Toward Structurally Novel and Metabolically Stable HIV-1 Capsid-Targeting Small Molecules.

HIV-1 capsid protein (CA) plays an important role in many steps of viral replication and represents an appealing antiviral target. Several CA-targeting small molecules of various chemotypes have been studied, but the peptidomimetic PF74 has drawn particular interest due to its potent antiviral activity, well-characterized binding mode, and unique mechanism of action. Importantly, PF74 competes against important host factors for binding, conferring highly desirable antiviral phenotypes. However, further development of PF74 is hindered by its prohibitively poor metabolic stability, which necessitates the search for structurally novel and metabolically stable chemotypes. We have conducted a pharmacophore-based shape similarity search for compounds mimicking PF74. We report herein the analog synthesis and structure-activity relationship (SAR) of two hits from the search, and a third hit designed via molecular hybridization. All analogs were characterized for their effect on CA hexamer stability, antiviral activity, and cytotoxicity. These assays identified three active compounds that moderately stabilize CA hexamer and inhibit HIV-1. The most potent analog (10) inhibited HIV-1 comparably to PF74 but demonstrated drastically improved metabolic stability in liver microsomes (31 min vs. 0.7 min t1/2). Collectively, the current studies identified a structurally novel and metabolically stable PF74-like chemotype for targeting HIV-1 CA.


Introduction
Human immunodeficiency virus 1 (HIV-1) encodes a Gag polyprotein which contains multiple protein domains for viral assembly and release: matrix (p17 MA), capsid (p24 CA), nucleocapsid (p7 NC), p6 and spacer peptides Sp1 and Sp2 [1]. Gag polyproteins assemble to form the immature viral capsid core, which upon protease cleavage rearranges into a fullerene-shaped mature capsid core [2] comprising approximately 250 CA hexamers and exactly 12 asymmetrically-distributed CA pentamers [3]. CA plays essential roles in viral assembly and multiple events during viral replication [4,5]. CA-CA interactions drive the assembly and disassembly of viral capsid, and capsid core stability is important for reverse transcription, nuclear entry, and cloaking of the viral DNA product from host nucleic acid sensing Despite the attractive antiviral profile and the unique bimodal mechanism of action, PF74 is not a viable drug candidate due primarily to its prohibitively poor metabolic stability [34]. In human liver microsomes (HLMs), the half-life (t1/2) of PF74 was less than 1 min [35,36]. This major deficiency strongly necessitates targeted efforts to search for structurally novel and metabolically stable small molecules capable of binding to the PF74 binding pocket. Along this line, the Zhan and Liu group recently reported PF74-like compounds where an easier and synthetically more accessible 1,2,3triazole ring was substituted for the indole moiety [35,36]. However, such compounds were considerably (> 10-fold) less potent than PF74 in antiviral assays and were essentially as unstable in HLMs [35,36]. Our current work features a pharmacophore-based shape similarity search approach based on PF74 (Figure 2A). Molecular similarity [37,38] is an important concept in medicinal chemistry and drug discovery. Hit generation using similarity search is based on the premise that similar chemical structure confers similar biological activity. In our search, we extracted the 3D conformation of PF74 from reported co-crystal structure (PDB: 4XFZ [25]) and defined 7 pharmacophore points ( Figure 2A) according to its mode of binding, including two H-bond donors (D1 and D2), two H-bond acceptors (A1 and A2), as well as three hydrophobic moieties (H1, H2 and H3). This 3D pharmacophore was then used to screen against a subset of the ZINC database [39] (100K compounds) using the program PHASE [40,41]. Hit ranking was based on the number of pharmacophore points satisfied and the lead like properties as predicted by the Lipinski's rule of 5 [42]. From this similarity search we selected two hits (2 and 22) for analog synthesis and SAR ( Figure  2B). Molecular hybridization [43] between hit 2 and PF74 also generated hit 11, which was also  [25]). The pocket formed around H3 and H4 of the CA NTD (cyan), and H8 and H9 of the adjacent CA CTD (green) accommodates the binding of multiple ligands, including host factors Nup153 and CPSF6, and small molecules PF74 and BI-2. Shown in the pocket is PF74 (pointed to with an arrow); (B) chemical structures of PF74 and BI-2.
Viruses 2020, 12, x FOR PEER REVIEW 4 of 16 subjected to analog synthesis ( Figure 2B). We report herein the analog synthesis and SAR on all three hits. In the end, our best compound inhibited HIV-1 with an EC50 of 1.6 μM, and more important exhibited a half-life (t1/2) 44-fold longer than PF74.

Materials and Methods
All analogs were synthesized using procedures described in Scheme 1 and 2, fully characterized with 1 H NMR, 13 C NMR, and HRMS, and displayed a purity of ≥95% as determined by HPLC. Detailed synthetic procedures and compound characterization data are included in Supplemental Materials.
The general synthetic strategy for major analogs  tested in this work is described here

