Hepatitis C Virus Subverts Human Choline Kinase-α To Bridge Phosphatidylinositol-4-Kinase IIIα (PI4KIIIα) and NS5A and Upregulates PI4KIIIα Activation, Thereby Promoting the Translocation of the Ternary Complex to the Endoplasmic Reticulum for Viral Replication

ABSTRACT In this study, we elucidated the mechanism by which human choline kinase-α (hCKα) interacts with nonstructural protein 5A (NS5A) and phosphatidylinositol-4-kinase IIIα (PI4KIIIα), the lipid kinase crucial for maintaining the integrity of virus-induced membranous webs, and modulates hepatitis C virus (HCV) replication. hCKα activity positively modulated phosphatidylinositol-4-phosphate (PI4P) levels in HCV-expressing cells, and hCKα-mediated PI4P accumulation was abolished by AL-9, a PI4KIIIα-specific inhibitor. hCKα colocalized with NS5A and PI4KIIIα or PI4P; NS5A expression increased hCKα and PI4KIIIα colocalization; and hCKα formed a ternary complex with PI4KIIIα and NS5A, supporting the functional interplay of hCKα with PI4KIIIα and NS5A. PI4KIIIα inactivation by AL-9 or hCKα inactivation by CK37, a specific hCKα inhibitor, impaired the endoplasmic reticulum (ER) localization and colocalization of these three molecules. Interestingly, hCKα knockdown or inactivation inhibited PI4KIIIα-NS5A binding. In an in vitro PI4KIIIα activity assay, hCKα activity slightly increased PI4KIIIα basal activity but greatly augmented NS5A-induced PI4KIIIα activity, supporting the essential role of ternary complex formation in robust PI4KIIIα activation. Concurring with the upregulation of PI4P production and viral replication, overexpression of active hCKα-R (but not the D288A mutant) restored PI4KIIIα and NS5A translocation to the ER in hCKα stable knockdown cells. Furthermore, active PI4KIIIα overexpression restored PI4P production, PI4KIIIα and NS5A translocation to the ER, and viral replication in CK37-treated cells. Based on our results, hCKα functions as an indispensable regulator that bridges PI4KIIIα and NS5A and potentiates NS5A-stimulated PI4KIIIα activity, which then facilitates the targeting of the ternary complex to the ER for viral replication. IMPORTANCE The mechanisms by which hCKα activity modulates the transport of the hCKα-NS5A complex to the ER are not understood. In the present study, we investigated how hCKα interacts with PI4KIIIα (a key element that maintains the integrity of the “membranous web” structure) and NS5A to regulate viral replication. We demonstrated that HCV hijacks hCKα to bridge PI4KIIIα and NS5A, forming a ternary complex, which then stimulates PI4KIIIα activity to produce PI4P. Pronounced PI4P synthesis then redirects the translocation of the ternary complex to the ER-derived, PI4P-enriched membrane for assembly of the viral replication complex and viral replication. Our study provides novel insights into the indispensable modulatory role of hCKα in the recruitment of PI4KIIIα to NS5A and in NS5A-stimulated PI4P production and reveals a new perspective for understanding the impact of profound PI4KIIIα activation on the targeting of PI4KIIIα and NS5A to the PI4P-enriched membrane for viral replication complex formation.

and reveals a new perspective for understanding the impact of profound PI4KIII␣ activation on the targeting of PI4KIII␣ and NS5A to the PI4P-enriched membrane for viral replication complex formation. KEYWORDS HCV, NS5A, hCK␣, PI4KIII␣, PI4P, ternary complex, ER translocation L ike many other positive-sense RNA viruses, the hepatitis C virus (HCV) exploits cellular lipids and reorganizes endoplasmic membranes to create a specialized viral replication membrane compartment, which is referred to as the "membranous web" (MW) (1,2). The MW serves as a compartmentation and concentration site for the HCV replication complex (RC), which is composed of the NS3 (nonstructural protein 3) to NS5B proteins, viral replicating RNA, and a wide array of host factors (3,4). The MW is an endoplasmic reticulum (ER)-derived membrane structure, highly phosphorylated and remodeled, that has heterogeneous vesicles ranging from 100 to 300 nm in diameter and is dedicated to HCV replication (5).
Phosphatidylinositol-4-kinases (PI4Ks) catalyze the phosphorylation of position 4 of the inositol ring in phosphatidylinositol (PI) to form PI-4-phosphate (PI4P) and play a crucial role in the regulation of the intracellular membrane trafficking process (6,7). PI4P is an essential precursor in the synthesis of PI(4,5)P2 and PI (3,4,5)P3, which are necessary for receptor-activated phospholipase C and phosphoinositide 3-kinase signaling (8), and controls cellular trafficking, especially in the Golgi apparatus and trans-Golgi network (8). PI4Ks exhibit distinct subcellular localization and affect the compartmentation of discrete PI4P pools. PI4KIII␣ localizes primarily to the ER and maintains a distinct population of PI4P that is critical for ER export when cells are responding to the increased ER cargo load (9,10). PI4KIII␣ also regulates the PI4P level at the plasma membrane (11,12). PI4KIII␣ can also be found in close association with the membranes of the Golgi vesicles and vacuoles (13) and in the nucleolus (14). PI4KIII␤ is the lipid kinase primarily responsible for the PI4P pool in the Golgi apparatus (15,16), and it also localizes on the lysosome to maintain lysosomal membrane integrity (17).
PI4K isoforms and their product PI4Ps play crucial roles in the replication of many positive-stranded RNA viruses in the Flaviviridae, Picornaviridae, and Coronaviridae families, shaping the architecture of their membranous replication compartments via the elevation of PI4P production (7). PI4KIII␣ has been identified as a crucial host factor in HCV replication (18)(19)(20)(21)(22)(23). Although PI4KIII␤ has been reported to play a role in the viral replication of genotypes 1a and 1b (19,22), its role in genotype 2a replication is less clear (18,24).
As a critical component of the viral RC, membrane-associated NS5A is present in both a hypophosphorylated p56 form and a hyperphosphorylated p58 form (25,26), which modulate HCV RNA replication and virus assembly (27). NS5A interacts intimately with PI4KIII␣ via the C-terminal part of domain 1 and recruits the kinase to the viral RC site, where NS5A stimulates kinase activity in conjunction with NS5B, thereby leading to the generation of elevated PI4P pools at intracellular membranes (24,28). PI4KIII␣ activation and PI4P redistribution are required for the changes in the ultrastructural architecture of HCV RCs embedded within the MW (18,21,24,29). Analogously to the PI4KIII␣ depletion phenotype, mutations in the seven amino acids within NS5A, which have been identified as a PI4KIII␣ functional interaction site (PFIS), not only interfere with PI4KIII␣ binding and attenuate PI4P production but also alter the morphology of viral replication sites (29). Conversely, PI4KIII␣ controls HCV RNA replication, presumably via alteration of the phosphorylation status of NS5A (29). Thus, NS5A interplays with PI4KIII␣ to promote PI4P production and endoplasmic membrane reorganization, which are necessary for viral replication. In addition, the accumulation of PI4P recruits lipid transport proteins, such as oxysterol-binding protein (30) or the four-phosphate adaptor protein (31), which regulate the trafficking of cholesterol and glycosphingolipids, respectively, to the viral MW for effective viral genome replication.
Recently, Harak et al. demonstrated that HCV depends on optimal levels of PI4KIII␣ activity for proper replication and showed that the replication of various genotypes requires PI4KIII␣ activation at a low PI4KIII␣ protein level, such as that observed for primary hepatocytes (32). Nevertheless, in cultured hepatoma Huh7 cells, which already express higher levels of PI4KIII␣ than primary hepatocytes, NS5A-induced PI4KIII␣ activation leads to excess PI4P production that is harmful to their replication (32). To manage the growth disadvantage in Huh7 cells, these isolates have evolved a mechanism through adaptive mutations, such as S2204R in NS5A and R2884G in NS5B, to limit excess PI4KIII␣ activation, thus ensuring their proper replication in Huh7 cells (32). Strikingly, the genotype 2a JFH1 virus is an exception, because it does not require adaptive mutations for robust replication in Huh7 cells (33). Thus, PI4KIII␣ activity delicately adapts to the viral and cellular microenvironment to ensure efficient viral replication. Nevertheless, whether the NS5A-mediated recruitment and activation of PI4KIII␣ require a yet-to-be identified cellular factor has not been clarified, and the mechanism underlying the regulation of PI4KIII␣ and NS5A trafficking to the PI4Penriched viral replication site has not been identified. Choline kinase (CK), the first enzyme in the CDP-choline (or Kennedy) pathway, catalyzes the phosphorylation of choline into phosphocholine. Previous genome-wide small interfering RNA (siRNA) screens have implicated human CK␣ (hCK␣) in HCV infections (24,34). As we demonstrated previously, hCK␣ activity enhances the translocation of the hCK␣-NS5A complex to the ER, where the hCK␣ protein mediates the NS5A-NS5B interaction, thereby promoting functional membranous viral RC assembly and viral RNA replication (35). However, whether hCK␣ interacts with PI4KIII␣ and NS5A to modulate the viral replication process has not been clarified. In addition, the mechanism underlying the regulation of hCK␣-mediated translocation of the hCK␣-NS5A complex to the ER remains to be elucidated.
