Foot-and-mouth disease virus replicates independently of phosphatidylinositol 4-phosphate and type III phosphatidylinositol 4-kinases

Picornaviruses form replication complexes in association with membranes in structures called replication organelles. Common themes to emerge from studies of picornavirus replication are the need for cholesterol and phosphatidylinositol 4-phosphate (PI4P). In infected cells, type III phosphatidylinositol 4-kinases (PI4KIIIs) generate elevated levels of PI4P, which is then exchanged for cholesterol at replication organelles. For the enteroviruses, replication organelles form at Golgi membranes in a process that utilizes PI4KIIIβ. Other picornaviruses, for example the cardioviruses, are believed to initiate replication at the endoplasmic reticulum and subvert PI4KIIIα to generate PI4P. Here we investigated the role of PI4KIII in foot-and-mouth disease virus (FMDV) replication. Our results showed that, in contrast to the enteroviruses and the cardioviruses, FMDV replication does not require PI4KIII (PI4KIIIα and PI4KIIIβ), and PI4P levels do not increase in FMDV-infected cells and PI4P is not seen at replication organelles. These results point to a unique requirement towards lipids at the FMDV replication membranes.

PI4P is generated at RO through the activity of lipid kinases. PI4KIIIb is localized at RO in enterovirus-infected cells, and inhibition or depletion of PI4KIIIb inhibits replication (Dorobantu et al., 2015;Greninger et al., 2012;Roulin et al., 2014;Sasaki et al., 2012;Spickler et al., 2013;Strating et al., 2015;van der Schaar et al., 2013;Wessels et al., 2006). In contrast, EMCV has been shown not to require PI4KIIIb, instead relying on PI4KIIIa to generate PI4P at RO (Dorobantu et al., 2015). The requirement for different isoforms of PI4K may reflect the differential use of ER and Golgi membranes as the initial replication site, as PI4KIIIa is predominantly located on the ER whereas PI4KIIIb is enriched on the Golgi (Balla, 2013;Godi et al., 1999). In addition, replication of certain HRV appears to require both PI4KIIIb and PI4KIIIa (Roulin et al., 2014).
FMDV is the prototype species of the Aphthovirus genus of the Picornaviridae and the causative agent of foot-and-mouth disease (FMD), which affects cattle, sheep, goats and pigs. FMD is endemic in many regions of the world and causes enormous economic losses to livestock industries (Knight-Jones & Rushton, 2013). Similarly to other picornaviruses, electron microscopy studies have shown that FMDV infection of (BHK) cells (Monaghan et al., 2004) and IBRS2 cells (Martin-Acebes et al., 2008) generates large numbers of single-and double-membrane vesicles, which are likely RO. The ER-Golgi intermediate compartment (ERGIC) and Golgi are disrupted early after infection and before major changes are detected in the ER (Midgley et al., 2013;Zhou et al., 2013). However, replication of FMDV appears to differ from that of the enteroviruses in a number of ways; for example, enterovirus replication requires an intact Golgi and is inhibited by Golgi disruption caused by inhibition of Arf1 or by brefeldin A (BFA) (Cuconati et al., 1998;Doedens et al., 1994;Gazina et al., 2002;Irurzun et al., 1992;Lanke et al., 2009;Maynell et al., 1992). In contrast, FMDV (along with the cardioviruses) is unusual among the picornaviridae by being resistant to BFA, and infection does not require an intact Golgi (Gazina et al., 2002;Martin-Acebes et al., 2008;Midgley et al., 2013;Monaghan et al., 2004;O'Donnell et al., 2001). Furthermore, we and others have shown that Golgi disruption appears to make FMDV replication more favourable as infection is enhanced by disruption resulting from either expression of dominant negative Arf1 or by BFA (Martin-Acebes et al., 2008;Midgley et al., 2013). These observations suggest that if FMDV requires membranes of the early secretory pathway to form RO they likely derive from pre-Golgi membranes such as the ER or ERGIC. The ER comes in to close apposition with other organelles, including the Golgi, at membrane contact sites (reviewed by Phillips & Voeltz, 2016). Lipid exchange at ER-Golgi contact sites by transport proteins such as OSBP is important for intracellular lipid homeostasis (Mesmin et al., 2013). As described above, this process has been hijacked by enteroviruses which utilize PI4K to synthesize PI4P at RO, which is then exchanged for cholesterol by OSBP, thus driving accumulation of cholesterol at RO (Roulin et al., 2014).
