Selective hepatitis B and D virus entry inhibitors from the group of pentacyclic lupane-type betulin-derived triterpenoids

Current treatment options against hepatitis B and D virus (HBV/HDV) infections have only limited curative effects. Identification of Na+/taurocholate co-transporting polypeptide (NTCP) as the high-affinity hepatic receptor for both viruses in 2012 enables target-based development of HBV/HDV cell-entry inhibitors. Many studies already identified appropriate NTCP inhibitors. However, most of them interfere with NTCP’s physiological function as a hepatic bile acid transporter. To overcome this drawback, the present study aimed to find compounds that specifically block HBV/HDV binding to NTCP without affecting its transporter function. A novel assay was conceptualized to screen for both in parallel; virus binding to NTCP (measured via binding of a preS1-derived peptide of the large HBV/HDV envelope protein) and bile acid transport via NTCP. Hits were subsequently validated by in vitro HDV infection studies using NTCP-HepG2 cells. Derivatives of the birch-derived pentacyclic lupane-type triterpenoid betulin revealed clear NTCP inhibitory potency and selectivity for the virus receptor function of NTCP. Best performing compounds in both aspects were 2, 6, 19, and 25. In conclusion, betulin derivatives show clear structure–activity relationships for potent and selective inhibition of the HBV/HDV virus receptor function of NTCP without tackling its physiological bile acid transport function and therefore are promising drug candidates.


Results
Screening assay for bile acid transport and myr-preS1 2-48 lipopeptide binding. The major goal of the present study was to analyze in parallel the effect of inhibitors on both, bile acid transport via NTCP and myr-preS1 2-48 lipopeptide binding to NTCP in a cell culture system. This was achieved by plating NTCP-HEK293 cells onto 96-well plates and by using tritium-labelled analytes, being [ 3

H]taurocholic acid ([ 3 H]TC) and [ 3 H]
myr-preS1 2-48 lipopeptide ([ 3 H]preS1), which were analyzed in a microplate scintillation counter. For validation of this assay, transport of 1 µM [ 3 H]TC first was inhibited by increasing concentrations of TC (concentrations ranging from 0.1 to 1,000 µM) itself and by myr-preS1 2-48 lipopeptide (concentrations ranging from 0.001 to 10 µM). In addition, binding of 5 nM [ 3 H]preS1 was inhibited by myr-preS1 2-48 lipopeptide itself (concentrations ranging from 0.001 to 10 µM) and by TC (concentrations ranging from 0.1 to 1,000 µM). As shown in Fig. 1a (Fig. 1b). These data clearly confirm the previously shown interference of the transport and receptor function of NTCP 10,18 and demonstrate the functionality of this screening assay. Furthermore, these data indicate that the myr-preS1  lipopeptide has an about 100-fold higher inhibitory potency (IC 50 = 0.35 µM / 0.55 µM) in both assays compared to the physiological transport substrate TC (IC 50 = 30 µM / 42 µM). Four reference compounds were analyzed (concentrations ranging from 0.1 to 1,000 µM) which had shown inhibition of the transporter and receptor function of NTCP before, namely ciglitazone 22 , troglitazone 22 , cyclosporine A [23][24][25] , and rapamycin 26 . Interestingly, all compounds showed comparable inhibition of [ 3 H]TC uptake and [ 3 H]preS1 binding with IC 50 values of 1 µM and 1.3 µM for ciglitazone (Fig. 1c), 1.9 µM and 2.5 µM for troglitazone (Fig. 1d), 14 µM and 26 µM for cyclosporine A (Fig. 1e), as well as 34 µM and 21 µM for rapamycin (Fig. 1f), respectively. These compounds therefore can be classified as inhibitors, non-selective for one of the two measured functions of NTCP. Among them, ciglitazone and troglitazone had a tenfold higher inhibitory potency compared to cyclosporine A and rapamycin, demonstrating that diverse compounds differ in their potency as NTCP inhibitors. Cyclosporine A, rapamycin and TC showed comparable inhibitory pattern and potency in both assays, as shown by similar IC 50 values.

