SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References
  8. Supporting Information

The hepatitis C virus (HCV) NS3-4A protease is not only an essential component of the viral replication complex and a prime target for antiviral intervention but also a key player in the persistence and pathogenesis of HCV. It cleaves and thereby inactivates two crucial adaptor proteins in viral RNA sensing and innate immunity, mitochondrial antiviral signaling protein (MAVS) and TRIF, a phosphatase involved in growth factor signaling, T-cell protein tyrosine phosphatase (TC-PTP), and the E3 ubiquitin ligase component UV-damaged DNA-binding protein 1 (DDB1). Here we explored quantitative proteomics to identify novel cellular substrates of the NS3-4A protease. Cell lines inducibly expressing the NS3-4A protease were analyzed by stable isotopic labeling using amino acids in cell culture (SILAC) coupled with protein separation and mass spectrometry. This approach identified the membrane-associated peroxidase GPx8 as a bona fide cellular substrate of the HCV NS3-4A protease. Cleavage by NS3-4A occurs at Cys 11, removing the cytosolic tip of GPx8, and was observed in different experimental systems as well as in liver biopsies from patients with chronic HCV. Overexpression and RNA silencing studies revealed that GPx8 is involved in viral particle production but not in HCV entry or RNA replication. Conclusion: We provide proof-of-concept for the use of quantitative proteomics to identify cellular substrates of a viral protease and describe GPx8 as a novel proviral host factor targeted by the HCV NS3-4A protease. (Hepatology 2014;59:423–433)

Abbreviations
DDB1

UV-damaged DNA-binding protein 1

ER

endoplasmic reticulum

Ero1α

ER oxidoreductin-1α

GFP

green fluorescent protein

GPx8

glutathione peroxidase 8

GST

glutathione S-transferase

HCV

hepatitis C virus

HCVcc

cell culture-derived HCV

HCVpp

HCV pseudoparticle

IFN-α

interferon-α

LC-MS/MS

liquid chromatography coupled with tandem mass spectrometry

mAb

monoclonal antibody

MAVS

mitochondrial antiviral signaling protein

MS

mass spectrometry

NS3-4A

nonstructural protein 3-4A

PDI

protein disulfide isomerase

POPC

1-palmitoyl-2-oleoyl-3-sn-glycero-3-phosphocholine

SDS-PAGE

sodium dodecyl sulfate-polyacrylamide gel electrophoresis

SILAC

stable isotopic labeling using amino acids in cell culture

TC-PTP

T-cell protein tyrosine phosphatase

TCID50

50% tissue culture infective dose.

Hepatitis C virus (HCV) infection is a leading cause of chronic hepatitis, liver cirrhosis, and hepatocellular carcinoma worldwide, with a disease burden predicted to increase further for the next 10 years.[1] HCV NS3-4A is a multifunctional complex composed of nonstructural protein 3 (NS3) and its cofactor NS4A. It harbors serine protease as well as NTPase/RNA helicase activities and is essential for viral polyprotein processing, RNA replication, and assembly.[2, 3] Given its central role in the viral life cycle, NS3-4A has been actively pursued as an antiviral target, and a first generation of NS3-4A protease inhibitors has been approved for the treatment of chronic HCV.[3]

The NS3-4A protease has been found to also target selected cellular proteins. Indeed, it cleaves and thereby inactivates Cardif[4] (also known as mitochondrial antiviral signaling protein [MAVS], IPS-1, and VISA) as well as TRIF,[5] two key adaptor molecules in the RIG-I and TLR3 viral RNA-sensing pathways, respectively (see Ref. [3] for review and abbreviations). In addition, NS3-4A has been reported to cleave T-cell protein tyrosine phosphatase (TC-PTP), thereby modulating growth factor signaling,[6] and UV-damaged DNA-binding protein 1 (DDB1), a core component of the E3 ubiquitin ligase complex.[7] Hence, NS3-4A is not only an essential component of the viral replication complex and a prime target for antiviral intervention but also a key player in the persistence and pathogenesis of HCV.

