|Home | About | Journals | Submit | Contact Us | Français|
The hepatitis C virus NS5A protein plays a critical role in virus replication, conferring interferon resistance to the virus through perturbation of multiple intracellular signaling pathways. Since NS5A is a phosphoprotein, it is of considerable interest to understand the role of phosphorylation in NS5A function. In this report, we investigated the phosphorylation of NS5A by taking advantage of 119 glutathione S-transferase-tagged protein kinases purified from Saccharomyces cerevisiae to perform a global screening of yeast kinases capable of phosphorylating NS5A in vitro. A database BLAST search was subsequently performed by using the sequences of the yeast kinases that phosphorylated NS5A in order to identify human kinases with the highest sequence homologies. Subsequent in vitro kinase assays and phosphopeptide mapping studies confirmed that several of the homologous human protein kinases were capable of phosphorylating NS5A. In vivo phosphopeptide mapping revealed phosphopeptides common to those generated in vitro by AKT, p70S6K, MEK1, and MKK6, suggesting that these kinases may phosphorylate NS5A in mammalian cells. Significantly, rapamycin, an inhibitor commonly used to investigate the mTOR/p70S6K pathway, reduced the in vivo phosphorylation of specific NS5A phosphopeptides, strongly suggesting that p70S6 kinase and potentially related members of this group phosphorylate NS5A inside the cell. Curiously, certain of these kinases also play a major role in mRNA translation and antiapoptotic pathways, some of which are already known to be regulated by NS5A. The findings presented here demonstrate the use of high-throughput screening of the yeast kinome to facilitate the major task of identifying human NS5A protein kinases for further characterization of phosphorylation events in vivo. Our results suggest that this novel approach may be generally applicable to the screening of other protein biochemical activities by mechanistic class.
Phosphorylation is a common posttranslational processing event in the synthesis of proteins, and changes in the degree of protein phosphorylation play an important role in modulating protein function. Many cellular processes, including signal transduction, transcription, protein synthesis, and cell growth and differentiation, are regulated by phosphorylation-dependent mechanisms (17, 19). Therefore, it is not surprising that the protein kinases, a family of enzymes that catalyze the phosphorylation of proteins, comprise one of the largest families of genes in eukaryotes. The protein kinase complement of the human genome (the human “kinome”) is estimated to comprise about 1.7% of all human genes, with a total of 518 genes identified according to a cataloging approach published recently (19). The presence of such a large number of kinases in the human genome represents a significant challenge to protein kinase studies, suggesting that it would be extremely powerful to develop high-throughput approaches for characterizing kinase-substrate reactions. Although a recent study describes the use of a microarray format for the high-throughput analysis of substrate specificity of yeast protein kinases (36), the use of such an approach to identify the human kinase(s) responsible for phosphorylation of a biological substrate has not been previously described.
There has been much interest in characterizing the role of protein phosphorylation in mechanisms associated with viral infection, pathogenesis, and persistence (16, 35). In this regard, it is well established that the hepatitis C virus (HCV) nonstructural 5A (NS5A) protein is a phosphoprotein and that phosphorylation of NS5A is highly conserved among HCV genotypes and other members of the Flaviviridae family (13, 23, 25). This led to the idea that phosphorylation may play a role in the flavivirus life cycle and inspired our interest in obtaining a better understanding of NS5A phosphorylation, particularly since this protein has been implicated in resistance of HCV to interferon treatment (4, 5, 27). NS5A is phosphorylated primarily on serine residues both in vitro and in vivo (13, 15, 22, 23, 25, 30). Although the effects of various kinase inhibitors have implicated a cellular serine/threonine kinase(s) from the CMGC kinase group (25), it should be noted that this group contains several kinase families, and these studies did not identify a particular kinase in the phosphorylation of NS5A. More recent in vitro kinase assays specifically demonstrated that NS5A can be phosphorylated by casein kinase 2 (CK2) (12, 14) and protein kinase A (PKA) (12).
The interferon-induced double-stranded RNA-activated protein kinase, PKR, is a key component of the antiviral and antiproliferative effects induced by interferon. NS5A forms a complex with PKR, disrupting PKR dimerization, which results in repression of PKR function and inhibition of PKR-mediated eIF-2α phosphorylation. Whether NS5A is a substrate of PKR is yet to be determined. There is no PKR homologue in the yeast kinome, but the yeast kinase GCN2 still has sequence homology with PKR. In this study, we were particularly interested in finding out whether GCN2 is a kinase of NS5A.
