Identification and assessment of the 1,6-dihydroxy-pyridin-2-one moiety as privileged scaffold for HBV ribonuclease H inhibition

Abstract

The Hepatitis B Virus (HBV) ribonuclease H (RNase H) although promising remains an unexploited therapeutic target. HBV RNase H inhibition causes premature termination of viral minus-polarity DNA strands, prevents the synthesis of the viral positive-polarity DNA strand, and causes accumulation of RNA:DNA heteroduplexes within viral capsids. As part of our ongoing research to develop more potent anti-HBV RNase H inhibitors, we designed, synthesized and analyzed a library of 18 novel compounds (17 N-hydroyxpyridinedione (HPD) imine derivatives and 1 barbituric acid analogue) as potential leads for HBV treatment development. In cell assays, fourteen HPDs showed significant anti-HBV activity with EC₅₀s from 1.1 to 2.5 μM and selectivity indices (SI) of up to 58. Three of them exhibited more than 3-fold improvement in the SI over the best previous HPD imine (SI=13). To gain insight to the interaction between the tested compounds and the active site of HBV RNase H, docking experiments were undertaken. In almost all binding poses, the novel HPDs coordinated both active site Mg²⁺ ions via their oxygen trident. Furthermore, the novel HPDs displayed high cell permeability and solubility as well as good drug-like properties. These results reveal that HPD imines can be significantly active and selective HBV inhibitors, and that the HPD scaffold merits further development towards anti-HBV agents.

Graphical Abstract

1. Introduction

Hepatitis B virus (HBV) is a small, hepatotropic DNA virus that replicates by reverse transcription (Seeger and Mason, 2000). HBV infections are a major health burden, with an estimated 296 million people who are chronically infected globally and 1.5 million new infections annually, resulting in around 820,000 deaths per year (WHO, 2023).

Current treatments for HBV infection are dominated by monotherapy with nucleos(t)ide analog drugs including lamivudine, adefovir, telbivudine, entecavir, and/or tenofovir, with the latter two drugs being the standard of care for most patients (Jeng et al., 2023; Prifti et al., 2021). The nucleos(t)ide analog drugs target the reverse transcriptase (RT) active site on the viral polymerase protein (P) and suppress HBV DNA elongation during reverse transcription. Despite their high potency and ability to control viremia and reduce hepatic inflammation, HBV clearance rates are only about 3−6% even after a decade or more of administration, so extended, often life-long, treatment is needed (van Bömmel et al., 2010; Woo et al., 2010; Wursthorn et al., 2010). Therapy is an economic burden on patients and can cause adverse events due to prolonged drug exposures. Therefore, identification of novel therapeutic targets is needed to combat HBV more effectively (Revill et al., 2019).

The HBV P protein has two enzymatic activities, the RT and the ribonuclease H (RNase H), that work together to copy the viral genome. The RT activity copies the RNA phase of the viral genome, the pregenomic RNA (pgRNA), into the negative polarity DNA strand [(−) DNA] within nascent viral capsids in the cytoplasm. After the pgRNA has been copied into (−) DNA, the RNase H domain destroys the pgRNA template strand. Because of steric hindrance and template strand occlusion by the pgRNA which stays intact and attached to the (−) DNA in extensive RNA:DNA heteroduplexes when RNase H activity is inhibited, synthesis of the plus polarity DNA strand [(+) DNA] cannot proceed. Consequently, inhibiting HBV RNase H prevents generation of infectious virions and replenishment of the nuclear form of the viral DNA, the covalently closed circular DNA (cccDNA), which is the template for HBV RNA transcription (Tavis et al., 2013). Despite the RNase H being a promising drug target, no drugs targeting the RNase H exist.

The lack of a structure of the HBV polymerase (P) has long precluded structure-guided drug design. We recently predicted and validated the structure of P (including its RNase H domain), enabling application of computational chemistry approaches to HBV RNase H drug development (Tajwar et al., 2022). This model revealed key aspects of the enzyme’s interaction with its heteroduplex substrate and details of its binding to the two Mg²⁺ ions that are essential for catalytic activity (Moianos et al., 2023).