Here we aimed to identify novel cellular substrates of the HCV NS3-4A protease. Searching for such cellular targets in an unbiased fashion represents a challenge. Bioinformatic approaches are likely to fail, as a vast number of cellular proteins display the NS3-4A consensus cleavage sequence D/E-X-X-X-X-C/T | S/A-X-X-X (P6-P1 | P1′-P4′). Moreover, the cellular substrates identified so far have less canonical cleavage sites (reviewed in Ref. [3]). In addition, our previous work has demonstrated that the processing and structural organization of NS3-4A are intrinsically linked to intracellular membranes and that proper positioning of the protease active site with respect to the membrane may represent a determinant of substrate selectivity.[8]

Based on these premises, we explored large-scale proteomics technologies to identify novel cellular substrates of the HCV NS3-4A protease. An approach was needed that permits accurate relative protein quantitation and at the same time could be coupled to a separation technique able to discriminate full-length proteins from their cleavage products. Hence, we employed stable isotopic labeling using amino acids in cell culture (SILAC), which has demonstrated high sensitivity, broad proteome coverage, and accuracy.[9] SILAC was coupled with molecular weight-based protein separation, mass spectrometry (MS), and a dedicated data analysis approach to detect and quantify comprehensively the species as well as the migration patterns of proteins from cells expressing or not the NS3-4A protease.

This quantitative proteomics approach followed by experimental validation identified the membrane-associated peroxidase GPx8 as a cellular substrate of the NS3-4A protease. GPx8 cleavage by NS3-4A occurs at Cys 11, removing the cytosolic tip of GPx8, and was observed in different experimental systems as well as in liver biopsies from patients with chronic HCV. Furthermore, overexpression and RNA silencing studies revealed that GPx8 is involved in viral particle production but not in HCV entry or RNA replication. Thus, we provide a proof-of-concept for the use of advanced proteomics for the identification of cellular substrates of a viral protease and describe GPx8 as a novel proviral host factor targeted by the HCV NS3-4A protease.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References
  8. Supporting Information
Reagents

Monoclonal antibodies (mAbs) C7-50 and 1B6 against HCV core and NS3, respectively, have been described.[10, 11] MAb Adri-1 against MAVS was from Adipogen (Epalinges, Switzerland). MAb ANTI-FLAG M2 and mAb AC-15 against β-actin were from Sigma-Aldrich. MAbs JS-81 and M-L13 against CD81 and CD9, respectively, were from BD Biosciences. Telaprevir was kindly provided by Johan Neyts (Rega Institute for Medical Research, Leuven, Belgium).

The U-2 OS human osteosarcoma-derived cell lines UNS3-4A-24 and UHCVcon-57.3, allowing tetracycline-regulated expression of the NS3-4A complex derived from the prototype HCV H77 (genotype 1a) clone[12] and of the entire polyprotein derived from the HCV H77 consensus clone,[13] respectively, have been described.[11, 14] Huh-7.5 cells were kindly provided by Charles M. Rice (Rockefeller University, New York, NY).[15]

SILAC, Protein Separation, and MS

UNS3-4A-24 cells were heavy isotope labeled by culture in modified RPMI 1640 medium in which 13C6-L-lysine and 13C615N4-L-arginine (Cambridge Isotope Laboratories) were present at 100 mg/L, whereas proline was supplemented at 180 mg/L. All other amino acids were present at standard concentrations. A parallel culture designated “light” was maintained in the same medium containing standard unlabeled amino acids. Dialyzed fetal bovine serum at a concentration of 10%, 4 mM L-glutamine, and 1 μg/mL tetracycline were added to all media.

Cell lines were cultured for six passages in the above medium to achieve complete labeling (heavy labeling >98% and arginine-to-proline conversion <5%). NS3-4A protease expression was induced by tetracycline withdrawal during 48 hours, with or without stimulation with interferon-α (IFN-α) 1000 IU/mL (Roche) for 24 hours before harvest. Cell extracts were mixed at a quantitative ratio of 1:1, and a total of 20 μg of protein was separated by gradient sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). After Coomassie staining, each lane was cut into either 56 (first screen) or 64 slices (second screen), followed by in-gel digestion with trypsin, as described.[16] The supernatants from each digestion reaction were concentrated and analyzed on a hybrid linear trap LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher) interfaced by way of a nanospray source to a Dionex RSLC 3000 nanoHPLC system (see Supporting Materials and Methods for details).

MS Data Analysis

MS data were analyzed and quantified with standard parameters using MaxQuant v. 1.0.13.13[17] and the IPI database (human subset) v. 3.52 (see Supporting Materials and Methods for details). Results from MaxQuant were filtered and processed further with custom-built software tools. The first was a Perl script to identify proteins present in at least two distinct gel slices with different SILAC ratios, as described in the Results section. Next, the evidence counts and normalized SILAC ratios for these selected proteins were visualized as a function of the gel slice with an R script (Supporting Fig. 1).

