Structural investigation of 2‐naphthyl phenyl ether inhibitors bound to WT and Y181C reverse transcriptase highlights key features of the NNRTI binding site

Abstract

Human immunodeficiency virus (HIV)‐1 remains as a global health issue that is primarily treated with highly active antiretroviral therapy, a combination of drugs that target the viral life cycle. One class of these drugs are non‐nucleoside reverse transcriptase inhibitors (NNRTIs) that target the viral reverse transcriptase (RT). First generation NNRTIs were troubled with poor pharmacological properties and drug resistance, incentivizing the development of improved compounds. One class of developed compounds are the 2‐naphthyl phenyl ethers, showing promising efficacy against the Y181C RT mutation. Further biochemical and structural work demonstrated differences in potency against the Y181C mutation and binding mode of the compounds. This work aims to understand the relationship between the binding mode and ability to overcome drug resistance using macromolecular x‐ray crystallography. Comparison of 2‐naphthyl phenyl ethers bound to Y181C RT reveal that compounds that interact with the invariant W229 are more capable of retaining efficacy against the resistance mutation. Additional modifications to these compounds at the 4‐position, computationally designed to compensate for the Y181C mutation, do not demonstrate improved potency. Ultimately, we highlight important considerations for the development of future HIV‐1 drugs that are able to combat drug resistance.

Short abstract

PDB Code(s): 6X47, 6X49, 6X4A, 6X4B, 6X4C, 6X4D, 6X4E and 6X4F;

Abbreviations

HAART

highly active antiretroviral therapy

NNRTI

non‐nucleoside reverse transcriptase inhibitor

RT

reverse transcriptase

1. INTRODUCTION

Human immunodeficiency virus (HIV) is a worldwide health issue, with 37.9 million people infected as of 2018. ¹ Untreated HIV leads to acquired immunodeficiency syndrome (AIDS), rendering the infected person susceptible as a host to other deadly diseases. Although there is not a cure for HIV, life‐long drug‐based therapies exist to suppress the virus to prevent progression of HIV into AIDS. Highly active anti‐retroviral therapy (HAART) is a combination therapy consisting of various drugs that target different stages of the HIV‐1 infection life cycle, improving the quality of life and life expectancy of patients. ² , ³ , ⁴ Non‐nucleoside reverse transcriptase inhibitors (NNRTIs) are vital components of HAART that target HIV‐1 reverse transcriptase (RT), which is responsible for the production of viral DNA. ⁵ , ⁶ Unlike nucleoside reverse transcriptase inhibitors (NRTIs) that act as replication chain terminators, NNRTIs inhibit RT activity by binding an allosteric site ~10 Å from the active site mediated through a conformational change that alters the rate limiting step in chemical catalysis. ⁷ , ⁸ , ⁹

HIV‐1 is highly mutable and drug‐resistant strains have emerged because RT is error‐prone, requiring the development of next generation drugs that remain efficacious despite the rise of resistance. ¹⁰ , ¹¹ One such mutation is Y181C in the allosteric pocket of RT, which has been demonstrated to reduce the potency of first‐generation NNRTIs such as nevirapine or delavirdine. ¹² , ¹³ , ¹⁴ To this end, second generation NNRTIs such as etravirine (ETV) and rilpivirine (RPV) were developed that are effective against WT and an array of resistance‐associated mutations. ⁵ , ¹⁵ Even with the improved second generation of compounds, further development of NNRTIs is needed as compounds such as ETV and RPV have poor pharmacological properties and the development of resistant mutations has been reported. ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹

In our previous work, we identified a set of catechol diether compounds with a 7‐cyano‐2‐naphthyl substituent (2‐naphthyl phenyl ethers) as potential drug candidates through computational modeling using the BOMB and MCPRO programs. ²⁰ , ²¹ After biochemical and structural characterization, we observed that our synthesized compounds demonstrated two distinct binding modes signified by the orientation of the 2‐naphthyl ring (Figure 1). The nitrile group of our parent compound (Figure 1a) projects toward Y181 while the nitrile of a derivative compound (Figure 1b) projects toward W229, likely due to a 1‐position substitution. Interestingly, Compound 1 has an EC₅₀ of 22 and 2,600 nM to WT and Y181C respectively, while Compound 2, containing 1‐methyl and 4‐chloro substituents has an EC₅₀ of 6.2 and 58 nM (Table 1). We hypothesized that the differences observed in Y181C EC₅₀ values are determined by whether the 2‐naphthyl ether largely interacts with W229 or Y181.

