Pyrrolyl Pyrazoles as Non-Diketo Acid Inhibitors of the HIV-1 Ribonuclease H Function of Reverse Transcriptase

Due to the biological liability of diketo acid (DKA) chain, we transferred this element of our previously reported anti-HIV-1 pyrrolyl derivatives to a non-DKA scaffold, obtaining a series of pyrro...


■ INTRODUCTION
The human immunodeficiency virus type 1 (HIV-1) is the etiological agent of one of the most threatening human diseases, referred to as acquired immune deficiency syndrome (AIDS). Currently, there are close to 40 million people living with HIV/AIDS worldwide, about 1 million AIDS-related deaths, and about 1.8 million new infections were estimated in 2017. 1 Even though the large therapeutic armory (ART, antiretroviral therapy) allowed reaching remarkable milestones, life-long treatments present multiple drawbacks. 2,3 Thus, there is still the pressing demand for a constant effort toward the identification of new targets and the development of new effective anti-HIV agents. 4 Along this line, the development of compounds able to inhibit crucial but unexplored viral processes is an attractive solution to raise the genetic barrier to resistance.
The conversion of the viral RNA into DNA is a critical step in the HIV-1 lifecycle initiated by the reverse transcriptase (RT), a well-established drug target. RT possesses two functional domains: an RNA-and DNA-dependent DNA polymerase domain (RDDP and DDDP, respectively) and a ribonuclease H domain (RNase H). All current anti-HIV drugs acting on RT are targeting the polymerase domain. The RNase H domain catalyzes the selective hydrolysis of the RNA strand of the RNA:DNA heteroduplex, an essential step for viral replication. 5−7 Indeed, several lines of evidence suggest that selectively inhibiting RNase H confers an antiviral phenotype, defining RNase H as an attractive target for rational anti-HIV-1 drug design. 8,9 The discovery of RNase H inhibitors (RHIs) is still in its infancy and none of them has reached the clinical pipeline. 8 Actually, many RHIs have been reported in the literature as effective inhibitors in enzyme assays; 5 nevertheless, they showed high toxicity or they did not exhibit selective inhibition over the human RNase H, 8 RT-associated RDDP, and/or integrase (IN) activities. 9−12 In this field of research, our group has previously reported DKA derivatives that prove to be dual inhibitors of IN and RH 13,14 or IN strand transfer inhibitors (INSTIs) endowed with marginal RH inhibition activities. 15,16 Among the dual inhibitors, notable examples are the phenylpyrrolyl derivative 1, 17 the pyrrolyl diketo hexenoic ester 2, which was the first DKA derivative reported as RHI to have an antiviral effect, 18 and its chlorine analogue 3, which was proven to inhibit viral replication by selectively blocking RNase H function (Figure 1). 19 Despite the relevance of the DKA branch in the inhibitory activity, it is widely reported that this group is biologically liable due to high metabolic turnover and because it shows poor cell membrane penetration. Furthermore, it has been proven to have a time-dependent decrease in the activity in solutions at room temperature, even during short periods of time. 20 Likewise, we observed a time-dependent decay of the inhibition of viral replication using the RHI 3 as an anti-HIV agent in cell-based assays. 19 For this reason, a variety of compounds were developed by mimicking the DKA element into scaffolds characterized by improved druglike qualities, thus overcoming the biological liability of the DKA moiety. A success case in the HIV field is the discovery of Raltegravir (4), a hydroxypirimidinone derivative, the first in class INSTI approved in 2007. 21 Another example reported in the literature are potential INSTIs designed by the conversion of the 1,3-diketo unit into heteroaromatic motifs, obtaining azole carboxylic acid derivatives. 22 In the end, in the field of RHIs, naphthyridinones, like compound 5, resulted in good anti-RH agents 8 but were inactive in multicycle viral replication assays due to high toxicity (Figure 1).
A further drawback reported for DKAs is their low selectivity vs enzyme targets. In fact, in general, they inhibit both IN and RH enzymes. Our works widely demonstrated that the quinolonyl and pyrrolyl DKA derivatives [13][14][15][16][17]19 were dual RH/IN inhibitors, or RH agents endowed with very poor selectivity, 2−3 fold at the best.
So the aim of this work is the design of new RHIs starting from the pyrrolyl DKAs 1−3 to obtain a new class of compounds endowed with improved stability and selectivity/ specificity against HIV RH vs IN enzyme.

