Identification of Novel Diarylpyrimidines as Potent HIV-1 Non-Nucleoside Reverse Transcriptase Inhibitors by Exploring the Primer Grip Region

HIV-1 reverse transcriptase (RT) plays a crucial role in the viral replication cycle, and RT inhibitors can represent a promising pathway in treating AIDS. To explore the primer grip region of HIV-1 RT, using -CH2O- as a linker, substituted benzene or pyridine rings were introduced into the left wing of diarylpyrimidines (DAPYs). A total of 17 compounds with new structures were synthesized. It showed that all compounds exhibited anti-HIV-1 (wild-type) activity values ranging from 7.6–199.0 nM. Among them, TF2 (EC50 = 7.6 nM) showed the most potent activity, which was better than that of NVP (EC50 = 122.6 nM). Notably, compared with RPV (CC50 = 3.98 μM), TF2 (CC50 > 279,329.6 nM) showed low cytotoxicity. For HIV-1 mutant strains K103N and E138K, most compounds showed effective activities. Especially for K103N, TF2 (EC50 = 28.1 nM), TF12 (EC50 = 34.7 nM) and TF13 (EC50 = 28.0 nM) exhibited outstanding activity, being superior to that of NVP (EC50 = 7495.1 nM) and EFV (EC50 = 95.1 nM). Additionally, TF2 also showed the most potent activity against E138K (EC50 = 44.0 nM) and Y181C mutant strains (EC50 = 139.3 nM). In addition, all the compounds showed strong enzyme inhibition (IC50 = 0.036–0.483 μM), which demonstrated that their target was HIV-1 RT. Moreover, molecular dynamics simulation studies were implemented to predict the binding mode of TF2 in the binding pocket of wild-type and K103N HIV-1 RT.