Screening of betulin derivatives for bile acid transport and [ 3 H]preS1 binding inhibition. Next,
30 derivatives of betulin were tested in both assays and IC 50  H]preS1 binding inhibition were calculated and are further referred to as selectivity index. The higher this index, the more selective the respective compound for myr-preS1 2-48 lipopeptide inhibition. Therefore, the data listed in Table 2 are not only helpful to identify potent myr-preS1 2-48 lipopeptide binding inhibitors but also to assess their selectivity. These values ranged from an index of 1, including more potent (e.g. 4) and less potent (e.g. 12) non-selective inhibitors, up to an index of > 60, including the highly selective and potent inhibitors 2 and 25.  www.nature.com/scientificreports/ (4, 9, 10, and 20), (II) poorly selective inhibitors (e.g. 21), and highly selective inhibitors, being 2 and 25. Also, from the group of less potent [ 3 H]preS1 binding inhibitors (IC 50 = 11-50 µM), some were poorly selective (e.g. 18) and others were highly selective (e.g. 13, 17, 19 and 30). Taken together, these data clearly indicate structure-activity relationships of the betulin derivatives for both, bile acid transport inhibition and myr-preS1 2-48 lipopeptide binding inhibition.
Most relevant modifications for structure-activity relationships. Closer analysis of the structure-activity relationships of the betulin derivatives revealed several interesting patterns that can be used to further optimize the potency and selectivity index of these compounds in order to end up with an even more suitable HBV/HDV entry inhibitor from the class of betulin derivatives. Of note, betulin (1) itself showed relatively low inhibitory potency for bile acid transport inhibition (IC 50 = 748 µM) and [ 3 H]preS1 binding inhibition (IC 50 = 109 µM) resulting in a selectivity index of 7. However, relatively moderate modifications at the betulin core structure revealed significantly improved potency and selectivity index of the individual compounds. Some of these modifications are depicted in more detail in Fig Acidic betulin derivatives are potent, but non-selective inhibitors of NTCP. All potent (IC 50 preS1 binding < 10 µM) and non-selective (selectivity index < 10) NTCP inhibitors from the group of betulin derivatives bear at least one acidic group. This applies to compounds 4, 8-12, 20, and 21 (Tab. 1). Based on their pKa values, these compounds are predominantly deprotonated and negatively charged at pH 7.4, similar to bile acids. In order to investigate the role of the pH-dependent negative charge of these compounds, the acidic derivatives 4 (pKa = 4.  (4) was used as a potent but non-selective inhibitor, 20,29-dihydrobetulin (2) as a potent and highly selective inhibitor, as well as lupenone (17) and betulonoyl dimethyl-L-aspartate (19) as less potent but still selective inhibitors. All four compounds significantly blocked in vitro HDV infection of the NTCP-HepG2 cells in a concentration-dependent manner (Fig. 5), clearly indicating that betulin derivatives represent an interesting novel compound class of HBV/HDV entry inhibitors. Of note, none of the compounds induced toxicity in the cells even after long term incubation at high (300 µM/600 µM) inhibitor concentrations (Fig. 5h). www.nature.com/scientificreports/ tion. These data confirm that also after preincubation with compound 17, NTCP remained at the plasma membrane.

Discussion
The identification of the hepatic bile acid transporter NTCP as the high-affinity receptor for HBV and HDV was the basis for the development of a novel class of anti-HBV/anti-HDV drugs, being cell-entry inhibitors targeting NTCP 9 . Since then, many studies have identified compounds that are capable to inhibit the virus receptor function of NTCP: Hepcludex 13 , cyclosporine A and other cyclosporine derivatives [23][24][25] , ezetimibe 10 , irbesartan 27 , ritonavir 28 , (-)-epigallocatechin-3-gallate 29 , vanitaracin A 30 , Ro41-5253 31 , proanthocyanidin 32 , zafirlukast 33 , sulfasalazine 33 , and Chicago Sky Blue 6B (an azo dye) 33 . Among them, Hepcludex is a peptide-based drug and cannot be administered orally 13 . Furthermore, it has correspondingly complex manufacturing, storage, and distribution requirements. Therefore, it is still desirable to develop a small molecular drug with HBV/HDV cell-entry inhibitor activity that can be orally applied and would be easier to handle. Many of the other mentioned compounds, e.g. cyclosporine A, were previously known to inhibit the bile acid transport function of NTCP 17 . Others were identified as novel HBV/HDV inhibitors and later revealed their interference with the physiological bile acid transport function of NTCP. Overall, this interference indicates that NTCP domains relevant for substrate binding and transport overlap with the domains mediating myr-preS1 2-48 lipopeptide / virus binding to NTCP. However, there are also some hints that both functions can be separated. In a previous study, we identified amino acid G158 of the human NTCP as absolutely essential for myr-preS1 2-48 lipopeptide binding and in vitro HBV and HDV infection. This amino acid is located at a domain (amino acids 157-165) that previously was shown to be involved in myr-preS1 2-48 binding to human NTCP 9,18 . Accordingly, G158R NTCP mutants were completely insusceptible for in vitro HBV/HDV infection, but still supported transport of bile acids. It was discussed that the larger amino acid side chain of arginine compared to glycine might sterically preclude myr-preS1 peptide binding, while bile acids can still bind to their respective binding pocket 34 . Furthermore, Shimura et al. 25 showed for the cyclosporine derivative SCY995 higher potency against in vitro HBV infection (IC 50 < 8 µM) than for bile acid transport inhibition (IC 50 > 25 µM).
