Antiviral Activity and Resistance Profile of the Novel HIV-1 Non-Catalytic Site Integrase Inhibitor JTP-0157602

ABSTRACT

HIV-1 integrase (IN) is an essential enzyme for viral replication. Non-catalytic site integrase inhibitors (NCINIs) are allosteric HIV-1 IN inhibitors and a potential new class of antiretrovirals. In this report, we identified a novel NCINI, JTP-0157602, with an original scaffold. JTP-0157602 exhibited potent antiviral activity against HIV-1 and showed a serum-shifted 90% effective concentration (EC₉₀) of 138 nM, which is comparable to those of the FDA-approved IN strand transfer inhibitors (INSTIs). This compound was fully potent against a wide range of recombinant viruses with IN polymorphisms, including amino acids 124/125, a hot spot of IN polymorphisms. In addition, JTP-0157602 retained potent antiviral activity against a broad panel of recombinant viruses with INSTI-related resistance mutations, including multiple substitutions that emerged in clinical studies of INSTIs. Resistance selection experiments of JTP-0157602 led to the emergence of A128T and T174I mutations, which are located at the lens epithelium-derived growth factor/p75 binding pocket of IN. JTP-0157602 inhibited HIV-1 replication mainly during the late phase of the replication cycle, and HIV-1 virions produced by reactivation from HIV-1 latently infected Jurkat cells in the presence of JTP-0157602 were noninfectious. These results suggest that JTP-0157602 and analog compounds can be used to treat HIV-1 infectious diseases.

IMPORTANCE Non-catalytic site integrase inhibitors (NCINIs) are allosteric HIV-1 integrase (IN) inhibitors that bind to the lens epithelium-derived growth factor (LEDGF)/p75 binding pocket of IN. NCINIs are expected to be a new class of anti-HIV-1 agents. In this study, we present a novel NCINI, JTP-0157602, which has potent activity against a broad range of HIV-1 strains with IN polymorphisms. Furthermore, JTP-0157602 shows strong antiviral activity against IN strand transfer inhibitor-resistant mutations, suggesting that JTP-0157602 and its analogs are potential agents for treating HIV-1 infections. Structural modeling indicated that JTP-0157602 binds to the LEDGF/p75 binding pocket of IN, and the results of in vitro resistance induction revealed the JTP-0157602 resistance mechanism of HIV-1. These data shed light on developing novel NCINIs that exhibit potent activity against HIV-1 with broad IN polymorphisms and multidrug-resistant HIV-1 variants.

INTRODUCTION

Extensive research has been conducted to develop anti-HIV-1 drugs since the discovery of HIV-1 (1). Multiple classes of HIV-1 drugs have been approved, and combination of antiretroviral therapy (cART) extends the life expectancy of HIV-1-infected patients (2). However, undesirable side effects from drugs and the emergence of resistant HIV-1 strains still pose problems. Furthermore, HIV-1-infected patients must take medication for life because of the persistence of latent proviral reservoirs such as HIV-1-infected resting CD4⁺ T cells and low-level replication under cART in sanctuary sites such as lymphoid tissues and central nervous system, which are less permeable to antiviral drugs (3, 4).

HIV-1 integrase (IN) is an essential enzyme for HIV-1 replication. IN catalyzes the integration of the viral cDNA into a host chromosome after viral cDNA synthesis from HIV-1 RNA by reverse transcriptase (RT) (5). IN has two catalytic functions. First, IN binds to the reverse-transcribed viral cDNA and then removes the CA dinucleotide from the 3′ end of both strands of long terminal repeat (LTR) sequences (3′ processing reaction). Subsequently, IN catalyzes the insertion of the viral cDNA into the host genome (strand transfer [ST] reaction) (6). During integration, IN binds to the host transcriptional coactivator, lens epithelium-derived growth factor (LEDGF)/p75 (7), which functions as a tethering factor to bring the IN-containing preintegration complex to the host chromatin (8). LEDGF/p75 binds with IN via its integrase-binding domain (IBD; residues 347 to 471), which is the minimal domain of LEDGF/p75 required for binding with IN (9). Furthermore, LEDGF/p75 affects the integration site selection (10, 11) and mediates integration into transcriptionally active sites (12–14).

Raltegravir (RAL) and elvitegravir (EVG) are first-generation IN strand transfer inhibitors (INSTIs) that share similar resistance profiles (15, 16). Dolutegravir (DTG) and bictegravir (BIC) are second-generation INSTIs that are effective against most variants with first-generation INSTI-resistant mutations (17, 18). However, patients with HIV-1 containing certain multiple IN mutations display reduced efficacy to DTG treatment (19, 20).

An alternative approach to inhibit IN function involves compounds binding the dimer interface formed by two IN catalytic core domains (CCDs), which is the binding site of LEDGF/p75 (9). These inhibitors were initially discovered as compounds called small molecules binding to the LEDGF/p75 binding site on integrase (LEDGINs) (21). These inhibitors are also termed non-catalytic site integrase inhibitors (NCINIs), allosteric integrase inhibitors (ALLINIs), multimerization selective integrase inhibitors (MINIs), or integrase-LEDGF allosteric inhibitors (INLAIs) (14, 22–30). This class of inhibitors (we use the term NCINIs in this article) that binds to the LEDGF/p75 binding site on IN was subsequently revealed to promote IN multimerization in vitro and impair IN binding to viral RNA (31), resulting in aberrant core morphology (28, 32–36). In the presence of NCINIs, newly produced virions cannot complete reverse transcription in target cells, thereby preventing de novo HIV-1 infection (33, 34). In this report, we identified a novel, potent NCINI with an original scaffold, JTP-0157602, and characterized the in vitro antiviral activity and the resistance profile of the compound.

RESULTS

Inhibition of IN-LEDGF interaction and IN enzyme assay.

NCINIs have been reported to bind to the LEDGF-binding pocket of IN and exhibit antiviral activity (14, 21–28). Here, we identified a novel NCINI, JTP-0157602 (Fig. 1A), that inhibited the interaction between LEDGF IBD and IN with a 50% inhibitory concentration (IC₅₀) of 4.2 nM (Fig. 2). A previously reported NCINI, GS-B, was also evaluated and showed an IC₅₀ of 8.1 nM. JTP-0157602 also inhibited the reaction in an IN assembly/ST assay (i.e., both IN-DNA assembly and the IN ST reaction) with an IC₅₀ of 8.9 nM (the IC₅₀ for GS-B was 12.0 nM) and a maximum inhibition level of approximately 100% (Fig. 2B). In contrast, JTP-0157602 did not completely inhibit the IN ST reaction (a maximum inhibition of ∼30%) (Fig. 2C), suggesting that JTP-0157602 primarily inhibits IN-DNA assembly and not the ST reaction, which is consistent with other NCINIs (22–24, 30).