FIGURE 1.

Crystal structures of Compound 1 (a) colored in yellow and Compound 2 (b) colored in magenta in complex with to WT RT. (c) Chemical structures of Compounds 1 and 2

TABLE 1.

Inhibitory activity (EC₅₀, nM) for HIV‐1 in MT‐2 cell assays

Compound	WT	Y181C	K103N/Y181C
1 a	22	2,600	4,000
2 a	6.2	58	280
3 a	7.8	60	890
4 a	1.9	28	410
5 a	18	900	1,200
6	23	2,100	1,700

^aSee Reference 20.

In this study, we have solved the crystal structures of Y181C RT in complex with a series of 2‐naphthyl phenyl ether compounds (Figure 2, Compounds 1 and 2) and compare them to wild type structures. These structures explain the mechanism by which certain 2‐naphthyl ethers may overcome the prevalent Y181C RT mutation. In addition, we solved structures of RT with derivatives of the parent compound (Compounds 3–6) to improve efficacy against the Y181C mutation, based on our previous results. Our structural analysis highlights key interactions that should be considered in the development of 2‐naphthyl ethers and future NNRTIs that combat drug resistant mutants of HIV‐1 while maintaining optimal pharmacokinetic properties.

FIGURE 2.

Chemical structures of Compounds 1–6

2. RESULTS

To deduce the structural consequences that result in the pronounced differences in activity between WT and Y181C RT observed for the 2‐naphthyl phenyl ethers, we have determined the structures of Compounds 1, 2, and 6 (Figure 2) bound to the Y181C mutant of RT using x‐ray crystallography (Figure 3). We have also solved the crystal structures of Compounds 3, 4, 5, and 6 bound to WT RT to examine the effects that various ring substitutions have on the binding of the 2‐naphthyl scaffold (Figure 4). Additionally, we have redetermined the structure of 1 bound to WT RT to a higher resolution than the structure reported previously ²⁰ for a more accurate comparison with the Y181C RT:1 complex structure presented here. Crystal structures were solved to resolutions of 2.50–2.86 Å and refined to final Free R‐factors between 0.2509 and 0.2685 and displayed good stereochemical and geometry statistics (Table S1). All structures have been deposited into the PDB. Calculated 2Fo‐Fc and Fo‐Fc omit electron density maps allowed for unambiguous modeling of inhibitors in both WT and Y181C protein structures (Figure S1). Consistent with previously reported structures of NNRTI complexes, no significant changes in the overall structure of RT were observed in either the WT or Y181C structures.

FIGURE 3.

(a) Superposition of 1 bound to WT RT (1 in lilac, protein in green) and Y181C RT (1 in yellow, protein in pink) (b) Superposition of 2 bound to WT RT (2 in lilac, protein in green) and Y181C RT (2 in yellow, protein in pink)

FIGURE 4.

(a) Superposition of 5 bound to WT RT (5 in yellow, protein in pink) with 2 bound to WT RT (2 in lilac, protein in green). (b) Superposition of 6 bound to WT RT (6 in yellow, protein in pink) with 2 bound to WT RT (2 in lilac, protein in green). (c) Superposition of 6 bound to WT RT (1 in lilac, protein in green) and Y181C RT (1 in yellow, protein in pink). (d) Superposition of bound structures of 5 (pink), and 6 (yellow) with the structure of the WT:1 complex (1 in lilac, protein in green)

The structure of Compound 1 bound to Y181C RT shows 1 binds in a similar conformation as it does in WT RT (Figure 3a). The cyano group of the naphthyl ring projects over Y181 and away from W229. As seen in the superposition with the WT structure, when bound to Y181C RT, both the naphthyl and catechol rings of 1 move 1.4 Å toward the empty space created by the vacated Y181 side chain. As a result of this shift, the naphthyl ring of 1 moves further away from W229 compared to its position in WT RT. Interestingly, the side chain of Y188 in the mutant compensates for this movement by turning toward C181 along with Compound 1. This compensatory movement allows Y188 to re‐optimize its stacking interaction with the displaced 1 in the Y181C RT mutant. Likewise, Compound 2 binds to the Y181C mutant in the same overall conformation previously observed in WT RT, in which the cyano group of the naphthyl ring points toward the W229 side chain. ²⁰ However, unlike Compound 1, the superposition of the WT and mutant structures clearly shows the binding of Compound 2 is identical in both WT and Y181C RT (Figure 3b). This suggests that the observed interactions of the edge of 2 with the face of W229, which results primarily from the different conformation of the 2‐naphthyl ring of 2, contributes much more to the binding of this compound than it does for Compound 1.