Materials and Methods
All analogs were synthesized using procedures described in Schemes 1 and 2, fully characterized with 1 H NMR, 13 C NMR, and HRMS, and displayed a purity of ≥95% as determined by HPLC. Detailed synthetic procedures and compound characterization data are included in Supplemental Materials.
The general synthetic strategy for major analogs  tested in this work is described here (Schemes 1 and 2) and the synthesis for others is outlined in supporting information (Schemes S1 and S2). Commercially available 1H-benzo[d] [1,3] oxazine-2,4-dione (a) was reacted with glycine methyl ester hydrochloride in the presence of Et 3  The synthesis of hit 22 analogs 22-39 was achieved with a straightforward approach described in Scheme 2. Hydrazides (e) reacted with ethyl isothiocyanatoacetate in ACN to give thiosemihydrazides (f) that were converted to triazole acid derivatives (g) through in situ cyclization and hydrolysis when treated with aqueous NaOH. Amide coupling of these triazole acid derivatives (g) with various amines in the presence of HATU and DIPEA in DMF resulted in analogs 22-39. Scheme  The synthesis of hit 22 analogs 22-39 was achieved with a straightforward approach described in Scheme 2. Hydrazides (e) reacted with ethyl isothiocyanatoacetate in ACN to give thiosemihydrazides (f) that were converted to triazole acid derivatives (g) through in situ cyclization and hydrolysis when treated with aqueous NaOH. Amide coupling of these triazole acid derivatives (g) with various amines in the presence of HATU and DIPEA in DMF resulted in analogs 22-39.

of 16
The synthesis of hit 22 analogs 22-39 was achieved with a straightforward approach described in Scheme 2. Hydrazides (e) reacted with ethyl isothiocyanatoacetate in ACN to give thiosemihydrazides (f) that were converted to triazole acid derivatives (g) through in situ cyclization and hydrolysis when treated with aqueous NaOH. Amide coupling of these triazole acid derivatives (g) with various amines in the presence of HATU and DIPEA in DMF resulted in analogs 22-39.

Biology Cells
TZM-GFP cells are a modified version of TZM-bl cells and contain an integrated nlsGFP reporter gene under the transcriptional control of the HIV-1 long terminal repeat (LTR) [44,45]. TZM-GFP cells were kindly provided by Dr. Marc Johnson (University of Missouri-Columbia, Columbia, MO) and cultured in DMEM supplemented with 10% fetal bovine serum (FBS; Hyclone). HEK293-FT cells were also cultured in DMEM supplemented with 10% FBS. MT-2 cells were grown in RPMI supplemented with 10% heat-inactivated FBS. All cells were grown at 37 °C and maintained in humidified atmosphere containing 5% CO2.

Biology Cells
TZM-GFP cells are a modified version of TZM-bl cells and contain an integrated nlsGFP reporter gene under the transcriptional control of the HIV-1 long terminal repeat (LTR) [44,45]. TZM-GFP cells were kindly provided by Dr. Marc Johnson (University of Missouri-Columbia, Columbia, MO) and cultured in DMEM supplemented with 10% fetal bovine serum (FBS; Hyclone). HEK293-FT cells were also cultured in DMEM supplemented with 10% FBS. MT-2 cells were grown in RPMI supplemented with 10% heat-inactivated FBS. All cells were grown at 37 • C and maintained in humidified atmosphere containing 5% CO 2 .

Thermal Shift Assays (TSAs) to Screen Compounds for Effect on HIV-1 CA Hexamer Stability
Compounds were screened for their effect on CA hexamer stability using purified covalently-crosslinked hexameric CA A14C/E45C/W184A/M185A (CA121). CA121 cloned in a pET11a expression plasmid was provided by Dr. Owen Pornillos (University of Virginia, Charlottesville, VA). Protein was expressed in E. coli BL21(DE3)RIL and purified as reported previously [26]. The TSA was conducted as previously described [46][47][48] using the PikoReal Real-Time PCR (Thermo Fisher Scientific, Waltham, MA, USA) or the QuantStudio 3 Real-Time PCR (Thermo Fisher Scientific) systems. Each reaction contained 10 µL of 15 µM CA121 (7.5 µM final concentration) in 50 mM sodium phosphate buffer (pH 8.0), 10 µL of 2× Sypro Orange Protein Gel Stain (Life Technologies, Carlsbad, CA, USA) in 50 mM sodium phosphate buffer (pH 8.0) and 0.2 µL of DMSO (control) or compound. Compounds were tested at a final concentration of 20 µM. The plate was heated from 25 to 95 • C with a heating rate of 0.2 • C/10 s. The fluorescence intensity was measured with an Ex range of 475-500 nm and Em range of 520-590 nm. The differences in the melting temperature (∆Tm) of CA hexamer in DMSO (T0) verses in the presence of compound (Tm) were calculated using the following formula: ∆Tm = Tm − T0.

Virus Production
The wild-type laboratory HIV-1 strain, HIV-1 NL4-3 [49], was produced using a pNL4-3 vector that was obtained through the NIH AIDS Reagent Program. HIV-1 NL4-3 was generated by transfecting HEK 293FT cells in a T75 flask with 10 µg of the pNL4-3 vector and FuGENE ® HD Transfection Reagent (Promega, Madison, WI, USA). Supernatant was harvested 48-72 h post-transfection and transferred to MT2 cells for viral propagation. Virus was harvested when syncytia formation was observed, which took 3-5 days. The viral supernatant was then concentrated using 8% w/v PEG 8000 overnight at 4 • C, followed by centrifugation for 40 min at 3500 rpm. The resulting viral-containing pellet was concentrated 10-fold by resuspension in DMEM without FBS and stored at −80 • C.

