Activity of vitamin D receptor agonists against dengue virus

Infections with the mosquito-transmitted dengue virus (DENV) are a pressing public health problem in many parts of the world. The recently released commercial vaccine for DENV has encountered some problems, and there is still no effective drug to treat infections. Vitamin D has a well characterized role in calcium and phosphorus homeostasis, but additionally has a role in the immune response to bacterial and viral pathogens. In this study a number of fused bicyclic derivatives of 1H-pyrrolo[1,2]imidazol-1-one with vitamin D receptor (VDR) agonist activity were evaluated for possible anti-DENV activity. The results showed that five of the compounds were able to significantly inhibit DENV infection. The most effective compound, ZD-3, had an EC50 value of 7.47 μM and a selective index of 52.75. The compounds were only effective when used as a post-infection treatment and treatment significantly reduced levels of infection, virus output, DENV protein expression and genome copy number. These results suggest that these VDR agonists have the potential for future development as effective anti-DENV agents.

Scientific RepoRtS | (2020) 10:10835 | https://doi.org/10.1038/s41598-020-67783-z www.nature.com/scientificreports/ C virus, influenza virus and HIV [15][16][17] . However, the association between vitamin D/VDR and dengue infection is not completely understood. However, it has been shown that there is a relationship between vitamin D levels and VDR polymorphism and the severity of DENV clinical manifestation 18,19 . Treatment of DENV infected monocytic U937 cells or hepatic Huh-7 cells with 1,25(OH) 2 D 3 resulted in decreased numbers of infected cells, reduced Toll-like receptors and lowered inflammatory cytokines 20 . Another study demonstrated that the presence of 1,25-dihydroxyvitamin D3 during macrophage proliferation restricted DENV infection and altered the proinflammatory cytokine response through reducing the expression of the C-type lectin mannose receptor, a DENV receptor protein 21 . In a recent study a novel class of VDR agonists were described 22 and this study sought to determine if these compounds had antiviral effects.

Results
Evaluation of cytotoxicity of VDR agonists. Prior to determining possible anti-DENV activity of the newly reported VDR agonists 22 , the cytotoxicity of the compounds to HEK293T/17 cells was determined by trypan blue staining and by the MTT assay. Additionally cell morphology was evaluated by observation under an inverted microscope. The trypan blue exclusion assay showed little cytotoxicity at concentrations up to 200 μM (Supplemental Figure S1A), while the MTT assay showed a dose dependent cytotoxicity (Supplemental Figure S1B). The calculated CC 50 values are shown in Table 1. Additionally observation of cell morphology showed signs of morphological changes at higher concentrations (Supplemental Figure S2). Based on all the data, further VDR agonist treatments were carried out with a concentration of 10 μM. Next, the effect of the seven VDR agonists on DENV infection was evaluated. HEK293T/17 cells were mockinfected or infected with DENV 2, followed by the incubation with 10 µM VDR agonists. After 24 h of treatment, the level of infection was determined by flow cytometry. Results (Fig. 1) showed that ZD-1, ZD-2, ZD-3, ZD-5, and ZD-6 treated cells showed highly significant reductions in the level of infection. For the remaining two compounds, treatment with ZD-4 resulted in infection levels being reduced by approximately 50%, while treatment with ZD-20 showed no significant reduction in the level of infection as compared to the control of DENV 2 infected cells treated with 0.01% DMSO control (Fig. 1A). Evaluation of the DENV 2 titer in the supernatant by standard plaque assay showed that treatment with ZD-1, ZD-2, ZD-3, ZD-5, and ZD-6 reduced virus production  In parallel, the supernatants of the treated cells were evaluated for virus titer by standard plaque assay. p value; *< 0.05, **< 0.01, ***< 0.001 for significance. The experiment was performed independently in triplicate with duplicate in plaque assay.