We previously identified inhibitors of the HBV RNase H in the low- to sub-micromolar range belonging to four compound chemotypes: α-hydroxytropolones (αHT), N-hydroxyisoqinonlinediones (HID), N-hydroxypyridinediones (HPD), and N-hydroxynapthyridinones (HNOs) (Cai et al., 2014; Edwards et al., 2017; Hu et al., 2013; Lomonosova et al., 2017a; Lomonosova et al., 2017b; Lu et al., 2015; Lu et al., 2016; Tavis et al., 2013; Villa et al., 2016) (and unpublished) (Fig. 1). These chemotypes are believed to bind to the active site in part by chelating the catalytic Mg²⁺ ions via a trident of oxygen or oxygen and nitrogen atoms. The compounds computationally dock into the HBV RNase H active site with the predicted coordination of the Mg²⁺ ions by the metal binding trident. Removing or blocking the metal-binding moieties eliminates inhibition of HBV replication (Edwards et al., 2019b). We previously reported that HPDs with modifications at the imine moiety increased the antiviral activity of the compounds from 0.69 μM (compound 208, (Edwards et al., 2017)) to 0.23 μM (compound A23, (Edwards et al., 2019a)). Here we report ongoing efforts to improve the anti-HBV activity of the imine HPD compound chemotype, to expand structure-activity understanding for the HPD chemotype, as well as improve drug-likeness properties and ADME-Toxicity profiles. Seventeen new imine HPDs and one structurally related barbituric acid analogue were synthesized and evaluated for inhibition of HBV replication to guide advancement of the HPD chemotype of HBV RNase H inhibitors.

Fig. 1. Example structures from the four chemotypes of HBV RNase H inhibitors.

1, N-Hydroxyisoquinolinediones (HIDs); 2, N-Hydroxynapthyridinones (HNOs); 3, N-Hydroxypyridinediones (HPDs); 4, α-Hydroxytropolones.

2. Methods

2.1. Molecular Modeling.

We generated the HBV RNase H structural model through Alphafold2 and then placed Mg⁺² ions into the active site by superposition of the DEDD active site motif of HBV RNase H onto the HIV RNase H co-crystal structure (PDB: 3K2P) (Tajwar et al., 2022). Ligands were prepared using LigPrep (Schrödinger LLC) including energy minimizations with OPSL4 force field, producing the different protonation states of the ligands using Epik at pH 7.5 +/−2, adding additional ionization states to ensure the full spectrum of metal binding states are represented, and desalting and tautomerization the compounds while retaining chirality. The RNase H model containing Mg²⁺ ions in the appropriate ionization states were prepared with the Schrödinger Protein Preparation wizard. The protein was protonated at pH 7.5 ± 2, hydrogen bonds were assigned with PROPKA (Schrödinger LLC) at pH 7.5, and energy minimization was done with OPSL4 force field. Induced fit docking (IFD) was performed to analyze the binding poses of HPDs. The docking grid around the RNase H active site was defined by placing β-thujaplicinol via superposition of the active site from the HIV RNase H co-crystal (PDB: 3K2P) onto the HBV model. The centroids of the ligand in the active site was used to create 10 Å receptor grid for docking. Protein refinement was carried out at a Van der Waals radius scaling factor of 0.7 for the protein and 0.5 for the ligand. Twenty poses were retained in the initial docking, residues were refined within 5.0 Å of the ligand poses, and redocking was performed with the best structures within 30.0 kcal/mol and the top 20 overall structures.

2.2. Cells and cell culture.

HepDES19 cells are HepG2 cells that carry a stably transfected HBV genotype D genome under control of a tetracycline repressible promoter.(Guo et al., 2007) The cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM/F12) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. Cells were maintained in the presence of 1 μg/mL tetracycline to repress the expression of the stably integrated HBV genome, and HBV replication was induced by withdrawal of tetracycline from the culture medium.

2.3. qPCR HBV replication inhibition assay.

