Since the identification of the human immunodeficiency virus (HIV) as the etiologic agent of acquired immunodeficiency syndrome (AIDS) (Barré-Sinoussi et al., 1983), the efforts of the scientific community have led to the approval of an originally unpredictable number of drugs belonging to six different classes, which, combined in several therapeutic regimens called highly active antiretroviral therapy (HAART), represent a unique success in the treatment of viral infections, turning a highly lethal syndrome into a chronic disease. HIV-1 infects CD4+ cells of the immune system, causing persistent infection. Moreover, it integrates its genome into the host cell genome and replicates, thus requiring a permanent treatment to control the viremia and to avoid progression to AIDS.

The management of this disease, however, is very difficult, and AIDS-related numbers are still impressively high, with 38.4 million people living with HIV, 1.5 million newly infected, and 650,000 deaths in 2021. Although AIDS-related mortality was reduced by the introduction of HAART, all the drugs on the market are susceptible to the selection of drug-resistant variants (Gupta et al., 2018), even the new first-line integrase (IN) inhibitors. The scientific community’s efforts are focused on the identification of innovative compounds able to block the circulating drug-resistant variants. In this context, the identification of agents capable of simultaneously blocking multiple viral functions (multitarget approach) represents a very promising strategy for the treatment of HIV (Ramsay et al., 2018). As a result, the replicative cycle of HIV has been studied in detail, and many of its steps have been validated as pharmacological targets (Sierra and Walter, 2012; Puhl et al., 2019).

In particular, two proteins play a key role in the early phases of the replication, namely, the reverse transcriptase (RT) (which converts the viral genome (ssRNA) into a proviral double-stranded DNA (dsDNA) (Sarafianos et al., 2009)) and the IN (which imports the viral genome into the host cell nucleus and integrates it into the host genome (Esposito and Craigie, 1999)). Once integrated, the provirus becomes a source of mRNAs that code for viral proteins and ssRNA genomes that together form the new viral particles.

RT is a multifunctional complex composed of two subunits of 66 and 51 kDa (p66/p51), with activities of DNA polymerase and ribonuclease H (RNase H). For DNA polymerization, RT uses either RNA (RNA-dependent DNA polymerase (RDDP) activity) or DNA (DNA-dependent DNA polymerase (DDDP) activity) as a template. The reverse transcription proceeds through an RNA/DNA hybrid, whose RNA must be removed to allow the synthesis of the second DNA strand. This process is operated by the RNase H-associated function through a sequence of highly specific hydrolytic events. Nevertheless, RT is extremely error-prone due to a missing draft control system, and the produced mutations can be amplified by recombination. This is one of the main reasons why the discovery of a vaccine is a hard task. Accordingly, due to their key role in viral replication, RDDP and RNase H are both valid targets for the identification of new inhibitors since both are essential to accomplish the reverse transcription, and none of them can be complemented by host proteins (Cristofaro et al., 2002; Corona et al., 2014a; Corona et al., 2014b).

However, all currently approved RT inhibitors only target the polymerase function, acting on the incorporation of nucleotides either as nucleoside analogs [nucleoside reverse transcriptase inhibitors (NRTIs)] or as non-nucleoside allosteric drugs [non-nucleoside reverse transcriptase inhibitors (NNRTIs)]. So far, no inhibitors of the RNase H function have been approved for clinical use. Thus, RNase H is the only viral enzyme function for which no inhibitor has undergone clinical development so far, making it a very promising and coveted target for the development of antiretroviral agents.

The DNA polymerase and RNase H activities are performed by two different domains on the p66 subunit. The DNA polymerase domain is located at the N-terminal and possesses the conformation of a “right hand”; conversely, the RNase H domain is located at the C-terminal portion, 60 Å far from the polymerase site. These two domains are connected by a connection domain. Long-range effects and a functional interdependence between the two active domains have been suggested, based on direct site mutagenesis studies that demonstrated how changes in the HIV-1 RT polymerase domain can influence the activity of RNase H, while deletions at the C terminal can decrease the efficiency of DNA polymerization (Li et al., 2016). This structural and functional interdependence is also supported by evidence showing that mutations in the RNase H domain influence resistance to NRTIs, while NNRTIs, such as nevirapine and efavirenz (EFV), increase the activity of the RNase H function (Julias et al., 2002; Corona et al., 2016b). In this scenario of close interdependence, the identification of a compound with the ability to inhibit both RT activities could represent a significant advance in the fight against drug resistance (Distinto et al., 2012; Distinto et al., 2013).