Plasmids and Recombinant Viruses

As described in more detail in the Supporting Materials and Methods, a GPx8 complementary DNA (cDNA) was cloned from U-2 OS cells and FLAG-tagged expression constructs were prepared in pcDNA3.1(+) (Invitrogen). The cDNA encoding GPx8 amino acids 38-209 was cloned into pGEX-6P-1 (GE Healthcare) for production of a glutathione S-transferase (GST) fusion protein. The cDNAs of wild-type and mutant GPx8 were cloned into pLPCX (ClonTech) for production of a recombinant retroviruses.

Plasmid pCMVNS3-4A and its catalytically inactive counterpart pCMVNS3S139A-4A have been described.[8, 11] pFKi389LucNS3-3′_dg_JFH[18] and pFK-JFH1J6C-846_dg[19] were kindly provided by Ralf Bartenschlager (University of Heidelberg, Germany).

In Vitro Transcription, Electroporation, Transient Replication, and Infection Assays

In vitro transcription of subgenomic replicon and full-length HCV RNA as well as electroporation were performed as described (Ref. [20] and references therein). RNA replication assay using a JFH1 subgenomic replicon harboring the firefly luciferase was performed as described.[18, 21] Jc1 cell culture-derived HCV (HCVcc) was produced as described.[19] The 50% tissue culture infective dose (TCID50) was determined as described.[22]

Retroviral Transduction

Recombinant retroviral particles were prepared in 293T cells by cotransfection of pLPCX-derived GPx8 constructs or pLPCX-GFP (kindly provided by Angela Ciuffi, University of Lausanne, Switzerland) and packaging constructs pMLV-NB and pMD.G (kindly provided by Didier Trono, Ecole Polytechnique Fédérale de Lausanne, Switzerland), as described.[23]

RNA Silencing

GPx8 small interfering RNAs (siRNAs) #1 (s54629) and #2 (s54630) were purchased from Ambion (Life Technologies). A nontargeted siRNA served as control (Silencer Select Negative Control #1, Ambion). RNAiMAX was used for siRNA transfection (Invitrogen).

HCV Pseudoparticle Production and Infection Assays

HCV pseudoparticles (HCVpp) were produced as described.[24]

Recombinant Protein Expression and Antibody Production

Polyclonal antibody IN103 against GPx8 was produced by immunization of a rabbit with GPx8[38-209] expressed as GST fusion protein in Escherichia coli. This antibody is now available from Adipogen (AG-25B-0028).

Liver Biopsies

Liver biopsy specimens from patients with chronic HCV and controls were obtained in the context of routine diagnostic workup and snap-frozen for research purposes if there was sufficient material for histopathological examination and the patient's written informed consent had been provided in accordance with the Ethics Committee of Basel.

Molecular Modeling

Sequences similar to that of the GPx8 N-terminal amino acid segment 1-38 were searched for in the Protein Data Bank by using the SSEARCH program in the IBCP website facilities (http://npsa-pbil.ibcp.fr/)[25] and structural models were constructed using Swiss PDB-viewer (http://www.expasy.org/spdbv/).[26]

Statistical Analyses

Group differences of continuous variables were assessed by means of Wilcoxon-Mann-Whitney U tests. P < 0.05 were considered significant.

Results

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References
  8. Supporting Information
Identification of GPx8 as a Novel Cellular Substrate of the HCV NS3-4A Protease

The UNS3-4A-24 cell line, allowing the tightly regulated expression of a functional HCV NS3-4A protease, was subjected to SILAC, followed by protein separation and MS, as illustrated in Fig. 1. Two screens were performed, a first one comparing cells expressing or not the NS3-4A complex and a second one comparing cells expressing or not the NS3-4A complex with the additional stimulation by IFN-α. Immunoblots of the cell lysates that were subjected to SDS-PAGE and MS demonstrate the tightly regulated expression of NS3-4A, cleavage of MAVS as a well-characterized cellular substrate of the NS3-4A protease, and upregulation of Stat1 as a control for the IFN-α stimulation (Supporting Fig. 2).

image

Figure 1. Quantitative proteomics using stable isotopic labeling with amino acids in cell culture (SILAC). Two populations of UNS3-4A-24 cells were fully labeled with either normal (light) or heavy isotopes. Expression of the HCV NS3-4A protease was induced in one of the two cultures by tetracycline withdrawal (-tet). After lysis and protein quantitation, the two samples were mixed at a 1:1 ratio, followed by gradient SDS-PAGE and cutting of the gel into multiple slices. Each gel slice was then submitted to trypsin digestion and LC-MS/MS. This results in identification and relative quantitation of proteins in each slice. The profiles along the gel are then reconstructed for each protein together with their abundance and heavy-over-light (H/L) ratios. Proteins with lower apparent molecular weight bands induced in protease-expressing samples were considered candidate NS3-4A protease substrates. m/z, mass-to-charge ratio.