HBV replication inhibition was measured following 3 days compound exposure in HepDES19 cells induced to replicate HBV by removal of tetracycline as previously described(Edwards et al., 2019a; Li et al., 2021). Quantitative PCR was performed with 40 cycles of 95 °C for 15 s and 60 °C for 1 minute employing the Kappa Probe Force universal PCR master mix. The primers and probe (IDT Inc.) for the (+) DNA strand were 5’CATGAACAAGAGATGATTAGGCAGAG3’, 5’GGAGGCTGTAGGCATAAATTGG3’, and 5’/56 FAM/CTGCGCACC/ZEN/AGCACCATGCA/3IABkFQ. The primers and probe for the (−) DNA strand were 5’GCAGATGAGAAGGCACAGA3’, 5’CTTCTCCGTCTGCCGTT3’, and 5’/56 FAM/AGTCCGCGT/ZEN/AAAGAGAGGTGCG/3IABkFQ. EC₅₀ values were calculated from the (+) DNA data with GraphPad Prism using the four parameter log(inhibitor) versus response algorithm with the bottom value set to zero.

2.4. MTS cytotoxicity assay.

Cell viability in the presence of compounds was assessed using the CellTiter 96TM Aqueous Non-Radioactive Cell Proliferation assay (MTS) (Promega, Madison, WI) in HepDES19 cells after 3 days compound exposure as described previously (Edwards et al., 2019a). The cytotoxic concentration 50% (CC₅₀) values were calculated with GraphPad Prism by using the four parameter variable response log(inhibitor) versus response algorithm with the bottom value set to zero.

2.5. Solubility limit assay.

Compound solubility limits were initially tested in DMEM-F12 without phenol red (Gibco) supplemented with 10% FBS (pH 7.2–7.4) to mimic cell culture experiments, as we have done previously (Edwards et al., 2019a). Solubility limits were also assessed in physiologically relevant fluids, fasted simulated gastric fluid (fassGF pH 1.6), fasted simulated intestinal fluid (fassIF, pH 6.5) and fed simulated intestinal fluid (fessIF pH 5). Physiologically relevant fluids were purchased from Biorelevant. Compounds were serially diluted in buffer (upper limit 200 μM, DMSO 1%) and read on a plate reader at 620 nm. Compound concentration (μM) was plotted against optical density (OD) and a solubility limit was determined by identifying an inflection point where there was a significant increase in optical density due to increasing turbidity; the concentration of compound at the inflection point was defined as the solubility limit. Two or more replicate assays were performed on different days for each compound at each pH.

2.6. Parallel artificial membrane permeability assay (PAMPA).

Apparent passive permeability (P_app) was assessed across artificial membranes in PAMPA studies (Sugano et al., 2001). An artificial membrane composed of 1% w/v lethicin/dodecane was added to the poly(vinylidene fluoride) (PVDF) membrane filter in a 96-well donor/acceptor cassette (Sigma Aldrich), which mimics the apical and basolateral sides of the small intestine epithelium. Once dry, compounds (200 μM) were diluted in buffer at pHs 7.54 and 5.0 and added to the donor side of the membrane. The same buffer was added to the acceptor side. The cassette was assembled and incubated at room temperature with shaking at 250 rpm for 2 h, after which 100 μL of the acceptor well was retrieved and compound absorbance was read on a plate reader between 200 and 600 nm, depending on the absorption spectrum of each compound. P_app values (cm/sec) were determined by normalizing to compound absorbance at equilibrium, incubation time, and membrane porosity (Sugano et al., 2001). Two or more replicate assays were performed on different days for each compound at each pH.

3. Results and discussion

3.1. Chemistry.

A three-step synthetic approach was used to synthesize the novel HPD compounds (Scheme 1). The key intermediate 5-acetyl-1-(benzyloxy)-6-hydroxy-4-methylpyridin-2(1H)-one 5 was synthesized with an improved yield compared with that of literature of 75% by refluxing a mixture of diketene (2 eq) and O-benzyl hydroxylamine (1 eq) in the presence of triethylamine (1 eq) in anhydrous toluene. Afterwards, the catalytic hydrogenolysis of the benzyl group over 10% palladium on carbon yielded the target compound 6 almost quantitatively. 5-Acetyl-1,6-dihydroxy-4-methylpyridin-2(1H)-one 6 was condensed with the suitable substituted aniline in absolute ethanol at reflux using sulfuric acid as catalyst in the presence of activated 4Å molecular sieves. The targeted compounds were isolated in good yields ranging from 60% to 70%, with compound 15 being the only exception as it was synthesized in an overall yield of 25%. Compound 23 was synthesized by coupling key intermediate 6 with benzylamine in absolute ethanol at reflux, without the use of sulfuric acid as catalyst. Finally, compound 24 was synthesized by a two-step reaction sequence starting from barbituric acid, which was acetylated in the presence of acetic anhydride at reflux. 5-Acetyl barbituric acid was coupled with aniline following the coupling reaction described for the synthesis of compound 23 to give target compound 24.

