Targeting HIV-1 Reverse Transcriptase Using a Fragment-Based Approach

Human immunodeficiency virus type I (HIV-1) is a retrovirus that infects cells of the host’s immune system leading to acquired immunodeficiency syndrome and potentially death. Although treatments are available to prevent its progression, HIV-1 remains a major burden on health resources worldwide. Continued emergence of drug-resistance mutations drives the need for novel drugs that can inhibit HIV-1 replication through new pathways. The viral protein reverse transcriptase (RT) plays a fundamental role in the HIV-1 replication cycle, and multiple approved medications target this enzyme. In this study, fragment-based drug discovery was used to optimize a previously identified hit fragment (compound B-1), which bound RT at a novel site. Three series of compounds were synthesized and evaluated for their HIV-1 RT binding and inhibition. These series were designed to investigate different vectors around the initial hit in an attempt to improve inhibitory activity against RT. Our results show that the 4-position of the core scaffold is important for binding of the fragment to RT, and a lead compound with a cyclopropyl substitution was selected and further investigated. Requirements for binding to the NNRTI-binding pocket (NNIBP) and a novel adjacent site were investigated, with lead compound 27—a minimal but efficient NNRTI—offering a starting site for the development of novel dual NNIBP-Adjacent site inhibitors.


Introduction
Human immunodeficiency virus type 1 (HIV-1) is a retrovirus that targets cells of the immune system of the human host. If untreated, HIV-1 infection leads to acquired immunodeficiency syndrome (AIDS) and eventually death. While advances in treatment have made HIV-1 a manageable chronic condition in many cases, the continued emergence of drug-resistant variants as well as drug intolerance and drug toxicity pose substantial threats to our current therapies [1]. Therefore, the proactive discovery of new and effective drug classes with distinct drug resistance profiles to existing antiretrovirals is crucial, Figure 1. Overview of fragment hit locations in newly identified sites by Bauman et al. [7]. Legend: HIV-1 RT p66 subunit with fingers (blue), palm (red), thumb (green), connection (yellow) and RNase H (orange); p51 shown in gray; rilpivirine bound to the NNRTI-binding pocket (NNIBP) is shown as yellow spheres; bound fragments are shown as cyan spheres. Each binding site is circled and color coded according to its name (i.e. purple corresponds to Knuckles). Site, compound, and potency information in Table S1. Created with PyMOL and BioRender.com.
In particular, the NNRTI Adjacent site is a potentially attractive target for the design of novel inhibitors targeting both this site and the NNRTI-binding pocket (NNIBP) since the key residues of the pocket are conserved [7]. Accordingly, the goal of this work was to follow a systematic fragment development process to elaborate the promising hit compound B-1 in an effort to target the NNRTI Adjacent site.
The Bauman et al. study reported compound B-1 to have an IC50 of 350 µM and a ligand efficiency (LE) of 0.34 kcal/mol [7] (Table S1). LE is an expression of binding energy accounting for a compound's size, whereby 0.30 kcal/mol or greater is considered efficient (further defined below). The fragment screen showed that this compound binds to the NNRTI Adjacent site, a region neighboring the NNIBP (Figure 1), where it makes a hydrogen bond with the backbone carbonyl oxygen of isoleucine (Ile180) and hydrophobic interactions with proline and isoleucine residues (Pro140 and Ile180). The side chains of two glutamine residues (Gln161 and Gln182) reposition to accommodate the fragment, allowing the formation of a hydrogen-bonding interaction with the glutamine (Gln182) backbone amide which was proposed to be responsible for RT inhibition by this fragment (see Figure 2b in [7]).
In this study, we aimed to optimize the binding and HIV-1 RT inhibitory activity of compound B-1 in pursuit of a novel RT drug candidate which overcomes drug resistance and minimizes side effects associated with current antiretroviral therapies. Three series of compounds were designed, synthesized and evaluated for their HIV-1 RT binding and inhibition. A derivative 27 with an alkyne extension at the 4-position yielded an IC50 value of 11 µM against wild-type (WT) RT and 34 µM against the K103N/Y181C mutant RT-a variant that is resistant to many NNRTIs [8].

Compound Synthesis
Beginning with compound B-1 as a core structure, we designed and synthesized three series of compounds ( Figure 2) to identify positions that could be substituted to  [7]. Legend: HIV-1 RT p66 subunit with fingers (blue), palm (red), thumb (green), connection (yellow) and RNase H (orange); p51 shown in gray; rilpivirine bound to the NNRTI-binding pocket (NNIBP) is shown as yellow spheres; bound fragments are shown as cyan spheres. Each binding site is circled and color coded according to its name (i.e., purple corresponds to Knuckles). Site, compound, and potency information in Table S1. Created with PyMOL and BioRender.com.
In this study, we aimed to optimize the binding and HIV-1 RT inhibitory activity of compound B-1 in pursuit of a novel RT drug candidate which overcomes drug resistance and minimizes side effects associated with current antiretroviral therapies. Three series of compounds were designed, synthesized and evaluated for their HIV-1 RT binding and inhibition. A derivative 27 with an alkyne extension at the 4-position yielded an IC 50 value of 11 µM against wild-type (WT) RT and 34 µM against the K103N/Y181C mutant RT-a variant that is resistant to many NNRTIs [8].