Scientific RepoRtS | (2020) 10:10835 | https://doi.org/10.1038/s41598-020-67783-z www.nature.com/scientificreports/ by 2-3Log 10 , while treatment with ZD-4 reduced virus output by slightly over 1Log 10 . Again, no significant effect was observed with treatment with ZD-20 (Fig. 1B). Overall there was consistency in the reductions seen in level of infection and final virus titer for all compounds. Given the significant reductions seen in virus production for the majority of compounds, we next determined the half-maximum effective concentration (EC 50 ) and selective index (SI) for all compounds. HEK293T/17 cells were mock-infected or infected with DENV 2, followed by incubation with various concentrations (0.1 nM to 10 µM for ZD-1,-2,-3,-5, and -6, and or 1-100 µM for ZD-4 and -20) of all VDR agonists. At 24 h post-treatment, virus titer in the supernatant was determined by plaque assay. Results (Supplemental Figure S3 and Table 1) showed that the EC 50 values ranged from 7.47 μM (ZD-3) to 151.72 μM (ZD-20) and the selective index values ranged from 52.75 (ZD-3) to 5.98 (ZD-20). ZD-1, ZD-2, ZD-3, ZD-5, and ZD-6 which showed the greatest antiviral effects also showed the highest SI values (Table 1). These five VDR agonists were therefore selected for further investigation. evaluation of virucidal activity of VDR agonists. To determine whether the compounds possessed direct virucidal activity, stock DENV 2 was directly incubated with 10 µM of VDR agonists at 37 °C for 1 h, after which the titer was determined by standard plaque assay. Results (Supplemental Figure S4) showed no reduction of virus titer after the incubation compared to the DMSO control, showing that the VDR agonists have no direct virucidal activity.
Effect of time-of-additional of VDR agonists. To assess the effect of VDR agonist treatment at a different time points during infection, both pre-and post-infection treatment was evaluated. In pre-treatment cells were treated with 10 µM of the five most effective VDR agonists (ZD-1, ZD-2, ZD-3, ZD-5, and ZD-6) for 6, 3 or 1 h before infection, and after 24 h both the level of infection and the virus titer in the supernatant were determined by flow cytometry and standard plaque assay respectively. In addition, cells were collected at the end of the experiment for subsequent protein isolation, while the remaining supernatant was used to determine genome copy number. For post-infection treatment, cells were infected with DENV 2 before being treated with the 5 VDR agonists at 0, 1, 3, 6 and 12 h post-infection (with 0 h representing treatment with the compounds immediately after the end of the infection step), and at 24 h post-infection the level of infection and the virus titer in the supernatant were again determined by flow cytometry and standard plaque assay respectively. Again, cells were collected at the end of the experiment for subsequent protein isolation, while the remaining supernatant was used to determine genome copy number. The results (Fig. 2) showed that the compounds had no effect when administered pre-infection on either level of infection or virus titer in the supernatant. Post-infection treatment of the compounds exerted considerable effects on both the level of infection and on the virus titer ( Fig. 2), with post-infection treatment resulting in highly significant reductions in both level of infection and virus titer. The largest effect was observed with treatment immediately after the infection step, and for up to 3 h post-infection, however, significant reductions in both the level of infection and virus titer were observed at 6 and 12 h post-infection treatment with compounds ZD-1 and ZD-5 (Fig. 2).
Proteins from the post-treatment experiments were used to determine the expression level of DENV 2 structural (E protein) and non-structural (NS1, NS3, and NS5) proteins by western blotting while RNA from the supernatant was used to establish DENV genome copy number by real-time RT-PCR. Results (Fig. 3) showed that expression of both structural and non-structural proteins was largely completely absent when treatment occurred at an early time point in infection, although reduced protein expression could be observed in cells treated with the VDR agonists as long as 12 h post-infection. Similarly, while large reductions were seen in the level of DENV genome in the supernatant (consistent with the plaque assay results), smaller, but still statistically significant results were seen in cells treated with all VDR agonists as late as 12 h post-infection (Fig. 3F).
VDR agonist activity as assessed by an immunofluorescence assay. To confirm the results of the earlier experiments, an immunofluorescence assay was performed. Mock infected and DENV 2 infected HEK293T/17 cells grown on coverslips were treated with 10 µM of the five VDR agonists for 24 hpi, after which the expression of DENV E protein and VDR detected after incubation with specific antibodies and examination under a confocal microscope. The results (Fig. 4) were consistent with the earlier observations, with DENV E protein being largely undetectable after treatment with the VDR agonists. Interestingly, cells infected with DENV 2 showed a markedly higher signal for VDR than mock infected cell, and apparently higher expression was observed in VDR agonist treated cells (Fig. 4). It is known that the binding of a ligand to the VDR can induce translocation of VDR into the nucleus, and the results appeared to be consistent with the increased localization of VDR to the nucleus in agonist treated cells. The co-localization between VDR and nuclear staining DAPI was captured using Airyscan with comparison between mock-infected, DENV 2 infected and DENV 2 infected and ZD-6 treated cells (Supplemental movie 1A-C). In combination the results support that the VDR agonists can induce the expression of VDR and also activate its nuclear translocation.
