expression in HEK293T cells (Fig. 3B to D), while C10 had no antiviral effect (Fig. 3E). The inhibition of HIV-1 infection was comparable to the inhibition of the RT-eEF1A interaction in the NanoBiT assay.

Oxazole-benzenesulfonamides bind HIV-1 RT but not eEF1A.

To determine if the compounds bind directly to either HIV-1 RT or eEF1A, biolayer interferometry (BLI) was used and biosensors were coated with the RTp66/p51 heterodimer (Fig. 4) or eEF1A (Fig. 5). RT was purified from Escherichia coli as described previously (21), and eEF1A was purified from HEK293T cells (Fig. 5A and B). The binding of the compounds to HIV-1 RT- or eEF1A-coated biosensors was measured by incubation with increasing concentrations of compounds on the Octet-Red system. A double-reference subtraction was performed where the response detected from the reference sensor (GST-His-coated sensors for RT or biocytin-coated sensors for eEF1A) incubated in compounds, as well as the response detected from the target sensor (RT- or eEF1A-coated sensors) incubated in buffer only, were subtracted from the response for the target sensor incubated with compounds. There was no specific dose-dependent response for the RT sensor incubated in any of the benzimidazole-pyrrolones (Fig. 6A to E), and therefore, these compounds were not characterized further. Delavirdine and all the oxazole-benzenesulfonamides (C6 to C9) except the negative control (C10) showed dose-dependent binding to the RT-coated probe (Fig. 4 and Fig. 6F and G). Under these conditions, the binding affinities (equilibrium dissociation constants [K_D]) calculated from the steady-state analysis were 15 μM for C7 (Fig. 4A, right), 29 μM for C8 (Fig. 4B, right), 34 μM for C9 (Fig. 4C, right), and 63 μM for delavirdine (Fig. 4D, right). Specific and dose-dependent binding to an eEF1A biosensor was not detected for C7 to C9 or delavirdine (Fig. 5C to F). The data show that the oxazole-benzenesulfonamides C7 to C9 were specific RT binding compounds, with no affinity for eEF1A.

FIG 4.

Oxazole-benzenesulfonamide series compounds bind HIV-1 RT in biolayer interferometry assay. Ni-NTA biosensors coated in HIV-1 RTp51/p66 heterodimer purified from E. coli were incubated in 3.125 μM (dark blue), 6.25 μM (red), 12.5 μM (light blue), 25 μM (green), 50 μM (yellow), and 100 μM C7 (A), C8 (B), C9 (C) or delavirdine (D). Data presented are the association and dissociation of the compounds with the RT-coated probes after a double-reference subtraction of RT biosensors incubated in buffer only, as well as GST-His-coated Ni-NTA biosensors incubated in compounds (left). The experiment was performed at least 3 times, with similar results, and a representative data set is shown. The K_D and r² shown were calculated from the steady-state analysis of responses (right).

FIG 5.

Oxazole-benzenesulfonamide compounds had no detectable binding to purified eEF1A. Human eEF1A purified from HEK293T cells were run on an SDS-PAGE gel and analyzed by Coomassie staining (A) and Western blotting using a mouse anti-eEF1A antibody (B). Super Streptavidin biosensors coated in biotinylated eEF1A purified from E. coli were incubated in C7 (C), C8 (D), C9 (E), or delavirdine (F) at concentrations of at least 100 μM. Data presented are the responses of the eEF1A-coated probes incubated in compounds after a double-reference subtraction of eEF1A biosensors incubated in buffer only, as well as Biocytin SSA biosensors incubated in compounds. The experiment was performed at least 2 times, with similar results, and a representative data set is shown.

FIG 6.

Benzimidazole-pyrrolone series compounds and C10 do not bind RT in biolayer interferometry. Ni-NTA biosensors coated in HIV-1 RTp51/p66 heterodimer purified from E. coli and were incubated in 3.125 μM (dark blue), 6.25 μM (red), 12.5 μM (light blue), 25 μM (green), 50 μM (yellow), and 100 μM (purple) C1 (A), C2 (B), C3 (C), C4 (D), C5 (E), C6 (F), or C10 (G) as indicated. Data presented are the responses of the RT-coated probes incubated in compounds after a double-reference subtraction of RT biosensors incubated in buffer only, as well as GST-His-coated Ni-NTA biosensors incubated in compounds. The experiment was performed at least 3 times, with similar results.

Oxazole-benzenesulfonamides inhibit reverse transcription during infection but not in vitro enzymatic activity.