Because PI4K isoforms play critical roles in intracellular membrane trafficking (6, 36) and PI4KIII␣ builds a PI4P-enriched membrane platform for effective HCV replication (37,38), we hypothesize that the hCK␣ activity-mediated ER translocation of hCK␣ and NS5A is tightly associated with PI4KIII␣ activation and that hCK␣ activity plays a crucial role in NS5A-stimulated PI4KIII␣ activation and in the translocation of PI4KIII␣ and NS5A to the ER, thereby promoting membranous viral RC assembly and viral replication (35).
In the present study, we demonstrate that HCV usurps hCK␣ functions to facilitate viral replication. hCK␣ acts as a novel upstream adaptor to bridge PI4KIII␣ and NS5A, thereby forming a PI4KIII␣ ternary complex. Additionally, the results obtained from a cell culture system and an in vitro kinase assay demonstrate that hCK␣ activity slightly enhances PI4KIII␣ basal activity but greatly potentiates NS5A-stimulated PI4KIII␣ activity, indicating that hCK␣ synergistically upregulates PI4KIII␣ activity in conjunction with NS5A. In turn, the massive production of PI4P pools facilitates the translocation of the ternary complex to the ER-derived, PI4P-enriched membrane for viral replication. Our study highlights the novel viewpoint that PI4KIII␣ activity and its translocation in conjunction with NS5A to viral replication sites are regulated by the cellular kinase hCK␣, even though the product of this kinase is not directly linked to membrane composition or structure.

RESULTS
hCK␣ colocalizes with NS5A and PI4P or PI4KIII␣ in HCV-expressing cells. We recently demonstrated that hCK␣ activity is critical for the intracellular trafficking of the hCK␣-NS5A complex to the ER membrane, where hCK␣ mediates the binding of NS5A to NS5B for viral RC assembly (35). In the present study, we further examined whether hCK␣ interacts with PI4KIII␣, the lipid kinase critical for the integrity of the HCV-induced MW, and NS5A to stimulate PI4P production and whether accumulated PI4P production facilitates the translocation of hCK␣ and NS5A, as well as PI4KIII␣, to the ER membrane, thereby promoting HCV replication.
To address this issue, we first searched for the intracellular colocalization of hCK␣, NS5A, and PI4P in HCV-infected Huh7 cells via confocal microscopy. As shown previously (35), a small fraction of hCK␣ colocalized with NS5A in HCV-infected cells (Fig. 1A, magenta puncta in the fluorescence image labeled "NS5A/hCK␣"). HCV infection also resulted in much more accumulation of PI4P in dot-like patterns scattered in the cytoplasm than that seen with mock infection, and a fraction of these PI4P puncta colocalized with NS5A (Fig. 1A), which represents viral replication sites, in HCV-infected FIG 1 Colocalization of hCK␣ with NS5A and PI4P or PI4KIII␣. (A) Huh7 cells either remained uninfected or were infected with HCVcc at an MOI of 1. At day 3 postinfection, the cells were fixed and successively immunostained with rabbit anti-hCK␣, a mouse anti-NS5A (9E10) MAb (IgG), and mouse anti-PI4P (IgM), followed by incubation with appropriate Alexa Fluor dye-conjugated secondary antibodies. The cells were then analyzed by confocal microscopy. Nuclei are demarcated by solid white lines. The images of the three fluorescence signals were merged, and the boxed areas were enlarged. The colocalization of two or three molecules as indicated in the fluorescence profiles was marked by white arrowheads or yellow arrows, respectively. (B) T7/Huh7 cells were cotransfected with or without pTM-NS3-5B, and the cells were analyzed by confocal microscopy as described for panel A. (C) T7/Huh7 cells were transfected with pTM-HA-PI4KIII␣ in the presence or absence of pTM-NS3-5B. The transfected cells were processed for confocal microscopy using rabbit anti-hCK␣, MAb 9E10, and goat anti-HA. (Left) Fluorescence profiles were analyzed as described for panel A. (Right) The colocalization of hCK␣ with HA-PI4KIII␣ in the presence or absence of NS3-5B expression was quantitated. The statistical significance of the values obtained in different experimental settings was calculated by applying the two-tailed, unpaired Student t test, and a P value of Ͻ0.05 was considered statistically significant. ****, P Ͻ 0.0001. cells (Fig. 1A, cyan puncta in the image marked "NS5A/PI4P"). Interestingly, a small proportion of hCK␣ colocalized with PI4P in mock-infected cells, whereas HCV infection augmented hCK␣ and PI4P colocalization, as evidenced by the increased accumulation of yellow puncta in infected cells (Fig. 1A, image labeled "hCK␣/PI4P"). Remarkably, a fraction of hCK␣ colocalized with NS5A and PI4P in HCV-infected cells (Fig. 1A, white puncta in the image marked "NS5A/hCK␣/PI4P").
To understand the involvement of hCK␣ in PI4KIII␣-catalyzed PI4P production, we then explored whether hCK␣ colocalized with NS5A and PI4KIII␣ in T7/Huh7 cells that overexpressed hemagglutinin (HA)-PI4KIII␣, i.e., PI4KIII␣ attached with an HA tag at its N terminus, with or without NS3-5B coexpression. The overexpressed HA-PI4KIII␣ displayed dot-like structures scattered in the cytoplasm of cells expressing proteins NS3 to NS5B (Fig. 1C, left), and certain HA-PI4KIII␣ puncta colocalized with those of NS5A (Fig. 1C, left). This observation was consistent with the current model in which PI4KIII␣ interacts with NS5A and is recruited to viral replication sites (24). A small amount of hCK␣ colocalized with HA-PI4KIII␣ in the absence of NS3-5B expression, whereas the expression of proteins NS3 to NS5B increased the colocalization of these two kinases (Fig. 1C, left). The percentage of the cell area that harbored fluorescent signals for both hCK␣ and HA-PI4KIII␣ with or without viral NS protein expression was quantitated as described previously (35,39). The results showed that the expression of NS3 to NS5B statistically upregulated the colocalization of hCK␣ with HA-PI4KIII␣ (Fig. 1C, right). Additionally, hCK␣ colocalized with NS5A and PI4KIII␣ in NS3-5B-expressing cells (Fig.  1C, left). Therefore, the results from Fig. 1 suggest that hCK␣ may impact viral replication through its involvement in NS5A-stimulated PI4KIII␣ activation in HCVexpressing cells.
NS5A-mediated PI4P accumulation requires hCK␣ activity. To explore whether hCK␣ contributes to PI4P production, we individually silenced hCK␣ and PI4KIII␣ and compared the effects on PI4P levels in HCV-infected cells. PI4P levels were quantified as described previously (24,29). As shown in Fig. 2A (top), HCV infection did not alter PI4KIII␣ and hCK␣ levels. Individual transfection with PI4KIII␣ and hCK␣ siRNAs knocked down PI4KIII␣ and hCK␣ levels to approximately 9% and 15% of those detected with control siRNA transfection; these reductions occurred concomitantly with reduced NS5A expression ( Fig. 2A, top). As shown in Fig. 1A, HCV infection augmented intracellular PI4P levels, and HCV-induced PI4P accumulation was ablated by PI4KIII␣ or hCK␣ knockdown ( Fig. 2A, bottom). These observations suggest a role for hCK␣ in PI4P production during HCV replication.
