Studies on anti-picornavirus compounds have revealed an essential role of a novel cellular pathway via host phosphatidylinositol-4 kinase III beta (PI4KB) and oxysterol-binding protein (OSBP) family I in poliovirus (PV) replication. However, the molecular role for this pathway in PV replication has yet to be determined. Here, viral and host proteins modulating production of phosphatidylinositol 4-phosphate (PI4P) and accumulation of unesterified cholesterol (UC) in cells were analyzed and the role of the PI4KB/OSBP pathway in PV replication characterized. Virus protein 2BC was identified as a novel interactant of PI4KB. PI4KB and VCP/p97 bind to a partially overlapped region of 2BC with different sensitivity to a 2C inhibitor. Production of PI4P and accumulation of UC were enhanced by virus protein 2BC, but suppressed by virus proteins 3A and 3AB. In PV-infected cells, a PI4KB inhibitor suppressed production of PI4P, and both a PI4KB inhibitor and an OSBP ligand suppressed accumulation of UC on virus-induced membrane structure. Inhibition of PI4KB activity caused dissociation of OSBP from virus-induced membrane structure in PV-infected cells. Synthesis of viral nascent RNA in PV-infected cells was not affected in the presence of PI4KB inhibitor and OSBP ligand; however, transient pre-treatment of PV-infected cells with these inhibitors suppressed viral RNA synthesis. These results suggest that virus proteins modulate PI4KB activity and provide PI4P for recruitment of OSBP to accumulate UC on virus-induced membrane structure for formation of a virus replication complex.
Dulbecco's modified Eagle medium
HEPES buffered saline
hepatitis C virus
nonstructural protein 5A
phosphatidylinositol-4 kinase III beta
proximity ligation assay
Poliovirus, a small non-enveloped virus with a single strand positive genomic RNA of about 7500 nt, belongs to the Enterovirus species C in the genus Enterovirus and the family Picornaviridae . PV causes poliomyelitis, destruction of motor neurons being mediated by the direct effects of PV in the cells [2, 3]. As an important target pathogen in public health preventable by vaccine, the World Health Organization started a global eradication program for poliomyelitis in 1988. In this eradication program, antivirals against PV are anticipated to have some roles in the post-eradication era in controlling a circulating vaccine-derived PV along with inactivated PV vaccine, treating patients chronically infected with PV, and for administering to persons exposed to PV . Accordingly, we have identified several candidate compounds with antiviral effects on PV by large-scale screening of potential anti-PV compounds in chemical libraries [5-9].
One major group of non-cytotoxic compounds, the enviroxime-like compounds, inhibits virus RNA replication and is defined as compounds that have in common a resistance mutation in the 3A-encoding region (G5318A [3A-Ala70Thr] mutation) and little structural similarity to the anti-picornavirus compound, enviroxime [5-11]. Enviroxime, which inhibits positive-strand RNA synthesis by preventing normal formation of replication complex [10, 12], was recently identified as a PI4KB inhibitor; thus, it is a host-targeting antiviral [8, 13]. One model of PV replication suggests that virus protein 3A acts as an indirect tether (3A/GBF1/ARF1) for recruiting PI4KB at the site of virus RNA replication, and that PI4KB provides PI4P for recruitment of virus RNA-dependent RNA polymerase (virus protein 3D) via direct interaction of PI4P and 3D protein on reorganized membrane vesicles for formation of virus replication complex . A role of ACBD3 in recruitment of PI4KB is proposed for Aichi virus replication [15, 16]; however, a role as suppressor is suggested for PV replication .
Enviroxime-like compounds are classified into at least two different groups in terms of the target host factors: PI4KB inhibitors (e.g., enviroxime, PIK93, GW5074 and T-00127-HEV1) and OSBP family I ligand (e.g., 25-HC, AN-12-H5 and T-00127-HEV2) [8, 9]. OSBP transfers cholesterol between endoplasmic reticulum and trans Golgi in a PI4P-dependent manner and contributes to homeostasis of cholesterol and lipid [18, 19]. HCV, which is a small enveloped virus with a single strand positive genomic RNA of about 9600 nt belonging to the genus Hepacivirus in the family Flaviviridae , uses PI4KA and OSBP for its RNA replication via the N-terminal domain I of NS5A, which activates PI4KA and binds to OSBP [21-23]. These observations underscore a cellular pathway via activation PI4 kinases and OSBP as a common machinery for virus RNA replication of some positive strand RNA viruses.