Download figure to PowerPoint

The SILAC workflow allowed the identification and relative quantification of up to 100 protein species per gel slice. Peptides from a total of 3,642 and 5,053 proteins were identified in the first and the second screen, respectively, demonstrating the high sensitivity and broad proteome coverage of SILAC-MS. To identify protein modifications induced by NS3-4A, we examined the protein quantification data obtained in each gel slice separately and compared the values for each protein across all gel slices. Indeed, the presence of the same protein in two distinct slices with opposite quantitative ratios indicates an NS3-4A-dependent shift in gel mobility, hence, possible substrates and products of the NS3-4A protease. As an important aid to interpretation and filtering of candidates we developed software tools in the Perl and R programming languages to visualize the complex patterns of bands and SILAC ratios observed (Supporting Fig. 1). These are available from the download page of http://www.unil.ch/paf.

The following criteria were applied to select for candidate substrates of the NS3-4A protease: (1) the protein had to be represented by at least two peptides per SDS-PAGE slice and (2) the protein had to be more abundant in a lower molecular weight slice in the sample derived from cells expressing NS3-4A, suggesting a substrate-product relationship. Based on these criteria, GPx8 was identified as top hit in both screens and further investigated in this study. Of note, MAVS as a well-characterized cellular substrate of the NS3-4A protease was identified in our screen while DDB1 did not show an obvious change upon protease expression and TRIF as well as TC-PTP were not detected (Supporting Table 2).

In the first screen, GPx8 was detected in the absence of NS3-4A only in SDS-PAGE slice #44 (Supporting Table 2, Supporting Fig. 1), harboring proteins of ∼24 kDa and hence the full-length form of GPx8. In cells cultured in the absence of tetracycline (i.e., expressing NS3-4A), GPx8 was detected also in slices #45 and 46, harboring proteins of ∼22 kDa. In the second screen, the two forms of GPx8 were identified in slices #42 versus 43 and 44 (Supporting Table 2).

GPx8 (glutathione peroxidase 8) is a 209-amino acid, 24-kDa protein with an N-terminal cytosolic tip, a predicted transmembrane segment, and a catalytic domain located in the endoplasmic reticulum (ER) lumen (Fig. 2A).[27] To experimentally validate GPx8 as candidate NS3-4A substrate, a GPx8 cDNA was cloned from U-2 OS cells, and N- or C-terminally FLAG-tagged versions of GPx8 were coexpressed with NS3-4A or the catalytically inactive control NS3S139A-4A. As shown in Fig. 2B (left), GPx8 is cleaved by NS3-4A but not by the catalytically inactive control or in the presence of the NS3-4A protease inhibitor telaprevir. Based on the migration pattern, cleavage was expected to occur within ∼2 kDa from the N terminus of the protein. Indeed, the N-terminally tagged construct was no longer detected after coexpression with functional NS3-4A (data not shown), whereas a second, slightly faster migrating band was found after coexpression of the C-terminally tagged construct with functional NS3-4A (Fig. 2B). As trans-cleavage by NS3-4A occurs after Cys residues, we replaced the most likely target residue in GPx8 position 11 by Ala (C11A). As shown in Fig. 2B (middle), this construct was no longer cleaved by NS3-4A. Finally, an N-terminally truncated GPx8 construct, in which the first 11 amino acid residues were deleted (S12), comigrated with the cleaved product derived from full-length GPx8 (Fig. 2B, right). Taken together, these results demonstrate that GPx8 is a bona fide substrate of the HCV NS3-4A protease and that cleavage occurs at Cys 11.