Anti-HIV-1 and Cytotoxicity Assays
Anti-HIV-1 activity of PF74 and related analogs was examined in TZM-GFP cells. The potency of HIV-1 inhibition by a compound was based on its inhibitory effect on viral LTR-activated GFP expression compared with that of compound-free (DMSO) controls. Briefly, TZM-GFP cells were plated at density of 1 × 10 4 cells per well in a 96-well plate. 24 h later, media was replaced with increasing concentrations of compound. 24 h post treatment, cells were exposed to an HIV-1 strain (MOI = 1). After incubation for 48 h, anti-HIV-1 activity was assessed by counting the number of GFP positive Cytotoxicity of each compound was also determined in TZM-GFP cells. Cells were plated at a density of 1 × 10 4 cells per well in a 96-well plate and were continuously exposed to increasing concentrations of a compound for 72 h. The number of viable cells in each well was determined using a Cell Proliferation Kit II (XTT), and 50% cytotoxicity concentration (CC 50 ) values were determined. All the cell-based assays were conducted in duplicate of at least two independent experiments and the average values were determined.
For calculation of EC 50 and CC 50 dose response curves, values were plotted in GraphPad Prism 5 and analyzed with the log (inhibitor) vs. normalized response-Variable slope equation. Final values were calculated in each independent assay and the average values were determined. Statistical analysis (calculation of standard deviation) was performed using Microsoft Excel.

Microsomal Stability Assay
The in vitro microsomal stability assay was conducted in duplicate in mouse and human liver microsomal systems, which were supplemented with nicotinamide adenine dinucleotide phosphate (NADPH) as a cofactor. Briefly, a compound (1 µM final concentration) was pre-incubated, in the absence or presence of 0.5 µM Cobicistat (CYP 3A inhibitor, purchased from medchemexpress.com and verified with LCMS), with the reaction mixture containing liver microsomal protein (0.5 mg/mL final concentration) and MgCl 2 (1 mM final concentration) in 0.1 M potassium phosphate buffer (pH 7.4) at 37 • C for 15 min. The reaction was initiated by addition of 1 mM NADPH, followed by incubation at 37 • C. A negative control was performed in parallel in the absence of NADPH to measure any chemical instability or non-NADPH dependent enzymatic degradation for each compound. At various time points (0, 5, 15, 30 and 60 min), 1 volume of reaction aliquot was taken and quenched with 3 volumes of acetonitrile containing an appropriate internal standard and 0.1% formic acid. The samples were then vortexed and centrifuged at 15,000 rpm for 5 min at 4 • C. The supernatants were collected and analyzed by LC-MS/MS to determine the in vitro metabolic half-life (t 1/2 ).

Molecular Modeling
Molecular modeling was performed using the Schrödinger small molecule drug discovery suite 2019-1 [50]. The crystal structure of native HIV-1 capsid protein in complex with PF74 [25] was retrieved from the protein data bank (PDB code: 4XFZ [25]). The above structure was analyzed using Maestro [51] (Schrödinger; LLC: New York, NY, USA) and subjected to a docking protocol that involves several steps including preparing protein of interest, grid generation, ligand preparation, and docking. The crystal structure was refined using the protein preparation wizard [52] (Schrödinger; LLC: New York, NY, USA.) wherein missing hydrogen atoms, side chains, and loops were added using Prime and minimized using the OPLS 3e force field [53] to optimize the hydrogen bonding network and converge the heavy atoms to an rmsd of 0.3 Å. The receptor grid generation tool in Maestro (Schrödinger; LLC: New York, NY, USA) was used to define an active site around the native ligand PF74 to cover all the residues within 12 Å. All the compounds were drawn using Maestro and subjected to Lig Prep to generate conformers, possible protonation at pH of 7 ± 2 that serves as an input for docking process. All the dockings were performed using Glide XP [54] (Glide: version 8.2) with the van der Waals radii of nonpolar atoms for each of the ligands were scaled by a factor of 0.8. The solutions were further refined by post docking and minimization under implicit solvent to account for protein flexibility. The residue numbers of HIV-1 capsid protein used in the discussion and the figures were based on the native HIV-1 capsid protein.

Results
All analogs synthesized for each hit were assessed first with a biophysical protein stability assay, or thermal shift assay, where the effect of the compound was measured by the change in protein melting point compared to the DMSO control (∆Tm). A positive value in ∆Tm indicates a stabilizing Viruses 2020, 12, 452 7 of 16 effect on the protein and a negative value indicates a destabilizing effect. Of note, the target CA protein is in a covalently crosslinked hexameric state. Thus, the ∆Tm values likely reflect local changes that may affect stabilization and exclude inter-hexamer effects, which are important correlates of overall capsid core stability. To simplify presentation of the data, we refer to the effects of these compounds as stabilization or destabilization of "CA hexamer." All compounds were then screened at 20 µM against HIV-1 in a cell-based assay to determine antiviral activity. Compounds demonstrating significant inhibition were further tested in a dose-response fashion for antiviral EC 50 values. All compounds were tested for cytotoxicity either by screening at 100 or 50 µM, or determination of CC 50 values. The benchmark compound PF74 was resynthesized and tested in these assays (1, ∆Tm = 7.4 • C, EC 50 = 0.61 µM, CC 50 = 76 µM). The best compound (10) was also tested for liver microsomal stability. Molecular modeling was performed for a few selected compounds to help understand the SAR.