Next, we confirmed the inhibitory effects of stable hCK␣ knockdown on PI4P production in cells with or without NS3-5B expression. In the absence of NS3-5B expression, PI4P levels were lower in hCK␣ stable knockdown cells than in control knockdown cells (Fig. 2B). As shown in Fig. 1B, NS3-5B expression upregulated PI4P production in control stable knockdown cells (Fig. 2B). Interestingly, stable depletion of hCK␣ in NS3-5B-expressing cells attenuated the increase in PI4P production induced by NS3-5B expression (Fig. 2B). Together, these results imply that hCK␣ participates in PI4P production in cells regardless of NS3-5B protein expression, and hCK␣ may function upstream of PI4KIII␣-catalyzed PI4P production. PI4KIII␣ and hCK␣ levels were first normalized to those of ␤-actin, and these levels in different settings were expressed as percentages relative to the levels detected in mock-infected cells, which were arbitrarily assigned the value of 100%. NS5A levels, after normalization to ␤-actin levels, were detected in HCV-infected, siRNA-transfected cells and were compared to levels detected in HCV-infected cells transfected with control siRNA, which were arbitrarily assigned the value of 100%. ****, P Ͻ 0.0001; n.s., not statistically significant. (Bottom left) Another set of cells was processed by confocal microscopy using MAb 9E10 and anti-PI4P (IgM) to visualize the intracellular localization of NS5A and PI4P. A representative set of PI4P distribution patterns under different conditions is shown. (Bottom right) PI4P levels were quantitated as described in Materials and Methods. In each group, the PI4P fluorescence intensity was expressed as mean arbitrary units (AU) Ϯ SD from 300 cells per condition. (B) Paired T7-control shRNA/Huh7 and T7-hCK␣ shRNA/Huh7 cells were transfected with the vector plasmid (Ϫ) or pTM-NS3-5B; the cells were monitored; and PI4P levels were quantitated. (C) (Left) Huh7 cells either remained mock infected or were infected with HCVcc, followed by treatment with DMSO as a vehicle (Ϫ), 100 M CK37, or 5 M AL-9. A set of cells (Continued on next page) CK37 is a computationally identified compound that specifically inhibits hCK␣ by targeting its choline-binding site, leading to the inhibition of enzymatic activity (40). We showed previously that CK37 impairs viral RNA replication as well as viral protein expression in a dose-dependent manner (35). We then compared in parallel the inhibitory effects of CK37 and AL-9 on viral protein expression in HCV-infected cells. AL-9 is a prototypical PI4KIII␣ inhibitor derived from 4-aminoquinazoline that targets PI4KIII␣ activity and impairs viral replication by depleting PI4P in the plasma membrane (41). CK37 at 100 M and AL-9 at 5 M greatly decreased NS5A expression but did not affect PI4KIII␣ and hCK␣ levels (Fig. 2C, left and center). Additionally, these two inhibitors did not influence the viability of HCV-infected cells (Fig. 2C, right).
To determine whether hCK␣ activity was critical for PI4P accumulation during HCV infection, we measured PI4P levels in mock-infected and HCV-infected cells treated with a vehicle, AL-9, or CK37. As expected, AL-9 greatly reduced HCV-induced PI4P accumulation (Fig. 2D), which is consistent with previous findings that PI4KIII␣ is primarily responsible for PI4P production in HCV-expressing cells (24,41,42). Similarly, CK37 also impaired PI4P accumulation in HCV-infected cells (Fig. 2D), suggesting that hCK␣ activity promotes a functional interaction with PI4KIII␣ and NS5A to regulate massive PI4P generation in HCV-expressing cells.
hCK␣ activity upregulates NS5A-mediated PI4P production by activating PI4KIII␣. To demonstrate the direct contribution of hCK␣ activity to elevated PI4P production, we evaluated the abilities of overexpressed wild-type hCK␣-R and an inactive D288A mutant hCK␣-R (35) to upregulate PI4P production in hCK␣ stable knockdown cells expressing proteins NS3 to NS5B. These two hCK␣-R clones express authentic wild-type and mutant hCK␣ molecules whose mRNAs are resistant to the restrictive effects exerted by the hCK␣ short hairpin RNA (shRNA) used to construct hCK␣ stable knockdown cells (35). Stable knockdown of hCK␣ decreased hCK␣ levels by 87% from those detected in control stable knockdown cells (Fig. 3A, top). The levels of both wild-type and D288A mutant hCK␣-R proteins overexpressed in hCK␣ stable knockdown cells were approximately 3-fold higher than the endogenous hCK␣ levels detected in control stable knockdown cells (Fig. 3A, top). Interestingly, overexpression of the wild-type, but not the D288A mutant, hCK␣-R upregulated PI4P levels in hCK␣ stable knockdown cells relative to hCK␣ stable knockdown cells transfected with the vector control plasmid, and these levels were comparable to those detected in control stable knockdown cells (Fig. 3A, bottom).
To determine whether hCK␣-regulated PI4P accumulation occurs via PI4KIII␣ activation in HCV-expressing cells, the inhibitory effect of AL-9 on wild-type hCK␣mediated PI4P accumulation in hCK␣ stable knockdown cells was examined. Active hCK␣-mediated PI4P production was greatly diminished by AL-9 to a level even lower than that detected in hCK␣ stable knockdown cells overexpressing D288A mutant hCK␣-R (Fig. 3B). The residual PI4P levels observed in hCK␣ stable knockdown cells overexpressing mutant hCK␣-R may be attributable to incomplete stable hCK␣ knockdown, resulting in susceptibility to AL-9 inhibition. Therefore, hCK␣ upregulates PI4P production solely via PI4KIII␣ activation during HCV replication. This notion supports the colocalization of hCK␣ with NS5A and PI4KIII␣ or PI4P (Fig. 1).
Compared with the previous findings that NS5A interacts with PI4KIII␣ and that NS5A and NS5B activate kinase activity at the viral replication site (24,28), our results further emphasize that NS5A alone is insufficient to activate PI4KIII␣ without the participation of hCK␣. Therefore, hCK␣ may function as an indispensable coordinator to modulate NS5A-stimulated PI4KIII␣ activation for massive PI4P production.

FIG 2 Legend (Continued)
was then analyzed by immunoblotting. (Center) Relative levels of the indicated cellular and viral proteins were quantified as described for panel A. (Right) Another set of cells was analyzed in a cell viability assay. Cell viability under different conditions was expressed as a percentage relative to that of mock-infected cells treated with the vehicle, which was arbitrarily designated 100%. (D) Huh7 cells either remained mock infected or were infected with HCVcc, followed by treatment with DMSO as a vehicle, AL-9, or CK37. The cells were then monitored, and PI4P levels were quantitated. hCK␣ in PI4P Production and ER Translocation of NS5A Journal of Virology hCK␣ activity upregulates PI4KIII␣ activity in the absence of viral NS protein expression. Because stable hCK␣ knockdown also decreased PI4P levels in cells without expression of viral proteins NS3 to NS5B, we next examined whether hCK␣ inactivation inhibits PI4P production and whether PI4KIII␣ overexpression compensates for the defective PI4P production induced by CK37 in the absence of viral NS protein expression. HA-PI4KIII␣ overexpression (Fig. 4A, top) ameliorated the inhibitory effects of CK37 on basal PI4P production (Fig. 4A, bottom), confirming that hCK␣ activity modulates basal PI4KIII␣ activity, even in the absence of NS3 to NS5B expression.
Next, we performed a rescue analysis to confirm the direct effect of hCK␣ activity on PI4KIII␣ activation. As a positive control, HA-PI4KIII␣ overexpression (Fig. 4B, top) rescued PI4P accumulation in hCK␣ stable knockdown cells (Fig. 4B, bottom), again indicating that hCK␣-induced PI4P production is mediated by PI4KIII␣. Similarly, overexpression of wild-type (but not D288A mutant) hCK␣-R (Fig. 4B, top) restored PI4P in hCK␣ stable knockdown cells to levels even higher than those detected in control stable knockdown cells (Fig. 4B, bottom). These findings illustrate that hCK␣ activity directly modulates PI4KIII␣ activity, even in the absence of viral NS protein expression. and either a control plasmid (Ϫ), wild-type hCK␣-R, or D288A mutant hCK␣-R. The cells were then subjected to Western blot analysis. (Top right) Relative protein levels were quantified. Red diamonds mark the levels of overexpressed wild-type or D288A mutant hCK␣-R relative to endogenous hCK␣ levels in control stable knockdown cells. Overexpressed and endogenous hCK␣ was detected with rabbit anti-hCK␣ in an immunoblot analysis. (Bottom) Another set of cells was monitored, and PI4P levels were quantified. (B) (Left) T7-hCK␣ shRNA/Huh7 cells were cotransfected with pTM-NS3-5B and wild-type or D288A mutant hCK␣-R and were treated with a vehicle (Ϫ) or AL-9 as indicated above the images. (Right) The cells were then quantitated to determine PI4P levels. ****, P Ͻ 0.0001; n.s., nonsignificant.
hCK␣ does not have a role in PI4KIII␣ expression. Because hCK␣ plays a crucial role in elevated PI4P production, we next investigated whether the inactivation or depletion of hCK␣ affects PI4KIII␣ expression and subsequently inhibits PI4P production. We showed above that CK37 and AL-9 did not affect PI4KIII␣ levels in HCV-infected cells (Fig. 2C). hCK␣ knockdown by specific siRNAs did not exert obvious effects on The cells were then analyzed by Western blotting (left), and relative protein levels were quantified (right). The red asterisk indicates total levels of overexpressed and endogenous PI4KIII␣, which was detected with rabbit anti-PI4KIII␣ by Western blotting, relative to the endogenous PI4KIII␣ levels obtained in cells without CK37 treatment and HA-PI4KIII␣ overexpression. **, P Ͻ 0.01; ***, P Ͻ 0.001; ****, P Ͻ 0.0001; n.s., nonsignificant. (Bottom) Another set of cells was monitored (left), and PI4P levels were quantitated (right). (B) Paired T7-control shRNA/Huh7 and T7-hCK␣ shRNA/Huh7 cells were transfected with either a control vector, a wild-type or D288A mutant hCK␣-R plasmid, or pTM-HA-PI4KIII␣. The cells were then analyzed by Western blotting (top), and PI4P levels were quantitated (bottom). The red asterisk indicates the total levels of overexpressed and endogenous PI4KIII␣ relative to endogenous PI4KIII␣ levels obtained in control stable knockdown cells, whereas the red diamonds indicate the levels of overexpressed wild-type or D288A mutant hCK␣-R relative to endogenous hCK␣ levels in control stable knockdown cells.
