Long Dissociation of Bictegravir from HIV-1 Integrase-DNA Complexes

The HIV integrase (IN) strand transfer inhibitor (INSTI) bictegravir (BIC) has a long dissociation half-life (t1/2) from wild-type IN-DNA complexes: BIC 163 h > dolutegravir (DTG) 96 h > raltegravir (RAL) 10 h > elvitegravir (EVG) 3.3 h. In cells, BIC had more durable antiviral activity against wild-type HIV after drug washout than RAL or EVG.

H IV integrase (IN) is essential for viral replication, and the IN strand transfer inhibitors (INSTIs) bictegravir (BIC), dolutegravir (DTG), elvitegravir (EVG), and raltegravir (RAL) are potent antiretroviral drugs (1). All currently approved INSTIs share a pharmacophore consisting of a metal binding scaffold that effectively chelates the two activesite catalytic Mg 21 ions and prevents terminal viral nucleotide 39-OH nucleophilic attack on the host DNA, thereby blocking integration (Table 1). These INSTIs show potent inhibition of HIV-1 replication in cells; however, there are distinct differences in their chemical structures, interaction with HIV-1 IN-DNA complexes, resistance profiles, and dissociation rates from complexes of IN bound to double-stranded DNA (IN-DNA complexes) (2)(3)(4)(5)(6)(7)(8)(9). The apparent dissociation rate of DTG from IN-DNA complexes was previously shown to be longer than those of RAL and EVG and was predicted to correlate with potent antiretroviral activity and a higher genetic barrier to resistance, but direct comparison to BIC has not been investigated (2,3). Here, the dissociation halflife (t 1/2 ) of BIC and other INSTIs were determined from wild-type (WT) and the clinically relevant G140S1Q148H drug-resistant mutant IN bound to IN-DNA complexes in vitro. In cells, INSTI dissociation was measured as viral replication after drug washout. Molecular models of BIC and DTG bound to IN-DNA complexes based on recent cryogenic electron microscopy (cryo-EM) structures provide structural insights into optimal binding geometry (2).
INSTI potency against WT HIV-1 has been well studied and, for the compounds described here, falls into the single-digit nanomole range (Table 1). BIC and DTG retain antiviral activity against many viral strains with INSTI resistance-associated mutations but show reduced activity against some complex mutants (4-9). The clinically relevant G140S1Q148H confers high-level resistance to RAL and EVG, 4.3-fold reduced susceptibility to DTG, and 2.1-fold reduced susceptibility to BIC. To further understand the role of the benzyl tail and bicyclic ring system of BIC, we studied two related analogs: compound 1 differs from BIC in that it contains a 2,4-difluorobenzyl group compared to a 2,4,6-trifluorobenzyl moiety in BIC, and compound 2 is the enantiomer of BIC (the bicyclic ring system is flipped in the opposite orientation). Compounds 1 and 2 showed more resistance to G140S1Q148H than BIC, and the orientation of the bicyclic ring system (compound 2 versus BIC) had a larger impact on mutant activity than the differences in the benzyl tail (compound 1 versus BIC).  Table 2). Data sets were analyzed using a single exponential decay equation as previously published for INSTIs (3). For WT IN, BIC had a substantially longer t 1/2 (163 6 31 h) than DTG (96 6 29 h), RAL (10 6 2 h), and EVG (3.3 6 0.9 h). The DTG, RAL, and EVG t 1/2 values were comparable to those previously published using this methodology (3). Both compound 1 and compound 2 had t 1/2 values that were intermediate to BIC and DTG. This suggests that both presence and orientation of the bicyclic ring system and the trifluorobenzyl tail are important for longer WT t 1/2 s.
The dissociation of INSTIs from G140S1Q148H mutant IN-DNA complexes was also studied and showed that BIC had a t 1/2 of 5.7 6 0.4 h compared to 1.9 6 0.2 h for DTG. There was high-level resistance to RAL and EVG with G140S1Q148H mutants, resulting in an insufficient level of binding of these INSTIs to IN-DNA complexes for determination of dissociation kinetics. The t 1/2 values of compounds 1 and 2 from G140S1Q148H IN were also different from those of BIC and DTG. Compound 1 retained a longer t 1/2 for this mutant than DTG, whereas compound 2 had a shorter t 1/2 , suggesting that both the presence and the specific orientation of the bicyclic ring system in BIC are more crucial than the benzyl tail modifications.
To probe the impact of the longer t 1/2 of BIC compared with other INSTIs in cells where all cellular intasome components are present, we used HIV-1 strain NL4.3 containing WT or G140S1Q148H to infect MT-2 cells and treated with BIC or other INSTIs. We then washed away the inhibitors 3 days postinfection and quantified the reverse transcriptase activity in the culture supernatant to evaluate viral growth as described on washout days 0, 4, and 8 (10,11). For BIC and DTG, the WT viral replication remained inhibited through at least 8 days after washout (Fig. 1A). EVG and RAL allowed for detectable viral replication 4 days after washout. These results are consistent with BIC and DTG having longer t 1/2 s from the IN-DNA complex than EVG and RAL. Longer experiments would be required to further differentiate BIC and DTG in washout experiments using WT virus.