Here, we have investigated the role of PI4P and PI4K in FMDV replication. Our results show that, in contrast to the enteroviruses, cardioviruses and kobuviruses, PI4P levels do not increase in FMDV-infected cells and PI4P is not present at RO. Consistent with these observations, we also show that FMDV infection does not require PI4KIIIa or PI4KIIIb. These results point to a unique requirement towards lipids at the FMDV replication membranes.

Investigating components of FMDV RO
There is little information describing formation of RO in FMDV-infected cells Martin-Acebes et al., 2008;Midgley et al., 2013;Monaghan et al., 2004). Therefore, our initial studies focused on identifying the viral proteins and host factors present at RO. For these studies we used IBRS2 cells. FMDV replication in IBRS2 cells is relatively fast, and infection results in cell death by 5-6 h. Therefore, we investigated RO at 3.5-4.0 h post-infection (p.i.), when replication has been shown to be maximal for IBRS2 cells (Midgley et al., 2013;Rodriguez Pulido et al., 2009). Consistent with these observations, at this time post-infection, labelling for dsRNA (which is produced during infection; Agol et al., 1999) is greatly enhanced in FMDV-infected cells (Fig. S1, available in the online Supplementary Material), as detected using mAb J2 (which recognizes lengths of dsRNA !40 bp). Fig. 1(a) shows infected cells co-labelled for dsRNA using mAb J2 and the viral non-structural proteins (nsp), 2C and 3D. These proteins are established markers of RO and might be expected to colocalize with dsRNA; however, for both proteins, little or no co-localization with dsRNA was observed. Fig. 1(b) also shows labelling for dsRNA combined with in-situ hybridization for positive-strand viral RNA. Again, although they appeared in close proximity, little co-localization was seen. To investigate this further, we determined co-localization of viral nsp with positive-strand RNA and observed extensive co-localization of positive-strand RNA with 2C and 3D (Fig. 2). The 3A protein was also co-localized with positivestrand RNA; however, significant 3A labelling was also seen distinct from the positive-strand signal. In addition, positive-strand RNA was extensively co-localized with the capsid protein VP1. These results show that in FMDV-infected cells, dsRNA is not localized with accepted markers of RO (i.e. viral nsp) or with positive-strand RNA. Furthermore, positive-strand RNA shows extensive co-localization with both nsp and capsid proteins suggesting that in FMDVinfected cells, dsRNA and other viral components are spatially segregated at RO.
PI4KIIIb is required for enterovirus replication and has been shown to co-localize at RO with viral nsp such as 3A and 3D (see Introduction). In IBRS2 cells, labelling for PI4KIIIb showed the expected peri-nuclear pattern consistent with a Golgi localization (data not shown). However, in contrast to enteroviruses, although labelling was dispersed in FMDV-infected cells, co-localization of PI4KIIIb at RO with nsp (2C or 3D) was not seen (Fig. 3). As a result of PI4KIIIb activity in enterovirus-infected cells, PI4P levels are increased and PI4P is also present at RO (Hsu et al., 2010). Similar to PI4KIIIb, labelling for PI4P was predominantly peri-nuclear in uninfected IBRS2 cells and dispersed by FMDV infection. Furthermore, PI4P was not co-localized at RO with the viral 2C protein (Fig. 4a). Similar observations were made for dsRNA, as co-localization with PI4P was not seen in infected cells (Fig. S2). Fig. 4(b) shows an analysis of intracellular PI4P levels for cells infected with FMDV or bovine enterovirus (BEV). Consistent with previous reports for enteroviruses, in BEV-infected cells PI4P was dispersed and PI4P levels were found to be elevated. However, in contrast to BEV, although PI4P was dispersed, PI4P levels did not appear altered by FMDV infection (Fig. 4b).
