Portal de Periódicos da CAPES

ACBD3-mediated recruitment of PI4KB to picornavirus RNA replication sites

2011; Springer Nature; Volume: 31; Issue: 3 Linguagem: Inglês

10.1038/emboj.2011.429

1460-2075

Jun Sasaki, Kumiko Ishikawa, Minetaro Arita, Koki Taniguchi,

Animal Virus Infections Studies

Article29 November 2011free access Source Data ACBD3-mediated recruitment of PI4KB to picornavirus RNA replication sites Jun Sasaki Corresponding Author Jun Sasaki Department of Virology and Parasitology, Fujita Health University School of Medicine, Aichi, Japan Search for more papers by this author Kumiko Ishikawa Kumiko Ishikawa Department of Virology and Parasitology, Fujita Health University School of Medicine, Aichi, Japan Search for more papers by this author Minetaro Arita Minetaro Arita Department of Virology II, National Institute of Infectious Diseases, Tokyo, Japan Search for more papers by this author Koki Taniguchi Koki Taniguchi Department of Virology and Parasitology, Fujita Health University School of Medicine, Aichi, Japan Search for more papers by this author Jun Sasaki Corresponding Author Jun Sasaki Department of Virology and Parasitology, Fujita Health University School of Medicine, Aichi, Japan Search for more papers by this author Kumiko Ishikawa Kumiko Ishikawa Department of Virology and Parasitology, Fujita Health University School of Medicine, Aichi, Japan Search for more papers by this author Minetaro Arita Minetaro Arita Department of Virology II, National Institute of Infectious Diseases, Tokyo, Japan Search for more papers by this author Koki Taniguchi Koki Taniguchi Department of Virology and Parasitology, Fujita Health University School of Medicine, Aichi, Japan Search for more papers by this author Author Information Jun Sasaki 1,‡, Kumiko Ishikawa1,‡, Minetaro Arita2 and Koki Taniguchi1 1Department of Virology and Parasitology, Fujita Health University School of Medicine, Aichi, Japan 2Department of Virology II, National Institute of Infectious Diseases, Tokyo, Japan ‡These authors contributed equally to this work *Corresponding author. Department of Virology and Parasitology, Fujita Health University School of Medicine, Dengakugakubo 1-98, Kutsukakecho, Toyoake, Aichi 470-1192, Japan. Tel.: +81 562 93 2486; Fax: +81 562 93 4008; E-mail: [email protected] The EMBO Journal (2012)31:754-766https://doi.org/10.1038/emboj.2011.429 PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Phosphatidylinositol 4-kinase IIIβ (PI4KB) is a host factor required for genome RNA replication of enteroviruses, small non-enveloped viruses belonging to the family Picornaviridae. Here, we demonstrated that PI4KB is also essential for genome replication of another picornavirus, Aichi virus (AiV), but is recruited to the genome replication sites by a different strategy from that utilized by enteroviruses. AiV non-structural proteins, 2B, 2BC, 2C, 3A, and 3AB, interacted with a Golgi protein, acyl-coenzyme A binding domain containing 3 (ACBD3). Furthermore, we identified previously unknown interaction between ACBD3 and PI4KB, which provides a novel manner of Golgi recruitment of PI4KB. Knockdown of ACBD3 or PI4KB suppressed AiV RNA replication. The viral proteins, ACBD3, PI4KB, and phophatidylinositol-4-phosphate (PI4P) localized to the viral RNA replication sites. AiV replication and recruitment of PI4KB to the RNA replication sites were not affected by brefeldin A, in contrast to those in enterovirus infection. These results indicate that a viral protein/ACBD3/PI4KB complex is formed to synthesize PI4P at the AiV RNA replication sites and plays an essential role in viral RNA replication. Introduction All known positive-strand RNA viruses utilize intracellular membranes, such as the endoplasmic reticulum (ER), the ER-Golgi intermediate compartment, the Golgi, endosomes, or lysosomes, for genome replication. Virus infection results in remodelling of intracellular membranes, and replication complexes, where viral RNA is replicated, are formed associated with membranes. Virus-induced membrane structures are thought to increase the local concentrations of components required for replication, to provide a scaffold for anchoring the replication complexes, to