Compound Synthesis
Beginning with compound B-1 as a core structure, we designed and synthesized three series of compounds ( Figure 2) to identify positions that could be substituted to improve affinity for HIV-1 RT. Series 1 investigated substitution at positions 4-7 around the pyrazolopyridine ring. Series 2 modified the ethyl ester. Substitution at the 4-position was found to better inhibit RT (discussed in the following inhibition and binding section), which led to the synthesis of a set of 4-alkynes (Series 3).  Pyrazolopyridines with substitution on the pyridine ring (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16) were prepared using the method by Kendall et al. [9,10], by N-amination of the substituted pyridines followed by 1,3-dipolar cycloaddition (Scheme 1). N-Amination of the pyridine starting materials was performed with commercially available O-(2,4-dinitrophenyl)hydroxylamine Pyrazolopyridines with substitution on the pyridine ring (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16) were prepared using the method by Kendall et al. [9,10], by N-amination of the substituted pyridines followed by 1,3-dipolar cycloaddition (Scheme 1). N-Amination of the pyridine starting materials was performed with commercially available O- (2,4-dinitrophenyl)hydroxylamine (DNPH) as it is more stable, easier to handle and more efficient than other reagents [11]. Acetonitrile was initially used as the solvent for this step following Kendall's method. A method described by Legault and Charette using THF and water mixture (1:1) gave improved yields and faster reaction times [11]. Other studies speculate that the presence of water as co-solvent increases reaction speed by stabilizing the transition state [12]. This led to the inclusion of water in this reaction where maximum conversion was observed in MeCN:H 2 O (1:1) at a temperature of 60 • C. The N-aminopyridines were used without any purification and elaborated to pyrazolopyridines by 1,3-dipolar cycloaddition reaction with ethyl propiolate. The pyrazolopyridines were obtained in moderate yields of 15-55%, depending on the reactivity of the substituted pyridine starting materials. It was found that pyridines with electron-withdrawing substituents were not reactive enough to undergo the N-amination reaction. Notably, two regioisomers were formed when 3-substituted pyridines were used as the starting material, providing mixtures of 4-and 6-substituted pyrazolo [1,5-a]pyridines. In these cases, the two regioisomers were separated by column chromatography and distinguished using 1 H NMR. Pyrazolopyridines with substitution on the pyridine ring (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16) were prepared using the method by Kendall et al. [9,10], by N-amination of the substituted pyridines followed by 1,3-dipolar cycloaddition (Scheme 1). N-Amination of the pyridine starting materials was performed with commercially available O- (2,4-dinitrophenyl)hydroxylamine (DNPH) as it is more stable, easier to handle and more efficient than other reagents [11]. Acetonitrile was initially used as the solvent for this step following Kendall's method. A method described by Legault and Charette using THF and water mixture (1:1) gave improved yields and faster reaction times [11]. Other studies speculate that the presence of water as co-solvent increases reaction speed by stabilizing the transition state [12]. This led to the inclusion of water in this reaction where maximum conversion was observed in MeCN:H2O (1:1) at a temperature of 60 °C. The N-aminopyridines were used without any purification and elaborated to pyrazolopyridines by 1,3-dipolar cycloaddition reaction with ethyl propiolate. The pyrazolopyridines were obtained in moderate yields of 15-55%, depending on the reactivity of the substituted pyridine starting materials. It was found that pyridines with electron-withdrawing substituents were not reactive enough to undergo the N-amination reaction. Notably, two regioisomers were formed when 3-substituted pyridines were used as the starting material, providing mixtures of 4-and 6-substituted pyrazolo [1,5-a]pyridines. In these cases, the two regioisomers were separated by column chromatography and distinguished using 1 H NMR. A selection of amides and esters were synthesized to investigate the importance of the ethyl ester present in compound B-1 (Scheme 2). Acid 17 was prepared by hydrolysis of ester 6. Methyl ester 18 was prepared by transesterification of ethyl ester 6 using dipotassium phosphate and methanol. Compound 19 was made by esterification of acid 17 using sulfuric acid and propanol. Amides 20-22 were synthesized from acid 17 using standard amide coupling conditions. A selection of amides and esters were synthesized to investigate the importance of the ethyl ester present in compound B-1 (Scheme 2). Acid 17 was prepared by hydrolysis of ester 6. Methyl ester 18 was prepared by transesterification of ethyl ester 6 using dipotassium phosphate and methanol. Compound 19 was made by esterification of acid 17 using sulfuric acid and propanol. Amides 20-22 were synthesized from acid 17 using standard amide coupling conditions. Finally, a series of 4-alkynyl analogues (23)(24)(25)(26)(27)(28)(29) was prepared using a modification of the method by Schilz [13], using tetrakis(triphenylphosphine)palladium(0) as the catalyst, copper iodide as the co-catalyst and diisopropylamine as the base and solvent (Scheme 3). Couplings were performed at 80 °C with complete conversion observed overnight. Finally, a series of 4-alkynyl analogues (23)(24)(25)(26)(27)(28)(29) was prepared using a modification of the method by Schilz [13], using tetrakis(triphenylphosphine)palladium(0) as the catalyst, copper iodide as the co-catalyst and diisopropylamine as the base and solvent (Scheme 3). Couplings were performed at 80 • C with complete conversion observed overnight. °C, overnight; (ii) LiOH. H2O, THF:MeOH:H2O (4:1:1), rt, 5 h. (iii) H2SO4, propanol, 85 °C, overnight; (iv) RNH2, HBTU, DIPEA, DMF, rt, 4 h.