Together, these results show that intracellular PI4P levels are not increased in FMDV-infected cells, and PI4KIIIb and PI4P are not associated with RO suggesting that, unlike the enteroviruses, FMDV does not require PI4KIIIb or PI4P for infection.

Investigating the role of PI4K in FMDV infection
To further investigate the role for PI4K in FMDV infection we used PIK93, which inhibits PI4KIIIb (Balla et al., 2008). For this study we included BEV, as enteroviruses require PI4KIIIb for replication and infection is inhibited by PIK93 (Roulin et al., 2014;van der Schaar et al., 2013). Consistent with observations with other enteroviruses, PIK93 had a potent inhibitory effect on BEV infection (Fig. 5a), including when added 1 h after infection. In contrast, PIK93 did not inhibit FMDV infection. We also compared the effect of PIK93 on FMDV replication using a luciferase-based subgenomic replicon (Fig. 5b). For comparison, we included a Coxsackievirus B3 (CVB3) replicon (Lanke et al., 2009) as CVB3 requires PI4KIIIb for replication and is inhibited by PIK93 (Hsu et al., 2010;van der Schaar et al., 2013). PIK93 inhibited CVB3 replication by~90 % at 4 µM, while FMDV replication was not inhibited at this concentration. However, PIK93 also inhibited FMDV replication, but to a lesser extent and only by~40 % at 8 µM. At higher concentrations, PIK93 is known to also inhibit PI4KIIIa (Balla et al., 2008), and the reduced luciferase signal from the FMDV replicon caused by higher concentrations of PIK93 could indicate a role for PI4KIIIa in replication. Therefore, we also investigated if FMDV infection requires PI4KIIIa using AL-9, which preferentially targets PI4KIIIa (Bianco et al., 2012). This analysis showed that at 10 µM, AL-9 had a small inhibitory effect on BEV infection but not on FMDV (Fig. 5c). Similarly, at 10 µM AL-9 had a small inhibitory effect on the FMDV (~25 %) replicon ( Fig. 5d). At this concentration, AL-9 also inhibited the CVB3 replicon by~65 %. However, although at low concentrations AL-9 preferentially targets PI4KIIIa, at 10 µM, AL-9 is reported to also inhibit PI4KIIIb (Bianco et al., 2012) suggesting that the effect on CVB3 may be mediated through inhibition of PI4KIIIb. This conclusion is supported by the observations that hepatitis C virus replication requires PI4KIIIa and is inhibited by much lower concentrations of AL-9 (IC 50 <1 uM) (Bianco et al., 2012). Together, the above observations suggest that the small inhibitory effects of PIK93 on FMDV infection/replication more likely result from non-specific, off-target effects and support the conclusion that FMDV does not require PI4KIIIb or PI4P for replication. Furthermore, our results suggest that FMDV replication is also independent of PI4KIIIa.