To determine whether and which stages of the HIV-1 life cycle can be affected by the oxazole-benzenesulfonamides, the optimal compound and virus concentration for inhibition in HeLa cells were determined to be used in a time-of-addition (TOA) assay. VSV-G-pseudotyped HIV-1 with an enhanced green fluorescent protein (eGFP) reporter in the Nef open reading frame (HIV-1_{pNL4-3-Δnef-eGFP}) was spinoculated with HeLa cells for 1.5 h at 16°C, and then unbound virus was removed and 0.39 μM or 0.78 μM C7 to C9 was added and incubated for 24 h. These concentrations were used because they were the lowest concentrations that produced 40% to 60% inhibition of HIV-1 infectivity in HEK293T cells (Fig. 3B to D). HeLa cells were used for TOA assays due to their superior attachment during the infection, washing, and compound addition stages and maintained almost 100% cell viability apart from C9 at 0.78 μM, which had 80% cell viability compared to DMSO-treated cells (Fig. 7A). Cells were fixed at 24 hpi and analyzed by flow cytometry, and the percentage of infected cells was determined by eGFP expression. C7 and C8 had no inhibition at 0.39 μM, while C9 inhibited infection by 15%; however, there was significant inhibition of the percentage of infected cells at 0.78 μM for C7, C8, and C9 (Fig. 7B) compared to DMSO-treated cells (P < 0.05). C7, C8, and C9 also inhibited nonpseudotyped HIV-1 in TZMBL cells (Fig. 8). The concentration of HIV-1 used in infection was titrated down, and C7 to C9 had increased inhibition in a virus 50% tissue culture infectious dose (TCID₅₀)-dependent manner, where C7 and C9 inhibited percent infection 3-fold and C8 by 35% at the lowest HIV-1 concentration (Fig. 7C). Therefore, HeLa cells were infected with HIV-1 to achieve approximately 40% infection (TCID₅₀ of 0.8) under DMSO treatment conditions, and 0.78 μM C7 to C9 was added in hourly intervals up to 8 hpi and then at 10 hpi and 12 hpi. Nevirapine (5 μM) and raltegravir (1 μM) addition was included as a control of RT and integrase (IN) inhibition. The percentage of inhibitory activity of the compound compared to DMSO at each time point was calculated so that the inhibitory activity of the compound was 100% when added at 0 hpi. Nevirapine still maintained 100% inhibitory activity when added up to 2 hpi; however, the inhibitory activity was sharply reduced when addition was delayed between 3 and 8 hpi (Fig. 7D). Since nevirapine loses its antiviral activity once reverse transcription is complete, the data indicate that reverse transcription was completed by 8 hpi. Susceptibility of HIV to raltegravir was approximately 2 h later than nevirapine, and the data indicate that integration was completed by 10 hpi (Fig. 7D). Delaying C7 addition to 8 hpi reduced the inhibitory activity of C7 by 50% (Fig. 7D), indicating that half of the antiviral activity of the compound is during uncoating and reverse transcription. However, C7 still maintained about 40% of its antiviral activity when added at 12 hpi, indicating that unknown postintegration effects exist. C8 and C9 displayed profiles similar to that of C7; however, inhibitory activity remained for slightly longer after infection (Fig. 7D).

FIG 7.

Oxazole-benzenesulfonamide series compounds inhibit HIV-1 reverse transcription in cells but not in vitro. (A) Cell viability of HeLa cells incubated with 0.39 μM or 0.78 μM C7, C8, or C9 for 24 h measured using MTS assay and normalized to cells incubated in DMSO only. (B) Percentage of infected HeLa cells (eGFP) when incubated in DMSO or 0.39 μM or 0.78 μM C7, C8, or C9 from 0 hpi for 24 h. This was repeated with 0.78 μM C7, C8, or C9 using a range HIV-1 concentrations to achieve a titration of TCID₅₀ (C). (D) Time-of-addition assay in which HeLa cells were infected with HIV-1 by spinoculation at 16°C to achieve a TCID₅₀ of approximately 0.8. The cells were washed and medium containing DMSO only, 0.78 μM C7, C8, or C9, 5 μM nevirapine, or 1 μM raltegravir was added at the indicated times postinfection. Cells were fixed at 29 hpi, and eGFP expression was measured by flow cytometry. Data were normalized to the percent infection of the DMSO-only sample; the percent inhibition was calculated and is the mean of 4 independent experiments performed in duplicate. (E) The in vitro RT enzymatic activity with incubation with DMSO only or 5 μM nevirapine, delavirdine, C7, C8, C9, or C10. (F and G) Strong-stop DNA copy number (F) or late DNA copy number (G) at 5 hpi in the presence of DMSO only, C7, C8, C9, or C10 as measured by qPCR. Data are the means of (D and E) or representative of (A, B, C, F, and G) at least 2 independent experiments performed in duplicate, and error bars represent SEMs.

FIG 8.

Inhibition of nonpseudotyped HIV-1 in TZMBL cells. TZMBL cells were spinoculated with nonpseudotyped HIV-1_{pNL4-3-Δnef-eGFP} at 1,200 × g and 37°C for 1.5 h. Cells were washed and medium containing DMSO or 0.78 μM C7, C8, or C9 was added, and cells were incubated for 24 h before fixation in 1% paraformaldehyde and flow cytometry for eGFP expression to determine the percentage of infected cells. Data are the means of duplicate samples and representative of those from two independent experiments.

