Acquired immune deficiency syndrome (AIDS) was identified in 1981 [1]. It is characterized by human immunodeficiency virus (HIV) infection leading to CD4⁺ T cell depletion and resulting in severe immunosuppression and increased incidence of opportunistic infections [2]. Globally, 1.1–2.1 million people were newly infected with HIV and about 480,000 to 1.0 million patients died of AIDS-related illnesses in 2020 based on data from the Joint United Nations Program on HIV/AIDS (UNAIDS) report [3]. HIV-1 is a retrovirus that utilizes the reverse transcriptase (RT) enzyme to convert the viral single-stranded RNA into double-stranded DNA. RT-targeted drugs can be divided into nucleoside (acid) RT inhibitors (NRTIs) and non-nucleoside RT inhibitors (NNRTIs). As an integral part of standard therapy in highly active antiretroviral therapy (HAART) regimens, NNRTIs are highly sought after due to their high antiviral potency, relatively low toxicity, and favorable pharmacokinetic properties. To date, eight NNRTIs have been approved for the treatment of HIV in the clinic, including Nevirapine (NVP), Delavirdine (DLV), Efavirenz (EFV), Etravirine (ETR), Rilpivirine (RPV), Elsulfavirine (ELV), Doravirine (DOR) and Ainuovirine (ANV) [4,5]. However, the rapid emergence of drug resistance and severe side effects of long-term administration has motivated medicinal chemists to develop a variety of potent NNRTIs with superior drug resistance and improved drug-like properties [6,7].

The 1-[(2-hydroxyethoxy)methyl]-6-(phenylthio)thymine (HEPT, Fig. 1), reported by Miyasaka and colleagues in 1989 as a kind of NNRTIs, showed moderate antiviral activity (EC₅₀ = 7 µmol/L) and weak selectivity index (SI = 106) [8]. Over the past three decades, considerable efforts have been made in the structural optimization of HEPT, leading to the identification of MKC442, a Phase Ⅲ clinical drug candidate, which was unfortunately terminated due to its undesirable metabolic properties [9]. Our research group focuses on the discovery of anti-HIV drugs and has performed many kinds of structural modifications on the unique structure of HEPT. Several naphthyl-containing HEPT analogs, such as compounds 3–5, were found to show potent inhibitory activity against WT HIV-1 (Fig. 1), but their activity toward clinically relevant mutants has not been determined [10-14]. Recently, we have developed a series of novel sulfinyl-containing HEPT analogs through a structure-based drug design strategy. The representative compound 6 in this series showed improved inhibitory activity (EC₅₀ = 0.59 µmol/L) but with an extremely poor selectivity index (SI = 9), compared to HEPT [15]. In addition, it was still inactive toward clinically relevant mutants. Therefore, further structural optimization is still needed to find more safe and effective drug candidates.

In the present study, we endeavored to optimize compound 6 by introducing CH-OH and CH-OTMS groups into the linker, respectively, aiming to further extend the SAR of HEPTs. The predicted docking studies of 6 with RT indicated that the central thymine skeleton of 6 was oriented by two hydrogen bonds between CONH moiety and the protein (Fig. 2B): (ⅰ) a hydrogen bond between the NH of thymine and the carbonyl oxygen of K101; (ⅱ) a water-mediated hydrogen bond between the NH of thymine and K101. The C6-phenyl ring of 6 was directed toward the hydrophobic aromatic pocket and formed π-π and hydrophobic interactions with W229, Y188 and Y181 residues, respectively. However, when the S=O linker of 6 was replaced with a CH-OH linker (14i), the orientation of the oxygen changed compared to 6 (Fig. 2C). The CH-OH linker would allow the molecule to retain good molecular flexibility and adopt multiple conformations within the NNIBP, which may be beneficial for being sensitive to mutant variations [13,16,17]. In addition, our previous work revealed that the CH-OH linker region was a typical metabolic site in DAPYs, which was responsible for their poor metabolic stability [18]. The introduction of a silicon atom to a known drug molecule can lead to significant changes in drug biological activity and metabolism [19]. For example, the introduction of a (trimethylsilyl)ethyl group into camptothecin significantly improved its broad-spectrum anticancer activity, and enhanced tissue penetration as well as bioavailability [20]. In addition, cytarabine was an anti-metabolite that is converted into a prodrug through one or more silicon-based protecting groups with improved lipophilicity [21]. Therefore, CH-OTMS-containing HEPT analogs as intermediates of CH-OH compounds were also investigated.