Evaluation of cell type specificity. Given that HEK293T/17 cells are not representative of a DENV target tissue, the antiviral activity of the VDR agonists was evaluated in cells representative of a primary target cell. For this reason, HepG2 cells were infected with DENV 2, followed by treatment with 10 μM of the five effective VDR agonists (ZD-1, ZD-2, ZD-3, ZD-5, and ZD-6) for 24 h, after which the level of infection was determined by flow cytometry. Results (Fig. 5) showed a significant reduction in infection for all compounds tested, albeit that the reduction seen with ZD-1 was modest as compared to the other four compounds. Agonist activity of VDR agonists. Upon translocation to the nucleus the VDR mediates gene transcription to induce the expression of CYP24A1 that functions to degrade active vitamin D, while suppressing the expression of CYP27B1 which functions to synthesize active vitamin D 23 . We therefore investigated the transcriptional response of VDR, CYP24A1 and CYP27B1 to the five VDR agonists investigated in this study. Results show that four of the five VDR agonists significantly increased expression of VDR, and all reduced expression of CYP27B1 (Fig. 7). Interestingly, all agonists also significantly suppressed expression of CYP24A1 (Fig. 7), in contrast to the natural action of vitamin D.
evaluation of a commercial VDR agonist. We additionally evaluated the activity of a commercial vitamin D analog, EB1089. HEK293T/17 cells were infected with DENV 2, followed by treatment with 20 μM EB1089. At 24 h.p.i., the treated cells and supernatant were collected to examine the levels of infected cell and the viral production, respectively. Treatment of EB1089 significantly reduced the levels of both infected cells and viral yield (Fig. 8A,B). Viral production was reduced by approximately 2Log 10 , consistent with the activity seen for the VDR agonists examined in this study. www.nature.com/scientificreports/ In addition, the regulation of VDR-mediated gene expression was evaluated by the incubation of HEK293T/17 cells with 20 µM EB1089 for 24 h. The relative gene expression was examined determined by qRT-PCR and normalized against a house keeping gene (actin) and a DMSO control. The transcriptional levels of VDR, CYP27B1 and CYP24A1 were significantly up regulated after treatment with EB1089 (Fig. 8C). As expected, EB1089 which an analog of 1,25-dihydroxyvitamin D3, successfully mimicked the action of active vitamin D3 by inducing the expression of CYP24A1, in contrast to the lack of induction of this gene seen with the VDR agonists used in this study.

Discussion
Finding specific and efficient anti-viral drugs is proving to be a challenge for DENV researchers. Host nutritional status and nutritional supplementation such as vitamin D, is one possible strategy for anti-DENV drug development 14 . Vitamin D exerts its activity through binding to VDR. In addition to having a role in calcium homeostasis 8 , vitamin D also plays a role in several biological systems, including the immune response and pathogen-defense mechanisms 10     www.nature.com/scientificreports/ A recent study identified fused bicyclic derivatives of 1H-pyrrolo [1,2]imidazol-1-one as potent regulators of VDR signaling 22 , and in this study a number of these compounds were shown to act as potent inhibitors of DENV infection. In a previous study it was proposed that vitamin D exerted its antiviral activity at the step of viral entry, as human macrophages differentiated in the presence of vitamin D showed reduced expression of the DENV receptor protein the C-type lectin mannose receptor 21 . However, this is not consistent with the results seen in this study as the most significant effect was seen with the compounds added immediately after the infection step, and for up to three hours post infection, while pre-treatment of cells with the VDR agonists had no effect on infection.  www.nature.com/scientificreports/ In monocytes, vitamin D has been proposed to regulate slightly less than 200 different genes to different extents 28 , and it is likely that similar numbers of genes are regulated in different cell types. Thus, knowing how the VDR agonists are specifically exerting their antiviral effect is going to be difficult. In their study Xu et al. 22 showed that the VDR agonists when administered together with calcitrol activated p62 expression as well as changed the LC3-II/LC3-I ratio, suggestive of the induction of autophagy. However, autophagy in DENV infection is generally proviral [29][30][31] , and as such the agonists are unlikely to be exerting their effect through induction of autophagy.