Scheme 1. Synthesis of target compounds 7–24.

Reagents and conditions: (a) Triethylamine (TEA) (1.0 eq), dry toluene, 65 °C, 4.5 h, Ar; (b) H₂, 40 psi, Pd/C (10%), 20 min, rt; (c) substituted aniline or primary amines, conc. H₂SO₄, EtOH, activated 4Å molecular sieves, 60 °C, 4–24 h, Ar; (d) C₆H₅CH₂NH₂, EtOH, activated 4Å molecular sieves, 20 h, 60 °C, Ar; (e) acetic anhydride, conc. H₂SO₄, reflux (60 °C), 48 h; (f) aniline, EtOH, activated 4Å molecular sieves, 72 h, 70 °C, Ar.

3.2. Efficacy against HBV replication and cytotoxicity.

Inhibition of HBV replication was assessed using our standard HBV replication inhibition assay (Li et al., 2021). HepDES19 cells that are stably transfected with a tetracycline-repressible HBV genomic expression cassette (Guo et al., 2007) were plated in a 96 well plate and grown for 48 hours in the absence of tetracycline to induce HBV replication. HBV expressing cells were then incubated with compound for three days and HBV nucleic acids were harvested. Accumulation of both DNA strands of the viral genome were assessed by qPCR because inhibiting the RNase H preferentially suppresses the (+) DNA strand (Li et al., 2021). Suppression of the (−) polarity strand was also monitored to provide a rough estimate of the proportion of the inhibition stemming from RNase H inhibition because preferential suppression of the (+) strand is a hallmark of RNase H inhibitors. In contrast, (−) strand inhibition correlates fairly well with cytotoxicity for most RNase H inhibitors we have tested. Twelve HPDs inhibited HBV replication at low micromolar levels, with EC₅₀s ranging from 1.1 to 2.9 μM (Table 1). To assess if the compounds had adverse effects on cellular health, we determined CC₅₀ values following three days compound exposure using an MTS assay that measures mitochondrial function. These assays measure the sum of cytotoxic and cytostatic effects of the compounds because the cells are plated at a low enough density that they grow continuously thought the three day compounds-exposure period. CC₅₀s ranged from 9.5 to 100 μM, and 11 of 18 compounds had CC₅₀s greater than 50 μM (Table 1). Active compounds were defined as those with a selectivity index (SI, CC₅₀/EC₅₀) >5.

Table 1.

Compound structures, physical properties, efficacy and cytotoxicity

a/a	Structure	logP1	logD2	tPSA3	Fsp4	Docking score5	EC50 (plus strand)6	EC50 (minus strand)6	CC506	SI7
7		0.91	0.69	74.83	0.14	−7.8	2.2 ± 1.09	66.7 ± 23.6	100 ± 0	46.2
8		0.85	0.08	93.29	0.25	−8.2	1.6 ± 0.86	50 ± 0	79.1 ± 15.8	49.5
9		1.60	1.05	74.83	0.14	−9.4	1.1 ± 0.52	50 ± 0	62.3 ± 11.1	57.7
10		1.97	1.63	74.83	0.20	−8.5	2.5 ± 0.99	66.7 ± 23.6	50.9 ± 20.8	20.3
11		−0.29	−0.65	74.83	0.45	−9.8	2.9 ± 1.1	23 ± 6.95	9.5 ± 6.82	3.3
12		0.82	−0.15	102.53	0.29	−8.3	1.8 ± 0.15	75 ± 25	47.8 ± 14.0	27.3
13		0.85	0.08	93.29	0.25	−8.8	2.4 ± 0.05	66 ± 16	38.5 ± 19.1	16.4
14		0.85	0.17	93.29	0.25	−8.3	3.3 ± 1.61	100 ± 0	37.6 ± 18.2	11.3
15		1.51	0.57	84.06	0.20	−8.6	6.7 ± 0.35	100 ± 0	30.2 ± 4.2	4.5
16		1.01	0.46	74.83	0.14	−8.7	2.1 ± 1.01	53.7 ± 36.3	100 ± 0	48.4
17		1.79	0.94	74.83	0.20	−9.8	2.6 ± 0.21	100 ± 0	81.5 ± 26.1	31
18		1.71	1.64	74.83	0.25	−7.6	2.0 ± 0.4	100 ± 0	58.2 ± 17.2	29.1
19		2.00	2.36	74.83	0.29	−8.2	35.7 ± 3.77	38.7 ± 8.01	26.5 ± 13.4	0.7
20		1.54	0.91	74.83	0.14	−8.2	2.1 ± 0.05	100 ± 0	60.7 ± 29.6	29.6
21		1.01	0.46	74.83	0.14	−8.6	2.5 ± 0.82	100 ± 0	72.8 ± 29.1	28.7
22		0.44	0.27	74.83	0.50	−9.2	30.1 ± 14.2	50 ± 0	55.4 ± 32.7	1.8
23		1.4	− 10.09	73.13	0.13	−9.0	22.4 ± 9.20	50 ± 0	13.8 ± 7.22	0.6
24		0.63	0.21	87.63	0.17	−4.7	100 ± 0	100 ± 0	100 ± 0	1.0
3		1.33	−9.00	73.13	0.14	−9.9	0.63 ± 0.085	24.2 ± 0	8.18 ± 7.64	13.0