image

Figure 2. GPx8 is a cellular substrate of the HCV NS3-4A protease. (A) Amino acid sequence of human GPx8 (GenBank accession number AAH29424). The NS3-4A cleavage site and the predicted transmembrane segment are highlighted by an arrowhead and a box, respectively. (B) U-2 OS cells were cotransfected with pCMVGPx8-FLAG, pCMVGPx8C11A-FLAG, or pCMVGPx8S12-FLAG and pcDNA3.1(+), pCMVNS3-4A, or pCMVNS3S139A-4A, as indicated. Treatment with the NS3-4A protease inhibitor telaprevir (TVR) served as additional control. Cell lysates were separated by 15% SDS-PAGE, followed by immunoblot using mAbs ANTI-FLAG M2, 1B6, and AC-15 against the FLAG tag, HCV NS3, and β-actin, respectively. Molecular weight standards are indicated on the left. (C) Molecular models of NS3-4A and GPx8. The model of membrane-associated NS3-4A has been described.[8] The NS3 serine protease domain is cyan, with side chain atoms of the catalytic triad (His 57, Asp 81, and Ser 139) highlighted as purple spheres. The NS3 helicase domain is gray and the N-terminal transmembrane (amino acids 1-20) and central (amino acids 21-32) segments of NS4A are orange and light orange, respectively. Homology models of the GPx8 cytosolic tip and transmembrane segment are green, while the crystal structure of the ER luminal domain is light green (PDB entry 3KIJ).[27] Side chain atoms of Cys 11, representing the site of NS3-4A protease-mediated cleavage, are represented as red spheres. The membrane is represented as a simulated model of a 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) bilayer (http://moose.bio.ucalgary.ca). Polar heads and hydrophobic tails of phospholipids (stick structures) are light yellow and gray, respectively. The figure was generated on the basis of the structure coordinates by using the Visual Molecular Dynamics program[34] and rendered with the POV-Ray software package.

Download figure to PowerPoint

While the crystal structure of the ER luminal domain of human GPx8 has been solved (GPx8[38-209]; PDB entry 3KIJ),[27] no structure is available for the N-terminal cytosolic tip and the predicted transmembrane segment of GPx8. However, GPx8 segment 1-17 exhibits 47% identity and 71% similarity with segment 155-171 of filamin A immunoglobulin-like repeat 10 (PDB entry 3RGH)[28] (Supporting Fig. 3). Moreover, a transmembrane α-helix was predicted by all tested methods for GPx8 segment 18-37, which exhibited 26% identity and 89% similarity to the transmembrane α-helix 20-41 of FXYD protein of the sodium-potassium ATPase (PDB entry 2ZXE, chain G)[29] (Supporting Fig. 3). These observations allowed us to construct a three-dimensional homology model for the N-terminal region 1-37 of GPx8, which was connected to the luminal domain and putatively positioned in a 1-palmitoyl-2-oleoyl-3-sn-glycero-3-phosphocholine (POPC) bilayer, as shown in Fig. 2C. This positioning was based on the assumption that the basic Lys residues at both helix ends interact with the negatively charged polar heads of phospholipids. As expected, this model shows that the α-helix has the right size to form a classical transmembrane segment. In addition, the position of GPx8 Cys 11 fits very well with the topology of the NS3-4A protease active site in the model of membrane-associated, trans-cleavage competent NS3-4A proposed by Brass et al.[8]

Endogenous GPx8 Is Cleaved in Models of HCV Infection and in Liver Biopsies From Patients With Chronic HCV

To investigate whether endogenous GPx8 is cleaved by HCV NS3-4A we prepared a polyclonal antibody against an E. coli-expressed GST fusion protein comprising the ER luminal domain of GPx8 (GPx8[38-209]). As shown in Fig. 3A, cleavage of endogenous GPx8 was observed in UNS3-4A-24 and UHCVcon-57.3 cells upon expression of the NS3-4A complex alone or in the context of the entire HCV polyprotein, respectively. As endogenous GPx8 expression was lower in Huh-7.5 cells as compared to U-2 OS cells and cleavage by the NS3-4A protease detectable but more difficult to illustrate (data not shown), we generated by retroviral transduction a Huh-7.5 cell line stably overexpressing GPx8, designated Huh-7.5-GPx8. As shown in the right panel of Fig. 3A, cleavage of GPx8 was observed in Huh-7.5-GPx8 cells upon transfection of a subgenomic HCV replicon or infection with HCVcc. Finally and most important, cleavage of GPx8 was also observed in liver biopsies from patients with chronic HCV, whereas only the full-length form of GPx8 was found in controls (Fig. 3B). Taken together, these data demonstrate that GPx8 is cleaved not only in heterologous overexpression systems but also in models of HCV infection as well as in the liver of patients with chronic HCV.

