Small-molecule inhibitors of the LEDGF/p75 binding site of integrase block HIV replication and modulate integrase multimerization

Abstract

Targeting the HIV integrase (HIV IN) is a clinically validated approach for designing novel anti-HIV therapies. We have previously described the discovery of a novel class of integration inhibitors, 2-(quinolin-3-yl)acetic acid derivatives, blocking HIV replication at a low micromolar concentration through binding in the LEDGF/p75 binding pocket of HIV integrase, hence referred to as LEDGINs. Here we report the detailed characterization of their mode of action. The design of novel and more potent analogues with nanomolar activity enabled full virological evaluation and a profound mechanistic study. As allosteric inhibitors, LEDGINs bind to the LEDGF/p75 binding pocket in integrase, thereby blocking the interaction with LEDGF/p75 and interfering indirectly with the catalytic activity of integrase. Detailed mechanism-of-action studies reveal that the allosteric mode of inhibition is likely caused by an effect on HIV-1 integrase oligomerization. The multimodal inhibition by LEDGINs results in a block in HIV integration and in a replication deficiency of progeny virus. The allosteric nature of LEDGINs leads to synergy in combination with the clinically approved active site HIV IN strand transfer inhibitor (INSTI) raltegravir, and cross-resistance profiling proves the distinct mode of action of LEDGINs and INSTIs. The allosteric nature of inhibition and compatibility with INSTIs underline an interest in further (clinical) development of LEDGINs.

Fig 1

Inhibition of recombinant HIV-1 IN strand transfer activity. Inhibition of the integrase strand transfer reaction by elvitegravir (○), raltegravir (■), and CX14442 (●) was measured under assay conditions where integrase was preincubated first with either LTR (left column) or compound (right column). The assays were performed using either Mg²⁺ (upper panel) or Mn²⁺ (lower panel) in the reaction buffer. (A) Graphic analysis of the inhibition. The compound concentration is plotted against % inhibition, which is determined by comparison to the amount of inhibition produced by 1 μM elvitegravir. Curves are representative of means ± SEM from 3 independent experiments. (B) Table with mean inhibitory constants and the 95% confidence interval. Superscript letters indicate the following: a, concentration (nM) required to inhibit in vitro protein-protein interaction by 50%, or IC₅₀ (95% CI); b, elvitegravir; c, raltegravir; d, CX14442.

Fig 2

LEDGINs increase the stability of HIV IN dimers. (A) Analysis of melting temperatures of HIV-1 integrase without a small-molecule ligand (▼; T_m = 48.1°C) and with CX05168 (●; T_m = 54.5°C), CX05045 (■; T_m = 57.3°C), or CX14442 (▲; T_m = 62.5°C) demonstrates that with increasing potency of compounds, the melting temperature is shifted to higher temperatures, indicating a stabilization of the HIV IN upon ligand binding. A representative experiment is shown. (B) Integrase dimerization assay: CX05168 (●; EC₅₀ = 1,134 ± 63 nM), CX05045 (■; EC₅₀ = 309 ± 178 nM), and CX14442 (▲; EC₅₀ = 123 ± 1 nM) increase the population of HIV IN dimers. The data represent triplicate experiments with averages and standard deviations.

Fig 3

Time of addition (TOA). After infection of MT-4 cells with HIV-1-IIIB inhibitors were added at indicated time points spanning from 1 to 26 h postinfection. Virus replication was determined by p24 antigen determination in the supernatant at 30 h after infection. We plotted relative inhibition (%) compared to results for the noninhibited replication control. We added reverse transcriptase inhibitors, such as zidovudine and tenofovir (NRTI), and inhibitors of late replication steps, such as the protease inhibitor ritonavir. CX14442 loses its inhibitory capacity if added later than 9 h after infection, matching the profile observed for raltegravir and elvitegravir.

Fig 4

LEDGINs impact HIV-1 infectivity without affecting virus production from chronically infected cells. (A) At 6 days post-addition of the 10× EC₅₀ of individual compounds (DMSO, noncompound control; ral, raltegravir; rit, ritonavir; CX05045, LEDGIN), the amount of p24 in the supernatant of chronically infected HUT78 cells was quantified by p24 ELISA. (B) The infectivity of virus produced in the presence 10× EC₅₀ of the indicated compounds was monitored by CPE scoring in MT4 cells. At 5 days postinfection, wells containing infected cells were scored for the presence of syncytia, and the TCID₅₀ was determined. Data represent mean values ± standard deviations of the TCID₅₀ as calculated by the Spearman-Karber equation (n = 3). The TCID₅₀ values were normalized to those obtained for the DMSO-treated cells, set as 100%. Bars represent the means of percentage values ± SD for 3 independent experiments.

Fig 5

LEDGINs are active against a broad spectrum of HIV-1 subtypes. Twenty-five different HIV-1 isolates from patients belonging to the HIV-1 subtype classes A, A1, AE, AG, B, BF, C, and D were analyzed for their susceptibilities to CX05045 (light gray) and raltegravir (dark gray) using the PhenoSense assay (Monogram Biosciences). The EC₅₀ for each compound was determined, and the fold change from the wild-type EC₅₀ was calculated.

Fig 6

MacSynergy II analysis of the anti-HIV-1 activity of LEDGINs in combination with raltegravir. (A) Synergy plot displaying the synergy at each concentration combination of CX14442 and raltegravir. The synergy ratio was calculated from the plot and is a representative experiment in which 4 replicates of each combination were present on each microtiter plate. Three independent experiments were conducted, and all gave similar values. (B). Mean synergy values with standard deviations from 3 independent experiments.

Fig 7

Cross-resistance profile of LEDGINs and INSTIs. (A) The central panel shows a cartoon of the IN catalytic core domain (CCD) dimer, with one monomer colored pale green and the other pale yellow. Gray and yellow squares indicate the locations of the IN catalytic and LEDGF/p75 binding sites, respectively. The panels on either side are colored accordingly and present close-up views of both. The IN active site (left, gray panel) is depicted with 3′-processed DNA (green) and Mg²⁺ (purple) bounds, as predicted by homology modeling based upon the prototype foamy virus (PFV) intasome (21). The catalytic DDE residues are shown as lines, while assayed resistance mutations are indicated as sticks colored by atom (red, carbon atoms). The LEDGF/p75 binding site (right, yellow panel) is rendered again in pale green and yellow colors for the CCD monomers and in red sticks colored by atom for the assayed resistance mutations. The HIV-1 intasome homology model was built using the software program MODELLER (39), based upon the corresponding PFV intasome structure (PDB ID 3L2R). The structures of the IN CCD dimers were taken from Protein Data Bank (PDB) identifiers 3L3U (40) and 2B4J (41). All structures were visualized in the software program PyMol (42). (B) Susceptibilities of HIV-1 strains containing integrase resistance mutations to various inhibitors. Data are represented as fold changes in EC₅₀ from those for the wild type. All compounds beginning with CX are LEDGINs. PF-03450074 is an HIV-1 capsid inhibitor. Data are presented as means from 3 independent experiments, with upper and lower 95% confidence intervals in parentheses. Superscript letters indicate the following: a, LEDGIN-resistant mutants; b, raltegravir-resistant mutants.