In this report, we further investigated the phosphorylation of NS5A using a kinome-scale high-throughput screening approach in order to identify human protein kinases that phosphorylate NS5A. The approach is based on a biochemical genomics strategy, originally described by Martzen et al. (20), for constructing an array of molecular constructs for expression of all yeast kinases (the yeast kinome) (11, 19). Our long-term interests are to gain insight into the role of phosphorylation in the ability of NS5A to modulate a multitude of cellular signaling pathways. Although there are already more than 518 known human kinases, more than four times the number of the kinases in yeast, we reasoned that the yeast kinome would represent a good model for beginning a global inspection of NS5A phosphorylation activity, since phosphorylation of NS5A occurs primarily at serine residues (13, 15, 22, 23, 25, 30) and the major families of serine/threonine kinases are conserved between yeast and humans. The use of the yeast model was based on the idea that the kinase(s) phosphorylating NS5A is very likely to be evolutionarily conserved. This is supported by our previous findings that Ser 2194 is a major phosphorylation site among all HCV genotypes and this site can be phosphorylated by yeast, insect, and mammalian kinases (13). We screened 119 glutathione S-transferase (GST)-tagged protein kinases purified from Saccharomyces cerevisiae in a global effort to identify yeast kinases capable of phosphorylating NS5A in vitro. BLAST searches were performed to identify the closest human sequence homologs, and commercially available sources of human protein kinases were used for subsequent in vitro NS5A kinase assays. We demonstrated that several of these human kinases were capable of phosphorylating NS5A in vitro, some of which play a major role in protein translation and antiapoptotic pathways that are regulated by NS5A. Furthermore, the phosphopeptide pattern of NS5A isolated from COS-1 cells suggests that the phosphorylation sites used in vivo were similar to those observed in vitro, allowing us to identify kinases of particular interest for further characterization of NS5A phosphorylation. Finally, rapamycin studies revealed that p70S6 kinase and potentially closely related family members phosphorylate NS5A inside the cell.
The human kinases AKT1, MEK1, MKK6, MKK7β1, SGK1, p70S6K, CHK2, PDK1, and the rat kinases CK1 delta and p90S6K were purchased from Upstate Biotechnology (Lake Placid, N.Y.). The purity and activity of these kinases were certified by the manufacturer but were not independently verified in this study. All materials used for GST-yeast kinase purification were obtained from Amersham Biosciences Corp. (Piscataway, N.Y.). Radiolabeled ATP, [γ-32P]ATP, and [32P]orthophosphoric acid in water were purchased from NEN Life Science Products, Inc. (Boston, Mass.). Dulbecco's modified eagle medium (DME), phosphate-free DME, Hanks' balanced salt solution, and calf serum were purchased from Gibco/Invitrogen Corp. (Grand Island, N.Y.). The S. cerevisiae protein kinases (119 constructs), cloned in the plasmid pYEX4T-1 and expressed in the EJ758 yeast strain (36), were provided by Eric M. Phizicky (University of Rochester, Rochester, N.Y.).
Colonies representing transformants of the EJ758 yeast strain were obtained on plates containing synthetic minimal (SD) minus Ura media. These colonies were then used to inoculate SD minus Ura liquid medium and grown for 8 h. At this time, each inoculum was split 1 to 10 into SD minus Ura minus Leu medium and grown for an additional 16 h. When the cultures reached an optical density at 600 nm of 0.8, the cells were induced with 0.5 mM CuSO4 for 2 to 3 h prior to harvesting by centrifugation at 4°C for 5 min at 1,500 × g. After induction, the cells were washed twice in two to four pellet volumes of ice-cold water. The washed cell pellet was resuspended in 1 ml of homogenization buffer (50 mM Tris-HCl [pH 7.5], 1 mM EDTA, 4 mM MgCl2, 1 mM dithiothreitol [DTT], 10% glycerol, 250 mM NaCl) containing 1× protease inhibitor mix, and extracts were made with glass beads (15 pulses, 30 s each at 4°C) using a multichannel vortexer. Extracts were transferred to 15-ml conical tubes and centrifuged at 4°C for 5 min at 450 × g. The supernatants were then centrifuged at 4°C for 30 min at 10,000 × g. The GST-kinases were purified from the resulting supernatants by glutathione-agarose chromatography (20). Briefly, the supernatants were incubated with prewashed reduced glutathione-agarose beads for 1 h at 4°C and then extensively washed in the same buffer (50 mM Tris-HCl [pH 7.5], 500 mM NaCl, 10% glycerol, 4 mM MgCl2). Aliquots of 33% slurries (150 μl each) were flash frozen in an ethanol-dry ice bath and stored at −80°C for later use.