prevent the activation of certain host defense mechanisms that can be triggered by dsRNA produced during virus RNA replication, and to provide certain lipids required for genome synthesis (reviewed in Miller and Krijnse-Locker, 2008). The family Picornaviridae is a group of non-enveloped, single-stranded positive-sense RNA viruses. Picornaviruses include many important pathogens for humans and animals, such as poliovirus, enterovirus 71, rhinoviruses, hepatitis A virus (HAV), and foot-and-mouth disease virus (FMDV). Each genome is 7200–8500 nucleotides in length, and has a single large open reading frame (ORF) consisting of a capsid-coding P1 region, and non-structural protein-coding P2 and P3 regions. Some viruses encode a non-structural protein, leader (L) protein, upstream of the P1 region. After a large polyprotein has been translated from a single ORF, the polyprotein is processed by virus-encoded proteases into 11–12 final cleavage products. Of the non-structural proteins, 2B, 2C, and 3A, and the cleavage intermediates, 2BC and 3AB, are membrane-associated proteins (Towner et al, 1996; Teterina et al, 1997; Knox et al, 2005; Moffat et al, 2005; Krogerus et al, 2007), and have been reported to be involved in membrane reorganization (Cho et al, 1994; Aldabe et al, 1996; Egger et al, 2000; Suhy et al, 2000). In addition, cellular factors are thought to be required for membrane reorganization for picornavirus replication. Many studies have shown the involvement of Golgi-specific Brefeldin A (BFA) resistance factor 1 (GBF1) and ADP-rybosilation factor 1 (Arf1), which participate in the cellular secretory pathway, in the replication of enteroviruses such as poliovirus and coxsackievirus B3 (CVB3) (Belov et al, 2005, 2007, 2008; Lanke et al, 2009; Wessels et al, 2006a, 2009b). Furthermore, the formation of membranous vesicles by enterovirus infection is also proposed to occur through COPII-mediated vesicle budding from the ER (Rust et al, 2001), or an autophagy-mediated process (Suhy et al, 2000; Jackson et al, 2005). Recently, a model by which enteroviruses remodel membranes for viral RNA replication was proposed (Hsu et al, 2010). According to the model, enterovirus RNA replication begins at the Golgi/trans-Golgi network (TGN). Viral protein 3A anchored to membranes binds and modulates GBF1/Arf1 to enhance recruitment of phosphatidylinositol 4-kinase IIIβ (PI4KB) to the sites for viral RNA replication on the membranes, over COPI. PI4KB catalyses the production of phophatidylinositol-4-phosphate (PI4P). The produced PI4P binds to soluble viral 3D RNA polymerase and recruits it to the membranes to facilitate viral RNA synthesis. Enhanced recruitment of PI4KB on the membranes decreases anterograde transport and leads to the emergence of PI4P-enriched organelles for enteroviral RNA replication adjacent to the ER exit sites. Aichi virus (AiV) is a member of the family Picornaviridae (Yamashita et al, 1998), and belongs to the genus Kobuvirus, a different genus from the genus Enterovirus, to which poliovirus and CVB3 belong. AiV was first isolated from patients with oyster-associated acute gastroenteritis in 1989 in Japan (Yamashita et al, 1991). The virus has been detected in gastroenteritis outbreaks or sporadic cases of diarrhoea not only in Japan, but also in other Asian countries, Brazil, Europe, and Africa, and is suggested to be a causative agent of gastroenteritis (Yamashita et al, 1995, 2000; Oh et al, 2006; Pham et al, 2007; Ambert-Balay et al, 2008; Goyer et al, 2008; Sdiri-Loulizi et al, 2008; Reuter et al, 2009; Yang et al, 2009). We have performed several studies to obtain an understanding of the mechanism of this virus replication, such as the characterization of some non-structural proteins and the cis-acting replication element at the 5′-terminus of the genome (Sasaki et al, 2003; Nagashima et al, 2005; Sasaki and Taniguchi, 2003, 2008; Ishikawa et al, 2010); however, the host factors involved in AiV replication remain to be elucidated. In this study, we demonstrated that PI4KB is an essential host factor for RNA replication of AiV as in the case of enteroviruses. However, we found that AiV utilizes a different strategy to recruit PI4KB to the