PI4KIII␣ inactivation impairs the colocalization of hCK␣ with PI4KIII␣ and NS5A on the ER membrane. We demonstrated previously that hCK␣ activity is required for the efficient translocation of hCK␣ and NS5A to the ER (35) and showed above that the inactivation of hCK␣ or PI4KIII␣ by CK37 or AL-9 reduced HCV-mediated PI4P accumulation (Fig. 2D). To provide evidence for the functional link between hCK␣ and PI4KIII␣, we examined whether PI4KIII␣ inactivation exhibited a phenotype similar to that achieved with hCK␣ knockdown, which impairs NS5A transport to the ER (35), and whether PI4KIII␣ inactivation also hindered PI4KIII␣ translocation to the ER, because PI4KIII␣ and NS5A localization sites on the ER are indicative of viral replication factories (24,42).
Treatment with AL-9 (or CK37 in hCK␣ inactivation) interferes with viral RNA replication and complicates the detection of NS5A on the ER; therefore, a transient pTM-NS3-5B expression model was employed. T7/Huh7 cells overexpressing NS3-5B were treated with a vehicle or AL-9, and the subcellular distribution of PI4KIII␣ and NS5A and their localization on the ER, which was immunostained using anti-calreticulin (also known as calregulin [CALR], an ER-resident marker), were examined via confocal microscopy. Quantitative analysis of 20 randomly selected cells indicated that AL-9 treatment not only hindered the distribution of NS5A on the ER but also reduced the localization of PI4KIII␣ on the ER from that achieved with vehicle treatment (Fig. 6A). In addition, AL-9 treatment hindered the ER colocalization of these two molecules (Fig. 6A). However, PI4KIII␣ still effectively colocalized with NS5A regardless of AL-9 treatment (Fig. 6A).
We then investigated whether AL-9 affected the colocalization of hCK␣ with overexpressed HA-PI4KIII␣ and NS5A. AL-9 decreased the colocalization of hCK␣ with NS5A or with HA-PI4KIII␣, as well as the colocalization of hCK␣ with both HA-PI4KIII␣ and NS5A (Fig. 6B). Because PI4KIII␣ activity was required for the efficient transport of PI4KIII␣ and NS5A to the ER membrane (Fig. 6A) as well as for the colocalization of hCK␣ with HA-PI4KIII␣ and NS5A (Fig. 6B), we sought to determine whether PI4KIII␣ activity is also important for the colocalization of hCK␣ with NS5A on the ER. As shown in Fig.  6C, treatment with AL-9 not only impeded hCK␣ localization on the ER but also interfered with the colocalization of hCK␣ and NS5A, as well as with the colocalization of these proteins on the ER.
Collectively, these data show that in addition to the PI4KIII␣-regulated translocation of PI4KIII␣ and NS5A to the ER, the trafficking of hCK␣ to the ER is regulated by PI4KIII␣ activity. These results also indicate that PI4KIII␣ activity is required for the effective colocalization of hCK␣ with PI4KIII␣ and NS5A on the ER, whereas it is not crucial for PI4KIII␣ and NS5A colocalization.
Inactivation of hCK␣ interferes with the translocation of hCK␣, PI4KIII␣, and NS5A to the ER membrane. To further address the functional relationship between hCK␣ and PI4KIII␣, we examined the effect of hCK␣ inactivation on the colocalization of PI4KIII␣ and NS5A on the ER. In agreement with our previous finding that hCK␣ activity is crucial for the transport of NS5A to the ER (35), CK37 treatment hindered the localization of PI4KIII␣ on the ER (Fig. 7A). Additionally, CK37 not only inhibited the colocalization of PI4KIII␣ with NS5A but also interfered with the colocalization of these proteins on the ER (Fig. 7A). In addition to inhibiting the colocalization of hCK␣ with NS5A, CK37 treatment also interfered with the colocalization of HA-PI4KIII␣ with hCK␣ and NS5A (Fig. 7B). Moreover, CK37 treatment impaired the colocalization of hCK␣ with NS5A on the ER (Fig. 7C). Therefore, the results shown in Fig. 7 collectively demonstrate that hCK␣ activity plays a critical role in the ER translocation of hCK␣ and NS5A and is also necessary for the effective translocation of PI4KIII␣ to the ER.
Inactivation of hCK␣ and PI4KIII␣ does not affect the localization of NS5B on the ER. We demonstrated previously that the ER localization of NS5A (but not that of NS5B or NS3) is inhibited by hCK␣ knockdown (35). We next investigated whether the inhibitory effects of hCK␣ or PI4KIII␣ inactivation on the translocation of NS5A to the ER are specific by evaluating the effects of hCK␣ or PI4KIII␣ inactivation on the ER localization of NS5B. In contrast to the inhibitory effects of AL-9 and CK37 on the localization of HA-PI4KIII␣ on the ER, AL-9 or CK37 did not affect the localization of NS5B on the ER (Fig. 8A). However, the colocalization of NS5B with HA-PI4KIII␣ and the colocalization of these proteins on the ER were inhibited by AL-9 or CK37 (Fig. 8A). As an internal control, Western blot analysis indicated that CK37 or AL-9 treatment did not affect levels of PI4KIII␣, hCK␣, or NS5B (Fig. 8B). Based on these observations, we conclude that NS5B localizes on the ER even when PI4KIII␣ is redirected to non-ER membranes due to the disruption of PI4P synthesis.
hCK␣ forms a ternary complex with PI4KIII␣ and NS5A. In order to understand the molecular functions underlying the role of hCK␣ in the production of PI4P and the ER trafficking of hCK␣, PI4KIII␣, and NS5A, we performed coimmunoprecipitation (co-IP) to examine whether these three functionally interrelated proteins form a complex in cells. Lysates from cells overexpressing NS3-5B and HA-PI4KIII␣ were incubated with a rabbit isotype IgG or anti-HA, and the precipitated proteins were subjected to immunoblot analysis. HA-PI4KIII␣ was specifically precipitated by anti-HA but not by control IgG (Fig. 9A, left). Anti-HA also specifically pulled down hCK␣ as well as NS5A (Fig. 9A,  left). Additionally, the precipitation of hCK␣ with rabbit anti-hCK␣ specifically cocaptured HA-PI4KIII␣ as well as NS5A (Fig. 9A, right). To confirm ternary complex formation in HCV-infected cells, lysates from HCVinfected cells were incubated with rabbit isotype IgG or anti-PI4KIII␣, followed by Western blotting. Precipitation of PI4KIII␣ with anti-PI4KIII␣ also specifically pulled down NS5A and hCK␣ (Fig. 9B, left). Additionally, anti-NS5A not only precipitated NS5A but also cocaptured PI4KIII␣ and hCK␣ (Fig. 9B, right). Thus, hCK␣, PI4KIII␣, and NS5A are interacting partners that form the ternary complex.
hCK␣ binds PI4KIII␣ independently of NS5A. Because hCK␣ was found to colocalize with PI4KIII␣ in the absence of NS5A, and because NS5A expression augmented the colocalization of hCK␣ and PI4KIII␣ (Fig. 1C), we examined whether hCK␣ formed a complex with PI4KIII␣ in the absence of NS5A and whether NS5A expression enhanced the binding of hCK␣ and PI4KIII␣. The precipitation of HA-PI4KIII␣ with anti-HA from cells coexpressing HA-PI4KIII␣ and hCK␣ tagged with a Flag epitope at its C terminus, i.e., hCK␣-Flag, with or without NS5A coexpression, pulled down similar amounts of hCK␣-Flag (Fig. 9C, left). Similarly, the precipitation of hCK␣-Flag with anti-Flag cocaptured comparable amounts of HA-PI4KIII␣ in the presence or absence of NS5A coexpression (Fig. 9C, right). These results indicate that NS5A does not have a crucial role in the hCK␣-PI4KIII␣ interaction.