Baculovirus recombinant NS5A (2 μl, 100 ng), purified from Sf9 cells as previously described (13), was added to 50 μl of purified GST-kinase bound to glutathione-agarose beads prepared as described above. The final amount of kinase added was normalized based on the relative recovery of each purified kinase as determined by Western blotting with an antibody to GST. The phosphorylation reaction was carried out at 30°C for 60 min in a final volume of 60 μl containing yeast kinase buffer (50 mM Tris-HCl [pH 7.2], 10 mM MgCl2, 1 mM DTT) supplemented with protease and phosphatase inhibitors and a mixture of [γ-32P]ATP and ATP with a predetermined specific activity (~5,000 dpm/pmol of ATP), such that the reaction mixture contained at least a 1,000-fold excess of ATP over the NS5A protein concentration. Human recombinant casein kinase 2 (hCK2) was used as a positive control to demonstrate NS5A phosphorylation. Glutathione-agarose beads, incubated with the yeast EJ758 strain expressing an empty glutathione vector, were used as a negative control to demonstrate that phosphorylation of NS5A resulted from the appropriate GST-kinase and not from the nonspecific binding of other proteins to either GST or the beads. The kinase reaction was terminated by flash freezing the samples in an ethanol-dry ice bath. An equal volume of 2× SDS protein loading buffer (125 mM Tris-HCl [pH 6.8], 20% glycerol, 2% SDS, 2% β-mercaptoethanol, 0.02% bromophenol blue) was added, and the samples were boiled at 95 to 100°C for 5 min prior to analysis by SDS-polyacrylamide gel electrophoresis (PAGE) (10% gel). After resolving the samples, the gel was dried using a slab gel dryer SGD 4050 (Savant) for 5 h at 80°C and then exposed to X-ray film at −70°C with an intensifying screen.
The phosphorylation reaction was carried out in the presence or absence of 100 ng of baculovirus recombinant NS5A (purified from Sf9 cells) in a final volume of 40 μl of human kinase buffer (20 mM morpholinepropanesulfonic acid [pH 7.2], 25 mM β-glycerol phosphate, 5 mM EGTA, 1 mM sodium orthovanadate, 1 mM DTT, 10 mM MgCl2) containing a mixture of [γ-32P]ATP and ATP with a predetermined specific activity (~5,000 dpm/pmol of ATP) such that the reaction mixture contained at least a 1,000-fold excess of ATP over the NS5A protein concentration. The appropriate kinase was added and assayed for phosphorylation of NS5A. The buffer condition used was that recommended by the enzyme supplier. The kinases assayed, and the amount used per reaction, were 20 mU of rat casein kinase 1 (rCK1 delta), 0.5 μg of hAKT1, 0.5 U of hMEK1, 0.05 μg of hMKK6, 0.2 μg of hMKK7β1, 0.025 μg of hSGK1, 10 mU of hp70S6K, 20 mU of rp90S6K, 10 ng of hPDK1, 100 mU of hCHK2, or 0.05 μg of hCK2. For samples containing rCK1 delta, hAKT1, hSGK1, hCHK2, hp70S6K, or hCK2 (positive control), the reaction was allowed to proceed for 10 min at 30°C. For samples containing hPDK1, rp90S6K, hMEK1, hMKK6, hMKK7β1, or hCK2 (positive control), the reaction was allowed to proceed for 30 min at 30°C.