Since C7 to C9 bind RT, an in vitro RT activity assay was performed to determine if these compounds also inhibit the catalytic activity of RT in vitro. HIV-1 RT, homopolymer poly(A) DNA template, and oligo(dT) primers were incubated in DMSO, nevirapine, delavirdine, or the oxazole-benzenesulfonamides at 5 μM for 3 h, and reverse transcription products were measured by a standard colorimetric enzyme-linked immunosorbent assay (ELISA). While nevirapine and delavirdine significantly inhibited reverse transcription, the oxazole-benzenesulfonamides did not affect the catalytic activity of RT in vitro (Fig. 7E). However, HIV-1 DNA products of viral reverse transcription in cells were measured in HeLa cell lysates by quantitative PCR (qPCR) after 5 hpi. This showed that C7, C8, and C9 at 5 μM inhibited strong-stop DNA (Fig. 7F) and late stage reverse transcription DNA products (Fig. 7G) at 5 hpi. This indicates that C7, C8, and C9 did not directly inhibit HIV-1 RT enzymatic activity in vitro but did inhibit reverse transcription in cells during infection through a different mechanism.

Oxazole-benzenesulfonamides inhibit infectivity and replication in human primary activated CD4⁺ T cells.

Human primary activated CD4⁺ T cells were infected with luciferase reporter HIV-1 in the presence of C7, C8, or C9 and lysed at 29 hpi, as described earlier for HEK293T cells. For C8 and C9 there was no inhibition of HIV-1 at 0.195 μM and approximately 20% more inhibition of luciferase signal from HIV-1 infection compared to cell viability measured by MTS assay at 0.78 μM and 1.56 μM (Fig. 9B and C). After 29 h of treatment with 0.39 μM C7 or C8, the cell viability was more than 90%, and while at this concentration C8 did not inhibit HIV-1 infection (Fig. 9B), C7 inhibited luciferase expression 3-fold (P < 0.001 compared to the value for cell viability) (Fig. 9A).

FIG 9.

Oxazole-benzenesulfonamide series compounds inhibit HIV-1 infection in human primary activated CD4⁺ T cells. Human primary activated CD4⁺ T cells infected with VSV-G pseudotyped HIV-1_{pNL4-3.Luc.R-E-} and incubated with 0.195, 0.39, 0.78, or 1.56 μM C7 (A), C8 (B), or C9 (C). Cell viability was determined by MTS assay; cells were lysed after 29 h, and RLU at each concentration were normalized to the DMSO-only treatment control. Data are the means from 2 independent experiments performed in triplicate, and error bars represent SEMs. A nonlinear semilog inhibition curve was fit using GraphPad Prism software. For replication assays, human primary activated CD4⁺ T cells were infected with HIV-1_pNL4-3 for 1.5 h at 37°C, and unbound virus was washed away. Cells were cultured in DMSO (1:1,000) or 0.195, 0.293, or 0.39 μM C7 for 6 days, cell culture supernatant was collected, and live (D) and dead (E) cells were counted using trypan blue staining. (F to I) Light microscopy of CD4⁺ T cells treated with DMSO (F) or 0.39 μM C7 (G) at 2 dpi at ×10 magnification and DMSO (H)- or 0.39 μM C7 (I)-treated cells at 4 dpi at a magnification of ×20. Total HIV-1 CA was measured in cell culture supernatant (top) collected on days 1 (J), 2 (K), 4 (L), and 6 (M) postinfection using a p24 ELISA, and this was normalized to total live cells (bottom). Data for replication assays are the means from 1 independent experiment performed in triplicate and are representative of 2 independent experiments. Error bars represent SEMs.

Since 0.39 μM C7 was the only concentration tested that significantly inhibited a single round of HIV-1 infection in human primary activated CD4⁺ T cells, this concentration as well as 0.293 μM or 0.195 μM was used to monitor HIV-1 replication over several days in CD4⁺ T cells. The live-cell count as determined by trypan blue staining showed that there were similar numbers of cells when treated with either DMSO (0.1%) and C7 at 0.39 μM for 1 day (Fig. 9D), matching the MTS data (Fig. 9A). However, the number of live cells was lower when treated with 0.39 μM C7 than for the DMSO-treated samples at 2, 4, and 6 days postinfection (dpi) (Fig. 9D), although there was a statistically significant difference only on day 4 (P < 0.01). Treatment of cells with 0.39 μM C7 did not lead to an increase in the number of dead cells (Fig. 9E), and cells maintained a round and translucent morphology similar to that of DMSO-treated cells throughout the experiment (Fig. 9F to I). Treatment of cells with 0.293 μM or 0.195 μM did not affect the number of live or dead cells compared to that with DMSO treatment (Fig. 9D and E).