In the present study, we analyzed viral and host proteins involved in the PI4KB/OSBP pathway in PV replication and identified virus protein 2BC as a novel interactant of PI4KB. PI4KB and VCP/p97, which is a host factor for PV RNA replication involved in cellular secretion pathway , bind to a partially overlapped region of 2BC with different sensitivity to 2C inhibitor. We found that production of PI4P and accumulation of UC are enhanced by virus protein 2BC and suppressed by virus proteins 3A and 3AB. We also found that, in PV replication, PI4KB is located upstream in the PI4KB/OSBP pathway and provides PI4P for recruitment of OSBP to accumulate UC on virus-induced membrane structure.
MATERIALS AND METHODS
Cells, viruses, plasmids, antibodies and compounds
RD cells (human rhabdomyosarcoma cell line) and HEK293 cells (human embryonic kidney cells) were cultured as monolayers in DMEM supplemented with 10% FCS. RD cells were used for titration of PV and cholesterol labeling assays. HEK293 cells were used for mammalian two-hybrid assay, immunofluorescence microscopy and flow cytometry. Expression vectors for PV proteins were constructed with pKS435, in which expression of virus proteins was controlled under the HEF-1α promoter. Antibody against PV 2B was raised in rabbits with peptides WLRKKACDVLEIPYVIKQ (amino acids 80–97 of PV 2B protein). Rabbit hyperimmune serum against PV 2 C protein was a kind gift from Tomoichiro Oka (Department of Virology II, National Institute of Infectious Diseases, Japan). Anti-PI4KB antibodies (mouse antibody [BD transduction laboratory; Franklin Lakes, NJ, USA]; rabbit antibody [Millipore, Billerico, MA, USA]; anti-VCP/p97 antibodies [mouse and rabbit antibodies; Abcam, Cambridge, UK] and anti-PI4P antibody [mouse IgM antibody; Echelon Biosciences, Logan, UT, USA]) were used for immunostaining. VCP/p97 inhibitors (EerI and XN) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). A PI4KB specific inhibitor T-00127-HEV1 (3-[3,4-dimethoxyphenyl]-2,5-dimethyl-N-[2-(4-morpholinyl) ethyl]pyrazolo[1,5-a]pyrimidin-7-amine) was supplied by Pharmeks (Moscow, Russia). An OSBP ligand T-00127-HEV2 ([3β,17β]-16,16-dimethyl-3-[(tetrahydro-2H-pyran-2-yl) oxy]-Androst-5-en-17-ol), was kindly provided by Hirotatsu Kojima (Open Innovation Center for Drug Discovery, University of Tokyo, Tokyo, Japan). BODIPY-cholesterol was purchased from Avanti Polar Lipids (Alabaster, AL, USA). siRNAs were purchased from Thermo Fisher Scientific (Pittsburgh, PA, USA) as a form of siGENOME SMART pools.
Mammalian two-hybrid assay
Mammalian two-hybrid assays were performed using a Checkmate mammalian two-hybrid system (Promega, Madison, WI, USA). VCP/p97 and PV proteins were expressed as fusion proteins in pFN10A(ACT) Flexi vector and pFN10A(BIND) Flexi vector. pBIND-VCP/p97 or pBIND-PI4KB was co-transfected with pACT-PV proteins expression vectors with a reporter plasmid (pGL4.31[luc2P/GAL4 UAS/Hygro]) into HEK293 cells. Compounds were added 3 hr p.t. in the indicated concentrations. Firefly luciferase activity (derived from reporter plasmid caused by binding of fusion proteins) and Renilla luciferase activity (derived from pBIND vectors for normalization of transfection efficiency) were measured 24 hr p.t. of DNA. As controls, pACT empty vectors with pBIND-VCP or with pBIND-PI4KB (a generous gift from Jun Sasaki, Department of Virology and Parasitology, Fujita Health School of Medicine, Aichi, Japan) were used . The ratio of firefly luciferase and Renilla luciferase was calculated, and then normalized by the controls.
HEK293 cells (8.0 × 105 cells) transfected with PV protein expression vectors or infected with PV were collected in 100 µL cell lysis buffer. SDS–PAGE, blotting and detection were performed as previously described [8, 14].