■ DESIGN OF NEW RHIS
In the present study we synthesized and biologically evaluated new compounds designed starting from our previously reported DKA inhibitors. In particular, starting from the structure of 1, we decided to mimic the α−γ-diketo unit of the DKAs with a pyrazole ring (Figure 2), to obtain potentially more stable pyrrolyl−pyrazole carboxylic acid derivatives (Chart 1).
This modification was suggested also by results reported by Zeng et al., 22 who described pyrazole carboxylate derivatives designed as DKA isosteres. These compounds proved to be very weak or null INSTIs and therefore their antiviral activity was supposed to be related to a mechanism of action different from the IN inhibition.
Thus, we designed the compounds reported in Chart 1, by keeping fixed the pyrrolyl−pyrazole carboxylic moiety, we conceived a new series of derivatives (6 and 7) characterized by one or more of the following modifications: (i) the removal of the substituent in 1-and/or 4-position of the pyrrole core; (ii) the introduction of various alkyl or arylalkyl substituents replacing the p-fluorobenzyl ring in position 1 of the pyrrole nucleus; (iii) the replacement of a naphth-2-yl group replacing the phenyl ring in position 4 of the pyrrolyl skeleton. Noteworthy, within this series, derivatives 6a and 7a, ester and acid, respectively, represent the non-DKA analogues of 1. Furthermore, a series of oxyphenylpyrrolyl−pyrazole derivatives (10 and 11) was designed mainly by introducing various substituents endowed with different degrees of freedom or steric hindrance on the oxyphenyl portion linked in 4-position to the pyrrole core and/or by replacing the p-fluorobenzyl ring with a methylnaphth-1-yl group on the pyrrolyl nitrogen. Notably, in order to define more extensive structure−activity relationships (SARs), we decided to include a small series of non-pyrrolyl derivatives (8 and 9) characterized by the replacement of the pyrrole core linked in 3-position to the pyrazole ring with a methyl, 23 phenyl, 24 or biphenyl substituent. The newly synthesized compounds have been evaluated in vitro for their ability to inhibit the specific enzymatic activity of recombinant RNase H and IN, for their cytotoxicity and antiviral activity against HIV-1 in human cells. In addition, a rationalization of the interaction with the biological target has been proposed on the basis of docking studies using the crystal structure of RNase H and validated by site-directed mutagenesis on the residues indicated as crucial for the binding. Finally, mass spectrometry (MS) experiments allowed probing the human serum stability of the most active pyrrolyl−pyrazole compounds in comparison with their diketobutanoic congeners.