For two-dimensional phosphopeptide mapping, the phosphorylated samples were resolved by SDS-10% PAGE and then transferred to a nitrocellulose membrane. Radiolabeled phosphoproteins were detected by autoradiography. The 32P-containing bands were excised, kept wet in water, and incubated in 0.5% polyvinylpyrrolidone-100 mM acetic acid for 1 h at 37°C in order to block nonspecific adsorption of proteases. Control slices were prepared from the same-molecular-weight region of lanes corresponding to phosphorylation reactions containing kinase only or NS5A only. The slices were then washed thoroughly with deionized water (five times, 1 ml per wash) followed by 50 mM ammonium bicarbonate, pH 8.3 (five times, 1 ml per wash), and finally resuspended in 80 μl of 50 mM ammonium bicarbonate, pH 8.3. Proteolytic digestion was accomplished by adding 1 μg of trypsin and incubating for 3 h at 37°C, followed by another 1-μg addition and incubation at 37°C overnight. The next day, a third addition of 1 μg of trypsin was made, and the samples were incubated for 2 h at 37°C prior to adding 1 μg of chymotrypsin and digesting for 3 h at 37°C. A second addition of 1 μg of chymotrypsin was made, and the samples were incubated at 37°C for 3 h or overnight. The supernatants containing soluble peptides were then transferred to a separate tube, and the slices were washed with 400 μl of water to extract any remaining peptides. The pooled supernatant and wash were dried with a speed vac, and the peptides were subsequently suspended in 10 μl of buffer (7.8% acetic acid, 2.2% formic acid). Phosphopeptide recovery was monitored at every step by Cerenkov counting. The resuspended samples were loaded onto cellulose thin-layer chromatography (TLC) plates (20 cm by 10 cm) and electrophoresed in the first dimension (20 cm) at 1,000 V for 60 min in reverse polarity (positive to negative). After the plates were dried overnight in the fume hood, separation in the second dimension was achieved by ascending chromatography in 37.5% n-butanol, 25% pyridine, and 7.5% acetic acid. The plates were dried extensively, and radiolabeled phosphopeptides were detected by autoradiography.
COS-1 cells (0.5 × 106) were cultured in a T-25 flask for 15 h prior to transfection with 10 μg of a pcDNA3 construct encoding the NS5A-1b protein. Transfections were performed using the SuperFect kit (Qiagen, Studio City, Calif.) as described by the manufacturer. A second flask was transfected with 10 μg of empty pcDNA3 vector for use as a negative control. After 15 h of gene expression, the cells were washed with phosphate-free DME prior to labeling. The cells were then labeled by incubating each T-25 flask with 1.5 ml of phosphate-free DME supplemented with 10% dialyzed calf serum (Gibco/Invitrogen Corp.) and [32P]orthophosphoric acid in water (1 mCi/T-25 flask) for 4 h at 37°C. Each flask was then washed twice with 1 ml of Hanks' balanced salt solution (Gibco/Invitrogen Corp.) prior to incubation in 80 μl of Triton lysis buffer (20 mM Tris-HCl [pH 7.5], 75 mM NaCl, 75 mM KCl, 20% glycerol, 4 mM MgCl2, 1 mM DTT, 1 mM EDTA, 0.5% TX-100) containing 1× protease inhibitor mix and phosphatase inhibitors (1 mM Na3VO4, 25 mM β-glycerophosphate) for 15 min at 4°C. The cells were then removed from the flasks with a cell scraper, and insoluble materials were pelleted by centrifugation at 16,000 × g for 1 min. The resulting supernatants were used for immunoprecipitation of NS5A as described below.
COS-1 cells (0.5 × 106, in a T-25 flask) were transfected as described under “In vivo [32P]orthophosphoric acid labeling.” The cells were washed with phosphate-free DME and then incubated with or without 20 nM rapamycin. After 1 h at 37°C, the cells were labeled with phosphate-free DME supplemented with 10% dialyzed calf serum (Gibco/Invitrogen Corp.) and [32P]orthophosphoric acid in water for 4 h. The cells were then processed as described under “In vivo [32P]orthophosphoric acid labeling.”
Protein lysates prepared from 0.5 × 106 cells were incubated with 1 μg of anti-NS5A monoclonal antibody for 3 h at 4°C and then mixed with 50 μl of protein G-agarose beads (Boehringer Mannheim) for 2 h at 4°C (final volume, 1 ml). The beads were then washed three times with Triton lysis buffer containing 1× protease inhibitor mix and phosphatase inhibitors (1 mM Na3VO4, 25 mM β-glycerophosphate) prior to suspension in 50 μl of 2× SDS protein loading buffer (20% [vol/vol] glycerol, 4% [wt/vol] SDS, 0.125 M DTT). The immunoprecipitated NS5A was then resolved by SDS-10% PAGE and transferred to a nitrocellulose membrane, and the NS5A phosphorylation band was detected by autoradiography.