Cells were fixed with 3% paraformaldehyde for 10 min at room temperature, and then permeabilized with 20 μM digitonin in HBS (21 mM HEPES buffer [pH 7.4], 1.8 mM disodium hydrogen phosphate, 137 mM NaCl, 4.8 mM KCl) (for PI4P staining) or with 0.1% Triton/3% FCS/HBS as indicated. The cells were stained by indirect immunofluorescence with primary antibodies against virus and host proteins, secondary antibodies conjugated with Alexa Fluor 488 or 594 dyes (Molecular Probes, Eugene, OR, USA), Hoechst 33342 (Molecular Probes) for counterstaining of nuclei, or with filipin III (Cayman, Ann Arbor, MI, USA) for staining of UC. Samples were observed with a confocal scanning laser microscope (FV1000, Olympus, Tokyo, Japan). Pearson's correlation coefficient values of more than 0.5 are highlighted with cyan in the figures.
HEK293 cells (8.0 × 105 cells) transfected with PV protein expression vectors or infected with PV at an MOI of 20 were collected 24 hr p.t. or 6 hr p.i. by using trypsin/EDTA (0.16% trypsin, 0.02% EDTA in PBS) in 0.8 mL of 10% FCS–DMEM. Cells were fixed with 3% paraformaldehyde for 10 min at room temperature and then permeabilized with 20 μM digitonin in HBS for 5 min. Cells were incubated with primary antibodies for 30 min at 37 °C. Anti-2B antibody was used to detect PV-infected cells. Cells were washed two times with HBS and then incubated with secondary antibodies conjugated with Alexa Fluor 647 and 488 dyes (Molecular Probes) for 20 min at 37 °C. UC in the cells was stained with filipin III (Cayman) at room temperature for 30 min as described previously . After washing, cells were suspended in 250 μL of HBS and then analyzed with a BD FACSCanto II Flow Cytometer (BD Biosciences, Franklin Lakes, NJ, USA).
The relative amounts of PI4P and UC are expressed as the ratio of geometric means of the signals of PI4P and UC in virus protein-expressing cells or PV-infected cells to those in non-expressing cells or non-infected cells. This ratio was 1 in non-expressing cells or non-infected cells. To evaluate the effects of test compounds on the amounts of PI4P and UC in the cells, the net increases of PI4P or UC induced by virus proteins or PV infection (%), which was defined as 100 × (relative amounts of PI4P and UC – 1). The net increases of PI4P or UC in virus protein-expressing cells or PV-infected cells in the absence of such compounds were taken as 100%.
pKS435-2BC (2BC expression vector) was co-transfected with VCP/p97-GFP-FLAG or FLAG-PI4KB expression vectors into HEK293 cells (4.0 × 105 cells) cultured in 24-well plates using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instruction. As a control, pKS435-2BC was transfected into HEK293 cells without FLAG-tagged protein expression vectors. At 24 hr p.t., cells were treated with the test compounds for 1 hr and then fixed with 1.8% paraformaldehyde for 10 min at room temperature. The cells were then lysed with 200 μL of cell lysis buffer (20 mM Tris buffer [pH 7.5], 150 mM NaCl, 2 mM EDTA, 10% glycerol, 1% Nonidet P-40, and protease inhibitor cocktail [complete Mini EDTA-free, Roche, Indianapolis, IN, USA]), after which they were sonicated for 30 s. Lysates were subjected to centrifugation at 13,000 g for 10 min at 4 °C and the supernatants collected. Supernatants (180 μL) were mixed with 20 μL of pre-washed anti-FLAG magnetic beads (ANTI-FLAG M2 magnetic beads, Sigma–Aldrich, St Louis, MO, USA), and then incubated for 1 hr on a rotator at 4 °C. The beads were then washed three times with washing buffer (20 mM Tris buffer [pH 7.5], 150 mM NaCl, 2 mM EDTA, 10% glycerol, and 1% Nonidet P-40), after which they were collected in SDS–PAGE sample buffer. Samples were decrosslinked by boiling for 30 min, and then subjected to western blot analysis with anti-2C and anti-FLAG antibodies.