Inhibition and Binding Affinity of Pyrazolo[1,5-a]pyridine Analogues
The inhibitory activity of all compounds was evaluated in vitro in assays that measure the DNA-dependent DNA polymerase (DDDP) activity of HIV-1 RT. Compounds were tested against recombinant WT and drug-resistant HIV-1 RT mutants (K103N, Y181C or K103N/Y181C). Inhibition of WT RT DDDP activity was initially evaluated by testing three concentrations of compound in two independent assays using nevirapine (an FDA approved NNRTI) as the positive control [3]. Compounds that demonstrated WT RT inhibitory activity at concentrations less than 1 mM were then subjected to a second assay to calculate the compound IC50 value from dose-response curves generated from at least five serial dilutions in at least two independent assays [14]. The IC50 value of nevirapine was determined in parallel to control for the performance of the assay. As part of the evaluation cascade, compounds with IC50 values lower than 200 µM against WT RT were also tested against NNRTI-resistant recombinant RT mutants, K103N, Y181C or K103N/Y181C. These RT mutations either reduce the rate of entry of NNRTIs to the binding pocket (K103N) or remove aromatic interactions that favor binding (Y181C) [3]. Initially, compound IC50 values were determined in an assay that quantifies DDDP activity mediated by HIV-1 RT using a radiolabeled 33 PdTTP substrate [14] and more recently using a nonradioactive PicoGreen-based spectrophotometric assay [15]. Accordingly, we first confirmed that the nevirapine IC50 value for inhibiting recombinant HIV-1 RT obtained in the non-radioactive assay was similar to the assay using 33 PdTTP. The nevirapine IC50 ± SEM for inhibiting WT RT using the nonradioactive DDDP assay was 0.44 ± 0.03 µM (n = 3), which was not significantly different compared to the IC50 value reported using the 33 PdTTP assay (0.44 ± 0.08 µM, p = 0.7, n = 3, Mann-Whitney test) [14], providing confidence that IC50 data generated by both assays are comparable. Inhibitory activities of Series 1 and 3 are summarized in Tables 1 and 2. The activities of Series 2, which was largely inactive, are shown in the supporting information (Table S2).
Compound binding to recombinant HIV-1 RT was measured by surface plasmon resonance (SPR). SPR is an optical biosensor detection method that involves passing an analyte (often a fragment) in a buffer solution over a gold surface chip with an immobilized protein (RT enzyme) and detecting the change in plasmon resonance angle which is dependent on the refractive index of the medium near the surface [16,17]. As compounds

Inhibition and Binding Affinity of Pyrazolo[1,5-a]pyridine Analogues
The inhibitory activity of all compounds was evaluated in vitro in assays that measure the DNA-dependent DNA polymerase (DDDP) activity of HIV-1 RT. Compounds were tested against recombinant WT and drug-resistant HIV-1 RT mutants (K103N, Y181C or K103N/Y181C). Inhibition of WT RT DDDP activity was initially evaluated by testing three concentrations of compound in two independent assays using nevirapine (an FDA approved NNRTI) as the positive control [3]. Compounds that demonstrated WT RT inhibitory activity at concentrations less than 1 mM were then subjected to a second assay to calculate the compound IC 50 value from dose-response curves generated from at least five serial dilutions in at least two independent assays [14]. The IC 50 value of nevirapine was determined in parallel to control for the performance of the assay. As part of the evaluation cascade, compounds with IC 50 values lower than 200 µM against WT RT were also tested against NNRTI-resistant recombinant RT mutants, K103N, Y181C or K103N/Y181C. These RT mutations either reduce the rate of entry of NNRTIs to the binding pocket (K103N) or remove aromatic interactions that favor binding (Y181C) [3]. Initially, compound IC 50 values were determined in an assay that quantifies DDDP activity mediated by HIV-1 RT using a radiolabeled 33 PdTTP substrate [14] and more recently using a nonradioactive PicoGreen-based spectrophotometric assay [15]. Accordingly, we first confirmed that the nevirapine IC 50 value for inhibiting recombinant HIV-1 RT obtained in the non-radioactive assay was similar to the assay using 33 PdTTP. The nevirapine IC 50 ± SEM for inhibiting WT RT using the nonradioactive DDDP assay was 0.44 ± 0.03 µM (n = 3), which was not significantly different compared to the IC 50 value reported using the 33 PdTTP assay (0.44 ± 0.08 µM, p = 0.7, n = 3, Mann-Whitney test) [14], providing confidence that IC 50 data generated by both assays are comparable. Inhibitory activities of Series 1 and 3 are summarized in Tables 1 and 2. The activities of Series 2, which was largely inactive, are shown in the supporting information (Table S2).
Compound binding to recombinant HIV-1 RT was measured by surface plasmon resonance (SPR). SPR is an optical biosensor detection method that involves passing an analyte (often a fragment) in a buffer solution over a gold surface chip with an immobilized protein (RT enzyme) and detecting the change in plasmon resonance angle which is dependent on the refractive index of the medium near the surface [16,17]. As compounds bind to the target protein, the refractive index changes proportionally to the extent of binding. This can then be used to measure the association and dissociation kinetics of binding as well as the compound's affinity for binding [17][18][19][20]. SPR is a label-free method that does not require large amounts of protein and is often used for primary screening of fragments. Binding affinities of compounds are summarized in Tables 1, 2 and S2 with SPR curves shown for compounds 2, 6 and 27 in Figure S1. the equation ΔG = -RTlnKi [6]. Using the RT inhibition (IC50), LE can be calculated using the equation LE = -1.4(logIC50)/N [22]. Compounds with LE values greater than 0.3 kcal/mol/HA are generally regarded as efficient binders [23][24][25]. LE values for binding to WT RT are shown in Tables 1 and 2. Notably, nearly all of the active compounds bind efficiently (LE > 0.3 kcal/mol/HA), and compound 27 binds with LE = 0.37 kcal/mol/HA.
The 50% inhibitory concentration (IC50) values ± SEM (standard error of the mean) were determined by assessing inhibition of HIV-1 RT DNA-dependent DNA Polymerase (DDDP) activity using the nonradioactive PicoGreen or 33 P radiolabeled assay. IC50 values were determined from at least n ≥ 2 independent assays. b Dissociation constants (KD) ± standard error in the dose-response fit, as calculated by Biacore S200 Evaluation Software] (n = 1 independent assays) were measured using surface plasmon resonance (SPR). c Ligand efficiency (LE) is expressed as kcal/mol/HA where HA represents the number of non-hydrogen atoms in the compound. d RT DDDP inhibitory activity determined using the PicoGreen assay. NB denotes no binding observed.-denotes not determined.
The 50% inhibitory concentration (IC 50 ) values ± SEM (standard error of the mean) were determined by assessing inhibition of HIV-1 RT DNA-dependent DNA Polymerase (DDDP) activity using the nonradioactive PicoGreen or 33 P radiolabeled assay. IC 50 values were determined from at least n ≥ 2 independent assays. b Dissociation constants (K D ) ± standard error in the dose-response fit, as calculated by Biacore S200 Evaluation Software] (n = 1 independent assays) were measured using surface plasmon resonance (SPR). c Ligand efficiency (LE) is expressed as kcal/mol/HA where HA represents the number of non-hydrogen atoms in the compound. d RT DDDP inhibitory activity determined using the PicoGreen assay. NB denotes no binding observed.-denotes not determined.
Due to their small size, fragments bind weakly; however, they are capable of binding efficiently to the target protein on a per-atom basis. Binding efficiency can be quantified by the ligand efficiency (LE) metric [21]. LE is the free energy of binding of a ligand (∆G) divided by the number of non-hydrogen (heavy) atoms (N) where ∆G is calculated using the equation ∆G = −RTlnK i [6]. Using the RT inhibition (IC 50 ), LE can be calculated using the equation LE = −1.4(logIC 50 )/N [22]. Compounds with LE values greater than 0.3 kcal/mol/HA are generally regarded as efficient binders [23][24][25]. LE values for binding to WT RT are shown in Tables 1 and 2. Notably, nearly all of the active compounds bind efficiently (LE > 0.3 kcal/mol/HA), and compound 27 binds with LE = 0.37 kcal/mol/HA. Table 1 shows RT inhibition and SPR dissociation constants for Series 1. The hit compound B-1 for the NNRTI Adjacent site was found to have a lower potency when tested in the DNA-dependent DNA polymerase (DDDP) activity assay used in this study compared to data from Bauman et al. which used an assay that detects inhibition of both DNA polymerase and RNase H activity [6]. Substitution at some sites on the pyridine greatly reduced inhibition. However, incorporation of bromo or methyl substituents at the 4position improved inhibition of the WT enzyme, and this activity was somewhat maintained in the mutant strains where no significant differences were observed for the compound 6 IC 50 value for WT RT compared to the K103N (p = 0.1, n ≥ 2) or the Y181C (p = 0.4, n ≥ 2 Mann-Whitney test) RT mutants. Notably, compounds 6 and 9 have IC 50 values of about 150 µM, where the IC 50 values of 5 and 6 were significantly different (p = 0.037, n = 3, Kruskal-Wallis test). These data suggest that the 4-position of the pyrazolo[1,5-a]pyridine core is important for inhibition of HIV-1 RT. in the 4-position, are shown in Table 2. A number of these compounds have IC50 v under 50 µM against recombinant WT RT as well as the K103N/Y181C NNRTI dr sistant mutant. Notably the 4-cyclopropylalkyne 27 had an IC50 value of 11 µM for W and was amongst the most potent compounds compared to 28 (p = 0.035, Kruskaltest), as well as demonstrating similar activity for inhibiting the K103N/Y181C RT m (p = 0.2, n ≥ 2, Mann-Whitney test). Compound 27 also demonstrated the strongest bi to both WT and mutant RT as determined by SPR (Table 2). Accordingly, this comp was selected for crystallographic studies.   Series 2 compounds, which investigated modifications of the ethyl ester present in compound B-1, were nearly all inactive (Table S2) in the RT inhibition assay. The fact that small structural changes (e.g., changing the ethyl ester to an ethyl amide or propyl or methyl ester) in this region greatly reduced the IC 50 values suggests that the presence of the ethyl ester is essential for RT inhibition.
The activities of the compounds in Series 3, which all contain an alkyne substituent in the 4-position, are shown in Table 2. A number of these compounds have IC 50 values under 50 µM against recombinant WT RT as well as the K103N/Y181C NNRTI drug-resistant mutant. Notably the 4-cyclopropylalkyne 27 had an IC 50 value of 11 µM for WT RT and was amongst the most potent compounds compared to 28 (p = 0.035, Kruskal-Wallis test), as well as demonstrating similar activity for inhibiting the K103N/Y181C RT mutant (p = 0.2, n ≥ 2, Mann-Whitney test). Compound 27 also demonstrated the strongest binding to both WT and mutant RT as determined by SPR (Table 2). Accordingly, this compound was selected for crystallographic studies.