Immunoblot analysis was performed as previously described (28). Briefly, blots were probed with a monoclonal antibody against NS5A (4 μg/ml; ViroStat, Portland, Maine), followed by a 1:20,000 dilution of donkey anti-mouse immunoglobulin G conjugated to horseradish peroxidase (Jackson ImmunoResearch Laboratories, Inc., West Grove, Pa.). NS5A was then visualized by enhanced chemiluminescence (Amersham Biosciences Corp.).
Phosphorylation of NS5A is believed to affect its biological properties. However, NS5A lacks intrinsic protein kinase activity, suggesting that its phosphorylation results from interaction with one or more cellular kinases (10, 12-15, 25). Previous in vitro studies demonstrated that CK2 and cAMP-dependent protein kinase (PKA) are able to phosphorylate NS5A (12, 14). Since NS5A performs a host of diverse functions, including the modulation of several signal transduction pathways (6, 8, 28), we expect that NS5A may be phosphorylated by additional cellular kinases. To further investigate the phosphorylation of NS5A, we have taken advantage of GST-yeast kinase fusion proteins to screen the S. cerevisiae kinome for kinases that phosphorylate NS5A in vitro.
The yeast genome is predicted to contain 124 protein kinases, and 119 of these have been successfully expressed as GST-kinase fusion proteins (36). Eight of these constructs were capable of phosphorylating NS5A under the conditions described here. Table Table11 provides a summary of the distribution by major groups of the GST-yeast kinases that phosphorylated NS5A in vitro, and autoradiographs from representative samples are presented in Fig. Fig.1.1. In order to eliminate the possibility that phosphorylation would be influenced by the presence of a tag used to facilitate protein purification, we used native NS5A that was purified to homogeneity by sequential DEAE-Sepharose and Mono Q anion-exchange chromatography (13). hCK2, which phosphorylates NS5A in vitro (14), was used as a positive control to demonstrate NS5A phosphorylation (Fig. (Fig.1B,1B, lanes 14 and 28).
The yeast kinases that phosphorylated NS5A did so with variable efficiencies. For example, comparison of lanes 2 and 3 demonstrates that the yeast kinase YPL204W (Fig. (Fig.1B,1B, lane 3) generated a stronger 32P signal associated with NS5A phosphorylation than did construct YNL154C (Fig. (Fig.1B,1B, lane 2). This is not surprising, since these kinases are likely to have different affinities for NS5A and may also have different requirements for optimal activity. Because of the vast number of kinases screened in this study, the phosphorylation reactions were performed under identical conditions.
Several of the reactions revealed additional bands of phosphorylation that were presumably associated either with autophosphorylation of the kinase or with contaminants copurifying with the respective kinase. For example, although the YOL128C construct was unable to phosphorylate NS5A, a higher-molecular-weight band, presumably corresponding to the autophosphorylated kinase, was detected (Fig. (Fig.1,1, lane 16). When the reaction was carried out in the presence of yeast extracts from cells transfected with an empty glutathione vector, no phosphorylation of NS5A was detected (Fig. (Fig.1,1, lanes 13 and 15). These results demonstrate that phosphorylation of NS5A is not associated with proteins nonspecifically bound to either GST or the glutathione-agarose beads. In addition, the yeast kinase GCN2, which has sequence homology with the interferon-inducible protein kinase PKR, did not phosphorylate NS5A (data not shown). This is consistent with our previous demonstrations that although NS5A can interact with PKR, it is not phosphorylated by this kinase (6).
We used the sequences of the eight yeast kinases (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi) that phosphorylated NS5A to perform a National Center for Biotechnology Information BLAST search (http://www.ncbi.nlm.nih.gov/BLAST) in order to identify human kinases with sequence homology. The human kinases identified by this approach included candidates from the CK1, CMGC, STE, CaMK, and AGC groups (Table (Table2).2). Consistent with previous studies demonstrating that CK2 (14) and PKA (12) phosphorylate NS5A, CK2 was identified as a top candidate for the CMGC group, represented by the yeast constructs YIL035C and YOR061W, and PKA was identified as a top candidate for the AGC group, represented by the yeast construct YDR466W. These observations support the validity of this approach and provided additional confidence that analysis of the human protein kinases present on this list would serve as a good starting point for subsequent in vitro NS5A phosphorylation studies. We therefore screened all the commercially available kinases corresponding to the top five nonredundant hits for each kinase group reported in Table Table2,2, including CK2, MEK1, MKK7β1, CHK2, AKT1, p70S6K, SGK1, and PDK1. Because human kinase homologues were not available for CK1 delta and p90S6K, we screened the respective rat kinases.