Proximity ligation assay
Proximity ligation assay was performed by using Duolink II reagents (Olink Bioscience, Uppsala, Sweden). HEK293 cells expressing 2BC protein were treated with the test compounds (GuHCl [2 mM] or T-00127-HEV1 [10 µM]) or with cell culture medium without these compounds (mock treatment) for 1 hr, and then fixed and permeabilized as described under the heading “Immunofluorescence microscopy”. For detection of PLA signals between 2BC and PI4KB or VCP/p97, anti-2C antibody (rabbit antibody) was used with anti-PI4KB or -VCP/p97 antibodies (mouse antibodies). Individual antibodies (anti-2C, -PI4KB and -VCP/p97) were used as negative controls for the PLA reaction. Next, cells were subjected to indirect immunofluorescence to detect 2BC, PI4KB or VCP/p97 with secondary antibodies (anti-rabbit or anti-mouse goat antibodies conjugated with Alexa Fluor 488). Samples were observed with a confocal scanning laser microscope (FV1000, Olympus).
Quantification of cholesterol
HEK293 cells (1.0 × 106 cells) were infected with PV at an MOI of 0 (mock-infected sample), 20 or 50, and then collected 6 hr p.i. The cells were washed with 0.8 mL of HBS three times, after which cholesterol was extracted in 200 μL of extraction buffer (chloroform:isopropanol:NP-40 = 7:11:0.1 volumes). The extracts were air dried and treated under vacuum for 30 min. Amount of UC in the extracts were quantified by using a Cholesteol/Choleteryl Ester Quantification Colorimetric/Fluorometric Kit (BioVision, San Francisco, CA, USA), according to the manufacturer's instructions.
Labeling of nascent viral RNA
Poliovirus-infected cells were treated with 10 µg/mL of actinomycin D at 3 hr p.i. for 1 hr, and were then transfected with BrUTP using a Fugene6 Transfection Reagent (Promega). The cells were incubated at 37 °C for 1 hr, and then stained with anti-bromodeoxyuridine antibody (Roche), followed by anti-mouse goat antibodies conjugated with Alexa Fluor 594 (Molecular Probes).
The results of experiments are shown as the means with SDs. Student's two-tailed t-test was performed with data obtained from three independent experiments. P values of less than 0.05 were considered significantly difference and are indicated by asterisks.
Identification of PV 2BC and 2C as novel interactants with PI4KB
To identify PV virus proteins that interact with PI4KB, mammalian two-hybrid assays were performed (Fig. 1a). VCP/p97 that had 2BC as the sole interactant in our previous assay was used as a control . In addition to 3A, which is a known interactant of PI4KB [14, 16], virus proteins 2B, 2BC and 2C were identified as novel interactants with PI4KB. Interaction of VCP/p97 and PI4KB with virus proteins was analyzed in the presence of a 2C inhibitor, GuHCl. GuHCl significantly enhanced interactions between VCP/p97 and 2BC and 2C (14- and 8.2-fold enhancement, respectively, P < 0.001), but not those with other virus proteins. GuHCl treatment did not enhance the interaction of PI4KB with any virus proteins; however, a PI4KB inhibitor, T-00127-HEV1, enhanced interactions between PI4KB and 2BC and 2C (8.3- and 4.7-fold enhancement, respectively, P < 0.01), but not those with other virus proteins. T-00127-HEV1 did not affect the interaction of VCP/p97 with any virus proteins. Expression of 2BC, PI4KB and VCP/p97 was not affected by GuHCl and T-00127-HEV1 treatment (data not shown). 2BC colocalized with PI4KB and VCP/p97, but not with GBF1 (Fig. 1b). The middle to C-terminal region of 2C was responsible for the interaction with VCP/p97 and the N- and C-terminal regions of 2C for interaction with PI4KB (Fig. 1c).
Interaction of 2BC with VCP/p97 was also confirmed by co-immunoprecipitation and by PLA in the presence or absence of the test compounds, including VCP/p97 inhibitors, EerI (D1 domain-interacting non-ATPase inhibitor) and XN (N domain-interacting inhibitor) [26-28] (Fig. 2). Both EerI and XN showed anti-PV activity (Fig. 2a). EerI and XN treatment significantly suppressed the interaction of VCP/p97 with 2BC (0.25- to 0.77-fold reduction, P < 0.01 to 0.001), but enhanced the interaction of PI4KB with 2BC (2.2- to 7.9-fold enhancement, P < 0.01 to 0.001), and that with 2 C (2.7- to 12-fold enhancement, P < 0.01). GuHCl treatment enhanced the interaction of VCP/p97 with 2BC and 2C (P < 0.001) and slightly, but significantly, suppressed the interaction of PI4KB with 2BC (P < 0.01). The amounts of co-precipitated 2BC or PLA signals between 2BC and VCP/p97 were enhanced in the presence of GuHCl by 2.6- and 2.8-fold, respectively. Treatment with other compounds did not affect the interaction of 2BC with VCP/p97 in co-immunoprecipitation and PLA.