Previously Hass et al. 32 have shown that Vitamin D can inhibit endoplasmic and oxidative stress 32 . Several studies have shown that ER stress is activated during DENV infection [33][34][35] , and studies have suggested that this activation is essential for DENV replication 33,35 . Hass et al. 32 additionally demonstrated that vitamin D could down regulate the ER stress induced up-regulation of GRP78. Several studies have shown the up-regulation of GRP78 in DENV infection [36][37][38] , and that this protein is essential for DENV replication 36,38 . Thus the VDR agonists exerting their effect through modulation of ER stress could have a major effect on DENV replication as observed here.
Typically, active vitamin D (1,25-dihydroxy vitamin D3) is generated by the action of 1,25-dihydroxylase (CYP27B1) in the kidney. Binding of active vitamin D to the VDR results in translocation to the nucleus where it can bind to specific vitamin D responsive elements (VDRE). These complexes can recruit transcription factors to mediate gene transcription 23 . As part of the transcriptional response, 24-hydroxylase, a catabolic enzyme for vitamin D that is encoded CYP24A1 is induced, and CYP27B1 is suppressed and VDR can also regulate transcription of the VDR gene 23 . The agonists investigated in this study all increased the transcription of VDR, and suppressed transcription of CYP27B1 (the enzyme responsible for active vitamin D synthesis) but markedly all also suppressed transcription of CYP24A1, the enzyme involved in catabolism of vitamin D. However, a commercial vitamin D agonist showed elevated expression of CYP241A, and thus, it is likely that the agonists used here would have a longer effective time than vitamin D itself, and this would be supported by the more than 3Log 10 inhibition seen with the agonists investigated in this study as compared with the approximately 2Log 10 inhibition seen with the commercial agonist.
The cell line used in this study, HEK293T/17 is derived from human embryonic kidney 39 , and although the kidney is not believed to be a target organ for DENV infection, the cell line is widely used in DENV studies. In this case the use of HEK293T/17 with a functioning VDR response pathway has been useful in highlighting some of the mechanism of action of the agonists used here. However, examination of the effect of the agonists in HepG2 cells also showed an effect, suggesting that the drug has broad cell specificity. Markedly, the agonists used here additionally had effects (albeit to a lesser extent) on other flaviviruses including JEV and ZIKV, although the strongest effects were seen with DENV, both for DENV 2, and other DENV serotypes.
Vitamin D insufficiency and deficiency are a global problem, with high levels in America (36% of the general population), Northern Europe (92% of the general population), Asia (45-99% of the general population), North Africa (60% of the general population), as well as in Canada, the Middle East and Australia 40 . Thus, it is likely that most people living in dengue endemic areas are vitamin D deficient, thus vitamin D deficiency may increase the presentation of the disease. However, clinical evidence on vitamin D status during DENV infection remains scant. An investigation into serum vitamin D levels found that levels were higher in DF and DHF patients as compared to normal controls 18 , while a much larger recent study found no difference in 25-hydroxy vitamin D www.nature.com/scientificreports/ levels between controls and cases 41 . However, one study suggested that a combination of oral calcium carbonate and vitamin D3 improved the clinical presentation and reduced signs and symptoms of dengue fever 42 , although the number of patients was extremely low.

conclusion
In this study seven fused bicyclic derivatives of 1H-pyrrolo [1,2]imidazol-1-one with vitamin D receptor (VDR) agonist activity were evaluated for possible anti-DENV activity. The results showed that five of the compounds were able to significantly inhibit DENV infection. The most effective compound, ZD-3, had an EC 50 value of 7.47 μM, a selective index of 52.75 and reduced virus production by more than 3Log 10 . These results suggest that these VDR agonists have the potential for future development as effective anti-DENV agents, and possibly as more broad-spectrum anti-flaviviral agents. ZD-20) used in this study were as previously described 22 . General information, including compound ID, chemical formula and formula weight, is provided in Table 1. Chemical structures are given in Supplemental Table S1. In addition, the commercial VDR agonist, EB1089 (Tocris Cookson Ltd., Bristol, UK) was also used in this study. All compounds were dissolved with 100% DMSO to a final concentration of 100 mM and kept at − 30 °C until used.