To further investigate the role of PI4KIIIa and PI4KIIIb in FMDV infection we used Huh7 cell lines where levels of PI4KIIIa were reduced by small hairpin RNA (shRNA) (Reiss et al., 2013), or expression of PI4KIIIb inhibited by gene knockout (Esser-Nobis et al., 2015). Western blot analysis confirmed the reduced levels of PI4KIIIa and PI4KIIIb in these cell lines (Fig. 6a, b). Next, we infected these cells with either FMDV or BEV and processed samples for confocal microscopy, labelling for viral proteins. For each matched pair of cells (parental and PI4K-depleted), the proportion of cells infected was determined and normalized to the parental cells. This analysis showed that, consistent with previous findings that show enteroviruses require PI4KIIIb for replication, the PI4KIIIb knockout cells were infected poorly by BEV (Fig. 6c). In contrast, infection by FMDV was reduced by only~30 % confirming that FMDV does not require PI4KIIIb for infection (Fig. 6c). Fig. 6 also shows that reduced expression of PI4KIIIa did not affect infection by either FMDV or BEV (Fig. 6d). Together, these results confirm that FMDV does not require the catalytic activity of PI4KIIIa or PI4KIIIb for infection, and demonstrate additionally that infection does not require any non-enzymic activities of these proteins.

DISCUSSION
To date, the best-studied picornaviruses are the enteroviruses and they have been invaluable as models to study picornavirus replication. However, the Picornaviridae is now recognized as one of the largest and most diversified virus families, and since 2012 the number of genera has increased from 12 (including 28 species) to the current 29, including at least 50 species (Knowles et al., 2012). It is now clear that, although picornaviruses share a common replication strategy, different host factors and cellular pathways contribute to the replication of different genera. Despite these differences, common themes to emerge from studies of RO formation are the requirements  Cells were labelled for intracellular PI4P using a PI4P-specific antibody following permeabilization with digitonin. Nuclei are blue. Bars, 10 µm. (a) FMDV-infected cells were double-labelled for intracellular PI4P (green, i) and for FMDV 2C protein (red, ii). Images shown are single optical sections through the middle of the cell. (b) Intracellular PI4P was quantified by immunofluorescence analysis. Mock-infected, FMDV-infected or BEV-infected IBRS2 cells (m.o.i.=0.6) were labelled for intracellular PI4P. FMDV-infected cells were identified by labelling for FMDV 3A, and BEV-infected cells were identified by labelling using a guinea pig antiserum raised against whole virus. Example images taken from mock-infected (i), FMDV-infected (ii) and BEV-infected samples (iii) are shown as maximum projections of z-stacks, with PI4P in green and infected cells distinguished by white arrows and by red labelling in insets. (iv) Intensities of intracellular fluorescent PI4P signals from whole-cell z-stacks (spacing 0.3 µm) were quantified from at least 10 cells per condition using Imaris image analysis software. Error bars show standard deviation. P values were calculated using one-way ANOVA. Significant difference (95 % confidence level using TUKEY-style multiple comparisons test) was found between FMDV and BEV (****P<0.0001), and control and BEV (****P<0.0001). There was no significant difference between FMDV and the control (not significant P>0.05). AU, Arbitrary units. for PI4K, PI4P and cholesterol. For example, in enterovirusinfected cells, recruitment of PI4KIIIb leads to the local production of PI4P at RO and the subsequent recruitment of viral (e.g. 3D) and host proteins (e.g. OSBP) that participate in viral genome replication (see Introduction). In addition, kobuviruses and cardioviruses have recently been shown to require PI4K (PI4KIIIb and PI4KIIIa, respectively) for replication (Dorobantu et al., 2015;Greninger et al., 2012;Sasaki et al., 2012), which likely generate PI4P at RO. Here we have shown that FMDV appears to use a novel replication strategy that does not require PI4KIIIa or PI4KIIIb. Furthermore, we have shown that intracellular levels of PI4P were not elevated in FMDV-infected cells and PI4P is not localized at RO. Recently, HAV was also shown to replicate independently from PI4KIIIa or PI4KIIIb (Esser-Nobis et al., 2015) suggesting that picornaviruses from diverse genera (FMDV and HAV) have evolved to replicate independently from type III PI4Ks.