Interaction of 2BC with PI4KB was also confirmed by PLA (Fig. 2d); however, no significant effects of the test compounds on the interaction were observed. Co-precipitation of 2BC with FLAG-tagged PI4KB was not detected (data not shown).
2BC induces UC accumulation via PI4KB and OSBP family I
To analyze the virus proteins involved in PI4P production, cellular PI4P and UC in the cells expressing each virus protein were assessed (Fig. 3). Expression of 2C-containing putative precursors, 2BC, 2BC3A and 2C3A, upregulated PI4P in the cells and accumulated UC. UC colocalized with these 2C-containing putative precursors in the cells.
Phosphatidylinositol 4-phosphate and UC in the cells expressing each virus protein were quantitated by flow cytometry analysis (Fig. 4a). Production of PI4P was enhanced by expression of 2BC, 2BC3A and 2C3A (1.4-, 2.0- and 1.3-fold enhancement, respectively), and substantially reduced by expression of 3 A and 3AB (0.73- and 0.80-fold reduction, respectively) (Fig. 4b). Apparent increases in UC by expression of 2BC, 2BC3A and 2C3A (1.1-, 1.3- and 1.2-fold enhancement, respectively), or reduction by expression of 3A and 3AB (0.90- and 0.93-fold reduction, respectively) compared to that in non-expressing cells were observed. In PV-infected cells, only 2BC was detected among these putative precursor proteins (Fig. 4c). The effects of GuHCl, T-00127-HEV1 and an OSBP family I ligand (T-00127-HEV2) and of siRNAs targeting PI4KB or OSBP on relative amounts of PI4P and UC by 2BC expression were analyzed (Fig. 4d). Expression of 2BC was not affected by treatment with these compounds (data not shown). Treatment with GuHCl did not affect the net increases in PI4P and UC induced by 2BC expression. Treatment with T-00127-HEV2, but not with T-00127-HEV1, suppressed upregulation of PI4P by 2BC expression. UC in 2BC-expressing cells was significantly reduced by treatment with both T-00127-HEV1 and T-00127-HEV2 (P < 0.01) and a synergic effect of T-00127-HEV1 and T-00127-HEV2 on PI4P expression was observed (P < 0.01). Observed effects of T-00127-HEV1 and T-00127-HEV2 were recapitulated by siRNA treatment targeting PI4KB and OSBP, respectively, but not by a control non-targeting siRNA. These results suggest that accumulation of UC by 2BC depends on PI4KB and OSBP.
PI4KB and OSBP accumulate UC in PV-infected cells
Phosphatidylinositol 4-phosphate and UC in PV-infected cells were quantitated by flow cytometry (Fig. 5). First, the effect of digitonin treatment on the measurement by cytometry of PI4P and UC in PV-infected cells was analyzed (Fig. 5a). Increase in PI4P signals in cells infected with EGFP-expressing PV was observed both with mock treatment and digitonin treatment. In contrast, UC signals in the infected cells were similar to those in non-infected cells without digitonin treatment. After digitonin treatment, UC signals of infected cells were increased whereas those in non-infected cells were reduced, possibly reflecting different localization and fugacity of UC in the cells. UC predominantly resides in the plasma membrane in non-infected cells (65–80% of cellular UC) , and is susceptible to digitonin extraction. Accumulation of UC in intracellular virus-induced membrane structure with PI4P would decrease resistance to digitonin extraction . This suggests that digitonin treatment allows quantification of accumulated intracellular UC by filipin III staining.
Next, the effects of PI4KB inhibitor and OSBP ligand on PI4P production and UC accumulation in PV-infected cells were analyzed (Fig. 5b). GuHCl, which was used as a control for replication inhibition, did not affect the net increase of PI4P, but slightly upregulated UC in PV-infected cells (P < 0.01). In contrast, treatment with T-00127-HEV1, but not with T-00127-HEV2, reduced the net increase of PI4P (P < 0.01). When added over the period in which their anti-PV activity is effective (3–5 hr p.i.), both T-00127-HEV1 and T-00127-HEV2 treatments reduced the net increase in UC (P < 0.01) . These results suggest that P14P production and UC accumulation are the targets of PI4KB inhibitor and OSBP ligand, respectively, for their anti-PV activities.