site of genome replication from that used by enteroviruses. We showed that AiV non-structural proteins, 2B, 2BC, 2C, 3A, and 3AB, interact with a Golgi resident protein, acyl-coenzyme A binding domain containing 3 (ACBD3). In addition, we found that ACBD3 interacts with PI4KB. No direct interaction between PI4KB and the viral proteins was detected. Knockdown of ACBD3 or PI4KB reduced virus RNA replication. Immunofluorescence microscopy revealed colocalization among the viral non-structural proteins, ACBD3, PI4KB, PI4P lipids, and dsRNA in viral RNA-replicating cells. Thus, these results indicate that a viral protein/ACBD3/PI4KB complex is formed to synthesize PI4P at the AiV RNA replication sites and plays an essential role in viral RNA replication. Results 2B, 2BC, 2C, 3A, and 3AB interact with ACBD3 To identify host proteins involved in AiV genome replication, we searched for host proteins that interact with the AiV 2B, 2C, and 3A proteins by screening a HeLa cell cDNA library in the yeast two-hybrid system. As a result, ACBD3 was identified as a binding partner of 3A. Using a mammalian two-hybrid system, all of the non-structural proteins including 3A were tested for binding to ACBD3. In this system, interaction between a transcription activation domain-fused protein expressed from a pACT construct and a DNA-binding domain-fused protein expressed from a pBIND construct results in transcription of the reporter firefly luciferase gene. Interaction between 3A and ACBD3 induced a 270-fold increase in luciferase activity compared with the negative control (Figure 1A). Interestingly, strong activation of luciferase expression (40- to 158-fold increase in luciferase activity) was observed also for 2B, 2BC, 2C, and 3AB. Figure 1.2B, 2BC, 2C, 3A, and 3AB interact with ACBD3. (A) The mammalian two-hybrid assay. The indicated combination of a pACT construct and a pBIND construct was transfected into Vero cells together with pG5luc encoding a firefly luciferase. Cell lysates were prepared at 48 h after transfection and assayed for firefly luciferase activity. Transfection efficiency was normalized by the activity of Renilla luciferase, which was simultaneously expressed from pBIND. The higher value of normalized luciferase activities obtained in cells transfected with the combination of the pBIND construct and empty pACT and with the combination of the pACT construct and empty pBIND was used as a negative control. The normalized firefly luciferase activity was represented as fold activation compared with a negative control. The experiment was repeated at least three times. Standard deviation bars are shown. (B) Co-immunoprecipitation of ACBD3 with 2B, 2BC, 2C, 3A, or 3AB. FLAG-tagged ACBD3 was co-expressed with HA-tagged L, 2B, 2BC, 2C, 3A, or 3AB. Proteins were immunoprecipitated with anti-FLAG (upper panel), anti-HA (lower panel) antibodies, or control IgG, and the resulting immunoprecipitates and whole cell lysates were analysed by immunoblotting with anti-FLAG and anti-HA antibodies. IB, immunoblotting; IP, immunoprecipitation. (C) MBP pull-down assay. MBP-fused viral proteins or MBP immobilized on amylose resin were mixed with purified GST–ACBD3, and proteins binding to the resin were analysed by SDS–PAGE, followed by immunoblotting with anti-GST antibody (upper panel). After immunoblotting, proteins on a PVDF membrane were stained with Coomassie brilliant blue to detect MBP-fused viral proteins or MBP (lower panel). Figure source data can be found in Supplementary data. Source Data for Figure 1c [embj2011429-sup-0001.pdf] Download figure Download PowerPoint To further confirm the interaction of these viral proteins with ACBD3, FLAG-tagged ACBD3 was co-expressed with HA-tagged L, 2B, 2BC, 2C, 3A, or 3AB in 293T cells, and then a co-immunoprecipitation assay was performed. Consistent with the results of the mammalian two-hybrid assay, 2B, 2BC, 2C, 3A, and 3AB were co-immunoprecipitated with ACBD3, but the L protein was not (Figure 1B). Furthermore, to examine whether these viral proteins interact with ACBD3 directly, a maltose binding protein (MBP) pull-down assay was performed using glutathione S-transferase (GST)-fused ACBD3 (GST–ACBD3) and MBP-fused viral proteins (MBP–2B, MBP–2C, MBP–3A, and MBP–3AB) expressed in Escherichia coli. Expression of MBP-fused 2BC was not enough to use for this experiment. GST–ACBD3 was pulled down with the MBP-fused viral proteins, but not with MBP (Figure 1C). These results indicate that 2B, 2BC, 2C, 3A, and 3AB have the ability to interact with ACBD3. 