FIG 1.

Chemical structures of JTP-0157602 (A) and GS-B (B).

FIG 2.

In vitro biochemical profile of JTP-0157602. (A) The LEDGF-IN binding assay was performed using the HTRF method to analyze the inhibitory activity of JTP-0157602 against the binding of His₆-tagged IN-F185H and FLAG-tagged LEDGF IBD. (B and C) Inhibition of the reaction in the IN assembly/ST assay (i.e., both IN-DNA assembly and the IN ST reaction) (B) or inhibition of the ST reaction step (C) was assessed using ELISA. The LEDGF-IN binding assay was conducted in quadruplicate, and the data shown represent mean values ± standard error from three independent experiments. The IN assembly/ST assay and ST assay were conducted in triplicate, and the data shown represent mean values ± standard error from three independent experiments. Error bars represent the standard error from three independent experiments. Filled black circles, JTP-0157602; open circles, GS-B; filled black triangles with dashed line, EVG; open triangles with dashed line, DTG.

Anti-HIV-1 activity of JTP-0157602.

The anti-HIV-1 activity of JTP-0157602 against HIV-1_IIIB was evaluated in a multicycle replication assay using peripheral blood mononuclear cells (PBMCs). JTP-0157602 gave a 50% effective concentration (EC₅₀) of 2.3 nM and an EC₉₀ of 7.4 nM (Table 1). In addition, JTP-0157602 showed a 50% cytotoxic concentration (CC₅₀) of 6,823 nM, and thus, JTP-0157602 demonstrated a favorable selectivity index (SI) of 2,967, while EVG and DTG showed SIs of 7,898 and 2,553, respectively. Next, the effect of human serum (HS) on the antiviral activity of JTP-0157602 was evaluated using H9 cells in the presence of 0%, 5%, 10%, 20%, 30%, or 40% HS, and the fold potency shift with 100% HS was estimated using an extrapolated EC₅₀ in the presence of 100% HS (25, 37, 38). The fold potency shift was predicted to be 18.7-fold, and therefore, the protein-binding adjusted EC₉₀ (PA-EC₉₀) of JTP-0157602 was estimated to be 138 nM, which is comparable to that of EVG or DTG (Table 1).

TABLE 1.

Antiviral activity of JTP-0157602

Antiviral	Activity in PBMCsa				Fold potency shift with 100% HSc	PA-EC90 (nM)d
	HIV-1IIIB		CC50 ± SE (nM)	Selectivity indexb
	EC50 ± SE (nM)	EC90 ± SE (nM)
JTP-0157602	2.3 ± 0.4	7.4 ± 1.0	6,823 ± 484	2,967	18.7	138
EVG	0.61 ± 0.08	2.7 ± 0.2	4,818 ± 675	7,898	65.0	176
DTG	0.51 ± 0.10	2.4 ± 0.2	1,302 ± 319	2,553	84.9	204

^aAnti-HIV-1 activities against HIV-1_IIIB were evaluated using PBMCs. The antiviral assay and the cytotoxicity assay using PBMCs were conducted in triplicate in one experiment, and data shown represent mean values ± standard error (SE) from three different donors.

^bSelectivity index = CC₅₀/EC₅₀.

^cFold potency shift with 100% HS was estimated using an extrapolated EC₅₀ in the presence of 100% HS, which was calculated by using EC₅₀ values evaluated with H9 cells infected with HIV-1_IIIB in the presence of 0%, 5%, 10%, 20%, 30%, or 40% HS. The antiviral assays using H9 cells were conducted in triplicate, and fold potency shift values were calculated using mean EC₅₀s from three independent experiments.

^dPA-EC₉₀ was predicted by multiplying the EC₉₀ in the PBMC assay by the fold potency shift value obtained in the H9 assay.

Anti-HIV-1 activity against recombinant viruses with IN polymorphisms.

HIV-1 IN is a polymorphic protein, and HIV-1 IN residues 124/125, which are located in the NCINI-binding pocket, have many polymorphisms (Table 2) (39–41). Some NCINIs displayed reduced antiviral activity against HIV-1 with residue 124/125 polymorphisms (36). In addition, V165I (proportion of polymorphism for HIV-1 subtype B, 4.9%) and N222K (0.8%) are polymorphisms that were reported to emerge as a result of in vitro resistance selection with NCINIs (24, 25, 35, 40, 42). D167 and K173 also possess polymorphisms (D167E, 5.3%; K173R, 1.2%) and V165/D167/K173 locate near to the LEDGF binding pocket of IN (35, 42, 43). V201I (44.0%) is one of the most frequent polymorphisms of HIV-1 IN (40). A compound that is potent against HIV-1 with multiple kinds of polymorphisms is desirable. Therefore, we determined the anti-HIV-1 activity of JTP-0157602 against eight HIV-1_NL4-3 variants containing residue 124/125 polymorphisms and five other IN polymorphisms (V165I, D167E, K173R, V201I, and N222K). As a result, JTP-0157602 was found to give EC₅₀s in the low nanomolar range (1.6 to 6.0 nM) against all viruses tested, including viruses containing residue 124/125 polymorphisms (Table 3).

TABLE 2.

Polymorphisms in HIV-1 IN amino acids 124/125^a

HIV-1 IN	aa 124/125	% of sequence
		Subtype B	All subtypes
WT	TT	36.4	17.1
Variants
T124A	AT	17.9	11.6
T125A	TA	11.3	8.7
T124N	NT	10.5	5.8
T124A/T125A	AA	7.5	42.6
T124N/T124A	NA	7.3	5.8
T125V	TV	1.8	1.2
T124A/T125V	AV	1.0	0.9
T124S/T125A	SA	0.8	2.0
T124S	ST	0.6	0.4

^aIN coding sequences of HIV-1 subtype B (n = 928) or HIV-1 of all subtypes (n = 2,206) registered in the Los Alamos National Laboratory HIV sequence database in and before 2011 were extracted. IN sequences were aligned, and those without amino acid information at amino acid (aa) 124 and/or aa 125 were excluded. Polymorphisms with proportions of more than 0.5% in HIV-1 subtype B are listed. Recombinant viruses with these polymorphisms except T124A/T125V were generated in the NL4-3 backbone. WT, wild type.

TABLE 3.