In non-infected cells, OSBP localizes in the cytoplasm and also at the Golgi apparatus, where interactions of its PH domain with PI4P and ARF protein are essential . In PV-infected cells, OSBP was co-localized with viral proteins 2B, 3 A and host proteins PI4KB and GBF1 (Fig. 6). Transient treatment with GuHCl or T-00127-HEV2 did not affect the localization of OSBP in PV-infected cells. In contrast, treatment with T-00127-HEV1 or with other PI4KB inhibitors (GW5074 and PI93) caused dissociation of OSBP from these proteins in PV-infected cells (Fig. 6, and data not shown). This suggests that localization of OSBP on virus-induced membrane structures depends on PI4P produced by PI4KB in PV replication.
Endogenous cholesterol accumulates on virus-induced membrane structure in the PI4KB/OSBP pathway
To identify the origin of accumulated UC in PV-infected cells, localization of pre-existent or imported cholesterol during PV infection was analyzed by labeling the cells with BODIPY-cholesterol. BODIPY-cholesterol added before PV infection (pre-existent cholesterol) or during PV infection (imported cholesterol) accumulated at a perinuclear region with 2B in PV-infected cells (Fig. 7a). In PV-infected cells, the signal of pre-existent BODIPY-cholesterol was far higher than that of imported BODIPY-cholesterol, whereas the signals of imported BODIPY-cholesterol appeared of similar strength in both PV-infected and non-infected cells. Importation of BODIPY-cholesterol was slightly suppressed in PV-infected cells (8% reduction) compared to that in non-infected cells as quantified by cytometry (Fig. 7b). Total amounts of UC in the cells were unchanged by PV infection (Fig. 7c). This suggests that, during PV infection, pre-existent rather than imported cholesterol is the major source of accumulated UC in PV-infected cells.
Accumulation of pre-existent cholesterol on virus-induced membrane structures was observed in PV-infected cells after 2 hr treatment with GuHCl, but significantly suppressed by treatment with T-00127-HEV1 and T-00127-HEV2 (Fig. 7d). BODIPY-cholesterol was detected both in the plasma membrane and cytoplasm of non-infected cells (Supplementary Data 1). These results suggest that PI4KB and OSBP are involved in accumulation of pre-existent cholesterol on virus-induced membrane structure in PV-infected cells.
The PI4KB/OSBP pathway is essential for formation of a virus replication complex rather than for viral RNA synthesis
To determine the target of the PI4KB/OSBP pathway in PV replication in terms of viral RNA synthesis, synthesis of viral RNA in PV-infected cells in the presence of PI4KB inhibitor and OSBP ligand was analyzed. nsRNA was found to be efficiently synthesized in PV-infected cells in the presence of PI4KB inhibitor or OSBP ligand, but completely suppressed in the presence of GuHCl (Fig. 8a, left panels). In contrast, pretreatment of PV-infected cells with these inhibitors significantly reduced nsRNA synthesis (Fig. 8a, right panels). Even when added from the middle to late stage of PV replication (the effective period of addition of these compounds is from −3 to 6 hr p.i.) , nsRNA synthesis in cells pretreated with GuHCl and T-00127-HEV1 was almost completely suppressed, or considerably reduced with T-00127-HEV2 pretreatment, consistent with their inhibitory effects on PV replication.
These results suggest that the PI4KB/OSBP pathway plays a role in formation of virus replication complex rather than being directly involved in viral RNA synthesis (Fig. 8b).
Lipids play essential roles in the whole life cycle of viruses, including entry, replication, progeny virus production and secretion of progeny virus . In virus RNA replication, modulation of lipids is essential for the formation of virus replication complexes [32, 33]. Rosette-like membrane structures in isolated replication complexes , or single to multilamellar membrane structures, which are possibly derived or initiated from cis Golgi apparatus, have been observed in cells infected by PV or coxsackievirus B3 (CVB3) [35, 36]. PI4P has been identified as a lipid produced on virus-induced membrane structures by PI4KB in PV replication (8, 14). Importation into PV-infected cells of exogenous long chain fatty acids, which are utilized in PC synthesis in the infected cells, has also been reported . Importation of fatty acids depends on the host factor ACSL3 and virus protein 2A, which is not a prerequisite for replication of PV , suggesting that, along with PI4P, PC might serve to optimize the lipid environment for PV replication. A recent study by Ilnytska et al. suggested that UC accumulation via clathrin-mediated endocytosis from the plasma membrane is required for enterovirus replication .