Based on this, the present study was designed to enable the identification of a selective HBV/HDV cell-entry inhibitor from the group of small molecules. Therefore, all test compounds were a priori tested for both, inhibition of [ 3 H]preS1 binding to NTCP as a surrogate parameter for HBV/HDV infection and inhibition of [ 3 H]TC transport via NTCP. In the present study, a novel chemical class of promising HBV/HDV cell-entry inhibitors was identified, represented by derivatives of the natural pentacyclic lupane-type triterpenoid betulin. Betulin can be isolated from various plant species, belonging to a variety of families such as Betulaceae, Platanaceae, Dilleniaceae, Rhamnaceae, Rosaceae, and Fagaceae 35,36 . Characteristics of the lupane skeleton of betulin are the five-membered E ring and the isopropylidene group 37 . Interestingly, nearly all betulin derivatives tested in the present study significantly inhibited Inhibition studies at pH 5.5 revealed additional structure-activity relationships regarding the charge of the tested compounds. All potent but non-selective inhibitors (Fig. 3b) bear at least one acidic group at R1 and/or R2 (see Tab. 1) and show pKa values between 4-5. Therefore, compounds from this group are predominantly deprotonated and negatively charged at physiological pH of 7.4, representing the standard condition in the inhibition assays. For comparison, also conjugated bile acids (pKa = 1-4) are predominantly deprotonated and negatively charged at physiological pH [38][39][40] . Based on this it can be speculated that acidic betulin derivatives may not only be

H]preS1 binding inhibition) and virus selectivity index (sel. ind., ratio IC 50 [ 3 H]TC transport inhibition / IC 50 [ 3 H]preS1 binding inhibition). (a) 3,28-di-O-acetylation (shown in orange), additional C 29 -hydroxylation (shown in purple), 3-O-modification with caffeoyl (shown in lightblue), 28-oxime generation (shown in red)
, and C 30 saturation (shown in green) of the betulin core structure improve the potency and/or selectivity index of the respective compounds. The respective inhibition pattern are illustrated by bar graphs for [ 3 H]TC uptake (shown in dark-blue) and [ 3 H]preS1 peptide binding (shown in red), respectively, in the presence or absence of a fix concentration (100 µM) of the respective betulin derivative. Graphs show means ± SD of quadruplicate determinations. *Significantly different with p < 0.01 (two-way ANOVA with Sidak's multiple comparison test); ns, not significant. (b) The diagram shows all IC 50 values for [ 3 H]TC uptake inhibition and [ 3 H]preS1 binding inhibition of the respective compounds. Effects (IC 50 shifts) of the structural modifications are illustrated by arrows, starting from the core structure betulin (compound 1). Compounds can be divided into three groups regarding their potency to inhibit [ 3 H]preS1 binding: potent inhibitors (IC 50 ≤ 10 µM), less potent inhibitors (10 µM < IC 50 < 100 µM) and poor inhibitors (IC 50 ≥ 100 µM). Compounds with a selectivity index of > 10 are classified as selective inhibitors. www.nature.com/scientificreports/ inhibitors but also transporter substrates of NTCP. This would explain, why these compounds (representatively shown for compounds 4 and 8) largely lost their [ 3 H]TC uptake inhibitory potency when the pH was reduced to 5.5 and, thereby reducing the relative charge of the molecules. In contrast, the [ 3 H]preS1 binding inhibitory potency was insensitive to this pH shift and only showed little differences between pH 7.4 and pH 5.5 conditions. This again indicates that the transporter and the receptor function of NTCP can be separated under certain experimental conditions and that blocking of the preS1 peptide binding site of NTCP does not require negative charge of the inhibitor. In contrast to compounds 4 and 8, betulin itself (compound 1) does not bear any acidic group. Consequently, betulin could be less active at the bile acid binding site, what