SAR of Hit 11 (R 1 , R 3 , and R 4 )
Hit 11 was designed from hit 2 and PF74 via molecular hybridization. As a result, the skeleton of 11 is very similar to that of PF74, except that the indole moiety was replaced with a bioisosteric quinazoline-2, 4-dione moiety. A general SAR observation was that when R 1 was a bulky alkyl group, compounds (14-21) did not show significant antiviral activity at 20 μM or cytotoxicity at 100 μM (Table 2), though most analogs weakly stabilized CA hexamer (ΔTm = 0.5-0.7 °C). However, when R 1 was a benzyl group to mimic PF74, the activity profiles changed substantially (analogs 11-13) and demonstrated a strong dependence on R 3 . When R 3 was a methyl group, analog 11 (ΔTm = 2.7 °C, EC50 = 6.9 μM) and analog 13 (ΔTm = 2.4 °C, EC50 = 8.0 μM) both moderately stabilized CA hexamer a Concentration of compound inhibiting HIV-1 replication by 50%, expressed as the mean ± standard deviation from at least two independent experiments. b Concentration of compound causing 50% cell death, expressed as the mean ± standard deviation from at least two independent experiments. c TSA: thermal shift assay. ΔTm: change of CA crosslinked hexamer melting point in the presence of compound minus DMSO control.

SAR of Hit 11 (R 1 , R 3 , and R 4 )
Hit 11 was designed from hit 2 and PF74 via molecular hybridization. As a result, the skeleton of 11 is very similar to that of PF74, except that the indole moiety was replaced with a bioisosteric quinazoline-2, 4-dione moiety. A general SAR observation was that when R 1 was a bulky alkyl group, compounds (14-21) did not show significant antiviral activity at 20 μM or cytotoxicity at 100 μM (Table 2), though most analogs weakly stabilized CA hexamer (ΔTm = 0.5-0.7 °C). However, when R 1 was a benzyl group to mimic PF74, the activity profiles changed substantially (analogs 11-13) and demonstrated a strong dependence on R 3 . When R 3 was a methyl group, analog 11 (ΔTm = 2.7 °C, EC50 = 6.9 μM) and analog 13 (ΔTm = 2.4 °C, EC50 = 8.0 μM) both moderately stabilized CA hexamer

SAR of Hit 11 (R 1 , R 3 , and R 4 )
Hit 11 was designed from hit 2 and PF74 via molecular hybridization. As a result, the skeleton of 11 is very similar to that of PF74, except that the indole moiety was replaced with a bioisosteric quinazoline-2, 4-dione moiety. A general SAR observation was that when R 1 was a bulky alkyl group, compounds (14-21) did not show significant antiviral activity at 20 μM or cytotoxicity at 100 μM (Table 2), though most analogs weakly stabilized CA hexamer (ΔTm = 0.5-0.7 °C). However, when R 1 was a benzyl group to mimic PF74, the activity profiles changed substantially (analogs 11-13) and demonstrated a strong dependence on R 3 . When R 3 was a methyl group, analog 11 (ΔTm = 2.7 °C, EC50 = 6.9 μM) and analog 13 (ΔTm = 2.4 °C, EC50 = 8.0 μM) both moderately stabilized CA hexamer

SAR of Hit 11 (R 1 , R 3 , and R 4 )
Hit 11 was designed from hit 2 and PF74 via molecular hybridization. As a result, the skeleton of 11 is very similar to that of PF74, except that the indole moiety was replaced with a bioisosteric quinazoline-2, 4-dione moiety. A general SAR observation was that when R 1 was a bulky alkyl group, compounds (14-21) did not show significant antiviral activity at 20 μM or cytotoxicity at 100 μM (Table 2), though most analogs weakly stabilized CA hexamer (ΔTm = 0.5-0.7 °C). However, when R 1 was a benzyl group to mimic PF74, the activity profiles changed substantially (analogs 11-13) and demonstrated a strong dependence on R 3 . When R 3 was a methyl group, analog 11 (ΔTm = 2.7 °C, EC50 = 6.9 μM) and analog 13 (ΔTm = 2.4 °C, EC50 = 8.0 μM) both moderately stabilized CA hexamer

SAR of Hit 11 (R 1 , R 3 , and R 4 )
Hit 11 was designed from hit 2 and PF74 via molecular hybridization. As a result, the skeleton of 11 is very similar to that of PF74, except that the indole moiety was replaced with a bioisosteric quinazoline-2, 4-dione moiety. A general SAR observation was that when R 1 was a bulky alkyl group, compounds (14-21) did not show significant antiviral activity at 20 μM or cytotoxicity at 100 μM (Table 2), though most analogs weakly stabilized CA hexamer (ΔTm = 0.5-0.7 °C). However, when R 1 was a benzyl group to mimic PF74, the activity profiles changed substantially (analogs 11-13) and demonstrated a strong dependence on R 3 . When R 3 was a methyl group, analog 11 (ΔTm = 2.7 °C, EC50 = 6.9 μM) and analog 13 (ΔTm = 2.4 °C, EC50 = 8.0 μM) both moderately stabilized CA hexamer

SAR of Hit 11 (R 1 , R 3 , and R 4 )
Hit 11 was designed from hit 2 and PF74 via molecular hybridization. As a result, the skeleton of 11 is very similar to that of PF74, except that the indole moiety was replaced with a bioisosteric quinazoline-2, 4-dione moiety. A general SAR observation was that when R 1 was a bulky alkyl group, compounds (14-21) did not show significant antiviral activity at 20 μM or cytotoxicity at 100 μM (Table 2), though most analogs weakly stabilized CA hexamer (ΔTm = 0.5-0.7 °C). However, when R 1 was a benzyl group to mimic PF74, the activity profiles changed substantially (analogs 11-13) and demonstrated a strong dependence on R 3 . When R 3 was a methyl group, analog 11 (ΔTm = 2.7 °C, EC50 = 6.9 μM) and analog 13 (ΔTm = 2.4 °C, EC50 = 8.0 μM) both moderately stabilized CA hexamer