Introduction
Human immunodeficiency virus type 1 (HIV-1), one of the viruses that pose a serious threat to human health worldwide, is the main pathogen of acquired immunodeficiency syndrome (AIDS) [1][2][3]. UNAIDS 2022 data indicated that 38.4 million people were living with HIV, 1.5 million people were newly infected with HIV, and 650,000 people have died from AIDS in 2021 [4]. Currently, the clinical treatment for AIDS often uses highly active antiretroviral therapy (HAART), which is a combination of two, three, or more drugs, typically including reverse transcriptase inhibitors (RTIs) [5,6]. Because of the high specificity, potent antiviral properties, and favorable pharmacokinetics, nonnucleoside reverse transcriptase inhibitors (NNRTIs) have proven to be an important part of HAART regimens. The NNRTIs are non-competitive inhibitors that act by allosteric inhibition of DNA Among them, ETR and RPV belong to diarylpyrimidines NNRTIs, exhibiting effec tive activity against the resistant mutations caused by the first-generation NNRTIs, such as Y181C, P236L and K103N [10,[14][15][16]. In spite of their higher genetic barrier, severa single, double or triple mutants have emerged with resistance towards these drugs, such as K101E, E138K, Y181V, Y181C + V179F and V179F + Y181C + F227C [17][18][19][20]. Therefore it is still urgent to develop novel NNRTIs with an improved anti-resistance profile.
The primer grip is a highly conserved structural motif consisting of the β12-β13 hairpin (F227-H235) ( Figure 3); it is responsible for locating the 3′-OH end of the primer strand at the polymerase catalytic site, making it important in the catalytic activity of RT [30,31]. The primer grip contains three amino acid residues in the NNRTI-binding pocket: F227, W229 and L234, and some mutations of F227 and L234 have been observed: F227C/L/I/V, L234/I. However, the mutation of the key W229 has never been observed. As for the other amino acid residues in the primer grip region, except for M230 (M230I/L), no other drugresistance mutations have been reported [32]. Targeting highly conserved amino acid residues is of great significance for the discovery of new compounds with improved drug resistance profiles [33], which inspired us to target the conserved residue W229 in the NNRTI-binding pocket for rational drug design and further explore the chemical space of the primer grip region. On the basis of the above analysis, with the privileged aminobenzonitrile moiety (right wing) and central pyrimidine ring (B-ring) unchanged, substituted benzene rings or pyridine rings were introduced to the left wing of diarylpyrimidines by a CH2O linker, expecting the newly introduced aromatic rings could develop stronger π-π interaction with the conserved residue (W229) and explore primer grip region to enhance the drug resistance profile (Figure 4). Here, we described the synthesis of these novel diarylpyrimidines as well as their anti-HIV-1 activity, and then we further discussed the preliminary structure-activity relationship (SAR) in detail and selected representative molecules for molecular dynamics simulation studies. The primer grip is a highly conserved structural motif consisting of the β12-β13 hairpin (F227-H235) ( Figure 3); it is responsible for locating the 3 -OH end of the primer strand at the polymerase catalytic site, making it important in the catalytic activity of RT [30,31]. The primer grip contains three amino acid residues in the NNRTI-binding pocket: F227, W229 and L234, and some mutations of F227 and L234 have been observed: F227C/L/I/V, L234/I. However, the mutation of the key W229 has never been observed. As for the other amino acid residues in the primer grip region, except for M230 (M230I/L), no other drug-resistance mutations have been reported [32]. Targeting highly conserved amino acid residues is of great significance for the discovery of new compounds with improved drug resistance profiles [33], which inspired us to target the conserved residue W229 in the NNRTI-binding pocket for rational drug design and further explore the chemical space of the primer grip region.  Figure 2. Illustration of the pharmacophore model of diarylpyrimidines.
The primer grip is a highly conserved structural motif consisting of the β12-β13 hairpin (F227-H235) ( Figure 3); it is responsible for locating the 3′-OH end of the primer strand at the polymerase catalytic site, making it important in the catalytic activity of RT [30,31]. The primer grip contains three amino acid residues in the NNRTI-binding pocket: F227, W229 and L234, and some mutations of F227 and L234 have been observed: F227C/L/I/V, L234/I. However, the mutation of the key W229 has never been observed. As for the other amino acid residues in the primer grip region, except for M230 (M230I/L), no other drugresistance mutations have been reported [32]. Targeting highly conserved amino acid residues is of great significance for the discovery of new compounds with improved drug resistance profiles [33], which inspired us to target the conserved residue W229 in the NNRTI-binding pocket for rational drug design and further explore the chemical space of the primer grip region. On the basis of the above analysis, with the privileged aminobenzonitrile moiety (right wing) and central pyrimidine ring (B-ring) unchanged, substituted benzene rings or pyridine rings were introduced to the left wing of diarylpyrimidines by a CH2O linker, expecting the newly introduced aromatic rings could develop stronger π-π interaction with the conserved residue (W229) and explore primer grip region to enhance the drug resistance profile (Figure 4). Here, we described the synthesis of these novel diarylpyrimidines as well as their anti-HIV-1 activity, and then we further discussed the preliminary structure-activity relationship (SAR) in detail and selected representative molecules for molecular dynamics simulation studies. On the basis of the above analysis, with the privileged aminobenzonitrile moiety (right wing) and central pyrimidine ring (B-ring) unchanged, substituted benzene rings or pyridine rings were introduced to the left wing of diarylpyrimidines by a CH 2 O linker, expecting the newly introduced aromatic rings could develop stronger π-π interaction with the conserved residue (W229) and explore primer grip region to enhance the drug resistance profile ( Figure 4). Here, we described the synthesis of these novel diarylpyrimidines as well as their anti-HIV-1 activity, and then we further discussed the preliminary structure-activity relationship (SAR) in detail and selected representative molecules for molecular dynamics simulation studies.

Chemistry
As indicated in Scheme 1, using 9 as the starting material, which we reported earlier [34][35][36], it was subjected to nucleophilic substitution reaction with 3,5-dimethyl-4-hydroxybenzaldehyde under the alkaline condition of potassium carbonate to obtain intermediate 10. Reduction of 10 with NaBH4 gave intermediate 11, which was brominated with PBr3 to give key intermediate 12. Finally, 12 was reacted with different substituted phenols and thiophenols to obtain the target compounds TF1-TF16 via nucleophilic substitution reactions. TF17 was obtained by removing the Boc protecting group of TF16.