PI4KIII␣ activity is not necessary for the binding of PI4KIII␣ to NS5A or hCK␣ and is not critical for hCK␣-Flag and NS5A interactions. In order to determine the molecular events underlying the effects of PI4KIII␣ inactivation on the transport of hCK␣, PI4KIII␣, and NS5A to the ER (Fig. 6), phenylarsine oxide (PAO), an inhibitor of PI4KIII␣ (28), was applied to investigate the binding of PI4KIII␣ to NS5A or hCK␣. PAO treatment did not affect the binding of HA-PI4KIII␣ to NS5A (Fig. 9D) or hCK␣-Flag (Fig.  9E). In addition, PAO did not affect the binding of hCK␣-Flag to NS5A (Fig. 9F), a finding consistent with the observation that PI4KIII␣ knockdown did not exert any apparent effects on the hCK␣-Flag and NS5A interaction (Fig. 9G). These results collectively indicate that PI4KIII␣ activity is not involved in the interactions between any two components of the ternary complex. and pTM-NS3-5B. The cell lysates were incubated with rabbit isotype-matched control IgG (Ϫ) or rabbit anti-HA (left) or with rabbit isotype control IgG or rabbit anti-hCK␣ (right). The precipitated proteins were subjected to immunoblot analysis. A portion of the cell lysates corresponding to 5% of the total proteins used for co-IP was also loaded as an input control in the Western blot analysis. (B) Lysates obtained from HCV-infected cells were incubated with rabbit control isotype IgG or anti-PI4KIII␣ (left) or with mouse control isotype IgG or anti-NS5A (right), and the precipitated proteins were subjected to Western blot detection. (C) T7/Huh7 cells were cotransfected with pTM-HA-PI4KIII␣ and pCMV6-hCK␣ with or without pTM-NS5A. The cell lysates were incubated with rabbit anti-HA (left) or rabbit anti-Flag (right). The isolated proteins were subjected to immunoblot analysis. (D and E) T7/Huh7 cells were cotransfected with pTM-HA-PI4KIII␣ and pTM-NS5A (D) or with pTM-HA-PI4KIII␣ and pCMV6-hCK␣ (E). The cells were treated with a vehicle or PAO, and the cell lysates were incubated with rabbit anti-HA prior to Western blot analysis. (F) T7/Huh7 cells were cotransfected with pTM-NS3-5B and pCMV6-hCK␣, followed by vehicle or PAO treatment. Cell lysates were precipitated with rabbit anti-Flag prior to immunoblot analysis. (G) T7/Huh7 cells were first transfected with nontargeted siRNA or PI4KIII␣ siRNA, followed by transfection with pCMV6-hCK␣ and pTM-NS5A. The lysates from transfected cells were immunoprecipitated with mouse anti-NS5A prior to Western blot analysis. In panels C to G, the results from three independent analyses did not demonstrate statistically significant differences between the vehicle and PAO treatment or between control and tested conditions. Therefore, representative sets of data are shown.
Although hCK␣ does not have an apparent effect on the interaction of hCK␣ with PI4KIII␣ or NS5A, its activity facilitates the binding of PI4KIII␣ to NS5A. To understand the molecular basis of the effects of hCK␣ activity on the ER translocation of hCK␣, PI4KIII␣, and NS5A (Fig. 7), we examined whether CK37 affected the binding of hCK␣ to PI4KIII␣ or NS5A. According to co-IP analyses, CK37 did not affect the binding of hCK␣-Flag to HA-PI4KIII␣ (Fig. 10A) or alter the hCK␣-Flag and NS5A interaction (Fig. 10B). Additionally, wild-type hCK␣-R and the D288A mutant bound comparably to HA-PI4KIII␣ (Fig. 10C), supporting the former observation. . The cells were then treated with a vehicle or CK37, and cell lysates were incubated with rabbit isotype IgG or anti-HA (A) or with rabbit isotype IgG or anti-Flag (B). The precipitated proteins were then analyzed by Western blotting. (C) T7/Huh7 cells were cotransfected with pTM-HA-PI4KIII␣ and wild-type or D288A mutant hCK␣-R. The cell lysates were then incubated with rabbit anti-Flag, and the precipitated proteins were subjected to Western blot analysis. (D) T7-control shRNA/Huh7 and T7-hCK␣ shRNA/Huh7 cells were cotransfected with pTM-HA-PI4KIII␣ and pTM-NS5A. The cell lysates were incubated with rabbit anti-HA (left) or an anti-NS5A MAb (right) prior to Western blot analysis. The binding of the two coimmunoprecipitated proteins, as indicated in the quantitation graphs, obtained from hCK␣ stable knockdown cells was expressed as a ratio relative to the binding obtained from control stable knockdown cells, which was arbitrarily set at a ratio of 1. (E) T7/Huh7 cells were cotransfected with pTM-HA-PI4KIII␣ and pTM-NS5A, followed by treatment with a vehicle or CK37 as indicated. The cell lysates were then incubated with rabbit anti-HA (left) or an anti-NS5A MAb (right), and the precipitated proteins were analyzed by immunoblot detection. The binding of the two coprecipitated proteins, as indicated in each quantitation graph, obtained from vehicle-treated cells was arbitrarily assigned a ratio of 1, and binding obtained from CK37 treatment was expressed as a relative ratio. In panels A to C, the results from three separate studies did not demonstrate statistically significant differences between vehicle and CK37 treatment or between the wild type and D288A mutant hCK␣-R expression. Therefore, representative sets of data are shown. **, P Ͻ 0.01; ***, P Ͻ 0.001; ****, P Ͻ 0.0001.
Next, we examined whether hCK␣ is involved in the PI4KIII␣-NS5A interaction. The precipitation of HA-PI4KIII␣ from lysates containing HA-PI4KIII␣ and NS5A with anti-HA pulled down a barely detectable NS5A band in hCK␣ stable knockdown cells, in contrast to the band obtained from control stable knockdown cells (Fig. 10D, left). Conversely, the precipitation of NS5A with anti-NS5A from hCK␣ stable knockdown cells cocaptured a barely visible HA-PI4KIII␣ band, in contrast to the band observed in control stable knockdown cells (Fig. 10D, right). Quantitation of the data from three independent studies indicated that hCK␣ stable knockdown reduced the ratios of NS5A binding to HA-PI4KIII␣ and HA-PI4KIII␣ binding to NS5A to approximately 0.24 and 0.28, respectively, relative to that for control stable knockdown cells, which was arbitrarily designated 1 (Fig. 10D).
Because CK37 reduces the colocalization of PI4KIII␣ and NS5A (Fig. 7A), we then determined whether the binding of PI4KIII␣ to NS5A requires hCK␣ activity. When cell lysates containing HA-PI4KIII␣ and NS5A were incubated with anti-HA or anti-NS5A, smaller amounts of NS5A or HA-PI4KIII␣ coprecipitated with HA-PI4KIII␣ or NS5A, respectively, in CK37-treated cells than in vehicle-treated cells (Fig. 10E, left and right, respectively). Quantitation of the data from three separate analyses indicated that CK37 treatment reduced the ratios of NS5A binding to HA-PI4KIII␣ and HA-PI4KIII␣ binding to NS5A to approximately 0.28 and 0.21 relative to that for vehicle treatment, which was arbitrarily designated 1 (Fig. 10E, left and right, respectively). These results indicate that hCK␣ activity promotes the recruitment of PI4KIII␣ to NS5A.
hCK␣ upregulates PI4KIII␣ activity in vitro in the presence or absence of NS5A. To further support the role of hCK␣ in PI4KIII␣ activation, the in vitro activity of recombinant PI4KIII␣ was assayed in the presence or absence of recombinant NS5A and/or hCK␣. Recombinant PI4KIII␣, obtained from the baculovirus expression system, comprises the amino acid residues between positions 1249 and 2102 in the kinase, including the kinase catalytic domain. Recombinant NS5A, purified from Escherichia coli, comprises amino acids 2061 to 2302 in the HCV polyprotein, i.e., the C-terminal portion of domain 1, low-complexity sequence 1, and the N-terminal portion of domain 2 of NS5A. Both recombinant proteins are fused with glutathione S-transferase (GST) at their N termini, and their expected molecular masses are 124 kDa (for PI4KIII␣) and 53 kDa (for NS5A). Recombinant hCK␣ was purified from 293T cells transfected with pCMV6-hCK␣. The authenticity of these proteins was confirmed by immunoblotting (Fig. 11A). As expected, the addition of NS5A to PI4KIII␣ increased PI4KIII␣ activity (Fig. 11B, compare bars 1 and 2). The addition of hCK␣ to PI4KIII␣ increased PI4KIII␣ basal activity to levels even higher than those activated by NS5A (Fig. 11B, compare bar 1 with bars 2 and 3). Remarkably, hCK␣, in conjunction with NS5A, greatly augmented PI4KIII␣ activation (Fig. 11B, compare bars 2 and 4). hCK␣-stimulated basal PI4KIII␣ activity was inhibited by both PAO (Fig. 11B, compare bars 3 and 5) and CK37 (Fig. 11B, compare  bars 3 and 6). Likewise, hCK␣-augmented PI4KIII␣ activity triggered by NS5A was ablated by PAO (Fig. 11B, compare bars 4 and 7) and CK37 (Fig. 11B, compare bars 4  and 8). Based on these results, we conclude that hCK␣ activity directly stimulates in vitro PI4KIII␣ activity and that ternary complex formation is essential for robust PI4KIII␣ activation.