SAR of Hit 11 (R 1 , R 3 , and R 4 )
Hit 11 was designed from hit 2 and PF74 via molecular hybridization. As a result, the skeleton of 11 is very similar to that of PF74, except that the indole moiety was replaced with a bioisosteric quinazoline-2, 4-dione moiety. A general SAR observation was that when R 1 was a bulky alkyl group, compounds (14-21) did not show significant antiviral activity at 20 μM or cytotoxicity at 100 μM (Table 2), though most analogs weakly stabilized CA hexamer (ΔTm = 0.5-0.7 °C). However, when R 1 was a benzyl group to mimic PF74, the activity profiles changed substantially (analogs 11-13) and demonstrated a strong dependence on R 3 . When R 3 was a methyl group, analog 11 (ΔTm = 2.7 °C, EC50 = 6.9 μM) and analog 13 (ΔTm = 2.4 °C, EC50 = 8.0 μM) both moderately stabilized CA hexamer

SAR of Hit 11 (R 1 , R 3 , and R 4 )
Hit 11 was designed from hit 2 and PF74 via molecular hybridization. As a result, the skeleton of 11 is very similar to that of PF74, except that the indole moiety was replaced with a bioisosteric quinazoline-2, 4-dione moiety. A general SAR observation was that when R 1 was a bulky alkyl group, compounds (14-21) did not show significant antiviral activity at 20 µM or cytotoxicity at 100 µM (Table 2), though most analogs weakly stabilized CA hexamer (∆Tm = 0.5-0.7 • C). However, when R 1 was a benzyl group to mimic PF74, the activity profiles changed substantially (analogs 11-13) and demonstrated a strong dependence on R 3 . When R 3 was a methyl group, analog 11 (∆Tm = 2.7 • C, EC 50 = 6.9 µM) and analog 13 (∆Tm = 2.4 • C, EC 50 = 8.0 µM) both moderately stabilized CA hexamer and inhibited HIV-1, without observed cytotoxicity (CC 50 >100 µM). By contrast, when R 3 was a hydrogen, the resulting analog (12) did not show activity in either assay. Overall, SARs around R 1 and R 3 within this series strongly indicate that both the benzyl group at R 1 and the methyl group at R 3 are important for potency. and inhibited HIV-1, without observed cytotoxicity (CC50 >100 μM). By contrast, when R 3 was a hydrogen, the resulting analog (12) did not show activity in either assay. Overall, SARs around R 1 and R 3 within this series strongly indicate that both the benzyl group at R 1 and the methyl group at R 3 are important for potency. and inhibited HIV-1, without observed cytotoxicity (CC50 >100 μM). By contrast, when R 3 was a hydrogen, the resulting analog (12) did not show activity in either assay. Overall, SARs around R 1 and R 3 within this series strongly indicate that both the benzyl group at R 1 and the methyl group at R 3 are important for potency. and inhibited HIV-1, without observed cytotoxicity (CC50 >100 μM). By contrast, when R 3 was a hydrogen, the resulting analog (12) did not show activity in either assay. Overall, SARs around R 1 and R 3 within this series strongly indicate that both the benzyl group at R 1 and the methyl group at R 3 are important for potency. and inhibited HIV-1, without observed cytotoxicity (CC50 >100 μM). By contrast, when R 3 was a hydrogen, the resulting analog (12) did not show activity in either assay. Overall, SARs around R 1 and R 3 within this series strongly indicate that both the benzyl group at R 1 and the methyl group at R 3 are important for potency. and inhibited HIV-1, without observed cytotoxicity (CC50 >100 μM). By contrast, when R 3 was a hydrogen, the resulting analog (12) did not show activity in either assay. Overall, SARs around R 1 and R 3 within this series strongly indicate that both the benzyl group at R 1 and the methyl group at R 3 are important for potency. and inhibited HIV-1, without observed cytotoxicity (CC50 >100 μM). By contrast, when R 3 was a hydrogen, the resulting analog (12) did not show activity in either assay. Overall, SARs around R 1 and R 3 within this series strongly indicate that both the benzyl group at R 1 and the methyl group at R 3 are important for potency. and inhibited HIV-1, without observed cytotoxicity (CC50 >100 μM). By contrast, when R 3 was a hydrogen, the resulting analog (12) did not show activity in either assay. Overall, SARs around R 1 and R 3 within this series strongly indicate that both the benzyl group at R 1 and the methyl group at R 3 are important for potency. and inhibited HIV-1, without observed cytotoxicity (CC50 >100 μM). By contrast, when R 3 was a hydrogen, the resulting analog (12) did not show activity in either assay. Overall, SARs around R 1 and R 3 within this series strongly indicate that both the benzyl group at R 1 and the methyl group at R 3 are important for potency.