Antiviral activities of JTP-0157602 against HIV-1 variants with IN polymorphisms^a

Substitution(s)	EC50 ± SE (nM) (fold change)b
	JTP-0157602	GS-B	EVG	AZT
WTc	3.6 ± 0.6	15 ± 0	1.7 ± 0.4	21 ± 2
T124A	2.6 ± 0.6 (0.7)	7.1 ± 2.4 (0.5)	1.5 ± 0.3 (0.9)	24 ± 4 (1)
T124N	4.7 ± 0.6 (1)	21 ± 6 (1)	2.0 ± 0.4 (1)	25 ± 4 (1)
T124S	3.8 ± 0.6 (1)	15 ± 1 (1)	1.1 ± 0.4 (0.6)	17 ± 2 (0.8)
T125A	5.8 ± 0.4 (2)	28 ± 6 (2)	1.3 ± 0.3 (0.8)	17 ± 4 (0.8)
T125V	1.6 ± 0.2 (0.4)	11 ± 3 (0.7)	0.84 ± 0.15 (0.5)	14 ± 3 (0.7)
T124A/T125A	4.8 ± 1.3 (1)	7.9 ± 2.1 (0.5)	0.90 ± 0.21 (0.5)	13 ± 3 (0.6)
T124N/T125A	5.8 ± 0.9 (2)	39 ± 6 (3)	1.6 ± 0.1 (0.9)	17 ± 2 (0.8)
T124S/T125A	5.5 ± 0.5 (2)	28 ± 7 (2)	2.0 ± 1.0 (1)	20 ± 6 (1)
V165I	5.0 ± 0.6 (1)	24 ± 5 (2)	1.0 ± 0.5 (0.6)	15 ± 3 (0.7)
D167E	3.3 ± 1.6 (0.9)	21 ± 4 (1)	0.73 ± 0.45 (0.4)	6.9 ± 2.9 (0.3)
K173R	6.0 ± 0.8 (2)	19 ± 0 (1)	1.1 ± 0.5 (0.6)	24 ± 5 (1)
V201I	5.1 ± 1.2 (1)	32 ± 14 (2)	1.2 ± 0.3 (0.7)	14 ± 8 (0.7)
N222K	5.2 ± 0.6 (1)	39 ± 11 (3)	1.1 ± 0.3 (0.6)	12 ± 5 (0.6)
CC50 ± SE (nM)	6,840 ± 402	>10,000	4,409 ± 220	>10,000

^aAnti-HIV-1 activities were evaluated in the multicycle replication assay using MT-4 cells. Assays were conducted in duplicate, and the data shown represent mean values ± standard error from three independent experiments.

^bFold changes were calculated as the ratio of EC₅₀ in HIV-1_NL4-3 variants to that of HIV-1_{NL4-3 WT}.

^cHIV-1_{NL4-3 WT} was used as the wild-type (WT) strain.

Anti-HIV-1 activity against recombinant viruses with INSTI-resistant mutations.

INSTIs are used widely in clinical settings, but several INSTI-resistant mutations have been reported in HIV-1-infected patients (19, 20). To determine whether JTP-0157602 sustained antiviral activity against variants with INSTI-resistant mutations, we evaluated the susceptibility of 15 INSTI-resistant viruses to JTP-0157602. EVG, a first-generation INSTI, displayed reduced antiviral activity against most variants with single mutations and those with multiple mutations. The second-generation INSTI DTG remained active against all variants with single mutations tested but showed reduced activity against those with multiple mutations that have emerged in clinical studies of RAL or EVG, such as E138K/Q148K, G140S/Q148H, and L74I/T124A/G140S/Q148H (20, 44, 45), and those that emerged in the VIKING clinical study (E92Q/G140S/Q148H, and G140S/Q148H/N155H) (46) (Table 4). However, none of the INSTI-resistant mutations affected the anti-HIV-1 activity of JTP-0157602 (Table 4).

TABLE 4.

Antiviral activities of JTP-157602 against HIV-1 variants with INSTI resistance-related mutations^a

Substitution(s)	EC50 ± SE (nM) (fold change)b
	JTP-0157602	GS-B	EVG	DTG	AZT
WTc	5.5 ± 0.3	19 ± 1	1.6 ± 0.3	2.0 ± 0.3	23 ± 5
T66I	5.1 ± 0.3 (0.9)	15 ± 1 (0.8)	23 ± 4 (14)	1.0 ± 0.3 (0.5)	40 ± 5 (2)
E92Q	4.4 ± 0.5 (0.8)	17 ± 1 (0.9)	54 ± 22 (34)	4.0 ± 1.2 (2)	29 ± 2 (1)
Y143C	6.1 ± 0.5 (1)	18 ± 1 (0.9)	1.9 ± 0.2 (1)	1.9 ± 0.3 (1)	40 ± 4 (2)
Q148H	3.9 ± 0.9 (0.7)	15 ± 2 (0.8)	4.7 ± 0.6 (3)	0.66 ± 0.20 (0.3)	13 ± 3 (0.6)
Q148K	4.0 ± 0.7 (0.7)	17 ± 1 (0.9)	141 ± 31 (88)	1.4 ± 0.5 (0.7)	18 ± 1 (0.8)
Q148R	4.6 ± 0.6 (0.8)	17 ± 0 (0.9)	132 ± 38 (83)	1.4 ± 0.5 (0.7)	17 ± 1 (0.7)
N155H	4.7 ± 0.5 (0.9)	16 ± 1 (0.8)	37 ± 12 (23)	3.1 ± 1.2 (2)	23 ± 3 (1)
E92Q/N155H	2.5 ± 0.5 (0.5)	16 ± 2 (0.8)	224 ± 53 (140)	3.5 ± 1.1 (2)	16 ± 1 (0.7)
T97A/Y143R	4.4 ± 0.6 (0.8)	14 ± 2 (0.7)	<18 (<11)	3.2 ± 0.8 (2)	39 ± 2 (2)
E138K/Q148K	2.9 ± 0.8 (0.5)	17 ± 2 (0.9)	734 ± 127 (459)	18 ± 4 (9)	19 ± 3 (0.8)
G140S/Q148H	5.1 ± 0.2 (0.9)	25 ± 7 (1)	NDd	11 ± 3 (6)	35 ± 3 (2)
G140S/Q148R	3.9 ± 0.8 (0.7)	17 ± 2 (0.9)	194 ± 30 (121)	6.6 ± 0.2 (3)	14 ± 2 (0.6)
E92Q/G140S/Q148H	5.2 ± 0.3 (0.9)	22 ± 4 (1)	NDd	105 ± 21 (53)	41 ± 6 (2)
G140S/Q148H/N155H	5.3 ± 0.4 (1)	25 ± 6 (1)	NDd	110 ± 20 (55)	29 ± 6 (1)
L74I/T124A/G140S/Q148H	5.6 ± 0.4 (1)	18 ± 1 (0.9)	NDd	28 ± 4 (14)	51 ± 3 (2)
CC50 ± SE (nM)	5,352 ± 373	>10,000	4,197 ± 429	>10,000	>10,000

^aAnti-HIV-1 activities were evaluated in the multicycle replication assay using MT-4 cells. All assays were conducted in duplicate, and the data shown represent mean values ± standard error from three independent experiments.