Crystallography
To understand the structural basis for the improved activity of compound 27, we determined a co-crystal structure of 27 in complex with RT52A. Initially, we attempted to co-crystallize compound 27 (anticipating its binding to the NNRTI Adjacent site) with the NNRTI rilpivirine. However, compound 27 was not observed in these experiments. Accordingly, we next solved the structure of compound 27 complexed with RT52A in the absence of an NNRTI. The structure was solved by molecular replacement using a highresolution RT-RPV complex [26], (PDB ID 4G1Q) as the search model, and was subsequently refined to 2.42 Å resolution (Table S3). Unexpectedly, the structure showed that 27 does not bind to the NNRTI Adjacent site [7] but instead occupies the NNRTI-binding pocket (NNIBP) of RT ( Figure 3). A Polder OMIT map (green mesh, Figure 3), calculated by excluding the bulk solvent around the omitted region, clearly showed the difference in electron density for 27 in the NNIBP. The cyclopropyl group of 27 is directed into the hydrophobic tunnel of the NNIBP, where it makes interactions with Y181, Y188 and W229. The N1 nitrogen of the pyrazolo[1,5-a]pyridine ring forms a hydrogen-bonding interaction with a water molecule mediating interaction with K103 ( Figure S3). The terminal methyl of the ethyl ester group forms an additional hydrophobic interaction with V179. The much greater affinity of RPV for the NNIBP compared to 27 (0.73 nM [27] vs. 11 µM) helps explain why RPV and not 27 was observed in co-crystallization experiments of RT with 27 and RPV. cordingly, we next solved the structure of compound 27 complexed with RT5 absence of an NNRTI. The structure was solved by molecular replacement usi resolution RT-RPV complex [26], (PDB ID 4G1Q) as the search model, and w quently refined to 2.42 Å resolution (Table S3). Unexpectedly, the structure sh 27 does not bind to the NNRTI Adjacent site [7] but instead occupies the NNRT pocket (NNIBP) of RT ( Figure 3). A Polder OMIT map (green mesh, Figure 3), by excluding the bulk solvent around the omitted region, clearly showed the di electron density for 27 in the NNIBP. The cyclopropyl group of 27 is directe hydrophobic tunnel of the NNIBP, where it makes interactions with Y181, Y188 The N1 nitrogen of the pyrazolo[1,5-a]pyridine ring forms a hydrogen-bondin tion with a water molecule mediating interaction with K103 ( Figure S3). The ter thyl of the ethyl ester group forms an additional hydrophobic interaction with much greater affinity of RPV for the NNIBP compared to 27 (0.73 nM [27] vs. 11 explain why RPV and not 27 was observed in co-crystallization experiments o 27 and RPV.   (Figure 4). Moreover, a favorable water bridge interaction is observed between the carbonyl of 27 and K103 ( Figure S3). Effectively, the NNIBP acts more as a "sink" likely providing a more favorable binding site for 27, though not as favorable as for RPV as noted above. Effectively, the NNIBP acts more as a "sink" likely providing a more favorable binding site for 27, though not as favorable as for RPV as noted above.