SAR of Hit 22 (R 5 , R 6 , and R 7 )
Compound 22 is structurally quite different from PF74 and the other two hits (2 and 11), though it satisfied multiple pharmacophore points (Figure 2A), and hence, could be similar to PF74 in 3D shape. In addition, the 1,2,4-trazole is a well-known bioisostere [55] of carboxamide, which renders 22 functionally similar to PF74. The SARs of this series were centered around the two aromatic rings (R 5 and R 7 ). The most prominent observation was that analogs within this series did not significantly inhibit HIV-1 at 20 μM, and did not show cytotoxicity at 100 μM (Table 3). However, with the exception of 23, 32, and 33, all analogs demonstrated an impact on CA hexamer stability, which essentially validates our hit generation approach as the shape similarity search is target-based. Very interestingly, a weak CA hexamer destabilizing effect, rather than stabilizing effect, was observed with hit 22 (ΔTm = −0.5 °C). A similar destabilizing effect was also observed with the only other analog featuring a pyridine ring at R 5 (compound 31), though the effect was significantly more prominent (ΔTm = −1.2 °C). All other analogs feature either a phenyl or a biphenyl ring at R 5 , and in stark contrast, produced CA hexamer stabilizing effect (ΔTm = 0.5-0.9 °C). This weak stabilizing effect was not impacted by functional group substitution at either R 5 or R 7 . Finally, N-methyl substitution of analog 23 (ΔTm = 0 °C) yielded a compound 39 stabilizing CA hexamer (ΔTm = 1.2 °C). These observations strongly indicate that 1) pyridine at R 5 destabilizes the CA hexamer whereas a phenyl or biphenyl moiety at R 5 stabilizes the CA hexamer; 2) a methyl group at the para position of the phenyl ring at R 7 reinforces the destabilizing effect (

SAR of Hit 22 (R 5 , R 6 , and R 7 )
Compound 22 is structurally quite different from PF74 and the other two hits (2 and 11), though it satisfied multiple pharmacophore points (Figure 2A), and hence, could be similar to PF74 in 3D shape. In addition, the 1,2,4-trazole is a well-known bioisostere [55] of carboxamide, which renders 22 functionally similar to PF74. The SARs of this series were centered around the two aromatic rings (R 5 and R 7 ). The most prominent observation was that analogs within this series did not significantly inhibit HIV-1 at 20 μM, and did not show cytotoxicity at 100 μM (Table 3). However, with the exception of 23, 32, and 33, all analogs demonstrated an impact on CA hexamer stability, which essentially validates our hit generation approach as the shape similarity search is target-based. Very interestingly, a weak CA hexamer destabilizing effect, rather than stabilizing effect, was observed with hit 22 (ΔTm = −0.5 °C). A similar destabilizing effect was also observed with the only other analog featuring a pyridine ring at R 5 (compound 31), though the effect was significantly more prominent (ΔTm = −1.2 °C). All other analogs feature either a phenyl or a biphenyl ring at R 5 , and in stark contrast, produced CA hexamer stabilizing effect (ΔTm = 0.5-0.9 °C). This weak stabilizing effect was not impacted by functional group substitution at either R 5 or R 7 . Finally, N-methyl substitution of analog 23 (ΔTm = 0 °C) yielded a compound 39 stabilizing CA hexamer (ΔTm = 1.2 °C). These observations strongly indicate that 1) pyridine at R 5 destabilizes the CA hexamer whereas a phenyl or biphenyl moiety at R 5 stabilizes the CA hexamer; 2) a methyl group at the para position of the phenyl ring at R 7 reinforces the destabilizing effect (31 vs. 22) but does not impact the stabilizing

SAR of Hit 22 (R 5 , R 6 , and R 7 )
Compound 22 is structurally quite different from PF74 and the other two hits (2 and 11), though it satisfied multiple pharmacophore points (Figure 2A), and hence, could be similar to PF74 in 3D shape. In addition, the 1,2,4-trazole is a well-known bioisostere [55] of carboxamide, which renders 22 functionally similar to PF74. The SARs of this series were centered around the two aromatic rings (R 5 and R 7 ). The most prominent observation was that analogs within this series did not significantly inhibit HIV-1 at 20 μM, and did not show cytotoxicity at 100 μM (Table 3). However, with the exception of 23, 32, and 33, all analogs demonstrated an impact on CA hexamer stability, which essentially validates our hit generation approach as the shape similarity search is target-based. Very interestingly, a weak CA hexamer destabilizing effect, rather than stabilizing effect, was observed with hit 22 (ΔTm = −0.5 °C). A similar destabilizing effect was also observed with the only other analog featuring a pyridine ring at R 5 (compound 31), though the effect was significantly more prominent (ΔTm = −1.2 °C). All other analogs feature either a phenyl or a biphenyl ring at R 5 , and in stark contrast, produced CA hexamer stabilizing effect (ΔTm = 0.5-0.9 °C). This weak stabilizing effect was not impacted by functional group substitution at either R 5 or R 7 . Finally, N-methyl substitution of analog 23 (ΔTm = 0 °C) yielded a compound 39 stabilizing CA hexamer (ΔTm = 1.2 °C). These observations strongly indicate that 1) pyridine at R 5 destabilizes the CA hexamer whereas a phenyl or biphenyl moiety at R 5 stabilizes the CA hexamer; 2) a methyl group at the para position of the phenyl ring at R 7 reinforces the destabilizing effect (