The synthetic route to targeted molecules 14a-14ar was depicted in Scheme 1. The nucleophilic substitution of trichloropyrimidines 7a-7c, prepared according to our previously reported protocol [15], with appropriate cyanides was conducted using sodium hydride as base in dry DMF, producing corresponding dichloropyrimidines 8a-8ar, as previously described by Loksha and coworkers [22]. Treating 8a-8ar with NaOMe in MeOH at 65 ℃ for 12 h afforded dimethoxypyrimidines 9a-9ar, which were then subjected to oxidation reaction by oxygen atmosphere in the presence of NaH in anhydrous THF to provide carbonyl pyrimidines 10a-10ar. 10a-10ar were reduced by NaBH₄ in MeOH at room temperature for 2–4 h to furnish alcohols 11a-11ar [23]. Demethylation of 11a-11ar with conc. HCl in MeOH at 65 ℃ overnight delivered thymine analogs 12a-12ar, which were next treated with benzyl chloromethyl ether in the presence of N,O-bis(trimethylsilyl)acetamide (BSA) and tetrabutylammonium iodide (TBAI) in DCM at room temperature for 4 h to afford 13a-13ar. Deprotection of 13a-13ar with TBAF in THF at room temperature for 0.5 h finally yielded desired products 14a-14ar in 68%–93% yields.

All the target compounds were evaluated for their in vitro anti-HIV activity and cytotoxicity in MT-4 cells infected with the WT HIV-1 (Ⅲ_B). The preliminary biological results were summarized in Table 1, taking compound 6, HEPT, NVP, and EFV as controls. Molecular docking was performed using Schrödinger Maestro 11.4 software to predict the binding modes of selected compounds to RT, the docking results of which were visualized and generated by PyMOL. When the linker was CH-OH, we explored the effect of the position and diversity of substituents on the R₂ phenyl ring on antiviral activity (14a-14y). The results in Table 1 indicated that most of these compounds showed moderate to good activity toward wild-type HIV-1 strain with EC₅₀ values ranging from 0.22 µmol/L to 47.76 µmol/L and selectivity index (SI) in the range of 2–446. Among them, the most active compound 14i exhibited significantly enhanced efficacy toward wild-type HIV-1 (EC₅₀ = 0.22 µmol/L), which was about 3-fold more potent than compound 6. The cytotoxicity of 14i was also improved with a much higher selectivity index (SI = 446) than 6 (SI = 9). Introduction of bulky substituents at the meta-position of the phenyl ring was detrimental to anti-HIV activity such as 14c, 14f and 14s. Besides, the introduction of different functional groups to the para-position of the benzene ring resulted in decreased antiviral activity (14d, 14g, 14j, 14m, 14p, 14r, 14w). For the mono-halogen-substituted compounds, the position priority abided by the rules as follows: meta- > ortho- > para-, for example, 14l > 14k > 14m, 14o > 14n > 14p, 14q > 14r. For these compounds with electron-donating group, their inhibitory activity mostly abided by the rules as follows: ortho- > meta- > para- (14b > 14c > 14d, 14e > 14f > 14g), whereas introducing a methyl group at meta-position on benzene ring (14i) led to the best anti-HIV-1 activity. Then, the di-substituted compounds were also investigated (14t-14v). The difluorinated compound 14t exhibited more favorable anti-HIV-1 activity, compared to its monofluorinated analogs (14k-14m). However, enlarging the bulk of the substituent resulted in decreased anti-HIV-1 activity, such as 14t > 14u > 14v. In addition, the substitution of the benzene ring with a naphthalene ring led to a 3-fold reduction in antiviral activity (14x vs. 14a). Interestingly, the α-naphthyl substituent was more favorable for anti-HIV potency than β-naphthyl substituent (14x > 14y).

Next, the effect of different sizes of R₁ substituent on the activity was also investigated. Comparing compounds 14a-14y with 14z-14ah in Table 1, substitution of the isopropyl group with a methyl group at C5 on the pyrimidine resulted in a significant reduction in antiviral activity. The isopropyl group was changed to ethyl group at C5 on pyrimidine (14ai-14ar), affording another potent compound 14aj, which exhibited more potent antiviral activity (EC₅₀ = 0.18 µmol/L) with higher SI value (SI = 907) than 14i. In addition, some intermediates of 14a-14ar were also evaluated for anti-HIV activity. The results illustrated in Table 1 showed that most of them exhibited decreased antiviral activity compared with their corresponding OH derivatives in 14a-14ar, for instance, 14a > 13a, 14k > 13k, 14l > 13l, 14q > 13q, 14s > 13s, 14t > 13t and 14ah > 13ah. Exceptionally, 13i exhibited comparable inhibitory activity (EC₅₀ = 0.20 µmol/L) to 14i (EC₅₀ = 0.22 µmol/L, SI = 446), but the selectivity of 13i was slightly improved (SI = 665).