HIV-1 p24 in the cell culture supernatant was measured by ELISA 0, 1, 2, 4, and 6 dpi. Human primary activated CD4⁺ T cells treated with 0.39 μM C7 had lower levels of p24 in the supernatant at every day postinfection measured, while treatment with 0.293 μM or 0.195 μM did not significantly inhibit p24 release into the supernatant (Fig. 9J to M, top). When the ratio of picograms of p24 per cell was calculated, there was less p24 released into the supernatant per cell when cells were treated with 0.39 μM C7 at each dpi (Fig. 9J to M, bottom); however, this was statistically significant only 4 and 6 dpi (P < 0.05). This suggests that C7 can inhibit HIV-1 infection and replication in human primary CD4⁺ T cells at 0.39 μM.

Oxazole-benzenesulfonamides inhibit infectivity NNRTI-resistant mutant HIV-1 in HEK293T cells.

To determine if the oxazole-benzenesulfonamides were able to inhibit HIV-1 mutants that are resistant to the NNRTIs nevirapine and delavirdine, luciferase reporter HIV-1 with drug resistance mutation K103N or Y181C in RT was used to infect HEK293T cells and compounds were added at the time of infection. Nevirapine and delavirdine strongly inhibited wild-type (WT) HIV-1 at concentrations between 0.098 μM and 1.56 μM, but nevirapine did not inhibit K103N or Y181C mutants at up to 0.39 μM, and 1.56 μM nevirapine inhibited these mutant strains only by 40% (Fig. 10A and B). Delavirdine did not inhibit K103N HIV-1 at 0.098 μM and had reduced inhibition compared to WT HIV-1 at the concentrations tested. However, C7 and C8 inhibited K103N and Y181C RT mutant HIV-1 at levels similar to WT HIV-1 (Fig. 10C and D). This indicates that the oxazole-benzenesulfonamides can inhibit the K103N and Y181C mutant NNRTI-resistant strains of HIV-1.

FIG 10.

Oxazole-benzenesulfonamide series compounds inhibit infection of NNRTI-resistant HIV-1 in HEK293T cells. HIV-1_{pNL4-3.Luc.R-E-} with WT RT or RT containing the K103N or Y181C mutation was used to infect HEK293T cells and incubated with 0.098, 0.39, or 1.56 μM nevirapine (A), delavirdine (B), C7 (C), or C8 (D). Cells were lysed at 29 hpi, and RLU at each concentration were normalized to the DMSO-only treatment control. Data are the means from 2 independent experiments performed in duplicate, and error bars represent SEMs.

DISCUSSION

We have previously demonstrated that the HIV-1 RT interaction with eEF1A is important for reverse transcription, uncoating, and replication (19–21). In this study, we have identified a subset of oxazole-benzenesulfonamide compounds that inhibited the interaction between RT and eEF1A in a live-cell NanoBiT assay. The high-throughput NanoBiT screening assay has the advantage of immediately measuring cell viability in the same cells after measuring luciferase signal, and it ensures that hits identified are bioavailable and tolerated by the cell at the first screen. By using a different protein-protein interaction as a control, such as the interaction between ZIKV NS2B and NS3pro, the specificity of compound inhibition is confirmed early in the screening process, and the identification of inhibitors of a second interaction is possible. The inhibitors of the RT-eEF1A interaction in NanoBiT assay, named C7, C8, and C9, specifically bound to HIV-1 RT and inhibited reverse transcription in HeLa cells and inhibited HIV-1 infection in HEK293T and HeLa cells, while C7 also inhibited replication in human primary CD4⁺ T cells. The oxazole-benzenesulfonamides also inhibited infection by common NNRTI-resistant HIV-1 mutants, and since NNRTIs inhibit in vitro RT enzymatic activity while our compounds do not, our data suggest that these compounds may represent a new class of RT-binding anti-HIV-1 inhibitors with a mechanism of action different from that of current NNRTIs. While the time-of-addition assays suggest that inhibition of HIV-1 by oxazole-benzenesulfonamides is complex, our data provide evidence that RT-binding compounds (Fig. 4) can inhibit RT interaction with eEF1A (Fig. 3) and reduced viral DNA synthesis postinfection and reduced infectivity (Fig. 7 to 9). It is notable that cytosolic sulfotransferase family 1A member 1 (SULT1A1) has been reported to be important during late stages of HIV-1 reverse transcription (27), and all oxazole-benzenesulfonamide series compounds are in the ChemDiv sulfotransferase compound library, but it is not known if these compounds have any antisulfotransferase activity.