Virus protein 3A has been proposed as the target of PI4KB in PV infection ; however, expression of 3A does not suppress the effect of PI4KB inhibitors . In this study, virus protein 2BC was identified as another interactant of PI4KB. The binding sites of PI4KB on 2BC partially overlapped with those of VCP/p97, and apparent competition of VCP/p97 and PI4KB over binding to 2BC was observed in the presence of GuHCl (Fig. 1). The affinity of 2BC to PI4KB seemed weaker than that to VCP/p97, and the interaction of 2BC with PI4KB was only detectable by some sensitive assays (mammalian two-hybrid assay and PLA); thus, it was not detected by immunoprecipitation (Figs,2 1). The importance of the interaction of 2BC with PI4KB in PV-infected cells remained to be further investigated. VCP/p97, an AAA + ATPase involved in various cellular pathways including protein degradation and homotypic membrane fusion , acts as a host factor for replication of PV and Drosophila C virus [24, 42]. Interaction of 2C/2BC with VCP/p97, but not with PI4KB, was apparently enhanced by GuHCl (Fig. 2), which reportedly suppresses ATPase activity of 2C . 2C belongs to the SFIII helicase family of AAA + ATPases, which cause conformational changes via their ATPase activity . This suggests that interaction of VCP/p97, but not of PI4KB, with 2C/2BC depends on ATPase activity and/or the conformation of 2C/2BC, and also that VCP/p97 might be an indirect target of GuHCl. Production of PI4P induced by 2BC or in PV infection was not affected by GuHCl (Figs,5 4), indicating that the ATPase activity of 2BC is not essential for PI4P production.
2C-containing putative precursors (2BC, 2BC3A and 2C3A), but not 2C or other virus proteins, activated PI4P production and UC accumulation (Figs,4 3). 2BC3A showed the highest activity in induction of PI4P production and UC accumulation. This suggests a negative correlation between induction of PI4P production and ATPase activity of 2C-containing precursors, among which 2C showed the strongest ATPase activity . Because only 2BC was detected in PV-infected cells (Fig. 4c), 2BC seems to act as an activator of PI4P production and UC accumulation in PV infection. Treatment with a PI4KB inhibitor or knockdown of PI4KB increased the relative amount of PI4P in 2BC-expressing cells compared with non-expressing cells, but suppressed UC accumulation induced by 2BC expression (Fig. 4d). This apparent enhancement of PI4P production might be caused by different sensitivity of endogenous PI4KB in 2BC-expressing and non-expressing cells to PI4KB inhibitor or knockdown, possibly via the interaction of 2BC with PI4KB. Interestingly, PI4KB knockdown does not affect relative amounts of PI4P in 3A-expressing cells (Arita, pers. comm., 2014). The ratio of geometric mean of PI4P signals in 3A-expressing cells to that in non-expressing cells was 0.79 by non-targeting siRNA#1 treatment vs. 0.74 by PI4KB-targeting siRNA treatment, P = 0.14 (a non-significant difference). Therefore, apparent enhancement of PI4P production in 2BC-expressing cells caused by PI4KB knockdown seemed to depend on some specific activity of 2BC. However, this phenomenon in 2BC-expressing cells made marked contrast to that in PV-infected cells, where PI4KB inhibitor substantially suppressed PI4P production in the infected cells (Fig. 4b). PI4P production was far greater in PV-infected cells than in 2BC-expressing cells (3.4-fold increase vs. 1.4-fold increase). Considering that the amount of production of PI4P in PV-infected cells was not recapitulated by expression of individual viral protein, a suite of viral proteins, possibly including 2BC and 3 A/3AB, might be required for modulation of PI4KB activity in infected cells. The properties, including stability, of PI4KB in infected cells might also be different from those in 2BC-expressing cells. UC accumulation seemed to be an event specific to virus infection or virus protein expression. Thus, it was not affected by the inhibitors or by knockdown of PI4KB or OSBP in non-infected cells. In contrast to 2BC, 3A and 3AB reduced PI4P and suppressed UC accumulation in the cells and no enhancement of the interaction with PI4KB was observed in the presence of PI4KB inhibitor (Fig. 1a). These observations suggest that 2BC and 3A/3AB have different roles in modulation of PI4P production and UC accumulation in PV replication. 3AB was recently shown to have the ability to induce formation of double-membrane liposomes in vitro . In HCV infection, virus proteins NS5A and NS5B bind to PI4KA, but only NS5A activates PI4KA . The role of NS5B (virus RNA-dependent RNA polymerase) in modulation of PI4KA activity remains unknown. The direct action of PV proteins on PI4KB activity remained to be further elucidated.