Consistent with previous studies (14), we found that hCK2 phosphorylated NS5A in vitro under the conditions described here (Fig. (Fig.2A,2A, lane 3). Similarly, rCK1 delta also phosphorylated NS5A (Fig. (Fig.2A,2A, lane 5). It should be noted that a faint lower-molecular-weight radiolabeled band was also detected in the phosphorylation reaction with hCK2. Since this band was also detected with hCK2 in the absence of NS5A in the kinase assay (Fig. (Fig.2A,2A, lane 2), we suspect that this corresponds to phosphorylation of a contaminant present in the hCK2 preparation.
We also found that NS5A was phosphorylated by members of the STE group, including the mitogen-activated protein kinase kinase (MAPK kinase) family members hMEK1 (Fig. (Fig.3A,3A, lane 4), hMKK6 (Fig. (Fig.3B,3B, lane 4), and hMKK7β1 (Fig. (Fig.3C,3C, lane 4). Although hMKK6 was not among the top five nonredundant hits, we chose to screen this kinase because it is nevertheless a closely related member of the MAPK family and was readily available. For each kinase tested, hCK2 was used as a positive control (Fig. 3A, B, and C, lanes 1). In each case, a second phosphorylated band was detected. Analysis of the kinases alone (Fig. 3A, B, and C, lanes 5) revealed that these bands were detected in the absence of NS5A, demonstrating that they are associated with the kinases. The observed molecular weights are consistent with those expected for each kinase, suggesting that these bands correlate to autophosphorylated kinases. This idea is supported by previous reports describing the autophosphorylation of these kinases (2, 3). However, we cannot rule out the possibility that these bands represent contaminants in the kinase preparations.
Two of the top five candidates from the AGC kinase group, hAKT1 (Fig. (Fig.3D,3D, lane 2) and hp70S6K (Fig. (Fig.3E,3E, lane 2), phosphorylated NS5A in vitro. As was the case with kinases from the STE group, a second band of 32P-labeled material that is likely to be an autophosphorylated form of the kinase was also detected (Fig. (Fig.3D3D and and3E,3E, lanes 1) (26, 32). The remaining kinases from the AGC group (hPDK1, hSGK1, and rp90S6K) did not phosphorylate NS5A (data not shown), nor did hCHK2, the only commercially available kinase among the top five nonredundant hits of the CaMK group (Fig. (Fig.3F,3F, lane 4). However, a member of the CaMK group not among the top five hits, Ca2+ calmodulin-dependent kinase II, was able to phosphorylate NS5A in vitro (data not shown). Although we did not independently verify whether the kinases that failed to phosphorylate NS5A were able to phosphorylate a known substrate, the presence of a signal associated with autophosphorylation suggests that these kinases were active.
The results described above demonstrated the utility of a high-throughput screen of the yeast kinome and revealed that human kinases from several major kinase groups were able to phosphorylate NS5A. In order to determine whether the sites of phosphorylation were unique for the different groups of kinases, we compared the phosphopeptide maps of in vitro-phosphorylated NS5A. Figure Figure2B2B shows the tryptic/chymotryptic phosphopeptide maps of NS5A phosphorylated with either rCK1 delta or hCK2. One labeled phosphopeptide detected in the presence of rCK1 delta, denoted by the arrow, was not detected with hCK2, suggesting that the phosphorylation sites may differ. Phosphopeptide maps generated with kinases from the STE kinase group (i.e., MAPK family member MEK1 or MKK6) and the AGC kinase group (AKT1 and p70S6K) are shown in Fig. Fig.44 and and5,5, respectively. The phosphopeptide patterns are quite distinct depending on the kinase used, demonstrating that NS5A can serve as a substrate for a variety of kinases, resulting in differential phosphorylation at multiple serine and threonine residues in vitro. Phosphopeptide maps were also generated from phosphorylation reactions carried out in the absence of NS5A in order to monitor for the presence of contaminating phosphopeptides (negative control). This was accomplished by cutting out and processing the same-molecular-weight region of the nitrocellulose membrane from lanes containing kinase only (see Materials and Methods). In some instances, we detected a significant amount of radiolabeled material in the absence of NS5A where no phosphopeptides were expected (i.e., negative control phosphopeptide maps with hAKT1 or hp70S6K) (Fig. (Fig.5).5). We suspect that this resulted from removal of some of the closely migrating kinase when the region of the filter corresponding to the molecular weight of NS5A was cut out for processing (see Fig. 3D and E, lanes 2). NS5A phosphopeptides detected in vitro that match with NS5A phosphopeptides detected in vivo are shown in Fig. Fig.66 and discussed in more detail below.