image

Figure 3. Cleavage of GPx8 in models of HCV infection and in liver biopsies from patients with chronic HCV. (A) UNS3-4A-24 and UHCVcon-57.3 cells, inducibly expressing the NS3-4A complex alone or in the context of the entire HCV polyprotein, respectively, were cultured for 48 hours in the presence or absence of tetracycline, followed by 15% SDS-PAGE and immunoblot using polyclonal antibody IN103 against GPx8. Arrowheads denote the position of full-length (FL) and cleaved (Cl.) GPx8. Similarly, Huh-7.5 cells stably overexpressing GPx8 (Huh-7.5-GPx8) were electroporated with a subgenomic replicon (sg2a, JFH1 strain) or infected with HCVcc (Jc1). Molecular weight standards are indicated on the left. (B) Lysates from liver biopsies of patients with chronic hepatitis C (CHC; lanes 3-5; B392, B402, and B419) as well as controls (lanes 6-7; B451 from a patient with alcoholic liver disease, B457 from a patient with chronic hepatitis B) were separated by 15% SDS-PAGE, followed by immunoblot using polyclonal antibody IN103 against GPx8. A total of 25 μg protein per lane was loaded for the lysates from UNS3-4A-24 cells cultured in the presence or absence of tetracycline (tet) while 100 μg protein per lane was loaded for the liver biopsy specimens. A selection from a larger panel of liver biopsies including eight samples from patients with CHC (of which GPx8 cleavage was detected in three) and six controls (of which GPx8 cleavage was detected in none) is shown. Arrowheads denote the position of full-length (FL) and cleaved (Cl.) GPx8.

Download figure to PowerPoint

GPx8 Facilitates HCV Particle Production

Overexpression and siRNA-mediated silencing were employed to probe for a role of GPx8 in the HCV life cycle. The HCVcc system allows investigation of the entire life cycle. Huh-7.5 cells were transduced with retroviruses expressing GPx8 or the green fluorescent protein (GFP) as a control, followed by infection with Jc1 HCVcc. Under these conditions, 90%-100% of the cells overexpressed GPx8 or GFP (data not shown). As shown in Fig. 4A, cells overexpressing GPx8 produced more infectious virus as compared to cells expressing GFP. In two independent experiments performed in triplicate each, TCID50 yields were 1.8 ± 0.7-fold higher upon GPx8 overexpression. Analogous results were obtained in Huh-7.5-GPx8 as compared to parental Huh-7.5 cells (2.3 ± 0.3-fold increase in TCID50 yields, n = 6; data not shown), confirming that overexpression of GPx8 slightly but very reproducibly enhances infectious HCV particle production.

image

Figure 4. GPx8 facilitates HCV particle production. (A) Huh-7.5 cells were transiently transduced with retroviruses (multiplicity of infection [MOI] = 10) expressing either GPx8 or green fluorescent protein (GFP), followed 2 days later by Jc1 HCVcc infection (MOI = 0.5). Extra- and intracellular infectivity were determined 3 days post-HCVcc infection by TCID50 titration. Total infectivity represents the sum of extra- and intracellular infectivity. Standard deviations of a representative experiment performed in triplicate are shown. *P for total infectivity in GFP versus GPx8 overexpressing cells <0.1. (B) Huh-7.5 cells were transfected with GPx8 siRNAs (GPx8#1, GPx8#2) or nontargeted control siRNA (nt), followed 48 hours later by Jc1 HCVcc infection (MOI = 0.5). TCID50 yields were determined as described in panel A as well as the Materials and Methods section. Standard deviations of a representative experiment performed in triplicate are shown. **P for total infectivity in cells transfected with nt versus GPx8 siRNAs <0.05. (C) Huh-7.5 cells overexpressing GPx8 or GFP following retroviral transduction (see panel A) were assayed for HCV RNA replication using a JFH1 subgenomic replicon harboring a firefly luciferase reporter gene, as described in the Materials and Methods section. Luciferase activity was measured at 4, 15, 24, and 48 hours postelectroporation, and relative light units (RLU) were normalized to the corresponding 4-hour value. (D) Huh-7.5 cells transfected with siRNAs as described in panel B as well as the Materials and Methods section were electroporated 72 hours later with subgenomic HCV replicon RNA and luciferase assay performed at 4, 15, 24, and 48 hours postelectroporation. The results were normalized to the corresponding 4-hour value. (E) To assess HCV entry, Huh-7.5 cells as well as Huh-7.5-GPx8 cells were transfected with siRNAs (see panel B) 24 hours prior to inoculation with HCVpp. Luciferase activity was measured 72 hours later. As controls for specificity, Huh-7.5 cells were incubated 1 hour prior to HCVpp infection either with monoclonal antibody (mAb) JS-81, blocking the HCV entry factor CD81, or mAb M-L13 against CD9 as a nonrelevant control.