^bFold changes were calculated as the ratio of EC₅₀ in HIV-1_NL4-3 variants to that of HIV-1_{NL4-3 WT}.

^cHIV-1_{NL4-3 WT} was used as the wild-type (WT) strain.

^dND, not determined.

Resistance selection assay.

In vitro selection of NCINI- and EVG-resistant viruses was performed using the dose escalation method to unveil the in vitro resistance profile of JTP-0157602. HIV-1_IIIB was serially passaged in the presence of each compound in MT-2 cells, and the concentration was increased 2-fold when the infected cells showed an intensive cytopathic effect (CPE). Viral infections were initiated in the presence of each compound at concentrations of 1- or 10-fold their EC₅₀ values. In the experiment starting with 1-fold the EC₅₀ of JTP-0157602, the amino acid substitution A128T was selected in the HIV-1 IN region at passage 4, accompanied by the A124T polymorphic amino acid substitution (Table 5). These substitutions converted to T174T at passage 9, and subsequently A128A/T and T174I were detected at the final passage. In the experiment starting from 10-fold the EC₅₀, T174I also emerged at passage 4, and the concentration increased to ∼2.5 μM with this single mutation. In the selection experiment with GS-B, the substitutions A129T and T174I emerged in the culture starting from both 1-fold and 10-fold the EC₅₀. Resistance selection using EVG starting from 1-fold the EC₅₀ leads to the emergence of the P145S mutation that reportedly confers more than 10-fold resistance to EVG (38). However, virus replication was not observed in the culture starting with 10-fold the EC₅₀.

TABLE 5.

In vitro resistance selection^a

Antiviral	Starting concn (nM)	Passage no.	Days of culture	Compound concn reached (nM)	Amino acid substitution(s)
JTP-0157602	3.9 (EC50 × 1)	P4	19	16 (EC50 × 4)	A124T, A128T
		P9	35	125 (EC50 × 32)	T174I
		P13	50	1,998 (EC50 × 512)	A128A/T, T174I
	39 (EC50 × 10)	P4	19	78 (EC50 × 20)	T174I
		P9	35	1,248 (EC50 × 320)	T174I
		P10	39	2,496 (EC50 × 640)	T174I
GS-B	6.1 (EC50 × 1)	P1	8	6.1 (EC50 × 1)	A124A/T
		P6	24	49 (EC50 × 8)	A124A/T, T174T/I
		P11	43	1,560 (EC50 × 256)	A124A/T, A129A/T, T174T/I
	61 (EC50 × 10)	P1	8	61 (EC50 × 10)	A124A/T, A129A/T, T174T/I
		P6	24	486 (EC50 × 80)	A124A/T, A129A/T
		P11	43	1,956 (EC50 × 320)	A124A/T, T174I, Q285Q/X
EVG	1.6 (EC50 × 1)	P4	19	3.2 (EC50 × 2)	None
		P9	35	6.4 (EC50 × 4)	P145P/S
		P14	54	51 (EC50 × 32)	V31V/I, A124A/T, P145S

^aIn vitro selection of NCINI- and EVG-resistant viruses was performed using the dose escalation method. Assays were performed singly in one experiment with two different starting concentrations. Sequencing results are shown as amino acid substitutions compared to the HIV-1_IIIB reference sequence. Mutations present in more than 25% of the total virus population are listed. Q285X (X = stop codon) was selected in the experiment with GS-B at a starting concentration of 10-fold the EC₅₀. The selection experiment with EVG at a starting concentration of 10-fold the EC₅₀ led to no increase in the drug concentration, and the proviral DNA was not amplified at passage 9.

Susceptibility to variants with NCINI-related mutations.

The antiviral activity of JTP-0157602 toward variants with NCINI-related mutations was determined using recombinant viruses encoding mutations that emerged during resistance selection (A128T, A129T, and T174I). In addition, other mutants (L102F, V126I, E170G, E170K, and H171T) that were selected previously by other NCINIs or LEDGF IBD were constructed (21, 22, 24, 27, 34, 35, 47). A128T and E170G conferred moderate resistance to JTP-0157602 (3-fold for both mutants) (Table 6). The H171T variant also showed reduced susceptibility to JTP-0157602 (9-fold). Moreover, significant decreases in susceptibility to JTP-0157602 of 87-fold and more than 384-fold for the L102F and T174I mutants, respectively, were observed.

TABLE 6.

Antiviral activities of JTP-0157602 against HIV-1 variants with NCINI resistance-related mutations^a

Substitution	EC50 ± SE (nM) (fold change)b
	JTP-0157602	GS-B	EVG	AZT
WTc	5.8 ± 0.1	19 ± 1	1.3 ± 0.3	30 ± 4
L102F	502 ± 27 (87)	3,050 ± 642 (161)	1.0 ± 0.4 (0.8)	18 ± 1 (0.6)
V126I	5.4 ± 0.2 (0.9)	19 ± 1 (1)	1.6 ± 0.2 (1)	44 ± 1 (1)
A128T	19 ± 1 (3)	38 ± 9 (2)	2.8 ± 0.7 (2)	47 ± 1 (2)
A129T	6.8 ± 0.4 (1)	5,740 ± 650 (302)	1.5 ± 0.3 (1)	27 ± 2 (0.9)
E170G	18 ± 0 (3)	68 ± 12 (4)	1.5 ± 0.3 (1)	35 ± 6 (1)
E170K	3.7 ± 0.5 (0.6)	9.9 ± 3.1 (0.5)	1.9 ± 0.2 (1)	45 ± 3 (1)
H171T	54 ± 4 (9)	1,910 ± 41 (101)	1.9 ± 0.5 (1)	27 ± 5 (0.9)
T174I	>2,228 (>384)	7,270 ± 230 (383)	1.6 ± 0.2 (1)	27 ± 4 (0.9)
CC50 ± SE (nM)	5,158 ± 471	>10,000	3,091 ± 606	>10,000

^aAnti-HIV-1 activities were evaluated in the multicycle replication assay using MT-4 cells. All assays were conducted in duplicate, and the data shown represent mean values ± standard error from three independent experiments.