Subsequently, 14i, 14aa, 14aj and 13i were selected to carry out molecular docking. As displayed in Fig. 3, 14i, 14aj and 14aa fitted well into the binding pocket of NNIBP and formed crucial hydrogen bonding interactions and π-π interactions with surrounding residues. It was worth noting that 14aa showed lower antiviral activity than 14i and 14aj, which may be due to the lack of π-π interaction with Y181. For 13i, it showed different binding model compared with above three compounds. Not only did it form key hydrogen-bonding interactions and water-mediated hydrogen-bonding interactions with K101, but it also formed hydrogen-bonding interactions with K103 and π-π interactions with F227 and W229, resulting in enhanced activity.

Some representative compounds were selected to further evaluate their inhibitory activity toward a panel of clinically relevant drug-resistant HIV-1 mutant strains (L100I, K103N, and E138K) as described in Table 2. Pleasingly, 14i, 14aj and 13i had superior activity to NVP toward K103N. 14i and 13i exhibited comparable potency to NVP toward E138K (EC₅₀ < 1 µmol/L). The inhibitory activity of 14n and 13k against the L100I mutant was comparable to that of NVP. Thereafter, we performed site mutations at L100, K103, and E138 of WT HIV-1 RT by computational simulation. Five compounds (14i, 14k, 14n, 14aa and 13i) were selected to perform molecular docking with the mutant RT, respectively. The docking results illustrated in Fig. 4 revealed that the π-π stacking interaction with Y181 toward the L100I mutant strain was lost in 14aa, interpreting the diminished activity of 14aa (Figs. 4A and B). By comparing 14k with 14aj, we could hypothesize that 14aj not only formed crucial hydrogen bonding interaction with K101 and water-mediated hydrogen bonding interaction with K101 but also formed hydrogen bonding interaction with N103, together resulting in higher K103N-associated mutant inhibitory than 14k (Figs. 4C and D). For the compound 14aa, its efficiency was lower than 13i and NVP, due to the deficiency of interaction with K103 toward the E138K mutant strain (Figs. 4E and F).

Furthermore, the representative compounds with good anti-HIV-1 activity were selected to evaluate their inhibitory activity toward WT HIV-1 RT to validate the binding target of these newly synthesized HEPTs (Table 3). All these tested compounds displayed moderate inhibitory activity toward WT HIV-1 RT with IC₅₀ values ranging from 0.57 µmol/L to 8.40 µmol/L. Among them, 14i (IC₅₀ = 0.57 µmol/L) and 13i (IC₅₀ = 0.65 µmol/L) showed the most potent activity against WT RT, being superior to NVP (IC₅₀ = 0.80 µmol/L) but inferior EFV (IC₅₀ = 0.011 µmol/L), which was consistent with their antiviral activity in MT-4 cells. The regression analysis of pEC₅₀ and pIC₅₀ displayed a good correlation (R² = 0.8505, Fig. S1 in Supporting information), indicating that these newly designed compounds could specifically bind to HIV-1 RT.

In conclusion, a novel series of structurally diverse CHOR-HEPT were synthesized and evaluated for their anti-HIV-1 activity. The majority of these compounds exhibited moderate to good activity against WT HIV-1 with EC₅₀ values ranging from 0.18 µmol/L to 51.88 µmol/L and SI values ranging from 4 to 907. For OH-substituted compounds, the most active compound 14aj with a CHOH linker displayed 3-fold increase in potency (EC₅₀ = 0.18 µmol/L) toward WT HIV-1 and 100-fold improvement in selectivity (SI = 907), compared to the lead compound 6 (EC₅₀ = 0.59 µmol/L, SI = 9). For CHOTMS-containing HEPT derivatives, compound 13i with a CHOTMS linker exhibited comparable activity to 14aj against WT HIV-1 (EC₅₀ = 0.20 µmol/L) but with a slight decrease in selectivity (SI = 665). Pleasingly, both 13i and 14aj had moderate inhibitory activity toward clinically relevant mutant strains (L100I, K103N, and E138K). Besides, compound 13i showed nanomolar inhibitory activity against WT RT (IC₅₀ = 0.65 µmol/L) while 14aj revealed low micromolar potency (IC₅₀ = 1.23 µmol/L). The molecular docking studies were fully leveraged to explain the SARs. Our findings will lay the structural foundation for further structural optimization of HEPT. Continued efforts around the HEPT structural modifications are still on the way aiming to search for safer and more effective non-nucleoside HIV-1 reverse transcriptase inhibitors with good druggability.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work was financially supported by National Natural Science Foundation of China (No. 22077018) and National Key R&D Program of China (No. 2017YFA0506000). We thank Ningxia Medical University for providing the sources of molecular modeling.

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.07.006.