Combination Assay
Compound 27 demonstrated a nonsignificant 3.0-fold decrease in susceptibility to the K103N/Y181C NNRTI-resistant mutant compared to WT RT (Table 2). Re-testing of compound B-1 in the non-radioactive PicoGreen RT assay revealed IC50 values of 96 µM and 208 µM against WT RT and the K103N/Y181C NNRTI drug-resistant RT mutant (p = 0.1, n = 3, Mann-Whitney test), respectively ( Figure S2). Thus, compound B-1 shows a nonsignificant decrease in susceptibility to the NNRTI drug-resistant mutant compared to WT RT. The PicoGreen RT assay uses a different DNA template compared to the radioactive assay, where the latter employs activated calf thymus DNA which may account for the difference observed for RT inhibitory activity of compound B-1 in the two assays. Taking together the X-ray crystallographic data and their ability to retain inhibitory activity against the NNRTI-resistant mutants, we hypothesized that compounds B-1 and 27 bind to more than one site on the HIV-1 RT.
To investigate this further, we determined whether compound B-1 and 27 in combination with the NNRTI, nevirapine, demonstrate additive, antagonistic or synergistic inhibition of HIV-1 RT. If compounds bind solely to the NNIBP, we would expect to see additive inhibition of RT activity in combination with nevirapine [8]. Combination assays were performed using the non-radioactive DDDP assay with drug combinations tested at fixed ratios. As a control for additivity, we tested the NNRTIs, nevirapine and doravirine (DOR) in combination [8]. While the combination of NVP and DOR was additive, the combinations of NVP and B-1 and NVP and 27 displayed moderate antagonism for inhibiting HIV-1 RT DDDP activity (Table 3). These data indicate that B-1 and 27 do not behave similarly to classical NNRTIs and that they may also bind to sites that are distinct from the NNRTI-binding pocket. Since these compounds displayed moderate antagonism, it is likely that binding to one site may impede the ability to bind to the second site. Given the

Combination Assay
Compound 27 demonstrated a nonsignificant 3.0-fold decrease in susceptibility to the K103N/Y181C NNRTI-resistant mutant compared to WT RT (Table 2). Re-testing of compound B-1 in the non-radioactive PicoGreen RT assay revealed IC 50 values of 96 µM and 208 µM against WT RT and the K103N/Y181C NNRTI drug-resistant RT mutant (p = 0.1, n = 3, Mann-Whitney test), respectively ( Figure S2). Thus, compound B-1 shows a nonsignificant decrease in susceptibility to the NNRTI drug-resistant mutant compared to WT RT. The PicoGreen RT assay uses a different DNA template compared to the radioactive assay, where the latter employs activated calf thymus DNA which may account for the difference observed for RT inhibitory activity of compound B-1 in the two assays. Taking together the X-ray crystallographic data and their ability to retain inhibitory activity against the NNRTI-resistant mutants, we hypothesized that compounds B-1 and 27 bind to more than one site on the HIV-1 RT.
To investigate this further, we determined whether compound B-1 and 27 in combination with the NNRTI, nevirapine, demonstrate additive, antagonistic or synergistic inhibition of HIV-1 RT. If compounds bind solely to the NNIBP, we would expect to see additive inhibition of RT activity in combination with nevirapine [8]. Combination assays were performed using the non-radioactive DDDP assay with drug combinations tested at fixed ratios. As a control for additivity, we tested the NNRTIs, nevirapine and doravirine (DOR) in combination [8]. While the combination of NVP and DOR was additive, the combinations of NVP and B-1 and NVP and 27 displayed moderate antagonism for inhibiting HIV-1 RT DDDP activity (Table 3). These data indicate that B-1 and 27 do not behave similarly to classical NNRTIs and that they may also bind to sites that are distinct from the NNRTI-binding pocket. Since these compounds displayed moderate antagonism, it is likely that binding to one site may impede the ability to bind to the second site. Given the proximity of the NNIBP to the NNRTI Adjacent site, it is possible that compounds are binding to both sites; however, this would need to be independently verified in a future work.