Biological Evaluation
These 17 newly synthesized compounds were tested in MT4 cells for their antiviral activity using the MTT method [37]. The five marketed drugs approved by the U.S. Food and Drug Administration: Zidovudine (AZT), NVP, EFV, ETR and RPV, were used as positive drugs. EC50, CC50 and SI were used to express the biological evaluation results, which were summarized in Tables 1-4.

Chemistry
As indicated in Scheme 1, using 9 as the starting material, which we reported earlier [34][35][36], it was subjected to nucleophilic substitution reaction with 3,5-dimethyl-4hydroxybenzaldehyde under the alkaline condition of potassium carbonate to obtain intermediate 10. Reduction of 10 with NaBH 4 gave intermediate 11, which was brominated with PBr 3 to give key intermediate 12. Finally, 12 was reacted with different substituted phenols and thiophenols to obtain the target compounds TF1-TF16 via nucleophilic substitution reactions. TF17 was obtained by removing the Boc protecting group of TF16.

Chemistry
As indicated in Scheme 1, using 9 as the starting material, which we reported earlier [34][35][36], it was subjected to nucleophilic substitution reaction with 3,5-dimethyl-4-hydroxybenzaldehyde under the alkaline condition of potassium carbonate to obtain intermediate 10. Reduction of 10 with NaBH4 gave intermediate 11, which was brominated with PBr3 to give key intermediate 12. Finally, 12 was reacted with different substituted phenols and thiophenols to obtain the target compounds TF1-TF16 via nucleophilic substitution reactions. TF17 was obtained by removing the Boc protecting group of TF16.

Biological Evaluation
These 17 newly synthesized compounds were tested in MT4 cells for their antiviral activity using the MTT method [37]. The five marketed drugs approved by the U.S. Food and Drug Administration: Zidovudine (AZT), NVP, EFV, ETR and RPV, were used as positive drugs. EC50, CC50 and SI were used to express the biological evaluation results, which were summarized in Tables 1-4.