The structures of 2‐naphthyl ether analogs 3 and 4 bound to WT RT have also been determined. The superposition of these structures with the Compound 2 complex show all three compounds bind to WT RT very similarly (Figure S2). The fluorine substitution at the 4‐position in 3 and the methyl group substitution on 4 make no impact on binding conformation of the 2‐naphthyl ring compared to 2, nor do they significantly alter activity of the compounds against either WT or mutant HIV (Table 1). ²⁰ The nearly identical bound conformations of 2, 3, and 4 most likely result from the shared methyl group substitution at the 1‐position of the naphthyl ring, which severely restricts the conformational freedom of the 2‐naphthyl compounds. ²⁰ The increased activity measured for 4 may result from the addition of the fluorine on the catechol ring as seen in other classes of catechol diether inhibitors. ²² , ²³

A series of 2‐naphthyl compounds containing larger substitutions off the naphthyl 4‐position have been previously reported. ²⁰ For our current analysis, we have determined the structures of two examples from this series, 5 and 6, bound to WT RT. The superpositions of the structures of 5 and 6 with 2 show the ligands bind similarly (Figure 4a,b). The 4‐cyclopropyl group of 5 and 4‐methyl ester group of 6 both bind within a small pocket surrounded by residues P95, Y181, and Y188 of RT. Interestingly, these three residues all shift from their relative positions in the Compound 2 bound structure (Figure 4a,b). Whereas P95 and Y188 move away from the bulky 4‐position substituents, Y181 moves toward the substituent to assist in forming the sub‐pocket that accommodates the cyclopropyl and methyl ester groups. In addition to these protein residue rearrangements, the 2‐naphthyl rings of 5 and 6 must also slightly shift toward F227, away from P95 and Y181, to create room for the 4‐position groups of each compound. Within its position in the P95 pocket, the methyl ester group of Compound 6 is unable to form any hydrogen bonds with the protein. Furthermore, the superposition of 6 bound to Y181C RT with that of the WT structure shows the binding of the compound is nearly identical in both structures (Figure 4c). The large methyl ester group off the 4‐positon continues to point toward P95 and does not move toward the empty space created by the missing tyrosine in the Y181C RT mutant.

3. DISCUSSION

The 2‐naphthyl phenyl ethers represent a promising class of NNRTIs that have been shown to have 1–10 nM potencies against WT HIV strains. ²⁰ However, initial analogs from this class, such as those presented here, showed weaker activity against clinical HIV strains that contain the Y181C RT mutant. Whereas Compound 2 displays a 10‐fold decrease in activity between WT and Y181C strains, Compounds 1 and 6 show up to 500‐ to 1,000‐fold diminished activity. The structures determined here provide insight into the structural basis that result in these reduced activities in the presence of the Y181C mutation in the NNRTI binding pocket.

Compounds 1 and 2 have previously been shown to bind WT RT with different conformations of the 2‐naphthyl ring dependent upon the substitution at the naphthyl 1‐position. When the 1‐position is substituted with a methyl group as in Compound 2, the 2‐naphthyl ring prefers to bind with a conformation such that the cyano group projects toward the vicinity of W229. Since this binding conformation takes advantage of a greater interaction between the naphthyl group and W229, it has been previously postulated that Compound 2 would show less of an impact when Y181 is substituted by cysteine. ²⁰ The structure of 2 bound to Y181C RT reported here does indeed reveal the interaction with W229 to be substantial enough to maintain Compound 2 in its WT position even in the absence of Y181. In contrast, we show the position of Compound 1 to be significantly altered in the presence of the Y181C mutation, and given the larger role Y181 plays in the binding of 1 to WT RT, this result is not unexpected. In the absence of Y181, the catechol and 2‐naphthyl rings of 1 shift toward C181 and away from W229. Whereas the interaction made by 1 with W229 is already suboptimal in WT RT, it becomes even less optimal in the presence of C181, as the naphthyl ring of 1 is displaced further away from W229 in order to occupy the space created by the vacated Y181. Thus, Compound 1 loses key contacts with both W229 as well as the absent Y181 in the Y181C RT mutant. Although Y188 slightly compensates for this loss by rotating toward the shifted naphthyl ring of 1 in the mutant, the activity of 1 still diminishes 1,000‐fold against Y181C HIV strain. In comparison, Compound 2 only loses interaction with Y181 and not W229 in the Y181C mutant and as a result, displays only a 10‐fold decrease in activity against the mutant strain. Four additional 2‐naphthyl analogs that did not have a substitution at the 1‐positon were also previously examined, and they similarly show 46‐163‐fold worse activities against the variant HIV strain. ²⁰