SAR
The initial fragment hit reported by Bauman et al., (ethyl pyrazolo[1,5-a]pyridine-3carboxylate, B-1) [7], was found in their RT activity assay to have an IC 50 value of 350 µM for the WT RT construct RT35A with a ligand efficiency of 0.34 kcal/mol/HA. In the current study, we have used a range of synthetic techniques to elaborate compound B-1. Three focused fragment libraries were designed, synthesized and characterized. Series 1 (Table 1) explored the effects of different substitutions around the pyridine ring. Compounds 2, 5, 6, 9 and 10 showed markedly improved binding and/or inhibitory activities, highlighting the importance of the 4-position. Series 2 (Table S2) investigated changes to the ethyl ester group. In this case, all changes to the ester (change in alkyl chain size, conversion to an amide) abolished inhibition and binding activities. This suggests that the ester group makes close contact and strong interactions with its binding site. Following on from the improved activities of 4-substituted compounds, Series 3 was designed. The 4-bromo substituted compound, 6, was used as a starting material to prepare a series of alkynes through Sonogashira couplings ( Table 2). Compounds 26, 27, 28 and 29 from this series demonstrated further improvements in activity, all having IC 50 values less than 100 µM, which is a >20-fold improvement on parent compound B-1, which had an IC 50 >1000 µM in the radioactive RT assay. Notably, compound 27 was among the most potent in both the RT inhibition and the SPR assay with an IC 50 of 11 ± 1 µM and 34 ± 8 µM (p = 0.2) and K D of 79 ± 7 µM and 187 ± 15 µM against WT HIV-1 RT and the K103N/Y181C double mutant, respectively. The 2-3-fold decrease in binding and non-significant change in RT inhibition mediated by 27 is encouraging, considering the small size of the compound. Compound 27 binds efficiently to WT RT with an LE of 0.40 kcal/mol/HA. To establish that the fragment series represented unique compounds, a Tanimoto similarity search was conducted against 1524 HIV-1 RT inhibitors recorded in the Binding Database (www.bindingdb.org, accessed on 9 November 2021). Of the 1524 RT inhibitors, no compound had a Tanimoto similarity score greater than 0.63. Compound 27 was selected as the lead compound to be further investigated, with crystallography and combination studies.

Structural Analysis
The X-ray structure of Bauman et al. demonstrated the ability of compound B-1 to bind in the NNRTI Adjacent site when the drug RPV is present in the NNIBP, while displaying significant inhibition of RT polymerase activity in the absence of RPV (Table S1) [7]. However, contrary to our initial expectations, the structure of the elaborated compound 27 in complex with RT52A (in the absence of RPV) showed that the compound bound in the NNIBP (Figure 3). We surmise that, with the exception of RPV or NNRTIs with central moieties extending towards the solvent, e.g., K07-15, PDB ID 7KWU [28], NNRTIs do not induce opening of the NNRTI Adjacent site, which stays occluded and prevents binding of ligands. Of note, mining of the literature revealed that the similar shaped fragment, bromoindanone, was found as a putative NNRTI hit in an SPR fragment screening campaign targeting HIV-1 RT [29].
To understand the nature of the NNRTI Adjacent site and whether it is open to compound binding in other structures, we compared the site when either FDA approved NNRTIs or other inhibitors were bound to the NNIBP of RT. Figure 5 shows the NNIBP (left) and the NNRTI Adjacent site (right) in various inhibitor-bound structures. Each inhibitor and the associated PDB are indicated underneath each panel. Compound B-1 is superimposed into the NNRTI Adjacent site in each structure, except for PDB ID 4KFB, which captures a bound state of RT52A/RPV/1. [7]. However, contrary to our initial expectations, the structure of the elaborated compound 27 in complex with RT52A (in the absence of RPV) showed that the compound bound in the NNIBP (Figure 3). We surmise that, with the exception of RPV or NNRTIs with central moieties extending towards the solvent, e.g., K07-15, PDB ID 7KWU [28], NNRTIs do not induce opening of the NNRTI Adjacent site, which stays occluded and prevents binding of ligands. Of note, mining of the literature revealed that the similar shaped fragment, bromoindanone, was found as a putative NNRTI hit in an SPR fragment screening campaign targeting HIV-1 RT [29].
To understand the nature of the NNRTI Adjacent site and whether it is open to compound binding in other structures, we compared the site when either FDA approved NNRTIs or other inhibitors were bound to the NNIBP of RT. Figure 5 shows the NNIBP (left) and the NNRTI Adjacent site (right) in various inhibitor-bound structures. Each inhibitor and the associated PDB are indicated underneath each panel. Compound B-1 is superimposed into the NNRTI Adjacent site in each structure, except for PDB ID 4KFB, which captures a bound state of RT52A/RPV/1. The surface representations help to visualize that the NNRTI Adjacent site remains closed in most of the NNRTI-bound structures and unavailable for compounds to bind. In contrast, inhibitors K07-15, 25a and K-5a2 ( Figure S5) revealed an NNRTI Adjacent site very similar to that observed in the RPV-bound structure ( Figure 5), in which the NNRTI Adjacent site is open. To further investigate this observation, a pocket analysis was performed using the PyVOL extension in PyMOL (Table 4) [30].  The surface representations help to visualize that the NNRTI Adjacent site remains closed in most of the NNRTI-bound structures and unavailable for compounds to bind. In contrast, inhibitors K07-15, 25a and K-5a2 ( Figure S5) revealed an NNRTI Adjacent site very similar to that observed in the RPV-bound structure ( Figure 5), in which the NNRTI Adjacent site is open. To further investigate this observation, a pocket analysis was performed using the PyVOL extension in PyMOL (Table 4) [30]. The pocket volume analysis confirmed the surface visual analysis. Compared to RPV, K07-15, 25a or K-5a2, other approved NNRTIs complexed with HIV-1 RT have smaller NNIBP volumes and no measurable pocket volume for the NNRTI Adjacent site. Additional analysis shows that the NNIBP and the Adjacent site pocket volumes correlate with the position of the thumb in RT, which is locked into a hyperextended orientation by NNRTIs and not conducive to reverse transcription. In general, structures of RT/NNRTI with smaller NNRTI pocket volumes have lesser hyperextension in the thumb region (e.g., nevirapine, NVP; etravirine, ETR; efavirenz, EFV). However, RPV and K07-15, 25a and K-5a2, the only compounds with measurable volume for the NNRTI Adjacent site and NNIBP pocket volumes larger than 500 Å 3 , are accompanied by greater thumb hyperextension ( Figure 6). This is consistent with further stabilization observed through NMR spectroscopy of the open conformation of RT (compared to closed in apo RT) when bound to diarylpyrimidine-based inhibitors (DAPYs), such as RPV, over EFV or NVP [31]. The pocket volume analysis confirmed the surface visual analysis. Compared to RPV, K07-15, 25a or K-5a2, other approved NNRTIs complexed with HIV-1 RT have smaller NNIBP volumes and no measurable pocket volume for the NNRTI Adjacent site. Additional analysis shows that the NNIBP and the Adjacent site pocket volumes correlate with the position of the thumb in RT, which is locked into a hyperextended orientation by NNRTIs and not conducive to reverse transcription. In general, structures of RT/NNRTI with smaller NNRTI pocket volumes have lesser hyperextension in the thumb region (e.g., nevirapine, NVP; etravirine, ETR; efavirenz, EFV). However, RPV and K07-15, 25a and K-5a2, the only compounds with measurable volume for the NNRTI Adjacent site and NNIBP pocket volumes larger than 500 Å 3 , are accompanied by greater thumb hyperextension ( Figure 6). This is consistent with further stabilization observed through NMR spectroscopy of the open conformation of RT (compared to closed in apo RT) when bound to diarylpyrimidine-based inhibitors (DAPYs), such as RPV, over EFV or NVP [31]. From these analyses, we conclude that there are two important structural features that may be relevant to design effective dual-targeting inhibitors binding both the NNIBP From these analyses, we conclude that there are two important structural features that may be relevant to design effective dual-targeting inhibitors binding both the NNIBP and NNRTI Adjacent site. First, an NNRTI must possess wing substituents that open the tunnel and groove channels deep enough to expand the NNIBP sufficiently to facilitate the opening of the NNRTI Adjacent pocket in the vicinity of the NNIBP solvent-exposed entrance region. Second, an aromatic extension at the 5-position of the pyrimidine ring in DAPY compounds may further open the Adjacent site and provide a bridge between the NNIBP and Adjacent site for dual-targeting inhibitor design [28,32,33]. While compound 27 in complex with RT described in this study is a relatively weak inhibitor of RT, an overlay with K07-15 suggests the pyrazolo[1,5-a]pyridine ring may serve as a minimal scaffold to elaborate in designing novel dual-targeting inhibitors merged with the 5-position extension of K07-15 (Figure 7). and NNRTI Adjacent site. First, an NNRTI must possess wing substituents that open the tunnel and groove channels deep enough to expand the NNIBP sufficiently to facilitate the opening of the NNRTI Adjacent pocket in the vicinity of the NNIBP solvent-exposed entrance region. Second, an aromatic extension at the 5-position of the pyrimidine ring in DAPY compounds may further open the Adjacent site and provide a bridge between the NNIBP and Adjacent site for dual-targeting inhibitor design [28,32,33]. While compound 27 in complex with RT described in this study is a relatively weak inhibitor of RT, an overlay with K07-15 suggests the pyrazolo[1,5-a]pyridine ring may serve as a minimal scaffold to elaborate in designing novel dual-targeting inhibitors merged with the 5-position extension of K07-15 (Figure 7). In summary, we determined a crystal structure of 27 complexed with RT after elaborating it from compound B-1, which was previously demonstrated to target the NNRTI-Adjacent site [7]. The NNRTI Adjacent site has previously been identified as a target for enhanced NNRTI inhibition, though all leads developed had molecular weight ≥ 500 Da and borderline bioavailability properties [32,34]. The elaborated 27 acts as an NNRTI with modest inhibitory activity but retains a high ligand efficiency (0.37 kcal/mol/HA) and displays favorable bioavailability properties according to SWISSADME analysis ( Figure S4) [35]. Compound 27 binding in the NNIBP is not accompanied by opening of the NNRTI Adjacent site, but crystallographic overlay with K07-15 shows a path for designing 27 derivatives to expand into the Adjacent site as a dual-targeting inhibitor with favorable pharmacokinetic properties. In this regard, the Arnold group has found additional NNRTI Adjacent fragment binders in a subsequent crystallographic fragment screening campaign that may aid in future dual-targeting inhibitor design [36].