2B, 2BC, 2C, 3A, and 3AB interact with the C-terminal region of ACBD3 To determine the region in ACBD3 required for binding to the viral proteins, the ability of various deletion mutants of ACBD3 to interact with 2B, 2BC, 2C, 3A, or 3AB was investigated in the mammalian two-hybrid system (Figure 2A and B). For 2B, 2C, 3A, or 3AB, mut3 and mut5, both of which contain a C-terminal region (aa 327–528), maintained the ability to induce a greater than ∼15-fold increase in luciferase activity compared with negative controls. Also for 2BC, 7- and 4-fold increases in luciferase activity were detected with mut3 and mut5, respectively. On the other hand, mutants without the C-terminal region lacked the ability to enhance luciferase expression. These results suggest that the C-terminal region of ACBD3 is essential for binding to 2B, 2BC, 2C, 3A, and 3AB. Further detailed experiments to identify the binding domain to these viral proteins were carried out using a series of deletion mutants of mut3 (Figure 2C and D). For the five viral proteins, mut3Δ1 with the deletion of aa 328–373 enhanced luciferase expression to a similar level to mut3, but the other mutants did not. This suggests that aa 374–528 of ACBD3 are important for binding to the viral proteins. Figure 2.The C-terminal region of ACBD3 interacts with 2B, 2BC, 2C, 3A, and 3AB. (A, C) Schematic representation of (A) full-length ACBD3 (WT) and its mutants (mut1–mut5) and (C) mut3 (amino acids 328–528) and its mutants (mut3Δ1–Δ5). ACBD3 contains characteristic domains as follows: PR, proline-rich domain; ACB, ACB region; CAR, charged amino acid-rich domain; QR, glutamine-rich domain; and GOLD, Golgi dynamic domain. Numbers indicate amino-acid positions. The region between amino acids 373 and 528 is required for binding giantin. (B, D) Mammalian two-hybrid analyses were performed to examine interactions (B) between the ACBD3 mutants (mut1–5) and 2B, 2BC, 2C, 3A, or 3AB, and (D) between the mut3 mutants (mut3Δ1–Δ5) and 2B, 2BC, 2C, 3A, or 3AB, and the results are represented as described in Figure 1A. All experiments were repeated at least three times. Standard deviation bars are shown. Download figure Download PowerPoint ACBD3 colocalizes with 2B, 2C, 3A, and dsRNA in AiV RNA-transfected cells To investigate whether ACBD3 colocalizes with viral proteins or replicating viral RNA, an AiV replicon RNA (AV-FL-Luc-5′rzm) containing a firefly luciferase gene was transfected into Vero cells by electroporation, and then an immunofluorescence assay was carried out. The replicating viral RNA was detected by staining double-stranded replicative intermediates and replicative forms using an antibody recognizing dsRNA in which the helix length is >40 bp, as carried out for other positive-stranded RNA viruses including picornavirus (Miller et al, 2006; Harwood et al, 2008; Knoops et al, 2008; Berger et al, 2009; DeWitte-Orr et al, 2009, Hyde et al, 2009). 2B, 2C, and 3A were detected as patchy clusters in the cytoplasm at 4 h after transfection (Figure 3A), and accumulated in the perinuclear region at 6 h (only the data for 3A are shown in Figure 3A). At 4 h, when viral RNA replicates actively (see Figure 5A or C), 2B, 2C, and 3A colocalized with dsRNA (Figure 3B). ACBD3, which also formed patchy clusters in the cytoplasm at 4 h after transfection, colocalized with 2B, 2C, 3A (Figure 3A), and dsRNA (Figure 3B). These results indicate that ACBD3, 2B, 2C, and 3A (and possibly also the precursor proteins 2BC and 3AB) are present in the viral RNA replication sites. Figure 3.