Over the years, three classes of RNase H inhibitors have been identified in preclinical studies: i) active site inhibitors, which chelate the two Mg²⁺ ions coordinated in the enzyme site by the DDE motif and used as cofactors in the catalysis; ii) allosteric inhibitors that bind the p66/p51 interface; and iii) dual allosteric inhibitors of both the RDDP and RNase H functions. None of them has reached clinical trials so far (Distinto et al., 2013; Corona et al., 2016a; Boyer et al., 2018; Martín-Alonso et al., 2022; Zhang et al., 2022).

Several HIV-1 RT RNase H inhibitors mostly belonging to active-site inhibitors showed promising activity against viral replication, with EC₅₀ values in the low micromolar range (Corona et al., 2020; Kang et al., 2023). Although some reports suggested that RNase H inhibitors could negatively impact the activity of NRTIs such as AZT (Davis et al., 2011), it is worthy of note that some RNase H inhibitors were also reported to retain full potency of inhibition against HIV-1 vectors that carry well-known INSTI, NNRTI, and NRTI resistance mutations in a single round of replication (Boyer et al., 2018) and have full potency against HIV-1 K103N-Y181C replicant virus conferring cytoprotective effect in a 5 days of cytopathic effect (CPE) assay (Corona et al., 2020), making them very appealing as potential therapeutics to treat the circulating drug-resistant strains. Allosteric inhibition of RNase H/RDDP has been described for scaffolds characterized by variously substituted aromatic portions linked by a functionalized spacer containing groups capable of donating or accepting hydrogen bonds, whether linear or made from heterocyclic rings (Distinto et al., 2013; Meleddu et al., 2014; Meleddu et al., 2015; Meleddu et al., 2017; Sonar et al., 2017; Esposito et al., 2019; Fois et al., 2021; Meleddu et al., 2021). It has been proposed that these compounds possess an innovative mechanism of action since they could bind two different allosteric RT pockets: the first one is located at the DNA polymerase domain (partially overlapping the NNRTI binding pocket), and the second one is under the active RNase site H near the interface between p66/p51 (Corona et al., 2016a).

Interestingly, both the catalytic sites of RNase H and IN belong to the polynucleotidyl transferase superfamily and exhibit a high degree of structural similarity, suggesting the development of dual-action drugs targeting both proteins (Yang and Steitz, 1995; Esposito et al., 2020).

HIV-1 IN is composed of three structurally distinct domains: the N-terminal domain (NTD), the catalytic domain (CCD) (which contains the catalytic DDE motif and coordinates two Mg²⁺ ions), and the C-terminal domain (CTD) (Esposito and Tramontano, 2014). IN acts in the context of a macromolecular assembly called the pre-integration complex, which is made up of viral DNA, viral proteins, and host cell proteins. It performs two different catalytic reactions. In the cytoplasm, IN processes the ends of the viral DNA 3′ (3′-processing), while in the nucleus, this viral DNA is a substrate for the transfer reaction (stand transfer) and is inserted into the chromosome. IN acts as a multimer: the dimeric form is required for the 3′ processing phase, while the tetrameric form forms, with the processed DNA, a stable complex called the intasome, allowing the strand transfer reaction, which takes place through the interaction with a cellular factor: the human lens epithelium-derived growth factor (LEDGF/p75) (Esposito and Craigie, 1999).