PI4KIII␣ overexpression rescues the impaired viral replication and PI4P production caused by hCK␣ inactivation. Because hCK␣ effectively interacted with PI4KIII␣ regardless of PI4KIII␣ or hCK␣ inactivation (Fig. 9E and 10A and C), we next examined the functional interaction between these two proteins and tested whether PI4KIII␣ overexpression counteracts the negative effects of CK37 on viral replication in HCVinfected cells. Total levels of overexpressed HA-PI4KIII␣ and endogenous PI4KIII␣ were approximately 4.7-fold greater than the endogenous PI4KIII␣ levels detected in mockinfected cells without CK37 treatment or HA-PI4KIII␣ overexpression (Fig. 12A). The overexpression of PI4KIII␣ in CK37-treated, HCV-infected cells resulted in levels of viral proteins, such as NS3, NS5A, and Core, higher than those in CK37-treated, HCV-infected cells without PI4KIII␣ overexpression (Fig. 12A). Additionally, the levels of these viral proteins were comparable to those detected in HCV-infected cells treated with a vehicle (Fig. 12A). Based on these results, we conclude that PI4KIII␣ overexpression compensates for the defects associated with hCK␣ inactivation during viral replication.
To determine whether hCK␣ modulates viral replication by activating PI4KIII␣ for mass PI4P production, we determined whether PI4KIII␣ overexpression could rescue PI4P production in CK37-treated HCV-infected cells. In agreement with the restoration of viral replication, the overexpression of PI4KIII␣ substantially augmented PI4P production in HCV-infected cells treated with CK37 (Fig. 12B). These findings, taken together, indicate that the deficiency in hCK␣ activity observed during suppressed viral replication proceeds via attenuation of PI4KIII␣-mediated PI4P production.
Expression of active (but not inactive) hCK␣ promotes the translocation of PI4KIII␣ and NS5A to the ER in hCK␣ stable knockdown cells. According to our results, hCK␣ inactivation interferes with the translocation of PI4KIII␣ and NS5A to the ER (Fig.  7); therefore, we next investigated whether hCK␣ activity directly upregulates the ER translocation of PI4KIII␣ and NS5A in hCK␣ stable knockdown cells. In agreement with the inhibitory effects of CK37 on the ER translocation of NS5A and PI4KIII␣, stable hCK␣ knockdown also reduced the localization of NS5A and HA-PI4KIII␣ on the ER (Fig. 13). Overexpression of wild-type (but not D288A mutant) hCK␣-R in hCK␣ stable knockdown cells not only restored NS5A localization on the ER, in agreement with our previous observations (35), but also rescued the translocation of HA-PI4KIII␣ to the ER (Fig. 13). Moreover, overexpression of wild-type (but not D288A mutant) hCK␣-R restored the colocalization of NS5A and HA-PI4KIII␣ on the ER (Fig. 13). These results, together with our previous findings (35), demonstrate that hCK␣ activity positively regulates the translocation of hCK␣ as well as NS5A and PI4KIII␣ to the ER.
Overexpression of active (but not inactive) PI4KIII␣ restores the translocation of PI4KIII␣ and NS5A to the ER in CK37-treated cells. Treatment with AL-9 or CK37 attenuated the production of PI4P (Fig. 2D) and the translocation of PI4KIII␣ and NS5A as well as hCK␣ to the ER (Fig. 6 and 7); therefore, we investigated whether PI4P production is essential for the ER translocation of PI4KIII␣ and NS5A. To address this issue, we investigated the abilities of active and kinase-deficient (KD) (K1792L mutant) HA-PI4KIII␣ to ameliorate the impaired ER translocation of PI4KIII␣ and NS5A caused by The in vitro activity of purified recombinant PI4KIII␣ was assayed in the presence or absence of recombinant NS5A, hCK␣, and/or inhibitors in an ADP-Glo kinase assay as indicated. Kinase activity, monitored as the conversion of ATP to ADP under different experimental conditions relative to that detected in PI4KIII␣ alone, which was arbitrarily set at a value of 1, was expressed as the fold change in luciferase activity. The dashed red line indicates the threshold for this in vitro kinase assay. The data shown are the quantitated results from four independent experiments. **, P Ͻ 0.01; ***, P Ͻ 0.001. CK37 in NS3-5B-expressing cells. The PI4KIII␣ mutant is unable to produce PI4P or to support viral replication (28). In agreement with the results shown in Fig. 7, CK37 reduced the translocation of PI4KIII␣ and NS5A to the ER (Fig. 14A), and overexpression of wild-type HA-PI4KIII␣, but not its kinase-deficient mutant (Fig. 14B), restored the translocation of PI4KIII␣ and NS5A to the ER in CK37-treated cells (Fig. 14A). These results strongly indicate that PI4KIII␣-mediated PI4P production is a prerequisite for the translocation of PI4KIII␣ and NS5A to the ER (also, see below).

DISCUSSION
We previously demonstrated the critical function of hCK␣ in promoting HCV membranous RC formation and viral replication (35). In our present study, we further elucidate the mechanism underlying the regulatory roles of hCK␣ in HCV replication. Like PI4KIII␣, hCK␣ played a crucial role in PI4P production and viral replication, as shown by siRNA knockdown and inhibitor analyses. Additionally, both hCK␣ and PI4KIII␣ activities were critical for the translocation of hCK␣, PI4KIII␣, and NS5A to the ER-derived membrane, indicating that these two kinases and NS5A function in concert to facilitate their own translocation to the ER. Furthermore, hCK␣ activity was essential for the generation of elevated PI4P pools in HCV-expressing cells. The overexpression of wild-type (but not D288A mutant) hCK␣-R restored PI4P production in hCK␣ stable knockdown cells, which is consistent with our previous finding that the active (but not and hCK␣ levels under the tested conditions relative to those detected in mock-infected cells without HA-PI4KIII␣ overexpression or CK37 treatment were determined. The red asterisk marks the total levels of overexpressed and endogenous PI4KIII␣, which were detected using anti-PI4KIII␣ in a Western blot, relative to endogenous PI4KIII␣ levels in mock-infected cells without overexpression or CK37 treatment, expressed as a percentage. The levels of NS3, NS5A, and core under different conditions, relative to the levels detected in HCV-infected cells without overexpression or CK37 treatment, are also shown. (B) Another set of cells, as described for panel A, was monitored for the intracellular localization of NS5A and PI4P (left), and PI4P levels were quantitated (right). **, P Ͻ 0.01; ***, P Ͻ 0.001; ****, P Ͻ 0.0001; n.s., nonsignificant. the D288A mutant) hCK␣-R can rescue viral replication (35). Moreover, wild-type hCK␣-R-mediated PI4P accumulation was abrogated by AL-9, a PI4KIII␣-specific inhibitor. These results collectively indicate that the profound PI4P synthesis mediated by hCK␣ occurs primarily via the upregulation of NS5A-stimulated PI4KIII␣ activation and that PI4P functions as a critical determinant for the effective trafficking of these three interacting partners (see below) to the ER membrane for productive viral replication.
hCK␣ is present predominantly as a cytosolic protein, and HCV infection or the expression of NS proteins facilitates the localization of hCK␣ to the ER (35). Given the multiple intracellular localization sites of NS5A in perinuclear membranes, such as those of the ER, Golgi complex, and lipid droplets (5,43), as well as the localization sites of PI4KIII␣ in the ER (9, 10), Golgi complex-associated membrane vesicles (13), and the nucleolus (14), it would be difficult to determine unambiguously where PI4KIII␣ and NS5A localize when PI4P synthesis is interrupted by CK37 or AL-9. Here we used an ER-residing marker, CALR, to investigate the transport or "translocation" of PI4KIII␣, NS5A, and hCK␣ to the PI4P-enriched, ER-associated membrane, because HCV replication increases their localization to the CALR-associated membrane. In contrast, AL-9 or CK37 hinders their colocalization with CALR, indicating that interference with PI4P hCK␣ in PI4P Production and ER Translocation of NS5A Journal of Virology production detours their ER translation to other membranes, which are collectively referred to as "non-ER membranes," because they are not associated with CALR. Although HCV replication induces the ER transport of only a portion of these three replication components, this process is biologically and virologically relevant, because their "ER translocation" correlates well with elevated PI4P production and viral replication.