The observation that FMDV does not require PI4KIIIa is consistent with a previous report that showed that ERAV infection was not inhibited by AL-9 (Dorobantu et al., 2015). It is possible that FMDV could use either PI4KIIa or PI4KIIb in place of type III PI4Ks used by other picornaviruses; however, this is unlikely since we have also shown that levels of intracellular PI4P were not elevated in FMDV-infected cells and PI4P is not localized at RO. For the enteroviruses and cardioviruses, the viral 3A protein has a key role in formation of the RO and recruits PI4K (Hsu et al., 2010), and 3A of PV, HRV and CVB3 have been shown to recruit PI4KIIIb to RO, while the 3A of EMCV recruits PI4KIIIa (Dorobantu et al., 2015;Greninger et al., 2012). Interestingly, a single point mutation in 3A of CVB3 confers resistance to drugs targeting PI4KIIIb and also enables the virus to replicate independently of PI4KIIIb and raised levels of PI4P on internal membranes, which are required by the parental virus (van der Schaar et al., 2012). Thus, these drug-resistant variants of CVB3 behave similarly to FMDV in that they demonstrate the ability to replicate independently of PI4K activity or elevated intracellular levels of PI4P. The FMDV 3A protein is much longer than that of enteroviruses and cardioviruses and appears functionally distinct as the enterovirus 3A protein inhibits protein secretion, whereas for FMDV, secretion is not blocked by 3A but instead by 2B and 2C (Moffat et al., , 2007. Given these differences, it is interesting to speculate that FMDV 3A may recruit alternative host factors that enable virus replication independently of PI4K and PI4P. Our results show that the membranes of FMDV RO lack PI4P and therefore have a different lipid composition to those found in enterovirus-infected cells. In enterovirus-infected cells PI4P is exchanged for cholesterol at RO by OSBP, and inhibition of OSBP by 25-hydroxycholesterol (25-HC) inhibits virus replication (Arita et al., 2013;Civra et al., 2014;Roulin et al., 2014). 25-HC targets cholesterol synthesis in addition to cholesterol/PI4P exchange by OSBP; however, enterovirus replication is unaffected by other cholesterol synthesis inhibitors such as AY9944 and compactin (Roulin et al., 2014), whilst remaining sensitive to the OSBP inhibitor OSW-1 (Albulescu et al., 2015a) and knockdown of OSBP by siRNA (small interfering RNA) (Roulin et al., 2014;Strating et al., 2015). This indicates that the effect of 25-HC on enterovirus replication is mediated by its impact on OSBP activity. Interestingly, our preliminary data suggests that FMDV is less sensitive than BEV to 25-HC (data not shown), and therefore may not require OSBP for replication. These observations are consistent with previous studies showing that ERAV (also genus Aphthovirus) replicates independently of OSBP and provides further evidence that aphthovirus RO form by different mechanisms to those of enteroviruses. However, further study will be required to determine the role of OSBP and cholesterol in FMDV replication if any.
Our results show that in FMDV-infected cells, positivestrand vRNA localized with 3A, 2C and 3D, which are accepted markers of picornavirus RO. However, neither positive-strand vRNA nor nsp was co-localized with dsRNA. This is a little surprising as dsRNA and the nsp might be expected to be present at RO. In contrast to our results, it has been reported that dsRNA is co-localized with viral nsp in Aichi virus- (Sasaki et al., 2012) and poliovirusinfected cells (Richards et al., 2014), and with PI4KIIIb in HRV-and echovirus-infected cells (Ilnytska et al., 2013). However, although in HRV-infected cells most PI4KIIIb was in close proximity with dsRNA, some dsRNA labelling appeared not to be associated with PI4KIIIb (Ilnytska et al., 2013). In addition, more recently, in cells infected with EMCV, dsRNA was shown to be in close association but not extensively overlapping with PI4KIIIa, a marker of RO that co-localizes extensively with EMCV 3AB (Dorobantu et al., 2015). This last report closely supports our data in suggesting that for some picornaviruses, including EMCV and FMDV, dsRNA labelling is spatially segregated from viral replication proteins as shown previously for coronaviruses (Hagemeijer et al., 2012;Knoops et al., 2008).