Part of the inhibition of RT-eEF1A NanoBiT signal and HIV-1 luciferase signal in HEK293T cells is likely due to partial cytotoxicity, and it is unclear to what extent this partial cytotoxicity affects the IC₅₀. The selectivity indices of these compounds were not calculated since they did not reach a CC₅₀ in HEK293T cells. The oxazole-benzenesulfonamides had increased toxicity at higher concentrations in CD4⁺ T cells than in HEK293T cells; however, toxicity was lower in CD4⁺ T cells incubated with 0.39 μM C7 and C8. C7 was at a high enough concentration (0.39 μM) to inhibit HIV-1 replication without adversely affecting cell viability in CD4⁺ T cells, while the higher concentrations of C8 and C9 required for antiviral activity in CD4⁺ T cells had significant cytotoxicity. The selectivity index for a single round of infection in human primary CD4⁺ T cells treated with C7 was 2.3, which was too low to perform extensive characterization in these cells. The enhanced potency of C7 compared to those of C8 and C9 in CD4⁺ T cells may reflect the increased binding affinity of C7 to RT in the BLI assay. However, delavirdine binding affinity in our BLI assays was significantly higher than has been previously reported (28). This is likely due to assay-specific differences, including dissociation buffer and temperature, and therefore indicates that the binding affinity of C7, C8, and C9 may be different under different assay conditions. However, the BLI assays were employed primarily to qualitatively determine compound binding to RT.

The oxazole-benzenesulfonamides C6 to C10 all share a 4-(oxazol-5-yl)benzenesulfonamide substructure; however, they have variable chemical substituents at both ends. The most active compounds have a distinctly kinked shape that arises from their unsymmetrical tertiary sulfonamide compared to the linear symmetrical structure of the inactive compound C10. The secondary sulfonamide and associated H bond donor NH group could be responsible for the lower binding affinity and antiviral potency of C9 in HEK293T cells, which is partially recovered with the close analogue C8, which has an N-methyl group; however, C9 was more potent than C8 in HeLa cells. The best compound, C7, contains the tertiary sulfonamide constrained by an additional 5-membered ring that rigidifies the kinked structure that appears to be favored for binding. In C6 the more acute angle conferred by the 6-membered ring appears to have directed the aromatic ring to an unfavorable position or caused a steric clash. The 2-cyclopropyloxazole also appears to be tolerated. Although only a very small series of compounds has been evaluated in this study, it is likely that further structure-active relationships could be exploited to enhance the antiviral activity and reduce the toxicity of these oxazole-benzenesulfonamides in future studies.

Previous studies in our lab have comprehensively characterized the importance of eEF1A in interaction with the HIV-1 RT, and the eEF1A-binding drug didemnin B did not inhibit the interaction but did inhibit HIV-1 reverse transcription (20). This indicated that the RT-eEF1A complex was important for HIV-1 replication and a potentially druggable target. The oxazole-benzenesulfonamides had a phenotype similar to that of RT containing the E300R mutation, which we previously reported did not inhibit RT enzymatic activity in an in vitro homopolymer DNA assay but did inhibit the RT interaction with eEF1A and impaired reverse transcription and replication in cells (21).

There are reports of other compounds with a benzenesulfonamide substructure having anti-HIV-1 activity. A “dihydrobenzo-1,4-dioxin-substituted analog of N-(3-aminoquinoxalin-2-yl)-4-chlorobenzenesulfonamide” (DQBS) compound bound directly to HIV-1 Nef protein and inhibited Nef-dependent major histocompatibility complex (MHC) class I downregulation and replication (29). The authors suggested that DQBS may inhibit Nef interaction with effector proteins required for Nef function. It is notable that Nef has also been shown to interact with eEF1A (30), but whether certain benzenesulfonamides affect this interaction is unknown. Another study found that 2,4-difluoro-N-8-quinolinyl-benzenesulfonamide derivatives inhibit HIV-1 Rev function and HIV-1 replication (31). The sulfonamide linker in this compound was not important for inhibition of Rev but was important for the anti-HIV-1 activity of the compound (31), suggesting that there may be another mechanism of HIV-1 inhibition other than anti-Rev activity. Kang et al. performed structure-based molecular hybridization and subsequent decorating to optimize thiophene[3,2-d]pyrimidine derivatives based on crystal structures of NNRTIs and RT (32). They synthesized 48 derivatives, and the most potent compound contained a primary benzenesulfonamide group (4-PhSO₂NH₂) that made H bonds with K104 and V106 in RT. An earlier study by Masuda et al. also found benzenesulfonamide derivatives with inhibitory activity against HIV-1 with WT or NNRTI-resistant mutant RT (33), while Pala et al. found benzenesulfonamide derivatives that inhibited HIV-1 RT RNase H (34). Together, these findings suggest that benzenesulfonamide-derived compounds may have more than one anti-HIV-1 activity and more than one anti-RT activity, and our data indicate that one of these activities inhibits RT interaction with cellular eEF1A.