Among the OSBP families, OSBP family I, which consists of OSBP and OSBP2, has been identified as a host factor for PV replication in a PI4KB pathway . Roles for OSBP in virus replication and virus particle release from infected cells were originally identified in HCV infection [23, 46]. In addition, antiviral effects of the endogenous OSBP ligand 25-HC and OSBP activity modulated by the interferon-induced transmembrane protein 3 targeting membrane fusion step of enveloped virus infection were recently reported [47, 48]. OSBP transfers cholesterol in a PI4P-dependent manner [19, 49] and overexpression of OSBP enhances cholesterol synthesis and decreases cholesterol esterification in cells . Structural analysis has suggested that sterol is a substrate for the few members of OSBP families that use PI4P as a common substrate . Treatment with the OSBP ligand 25-HC reportedly enhances esterification of cholesterol and decreases UC in cells . These observations suggest that UC is important in PV replication. In this study, accumulation of UC, but not increased amounts of UC, was observed in PV-infected cells (Figs,7 5). PI4P production in PV-infected cells was suppressed by a PI4KB inhibitor, whereas UC accumulation was suppressed by both PI4KB inhibitor and OSBP ligand (Fig. 5), suggesting that accumulation of UC is a downstream event of the PI4KB/OSBP pathway in PV replication. Consistently, treatment with PI4KB inhibitor caused dissociation of OSBP from virus-induced membrane structure in the infected cells (Fig. 6), suggesting that PI4KB is located upstream in the PI4KB/OSBP pathway and provides PI4P for recruitment of OSBP to accumulate UC during PV replication. A recent study by Mesmin et al. on the mechanism of UC transport by OSBP indicated that OSBP transfers UC to membranes by consuming PI4P on them , suggesting that one role of PI4P is as a “fuel” for accumulating UC on membranes during PV replication. Neither PI4KB inhibitor nor OSBP ligand suppressed synthesis of viral RNA on pre-existent viral replication complex; suppression of viral nsRNA synthesis was observed only after pretreatment with these compounds before measurement of viral RNA synthesis (Fig. 8a). These findings indicate that the PI4KB/OSBP pathway has a functional role in accumulating UC on virus-induced membrane structures during PV replication. PI4P produced by PI4KB recruits OSBP on virus-induced membrane structures, after which OSBP accumulates UC on the membranes at the expense of PI4P for formation of a viral replication complex rather than for viral RNA synthesis (Fig. 8b).
Phosphatidylinositol-4 kinase III beta inhibitors have been identified as non-cytotoxic compounds in vitro cultured cells [5, 8, 10-12, 53, 54]. However, they have antiproliferative effect in lymphocytes of mice and rats, and are lethal to SJL mice [53, 54]. Dissection of the PI4KB/OSBP pathway might help to elucidate the mechanism(s) of these adverse effects and to alleviate deleterious effects of PI4KB inhibitors.
In summary, these results provide a molecular basis for a PI4KB/OSBP pathway in PV replication.
We thank Junko Wada for her excellent technical assistance and Hiroyuki Shimizu and Takaji Wakita for their kind support and organization. We are grateful to Tadaki Suzuki for kindly introducing flow cytometry analysis, Jun Sasaki for his generous gift of pBIND-PI4KB vector, Tomoichiro Oka for anti-PV 2C antibody, and Hirotatsu Kojima for kind supports with precious compounds. This study was supported in part by Grants-in-Aid for the Promotion of Polio Eradication and Research on Emerging and Re-emerging Infectious Diseases from the Ministry of Health, Labor and Welfare, Japan, a grant from the World Health Organization for a collaborative research project of the Global Polio Eradication Initiative, and by JSPS KAKENHI Grant Number 25460579.
The author declares no conflict of interest.