In agreement with our previous observation that the ER localization of NS5B is not affected by hCK␣ knockdown (35), NS5B localization on the ER, as shown by its colocalization with the ER marker CALR, was not influenced by AL-9 or CK37. However, the colocalization of NS5B and PI4KIII␣ and their colocalization on the ER were inhibited by AL-9 or CK37. Based on these observations, we conclude that NS5B still localizes on the ER membrane when PI4P synthesis is impaired by inhibitors, while PI4KIII␣ localizes to non-ER membranes. Therefore, the transport of NS5A and NS5B to the ER is differentially regulated; the route that delivers NS5A to the ER is strictly dependent on PI4P production, whereas the localization of NS5B on the ER is independent of PI4P. Therefore, HCV replication specifically "redirects" PI4KIII␣, NS5A, and hCK␣ to the ER.
Furthermore, we performed a series of co-IP analyses to elucidate the molecular basis of the function of hCK␣ in NS5A-stimulated PI4KIII␣ activation and the translocation of these three molecules to the ER. The inactivation of PI4KIII␣ by PAO or of hCK␣ by CK37 did not have an apparent effect on PI4KIII␣-hCK␣ binding. In support of this finding, the wild-type and D288A mutant hCK␣-R clones interacted comparably with PI4KIII␣. Thus, the inability of the D288A mutant to rescue PI4P generation in hCK␣ stable knockdown cells expressing NS3-5B proteins is attributable to its lack of hCK␣ activity rather than to its inability to bind PI4KIII␣. In the present study, as well as in previous studies (35), we observed that D288A mutant hCK␣ still possesses wild-typelike NS5A-binding ability (35), and PAO or CK37 did not affect the interaction between these two molecules. In addition, hCK␣ interacted effectively with NS5A, even when PI4KIII␣ was knocked down. Collectively, these results imply that PI4P levels do not influence hCK␣ binding to PI4KIII␣ or NS5A. Nevertheless, AL-9 or CK37 reduced the colocalization of hCK␣ with NS5A or PI4KIII␣ on the ER membrane. Thus, in the absence of massive PI4P production, the physical interactions between hCK␣ and PI4KIII␣ and between hCK␣ and NS5A do not ensure the correct translocation of hCK␣ with PI4KIII␣ or NS5A to the same ER membrane; instead, these paired molecules are spatially segregated into different non-ER membrane compartments (Fig. 15A, center and right).
Although AL-9 did not affect PI4KIII␣ and NS5A colocalization, it inhibited the colocalization of these two molecules on the ER. This observation is consistent with the finding that the PI4KIII␣-NS5A interaction was not influenced by PAO, as shown in the present study, or by AL-9, as reported previously (29). Thus, in the absence of PI4KIII␣ activity, PI4KIII␣ and NS5A still colocalize and interact with each other on the same non-ER membrane (Fig. 15A, center). In contrast, CK37 inhibited the colocalization of these two proteins, a finding compatible with the co-IP results showing that hCK␣ knockdown or inactivation by CK37 impaired PI4KIII␣-NS5A binding. In the absence of hCK␣ activity, hCK␣, PI4KIII␣, and NS5A appear to segregate into different non-ER membrane compartments in which hCK␣ still binds to PI4KIII␣ and NS5A (Fig. 15A,  right). Conversely, the paradoxical discrepancy between the differential effects of NS5A on the colocalization of PI4KIII␣ with hCK␣ and the binding of PI4KIII␣ to hCK␣ may be interpreted as follows. In the absence of NS5A, hCK␣-PI4KIII␣ complexes do not accumulate on membranes devoid of PI4P, thus increasing the difficulty in visualizing these complexes via microscopy. However, NS5A expression enhances the translocation of the hCK␣-PI4KIII␣-NS5A complex to the ER, resulting in the formation and accumulation of viral RCs on the PI4P-enriched membrane (Fig. 15B) and increasing the visibility of these complexes by confocal microscopy over that of the single hCK␣-PI4KIII␣ complex.
The differential effects of CK37 and PAO on the colocalization and interaction of PI4KIII␣ and NS5A are not attributable to the influence of PI4P production on the PI4KIII␣-NS5A interaction. Instead, they may be ascribed to the specific function of hCK␣ activity in PI4KIII␣-NS5A binding. This proposal is supported by the co-IP results in which hCK␣ activity facilitates the binding of PI4KIII␣ to NS5A. Because hCK␣ shRNA reduces hCK␣ levels to approximately 13% of that detected in control stable knockdown cells, and CK37 does not completely attenuate hCK␣ activity, as reflected by the low levels of viral protein expression in CK37-treated, HCV-infected cells, residual hCK␣ protein and/or activity conceivably may still bridge PI4KIII␣ and NS5A, resulting in incomplete abrogation of the NS5A-PI4KIII␣ interaction.
We envision that the interfering effects exerted by CK37 on PI4P production and viral replication depend on the molar ratio of CK37 molecules to intracellular PI4KIII␣ molecules in a defined context. Since the total level of overexpressed HA-PI4KIII␣ and endogenous PI4KIII␣ is approximately 4.3-to 4.7-fold that of endogenous PI4KIII␣ expressed in cells lacking overexpression, unsurprisingly, functional HA-PI4KIII␣ overexpression enables PI4KIII␣ to outcompete CK37 molecules and "neutralize" the inhibitory effects of CK37, thereby restoring PI4P production, the ER translocation of PI4KIII␣ and NS5A, and viral replication. Effect of CK37 on the segregation of hCK␣, PI4KIII␣, and NS5A into three distinct non-ER membranes. In this case, hCK␣ still binds PI4KIII␣ and NS5A; however, PI4KIII␣ and NS5A localize to different non-ER membranes, and their interaction is also disrupted. (B) Model illustrating the regulatory roles of hCK␣ in NS5A-stimulated PI4P production and the subsequent ER translocation of the hCK␣, PI4KIII␣, and NS5A ternary complex. hCK␣ first interacts with PI4KIII␣ at a non-ER membrane, where hCK␣ recruits and bridges NS5A and PI4KIII␣ via hCK␣ activity, thus forming a ternary complex. The activity of PI4KIII␣ in the ternary complex is synergistically upregulated by hCK␣ and NS5A, resulting in pronounced PI4P production. Additionally, hCK␣ activity directly upregulates PI4KIII␣ activation without the participation of NS5A. These events then facilitate the targeting of hCK␣, PI4KIII␣, and NS5A to the ER-derived membrane, where hCK␣ mediates the binding of NS5A to NS5B and, along with NS5A, activates PI4KIII␣, resulting in the formation of the PI4P-enriched platform for the continued viral replication complex (RC) assembly and membranous web (MW) formation necessary for viral replication.
One possible mechanism by which hCK␣ activity modulates the PI4KIII␣-NS5A interaction is via modifications to the PI4KIII␣ or NS5A structure and subsequent enhancement of the PI4KIII␣-NS5A interaction, resulting in PI4KIII␣ activation in the ternary complex (Fig. 15B). This notion is supported by the finding that hCK␣, in conjunction with NS5A, vigorously upregulates PI4KIII␣ activity in vitro through hCK␣ activity. Therefore, the main function of hCK␣ during HCV replication is to recruit and closely bridge PI4KIII␣ and NS5A, leading to the formation of a PI4KIII␣ ternary complex. Then hCK␣, in conjunction with NS5A, synergistically upregulates PI4KIII␣ activity to facilitate abundant PI4P production and viral replication (Fig. 15B). Thus, the "trio" acts as a critical module to coordinately boost PI4KIII␣ activation to achieve so as the massive PI4P production required to maintain the integrity of the MW structure and viral replication.