Download figure to PowerPoint

Both GPx8 siRNAs, #1 and #2, efficiently and specifically silenced GPx8 expression (Supporting Fig. 4). Huh-7.5 cells were transfected with GPx8 siRNAs #1 or #2 or a nontargeting control, followed by infection with Jc1 HCVcc. As shown in Fig. 4B, cells produced less infectious virus following GPx8 silencing. In two independent experiments performed in triplicate each, TCID50 yields were reduced by 62.1 ± 21.2% upon GPx8 silencing. Of note, intracellular HCV RNA levels were not affected significantly in cells silenced for GPx8 (data not shown), suggesting that GPx8 does not affect HCV entry and RNA replication.

To confirm the latter notion, Huh-7.5 cells overexpressing GPx8 or treated with GPx8 siRNAs were transfected with subgenomic HCV replicons or infected with HCVpp, respectively. As shown in Fig. 4C, the efficiency of HCV RNA replication was similar in Huh-7.5 cells overexpressing GPx8 as compared to cells expressing GFP as a control. In addition, there was no appreciable effect of GPx8 silencing on HCV RNA replication (Fig. 4D). Moreover, no consistent difference in the efficiency of HCVpp entry into cells overexpressing GPx8 as compared to parental Huh-7.5 cells or cells treated with GPx8 siRNAs as compared to the nonrelevant control siRNA was observed (Fig. 4E).

To gain further insight into the role of GPx8 in the HCV life cycle, the uncleavable mutant (GPx8C11A), the N-terminally truncated form corresponding to cleaved GPx8 (GPx8S12) or a mutant harboring a substitution of predicted active site residue Cys 79[27] (GPx8C79T) were overexpressed in Huh-7.5 cells, followed by HCVcc infection and TCID50 determination. As shown in Supporting Fig. 5 and in line with the previous report by Nguyen et al.,[27] GPx8 displayed a subcellular distribution typical of the ER, with partial colocalization with NS3 in HCV-infected cells. There was no appreciable difference in the subcellular localization of the different GPx8 mutants. As shown in Fig. 5, cells overexpressing GPx8, GPx8C11A, and GPx8S12 produced significantly more infectious virus as compared to cells expressing GFP, while GPx8C79T did not increase TCID50 yields, indicating that the enzymatic activity of GPx8 is required to enhance HCV particle production.

image

Figure 5. The catalytic activity of GPx8 is required to enhance HCV particle production. (A) Huh-7.5 cells were transiently transduced with retroviruses expressing either GFP or wild-type (GPx8), uncleavable (GPx8C11A), N-terminally truncated (cleaved form; GPx8S12), or catalytically inactive GPx8 (GPx8C79T), followed by Jc1 HCVcc infection and TCID50 determination, as described in the legend to Fig. 4 as well as the Material and Methods section. (B) Cell lysates from the experiment shown in panel A were analyzed by 12% SDS-PAGE and immunoblot using mAbs C7-50 and AC-15 against HCV core and β-actin, respectively, as well as polyclonal antibody IN103 against GPx8. Arrowheads denote the position of full-length (FL) and cleaved (Cl.) GPx8. Standard deviations of a representative experiment performed in triplicate are shown. **P for total infectivity in cells overexpressing GFP versus GPx8, GPx8C11A and GPx8S12 <0.05. There was no significant difference in cells overexpressing GFP versus GPx8C79T (P = 0.28).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References
  8. Supporting Information

SILAC coupled with molecular weight-based protein separation and MS identified the membrane-associated peroxidase GPx8 as a novel cellular substrate of the NS3-4A protease. Cleavage by NS3-4A occurs at Cys 11, removing the cytosolic tip of GPx8. It was observed in models of HCV infection involving different viral genotypes as well as in liver biopsies from patients with chronic HCV. Furthermore, overexpression and RNA silencing studies revealed that GPx8 is involved in viral particle production but not in HCV entry or RNA replication.