would explain the relatively low potency for TC uptake inhibition (IC 50 = 748 µM) at neutral and acidic pH values. However, no uniform legality could be established for the other chemical modifications of betulin, which would precisely predict the potency and selectivity pattern of the respective betulin derivative. Nevertheless, based on the structure-activity relationships described in the present study it is most likely that even more potent betulin derivatives can be identified, while sparing the cross-reactivity on the bile acid transport function of NTCP. Among the betulin derivatives, betulinic acid (4) particularly underwent intensive toxicological assessment 35,[41][42][43] . In vitro and in vivo model systems consistently categorized betulinic acid (4) as a safe compound with low toxicity. More precisely, no toxic effects were observed in rats that obtained single i.p. application of 400 mg/kg or in mice receiving 100 mg/kg i.p. every 3-4 days for a total of six treatments, or 500 mg/kg i.p. once 35,42,43 . Unfortunately, betulinic acid (4), although potent in [ 3 H]preS1 binding inhibition (IC 50 = 2 µM), is not selective and showed equipotent inhibition of the bile acid transport function of NTCP at least at physiological pH. Therefore, betulinic acid (4) is not regarded as the optimal HBV/HDV entry inhibitor candidate from the betulin class of compounds. Apart from betulinic acid (4), betulin (1) also has a good safety profile. Betulin (1) is approved for medical use in humans and is currently available as a betulin containing gel (Oleogel-S10, Episalvan). It was authorized in 2016 by the European Medicines Agency (EMA) for treatment of partial thickness wounds in adults. This gel contains an extract from birch bark of Betula pendula Roth, Betula pubescens Ehrh., and hybrids of both species, obtained with n-heptane as the extraction solvent. The resulting betulin (1) concentrations are in the range of 72-88 mg betulin (1) per 100 mg extract. The gel itself is composed of 10% birch bark extract in 90% sunflower oil. Moreover, Episalvan is currently undergoing a phase III efficacy and safety study in patients with inherited epidermolysis bullosa (EB) that started in 2017 with an estimated enrollment of 250 participants (NCT03068780). Based on this good safety profile and the established use in human medicine, betulin (1) is an attractive early lead candidate for selective HBV/HDV cell-entry inhibition. However, despite its superior selectivity in comparison to betulinic acid (4), the potency for [ 3 H]preS1 binding inhibition is relatively www.nature.com/scientificreports/ low (IC 50 = 109 µM). Nevertheless, the present study clearly demonstrates that even small modifications of the lupane skeleton of betulin can improve the potency and selectivity of the respective compounds significantly. Therefore, it seems likely that advanced rational drug design can further improve this compound class to end up with a potent and selective HBV/HDV cell-entry inhibitor. Apart from the above-mentioned clinical application, several previous studies showed that betulin (1) and betulin derivatives have a broad spectrum of pharmacological effects, including anti-cancer, anti-viral, antimicrobial, anti-inflammatory, and anti-fibrotic effects [44][45][46][47][48] . Interestingly, betulinic acid (4) was also used in HBV infection studies before, which showed inhibition of HBV replication by suppression of manganese superoxide dismutase (SOD2) expression with subsequent mitochondrial reactive oxygen species (ROS) over-generation 49 . Based on this and the results from the present study, betulinic acid (4) might have additive HBV/HDV antiviral properties by addressing different cellular targets involved in cell-entry (NTCP) and replication (SOD2). Therefore, further studies with betulin derivatives should take selective NTCP inhibition and potential suppression of SOD2 into account, which was beyond the scope of the present study.