Conclusions
In this work, using a fragment-based approach, we have developed a novel NNRTI minimal chemotype with high ligand efficiency. While the initial hit compound B-1 was In summary, we determined a crystal structure of 27 complexed with RT after elaborating it from compound B-1, which was previously demonstrated to target the NNRTI-Adjacent site [7]. The NNRTI Adjacent site has previously been identified as a target for enhanced NNRTI inhibition, though all leads developed had molecular weight ≥ 500 Da and borderline bioavailability properties [32,34]. The elaborated 27 acts as an NNRTI with modest inhibitory activity but retains a high ligand efficiency (0.37 kcal/mol/HA) and displays favorable bioavailability properties according to SWISSADME analysis ( Figure S4) [35]. Compound 27 binding in the NNIBP is not accompanied by opening of the NNRTI Adjacent site, but crystallographic overlay with K07-15 shows a path for designing 27 derivatives to expand into the Adjacent site as a dual-targeting inhibitor with favorable pharmacokinetic properties. In this regard, the Arnold group has found additional NNRTI Adjacent fragment binders in a subsequent crystallographic fragment screening campaign that may aid in future dual-targeting inhibitor design [36].

Conclusions
In this work, using a fragment-based approach, we have developed a novel NNRTI minimal chemotype with high ligand efficiency. While the initial hit compound B-1 was bound in the NNRTI Adjacent site, crystallography has revealed that binding of lead compound 27 occurs in the NNIBP. Combination experiments suggest that compounds B-1 and 27 could also be binding to additional sites apart from the NNBIP that likely involve the NNRTI Adjacent site. Related to the latter, inhibitors targeting both the NNIBP and the NNRTI Adjacent site are appealing, but so far DAPY-based leads have a very large molecular weight and poor bioavailability. Structural analysis shows that compound 27-a minimal but efficient NNRTI-offers an attractive possibility to develop a new drug class of dual NNIBP-Adjacent site inhibitors, with potential to overcome drug resistance and side effects associated with current therapeutics.