In summary, we have identified new RT-binding compounds that inhibit the HIV-1 RT interaction with eEF1A and inhibit HIV-1 replication. The oxazole-benzenesulfonamides identified in this study may represent a completely new class of anti-HIV-1 compounds, as they bind RT but do not inhibit in vitro enzymatic activity. Resistance to anti-HIV-1 drugs is an increasing problem (9), and our potentially new class of compounds inhibit common NNRTI-resistant HIV-1 strains. It is also possible that if the oxazole-benzenesulfonamides bind at the interaction interface, then a resistance mutation that affects this interface may have a fitness cost that could inhibit HIV-1 replication and prevent emergence of resistant strains. However, the oxazole-benzenesulfonamides may act through an allosteric effect at an RT region where mutation may not impose a fitness cost. The potential for novel compounds to inhibit the interaction between an HIV-1 protein and a host cell protein has been demonstrated with the development of IN/lens epitheium-derived growth factor (LEDGF) inhibitors, called LEDGINs, that are in preclinical trials (35). Structure determination of the HIV-1 RT-eEF1A complex by cocrystallization or cryogenic electron microscopy (cryo-EM) could provide valuable information for drug design and compound optimization. Hit-to-lead discovery of new compounds based on the structure of C7, C8, and C9 using medicinal chemistry, pharmacophore modeling, and molecular docking studies may progress a compound to preclinical trials as a new type of antiretroviral agent.

MATERIALS AND METHODS

Cell lines and virus culture.

HEK293T (ATCC) cells were maintained in Dulbecco’s modified Eagle medium (DMEM) (Gibco) supplemented with 10% heat-inactivated fetal calf serum (FCS), penicillin (100 IU/ml), and streptomycin (100 μg/ml). Opti-MEM medium supplemented with 10% heat-inactivated FCS, penicillin (100 IU/ml), streptomycin (100 μg/ml), and 1 mM HEPES was used for HEK293T cells in the NanoBiT assay. HeLa (clone CCL-2; ATCC) cells were maintained in RPMI 1640 supplemented with 10% heat-inactivated FCS, penicillin (100 IU/ml), and streptomycin (100 μg/ml).

Ficoll density centrifugation was used to purify peripheral blood mononuclear cells (PBMCs) from buffy coat supplied by the Australian Red Cross Blood Service. Magnetic isolation of CD4⁺ T cells was performed using the EasySep human CD4⁺ T cell isolation kit as per the manufacturer’s instructions (STEMCELL Technologies). Purified CD4⁺ T cells were activated by culturing in plates coated with anti-human CD3 (clone HIT3a) and anti-human CD27 (clone CD28.2) (BioLegend) antibodies in RPMI 1640 supplemented with 20% heat-inactivated FCS, 2 ng/ml of interleukin-2 (IL-2; Miltenyi Biotec), penicillin (100 IU/ml), and streptomycin (100 μg/ml) for 2 days. All cells were incubated at 37°C in 5% CO₂.

Stocks of VSV-G-pseudotyped HIV-1_{pNL4-3-Δnef-eGFP} or HIV-1_{pNL4-3.Luc.R-E-} with wild-type or K103N or Y181C mutant RT or WT HIV-1_pNL4-3- were generated by transfection of the proviral plasmid in to HEK293T cells with or without a VSV-G expression vector using X-tremeGENE HP transfection reagent (Roche). Cell culture supernatants were collected 48 h posttransfection, filtered through a 0.45-μm filter, and stored at −80°C. HIV-1 CA in virus stocks and other viral samples was measured using a RETROtek HIV-1 CA antigen ELISA (Zaptometrix, USA) as per manufacturer instructions.

Plasmids.

HIV-1 RTp66 was inserted into the NanoBiT pBiT1.1-C[TK/LgBiT] vector using the XhoI and EcoRI restriction sites to fuse the NanoBiT LgBiT subunit of NanoLuc luciferase to the C terminus of HIV-1 RTp66. The ZIKV NS2B gene was inserted into the pBiT1.1-N[TK/LgBiT] vector using the XhoI and EcoRI restriction sites to fuse the NanoBiT LgBiT subunit of NanoLuc luciferase to the N terminus of ZIKV NS2B. A codon-optimized eEF1A gene and the ZIKV NS3pro gene were inserted into the pBiT2.1-C[TK/SmBiT] using the XhoI and EcoRI restriction sites to fuse the NanoBiT SmBiT subunit of NanoLuc luciferase to the C termini of eEF1A and ZIKV NS3pro, respectively. Inverse PCR was used to insert a FLAG tag at the C terminus of ZIKV LgBiT-NS2B and an HA tag at the N terminus of NS3pro-SmBiT using Phusion polymerase (New England Biolabs) as per the manufacturer instructions. The HIV-1_{pNL4-3.Luc.R-E-} plasmid was used as a template in inverse PCR using CloneAmp HiFi PCR premix (TaKaRa Bio USA Inc.) and primers with nucleotide changes to introduce the K103N or Y181C mutation in RT, and the mutations were confirmed by sequencing.

Western blot and Nano-Glo HiBiT blotting system.