^bFold changes were calculated as the ratio of EC₅₀ in HIV-1_NL4-3 variants to that of HIV-1_{NL4-3 WT}.

^cHIV-1_{NL4-3 WT} was used as the wild-type (WT) strain.

Structural modeling of the interaction between the IN central core domain and JTP-0157602.

We performed structural modeling of the HIV-1 IN–JTP-0157602 complex (Fig. 3) using the structure of the IN-CCD monomer in complex with GS-B (PDB ID 4E1N) (24). According to the modeling, the carboxylic acid of JTP-0157602 forms hydrogen bonds with the side chain oxygen of T174 and the backbone amide groups of E170 and H171. Furthermore, A128 was found to locate in the vicinity of the indole core of JTP-0157602. The cyclohexene ring of JTP-0157602 occupies a region in close proximity to the side chain of L102, which is located at the bottom of the LEDGF-binding pocket. These structural observations are consistent with the measured decrease in antiviral activity of JTP-0157602 against these mutants.

FIG 3.

Structural modeling of HIV-1 IN with JTP-0157602. Magenta dots represent hydrogen bonds between the carboxylic acid of JTP-0157602 with (i) the side chain oxygen of T174, (ii) the main chain amide group of E170, and (iii) the main chain amide group of H171.

Inhibitory activity against the late-phase replication cycle of HIV-1.

The inhibitory mechanism of JTP-0157602 was further elucidated by determining the antiviral activity during the early phase of the HIV-1 life cycle using HeLa-CD4/LTR-β-gal (MAGI)/LTR-NanoLuc indicator cells that exhibit NanoLuc luciferase activity under the control of the HIV-1 LTR promoter. EVG and 3′-azido-3′-deoxythymidine (AZT; nucleotide reverse transcriptase inhibitor) were used as control compounds to selectively inhibit the early phase of the replication cycle (22). EVG and AZT gave comparable EC₅₀s both in the early phase of the replication assay and in the multicycle replication assay. Darunavir (DRV; protease inhibitor) was used as a control to selectively inhibit the late phase of replication, and DRV did not exhibit any inhibition during the early phase, as expected (22, 48). Moderate early-phase inhibitory activity of JTP-0157602 was detected, with the EC₅₀ reduced by ∼500-fold in comparison with the results from the multicycle replication assay (Fig. 4).

FIG 4.

Inhibition of the early phase of the HIV-1 replication cycle. MAGI/LTR-NanoLuc cells were infected with HIV-1_{NL4-3 WT} at an MOI of 0.0003 in the presence of DEAE-dextran (20 μg/mL) for 2 h. The medium was replaced with medium containing each compound, and the culture plates were incubated for a further 2 days. NanoLuc luciferase activity was analyzed. The assay was conducted in triplicate, and the data shown represent mean values ± standard error from three independent experiments. Error bars represent the standard error from three independent experiments. Filled black circles, JTP-0157602; open circles, GS-B; filled black triangles, EVG; gray circles, AZT; gray triangles, DRV.

Inhibition of the late phase of the replication cycle by JTP-0157602 was examined. J_NLLat cells, an HIV-1 latency model cell line, were reactivated with tumor necrosis factor alpha (TNF-α) or romidepsin in the presence of each compound, and then the infectivity of the viruses produced by reactivation was determined by infecting MAGI/LTR-NanoLuc cells with these viruses (Fig. 5). EVG and AZT did not exhibit inhibitory activities, whereas JTP-0157602 showed a potency of inhibition similar to that in the multicycle replication assay (Fig. 5A). Next, the p24 amount and RT activity in culture supernatants from reactivated J_NLLat cells were evaluated (Fig. 5B and C). Although DRV inhibited both the amount of p24 and the RT activity, JTP-0157602 did not show significant inhibition in either assay. These results suggest that JTP-0157602 inhibits the late phase of HIV-1 replication and suppresses the infectivity of newly produced virions without affecting their p24 amount and RT activity.

FIG 5.

Impaired infectivity of HIV-1 produced by reactivation from latently infected Jurkat cells. (A) J_NLLat cells (Jurkat cells latently infected with HIV-1_{NL4-3 WT}) were reactivated with TNF-α (10 ng/mL) or romidepsin (100 nM) in the presence of each compound. Culture supernatants were collected as newly produced viruses. After concentration, the virus particles were used to infect MAGI/LTR-NanoLuc cells, and NanoLuc luciferase activity was analyzed. (B and C) J_NLLat cells were reactivated with TNF-α (10 ng/mL) or romidepsin (100 nM) in the presence of each compound. Culture supernatants were collected as virus preparations, and then the RT activity (B) and the amount of p24 (C) were quantified. All assays were conducted in triplicate, and the data shown represent mean values ± standard error from three independent experiments. Error bars indicate the standard error from three independent experiments. Filled black circles, JTP-0157602; open circles, GS-B; filled black triangles, EVG; gray circles, AZT; gray triangles, DRV.

DISCUSSION

In this report, we characterized the in vitro profile of a novel NCINI, JTP-0157602. The EC₉₀ of JTP-0157602 was 7.4 nM in PBMCs, and the PA-EC₉₀ was predicted to be 138 nM, which is comparable to that of FDA-approved INSTIs (Table 1). The mechanism of action of JTP-0157602 was also investigated. JTP-0157602 demonstrated stronger antiviral activity during the late phase of the replication cycle than in the early phase (Fig. 4 and 5A), which is consistent with other NCINIs (49).

Additionally, JTP-0157602 exhibited a preferable antiviral activity profile against a wide variety of HIV-1_NL4-3 variants containing amino acid polymorphisms at residues 124/125 (39–41), which affects the susceptibility to some NCINIs (36). Also, V165I, D167E, and K173R polymorphisms (40), which are located near the LEDGF-IN binding pocket, were found not to affect susceptibility to JTP-0157602. Thus, JTP-0157602 has potent activity against a wide variety of HIV-1 polymorphisms (i.e., fold change of ≤2 against all polymorphisms tested), indicating that JTP-0157602 should be effective against a broad range of HIV-1 strains.