The 2‐naphthyl analogs that contain larger substituents off the naphthyl ring 4‐position were initially designed with the expectation that the larger groups would occupy a portion of the space created in the Y181C mutant and have improved potency against mutant HIV strains. The initial compounds, however, were less active against both WT and variant HIV strains compared to the parent compound that possesses only a methyl group substitution at the 4‐position. ²⁰ We speculate that the structural rearrangement of residues P95, Y181, and Y188 that are necessitated to accommodate the larger 4‐position groups most likely incurs an energetic penalty that results in the decreased activity exhibited by these compounds. Furthermore, in the presence of the Y181C mutation, the bulky group of 6 maintains its WT‐bound conformation and does not move toward the vacated space of Y181. This inflexibility may explain why these compounds do not fare better against HIV variants as initially hypothesized. In the case of Compound 6, the methyl ester group is unable to form any hydrogen bonds with the surrounding residues, and the resulting desolvation penalty may further contribute to its weaker activity. Interestingly, the superposition of the 5 and 6 structures with the structure of 1 shows the cyclopropyl and methyl ester substituents to occupy the same space as that of the 2‐naphthyl ring of 1 (Figure 4d). The poorer activities displayed by both the 4‐position substituted analogs and 1 against WT HIV (Table 1) suggest that the P95 sub‐pocket can be a difficult region of the NNRTI binding site to target successfully in design studies.

Although the initially designed set of 2‐naphthyl compounds containing large 4‐position substituents described here did not show improved activity against either WT or mutant HIV strains, subsequent analogs have been successfully utilized in the design of covalent inhibitors against Y181C RT. ²⁴ These compounds contain either an α‐halo amide or acrylamide group off the naphthyl 4‐position that act as an electrophilic warhead, and they provide an interesting comparison to the 4‐position substituted examples presented here. Upon binding to WT RT, the catechol ring of these warhead‐containing compounds shifts away from Y181 by 1.2 Å to provide room for the large acrylamide group from the 4‐position, much like what is observed in the structures of 5 and 6 RT complexes. As a result, these warhead‐containing 2‐naphthyls similarly display a 100‐fold decrease in activity against WT HIV compared to Compound 1. However, the electrophilic functional groups contained in these compounds have been shown to covalently modify the C181 side chain and irreversibly inhibit activity of the resistant Y181C RT mutant. ²⁴ This special feature allows these 4‐substituted 2‐naphthyl inhibitors to retain their activity against Y181C mutant strains. Whereas placing bulky groups off the 4‐position that look to solely replace the vacated Y181 residue may not be sufficient to overcome Y181C resistance, success can be achieved by instead utilizing functional groups at this position that are capable of reacting with the substituted C181 side chain.

Based on the structural results presented here, the interaction between NNRTI compounds with W229 appears to be a major driving force for overcoming Y181C RT strains. Compound 1 does not interact as strongly with W229 in WT RT (Figure 5a) and displays a 1,000‐fold reduction of activity against Y181C RT. In comparison, Compound 2 interacts much more strongly with W229 (Figure 5b) and only shows a 10‐fold reduction against the mutant strain. Furthermore, a compound from the 1‐naphthyl class of catechol diethers, which has been shown to make an even more extended edge to face interaction with W229 than that observed for Compound 2 (Figure 5c), ²² displays only a threefold reduction in activity against Y181C HIV. ²³ This comparison suggests that the more a compound interacts with W229 within the NNRTI binding site, the more likely it will have the ability to overcome the Y181C resistance mutation in HIV. Given that W229 does not appear to be a very mobile residue when comparing the many RT structures solved and that W229 mutations that yield resistance have yet to be identified in the clinic further suggest that targeting W229 in drug design studies may lead to the discovery of more efficacious NNRTI drugs.

FIGURE 5.