In order to investigate the biological relevance of in vitro phosphorylation of NS5A by the kinases identified above, we next examined the phosphorylation of NS5A in cell culture. Briefly, COS-1 cells were transfected with a plasmid encoding the isogenic NS5A used for our in vitro studies, and in vivo-labeled NS5A was generated using [32P]orthophosphate. Figure Figure6A6A shows an autoradiograph of NS5A immunoprecipitated with antibody specific to NS5A, resolved by SDS-10% PAGE, and transferred to a nitrocellulose membrane. Consistent with previous findings (22, 30), two phosphorylated species of NS5A were detected. The identity of these bands was confirmed by immunoblot analysis with NS5A-specific monoclonal antibody (Fig. (Fig.6B).6B). In addition, a third higher-molecular-weight species (150 kDa) was also detected during immunoprecipitation with cell lysates expressing NS5A (Fig. (Fig.6A,6A, lane 1). We suspect that this may represent a binding partner of NS5A, since it was not detected in cell lysates transfected with a control vector lacking NS5A (Fig. (Fig.6A,6A, lane 2). Each of the phosphorylated species of NS5A was subsequently processed for two-dimensional phosphopeptide mapping. Since the phosphopeptide maps generated from the two species of phosphorylated NS5A were nearly identical, Fig. Fig.6C6C shows the tryptic/chymotryptic phosphopeptide map resulting from the lower-molecular-weight species. Immunoprecipitates from protein lysates of COS-1 cells transfected with a control vector lacking NS5A were run in parallel, and the same-molecular-weight region of the nitrocellulose membrane was processed to monitor for the presence of contaminating phosphopeptides (Fig. (Fig.6C,6C, negative control). Phosphopeptides unique to NS5A, as determined by comparison with the negative control, are labeled a to g.
The autoradiography films from the in vivo and in vitro phosphopeptide maps were then overlaid for comparison. The comigrating spots observed in the in vitro mapping experiments are given the same letter as for the in vivo map (compare Fig. Fig.44 and and55 with Fig. Fig.6).6). Interestingly, six of the seven phosphopeptides (labeled a to f) detected in vivo were also detected during in vitro phosphorylation of NS5A with hp70S6K (Fig. (Fig.5).5). Four of these phosphopeptides (labeled b, d, e, and f) were also detected when NS5A was phosphorylated with hAKT1 (Fig. (Fig.5).5). By contrast, only a small number of the peptides detected in vivo could also be seen during in vitro phosphorylation with either MEK1 or MKK6 (Fig. (Fig.4,4, phosphopeptide c for MEK1; phosphopeptides b, c, and d for MKK6).
One in vivo phosphopeptide (labeled g) was not detected in any of the in vitro phosphopeptide maps presented here. It is possible that this phosphopeptide resulted from the phosphorylation of NS5A by another kinase not characterized by two-dimensional phosphopeptide mapping in this study. Last, it is important to note that the in vivo phosphopeptide map reported here is more complex than what we previously observed (13). The conditions used here for enzymatic digestion of NS5A were more aggressive than those used in our previous study and more closely reflect those used to generate the distinct and complex phosphopeptide patterns reported for NS5A-1a (25). Therefore, we suspect that the increased complexity of the phosphopeptide maps observed for NS5A-1b is attributable to more complete enzymatic digestion.