SAR of Hit 22 (R 5 , R 6 , and R 7 )
Compound 22 is structurally quite different from PF74 and the other two hits (2 and 11), though it satisfied multiple pharmacophore points (Figure 2A), and hence, could be similar to PF74 in 3D shape. In addition, the 1,2,4-trazole is a well-known bioisostere [55] of carboxamide, which renders 22 functionally similar to PF74. The SARs of this series were centered around the two aromatic rings (R 5 and R 7 ). The most prominent observation was that analogs within this series did not significantly inhibit HIV-1 at 20 μM, and did not show cytotoxicity at 100 μM (Table 3). However, with the exception of 23, 32, and 33, all analogs demonstrated an impact on CA hexamer stability, which essentially validates our hit generation approach as the shape similarity search is target-based. Very interestingly, a weak CA hexamer destabilizing effect, rather than stabilizing effect, was observed with hit 22 (ΔTm = −0.5 °C). A similar destabilizing effect was also observed with the only other analog featuring a pyridine ring at R 5 (compound 31), though the effect was significantly more prominent (ΔTm = −1.2 °C). All other analogs feature either a phenyl or a biphenyl ring at R 5 , and in stark contrast, produced CA hexamer stabilizing effect (ΔTm = 0.5-0.9 °C). This weak stabilizing effect was not impacted by functional group substitution at either R 5 or R 7 . Finally, N-methyl substitution of analog 23 (ΔTm = 0 °C) yielded a compound 39 stabilizing CA hexamer (ΔTm = 1.2 °C). These observations strongly indicate that 1) pyridine at R 5 destabilizes the CA hexamer whereas a phenyl or biphenyl moiety at R 5 stabilizes the CA hexamer; 2) a methyl group at the para position of the phenyl ring at R 7 reinforces the destabilizing effect (31 vs. 22) but does not impact the stabilizing >20 >100 0.6 a Concentration of compound inhibiting HIV-1 replication by 50%, expressed as the mean ± standard deviation from at least two independent experiments. b Concentration of compound causing 50% cell death, expressed as the mean ± standard deviation from at least two independent experiments. c TSA: thermal shift assay. ∆Tm: change of CA crosslinked hexamer melting point in the presence of compound minus DMSO control.

SAR of Hit 22 (R 5 , R 6 , and R 7 )
Compound 22 is structurally quite different from PF74 and the other two hits (2 and 11), though it satisfied multiple pharmacophore points (Figure 2A), and hence, could be similar to PF74 in 3D shape. In addition, the 1,2,4-trazole is a well-known bioisostere [55] of carboxamide, which renders 22 functionally similar to PF74. The SARs of this series were centered around the two aromatic rings (R 5 and R 7 ). The most prominent observation was that analogs within this series did not significantly inhibit HIV-1 at 20 µM, and did not show cytotoxicity at 100 µM (Table 3). However, with the exception of 23, 32, and 33, all analogs demonstrated an impact on CA hexamer stability, which essentially validates our hit generation approach as the shape similarity search is target-based. Very interestingly, a weak CA hexamer destabilizing effect, rather than stabilizing effect, was observed with hit 22 (∆Tm = −0.5 • C). A similar destabilizing effect was also observed with the only other analog featuring a pyridine ring at R 5 (compound 31), though the effect was significantly more prominent (∆Tm = −1.2 • C). All other analogs feature either a phenyl or a biphenyl ring at R 5 , and in stark contrast, produced CA hexamer stabilizing effect (∆Tm = 0.5-0.9 • C). This weak stabilizing effect was not impacted by functional group substitution at either R 5 or R 7 . Finally, N-methyl substitution of analog 23 (∆Tm = 0 • C) yielded a compound 39 stabilizing CA hexamer (∆Tm = 1.2 • C). These observations strongly indicate that (1) pyridine at R 5 destabilizes the CA hexamer whereas a phenyl or biphenyl moiety at R 5 stabilizes the CA hexamer; (2) a methyl group at the para position of the phenyl ring at R 7 reinforces the destabilizing effect (31 vs. 22) but does not impact the stabilizing effect (28 vs. 24, 30 vs. 36); and (3) a methyl group at R 6 confers a CA hexamer stabilizing effect (39 vs. 23). Table 3. Anti-HIV-1 activity, cytotoxicity, and CA hexamer stability profiles of 22-39 (R 5 , R 6 , and R 7 ).

Metabolic Stability
To characterize the drug-like property of 10, our most potent antiviral compound from the current work, we conducted metabolic stability assays in both HLMs and mouse liver microsomes (MLMs). Metabolic stability is a major absorption, distribution, metabolism and excretion (ADME) property that profoundly impacts drug bioavailability [56]. Peptidomimetics are particularly susceptible to phase I metabolism, presumably because they are good substrates [57] for liver metabolizing enzyme subfamily cytochrome P450 3A (CYP3A), which is responsible for the metabolism of at least 50% of all current drugs [58]. It is known that PF74 is a severely flawed antiviral lead due to its prohibitively low metabolic stability [34]. This was confirmed in our metabolic stability assays, where the half-life (t 1/2 ) of PF74 is less than 1 min in both HLMs and MLMs (Table 4). By contrast, our compound 10 was decisively more stable, particularly in HLMs where its half-life (t 1/2 = 31 min) was 44-fold longer than that of PF74 (t 1/2 = 0.7 min). When tested in combination with a CYP3A inhibitor Cobicistat (Cobi) [59], 10 still exhibited significantly longer half-life than that of PF74. A t 1/2 >30 min in HLM generally indicates good in vivo metabolic stability and oral bioavailability. Collectively, these observations support our compound 10 as a viable antiviral hit, and corroborate the hypothesis that the poor metabolic stability of PF74 is due to CYP3A-mediated phase I metabolism.

