Cutting into the Substrate Dominance: Pharmacophore and Structure-Based Approaches toward Inhibiting Human Immunodeficiency Virus Reverse Transcriptase-Associated Ribonuclease H

Abstract

Human immunodeficiency virus (HIV) reverse transcriptase (RT) contains two distinct functional domains: a DNA polymerase (pol) domain and a ribonuclease H (RNase H) domain, both of which are required for viral genome replication. Over the last 3 decades, RT has been at the forefront of HIV drug discovery efforts with numerous nucleoside reverse transcriptase inhibitors (NRTIs) and non-nucleoside reverse transcriptase inhibitors (NNRTIs) approved by the FDA. However, all these RT inhibitors target only the pol function, and inhibitors of RT-associated RNase H have yet to enter the development pipeline, which in itself manifests both the opportunity and challenges of targeting RNase H: if developed, RT RNase H inhibitors would represent a mechanistically novel class of HIV drugs that can be particularly valuable in treating HIV strains resistant to current drugs. The challenges include (1) the difficulty in selectively targeting RT RNase H over RT pol due to their close interplay both spatially and temporally and over HIV-1 integrase strand transfer (INST) activity because of their active site similarities; (2) to a larger extent, the inability of active site inhibitors to confer significant antiviral effect, presumably due to a steep substrate barrier by which the pre-existing substrate prevents access of small molecules to the active site. As a result, previously reported RT RNase H inhibitors typically lacked target specificity and significant antiviral potency. Achieving meaningful antiviral activity via active site targeting likely entails selective and ultrapotent RNase H inhibition to allow small molecules to cut into the dominance of substrates. Based on a pharmacophore model informed by prior work, we designed and redesigned a few metal-chelating chemotypes, such as 2-hydroxyisoquinolinedione (HID), hydroxypyridonecarboxylic acid (HPCA), 3-hydroxypyrimidine-2,4-dione (HPD), and N-hydroxythienopyrimidine-2,4-dione (HTPD). Analogues of these chemotypes generally exhibited improved potency and selectivity inhibiting RT RNase H over the best previous compounds and further validated the pharmacophore model. Extended structure-activity relationship (SAR) on the HPD inhibitor type by mainly altering the linkage generated a few subtypes showing exceptional potency (single-digit nanomolar) and excellent selectivity over the inhibition of RT pol and INST. In parallel, a structure-based approach also allowed us to design a unique double-winged HPD subtype to potently and selectively inhibit RT RNase H and effectively compete against the RNA/DNA substrate. Significantly, all potent HPD subtypes consistently inhibited HIV-1 in the cell culture, suggesting that carefully designed active site RNase H inhibitors with ultrapotency could partially overcome the barrier to antiviral phenotype. Overall, in addition to identifying our own inhibitor types, our medicinal chemistry efforts demonstrated the value of pharmacophore and structure-based approaches in designing active side-directed RNase H inhibitors and could provide a viable path to validating RNase H as a novel antiviral target.

Figure 1.

Structure of RT (created with PyMOL based on PDB code 4PQU). The active site of pol is shown in pink and that of RNase H in cyan. The RNA (red) / DNA (blue) heteroduplex engages with both active sites. CN = connection domain. Adapted from ref 14. Copyright 2017 American Chemical Society.

Figure 2.

Two-metal mechanism for HIV-1 RNase H activity.

Figure 3.

Schematic description of modes of HIV-1 RT RNase H cleavage (left, adapted with permission from ref. 6. Copyright 2008 Elsevier) and the corresponding RNA/DNA substrates for measuring each cutting mode (right).

Figure 4.

Active site inhibitor types reported prior to our work. (A) Early inhibitor types 1-3 were mostly metal chelating fragments. (B) Structurally more elaborate inhibitor types 5-7 generally conferred better potency and selectivity. (C) Prior research informed a pharmacophore model consisting of a chelating triad built around a monocyclic or bicyclic heterocycle, a hydrophobic aryl or biaryl moiety, and a one or two atoms flexible linker.

Figure 5.

Redesigned HID subtype represented by compound 8. The design was based on the pharmacophore model. (A) Activity profile of compound 8. (B) Binding mode of 8. Left: structure of full length RT with two subunits p66 and p51. Right: a close-up view of RNase H active site with predicted binding mode of the reference compound 7 (green) and our inhibitor 8 (cyan). Adapted from ref 33. Copyright 2015 American Chemical Society.

Figure 6.

Structures and activity profiles of two additional HID subtypes 9 and 10.

Figure 7.

(A) Pharmacophore-based design of HPCA inhibitor types 14-16. (B) X-ray crystal structure of HIV RT in complex with 16. Cross-eyed stereoview of 16 (cyan) bound at the RNase H active site of HIV RT. The RNase H domain of RT is shown in orange, the p51 in light gray. Conserved active site residues are shown as sticks, and Mg²⁺ ions are shown as magenta spheres. Adapted from ref 38. Copyright 2016 American Chemical Society.

Figure 8.

The design of the original HPD chemotype and its inhibition of RT RNase H: a simple N-hydroxylation converts a potent NNRTI (17) into an RT pol/ INST dual inhibitor (18), which was later found to also inhibit RT RNase H. As was the case with INST inhibition, the OH group on N-3 was required for RNase H inhibition. HBD: hydrogen bond donor.

Figure 9.

Two early HPD subtypes (19, 20) derived from 18, both showing low μM inhibition of RT RNase H of RT, yet different biochemical and antiviral inhibitory profiles. Structural modifications included mainly the C6 hydrophobic moiety (HPD-2) and the linker (HPD-3).

Figure 10.

Redesigned HPD subtypes. (A) The original HPD (18) inhibited RT RNase H moderately and non-selectively. (B) Pharmacophore-based approach led to the design of more potent and selective RNase H inhibitor types 21-23 by eliminating or minimizing N-1 and C-5 substituents. (C) Further SAR identified subtypes 24-27 with optimized RNase H inhibitory profile and significant antiviral activity.

Figure 11.

Design of the HTPD inhibitor types. Hybridization of HID and HPD designed inhibitor type 28, which upon bioisosterism inhibitor led to improved HTPD inhibitor types 29 and 30.

Figure 12.

HPD subtype represented by compound 31 fits the pharmacophore of both INST and RT RNase H inhibitors, and inhibited both enzymatic functions with high potency.

Figure 13.

Design of double-winged HPD subtype 33. Docking of single-winged subtype 21a into RNase H active site without (A) or with (B) substrate. With the substrate binding to the active site, the wing of 21a is forced to flip and the key interaction with H539 is lost. Introducing a second wing (pink) at the C-5 position of HPD led to the design of 32 which allows interactions with H539 and nucleic acid substrate (C, docking of 32). Unsymmetrically double-winged subtype 33 is designed for synthetic accessibility. Adapted from ref 14. Copyright 2017 American Chemical Society.

Figure 14.

Evolution of our designed inhibitor types. The best previous compound (7) was used for benchmarking. Pharmacophore-based design (PBD) led to the 1^st Gen inhibitor types (9, 10, and 21) with improved selectivity profile; further design by altering the linker identified 2^nd Gen HPD subtypes (24, 25, and 27) exhibiting optimal RNase H inhibition and significant antiviral activity. A parallel structure-based design (SBD) generated a double-winged subtype 33 with excellent overall activity profile. A dual pharmacophore (DP) approach allowed the design of 31 which fits the pharmacophores of both RT RNase H and INST.