Biological Evaluation
These 17 newly synthesized compounds were tested in MT4 cells for their antiviral activity using the MTT method [37]. The five marketed drugs approved by the U.S. Food and Drug Administration: Zidovudine (AZT), NVP, EFV, ETR and RPV, were used as positive drugs. EC 50 , CC 50 and SI were used to express the biological evaluation results, which were summarized in Tables 1-4. duction of hydroxymethyl benzene ring may be detrimental to the improvement of the toxicity of compounds.
In addition, 3-pyridine-containing and substituted 3-pyridine-containing compounds (TF11-TF14) were also synthesized. Compared to TF11 (EC50 = 199.0 nM) with an unsubstituted pyridine ring, the activity of TF12~TF14 showed different degrees of increase, among which TF12 (EC50 = 7.8 nM) was the most active compound. duction of hydroxymethyl benzene ring may be detrimental to the improvement of the toxicity of compounds. In addition, 3-pyridine-containing and substituted 3-pyridine-containing compounds (TF11-TF14) were also synthesized. Compared to TF11 (EC50 = 199.0 nM) with an unsubstituted pyridine ring, the activity of TF12~TF14 showed different degrees of increase, among which TF12 (EC50 = 7.8 nM) was the most active compound. duction of hydroxymethyl benzene ring may be detrimental to the improvement of the toxicity of compounds. In addition, 3-pyridine-containing and substituted 3-pyridine-containing compounds (TF11-TF14) were also synthesized. Compared to TF11 (EC50 = 199.0 nM) with an unsubstituted pyridine ring, the activity of TF12~TF14 showed different degrees of increase, among which TF12 (EC50 = 7.8 nM) was the most active compound. duction of hydroxymethyl benzene ring may be detrimental to the improvement of the toxicity of compounds. In addition, 3-pyridine-containing and substituted 3-pyridine-containing compounds (TF11-TF14) were also synthesized. Compared to TF11 (EC50 = 199.0 nM) with an unsubstituted pyridine ring, the activity of TF12~TF14 showed different degrees of increase, among which TF12 (EC50 = 7.8 nM) was the most active compound.  According to the results in Table 1, all compounds (EC 50 = 7.6~199.0 nM) had potent inhibitory activity against WT HIV-1. TF2 (EC 50 = 7.6 nM), TF4 (EC 50 = 7.8 nM) and TF12 (EC 50 = 7.8 nM) exhibited the best anti-HIV-1 activity, being comparable to that of the second-generation drug ETR (EC 50 = 3.0 nM) and much superior to those of AZT (EC 50 = 27.3 nM) and NVP (EC 50 = 122.6 nM). Moreover, at a concentration of 279,329.6 nM, TF2 exhibited no cytotoxicity, which contributes to its higher SI values (SI > 36,610.9) toward HIV-1 IIIB. None of the compounds had inhibitory activity against HIV-2, which demonstrated that these compounds belong to typical HIV-1 inhibitors.  The preliminary SAR was summarized as follows. Inhibitory activity differed depending on the types and positions of substituents on the benzene ring introduced to the left wing by a CH 2 O linker. For the compounds containing para-substituted benzene ring, the order of potency was as follows: -CN (TF2, EC 50 = 7.6 nM) > -OCH 3  Furthermore, we found that the substitution of different positions on the introduced benzene rings had a certain effect on cytotoxicity. Compared with compounds containing orthoor meta-substituted benzene rings, TF1 (-H, CC 50 > 295,865.0 nM), TF2 (para-CN, CC 50 > 279,329.6 nM), TF5 (para-OCH 3 , CC 50 = 253,718.1 nM), and TF15 (para-NO 2 , CC 50 = 215,001.1 nM) showed low cytotoxicity than that of RPV (CC 50 = 3.89 µM). This suggested that the introduction of the benzene ring and para-substituted benzene ring may help reduce cytotoxicity. However, all three compounds containing hydroxymethyl substituted benzene rings, TF8 (CC 50 = 2917.9 nM), TF9 (CC 50 = 3458.6 nM), TF10 (CC 50 = 4804.6 nM), showed a sharp increase in cytotoxicity compared to TF1, which suggested that the introduction of hydroxymethyl benzene ring may be detrimental to the improvement of the toxicity of compounds.
To verify the target of synthesized compounds (TF1-TF17), they were further tested for enzyme inhibitory activity by the ELISA method [38], and the results are shown in Table 4. All compounds (IC 50 = 0.038~0.438 µM) exhibited potent inhibitory activity against WT HIV-1 RT, being superior to that of NVP (IC 50 = 0.568 µM). The three most potent compounds against WT HIV-1 mutant, TF2 (IC 50 = 0.055 µM), TF4 (IC 50 = 0.050 µM) and TF12 (IC 50 = 0.038 µM), displayed higher enzyme inhibitory activity. The weaker inhibitors of the WT HIV-1 strain showed reduced inhibitory activity, such as TF11 and TF16 (IC 50 = 0.483 and 0.160 µM, respectively). It is noteworthy that the antiviral activity of some compounds was inconsistent with their enzyme-inhibitory potency to some extent. This discrepancy may be caused by the variations in the HIV-1 RT-substrate binding affinities and polymerase processivity on different nucleic acid templates [39]. Nonetheless, these novel synthesized compounds functioned as traditional NNRTIs.