(A) Colocalization of ACBD3 with 2B, 2C, and 3A. Vero cells were electroporated with replicon RNA, AV-FL-Luc-5′rzm. At 4 or 6 h after electroporation, the cells were fixed and double stained with rabbit anti-ACBD3 and guinea pig anti-3A, anti-2B, or anti-2C antibodies. (B) Colocalization of dsRNA with ACBD3, 2B, 2C, or 3A. Vero cells were electroporated with replicon RNA. At 4 h, the cells were fixed and double stained with anti-dsRNA (mouse) and anti-2B, anti-2C, anti-3A, or anti-ACBD3 antibodies (rabbit). Bars represent 30 μm. (C) Effect of knockdown of ACBD3 on AiV replication. Vero cells were transfected with 80 nM of either control siRNA or siRNA against ACBD3. At 72 h after transfection, lysates were prepared and subjected to immunoblotting to assess the levels of ACBD3 and α-tubulin (left panel). At 72 h post transfection with siRNAs, the cells were transfected with replicon RNA and then luciferase activity in cell lysates harvested at the indicated times was measured (right panel). The maximum value obtained for cells treated with control siRNA was taken as 100%. The experiment was repeated at least three times. Standard deviation bars are shown. Download figure Download PowerPoint Knockdown of ACBD3 inhibits AiV RNA replication To investigate whether ACBD3 is involved in AiV RNA replication, replication of AV-FL-Luc-5′rzm RNA was examined in Vero cells treated with small interfering RNA (siRNA) targeting ACBD3 or control siRNA. At 72 h after treatment with siRNA targeting ACBD3, the amount of ACBD3, but not that of α-tubulin, was apparently decreased (Figure 3C). An effect of knockdown of ACBD3 on cell viability was not observed (data not shown). At 72 h after treatment with siRNA targeting ACBD3 or control siRNA, cells were transfected with AV-FL-Luc-5′rzm RNA by lipofection, and then cell lysates were prepared at various times after transfection and subjected to the luciferase assay. Knockdown of ACBD3 resulted in a decrease in luciferase activity. At 10 h after transfection, luciferase activity was decreased by ∼70% (Figure 3C). Here, Vero cells were used to exclude the effect of siRNA-induced interferon response, but a similar result was obtained using HeLa cells (data not shown). These results indicate that ACBD3 plays an important role in AiV RNA replication. Localization of other Golgi proteins in AiV RNA-transfected cells ACBD3 localizes to the Golgi through interaction with giantin, a Golgi protein, and the giantin-binding domain of ACBD3 has been mapped to the C-terminal region (aa 373–528) (Sohda et al, 2001). This region overlaps with the region important for binding to the AiV proteins (aa 374–528) (Figure 2). We then examined localization of giantin in viral RNA-replicating cells (Figure 4A). To distinguish RNA-transfected cells, a viral non-structural protein, the L protein, was immunostained. In mock-transfected cells, both ACBD3 and giantin were located in the Golgi. At 2 h after electroporation with viral RNA, the dispersion of ACBD3 and giantin was observed in some cells; however, the L protein could not be detected, because of its insufficient accumulation. At this time point, ACBD3 colocalized with giantin. As infection progressed, the redistribution of ACBD3 and giantin from the Golgi to the cytoplasm became more apparent. Importantly, at 4 h, giantin was dispersed throughout the cytoplasm, whereas ACBD3 formed clusters. As shown in Figure 3, ACBD3 colocalized with dsRNA or the viral proteins in such structures. At 4 h after electroporation, the intensity of immunofluorescence for giantin seems to be decreased compared with that in mock-transfected cells; however, no degradation of giantin was observed in an immunoblot analysis using anti-giantin antibody (data not shown). Figure 4.Dynamics of different Golgi proteins during AiV RNA replication. Vero cells were mock electroporated or electroporated with the replicon RNA, AV-FL-Luc-5′rzm. At 2 or 4 h after electroporation, the cells were fixed and triple stained with rabbit anti-ACBD3, guinea pig anti-L, and (A) mouse anti-giantin, (B) mouse anti-GM130, or (C) sheep anti-TGN46 antibodies. Bars represent 30 μm. Download figure Download PowerPoint We also examined the localization of other Golgi proteins, GM130, a cis-Golgi marker, and TGN46, a trans-Golgi marker. At 2 h after transfection with replicon RNA, dispersion of GM130 and TGN46 appears to begin like giantin, and they colocalized with ACBD3 (Figure 4B and C). GM130 was redistributed to form clusters in the cytoplasm at 4 h after transfection, but did