Development of a novel anti-hepatitis B virus agent via Sp1

Nucleos(t)ide analog (NA) therapy has proven effective in treating chronic hepatitis B. However, NAs frequently result in viral relapse after the cessation of therapy. This is because NAs cannot fully eliminate the viral episomal covalently closed circular DNA (cccDNA) in the nucleus. In this study, we identified small molecular compounds that control host factors related to viral replication using in silico screening with simulated annealing based on bioinformatics for protein-ligand flexible docking. Twelve chemical compound candidates for alpha-glucosidase (AG) inhibitors were identified from a library of chemical compounds and used to treat fresh human hepatocytes infected with HBV. They were then monitored for their anti-viral effects. HBV replication was inhibited by one candidate (1-[3-(4-tert-butylcyclohexyl)oxy-2-hydroxypropyl]-2,2,6,6-tetramethylpiperidin-4-ol) in a dose-dependent manner. This compound significantly reduced ccc DNA production, compared to Entecavir (p < 0.05), and had a lower anti-AG effect. Gene expression analysis and structural analysis of this compound showed that its inhibitive effect on HBV was via interaction with Sp1. The nuclear transcription factor Sp1 acts on multiple regions of HBV to suppress HBV replication. Identifying candidates that control nuclear transcription factors facilitate the development of novel therapies. Drugs with a mechanism different from NA are promising for the elimination of HBV.

Screening for anti-HBV activity and cell toxicity of human alpha-glucosidase inhibitor candidates. PXB cells infected with HBV were used to observe anti-viral activity and cell toxicity. AGI1-AGI3 and AGI4-AGI15 were dissolved in H 2 O and DMSO, respectively. AGI10 was not used in this study because it was insoluble in DMSO and H 2 O. Entecavir (ETV) was used as a positive control for the anti-HBV drug. AGI1-AGI3 showed no observable anti-viral effect or cell toxicity ( Supplementary Fig. 2). AGI5, AGI6, AGI7, AGI13, and AGI14 reduced the concentration of HBV-DNA in the supernatant in a dose-dependent manner ( Fig. 1 and Supplementary Table 1). AGI14 reduced HBsAg in the supernatant in a dose-dependent manner; however, there was no noticeable reduction following treatment with AGI5, AGI6, AGI7, and AGI13 ( Fig. 2A). There was also no noticeable change in human albumin concentration in the supernatant following treatment with these five AGI candidates (Fig. 2B). The IC 50 of AGI5, AGI6, AGI7, AGI13, and AGI14 were 4.9 μM, 6.9 μM, 1.9 μM, 3.8 μM, and 296 nM, respectively (Fig. 2C). Based on the value of their IC 50 , AGI7 and AGI14 were selected for analysis of anti-cccDNA activity. The amount of cccDNA remaining after treatment with high concentrations of AGI14 without cell toxicity was significantly lower than that treated with ETV (Fig. 2D). Multiple experiments were conducted at different times, as described in the figure legends, confirming the reproducibility of the experimental results.
AG activity in novel alpha-glucosidase inhibitor candidates. The ability of five novel candidates to inhibit alpha-glucosidase was verified. It was observed that AGI5, AGI6, and AGI13 mildly suppressed the activity of alpha-glucosidase in a dose-dependent manner, and conversely, the suppressive effect of AGI7 and AGI14 increased in a dose-dependent manner ( Supplementary Fig. 3).
investigating the mechanism of the alpha-glucosidase inhibitor candidates on HBV replication. Since the novel candidates had little inhibitory effect on alpha-glucosidase, an alternative mechanism of the anti-viral activity of these compounds was investigated. Gene expression patterns among non-treated cells, AGI7-treated cells, and AGI14-treated cells were compared using a next-generation sequencer (NGS). Principal component analysis (PCA) showed that 194 and 208 genes in PXB cells were commonly increased and decreased when treated with AGI7 and AGI14, respectively (Fig. 3A) 16 . To clarify the function of these genes, a G-profiler was used to analyze which regulatory DNA elements were controlled by gene clusters with common expression patterns (https://biit.cs.ut.ee/gprofiler/index.cgi).
The G-profiler showed that the GGCGSG motifs of the two E2F genes (M00803_0) and (M00803_1) were regulated by 160 and 140 of the 194 genes that were upregulated, respectively. The GGCGGGN motif of the E2F3 genes (M02089_1) was also regulated by 129 of the 208 genes that were downregulated. The NGGGGGCGGGGYN motif of Sp1 (M00196_0) was also regulated by 147 of the commonly downregulated genes (Supplementary Table 2). Since many genes recognize the promoter region of E2F3 and Sp1, it was concluded that AGI7 and AGI14 controlled viral replication via E2F3 or Sp1 (Fig. 3B).