INSTI-based regimens are recommended for antiretroviral-naive patients, but it has been reported that the accumulation of multiple mutations in clinical trials leads to resistance to the second-generation INSTI DTG (19, 20). Therefore, the susceptibility of these INSTI-resistant mutants to JTP-0157602 was investigated. JTP-0157602 was shown to be fully active against these INSTI-resistant mutants, indicating that JTP-0157602 should be effective for HIV-1-infected patients in which an INSTI-based treatment regimen has failed.

The resistance selection experiment starting from 1-fold the EC₅₀ of JTP-0157602 led to the emergence of A128T initially. Subsequently, the proportion of T174I increased at later passages probably because T174I (>384-fold change) confers severe resistance to JTP-0157602. The resistance selection experiment starting from 10-fold the EC₅₀ of JTP-0157602 also resulted in the emergence of T174I, which may reflect the large fold change. A128 and T174 are highly conserved across multiple subtypes of HIV-1 (41) and were reported previously to be selected by other NCINIs (21, 24, 25, 35). Structural modeling of HIV-1 IN-CCD with JTP-0157602 was performed, and these two amino acids are located at the LEDGF-binding pocket of IN (Fig. 3). The structural model revealed that the carboxylic acid of JTP-0157602 forms a hydrogen bond with the side chain oxygen of T174, indicating that this hydrogen bond is lost by the T174I mutation, which weakens the interaction of JTP-0157602 with IN. The indole ring of JTP-0157602 is located close to A128, suggesting that the A128T mutation hampers the binding of the compound to IN. In addition, L102F conferred a large decrease in the antiviral activity of JTP-0157602. The cyclohexene ring of JTP-0156702 occupied a region in close proximity to the side chain of L102, which is positioned at the bottom of the LEDGF-binding pocket. Thus, the L102F substitution, a change to a larger side chain, may impede binding of the compound with IN. Collectively, the structural model explains the possible mechanism of resistance acquisition.

In conclusion, JTP-0157602 has potent antiviral activity against HIV-1, including variants with broad IN polymorphisms. Induced in vitro drug resistance and structural modeling revealed the JTP-01507602-resistant mechanisms of HIV-1 IN. JTP-0157602 showed no cross-resistance with INSTIs. Therefore, JTP-0157602 and its analog compounds represent potential options for treating HIV-1 infections. These data shed light on developing NCINIs which exhibit potent activity against HIV-1 with broad IN polymorphisms and multidrug-resistant HIV-1 variants.

MATERIALS AND METHODS

Antiviral agents.

JTP-0157602 (enantiomer), GS-B (24, 34), EVG (50, 51), and DTG (52) were synthesized at the Central Pharmaceutical Research Institute of Japan Tobacco, Inc. (Osaka, Japan). The structures of JTP-0157602 and GS-B are shown in Fig. 1. AZT was purchased from Sigma-Aldrich (St. Louis, MO). DRV was purchased from Toronto Research Chemicals (North York, Canada). Spectral data (1H NMR) for JTP-0157602 was obtained to confirm compound identity. Experimental details of chemical synthesis and a copy of nuclear magnetic resonance (NMR) spectra for JTP-0157602 are included in the supplemental material.

Viruses.

As a laboratory strain, HIV-1_IIIB was used in this study. The HIV-1 molecular clone pNL4-3 (Southern Research Institute, Frederick, MD) was used to generate recombinant HIV-1 variants (53). HIV-1_NL4-3 with IN polymorphisms, INSTI resistance-related mutations, or NCINI resistance-related mutations were introduced by site-directed mutagenesis (25) or mutagenesis using an in-fusion technique (54), except for pNL4-3 T66I, E92Q, and N155H (Southern Research Institute). Site-directed mutagenesis was performed with QuikChange II site-directed mutagenesis kits (Agilent Technologies, Santa Clara, CA), and mutations were introduced into a IN_NL4-3-containing vector. Then, IN_NL4-3 carrying each substitution(s) was cloned into pNL4-3 using the AgeI-SalI restriction site. The pNL4-3 mutants were transfected into 293T cells. After 48 h of incubation, the supernatants were stored as viral stocks at −80°C. The viral titer was determined as reported previously (55) with some modifications. Briefly, MT-4 cells were infected with serial log dilutions of virus stocks, and the 50% tissue culture infectious dose (TCID₅₀) was determined by measuring the RT activity of the culture supernatant at day 5. The TCID₅₀ was calculated by the Spearman-Karber method (55).

Plasmid.

A NanoLuc expression plasmid, pNL3.2 NF-κB-RE (NlucP/NF-κB-RE/Hygro) (Promega, Madison, WI), was modified as described below and used to introduce LTR and NanoLuc-PEST into MAGI cells (Immuno Diagnostics, Inc., Woburn, MA) (56). Since hyg^r had already been introduced into the MAGI cell line, hyg^r (BamHI to SalI in pNL3.2 NF-κB-RE [NlucP/NF-κB-RE/Hygro]) was replaced with the coding sequence of a puromycin resistance gene (BamHI to SalI in pGL4.20 [Luc2/puro]; Promega). The resultant plasmid was designated pNL3.2-puro. Next, the 5′ LTR sequences (634 bp) of pNL4-3 (GenBank accession number AF324493.2) with an 18-bp extension at its 5′ end, which is homologous to the KpnI cleavage site of pNL3.2-puro, and a 15-bp extension at its 3′ end, which is homologous to the 15-bp overhang at the 5′ end of the NanoLuc PCR product, were amplified with two primers (LTR-Fw, 5′-GGC CTA ACT GGC CGG TAC TGG AAG GGC TAA TTT GGT CCC A-3′; LTR-Rv, 5′-CAG TAC CGG ATT GCC TGC TAG AGA TTT TCC ACA CT-3′). The NanoLuc-PEST sequences of pNL3.2 with a 15-bp extension at its 5′ end, which is homologous to the 15-bp overhang of the 3′ end of the LTR PCR product, and a 20-bp extension at its 3′ end, which is homologous to the FseI cleavage site of pNL3.2, were also amplified with two primers (Nanoluc_Fw, 5′-GGC AAT CCG GTA CTG TTG GTA AAG CCA CCA T-3′; NanolucP_Rv, 5′-TGT CTG CTC GAA GCG GCC GGC CGC CCC GAC TCT AG-3′). Then, the two PCR products of LTR and NanoLuc-PEST sequences were simultaneously inserted into pNL3.2-puro by the In-Fusion HD kit (TaKaRa, Shiga, Japan) after removal of the original NanoLuc-PEST sequence from pNL3.2 using KpnI-FseI restriction sites. The resultant plasmid (pNL3.2-puro-LTR-NanoLuc-PEST) was introduced into MAGI cells as described below.