not colocalize with ACBD3 (Figure 4B). TGN46 also became dispersed in the cytoplasm with the progression of infection, and did not colocalize with ACBD3 (Figure 4C). Thus, in viral RNA-replicating cells, other Golgi proteins examined did not colocalize with ACBD3. AiV RNA replication requires PI4KB activity It was recently reported that PI4KB is an essential host factor for enterovirus RNA replication (Hsu et al, 2010; Arita et al, 2011). We investigated whether PI4KB is also important for AiV replication. First, we examined the effect of a PI4KB-specific inhibitor, T-00127-HEV1 (Arita et al, 2011), on viral RNA replication. Vero cells were electroporated with AV-FL-Luc-5′rzm RNA, and then cultured in medium containing 0, 1, or 5 μM T-00127-HEV1. T-00127-HEV1 inhibited AiV RNA replication in a dose-dependent manner, and the replication was almost completely inhibited at 5 μM (Figure 5A). Cell viability was not affected by treatment with 5 μM (data not shown). It is noted that at 1 h after electroporation, the level of luciferase activity observed for T-00127-HEV1-treated cells was similar to that for cells treated with DMSO, and was significantly higher than that for mock-transfected cells. At this time point, viral RNA replication does not occur, and the obtained luciferase activity mainly results from translation of input RNA (Sasaki and Taniguchi, 2008). Thus, this result suggests that PI4KB activity is important for AiV RNA replication, but not for translation. Figure 5.(A) T-00127-HEV1 inhibits AiV RNA replication. Vero cells were mock electroporated or electroporated with the replicon RNA, AV-FL-Luc-5′rzm, and then treated with 0 (DMSO), 1 or 5 μM T-00127-HEV1. After incubation for the indicated times, the cells were assayed for luciferase activity. (B) Depletion of endogenous PI4KB severely reduces AiV RNA replication. Vero cells were transfected with 40 nM control siRNA or siRNA against PI4KB. At 72 h after transfection, the levels of PI4KB and α-tubulin were determined by immunoblotting (right panel), and the subsequent analysis was performed as in Figure 3C (left panel). (C) AiV RNA replication is insensitive to BFA. Vero cells were electroporated with poliovirus (PV) (upper panel) or AiV (lower panel) replicon RNA and then incubated with or without 10 μg/ml of BFA. After incubation for the indicated times, the cells were assayed for luciferase activity. All experiments were repeated at least three times. Standard deviation bars are shown. Download figure Download PowerPoint Furthermore, we investigated viral RNA replication in PI4KB knockdown cells. Vero cells were treated with siRNA targeting PI4KB or control siRNA. At 72 h after treatment with siRNA targeting PI4KB, a decrease in the amount of PI4KB was observed (Figure 5B). Cell viability was not affected by treatment with siRNA targeting PI4KB (data not shown). At 72 h after siRNA treatment, AV-FL-Luc-5′rzm RNA was transfected by lipofection, and RNA replication was examined by a luciferase assay. Knockdown of PI4KB almost completely inhibited AiV RNA replication, and at 10 h after transfection, luciferase activity was decreased by 99% (Figure 5B). A similar result was obtained using HeLa cells (data not shown). These results indicate that PI4KB activity is essential for AiV RNA replication. PI4KB interacts with ACBD3 BFA is an inhibitor of enterovirus RNA replication (Irurzun et al, 1992; Maynell et al, 1992). BFA inactivates GBF1 (Peyroche et al, 1999) and inhibits recruitment of GBF1/Arf1 to 3A, resulting in inhibition of recruitment of PI4KB to the replication sites of enteroviruses (Belov et al, 2008; Hsu et al, 2010). We investigated whether BFA inhibits AiV replication. Treatment with 10 μg/ml of BFA inhibited poliovirus RNA replication, as known, but did not affect AiV RNA replication at all (Figure 5C). This result suggests that PI4KB is recruited independently of GBF1/Arf1 to the AiV replication sites. As described above, ACBD3 was localized to the Aichi viral RNA replication sites through binding to some viral non-structural proteins. We examined whether ACBD3 interacts with PI4KB. In the mammalian two-hybrid assay, interaction between ACBD3 and PI4KB resulted in a 138-fold