Space filling models depicting the interactions between Compounds 1 (a, 1 in yellow), 2 (b, in lilac), and a 1‐naphthyl analog (c, in pink, PDB code: 6OE3) and Y181 and W229 (in green)

4. CONCLUSIONS

We have determined the structures of six 2‐naphthyl phenyl ether inhibitors bound to WT RT and Y181C RT. The comparison of these 2‐naphthyl complex structures along with their measured antiviral activity EC₅₀ values for HIV‐1 infected MT‐2 cellular assays highlight important aspects of the NNRTI binding site. As in the case of Compound 1, a decrease in activity against variant HIV strains arises not only from the loss of interaction with the substituted residue, but from also the loss of interaction with other non‐substituted residues that result from ligand displacement from its WT position. Also, building off the 4‐position of the 2‐naphthyl ring to interact with the P95 pocket does not offer any advantage against either WT or Y181C strains. However, optimizing interactions with residues that are immobile and cannot be mutated by the virus, such as W229, offers significant advantages in overcoming Y181C resistance. Comparative structural analyses such as these help to identify the key residues to target for inhibitor design, and this information will aid in the future design of NNRTI compounds that are effective against both WT and resistant strains of HIV.

5. MATERIALS AND METHODS

5.1. Expression, purification, and crystallization

Recombinant WT and mutant RT52A HIV‐1 RT was expressed and purified as previously described. ²⁵ , ²⁶ , ²⁷

5.2. Data collection, processing, and structure determination and refinement

Apo crystals used for inhibitor soaking were grown by hanging drop vaporization in 50 mM imidazole or Hepes (pH 6.5–7.0), 16–20% (wt/vol) PEG 8000, 100 mM ammonium sulfate, 15 mM magnesium sulfate, and 5 mM spermine with an initial protein concentration of 10 mg/ml. Inhibitor was added to final concentration of 0.5 mM to drops containing suitable crystals for overnight soaking. Crystals of the WT:3 complex were obtained by cocrystallization, in which 20 mg/ml of protein was first incubated with 0.5 mM inhibitor on ice for 1 h before crystallization. Cocrystals of WT:3 were then grown in 50 mM MES pH 6.0, 14% (wt/vol) PEG 8000, 100 mM ammonium sulfate, 15 mM magnesium sulfate, and 5 mM spermine. For cryoprotection, all crystals were transferred to a solution containing 27% (vol/vol) ethylene glycol and 0.5 mM inhibitor before being flash frozen in liquid nitrogen. X‐ray diffraction data sets on frozen crystals were collected at the Advanced Photon Source on NE‐CAT beam line 24‐ID‐E. Data sets were processed with either HKL2000 ²⁸ or XDS. ²⁹ Phases were determined by molecular replacement with the program PHASER ³⁰ using the previously solved structure PDB 5TER ²⁰ as the initial search model. All model building into electron density was performed with COOT ³¹ and the structures were refined using Phenix Refine. ³² For each refinement, 5% of all reflections were omitted and used for the calculation of R_free. Successive rounds of simulated annealing, XYZ coordinate, and individual B‐factor refinement were performed until acceptable R factors, geometry statistics, and Ramachandran statistics were achieved. Data collection, diffraction, and refinement statistics can be found in Table S1. Iterative build, composite omit electron density maps shown in Figure S1 were calculated using Phenix Autobuild. ³³ All atomic coordinates and structure factors and have been deposited in the Protein Data Bank with PDB ID codes 6X47, 6X49, 6X4A, 6X4B, 6X4C, 6X4D, 6X4E, and 6X4F. All figures were prepared by PYMOL. ³⁴ Crystallography programs were compiled by SBGrid. ³⁵

5.3. MTT cellular viability assays

A tetrazolium‐based colorimetric assay was used to determine the in vitro anti‐HIV‐1 activity of compounds in MT‐2 cells as described previously. ²⁴

AUTHOR CONTRIBUTIONS

Vincent N. Duong: Funding acquisition; investigation; visualization; writing‐original draft; writing‐review and editing. Joseph A. Ippolito: Formal analysis; visualization; writing‐original draft; writing‐review and editing. Albert H. Chan: Investigation. Won‐Gil Lee: Investigation. Krasimir A. Spasov: Formal analysis; investigation. William L. Jorgensen: Conceptualization; funding acquisition; project administration; supervision; writing‐review and editing. Karen S. Anderson: Conceptualization; funding acquisition; project administration; supervision; writing‐review and editing.

Supporting information

Figure S1: The σ_A–weighted 2mF_o‐F_c electron density maps of the RT‐inhibitor complexes

Figure S2: Complexes of RT with compounds 3 and 4

Table S1: Crystallographic statistics table

Duong VN, Ippolito JA, Chan AH, et al. Structural investigation of 2‐naphthyl phenyl ether inhibitors bound to WT and Y181C reverse transcriptase highlights key features of the NNRTI binding site. Protein Science. 2020;29:1902–1910. 10.1002/pro.3910