Cells.

MAGI cells were used to generate an indicator cell line, MAGI/LTR-NanoLuc, which was designed to express NanoLuc luciferase (Promega) under the control of the HIV-1 LTR promoter. pNL3.2-puro-LTR-NanoLuc-PEST was transfected into MAGI cells (56), and the diluted cells were cultured under puromycin selection to obtain a cell clone that was stably introduced with LTR-NanoLuc. Then, a clone that showed high NanoLuc luciferase activity after HIV-1 infection was selected as the MAGI/LTR-NanoLuc cell line.

An HIV-1 latency model cell line, J_NLLat, was originally established as follows. Briefly, Jurkat cells were infected with HIV-1_{NL4-3 WT} at a multiplicity of infection (MOI) of 1, and then limiting dilution was performed. Cloned cells were stimulated with latency-reversing agents (LRAs), including TNF-α and romidepsin. One clone that showed low-level expression of HIV-1 without LRA treatment but exhibited substantial HIV-1 expression after LRA treatment was selected as the J_NLLat cell line.

H9, MT-2, MT-4, and J_NLLat cells were grown in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 μg/mL streptomycin. 293T cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS, 100 U/mL penicillin, and 100 μg/mL streptomycin. MAGI/LTR-NanoLuc cells were grown in DMEM supplemented with 10% newborn calf serum, 100 U/mL penicillin, 100 μg/mL streptomycin, 200 μg/mL G418, 100 μg/mL hygromycin B, and 0.1 μg/mL puromycin. PBMCs (Biological Specialty Corporation, Colmar, PA) in RPMI 1640 medium containing 20% FBS, 100 U/mL penicillin, 100 μg/mL streptomycin, and 20 ng/mL human interleukin-2 (PeproTech, Cranbury, NJ) were stimulated with 5 μg/mL phytohemagglutinin (PHA) (Sigma-Aldrich) for 3 days and used for subsequent experiments.

Structural modeling.

Structural modeling of HIV-1 IN with JTP-0157602 was performed. All calculations were carried out with Schrodinger software (version 2020-3; Schrodinger, LLC, New York, NY), and figures were prepared using PyMOL (Schrodinger, LLC; http://www.pymol.org). The crystal structure of the IN-CCD monomer with bound GS-B (PDB ID 4E1N) (24) was downloaded from the Protein Data Bank, and the IN-CCD dimer structure was generated with PyMOL. The docking template of the IN-CCD dimer structure was prepared using the Protein Preparation Wizard in Maestro (Schrodinger, LLC) with default options. The starting conformation of JTP-0157602 for Glide docking was prepared by a MacroModel conformational search with default settings to obtain a global minimum conformation of the compound. All dockings were performed in Glide using the SP mode and default settings. Grids for docking were prepared using the docking template described above by selecting the bound ligand to define the binding site with a length of 30 Å. The docking pose with the best docking score was obtained and further optimized using the MacroModel embrace minimization function (residues 4 Å from the ligand were allowed to move freely, residues an additional 4 Å from the ligand were tethered, and residues more than an additional 8 Å from the ligand were fixed). The obtained minimized pose was selected for generating PyMOL figures.

LEDGF-IN binding assay.

The LEDGF-IN binding assay was performed as reported previously (57) with some modifications. The homogeneous time-resolved fluorescence (HTRF) assay was carried out using a 384-well plate in reaction buffer consisting of 25 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM MgCl₂, 0.01% Tween 20, and 0.1% bovine serum albumin (BSA). His₆-tagged IN containing the F185H mutation (IN-F185H; mutated IN with improved solubility) (30 nM final concentration) (58, 59) was preincubated with compounds for 45 min at room temperature. Next, the FLAG-tagged LEDGF IBD (10 nM final concentration) and a mixture of anti-FLAG-M2-Tb cryptate and anti-His₆ d2 antibody (Cisbio, Codolet, France) (0.65 nM and 30 nM final concentrations, respectively) was added to the plate. Finally, the plate was incubated for a further 3 h at room temperature, and the luminescence was measured using an EnVision plate reader (Perkin Elmer, Waltham, MA).

IN enzymatic assay.

An IN assembly/ST assay, which evaluates the inhibitory activity in both IN-DNA assembly and the ST reaction, was performed according to the method described previously (60), with some modifications. Reactions were performed using a 96-well plate and a reaction buffer consisting of 30 mM morpholinepropanesulfonic acid (MOPS; pH 7.2), 5 mM MgCl₂, 3 mM dithiothreitol (DTT), 0.1 mg/mL BSA, 5% glycerol, 0.01% Tween 20, and 10% dimethyl sulfoxide (DMSO). 3′-end-processed donor DNA containing biotin at the 5′ end was prepared by annealing D1 (5′-biotin-ACC CTT TTA GTC AGT GTG GAA AAT CTC TAG CA-3′) and D2 (5′-ACT GCT AGA GAT TTT CCA CAC TGA CTA AAA G-3′) (complementary sequences are underlined) and then immobilized on streptavidin-coated 96-well plates. Compounds and IN-F185H (60 nM final concentration) were added and incubated for 60 min at 37°C. Plates were washed, and the ST reaction was conducted for 20 min at 37°C by adding a target DNA containing digoxigenin at the 3′ end (2.5 nM final concentration) that was prepared by annealing T1 (5′-TGA CCA AGG GCT AAT TCA CT-digoxigenin-3′) and T2 (5′-AGT GAA TTA GCC CTT GGT CA-digoxigenin-3′) (complementary sequences are underlined). Thereafter, the conjugate of target and donor DNA was quantified by detecting digoxigenin using an Infinite 200 microplate reader (TECAN, Männedorf, Switzerland).

For the IN ST assay, which evaluates the inhibitory activity only in the ST reaction, the experiment was performed in the same manner except that the compounds were added to the plate after incubation of IN-F185H with donor DNA for 60 min and that the plates were washed subsequently.

Antiviral assay in PBMCs.

