Inhibition of the Interferon- Inducible Protein Kinase PKR by HCV E2 Protein

See allHide authors and affiliations

Science  02 Jul 1999:
Vol. 285, Issue 5424, pp. 107-110
DOI: 10.1126/science.285.5424.107


Most isolates of hepatitis C virus (HCV) infections are resistant to interferon, the only available therapy, but the mechanism underlying this resistance has not been defined. Here it is shown that the HCV envelope protein E2 contains a sequence identical with phosphorylation sites of the interferon-inducible protein kinase PKR and the translation initiation factor eIF2α, a target of PKR. E2 inhibited the kinase activity of PKR and blocked its inhibitory effect on protein synthesis and cell growth. This interaction of E2 and PKR may be one mechanism by which HCV circumvents the antiviral effect of interferon.

Hepatitis C virus (HCV) is an emerging virus of great medical importance and almost always causes chronic infections. The only available therapy is interferon (IFN), but most HCV isolates are resistant (1). IFN normally acts by inducing transcription of several antiviral genes, including the double-stranded RNA-activated protein kinase (PKR), which inhibits protein synthesis by phosphorylation of the translation initiation factor eIF2α. Many viruses have evolved strategies to overcome the antiviral effects of PKR (2). One of the HCV nonstructural proteins, NS5A, has been shown to bind to and inhibit PKR (3); however, the NS5A-PKR interaction can explain only those cases of IFN resistance that exhibit NS5A quasispecies divergence after IFN therapy (4). Such divergence has not been observed in most patients.

Our investigation of the IFN resistance of HCV focused on E2, an outer protein of the viral envelope that may participate in virus binding to target cells. E2 shows considerable sequence variation among different HCV isolates, and it induces neutralizing antibodies. E2 from most HCV isolates contains a 12–amino acid sequence that is similar to the PKR autophosphorylation site (5) and the eIF2α phosphorylation site, a target of PKR. The sequences of the PKR-eIF2α phosphorylation homology domain (PePHD) (Fig. 1A) of the more IFN-resistant HCV genotypes (1a and 1b) more closely resemble the sequences of PKR and eIF2α than do the corresponding sequences of the less resistant HCV genotypes (2a, 2b, and 3a) (Fig. 1B). This sequence is highly conserved within each genotype.

Figure 1

HCV E2 shares sequence identity with eIF2α and PKR. (A) Structural diagram of HCV E2. (B) The sequences surrounding the PKR phosphorylation site in eIF2α (amino acids 48 to 54) and surrounding the autophosphorylation sites (amino acids 79 to 90) in the RNA-binding domain of PKR are shown (24). E2 sequences from the various HCV genotypes are aligned. Symbols: *, similar amino acid to PKR; !, identical to PKR; underlined, autophosphorylation sites in PKR and eIF2α phosphorylation site by PKR; and •, sequence identity with HCV-1a. IFN res. and IFN sen. designate IFN resistant and sensitive, respectively.

To determine if E2 physically interacts with PKR, we performed an in vitro binding assay with histidine (His)–tagged PKR and [35S]methionine-labeled E2, synthesized from a construct derived from the HCV genotype 1a strain (6) (Fig. 2A). E2 bound to both wild-type PKR (lane 2) and a mutant PKR (K296R) that has a mutation in the adenosine triphosphate (ATP)–binding domain and is catalytically inactive (7) (lane 3), suggesting that autophosphorylation is not required for the E2-PKR interaction. Under the same conditions, only a trace amount of an E2 deletion mutant E2ΔC, which lacks the PePHD, bound to PKR (six times less as compared with the wild-type E2), suggesting that the PePHD region is the major, if not the only, PKR-binding sequence in E2. The cellular protein hnRNPK, which has a strong protein-protein interacting domain (8), did not bind to any of the proteins tested, and none of the proteins bound to the GFP-His control (lane 4).

Figure 2

E2 binds to and inhibits PKR in vitro. (A) In vitro–translated, 35S-labeled E2, E2-ΔC, or hnRNPK was incubated with His-tagged proteins coupled to nickel agarose beads as described in (10). Lane 1 represents 10% of the input volume of in vitro translation products used for the binding reactions. Wt, wild type; GFP-His, His tagged green fluorescent protein control. (B) Immunoblot of the His-tagged proteins used in the binding reactions, probed with antibody to His (Qiagen). (C) Protein kinase assay of PKR immunoprecipitated from transfected HEK-293 cells using rabbit polyclonal antibody to PKR (23) and protein A– Sepharose as described in (9). These experiments were repeated three to six times with similar results. (D) Protein kinase assay with bacterially expressed, partially purified, His-tagged PKR (10).