¹LogP (<5) is the log of the partition coefficient of a solute between octanol and water. Predicted with FAF4 online server (Lagorce et al., 2017).

²LogD is the log of the partition coefficient of a solute between 1-octanol and water at pH 7.4, or physiological pH.

³tPSA (<140Ų) is the Topological Polar Surface Area (Ų)

⁴Fsp³, the number of sp³ hybridized carbons/total carbon count

⁵Induced fit docking score to the HBV RNase H active site; kCal/Mol

⁶Values in μM

⁷CC₅₀/EC₅₀

3.3. ADME-toxicity profilinge, solubility and apparent passive permeability.

The compounds’ drug-likeness properties and ADME-Tox profile were predicted with the FAF4 online server (Table S1a, Table S1b, Table S2a and Table S2b) (Lagorce et al., 2017) to give an initial evaluation of their pharmacological profiles.

Compound absorption through the gastrointestinal tract (GI) is directly proportional to its solubility and permeability. Therefore, we first assessed the solubility limits of four compounds. Derivatives 7, 9, and 10 were chosen as they are structurally representative of the compound series and include the most active imine HPDs with high SI values, while 24 was chosen for comparison as it does not inhibit HBV replication. Solubility limits were tested for each compound at physiologically relevant pHs, DMEM-F12 without phenol red (7.4) and three simulated bodily fluids, fasted gastric (pH 1.6), fasted intestinal (pH 6.5), fed intestinal (pH 5). Solubility limits were classified as low (< 75 μM), medium (75 – 100 μM) and high (> 100 μM). All compounds had high solubility limits at each pH tested (Table 2).

Table 2.

PAMPA passive diffusion and solubility of the most active derivatives

	PAMPA passive Diffusion1		Solubility2
Compound	pH 7.4	pH 5	pH 1.6	pH 5	pH 6.5	pH 7.4
7	9.42 ×10−07	1.29E × 10−04	200	200	200	133.3
9	4.09 × 10−06	4.21E × 10−04	200	200	200	200
10	2.56 × 10−06	6.07 × 10−05	200	200	200	133.3
24	1.33 × E−06	2.00 × 10−04	200	116.7	150	200

¹PAMPA results are in cm/sec. P_app ≥ 1×10⁻⁶ indicates high permeability and P_app < 1×10⁻⁶ indicates poor permeability

²Solubility limits are in μM with an upper limit in the assay of 200 μM. All assays were done in duplicate or triplicate

Passive permeability is the most common mode of drug transport across the GI tract epithelium (Luo et al., 2021). Therefore, the apparent rate of passive diffusion (P_app) across artificial membranes was assessed for all four compounds at pHs 7.4 and 5 (Table 2). Passive permeability was classified as high (> 1×10⁻⁶ cm/sec) or low (< 1×10⁻⁶ cm/sec) according to an industry-standard cutoff (Zhu et al., 2002). Compounds 9, 10, and 24 had high P_apps at both pHs tested, while 7 had a low P_app at pH 7.4 but a high P_app at pH 5, suggesting that apparent passive permeability is compound-specific and is not directly tied to anti-HBV potency.