The antiviral assay was performed as described previously (61), with some modifications. PHA-stimulated PBMCs were infected with HIV-1_IIIB at an MOI of 0.001 for 2 h. The cells were then washed and plated onto 96-well culture plates (8 × 10⁴ cells/200 μL) with each compound for 7 days. Ninety microliters of culture supernatant was collected. Fifteen microliters of the supernatant and 10 μL of reaction buffer consisting of 16 mM Tris-HCl (pH 7.4), 13 mM MgCl₂, 16 mM DTT, 0.8% Triton X-100, 40 mM EGTA, 0.08 mg/mL poly(rA), 0.04 mg/mL oligo(dT)₁₅, and 67 μCi/mL [³H]dTTP were added to the 96-well plate. The plate was incubated at 37°C for 2 h, and the reaction was terminated by adding 150 μL of stop buffer consisting of 1% trichloroacetic acid and 0.1% sodium pyrophosphate. Reverse-transcribed DNAs radiolabeled with [³H]dTTP were trapped on a UniFilter-96 GF/C, a microplate with glass fiber filters (PerkinElmer), and the radioactivity was measured using TopCount NXT (PerkinElmer). The EC₅₀ and EC₉₀ were then calculated.

Antiviral assay in H9 cells.

The antiviral assay was performed as reported previously (38), with some modifications. Briefly, H9 cells were infected with HIV-1_IIIB at an MOI of 0.05 in medium containing 0%, 5%, 10%, 20%, 30%, or 40% HS (Interstate Blood Bank, Memphis, TN). Cells were washed and resuspended again in medium containing the respective concentrations (0% to 40%) of HS. Cells were plated onto 96-well culture plates (1 × 10⁴ cells/200 μL) with each compound for 5 days. Then, 90 μL of culture supernatants was collected, and the RT activity was measured according to the method described for the PBMC assay. The EC₅₀ values in the presence of the respective HS concentrations were calculated.

Effect of HS on antiviral activity.

The effect of HS on the antiviral activity was estimated according to a method reported previously (38), with some modifications. The reported EC₅₀ values in the presence of 0% to 40% HS (see above) were used to extrapolate to the EC₅₀ in the presence of 100% HS. The fold potency shift was calculated by dividing the extrapolated EC₅₀ in the presence of 100% HS by the EC₅₀ in the absence of HS. The PA-EC₉₀ was estimated by multiplying the EC₉₀ in the PBMC assay by the fold potency shift value determined from the H9 assay.

Antiviral assay in MT-4 cells.

The antiviral assay was performed as reported previously (62), with some modifications. Briefly, MT-4 cells (5 × 10³ cells) were infected with HIV-1_{NL4-3 WT} or its variants at an MOI of 0.05 and incubated in 96-well culture plates at 37°C for 5 days with each compound. After incubation, CellTiter-Glo reagent (Promega) was added to determine cell viability, and the chemiluminescence was measured using a SpectraMax L microplate reader (Molecular Devices, San Jose, CA). The EC₅₀ was then calculated.

In vitro resistance selection experiment.

Resistance selections were performed by using the dose-escalating method (51), and the emerged mutations were analyzed by population-based sequencing (63–65) as reported previously, with some modifications. MT-2 cells were infected with HIV-1_IIIB (MOI of 0.05) in the presence of each compound at a starting concentration of 1-fold or 10-fold the EC₅₀. The concentration of the compound was increased 2-fold when an extensive CPE was observed. Genomic DNA including proviral DNA was extracted from the infected cells using DNAzol (ThermoFisher Scientific, Waltham, MA). The IN region of proviral DNA was then amplified with first-PCR primers INT-Fw (5′-GTA AAC ATA GTA ACA GAC TCA CAA T-3′ [corresponding to positions 4026 to 4050]) and INT-Rv (5′-CTC CTG TAT GCA GAC CCC AAT ATG T-3′ [corresponding to positions 5242 to 5266]), followed by the second-round nested PCR with primers Seq5-0 (5′-ATG CAT TAG GAA TCA TTC AA-3′ [corresponding to positions 4051 to 4070]) and Seq3-1 (5′-GGA TGT GTA CTT CTG AAC TT-3′ [corresponding to positions 5193 to 5212]). Primer positions correspond to the positions in HIV-1_HXB2 (GenBank accession number K03455.1). Second-round PCR products were then sequenced with an ABI Prism 3100 genetic analyzer (Applied Biosystems, Foster City, CA) using a BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems).

Inhibition of early-phase replication.

MAGI/LTR-NanoLuc cells (1.5 × 10⁴ cells) were seeded onto 96-well culture plates and incubated at 37°C for 1 day. The culture medium was replaced with a new medium containing DEAE-dextran (final concentration, 20 μg/mL), and the cells were infected with HIV-1_{NL4-3 WT} at an MOI of 0.0003 for 2 h. Thereafter, the virus-containing medium was washed out, and the culture medium containing compounds was added. The plates were incubated at 37°C for 48 h. Then, the Nano-Glo luciferase assay reagent (Promega) was added, and the chemiluminescence was measured using a SpectraMax L microplate reader (Molecular Devices).

Inhibition of late-phase replication.

J_NLLat cells (1.0 × 10⁶ cells) were seeded onto 6-well culture plates. The cells were then preincubated with each compound at 37°C for 3 h. After the preincubation, TNF-α (R&D Systems, Minneapolis, MN) or romidepsin (Sigma-Aldrich) (10 ng/mL or 100 nM final concentration, respectively) was added to the plates to stimulate HIV-1 gene expression, and the plates were cultured at 37°C for 16 h. The cell-free supernatants were diluted with PBS and concentrated using Amicon Ultra-4 100-kDa centrifugal filter devices (Millipore, Darmstadt, Germany) to remove free compounds (34). This process was repeated two times to give an estimated 2,200 dilution of compounds. The virus preparations were stored at −80°C until use. MAGI/LTR-NanoLuc cells (1.5 × 10⁴ cells) were seeded onto 96-well culture plates and incubated at 37°C for 1 day. The concentrated virus preparations were further diluted 50-fold and used to infect MAGI/LTR-NanoLuc cells. The infectivity was quantified as described above.

RT assay and p24 ELISA.

J_NLLat cells were preincubated with each compound at 37°C for 3 h, and the cells were treated with TNF-α or romidepsin as described above. The cell-free supernatants were then collected. The RT assay was performed according to the method described for the PBMC assay except that the reaction buffer consisted of 9.7 mM Tris-HCl (pH 7.4), 7.8 mM MgCl₂, 9.7 mM DTT, 0.5% Triton X-100, 24.3 mM EGTA, 0.05 mg/mL poly(rA), 0.03 mg/mL oligo(dT)₁₅ and 40 μCi/mL [³H]dTTP. The RT reaction was performed at 37°C for 2 h. The amount of p24 was quantified using RETROtek HIV-1 p24 antigen ELISA 2.0 (ZeptoMetrix Corporation, Buffalo, NY) according to the manufacturer’s protocol, and the absorbance was measured using a VMax microplate reader (Molecular Devices).