In summary, our data show that FMDV uses a novel approach to generate RO. In contrast to enteroviruses and cardioviruses, FMDV does not require PI4KIIIa or PI4KIIIb for infection and PI4P is not elevated or localized to RO in infected cells, demonstrating the possibility of a variety of different RO lipid compositions in the family Picornaviridae.

Antibodies and reagents
Murine mAbs against dsRNA (mAb J2), PI4KIIIb and PI4P were purchased from English and Scientific Consulting, BD Transduction Laboratories, and Echelon biosciences, respectively. Rabbit antibody against PI4KIIIa for Western blotting was obtained from Cell Signalling Technology. The murine mAb against actin was from Sigma.

Immunofluorescence confocal microscopy
Infection of cells. Cells grown on glass coverslips were inoculated with the indicated m.o.i. of FMDV O1K or BEV-1 diluted in cell culture medium containing 1 % FCS (FMDV) or 1 % porcine serum (BEV-1) for 1 h at 37 C. Unbound virus or capsid was removed by washing, and the cells were returned to 37 C in cell culture medium with reduced serum for the remainder of the infection time. At the end of the infection time, cells were fixed in 4 % paraformaldehyde for 40 min.
Antibody labelling. Following fixation, cells were permeabilized for 5 min using digitonin (50 µg ml À1 ) when labelling for PI4P, or for 15 min with 0.1 % Triton X-100 in PBS in all other cases, including labelling of dsRNA using mAb J2. Coverslips were then washed, and incubated in block buffer (10 mM Tris, 150 mM NaCl, 1 mM CaCl 2 , 0.5 mM MgCl 2 , 10 %, v/v, normal goat serum and 1 %, v/v, fish skin gelatin) for 0.5 h. Primary antibodies were added for 1 h in block buffer. The cells were washed and incubated with Alexa-Fluor-conjugated secondary antibodies in block buffer for 45 min. After washing, the coverslips were mounted onto microscope slides using Vectashield mounting medium with DAPI (Vector Laboratories).
Simultaneous in situ hybridization and antibody labelling.
Following fixation, cells were permeabilized for 1 h with 70 % ethanol at 4 C, and washed with wash buffer (300 mM NaCl, 30 mM sodium citrate dihydrate, 10 %, v/v, formamide). Cells were then incubated for 4 h at 37 C with hybridization buffer [wash buffer supplemented with 10 %, w/v, dextran sulphate (Mr, 500 000)] containing a Cal-Fluor 590 labelled Stellaris in situ hybridization probe set (125 nM) and primary antibody against an antigen of interest. The probe set is designed to recognize positive-sense RNA from FMDV O1K and contains 48 different fluorescently labelled probes designed to hybridize with the nonstructural protein region (2B-3D) of the RNA. The coverslips were then twice incubated for 30 min at 37 C in wash buffer containing secondary antibody, washed once with 2Â SSC buffer (300 mM NaCl, 30 mM sodium citrate dihydrate), twice with PBS, and mounted onto microscope slides using Vectashield mounting medium with DAPI.

Imaging and image analysis
Cells were viewed using a Leica SP2 laser-scanning confocal microscope and optical sections recorded using either the Â63 or Â40 oil-immersion objective with a numerical aperture of 1.4 and 1.25, respectively. The data are shown as single optical sections through the middle of the cell with the exception of Fig. 4(b), which shows maximum projections of z-stacks (spacing 0.3 µm). All data were collected sequentially to minimize cross-talk between fluorescent signals. Images were processed using Adobe Photoshop software.
For quantification of PI4P using Imaris image analysis software (Bitplane Scientific Software), images were recorded in sequential scanning mode. Three-dimensional datasets of cells labelled for PI4P and FMDV 3A, or PI4P and BEV-1 virions, were acquired using the Leica SP2 stack function (spacing 0.3 µm). PI4P labelled structures were detected with the spot function of Imaris. The sum fluorescent intensity for each spot (sum of the intensity of all voxels in the spot) was exported in to Microsoft EXCEL, and the sum PI4P fluorescent intensity was then calculated for each cell.