Because the translocation of hCK␣, PI4KIII␣, and NS5A to the ER membrane is correlated with hCK␣-and PI4KIII␣-stimulated PI4P generation, we propose that massive PI4P production drives the translocation of the hCK␣-PI4KIII␣-NS5A ternary complex to viral replication sites on the ER membrane (Fig. 15B). This notion is supported by the findings that overexpression of wild-type (but not D288A mutant) hCK␣-R upregulated the PI4P level and enhanced the ER transport of hCK␣, PI4KIII␣, and NS5A in hCK␣ stable knockdown cells. The D288A single mutation in hCK␣-R likely does not account for the impaired ER translocation of this mutant. In addition, the inhibited translocation of authentic PI4KIII␣ and NS5A to the ER when the mutant hCK␣-R was overexpressed was caused by the inability of this mutant to mediate elevated PI4P production. Because a single point mutation in the PI4KIII␣ KD mutant likely does not drastically alter the ER translocation of this mutant, the redirection of PI4KIII␣ and NS5A to the ER membrane via the overexpression of wild-type (but not KD mutant) PI4KIII␣ in CK37-treated cells further supports a scenario in which mass PI4P synthesis occurs prior to PI4KIII␣ and NS5A translocation to the ER (Fig. 15B).
Our previous (35) and present findings suggest that hCK␣ engages in multiple, complex protein-protein interactions with NS5A, PI4KIII␣, and NS5B during viral RC assembly on the ER membrane. The results obtained from the cell culture system and the PI4KIII␣ in vitro activity assay presented here reveal the essential nature of the protein-protein interactions that lead to ternary complex formation to achieve robust PI4KIII␣ activation. Further studies aimed at understanding the domains, motifs, and/or residues within hCK␣, PI4KIII␣, and NS5A that are responsible for mediating the interactions between each set of paired counterparts and at characterizing JFH1 virus mutants harboring these mutations will not only elucidate the underlying molecular bases and allow us to further understand how protein-protein interactions impact the formation of the ternary complex but also reveal how hCK␣ acts as a key modulator to reshape basal and NS5A-stimulated PI4P production and promote viral replication.
Regarding the critical involvement of finely tuned PI4KIII␣ activation in HCV replication (32), our study provides novel insight into the modulatory role of hCK␣ during HCV replication. hCK␣ functions as an indispensable, upstream coordinator of PI4KIII␣ and NS5A by bridging them, reshaping their functions, and retargeting their intracellular localization to the PI4P-enriched, ER-derived membrane platform (Fig. 15B). On the PI4P-enriched membrane, the hCK␣ protein mediates the NS5A-NS5B interaction to assemble a functional RC (35), and PI4KIII␣ activation continues to build the PI4Penriched MW structure to harbor more viral RCs, leading to productive viral replication (Fig. 15B).
HCV production and infection. JFH1 RNA was generated by in vitro RNA transcription and electroporated into Huh7 cells to produce cell culture-derived infectious HCV (HCVcc). The titration and infection of HCVcc at the multiplicity of infection (MOI) of 1 were performed based on a previously described procedure (35). Cells infected with HCVcc were harvested 72 h postinfection for Western blot analysis or confocal microscopy.
Plasmid DNA transfection. For Western blot analysis, confocal microscopy, and PI4P measurements, 2 ϫ 10 5 Huh7, T7/Huh7, T7-control shRNA/Huh7, or T7-hCK␣ shRNA/Huh7 cells were seeded in 6-well plates overnight and were then transfected with 4 g of each plasmid using Lipofectamine 2000 based on a previously described procedure (35). For co-IP analyses, 1 ϫ 10 6 T7/Huh7, T7-control shRNA/Huh7, or T7-hCK␣ shRNA/Huh7 cells were seeded in 6-cm plates overnight and were then transfected with 8 g of each plasmid examined. The appropriate vectors were added to the transfection reactions, and the total DNA amounts in all transfections in a set were identical. The cells were harvested 48 h posttransfection and were lysed with phosphate-buffered saline (PBS) (pH 7.4) containing 1% NP-40, 1% sodium deoxycholate, and a protease inhibitor cocktail for the Western blot analyses or with Tris-HCl (pH 7.4) containing 150 mM NaCl, 1% CHAPS, and a protease inhibitor cocktail for co-IP analysis. Alternatively, the transfected cells were examined by confocal microscopy for the intracellular localization of the proteins or the assessment of PI4P levels. In certain cases, the siRNA-transfected cells were transfected with the plasmids examined. Additionally, the infected cells were transfected with the plasmids and then subjected to inhibitor treatment studies.
siRNA silencing and inhibitor studies and cell viability assay. For the RNAi silencing studies, the cells grown in 6-well plates were transfected with 100 pmol of siRNAs using Lipofectamine RNAiMax (Invitrogen) according to the manufacturer's protocol. At 24 h post-siRNA transfection, the cells were left uninfected or infected HCVcc and were then subjected to a Western blot analysis. Alternatively, the siRNA-transfected cells were transfected with plasmid DNAs prior to Western blotting or co-IP analysis.
For the inhibitor studies, the cells seeded in 6-well plates were infected with or without HCVcc. The inhibitors CK37 and AL-9 were prepared in DMSO and were added to cultured cells at final concentrations of 100 M and 5 M, respectively, at 48 h post-HCVcc infection or 24 h post-DNA transfection. The cells treated with DMSO (vehicle) at the same concentration as that in the inhibitor treatment were used as controls. The cells were harvested 24 h after inhibitor treatment for Western blot analysis or confocal microscopy. Alternatively, the transfected cells were treated with CK37 or PAO at a final concentration of 100 M or 5 M and were subjected to co-IP analysis. The viability of the HCV-infected cells treated with a vehicle or different inhibitors was ascertained using a CellTiter-Glo luminescent cell viability assay kit.
Cycloheximide chases. For the cycloheximide chases, the Huh7 cells grown in 6-well plates were transfected with the control or hCK␣ siRNA. At 48 h after siRNA transfection, the cells were treated with cycloheximide at a final concentration of 100 g/ml and were then collected at different time points post-cycloheximide addition for Western blot analysis. The percentage of PI4KIII␣ detected at different times relative to that detected at time zero in each siRNA transfection, which was arbitrarily assigned a value of 100%, was expressed.
SDS-PAGE, immunoblotting, and co-IP analyses. SDS-PAGE, Western blotting, and immunoblot visualization were performed as indicated previously (35). The levels of the proteins indicated in the immunoblots were quantified using ImageQuant TL software (GE Healthcare), and the relative protein levels were determined. For co-IP analysis, 1 g of the antibody was first incubated with protein A Mag Sepharose beads (28-9670-62; GE Healthcare) at 4°C for 2 h, followed by incubation with cell lysates at 4°C for 6 h. After the beads were washed 5 times in PBS, they were boiled with sample loading buffer and were subjected to SDS-PAGE and immunoblot analyses.
Confocal laser scanning microscopy and PI4P quantification. For the confocal microscopy analysis and quantification of the colocalization of the proteins, the HCV-infected or DNA-transfected cells were fixed, immunostained with specific antibodies, and examined by confocal microscopy based on previously described procedures (35). The images were edited using Zen 2011 software. The colocalization of the indicated proteins was quantified using ImageJ or MBF ImageJ from 20 randomly selected cells as previously described (35).
For the PI4P quantification, the cells were prepared as described previously (24). Z stack images of cells were acquired with a Zeiss LSM 780 confocal laser scanning microscope using a 40ϫ objective lens. Then the z projections of the z stack image were generated with the "sum slices" option in ImageJ, followed by thresholding the signal intensity of PI4P staining. The PI4P amount in 300 cells was determined by defining the cell area, and the integrated density (IntDen) value of the PI4P signal in each cell was obtained by the "analyze particles" function in ImageJ.
In vitro PI4KIII␣ assay. PI4KIII␣ in vitro activity was determined with 30 ng of purified recombinant PI4KIII␣ in the presence or absence of 30 ng each of purified recombinant NS5A and/or hCK␣ according to the method previously described by Bianco et al. (41) using an ADP-Glo kinase assay kit. Bovine serum albumin was added into the kinase reaction mixture so that all reaction mixtures contained the same total amounts of protein. Reactions were also performed without the PI:PS substrate to detect contaminating ATPase activity present in the test conditions. This activity was subtracted from the measured kinase activity. Kinase activity was expressed as the fold change in the luciferase activity under test conditions relative to that detected in the presence of PI4KIII␣ alone, which was arbitrarily assigned a value of 1.
Data and statistical significance analyses. The results from the Western blot analysis and the relative protein quantification in the immunoblots were obtained from three independent studies. The Western blot data for representative sets are shown in the figures. All data are presented as the means Ϯ standard deviations (SD), and statistical analyses were performed using a two-tailed, unpaired Student t test. Differences between the two indicated settings were considered statistically significant at a P value of Ͻ0.05.