The interaction of AGI7 and/or AGI14 with Sp1. Sp1 has a binding motif for E2F1 (amino acids 102-125) contained within amino acids 622-668. Both E2F2 and E2F3 in the N-terminal domains, which have sequences similar to E2F1, interacted with Sp1 17,18 . Based on the gene expression and G-profiler analysis, AGI7 and AGI14 were hypothesized to have a binding affinity for Sp1. Therefore, a structural analysis was performed to confirm how AGI7 and AGI14 interact with Sp1. Using the LD method for AGI14 construction, docking between AGI14 and Sp1 was interfaced with two glycerol fingerprints 19 . However, while AGI7 could also be docked to the Sp1 protein via two molecules of glycerol, the binding site was different from that of AGI14 (Fig. 4).
The anti-viral effect of siRNA for Sp1 and E2F3. We investigated whether siRNA that targets Sp1 and E2F3 controlled HBV replication in Hep38.7 cells. Due to a higher gene-introduction efficiency, a gene knockdown experiment was performed with Hep38.7 cells instead of PXB cells. It was observed that AGI7 and AGI14 did not inhibit HBV replication in Hep38.7 cells (data not shown). Administering siRNA for Sp1 significantly inhibited HBV replication; however, siRNA for E2F3 did not show inhibition of viral replication ( Supplementary Fig. 4).  www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
In this study, novel anti-viral agents that potentially control replication intermediates via nuclear transcription factors were identified. Alpha-glucosidase inhibitors are standard drugs for treating diabetes; they also exhibit the potential to act against HBV [2][3][4][5] . The novel alpha-glucosidase inhibitor candidates exhibited little inhibition of alpha-glucosidase. Assuming that they had different mechanisms for inhibiting viral replication, how a novel candidate modified gene expression was clarified. Interestingly, most genes that had a similar expression pattern after treatment with AGI7 and AGI14 had a binding affinity with the Sp1 and E2F transcription factor families (E2F1, E2F2, and E2F3). SP1 and E2F families bind to each other when acting as transcription factors 17 . Structural analysis revealed that AGI7 and AGI14 had a high affinity for Sp1, and HBV replication was suppressed, even in experiments using siRNA.
In Hep38.7 cells, AGI7 and AGI14 did not exhibit anti-viral effects (data not shown) due to the difference between PXB and Hep38.7 cells. Since PXB cells do not require passage, the effect of candidate drugs using PXB cells  www.nature.com/scientificreports www.nature.com/scientificreports/  www.nature.com/scientificreports www.nature.com/scientificreports/ was determined after 22 days, and therefore, they are suitable for relatively long-term studies; however, Hep38.7 cells require passage once every three days, and therefore, they are only suitable for short-term observations of anti-viral effects. The lack of passaging of PXB cells indicated that gene transfer efficiency using liposomes was very poor; therefore, experiments with siRNA in PXB cells failed to reproduce the experimental results of Hep38.7 cells.
The expression of HBV genes was regulated by several transcription factors. Sp1-binding sites in the HBV core promoter are important in regulating the transcription of the core and precore RNA 20 . An HBV-transgenic mouse study showed that cyclin D2 was upregulated in HBV-expressing cells and liver tissues. Cyclin D2 regulated HBV replication by enhancing the activity of HBV core and Sp1 promoters by targeting the transcription factor CREB2 14 . HBx upregulates C4b-binding protein α (C4BPα) by activating transcription factor Sp1, and this protects liver cancer cells from a complement attack 21 . HBx 43-154 upregulates the activity of HBV enhancer II; moreover, CCAAT/Enhancer Binding Protein (C/EBP) and Sp1 sites on enhancer II are required for enhancer II activation by HBx 43-154 15 . Association between HBx mutated HBV, and several chromatin-modifying enzymes also influenced nuclear cccDNA 22 . HBx is essential for the initiation and maintenance of transcription from cccDNA. Therefore, HBx expression levels exactly reflect HBV RNA transcription 23 .
Taken together, AGI14 bound to Sp1 and suppressed its function. Sp1 is involved in HBV replication at multiple points 15,20 . Downregulated Sp1 suppressed HBV replication via the downregulation of the HBV core promoter; the absence or silence of Sp1 suppressed HBx activity then induced the low stability of HBV episomal nuclear localization, and finally, downregulated cccDNA (Fig. 5).