Molecular Dynamics Simulation Studies
In order to predict the binding modes of these novel synthesized diarylpyrimidines in the NNRTI-binding pocket and initially explain the anti-resistance profiles to the K103N strain of TF2, MD simulations were performed in detail using the software Schrödinger [40]. The co-crystal structures of HIV-1 WT RT in complex with RPV (PDB code:2ZD1) [22] and HIV-1 K103N RT in complex with RPV (PDB code:3MEG) [27] were chosen as templates for docking studies. The binding modes of TF2 to the allosteric pocket in WT and K103N HIV-1 RT were investigated by running a 500 ns MD simulation. Figure 5 shows that the root-mean-square deviation (RMSD) from the initial structure of TF2 was computed throughout the MD simulation for all systems. Figure 5A illustrates the RMSD of ligand TF2; the plot showed that the inhibitor in the WT and K103N mutant strains had deviated from the starting structure. Figure 5B illustrates the RMSD of the complex of RT and ligand TF2, which formed different conformational during the MD time. As a result of the narrow range of RMSD values, these conformational ensembles were structurally similar. The binding modes of TF2 to the allosteric pocket in WT and K103N HIV-1 RT were investigated by running a 500 ns MD simulation. Figure 5 shows that the root-meansquare deviation (RMSD) from the initial structure of TF2 was computed throughout the MD simulation for all systems. Figure 5A illustrates the RMSD of ligand TF2; the plot showed that the inhibitor in the WT and K103N mutant strains had deviated from the starting structure. Figure 5B illustrates the RMSD of the complex of RT and ligand TF2, which formed different conformational during the MD time. As a result of the narrow range of RMSD values, these conformational ensembles were structurally similar. Compound TF2 bound to WT and K103N RTs in similar binding modes, the binding modes of TF2 within the NNRTI-binding pocket resembled other diarylpyrimidines as the typical conformation (U shape), as noted in the snapshots collected at the end of the MD trajectories ( Figure 6). However, there were some differences in the orientation of some groups and the distance between them and several remarkable features and wellknown interactions were delineated here.
In the WT RT protein ( Figure 6A), the right benzonitrile moiety was situated in the tolerant region I, where the cyano group could form a water-mediated hydrogen bond with K103. Stable hydrogen-bonding interactions were formed between the 2-aminopyrimidine ring of TF2 and the backbone (both NH and C = O units) of K101. The left 2,4,6trisubstituted moiety remained stable in a hydrophobic sub-pocket formed by Y181, Y188, and W229. Moreover, the newly introduced p-cyanobenzonitrile group formed π-π stacking interaction with Y188 and hydrogen-bonding interaction with an extra amino acid residue H221. In the K103N RT protein ( Figure 6B), compared to the binding mode in WT RT, the overall structure of TF2 was shifted, causing it to move away from K101 and affecting hydrogen-bonding interactions. A hydrogen bond was formed between the pyrimidine-bound NH of TF2 and the carbonyl oxygen of K101, but the N atom on the pyrimidine ring formed water bridge-mediated hydrogen bonds with K101 and E138. Besides, the left 2,4,6-trisubstituted moiety was also positioned in the hydrophobic subpocket, and the phenyl group formed π-π stacking interaction with the indole ring of W229. Similarly, the p-cyanophenyl moiety extended out of the hydrophobic channel, but no additional interactions with surrounding amino acids were found.
MMGBSA calculations [41] were used to determine the relative binding strengths of the compound TF2 to the WT and K103N HIV-1 RTs. These estimations are shown in Table 5. Accordingly, the binding affinity (ΔGbind) of TF2 to WT RT was higher than that to K103N RT, supporting the experimental EC50 values. Two important interactions observed in the complex of TF2 and WT were the hydrogen-bonding interaction with an extra residue H221 and a water-mediated hydrogen bond with K103. The H bond (ΔGHbond) contributed more to TF2 binding to WT RT comparing K103N RT, which was consistent with the MD studies. Compound TF2 bound to WT and K103N RTs in similar binding modes, the binding modes of TF2 within the NNRTI-binding pocket resembled other diarylpyrimidines as the typical conformation (U shape), as noted in the snapshots collected at the end of the MD trajectories ( Figure 6). However, there were some differences in the orientation of some groups and the distance between them and several remarkable features and well-known interactions were delineated here.
In the WT RT protein ( Figure 6A), the right benzonitrile moiety was situated in the tolerant region I, where the cyano group could form a water-mediated hydrogen bond with K103. Stable hydrogen-bonding interactions were formed between the 2-aminopyrimidine ring of TF2 and the backbone (both NH and C = O units) of K101. The left 2,4,6-trisubstituted moiety remained stable in a hydrophobic sub-pocket formed by Y181, Y188, and W229. Moreover, the newly introduced p-cyanobenzonitrile group formed π-π stacking interaction with Y188 and hydrogen-bonding interaction with an extra amino acid residue H221. In the K103N RT protein ( Figure 6B), compared to the binding mode in WT RT, the overall structure of TF2 was shifted, causing it to move away from K101 and affecting hydrogenbonding interactions. A hydrogen bond was formed between the pyrimidine-bound NH of TF2 and the carbonyl oxygen of K101, but the N atom on the pyrimidine ring formed water bridge-mediated hydrogen bonds with K101 and E138. Besides, the left 2,4,6-trisubstituted moiety was also positioned in the hydrophobic sub-pocket, and the phenyl group formed π-π stacking interaction with the indole ring of W229. Similarly, the p-cyanophenyl moiety extended out of the hydrophobic channel, but no additional interactions with surrounding amino acids were found.
MMGBSA calculations [41] were used to determine the relative binding strengths of the compound TF2 to the WT and K103N HIV-1 RTs. These estimations are shown in Table 5. Accordingly, the binding affinity (∆G bind ) of TF2 to WT RT was higher than that to K103N RT, supporting the experimental EC 50 values. Two important interactions observed in the complex of TF2 and WT were the hydrogen-bonding interaction with an extra residue H221 and a water-mediated hydrogen bond with K103. The H bond (∆G Hbond ) contributed more to TF2 binding to WT RT comparing K103N RT, which was consistent with the MD studies.  Overall, a comparison with WT RT revealed that K103N mutation would destabilize the binding mode of TF2 by partially disrupting the protein-ligand hydrophobic and hydrogen bond interactions, which could also contribute to the reduced activity against this strain. Moreover, the newly introduced p-cyanophenyl moiety reached the primer grip region, but no interaction with the key amino acid here was observed. It pointed to the outside of the β12, which may not be an ideal location to interact with amino acids in the primer grip region. This also inspired us to pay attention to the selection of linkers to make it more oriented to explore the primer grip region.