One of the most interesting and important kinases identified by our approach was p70S6K. It is well described in the literature that the mammalian target of rapamycin (mTOR) pathway is a major pathway activating p70S6K (21, 29, 33) and that rapamycin (21, 29, 33) is commonly used to study p70S6K activity. We therefore used rapamycin to investigate the in vivo contribution of p70S6K to the phosphorylation of NS5A. We already observed that on a one-dimensional SDS-PAGE gel, the phosphorylation of NS5A was significantly reduced by addition of rapamycin (data not shown), suggesting that the p70S6K activated by mTOR could play a role in the phosphorylation of NS5A. We then analyzed the phospho-NS5A-peptide map obtained from cells treated with or without rapamycin, and as shown in Fig. Fig.6D,6D, the phosphorylation of three of the major in vivo NS5A peptides phosphorylated by p70S6K (spots b, d, and f) was significantly reduced. However, as shown in Fig. Fig.6D,6D, rapamycin does not eliminate all the NS5A phosphopeptides (for example, spots a, c, and g), which suggests that this inhibitor is acting with specificity. This provides additional critical evidence that p70S6K and potentially closely related members phosphorylate NS5A in vivo.
Based on our findings, we suggest that hp70S6K is a human kinase responsible for phosphorylating NS5A in vivo. An examination of the sequence of the NS5A-1b species used here revealed sequences closely related to the consensus phosphorylation motifs of p70S6K (K/R-X-R-X-X-S/T-X). In addition, a slight variation in this motif is present around serine 2194, a highly conserved major phosphorylation site in NS5A-1b and in NS5A proteins from other HCV genotypes (13). In contrast, we were unable to detect this motif near serine 2321, a major phosphorylation site in the HCV 1a genotype (24). Further experiments using specific inhibitors or RNA interference (RNAi) strategies aimed at inhibiting or down-regulating the expression of the various kinases identified may provide further insight into the relative contributions of these kinases to the in vivo phosphorylation of NS5A.
It is likely that phosphorylation of NS5A plays a role in regulating the diverse functions of this protein. In this regard, we have been particularly interested in understanding the significance of NS5A phosphorylation in the ability of HCV to interfere with cellular signaling. For example, the interaction between NS5A and Grb2 may inhibit the MAPK pathway, resulting in down-regulation of cellular mRNA translation (22). In addition, NS5A plays a role in regulating viral mRNA translation initiation from the HCV internal ribosome entry site (9). Given the ability of NS5A to regulate mRNA translation, it is intriguing that NS5A was phosphorylated by both hAKT1 and hp70S6K, cellular kinases that themselves play important roles in translational control (1, 31, 34). Phosphorylation of NS5A by AKT and p70S6K could also induce HCV internal ribosome entry site-mediated translation, thereby facilitating HCV replication and persistence.
Clearly, the complexity of the in vivo phosphorylation patterns observed here warrants further experimentation to determine the significance of these findings. It is likely that multiple kinases, in addition to p70S6K, contribute to NS5A phosphorylation, including AKT, MEK1, and MKK6. In addition, although it is possible to positively identify kinases that phosphorylate NS5A in vivo, it is more difficult to rule out the possibility that other kinases may also contribute to NS5A phosphorylation. In this regard, it is also important that some of the kinases that phosphorylated NS5A in vitro may have no relevance in vivo and that in vitro assays may not detect biologically relevant NS5A kinases that require additional regulatory factors normally present in vivo. Nevertheless, our approach identified several kinases capable of phosphorylating NS5A. This information should facilitate the identification of the protein kinases that phosphorylate NS5A in liver cells, the natural site of HCV replication (7). Further studies aimed at evaluating the role of the identified NS5A kinases using in vivo model systems, such as the HCV replicon (18), may provide a better understanding of the role of NS5A phosphorylation during the HCV life cycle.
In summary, this study illustrates the use of a kinome-scale high-throughput screening approach for identifying protein kinases of interest. We propose that this approach represents an ideal model for other studies aimed at identifying protein kinases for further investigation of functionally relevant phosphorylation events. The development of similar approaches for characterizing gene function directly at the protein level should contribute significantly to the success of efforts in functional genomics, and such approaches may be generally applicable to the high-throughput purification and biochemical analysis of other proteins by mechanistic class.
This work was supported by Public Health Service grant R01-AI47304 from the National Institute of Allergy and Infectious Diseases and grant P30DA015625 to the Center for Functional Genomics & HCV-Associated Liver Disease from the National Institute on Drug Abuse.
We thank Eric M. Phizicky at the University of Rochester, Rochester, N.Y., for providing us with the S. cerevisiae protein kinase constructs cloned in the plasmid pYEX4T-1 and expressed in the EJ758 yeast strain.