3.4. Computational molecular docking.

Induce fit docking (IFD) experiments were done to observe the binding poses of HPD compounds into the active site of the HBV RNase H to gain insight into the structure-activity relationships for HBV RNase H inhibition and to suggest possible chemical modifications for optimization of the compound series. These docking studies were conducted after compound design because the molecular model had not yet been generated during the design phases. IFD generated multiple binding poses for each compound as it considered both receptor and ligand to be flexible. In almost all binding poses, the compounds coordinated both active site Mg²⁺ ions, as illustrated by compound 17 in Fig. 2A. The hydroxyl group at position 2 of the HPD core ring coordinated both Mg²⁺ ions by making salt bridges, while the OH group at position 1 either chelated both Mg²⁺ ions or only one Mg²⁺ ion via a salt bridge. In the case of the barbituric acid analogue 24, three carbonyl groups are present on the HPD core ring at alternative positions, and they failed to chelate either Mg²⁺ ions (Fig. 2B). It seems the positioning of carbonyl and OH groups adjacent to each other on the HPD core ring, like an oxygen trident, is key in order to chelate both metal ions. Compound 24 is inactive (EC₅₀ value of >100 μM) and has a poor docking score (−4 kcal/mol) compared to other compounds (−9.8 to −7.6 kcal/mol), implying that it has the least binding affinity for the RNase H active site.

Fig. 2. Metal chelation by HPDs.

A. Surface diagram and ligand interaction map are shown for compound 17 chelating both Mg²⁺ ions. B. Surface diagram and ligand interaction map for compound 24 reveal that it fails to chelate the Mg²⁺ ions. Left, binding poses; Right, interaction maps.

Positioning of side chains of compounds around the active site of RNase H defined three binding pockets (S1-S3) (Fig. 3A). Pocket S1 consists of residues A701, T702, P703, T704, S791 and R792. Pocket S2 is lined with residues P703, T704, L723, P724 and I725, while pocket S3 is formed by residues N749, Y779, V780, P781, S782 and A783 (numbering based on the HBV genotype B polymerase sequence). Most of the side chains of compounds in the binding pockets were solvent exposed, and a few common interactions were observed for the side chain in the pockets. These include the side chains of compounds 14, 11 and 8 making H-bonds with T702 in pocket S1 (Fig.3B), compound 21 making pi-cation interactions with residue L723 in pocket S2 (Fig. 3C), and compound 10 making halogen bonds with residues E748 and N749 and compound 22 making H-bonds with residue S782, respectively, in pocket S3 (Fig. 3D).

Fig. 3. Compound docking into the HBV RNase H active site.

A. Binding pockets S1-S3 defined by positioning of inhibitor side chains around the active site of the HBV RNase H. B. Compounds 14 (green), 11 (purple) and 8 (blue) making hydrogen bonds with T702 in pocket S1. C. Compound 21 (pink) making pi-cation interactions with L723 in pocket S2. D. Compound 10 (white) making halogen bonds with E748 and N749 while compound 22 (orange) making a hydrogen bond with S782, respectively, in pocket S3. Right side panels showing ligand interaction maps of compounds 14, 21 and 22.

Most amino acid residues near the active site are not conserved between the HBV RNase H and other RNases H, but HBV RNase H residue R792 is conserved with human RNase H1 and the HIV RNase H. It makes a salt bridge with active site residue D788 of the DEDD active site motif similar to its position in the HIV RNase H and human RNase H1. In almost all the binding poses for HPDs into the HBV RNase H active site, the conserved R792 residue makes either a H-bond or salt bridge with an OH group or an H-bond with the carbonyl group on the HPD core ring (Fig. 3). This same interaction was also observed between OH group of an α-hydroxytropolone RNase H inhibitor with a conserved arginine residue in the crystal structure of HIV RNase H co-crystal with the α-hydroxtropolone β-thujaplicinol (Himmel et al., 2009).

4. Discussion and summary

Our past studies with the HPDs versus HBV replication identified the amine 3 (EC₅₀=0.63, CC₅₀=8.18 μM, SI=13 (Edwards et al., 2017)) as the best and most selective among the imine subclass of HPD compounds, and therefore it was the ‘hit’ for investigating the anti-HBV potential of the imine HPDs. Seventeen new HPD derivatives and one structurally related barbituric acid analogue were designed, synthesized, and tested for anti-HBV activity and selectivity.