To assess the effect of E2 on PKR activity in vitro, we performed kinase assays with immunopurified PKR from transfected human embryonic kidney 293 (HEK-293) cells. The immunoprecipitates were incubated with purified glutathione S-transferase (GST) or GST-E2 fusion proteins, using histone H2a as a substrate (9) (Fig. 2C). GST-E2, but not GST, inhibited histone phosphorylation by 90 to 98% (lanes 7 to 9), but PKR autophosphorylation was only slightly inhibited. The lack of robust inhibition of PKR autophosphorylation by GST-E2 was probably due to the highly activated state of the transfected human PKR even in the absence of poly(I·C) (Fig. 2C, compare lane 1 to lane 2). We also used recombinant His-tagged PKR purified from Escherichia coli, which exhibits only a low level of activation in the absence of double-stranded RNA (Fig. 2D, compare lane 1 to lane 2), in the kinase assay (10). GST-E2, but not GST, almost completely inhibited PKR autophosphorylation induced by the double-stranded RNA activator (lanes 5 and 6). The presence of GST-E2 (66 kD, similar to PKR) did not result in the appearance of any additional phosphorylated proteins, indicating that GST-E2 was not phosphorylated. These results suggest that E2 may serve as a pseudosubstrate of PKR and inhibit the function of PKR in vitro.

We next investigated whether E2 inhibits cellular PKR activity in vivo, which would lead to enhanced protein synthesis. We cotransfected a reporter plasmid encoding the chloramphenicol acetyl transferase (CAT) gene under control of a cytomegalovirus (CMV) promoter and a plasmid encoding E2L (E2 with the leader peptide) or an empty vector (pcDNA3) into HeLa cells and measured CAT enzyme activity to determine the amount of protein synthesis (11). CAT activity was a factor of 5 higher in the presence of E2L (Fig. 3A). The enhancement of CAT activity correlated with the amount of transfected E2L plasmid (Fig. 3B).

Figure 3

HCV-1 E2 enhances protein synthesis in mammalian cells. (A) HeLa cells were transfected with CAT-expressing plasmids (20 μg) and pcDNA3 or E2L-pcDNA3 (20 μg). (B) CAT assays were performed on lysates from human hepatocellular carcinoma (Huh7) cells that were cotransfected with 15 μg of CAT-pcDNA3 and increasing amounts of E2L-pcDNA3, or GFP (2 μg). (C) Ribonuclease protection analysis (Hybspeed RPA; Ambion) was performed on RNA (50 μg per sample) prepared from cells in (B). Lanes 1 and 7, CAT probe after ribonuclease digestion; lanes 2 to 6, decreasing amounts of T7 RNA polymerase–transcribed CAT RNA used for quantitation of RNA in lanes 9 to 13. Lanes 9 to 13, RNA from cells transfected with increasing amounts of E2L-pcDNA3, 0 to 20 μg. All lanes are from the same autoradiogram. (D) CAT activity of cells transfected with CAT-pcDNA3 and various E2 constructs in pcDNA3. NS5A was from the HCV-1a strain (6). CAT assay experiments were performed three to six times with each sample done in triplicate. Error bars display average and standard deviation of three independent experiments. (E) Structural diagram of the E2 proteins tested in (D) (24). The hatching on the left represents the E2 leader sequence. (F) HeLa cells were transfected with HCV-1 E2L-pcDNA3 (+) or with the empty vector (; pcDNA3). (Top) Cells were metabolically labeled with [32P]orthophosphate, and lysates were immunoprecipitated with polyclonal antibody to PKR (23); (middle) the same as in the top panel except that labeling was done with [35S]methionine; (bottom) immunoblot of unlabeled cell lysates from the same transfection. (G) Metabolic labeling with [32P]orthophosphate of cells transfected with pcDNA3 and wild-type and mutant E2-pcDNA3 constructs. (Top) Lysates were precipitated with antibody to PKR. Parallel unlabeled lysates were resolved on a gel for immunoblot analysis; (middle) PKR; (bottom) actin. PhosphorImage analysis of the 32P-labeled PKR is summarized in a histogram. This experiment was repeated three times with similar results. (H) Colocalization of HCV E2L and PKR in HEK-293 cells by immunofluorescent staining (11).

To establish that this enhanced CAT activity was due to increased translation rather than transcription, we measured CAT RNA levels in the transfected cells. The cells transfected with various amounts of E2L plasmid contained equal amounts of CAT RNA (about 20 pg of RNA per sample, as determined by comparing with lanes 2 through 6 in Fig. 3C), indicating that the E2L plasmid did not enhance transcription. Furthermore, CAT activity was enhanced only when E2 contained the leader sequence (E2L, Fig. 3D), suggesting that E2 needs to be associated with the endoplasmic reticulum (ER) to inhibit PKR. Truncation of the COOH-terminus of E2 (E2L-ΔC), which removes the PePHD, abolished the enhancement of CAT activity. Furthermore, mutation of PePHD to a sequence derived from HCV genotypes 2 and 3 (E2L-clu1) abolished enhancement of CAT activity. Finally, HCV NS5A, which also inhibits PKR activity (3), enhanced CAT activity by a factor of 2.5. These results suggest that E2 can enhance the translation of cellular mRNAs, most likely through its PePHD.