■ RESULTS AND DISCUSSION
The synthesis of derivatives 6a−h and 7a−h is outlined in Scheme 1.
Scheme 2 describes the synthetic route to obtain derivatives 8c and 9c.
Synthesis of oxyphenylpyrrolyl−pyrazole derivatives 10a−h and 11a−h were performed as reported in Scheme 3. For a detailed description of all synthetic procedures and characterization of final compounds and intermediates, see the Supporting Information.
In Vitro Screening for RNase H Inhibitory Activity. All the newly synthesized compounds 6a−h, 7a−h, 8a−c, 79a−c, 10a−i, and 11a−i were tested in vitro against recombinant    To determine the effect of compounds on viral replication four acid−ester couples of compounds (6a,h−7a,h; 10a,b− 11a,b) were chosen to test the antiviral activity in HeLa-CD4-LTR-β-gal cells. This was initially assessed using 3 (IC 50 value against RNase H of 7.3 μM) 18 25 was employed, whose accuracy in correctly predicting binding poses and energies was already demonstrated in our previous works. 26 The high-resolution crystal structure of HIV-1 RT in complex with a small molecule RHI (PDB code 3QIP) 27 was chosen as macromolecule for docking calculation. Compounds 6a, 6h, 10a, and 10b were selected for in silico experiments, given their promising anti-RNase H activity in vitro. The docked pose for 6a (Figure 3a) indicates that the pyrazole nitrogens and the ester moiety chelate the two magnesium ions in a position that is consistent with other cocrystallized RNase H ligands. 27,28 This pose allows for an H-bonding interaction between the ester group and H539. Of note, the p-fluorobenzyl group is pointing toward the positively charged K540, probably engaging in a cation−π interaction. Moreover, the adjacent W535 also establishes a T-shaped charge-transfer interaction with the same p-fluorobenzyl moiety. Also, the phenyl moiety is lodged in a cleft lined by residues S499, Y501, Q475, E478, N474, and Q475. Consistently with previous mutagenesis studies with ligands featuring comparable binding modes (compound 1), 18 a parallel-displaced π interaction takes place between Y501 and the phenyl moiety. The latter would also engage in a cation−π interaction with R448. In agreement with  RNase H inhibitory activity, it could be postulated that two aromatic rings, one in the cleft comprising S499, Y501, Q475, E478, N474, and Q475, and the other in the gorge with W535 and K540, represent critical structural requirements for RNase H inhibition in our compounds. This might account for the lack of activity shown by compounds devoid of one or both aromatic rings. As inferred in our previous studies, 18 the ester group featured in the compounds of the series 6 and 10, in place of the carboxyl acid in the series 7 and 11, contributes to the charge neutrality of the binding site, in some cases favorably influencing the stability of the ligand/protein complex. This trend is verified for 6a/7a, 6f/7f, 6h/7h, 10f/11f, and 10g/ 11g ester/acid couples. Regarding compound 6h, the binding mode of the inhibitor is similar to that observed for compounds 6a (Figure 3b). One key difference is to be found in the positions of naphthyl and p-fluorobenzyl groups, which appeared switched compared to the corresponding rings in 6a. The bulky naphthyl group can be better accommodated in the cleft harboring the RNase H residues W535 and K540. The enhancement of the interaction with these latter residues could explain the higher activity of compounds carrying this substituent. In the same vein, compound 10a shares a similar binding mode compared to that of 6a, with the flexible pendant benzyloxy portion lodged closer to the RNase H residue R448, thereby enhancing the strength of the cation−π interactions (Figure 3c). Finally, thanks to its long and flexible phenylbutoxy chain, compound 10b delves into an additional pocket, adjacent to the main binding site common with 6a and 10a. Thus, 10b establishes additional contacts with RNase H residues N474, Q475, T473, A446, and R448. In addition to the aforementioned cation−π interaction with R448, also K451 might favorably interact with phenylbutoxy moiety of 10b.
Site-Directed Mutagenesis Experiments. To experimentally verify the binding model suggested by computational studies, seven-point mutants of HIV-1 RNase H were generated by introducing an alanine substitution at residues R448, K451, N474, Q475, Y501, W535A, and K540. Compounds 6a, 6h, 10a, and 10b were tested for their ability to inhibit these mutated RT variants. In parallel, we used RHI 3 12 as control ( Figure 4). In agreement with computational calculations, mutations N474A, Q475A, Y501A, and W535A induced drastic reductions of inhibitory potency (generally more than 5 folds) of all compounds (Figure 4 and Table S2 in  the Supporting Information).
Notably, all these amino acids are highly conserved among HIV-1 subspecies, even in drug-resistant variants isolated from patients. 10 In the case of substitutions K540A and R448A, the loss of potency was less prominent but still significant for all the non-DKA pyrrolyl derivatives, while it was not significant for control 3 that lacks the aromatic substituent in position 4 of the pyrrole ring. These data are consistent with the structural requirements postulated in computational studies suggesting the need for a second aromatic ring in the gorge formed by the RNase H residues W535 and K540. In particular, the interaction with the aforementioned residues should compensate for the absence of DKA moiety (3), which should per se provide higher inhibitory potency than the pyrrole group. In this respect, it is worth noting that compound 6h, carrying the naphthyl group in the 4-position, is the one that exhibited the most significant variation in potency of inhibition against K540A (p-value <0.0002), as predicted. In the case of the K451A substitution, only moderate, but still significant, increases in IC 50 values were observed for 6h and 10a and no effect was observed for 6a, while compound 10b, with the long and flexible phenylbutoxy chain, exhibited the most significant variation in potency (p-value <0.0008) as predicted. The presence of K451A substitution had no effect on the potency of inhibition of compound 3, as expected.
In Vitro Serum Stability. To probe whether our strategy of replacing the biologically labile diketobutanoic scaffold with the pyrrolyl−pyrazole carboxylic moiety resulted in more biologically stable compounds, liquid chromatography−mass spectrometry experiments were performed. In particular, compound 1, its pyrazole derivative 6a, and the corresponding carboxylic acid derivative 7a were assessed for their serum stability by incubating them in 90% human serum at 37°C according to a previously described protocol (see Supporting Information for detailed experimental methods). 29 As shown in Figure 5, 1 starts to be significantly degraded after 3 h while its carboxylic analogue is instantly degraded upon serum exposure. Interestingly, we were able to identify a common degradation product for 1 and its carboxylic analogue, which should be ascribed to the retro-Claisen adduct (see Supporting  Information Table S3 and Figure S1). Similar results were already obtained by other authors for structural congeners of 1. 20 Conversely, in our assay both 6a and 7a remained unaltered for the entire course of the experiment, confirming  that the employed lead-optimization strategy was successful to yield more biologically stable RHIs.

■ CONCLUSIONS
In this work, we used the pyrazole ring as a non-DKA scaffold replacing the DKA chain to overcome its biological limitations. We reached the first goal of this work, obtaining non-DKA pyrrolyl derivatives with selective RNase H activity. Among the newly synthesized derivatives, the oxyphenylpyrrolyl−pyrazole subseries exhibited the best inhibitory potencies, being active within the low micromolar/submicromolar range. The most potent compound was 11b that inhibited RNase H with an IC 50 = 0.27 μM. The selected compounds that have been tested in cell-based assays against HIV, showed low cytotoxicity, with moderate SIs.
Interestingly, all tested compounds showed up to 2 orders of magnitude of selectivity for RNase H vs IN. Docking studies, coupled with site-directed mutagenesis, allowed us to identify the structural requirements that could confer the possibility to establish specific interactions within the RNase H domain, retaining good selectivity against IN. Indeed, the presence of two aromatic rings on the pyrrole nucleus of non-DKA derivatives represents a critical requirement for RNase H inhibition. It is also worth noting that our non-DKA derivatives showed the ability to interact with residues highly conserved among HIV-1 subspecies isolated among patients carrying drug-resistant variants. In the end, the newly discovered pyrazole carboxylic acid derivatives, feature promising serum stability with respect to their corresponding DKAs, thereby confirming the achievement of the second goal of our design strategy.