increase in luciferase activity, indicating a strong interaction between the two proteins (Figure 6A). No apparent activation of luciferase expression was detected for viral non-structural proteins (3.5-fold increase for 3A and less for other proteins). To further confirm the interaction between ACBD3 and PI4KB, HA-tagged PI4KB was co-expressed with FLAG-tagged ACBD3 or FLAG-tagged 3A in 293T cells, and then co-immunoprecipitation analysis was performed. As shown in Figure 6B, PI4KB was co-immunoprecipitated with ACBD3, but not with 3A, confirming the interaction between ACBD3 and PI4KB. Figure 6.PI4KB interacts with ACBD3. (A) The mammalian two-hybrid assay was performed to examine interactions between PI4KB and ACBD3 or the virus proteins and the results are represented as in Figure 1A. (B) Co-immunoprecipitation of PI4KB with ACBD3. HA-tagged PI4KB was co-expressed with FLAG-tagged ACBD3 or FLAG-tagged 3A in 293T cells, and the subsequent analysis was performed as in Figure 1B. (C) Mapping of the PI4KB-binding region of ACBD3. Schematic representation of full-length ACBD3 (WT) and its mutants (mut6 and mut7) (upper panel). Mammalian two-hybrid analysis was performed to examine interactions between PI4KB and ACBD3 WT or its mutants (mut1–7), and the results are represented as described in Figure 1A (lower panel). Figure source data can be found in Supplementary data. Source Data for Figure 6b [embj2011429-sup-0002.pdf] Download figure Download PowerPoint PI4KB interacts with the central part of ACBD3 To determine the PI4KB-binding domain of ACBD3, interactions of deletion mutants of ACBD3 with PI4KB were investigated in the mammalian two-hybrid system (Figure 6C). Of the five mutants shown in Figure 2A, mut2 and mut5 activated luciferase expression strongly (∼15- and 45-fold, respectively), implying the importance of a wider region including aa 171–327 of ACBD3 for binding to PI4KB. Additional two mutants, mut6 and mut7, which contain aa 116–327 and aa 116–428, respectively, exhibited the ability to strongly activate luciferase expression (24- and 32-fold, respectively). These results suggest that the central part of ACBD3 (aa 116–327) is important for binding to PI4KB. In addition, the PI4KB-binding domain was suggested not to overlap with the binding domain for 2B, 2BC, 2C, 3A, and 3AB (aa 374–528). PI4KB colocalizes with ACBD3, 2B, 2C, 3A, and dsRNA in AiV RNA-transfected cells We examined whether PI4KB localizes to the viral RNA replication sites. In mock-transfected cells, PI4KB mainly localized to the Golgi, and colocalized with ACBD3 (Figure 7A). When ACBD3 expression was knockdown by using siRNA, PI4KB was dispersed in the cytoplasm and did not localize to the Golgi (Figure 7B), strongly suggesting a critical role of ACBD3 in Golgi recruitment of PI4KB. In viral RNA-replicating cells, PI4KB colocalized with ACBD3, 2B, 2C, 3A, and dsRNA (Figure 7A and C). Treatment of mock-transfected cells with BFA resulted in dispersion of ACBD3 and PI4KB to the cytoplasm (Figure 7D). In contrast, in viral RNA-transfected cells with BFA treatment, ACBD3 and PI4KB formed clusters in the cytoplasm, and colocalized with dsRNA (Figure 7E). These results indicate that PI4KB is present in the viral RNA replication sites through interaction with ACBD3, which is recruited through interaction with 2B, 2BC, 2C, 3A, and 3AB. In addition, it was shown that recruitment of PI4KB and ACBD3 to the AiV RNA replication sites was not affected by BFA. Figure 7.PI4KB is recruited to the AiV RNA replication sites through interaction with ACBD3 associated with the viral proteins. (A) Colocalization of PI4KB with ACBD3. Vero cells were mock electroporated or electroporated with the replicon RNA, AV-FL-Luc-5′rzm. At 4 h after electroporation, the cells were fixed and triple stained with mouse anti-PI4KB, rabbit anti-ACBD3, and guinea pig anti-L antibodies. (B) Effect of ACBD3 knockdown on the localization of PI4KB. Vero cells were transfected with siRNA against ACBD3. At 72 h after transfection, cells were fixed and double stained with anti-ACBD3 and anti-PI4KB. Asterisks indicate cells where knockdown of the expression of ACBD3 was observed. (C) PI4KB colocalizes with 2B, 2C, 3A, or dsRNA.