HEK293T cells were transfected with RTp66-LgBiT, eEF1A-SmBiT, HaloTag-SmBiT, LgBiT-NS2B-FLAG, or HA-NS3pro-SmBiT for 24 h, and cells were lysed using Glo lysis buffer (Promega). Cell lysates were run on SDS-PAGE gels, and protein was transferred to an Immobilon-P polyvinylidene difluoride (PVDF) membrane (Merck) using the Trans-Blot SD semidry transfer cell (Bio-Rad). Where indicated, Western blots were probed with mouse anti-HIV-1 RT (Santa Cruz Biotechnology), mouse anti-FLAG (Cell Signaling), rabbit anti-HA (Cell Signaling), or mouse anti-eEF1A (Santa Cruz Biotechnology). To detect the SmBiT proteins, the Nano-Glo HiBiT blotting system (Promega) was used as per the manufacturer’s instructions.

Compounds.

During the screening process, compounds were supplied in assay-ready format from Compounds Australia, Griffith Institute for Drug Discovery (GRIDD), Griffith University, Nathan QLD, Australia. One compound from each of the 1,219 scaffolds from the Open Collection Scaffolds library were selected at random and plated into wells of a 96-well white opaque plate (Costar) in 0.5 μl of DMSO at 5 mM by the Assay Ready Plate BioCel system using Echo liquid handling. Each scaffold contains 60 or fewer structure-activity relationship (SAR)-meaningful compounds selected by the GRIDD and Queensland Facility of Advanced Bioinformatics (QFAB). Compound plating was completed 24 to 48 h before use in the NanoBiT assay, and the plates were sealed and stored at room temperature under nitrogen. Once scaffolds of interest were identified using NanoBiT, all SAR-meaningful compounds within that scaffold were provided by Compounds Australia in the same assay-ready format. The following identified “hits” were purchased from ChemDiv in powder form and dissolved in DMSO: compounds D086-0046, D086-0114, D086-0148, D487-0468, D487-0490, G408-0087, G408-2173, G408-2184F, G408-2081, and G408-0073. These compounds were given “C series” codes and are described further in Table 2. Delavirdine mesylate, nevirapine, and raltegravir powder (Sigma-Aldrich) were dissolved in DMSO and diluted to the required concentrations.

NanoBiT.

HEK293T cells were seeded in 10-cm tissue culture plates overnight to reach approximately 95% confluency the following day, and 5 μg each of LgBiT and SmBiT NanoBiT plasmids was transfected using X-tremeGENE HP transfection reagent (Roche) and incubated for 6 h at 37°C. HIV-1 RTp66-LgBiT was cotransfected with eEF1A-SmBiT, and ZIKV LgBiT-NS2B was cotransfected with ZIKV NS3pro-SmBiT. After the 6-h transfection, cells were trypsinized and 40,000 were seeded into wells of 96-well plates containing a 0.5-μl droplet of 5 mM compound or 0.5 μl of DMSO in a 100-μl final volume of Opti-MEM medium to dilute the compound to a final concentration of 25 μM. For the dose-response experiments, compounds were serially diluted in Opti-MEM (Gibco) medium in the 96-well plate before cells were added. Transfected cells were incubated in compounds overnight. NanoBiT substrate was added to cells at a final dilution of 1:100 and incubated at room temperature for 5 min before the luminescence signal was measured using a Biotek Synergy (Winooski, VT) H4 multimode plate reader at approximately 24 h posttransfection.

Purification of eEF1A from HEK293T cells.

Human eEF1A was purified from HEK293T cells using a slightly modified published protocol (36). Briefly, HEK293T cells were lysed in purification buffer (30 mM potassium phosphate [pH 7.5], 1 mM MgCl₂, 15% glycerol, 6 mM β-mercaptoethanol) supplemented with 1 mM phenylmethane sulfonyl fluoride (PMSF) using a Dounce homogenizer. After centrifugation at 10,000 × g and 4°C for 30 min, the supernatant was loaded into a preequilibrated 4-ml HiTrap DEAE column (GE Healthcare) and allowed to flow through. The DEAE column was then washed using purification buffer. The unbound proteins were collected and loaded into a preequilibrated 1-ml HiTrap SP column (GE Healthcare). The SP column was washed using buffer A with increasing concentrations of KCl (50 mM, 100 mM, and 150 mM) and purified protein was then eluted in buffer A containing 300 mM KCl. The elution was concentrated and dialyzed using a Ultra-0.5 centrifugal filter unit with an Ultracel-3K membrane (Amicon). The concentrated protein was stored in dialysis buffer (5 mM NaH₂PO₄/Na₂HPO₄ [pH 7.5], 1 mM MgCl₂, 25% glycerol) at −80°C. The concentration of eluted proteins was determined by the Bradford assay (Bio-Rad), the purity of eEF1A was analyzed by 10% SDS-PAGE with Coomassie blue staining, and eEF1A was detected by Western blotting using a mouse anti-eEF1A antibody (Santa Cruz Biotechnology).

Biolayer interferometry (BLI).