Materials and methods
Cell viability assessment by cell morphology alteration, trypan blue staining, and MTT assay. To assess the toxicity of the VDR agonists, HEK293T/17 cells were cultured in 6-well plates until 70% confluence was obtained under standard conditions. The medium was replaced with the VDR agonists diluted with complete medium to various concentrations (1-200 µM). After 24 h of incubation, the cell morphology of the treated cells was observed under an inverted microscope, following which the treated cells were trypsinized and stained with 0.4% trypan blue solution. Cells were counted using a hemocytometer and the percentage cell viability determined. To determined cell viability by the MTT assay (Thermo Fisher Scientific Inc., Waltham, MA), HEK293T/17 cells were cultured for 24 h on 96-well plates until 70% confluence was reached under the standard conditions. The cell culture medium was incubated with 100 µl of various concentrations of VDR agonists (1-800 µM) diluted with complete DMEM in parallel with DMSO control. After 24 h of incubation, MTT dye was added into each well and the formazan precipitant was dissolved with DMSO. Optical density at 570 nm was determined using a standard microplate reader. The percentage of cell viability was calculated from the average measurement of four replicates as compared with the negative control (cells treated with complete medium).
Virucidal assay. Stock DENV 2 was incubated directly with medium only or with DMSO control or with 10 µM of VDR agonists in a final volume of 100 µl for 1 h at 37 °C, after which infectious virus titer was determined by standard plaque assay, essentially as described elsewhere 43 . All experiments were undertaken independently in triplicate with duplicate plaque assay. VDR agonist treatment. HEK293T/17 cells were cultured on 6-well plates until 70% confluence was reached under standard conditions. The cells were mock-infected or infected with DENV 2 at a multiplicity (MOI) of 5 for 2 h. After infection, the medium was removed and the infected cells were incubated with appropriate concentrations of the VDR agonists until 24 h.p.i., in parallel with DMSO treated control cells. Experiments were performed independently triplicate.
For time course experiments, under conditions of pre-infection treatment, HEK293T/17 cells were pre-treated with an appropriate concentration of VDR agonists for 1, 3 or 6 h before infection, followed by incubation in complete medium. For treatment post-infection, infected cells were treated with VDR agonists at 0, 1, 3, 6 and 12 h.p.i. The infected cells and supernatant for both pre-and post-treated samples were collected at 24 h.p.i. Experiments were performed independently triplicate.
For viruses other than DENV 2, HEK293T/17 cells were infected with DENV 1 (strain 16007), DENV 3 (strain 16562), DENV 4 (strain 1036) and ZIKV (strain SV0010/15) at MOI 10 and with JEV (strain BJ1) at MOI 1. At 2 h.p.i., the virus was removed and the cells were incubated with 10 μM of VDR agonist (ZD-6) in complete medium for 24 h, after which the supernatant was collected, and the virus titer was determined by standard plaque assay. Experiments were performed independently triplicate.
Determination of viral infectivity by flow cytometry. HEK293T VDR-regulated gene expression by quantitative real-time PCR. HEK293T/17 cells were incubated with either 10 µM of VDR agonists or 20 μM of commercial VDR agonist EB1089 (Tocris Cookson Ltd., Bristol, UK) for 24 h. Subsequently, the treated cell lysate was collected to determine the mRNA levels of the VDR gene as well as genes regulated by VDR. Total RNA was extracted and cDNA was synthesized as described above. Amplification was performed using specific primers, namely VDR-F: Immunofluorescence assay. HEK293T/17 cells grown on glass coverslips were mock-infected or infected with DENV 2 followed by treatment with 10 µM of VDR agonists for 24 h, after which time cells were fixed with 4% ice-cold paraformaldehyde. Then the cells were washed twice with 1X PBS/IFA and blocked with 10% goat serum. After washing twice with 0.03% Triton X-100 in PBS/IFA, the cells were permeabilized with 0.3% Triton X-100 in PBS. The cells were incubated with a mouse monoclonal anti-dengue serotype 1-4 antibody (MA1-27093; Thermo Fisher Scientific Inc., MA) and a rabbit monoclonal anti-VDR antibody (Abcam: ab3508), both at a dilution of 1:200. Cells were subsequently incubated with an Alexa 488-conjugated donkey anti-mouse IgG antibody (A21202, Thermo Fisher Scientific Inc., Waltham, MA) and an Alexa 647-conjugated donkey anti-rabbit IgG antibody (A31573, Thermo Fisher Scientific Inc., Waltham, MA), together with a 1:500 dilution of DAPI for 1 h at room temperature. The coverslips were subsequently mounted onto glass slides using Prolong Gold anti-fade reagent (Invitrogen) before visualization under a LSM 800w Airy scan confocal microscope (ZEISS, Oberkochen, Germany).

Data availability statement
All data generated or analysed during this study are included in this published article (and its Supplementary  Information files).