The EC₅₀ values for the novel HPDs by the three-day qPCR replication inhibition assay ranged from 1.1 to 35.7 μM, and the CC₅₀s ranged from 9.5 to >100 μM, resulting in SIs from 1 to 57.7 (Table 1). To identify the structural features of the HPDs required for potent anti-HBV activity, we modified the imine moiety first by incorporating electron donating (-NH₂, −OCH₃) and electron withdrawing groups (-F, −Cl) in different positions of the aromatic ring to enhance potency, eliminate cytotoxicity and improve physiochemical and druglike properties. The position and the number of substitutions on the aromatic ring seem to play a role in activity. Specifically, the meta fluoro- and meta bromo were the most promising substitutions (compound 16 vs 21), based on SIs (48.4 and 28.7 respectively), while the monosubstituted analogues were more potent than the disubstituted ones (compound 9 vs. 10).

Insertion of a one carbon chain between the aromatic ring and the HPD scaffold (compound 23 vs 7) or replacement of the aromatic moiety with an alkyl chain or a carbocyclic ring (compound 22vs 7 and 11 vs 7 respectively) dramatically reduced activity against HBV. This may be due to specific stacking interactions between the aromatic ring and the protein plus proper orientation of the pharmacophoric oxygen trident. Docking studies strengthen this hypothesis, e.g. compound 21 makes pi-cation interactions with residue L723 in pocket S2 (Fig. 3C). These observations suggests that further improvements in potency may be had by optimizing the extended aromatic system of the imine linker.

Finally, the lack of inhibitory activity for compound 24, in which we replaced the HPD ring with the pharmacophore ring of barbituric acid, was not surprising since the oxygen trident of the N-hydroxyimide group is a prerequisite for the coordination of the metal ions in the RNase H active site. Binding assays using our model of HBV RNase H revealed that compound 24 failed to coordinate either Mg²⁺ ions (Fig. 2B) and highlighted the importance of hydroxyl group at position 2 of the HPD core ring on anti-HBV activity.

5. Conclusion

We tested 18 novel compounds, 17 HPDs and one barbituric acid derivative, for activity against HBV replication in culture. Fourteen HPDs were significantly active against HBV replication, 2 compounds (19 and 22) demonstrated marginal activity, and the barbituric acid derivative 24 was inactive. Although potency of this series was reduced relative to the hit compound 3, most of the novel HPD imines had SIs better than the hit compound. Specifically, compounds 16 (SI=48), 8 (SI=50) and 9 (SI=58) exhibited more than 3-fold improvement in the SI over the best previous HPD imine (3, SI=13) (Edwards et al., 2017). The major conclusions are: (i) the oxygen trident of the HPDs is critical for the coordination of the metal ions in the RNase H active site and thus for the anti-HBV activity, (ii) incorporation of electron withdrawing groups (-F, −Cl) in different positions of the aromatic ring enhanced the potency and improved the selectivity of the tested compounds, (iii) aromatic substitution at the side chain of the HPDs is another essential structural feature of this class of compounds, (iv) increased flexibility of the side aromatic chain reduced favorable aromatic interactions, resulting in lower activity and selectivity compared to the more conformationally constrained counterparts, and (v) the selectivity indices observed for most of the tested analogs are better than the hit compound (3). Moreover, the novel HPDs exhibited high cell permeability and solubility (Table 2) and good drug-like properties (Supplemental file 1). The structure-activity information gained in this study should be fully applicable to the more potent oxime HPD subclass (Edwards et al., 2019a), so will help advance the HPD chemotype more broadly. With further structural optimization, these studies indicate that HBV RNase H drugs, when combined with other inhibitors with different mechanisms of action, should result in better treatment options for the hundreds of million chronically infected people worldwide.

Supplementary Material

Supplemental information

Supplemental file 1. Chemical synthesis (Preparation procedures and characterization data of compounds).

Table S1a, predicted drug-likeness properties of compounds 7–14

Table S1b, predicted drug-likeness properties of compounds 15–24

Table S2a, predicted ADME-Tox profile of compounds 7–14

Table S2b, predicted ADME-Tox profile of compounds 15–24