We next determined whether the translational enhancement by E2L was mediated through reduction of PKR activity. We transfected a pcDNA3 plasmid encoding E2L into HEK-293 cells. Expression of PKR was induced by the addition of IFN-α, and the kinase was activated by infection with adenovirus mutant dl331, which does not express the PKR inhibitor VA RNAI (12). The results showed that the amount of phosphate incorporated into PKR was reduced by almost 50% in E2L-transfected cells (Fig. 3F, compare lanes 2 and 4), whereas the amount of PKR protein (as determined by [35S]methionine labeling) was largely unaffected. Because slightly less than 50% of cells were transfected with this plasmid, and 100% were infected with adenovirus dl331, this result suggests that PKR autophosphorylation was almost completely inhibited in E2L-expressing cells. Examination of E2 mutants (Fig. 3G) revealed that E2 (without the leader peptide) reduced PKR activation by 20%, whereas—for reasons that remain to be determined—E2L-ΔC and E2L-clu1 activated PKR activity. Thus, only E2L efficiently inhibits PKR activation. HCV E2 is localized to the perinuclear ER region of the cytoplasm (13). Immunofluorescence studies showed that E2L colocalized with the endogenous PKR in the cells (Fig. 3H), consistent with the requirement for the E2 leader sequence in PKR inactivation. PKR has previously been reported to be associated with ribosomes (14–16), including most likely ER-bound ribosomes.

We next expressed E2 in a GCN2-deficient strain of the yeastSaccharomyces cerevisiae. GCN2 is an eIF2α kinase and a functional homolog of PKR (17). In gcn2(−) yeast, expression of PKR stimulates eIF2α phosphorylation, which in turn inhibits protein synthesis and retards cell growth (17). To study the effects of E2 on PKR in yeast, we used strain RY1-1, which lacksgcn2 but contains two chromosomally integrated copies ofpkr under the control of the galactose-induciblegal1-cyc1 promoter (18). We transformed a DNA construct containing the E2 gene under the control of thegal1 promoter (E2-pYES2) into this yeast strain. Yeast transformed with the vector alone (pYES2) grew well on synthetic dextrose minimal (SD) medium but poorly on synthetic galactose (SGAL) medium (Fig. 4A). This was expected because gal-induced PKR expression would result in eIF2α phosphorylation, leading to inhibition of translation and cell growth (17, 18). In contrast, yeast that express the dominant-negative PKR mutant, PKRΔ6, grew well on both SD and SGAL (Fig. 4A). Cells transformed with E2 grew on both SD and SGAL, indicating that E2 inhibits PKR activity in this yeast model (Fig. 4, A and B). In contrast, yeast transformed with E2-clu1 and E2-ΔC, both of which failed to stimulate translation in mammalian cells (Fig. 3D), did not grow on SGAL, suggesting that the E2 PePHD is responsible for PKR inhibition in yeast. As previously demonstrated (3), yeast transformed with NS5A also grew well on both SD and SGAL media. PKR protein levels were comparable in all transformants (Fig. 4C); thus, the E2-mediated stimulation of cell growth was due to the inhibition of PKR kinase activity rather than the inhibition of PKR synthesis. When E2 was expressed in yeast strain H1816, which encodes wild-type eIF2α but no PKR, or strain H1817, which encodes a nonphosphorylatable mutant of eIF2α (S51A), neither E2 nor the E2 mutants altered the growth of yeast on SD or SGAL (19), indicating that E2 did not alter the growth of yeast cells in the absence of PKR.

Figure 4

HCV-1 E2 inhibits PKR in yeast. pYES 2 plasmids were introduced into yeast strain RY1-1 (25). (Aand B) Yeast were transformed with the indicated plasmids, and transformants were streaked on SD or SGAL plates and grown at 30°C for 3 to 5 days. (C) Expression of PKR from RY1-1 strains transformed with various plasmids and monitored by protein immunoblotting with monoclonal antibody to PKR (26). pYES 2, vector alone. PKRΔ6, a dominant-negative mutant of PKR.

Our results suggest that the E2 protein of HCV genotype 1 binds to and inhibits PKR in vitro and does so in mammalian cells and yeast because of sequence homology to the PKR and eIF2α phosphorylation sites. These effects correlate with the relative resistance of HCV genotype 1 to IFN. Thus, HCV may have evolved a two-level attack, namely NS5A and E2, on PKR to interfere with IFN action. Although the NS5A-PKR interaction may contribute to development of resistance during IFN therapy (20), the E2-PKR interaction may account for the intrinsic IFN resistance of HCV genotypes 1a and 1b. The combined effects of NS5A and E2 may explain why most HCV infections become persistent. Another potential outcome of PKR inhibition is the promotion of cell growth, which may contribute to HCV-associated hepatocellular carcinoma.

The PKR-inhibitory activity of HCV E2 protein in mammalian cells is enhanced by the presence of the leader sequence, suggesting that E2 and PKR likely interact with each other at the ER. An ER-resident eIF2α kinase (PERK) has recently been described, which shares homology with PKR and recognizes the same substrate, eIF2α (21). This may provide another target for E2.

  • * To whom correspondence should be addressed. E-mail: michlai{at}


View Abstract

Navigate This Article