Inhibitors of IN represent the latest innovation in the repertoire of drugs for the treatment of HIV-1. Those approved so far, raltegravir (RAL), elvitegravir (EVG), and dolutegravir (DGV), all act in the active site of IN simultaneously interacting with the protein, the Mg cofactors, and the DNA by inhibiting the strand transfer reaction (INSTI). Unfortunately, although extremely effective in the management of HIV, all INSTIs are susceptible to failure through the selection of mutations that often also confer cross-resistance (Malet et al., 2015; Dekker José et al., 2022; Richetta et al., 2022; Dekker et al., 2023). For this reason, the interest of the scientific community is growing in the identification of allosteric inhibitors. Recently, molecular inhibitors of the IN site that binds LEDGF/p75 (LEDGIN), which block HIV-1 replication and modulate the oligomerization of IN, have been identified (Demeulemeester et al., 2014). Furthermore, a different single compound, Kuwanon-L, has been reported to bind a second allosteric site, inhibiting the catalytic activity of the enzyme and the interaction IN/LEDGF/p75, promoting and stabilizing the multimerization of IN, reducing its flexibility, and ultimately leading to the inhibition of the catalytic process (Esposito et al., 2015). Recently, it has been shown that Kuwanon-L also inhibits the activity of RT (Martini et al., 2017), opening the way to the development of dual allosteric inhibitors that inhibit both RT and IN. Thanks to the availability in public databases of structural information arising from both RT 3D high-resolution crystalline structures and ligands, both the ligand-based approach and the structure-based approach have been widely used to identify in silico new ligands (Frey et al., 2022). In addition, the resolution of the HIV-1 intasome (Passos et al., 2017; Engelman and Cherepanov, 2021) and the availability of several structures containing INSTI complexed to the Prototype Foamyvirus intasome, which presents a very high level of structural homology with HIV-1, make the study of molecules acting on both systems possible. The ((Z)-3-(2-(4-(3,4-dihydroxyphenyl)thiazol-2-yl)hydrazineylidene)indolin-2-one (1, Figure 1 ) structure was identified as the initial prototype for dual RT inhibitors through an in silico screening based on the shape-based method (Distinto et al., 2012). On these bases, the design and synthesis of differently substituted analogs were performed, which demonstrated a stable inhibitory capacity toward both RNase H and RDDP (Meleddu et al., 2014; Meleddu et al., 2015; Corona et al., 2016b; Meleddu et al., 2017; Meleddu et al., 2021) ( Figure 1 ).

A small library of 5-nitro- and 5-chloro-indolinone derivatives, namely, compounds 1a–12a, 1b–12b, and 1c-12c, was designed and synthesized to explore their ability to inhibit both associated functions of HIV RT. According to a previously described procedure (Meleddu et al., 2015; Meleddu et al., 2017), the synthesis consists of a multi-step approach, starting from the reaction of the 5-nitro or 5-chloro-substituted indolinone with hydrazinecarbothioamide, as in the case of compounds 1a–12a and 1b–12b, and N-methylhydrazine carbothioamide, as in the case of compounds 1c–12c, to give the intermediate thiosemicarbazones and methyl-thiosemicarbazones, which were further reacted with the appropriate alpha halogen-aryl ketones to furnish the desired compounds. Therefore, three families of compounds were synthesized, namely, two series of hydrazothiazoles 1a–12a and 1b–12b and one series of hydrazo-N-methyl-dihydrothiazoles, namely, compounds 1c–12c. All compounds were purified by either crystallization or column chromatography and submitted to biological testing.

When tested on the two enzymatic RT functions, all the new compounds exhibited dual activity with very few exceptions, as for compounds 6c and 8c, which were active only toward the RDDP function (Table 4). However, two compounds of the N-methyl-dihydrothiazole series, 4c and 9c, were the most potent inhibitors of the RNase H function, with an IC₅₀ value of 1.9 μM. Compound 4c was also the most active derivative toward the RDDP function, with an IC₅₀ value of 6.0 μM. According to the enzymatic evaluation results, the N-methyl-dihydrothiazole series could be considered the most promising one among the three tested families.

Furthermore, all compounds with an IC₅₀ value lower than 20 μM against both enzymatic functions were tested for their ability to inhibit viral replication in cell cultures (Table 5). Unfortunately, although interesting in the enzymatic assays, N-methyl-dihydrothiazole showed greater cytotoxicity when compared to the hydrazothiazole congeners. Among the hydrazothiazole derivatives, four compounds were found active in the replication assay, namely, compounds 1a, 3a, 10a, and 9b, with EC₅₀ values of 25, 32, 15.8, and 13.2 μM, respectively. Interestingly, none of these four derivatives exhibited toxicity up to 100 μM (the highest concentration tested) against MT4 cells. Compounds 10a and 9b were the most potent and promising derivatives. However, considering that nitro-indole derivatives were generally better performing in the cell-based replication assays, compound 10a was chosen for further investigation on the mechanism of action, and 9b was flanked to the comparative study as a selective inhibitor for RT.

Accordingly, we measured quantitatively the formation of various viral DNA species during a single HIV-1 replication cycle in the presence of increasing concentrations of 9b and 10a (Figure 3), using the NNRTI EFV as a control for retrotranscription and INSTI RAL as a control for integration. Both compounds were able to cause a decrease in total viral DNA, a product of retrotranscription, confirming an inhibitory effect exerted on RT (Figure 3A). Moreover, the integrated viral DNA was quantified (Figure 3B), on which 10a exerted a significantly more marked effect than 9b.