The G140S1Q148H mutations cause varied degrees of resistance to INSTIs. In previous washout experiments, G140S1Q148H-containing virus was able to resume replication and integration after washout of RAL or EVG (10). Even though very low levels of replication were observed under constant BIC pressure, BIC was able to maintain .80% of suppression of the G140S1Q148H virus after 8 days of its washout (Fig. 1B). In contrast, there was  low replication under constant DTG pressure and significant viral replication 4 and 8 days after DTG washout. These data obtained in the more complex cell-based assay support the biochemical t 1/2 results. BIC maintained antiviral activity after washout for several days against WT and mutant HIV, which provides further evidence that long t 1/2 s may prevent viral rebound and emergent resistance in vivo after missing doses of drug. These findings are consistent with the resistance profiles reported in Table 1.  Fig. 2A to C), it is evident that the bicyclic ring of BIC makes additional van der Waals contacts with the b4-a2 loop (see G118 in Fig. 2) of WT IN and 39-dA vDNA compared to DTG, where the corresponding monocyclic ring of DTG has only partial contact with this region of HIV-1 IN. This additional interaction of the bicyclic ring of BIC with the b4-a2 loop of IN and 39-dA vDNA may contribute to its tighter binding and explain why we observed the longer t 1/2 . These models also agree well with other homology models based on cryo-EM structures of HIV-1 IN with BIC (PDB accession no. 6PUW) (2,12). A recent work using explicit solvent molecular dynamics simulations of WT and G140S1Q148H bound with BIC and a compound without a bicyclic ring system and containing 2,4-difluorobenzyl also shows the importance of additional interactions with the IN b4-a2 loop (2). A wider distribution of the atomic displacements in the case of this truncated analog compared to BIC in WT and G140S1Q148H supports the importance of this additional anchoring interaction.
Understanding the interactions between INSTIs and their targets is clinically relevant because INSTI-based regimens are the recommended choice for initial HIV therapy. The INSTIs BIC and DTG are highly potent and have high genetic barriers to resistance. Emergent resistance to three-drug BIC-and DTG-based regimens has not occurred in clinical trials but has occurred in rare cases in clinical practice (13). EVG-and RAL-based regimens, which have lower genetic barriers to resistance, select for IN mutations at codons  (23). The sequence alignment between SIVrcm IN (template) and HIV-1 IN (query) was optimized using ClustalW (24), yielding a sequence identity of 73%, a sequence similarity of 83%. The residues that were similar between the two sequences were retained while building the model. The side chains of the residues that were not part of the template were iteratively sampled using a coarse library of rotamers derived from known PDB structures until no clashes remained. The coordinates of all atoms not derived directly from the template itself were then minimized, producing the final refined model. BIC, DTG, and compound 2 were prepared with LigPrep with the metal binding states option using Schrodinger Suite 2019 (Schrödinger). INSTIs were docked using Glide XP docking protocol with expanded sampling method (25). Solvent molecules that coordinate to active site Mg 21 ions were kept in place during docking. Top three docked poses for both the INSTIs were saved as an output. 92, 143, 155, and 148. A long INSTI residence time may lead to higher tolerance to missed doses, but antiviral support by two nucleoside RT inhibitors of BIC-and DTG-based threedrug regimens is also important. There are some concerns around the resistance risk of potentially less forgiving two-drug INSTI-based regimens (DTG plus either lamivudine or rilpivirine) or three-drug RAL-or EVG-based regimens, which have lower resistance barriers (9,(14)(15)(16)(17). Single-drug therapy, such as DTG monotherapy, has given rise to unacceptably high rates of virological failure with INSTI resistance (16). Likewise, cases of failure with emergent M184V and R263K were reported for DTG-plus-lamivudine dual therapy in the clinical trials GEMINI and ACTG5353 (18,19). Clinically, the long residence times of BIC and DTG may help support their high barriers to resistance by maintaining viral inhibition during periods of low drug levels, such as when several consecutive doses are missed.
We performed enzymatic, tissue culture, and structural studies to investigate the mechanistic basis for the more favorable resistance profiles of BIC and DTG compared with EVG and RAL and to investigate the differences between BIC and DTG. Drug washout/release studies are consistent with a longer t 1/2 of BIC compared to DTG. Structural studies of the INSTI interactions with integrase intasomes (2,12,20,21) and our models show that BIC forms a stable complex with HIV-1 IN bound to HIV-1 long-terminalrepeat DNA. The t 1/2 of BIC from HIV-1 IN-DNA complexes was longer than the t 1/2 of DTG, RAL, and EVG. The long t 1/2 s of INSTIs with the integrase-DNA complex have been correlated with potent antiretroviral activity against WT HIV-1 integrase and a high barrier to resistance in vitro (3). BIC also dissociated more slowly than DTG from the resistant mutant G140S1Q148H IN-DNA complex, which is consistent with the greater in vitro activity of BIC than of other INSTIs against this mutant (2-9). Long t 1/2 and more optimal binding-site complementarity of the INSTI into its binding site may also translate into more favorable resistance profiles when the INSTI may tolerate small perturbations in the pocket induced by amino acid mutations. The impact of a longer BIC t 1/2 due to more optimal interactions with HIV-1 IN and vDNA may lead to greater tolerance of binding-site perturbations by drug resistance mutations and forgiveness of missed doses. These in vitro data further our understanding of the high efficacy and resistance barrier of BIC in the treatment of people living with HIV.