conclusion
This study uncovered a novel class of anti-HBV agents with high anti-viral effects and low cell toxicities. These compounds did not directly target cccDNA; instead, they regulated cccDNA via nuclear transcription factors. Controlling cccDNA, which is the cause of latent infection, requires promising drugs with a low recurrence rate after the patient has completed treatment.

Methods
In silico screening and chemical compounds. In silico screening to develop a novel human α-glucosidase inhibitor has previously been reported 24,25 , except that here Q14697.3 in UniprotKB/Swissprot was used to analyze the amino acid sequence in the alpha-glucosidase inhibitor. A chain of PDB (protein data bank) ID 3LU4 was selected as the reference protein based on the alignment results. In total, 13 molecules were used as the fingerprints for in silico screening with chooseLD. The models constructed by the full automatic modeling system  www.nature.com/scientificreports www.nature.com/scientificreports/ is registered with the ChEMBL database. Several low weight molecules were selected from the AKos database in order to fit alpha-glucosidase, the details of which have been described in previous papers 26,27 . Entecavir was purchased from Toronto Research Chemicals (Toronto, ON); AG1-AG3 and AG4-AG15 compounds were purchased from Tokyo Chemical industry (Tokyo, Japan) and AKos (Steinen, Germany), respectively (Table 1).
In vitro HBV infection. The HBV-infected serum sample (genotype C) was obtained from a 54-year-old male patient. He was negative for HIV and HCV. The patient provided written informed consent, and Osaka City University Graduate School and the Faculty of Medicine's Ethics Committee approved the study in accordance with the Helsinki Declaration 2013.
The transfection procedure, DNA extraction, quantification of HBV-DNA, HBsAg, and albumin have all been described previously 26,27 . The HBV-DNA in the sample medium was quantified by real-time qPCR (Roche Diagnostic, Tokyo, Japan) by comparing serially diluted HBV/C1.24 (HBV-DNA containing plasmid) obtained from Prof. Yasuhito Tanaka of Nagoya City University 28 .
The quantification of cccDNA. Real-time PCR was performed with Step one plus (Applied Biosystems) using 10 μl of the sample and 20 μl of PCR mixture (Roche MasterMix, Roche Diagnostics, Almere, The Netherlands). The following target probes were used: 5′-6FAM-CGTCGCATGGARACCACCGTGAACGCC-BHQ1-3′ and IC-DNA probe: 5′-TBRCCCTTTACATCTTTCTGAAGTAGGG-3′ 29 . Primer concentrations were 0.9 μM for the target and IC-DNA primers. Probe concentrations were 0.4 μM for cccDNA and 0.2 μM for IC. Amplification was performed as follows: 50 °C for 2 min, then 95 °C for 10 min, followed by 55 cycles at 95 °C for 10 sec, 58 °C for 5 sec, 63 °C for 10 sec, and 72 °C for 20 sec. Every run included a negative plasma sample, water, and a positive control with a known concentration of cccDNA (10 3 copies/PCR) 30 .
Alpha D-glucosidase activity assay. Where presented, IC 50 values were determined by a non-linear, least squares regression analysis using MathIQ TM (Business Solutions Ltd., UK). Where inhibition constants (Ki) are presented, the Ki values were calculated using the equation of Cheng and Prusoff 31 using the observed IC 50 of the tested compound, the concentration of radioligand employed in the assay, and the historical values for the K D of the ligand (obtained experimentally at Eurofins Panlabs, Inc.). Where presented, the Hill coefficient (n H ), defining the slope of the competitive binding curve, was calculated using MathIQ TM . Hill coefficient values significantly greater than 1.0 may suggest that the binding displacement did not follow the laws of mass action with a single binding site. Where IC 50 , Ki, and/or n H data are presented without a standard error of the mean (SEM), the data were insufficient to be quantitative, and the values presented (Ki, IC 50 , n H ) should be interpreted with caution 32 .
Next-generating sequencing. HBV infection and alpha-glucosidase inhibitor treatment were performed in PXB cells at the same time. The medium was changed on days one and two, and PXB cells were harvested on day seven. Total RNA was extracted using a miRNeasy Mini kit (Qiagen, Hilden, Germany). Extracted RNA from Figure 5. Summary of the anti-HBV effect of Sp1. AGI14 binds to Sp1 (see figure legend 4C), and as a result, Sp1 activity decreases, and HBV enhancer II initially increases in the presence of Sp1 and enhances HBV replication. However, it is assumed that HBV replication decreases due to decreased Sp1 activity. Moreover, when the promoter activity of Sp1 decreases, the activity of the HBV core promoter decreases, and it is expected that HBV replication will also decrease. Although the direct effect on cccDNA is unknown at this time, the possibility of reduced production of cccDNA with reduced viral replication capacity is also indicated.