To verify if this effect was the consequence of the inhibitory activity of 10a against IN, we measured the levels of DNA 2LTRc, a sign of increased events of DNA circularization due to inhibition of its integration into the host genome. A marked increase of DNA 2LTRc was observed in cells treated with 10a, while no increase was observed when 9b was used. Prompted by these interesting results, we investigated the nature of the interaction between 10a and IN in a biochemical assay using RAL as an active site inhibitor and Kuwanon-L operator as positive control (Table 6).

For comparison, 9b was also tested. The results show that, as expected, 9b is not able to inhibit IN. Conversely, 10a acts with a singular mechanism, different from the compounds hitherto known. Indeed, 10a inhibits the enzymatic activity and the interchange of the subunits of IN without interfering with the binding of LEDGF, contrary to Kuwanon-L, which has been predicted to bind the same site but, being a bulkier compound, interferes also with LEDGEF binding site. According to our findings, compound 10a might act by an innovative allosteric mechanism (Shkriabai et al., 2004; Sippel and Sotriffer, 2010), interfering with the IN–IN dimer subunit exchange, which is essential for the enzymatic activity, and the IN multimerization, referred to in literature for other non-catalytic site integrase inhibitors (NCINI), e.g., 1-pyrrolidineacetamide derivatives (Du et al., 2008) (2E)-3-[3,4-bis(acetoxy)phenyl]-2-propenoate-N-[(2E)-3-[3,4-bis(acetyloxy)phenyl]-1-oxo-2-propenyl]-l-serine methyl ester (Kessl et al., 2009). This mechanism is promising because it would offer an option against RAL-resistant strains since most mutations are located close to the DDE motif (Cooper et al., 2008).

Therefore, to better understand the binding of 10a, docking experiments using the enzymes RT and IN were performed.

The interactions between compound 10a (Figures 2A, B) and both NNRTI and RNAse H allosteric sites (Himmel et al., 2009) can explain the dual inhibition of both associated catalytic functions. These include π–π with Tyr188, Phe227, Trp229, and Tyr318; a hydrogen bond (HB) with Lys101; and several hydrophobic interactions with Val106, Val108, Phe227, and Trp229.

The complex IN dimer 10a (Figures 2C, D) is stabilized by multiple van der Waals contacts with Val88, Ile89, and Pro90 of both dimer chains, as well as cation–π interactions with Lys173 of chains A and B, HB with Tyr99. The acquired knowledge can help to optimize the multitarget activity of this compound and its pharmacokinetic properties.

In order to better characterize the potential against circulating drug-resistant variants carrying mutations within the NNRTI binding pocket and close to the binding site predicted for compounds 10a and 9b, both compounds were tested against the RNase H and RDDP activities of HIV-1 RTs bearing K103N, Y181C, and Y188L amino acid substitutions, the most frequent resistance mutations described for NNRTIs (Wensing et al., 2019). For RNase H and RDDP inhibition, compound b-thujaplicinol (BTP) and the NNRTI EFV were used as positive controls, respectively. The results showed that all the enzymes retained almost full sensitivity to both compounds vs. the RT-associated RNase H function, with a slight increase in IC₅₀ values that were in all the cases less than threefold and significant only for compound 9b against K103N ( Figure 4A ). A similar result was obtained against the RDDP activity ( Figure 4B ). Also, in this case, the increase in IC₅₀, although found to be significant in most of the cases, ranged from 1- to 2.6-fold. For EFV, the increase in IC₅₀ values was remarkable, which was 3.5-fold against Y181C, 9.1-fold against K103N, and 14-fold against Y188L. These data suggest that compounds 9b and 10a would retain a good potency of inhibition against circulating variants bearing these mutations.

These data indicate that the 5-nitro-3-(2-(4-arylthiazol-2-yl)hydrazineylidene)indolin-2-one and 5-chloro-3-(2-(4-arylthiazol-2-yl)hydrazineylidene)indolin-2-one scaffolds have unique potential for the design of both dual RT inhibitors, able to block the replication of HIV-1 acting in an allosteric manner on both catalytic activities of RT, and, for multitarget inhibitors, capable of blocking the replication of HIV-1 by acting in an allosteric manner on both RT and IN. The synthesized molecules have been characterized in detail. The best derivatives undoubtedly deserve further studies to fully characterize the modalities of interaction with the two viral enzymes and the mechanism of action. Structural optimization will hopefully lead to increased inhibitory activity, making a significant step toward the development of a new class of HIV-1 inhibitors.

The raw data supporting the conclusions of this article will be made available by the authors without undue reservation.

All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.