Our [ 3 H]preS1 binding assay successfully predicted the inhibitory effect of the betulin derivatives on in vitro HDV infection. A minor but expected limitation is that absolute IC 50 values varied between both experimental setups. While the qualitative activity pattern of the betulin derivatives remained the same between the two assays, the absolute IC 50 values for in vitro HDV infection inhibition were about one to two orders of magnitude higher than predicted from the [ 3 H]preS1 binding experiments. Most likely, several factors contribute to this effect. One factor could be the different incubation times for [ 3 H]preS1 binding (minutes) and in vitro HDV infection experiments (hours). Another factor might be the use of different cell lines and incubation media. Whereas, the [ 3 H]preS1 binding assays were performed in NTCP-HEK293 cells in pure DMEM, the infection experiments needed to be performed with NTCP-HepG2 cells in HGM containing significant amounts of bovine serum albumin (BSA) as well as polyethylene glycol (PEG). The latter medium conditions potentially alter free drug concentrations due to protein binding. Lowering of the inhibitory potency of betulinic acid (4) due to protein binding has already been identified in a previous study, where this compound was used as inhibitor for the organic anion transporting polypeptide OATP1B3 50 . A further factor could be the different stoichiometry of the single preS1 peptide molecule and the multitude of preS1 domains of the large envelope proteins of HBV/HDV virions. The latter one could need a higher degree of blocked NTCP receptors for successful virus binding and cell-entry inhibition. On this background, it could be beneficial if an NTCP inhibitor would not only be selective, but would also irreversibly block the preS1/virus binding sites at NTCP. However, this was not achieved with the betulin derivatives analyzed in the present study. The data obtained on compounds 4 and 17 more suggest a reversible mode of inhibition for myr-preS1 2-48 peptide binding to NTCP. Their inhibitory effect on [ 3 H]preS1 binding was completely abrogated by washing the cell surface with inhibitor-free medium (see Fig. 5e,f). But, these observations clearly indicate that the inhibitory effects of compounds 4 and 17 on HDV infection cannot be attributed to drug-induced internalization of NTCP.
Of interest are also the long-term effects of compounds 4 and 17 on the transporter and receptor function of NTCP. In the case of compound 4 preincubation with 600 µM significantly and selectively reduced the transporter function of NTCP, while NTCP still was completely preS1 peptide binding competent. As compound 4 (betulinic acid) belongs to the group of acidic betulin derivatives one can speculate that betulinic acid is transported by NTCP, accumulates inside the cell during preincubation and so by trans-inhibiton blocks substrate binding but not preS1 peptide binding to the outer surface of the NTCP protein. This again would support the idea that bile acid and preS1 peptide binding to NTCP occur at separate, but overlapping domains. In contrast, compound 17 showed a paradoxical effect after long term preincubation and increased the capacity for [ 3 H]preS1 peptide binding to approximately 130%, while the [ 3 H]TC uptake rate was slightly reduced. This could be explained by longer lasting allosteric effects of this compound that differently affects the transport and receptor function of NTCP. The mechanism behind these effects could not be clarified in the present study and needs further investigation, including LC-MS/MS supported direct transport experiments with key betulin derivatives.
In conclusion, betulin derivatives show clear structure-activity relationships for potent and selective inhibition of the HBV/HDV virus receptor function of NTCP without tackling its physiological bile acid transport function. Therefore, betulin derivatives are promising candidates for further development of HBV/HDV cellentry inhibitors. Further studies with this compound class should additionally focus on the mode of inhibition (reversible vs. irreversible) and should be extended to all relevant HBV genotypes.

NTCP-expressing cell lines. Human embryonic kidney (HEK293) cells were stably transfected with
human NTCP, C-terminally tagged with the FLAG epitope (further referred to as NTCP-HEK293 cells) as reported before 51 . Cells were maintained at 37 °C, 5% CO 2 and 95% humidity in DMEM/F-12 medium (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% fetal calf serum (Sigma-Aldrich, St. Louis, MO, USA), 4 mM L-glutamine (PAA, Cölbe, Germany) and penicillin/streptomycin (PAA). HepG2 cells stably transfected with NTCP-FLAG (further referred to as NTCP-HepG2 10 ) were cultured under the same conditions in DMEM with all supplements listed above, except for L-glutamine. For induction of the transgene, the medium was supplemented with 1 µg/ml tetracycline (Roth, Karlsruhe, Germany) in the case of the NTCP-HEK293 cells or with 2 µg/ml doxycycline (Sigma-Aldrich) in the case of the NTCP-HepG2 cells.  (4)
Preincubation studies in NTCP-HepG2 cells. NTCP-expressing HepG2 hepatoma cells were seeded onto rat tail collagen coated 24-well plates, induced with 2 µg doxycycline per ml, and grown to confluence over 72 h at 37 °C. For experiments, cells were preincubated with a fix concentration (600 µM) of compound 4 and