FMDV infection assays 96-Well plate assay. The assay to quantify infection has been described in detail previously (Berryman et al., 2005). Briefly, cells in 96well tissue culture plates were grown until approximately 90 % confluent. The cells were incubated for 1 h at 37 C with FMDV or BEV-1 at an m.o.i of 0.3 p.f.u. cell À1 . The monolayers were washed and incubated with serum-free medium at 37 C for a further 3 h (FMDV) or 4 h (BEV). Infection was stopped, and the cells were fixed with 4 % paraformaldehyde. The cells were permeabilized with 0.1 % Triton and incubated with blocking buffer [10 mM Tris-HCl (pH 7.5), 140 mM NaCl, 1 mM CaCl 2 , 0.5 mM MgCl 2 , 10 %, v/v, normal goat serum, 1 %, v/v, fish gelatin]. Infected cells were identified by sequential incubation with MAb 2C2 (FMDV) or whole virus antiserum (BEV), biotinylated goat anti-mouse or guinea pig IgG secondary antibody (Southern Biotechnologies) and streptavidin-conjugated alkaline phosphatase (Invitrogen), each for 1 h at room temperature, followed by alkaline phosphatase substrate (Bio-Rad) for 10 min. The cells were then washed with distilled water. Infected cells stained dark blue and were counted using an ELISA plate reader (Zeiss). Inhibitors were added to the cells 1 h prior to the addition of virus and remained present until fixation. Each experiment was carried out with triplicate samples for each condition.
Quantification of infection of Huh7 cell lines by confocal microscopy. Huh7-Lunet T7 cell lines were grown on glass coverslips, infected with BEV or FMDV (m.o.i.=0.6) for 4 h, fixed and processed for confocal microscopy as above, using mAb 2C2 to label FMDV-infected cells or guinea pig polyclonal serum to detect BEV-1-infected cells. The proportion of infected cells was scored and normalized to control cells.

Replicon assays
An FMDV renilla luciferase replicon was derived from the infectious copy plasmid pT7S3 (Ellard et al., 1999) by replacement of VP4 residue 19 to VP1 residue 213 (encompassing all of capsid proteins VP2 and VP3 and the majority of VP1 and VP4) by renilla luciferase. A CVB3 renilla luciferase replicon (Lanke et al., 2009) was obtained from Frank van Kuppeveld (University of Utrecht, Netherlands). Replicon plasmids were linearized and FMDV and CVB3 replicon RNA was transcribed using the T7 Megascript kit (Ambion) and purified using the Megaclear Kit (Ambion). Cells in 96-well plates (white walls, clear base; Greiner) were grown to 70 % confluency in phenol-red-free cell culture medium, and were transfected with 90 ng of replicon RNA using the TransIT-mRNA transfection kit (Mirus). Transfected cells were incubated at 37 C in the presence of 60 µM Enduren Live Cell renilla luciferase substrate (Promega) for the times indicated, and luminescence was measured at 37 C using a Hidex Chameleon plate reader. Chemical inhibitors were added at the time of transfection and were present throughout the assay.

Western blotting
Huh7-Lunet T7 cell lines grown in six-well plates were lysed in lysis buffer [120 mM NaCl, 50 mM Tris, pH 7.5, 0.5 %, v/v, octylphenoxypolyethoxyethanol (IGEPAL)], and the protein content of the lysates assayed using a BCA assay kit (Pierce) using BSA standards. Equal quantities of the lysates were added to 3Â SDS-PAGE sample buffer (New England BioLabs), and proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes. The membranes were probed with antibodies against PI4KIIIa, PI4KIIIb or actin.