Molecular Modeling
To understand some of the aforementioned SAR, we performed molecular modeling with selected compounds based on the co-crystal structure of native HIV-1 capsid protein bound to PF74 (PDB code: 4XFZ [25]). Compound 11 ( Figure 3A) bearing a quinazoline-2,4(1H,3H)-dione core in place of the indole ring of PF74 interacted with the same key residues in the CA hexamer as PF74. Observed key interactions included (i) hydrogen-bonding between N57 and the NH and carbonyl groups of the phenylalanine fragment on compound 11, between K70 and the carbonyl next to the phenylalanine fragment of 11, and between Q63 and the free NH of the quinazoline-2,4(1H,3H)-dione core of 11; (ii) cation-π interactions between protonated K70 and quinazoline-2,4(1H,3H)-dione aromatic ring. However, despite sharing the same binding site and similar key interactions, compound 11 was significantly less potent than PF74. In the meantime, a complete loss in potency was observed for leucine derived analog 14, which could be attributed to its shifting away from the binding site (circled), possibly in a binding mode shown in Figure 3B, and hence the loss of key interactions. This observation signified the importance of the phenylalanine core of these compounds for potency. In particular, the benzyl group of the phenylalanine core could play an important role in keeping a molecule inside the PF74-bound cavity through hydrophobic interactions [25] with surrounding residues L56, M66, and L69, as evident from the observed binding mode for compound 11 ( Figure 3A). This was consistent with the significant increase in the potency of compound 10 as well, which bears a benzyl group and an additional L-proline core. Like the PF74 backbone, the benzyl group of the phenylalanine fragment forces compound 10 to nestle in the cavity in such a way that the aniline ring of the L-proline core is extended to the adjacent CA NTD domain, resulting in an additional hydrogen-bonding between the carbonyl of the L-proline and NH 2 of N53, along with the typical H-bonding with N57, Q63, and K70 ( Figure 3C). Additionally, pyrrolidine core of L-proline is oriented in such a way that it has maximum interaction with A105, T107, and Y130. Expectedly, introduction of an L-cysteine derived core having an elongated backbone in place of phenylalanine in compound 8 forces the entire molecule out of the cavity, diminishing it potency ( Figure 3D). Compound 8 was found to be closer to the adjacent CA CTD and CA NTD domains, interacting with P34, R173, Q179, and K182 through H-bonding.
Viruses 2020, 12, x FOR PEER REVIEW 12 of 16 Compound 8 was found to be closer to the adjacent CACTD and CANTD domains, interacting with P34, R173, Q179, and K182 through H-bonding. In Figure A and D, H-bond and cation-π interactions are depicted as black dotted lines and double headed arrow, respectively. In (A) and (B), the capsid protein chain is colored cyan and key residues around binding site are colored olive. In (C) and (D), CANTD around helices H3 and H4 in chain A are colored cyan, and CACTD around helices H8 and H9 in the adjacent chain B are colored green. The nitrogen, oxygen, and sulfur atoms are colored blue, red, and yellow, respectively.

Discussion
Despite the approval of many HIV-1 antiviral regimens [60], a curative therapy remains elusive and HIV-1 continues to pose a global healthcare challenge. There is a need to develop new classes of HIV-1 drugs with distinct mechanisms of action to manage HIV-1 strains resistant to current drugs. The multifunctional HIV-1 CA represents an attractive target for novel antiviral discovery. PF74 is a CA-targeting small molecule which binds to a unique pocket between a viral CANTD and the adjacent CACTD, and competes against a few host factors important for viral replication. The antiviral profile of PF74 and its mode of CA binding are well characterized. However, PF74 is not a viable antiviral lead as it suffers from extremely low metabolic stability. Aiming to identify mechanistically similar yet structurally distinct small molecules with improved metabolic stability, we performed a pharmacophore-based shape similarity search based on PF74. Subsequently, we conducted analog  Figure A and D, H-bond and cation-π interactions are depicted as black dotted lines and double headed arrow, respectively. In (A) and (B), the capsid protein chain is colored cyan and key residues around binding site are colored olive. In (C) and (D), CA NTD around helices H3 and H4 in chain A are colored cyan, and CA CTD around helices H8 and H9 in the adjacent chain B are colored green. The nitrogen, oxygen, and sulfur atoms are colored blue, red, and yellow, respectively.

Discussion
Despite the approval of many HIV-1 antiviral regimens [60], a curative therapy remains elusive and HIV-1 continues to pose a global healthcare challenge. There is a need to develop new classes of HIV-1 drugs with distinct mechanisms of action to manage HIV-1 strains resistant to current drugs. The multifunctional HIV-1 CA represents an attractive target for novel antiviral discovery. PF74 is a CA-targeting small molecule which binds to a unique pocket between a viral CA NTD and the adjacent CA CTD , and competes against a few host factors important for viral replication. The antiviral profile of PF74 and its mode of CA binding are well characterized. However, PF74 is not a viable antiviral lead as it suffers from extremely low metabolic stability. Aiming to identify mechanistically similar yet structurally distinct small molecules with improved metabolic stability, we performed a pharmacophore-based shape similarity search based on PF74. Subsequently, we conducted analog synthesis and SAR for two hits (2 and 22) generated from the shape similarity search, as well as a third hit (11) designed via molecular hybridization. Overall, most of analogs exhibited weak yet discernible impacts on the stability of CA hexamer, which largely validates our hit generation approach. Three of the analogs (10, 11, and 13) showed moderate CA hexamer stabilizing effects and significant anti-HIV-1 activities. Particularly, compound 10 inhibited HIV-1 with an EC 50 of 1.6 µM, which is only 2.5-fold less potent that PF74. More importantly, compound 10 demonstrated drastically improved metabolic stability over PF74 in HLMs (t 1/2 = 31 min for 10 vs. 0.7 min for PF74). Molecular modeling indicates that our compound 10 binds comfortably in the PF74 binding pocket. Collectively, our data support 10 as a potent and metabolically stable HIV-1 CA-targeting antiviral lead.