General Chemistry Methods
Starting materials and reagents were purchased from commercial suppliers (Sigma-Aldrich, Merck, BDH laboratories, Ajax Finechem, ChemSupply, Matrix Scientific, Alfa Aesar and Chem-Impex) and were used without further purification unless otherwise stated. Anhydrous solvents were obtained from MBraun SPS-800 Solvent purification system. TLC plates were visualized under UV illumination at 254 nm. Column chromatography was achieved using Davisil silica gel-LC60A (40-63 microns).
The 1 H NMR, 13 C NMR and 19 F NMR spectra were recorded on a Bruker Advance Nanobay III 400 MHz Ultrashield Plus spectrometer at 400.13, 100.61 and 376.50 MHz, respectively. Chemical shifts (δ) are reported in parts per million (ppm), and multiplicities are reported as follows: singlet (s), doublet (d), triplet (t), quartet (q), sextet (sex), doublet of doublets (dd) and multiplet (m). All samples were dissolved in chloroform-d (CDCl 3 ), and the residual solvent signals were used as the internal reference. The 13 C NMR assignment of carbon environments based on APT phasing is as follows: C = quaternary carbon, CH = methine carbon, CH 2 = methylene carbon and CH 3 = methyl carbon.
Liquid chromatography-mass spectrometry (LC-MS) was performed using either system A or B. System A: an Agilent 6100 Series Single Quad coupled to an Agilent 1200 Series high-performance liquid chromatography (HPLC) system using a Phenomenex Luna C8 (2) 50 × 4.6 mm 2 , 5 µm column. The following buffers were used: buffer A: 0.1% formic acid in H 2 O; buffer B: 0.1% formic acid in MeCN. Samples were run at a flow rate of 0.5 mL/min for 10 min: 0-4 min 5-100% buffer B in buffer A, 4-7 min 100% buffer B, 7-9 min 100-5% buffer B in buffer A and 9-10 min 5% buffer B in buffer A. Mass spectra were acquired in positive-and negative-ion modes with a scan range of 100-1000 m/z. UV detection was carried out at 254 nm. System B: an Agilent 6120 Series Single Quad coupled to an Agilent 1260 Series HPLC using a Poroshell 120 EC-C18 50 × 3.0 mm 2 , 2.7 µm column. The following buffers were used: buffer A: 0.1% formic acid in H 2 O; buffer B: 0.1% formic acid in MeCN. Samples were run at a flow rate of 0.5 mL/min for 5 min: 0-1 min 5% buffer B in buffer A, 1-2.5 min 5-100% buffer B in buffer A, 2.5-3.8 min 100% buffer B, 3.8-4 min 100-5% buffer B in buffer A and 4-5 min 5% buffer B in buffer A. Mass spectra were acquired in positive-and negative-ion modes with a scan range of 100-1000 m/z. UV detection was carried out at 214 and 254 nm.
High resolution mass spectrometry was performed on an Agilent 6224 TOP LC/MS coupled to an Agilent 1290 Infinity. All data were acquired, and reference mass was corrected by a dual-spray electrospray ionization (ESI) source where the chromatographic separation was performed using an Agilent Zorbax SB-C18 Rapid Resolution HT 2.1 × 50 mm, 1.8 µm column using an acetonitrile gradient (5% to 100%) over 3.5 min at 0.5 mL/min. Analytical high-performance liquid chromatography (HPLC) was carried out on Agilent 1260 Infinity Analytical HPLC with a Zorbax Eclipse Plus C18 rapid resolution 4.6 × 100 mm 3.5-Micron column. PP gradient was performed using a flow rate of 1 mL/min and a gradient of 5-100% B over 9 min followed by 100% B over 1 min. LC/MSD Chemstation Rev.B.04.03 coupled with Mass Hunter Easy Access Software managed the running and processing of samples. Preparative reverse-phase HPLC was carried out using a Waters Associates liquid chromatography system (Model 600 Controller and Waters 486 Tunable Absorbance Detector) with a Phenomenex Luna C8(2) 100 Å, 10 µm, 250 × 21.2 mm column with UV detection at 254 nm. A routine slow hydrophobic run was performed using a flow rate of 10 mL/min and a gradient of 20-100% B over 35 min (solvent A: 0.1% TFA in water, solvent B: 0.1% TFA in acetonitrile). To a solution of substituted pyridine (1.6 mmol) in DCM (10 mL) was added o-(2,4dinitrophenyl)hydroxylamine (300 mg, 1.5 mmol) at 0 • C then stirred at room temperature for 2 h. The crude material was concentrated and taken up in DMF (5 mL). Ethyl propiolate (168 µL, 1.7 mmol) and potassium carbonate (260 mg, 1.9 mmol) were added and stirred at room temperature overnight. The reaction mixture was diluted with water (100 mL), extracted with ethyl acetate (2 × 100 mL), the combined organic layers were washed with water (3 × 100 mL), brine (100 mL), dried over magnesium sulfate, filtered and concentrated in vacuo to give a gel-like brown crude product. This was purified using silica gel chromatography (hexanes:ethyl acetate, 85:15) to give the desired products.

Ethyl 6-Bromopyrazolo[1,5-a]pyridine-3-carboxylate
Diisopropylamine (~2 mL) and substituted alkyne (1.5-2 eq.) were added to the reaction mix, degassed and then stirred under nitrogen for 20 h at 80 • C. Crude material was taken up into ethyl acetate (20 mL), filtered through Celite™ and washed with water, and the aqueous layer was washed with ethyl acetate. The combined organic layers were washed with water (3 × 30 mL) and brine (20 mL), dried over magnesium sulphate, filtered and concentrated in vacuo to give crude product. This was purified using silica gel chromatography (hexanes:ethyl acetate, 9:1) to give the desired product. Data Availability Statement: Data and protocols associated with this publication not included in the manuscript and supplementary can be made available upon request. The RT/compound 27 structure has been deposited in the Protein Data Bank (PDB ID 8FFX). Atomic coordinates and experimental data will be released upon article publication.