Synthesis of Compounds
On a column filled with Silica Gel GF254 for thin-layer chromatography (TLC) and spots were observed using UV radiation at 254 and 365 nm wavelengths. The melting  Overall, a comparison with WT RT revealed that K103N mutation would destabilize the binding mode of TF2 by partially disrupting the protein-ligand hydrophobic and hydrogen bond interactions, which could also contribute to the reduced activity against this strain. Moreover, the newly introduced p-cyanophenyl moiety reached the primer grip region, but no interaction with the key amino acid here was observed. It pointed to the outside of the β12, which may not be an ideal location to interact with amino acids in the primer grip region. This also inspired us to pay attention to the selection of linkers to make it more oriented to explore the primer grip region.

Synthesis of Compounds
On a column filled with Silica Gel GF254 for thin-layer chromatography (TLC) and spots were observed using UV radiation at 254 and 365 nm wavelengths. The melting points (mp) of compounds were determined using a micro melting point meter. Flash column chro-matography was carried out on a column filled with Silica Gel 60 (200-300 mesh). 1 H NMR and 13 C NMR spectra were acquired on a Bruker AV-400 spectrometer with tetramethylsilane as the internal standard and DMSO-d 6 as solvent. The relevant mass spectrometry data were measured by a Standard G1313A Autosampler instrument. Reagents used in this work required no further purification and were bought from commercial sources.
Author Contributions: T.Z. and Z.Z. designed and carried out the experiments and wrote the paper. C.P. and E.D.C. carried out the bio-experiments. F.Z. carried out dynamic simulation study. Z.S. and F.Z. modified the paper. X.L., P.Z. and D.K. provided study design and guidance. All authors have read and agreed to the published version of the manuscript.