Nickel-nitrilotriacetic acid (Ni-NTA) biosensors (Pall ForteBio, CA) were coated with recombinant 6× His-tagged HIV-1 RTp66/p51 heterodimer purified from E. coli as previously described (21). Super Streptavidin (SSA) biosensors (Pall ForteBio) were coated with eEF1A1 purified from HEK293T cells and biotinylated using EZ-Link-Sulfo-NHS-biotin following the manufacturer’s instructions (Pierce Biotechnology, IL). Sensors for background reference subtraction were coated in recombinant purified glutathione S-transferase (GST)-His for Ni-NTA biosensors or Biocytin (Sigma-Aldrich) for SSA biosensors. Biosensors were incubated in a standard kinetic buffer (1 mM phosphate, 15 mM NaCl, 0.002% Tween 20, and 0.1 mg/ml of gelatin) with 2.5% DMSO or 3.125 μM, 6.25 μM, 12.5 μM, 25 μM, 50 μM, or 100 μM compounds for 120 s to allow association (k_a) and then washed in a standard kinetic buffer for 120 s to allow dissociation (K_D). A double-reference subtraction was performed by subtracting the response for GST-His- or Biocytin-coated sensors in compounds and HIV-1 RT- or eEF1A-coated sensors in buffer only. The experiment was performed at 30°C. The curves were fitted using a 1:1 model, global curve fitting grouped by color, and Rmax (maximal signal response upon saturating binding of the partner to the immobilized protein) linked. The K_D and r² were calculated from the steady-state analysis of responses.

Time-of-addition assay.

HeLa cells were seeded at 20,000 per well in 96-well plates overnight. Cells were infected with equal amounts of VSV-G-pseudotyped HIV-1_{pNL4-3-Δnef-eGFP} by spinoculation at 1,200 × g for 1.5 h at 16°C so that approximately 30 to 50% of cells were infected at 24 hpi (TCID₅₀, 0.6 to 1). Virus medium was removed, cells were washed once with phosphate-buffered saline (PBS), and warm medium supplemented with DMSO, 0.78 μM C7, C8, or C9, 5 μM nevirapine, or 1 μM raltegravir (final DMSO concentration, 0.1% [vol/vol]) was added at the desired times postinfection. Cells were incubated for 24 h and fixed in 1% paraformaldehyde, and eGFP expression was analyzed by flow cytometry. eGFP expression (percent infection) was normalized to the percent infection of the DMSO-only addition sample at each time point, and percent inhibition was calculated at each concentration using the formula 100 × [1 – (normalized percent infected cells of sample – normalized percent infected cells at highest concentration)/(100 – normalized percent infected cells at highest concentration)].

In vitro RT enzymatic activity assay.

The effects of compounds on the in vitro enzymatic activity of RT were determined by incubating 5 μM compounds with recombinant HIV-1 RT, poly(A)·oligo(dT)15 template, and primer for 3 h. The colorimetric reverse transcriptase assay (Roche) was performed as per manufacturer instructions.

Virus infection and luciferase assays.

HEK293T cells were seeded overnight in 12-well plates at 200,000 per well. Medium was replaced the following day with medium containing the indicated concentration of compounds and 8 μg/ml of Polybrene and incubated for 1 h at 37°C. HIV-1_{pNL4-3.Luc.R-E-} with WT, K103N, or Y181C RT was added to the cells and incubated for 29 h before lysis in Glo lysis buffer (Promega). For analysis of reverse transcription of DNA products at 5 hpi in the presence of 5 μM C7, C8, or C9, qPCR was performed as previously described (21).

For single-round infection assays in CD4⁺ T cells, 300,000 human primary activated CD4⁺ T cells were infected with equal amounts of VSV-G-pseudotyped HIV-1_{pNL4-3.Luc.R-E-} for 1 h at 18°C and C7, C8, or C9 was added to a final concentration of 0.195, 0.39, 0.78, or 1.56 μM. Cells were incubated at 37°C for 29 h, cell viability was determined by MTS assay, and cells were lysed in Glo lysis buffer (Promega). The lysate was mixed with firefly luciferase assay substrate (Promega), and luminescence was measured using a Biotek (Winooski, VT) Synergy H4 multimode plate reader and normalized to the DMSO-only treatment control.

For replication assays, human primary activated CD4⁺ T cells were infected with HIV-1_pNL4-3 for 1.5 h at 37°C. Cells were washed four times in PBS, and the day 0 supernatant sample was taken at the last wash. Cells were resuspended in medium and split equally into wells of 24-well plates initially, and when necessary cells were transferred to 12- or 6-well plates. Medium was supplemented with DMSO (1:1,000) or 0.195, 0.293, or 0.39 μM C7. Fresh medium containing C7 was added every 24 or 48 h when required. Cells were incubated at 37°C and counted, cell culture supernatant was collected at 1, 2, 4, and 6 dpi, and p24 was measured by ELISA.

Statistical analysis.

An unpaired Student’s t test was performed for data in Fig. 1E and F. An ordinary two-way analysis of variance (ANOVA) with multiple comparisons was performed for Fig. 7B and C and Fig. 9A to E. An ordinary one-way ANOVA with multiple comparisons comparing each column to DMSO was performed in Fig. 9J to M. The statistical significance of differences between data is indicated as follows: *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.