A nonhepatic cell line allowing the tightly regulated expression of the NS3-4A complex was used in this study. To our knowledge, a similarly robust and well-controlled cellular expression system for NS3-4A is currently not available in a hepatocyte-derived cell line. Alternative approaches, such as the use of naïve versus replicon-harboring or HCVcc-infected Huh-7 or Huh-7.5 cells appeared less suitable for our purpose, as a vast number of proteins are likely to be up- or down-regulated in these settings, rendering the identification of bona fide NS3-4A protease substrates more difficult. Moreover, transfection or retroviral transduction of Huh-7 or other liver-derived cell lines may introduce clonal artifacts, which may complicate analyses. Based on these considerations, we have opted for the use of the well-characterized and tightly regulated U-2 OS-derived cell line UNS3-4A-24 for screening and to include Huh-7.5 cells transfected with subgenomic HCV replicons or infected with HCVcc as well as liver biopsies from patients with chronic HCV in the subsequent validation.

A crucial parameter of our “slice-SILAC” approach was the resolution of molecular weight-based separation, which is determined by the number of SDS-PAGE slices taken. Indeed, the high number of gel slices taken in this study allowed for the detection of molecular weight shifts as small as 2 kDa, as was the case for GPx8, which would have been missed with a less fine fractionation.

Similar strategies have been described recently to identify caspase substrates[30] or for the analysis of protein isoforms and their turnover.[31] The coupling of SILAC with SDS-PAGE separation applied in this study is in fact not specific for proteolytic cleavage events but can detect any change in gel electrophoretic mobility. Apart from proteolysis, such shifts can also result from changes in posttranslational modifications (e.g., ubiquitination or glycosylation) or a switch between splice variants of the same protein, possibly explaining why some of the candidates identified in our study, such as Rab34, cytochrome b5, and Rab13, could not be confirmed as proteolytic substrates of the NS3-4A (K.M., J.G., H.T.L.T., M.Q., and D.M., unpublished data).

As the other cellular substrates of the NS3-4A protease reported thus far, i.e., MAVS, TRIF, TC-PTP, and DDB1 (see introduction and Ref. [3]), GPx8 displays a noncanonical NS3-4A trans-cleavage site. Indeed, GPx8 carries an Ala in the P6 position which in the HCV polyprotein is occupied by an acidic residue (Asp or Glu). We believe that substrate specificity may also be determined by the topology of the NS3-4A protease active site, which is defined by NS3 N-terminal amphipathic helix α0, the N-terminal transmembrane segment of NS4A, and NS3 loop 38-40, resulting in tripod-like, strict positioning of the protease active site on the membrane.[8] Indeed, as illustrated in Fig. 2C, the GPx8 cleavage site fits perfectly with our previous prediction.[8] These observations underpin the use of a cellular system to search for novel substrates of the NS3-4A protease.

GPx8 belongs to the glutathione peroxidase family but it has recently been shown to have protein disulfide isomerase (PDI) rather than glutathione peroxidase activity.[27] It interacts with the sulfhydryl oxidase Ero1α (ER oxidoreductin-1α) in the ER lumen and facilitates the use of peroxide produced by Ero1α during disulfide bond formation.[27]

The results obtained with GPx8C79T, harboring a substitution of predicted active site residue Cys 79,[27] indicate that the catalytic activity of GPx8 is required to facilitate HCV particle production. However, cleavage of GPx8 does not appear to be required, at least in Huh-7.5 cells in vitro, as uncleavable mutant GPx8C11A and N-terminally truncated mutant GPx8S12 also enhanced HCV particle production in this experimental setting. Therefore, the role of GPx8 cleavage in the HCV viral life cycle remains currently unknown. Clearly, this cleavage may be required in natural HCV infection in the liver in vivo but not in Huh-7.5 cells in vitro. Future studies will be aimed at determining whether GPx8 acts through a direct mechanism, e.g., by facilitating HCV envelope glycoprotein disulfide bond formation and folding in the ER lumen, or through indirect mechanism(s), e.g., by limiting HCV-induced ER stress and the unfolded protein response. In addition, cell-based methods to assess the function of full-length versus the N-terminally truncated form of GPx8 will have to be developed to further investigate the role of GPx8 and its cleavage in the viral life cycle. Such studies may also provide insights into the role of the cytoplasmic tip and the transmembrane segment which are unique to GPx8 as opposed to other GPx family members.

In conclusion, using quantitative proteomics we identified the membrane-associated peroxidase GPx8 as a cellular substrate of the HCV NS3-4A protease. Identification of such novel targets should enhance our understanding of the life cycle as well as pathogenesis of HCV and may reveal new angles for therapeutic intervention in the future. In addition, the methodology described in this study may be applied to the discovery of host targets of other viral proteases, e.g., those of related Flaviviridae family members, the hepatitis A[32, 33] and E viruses, as well as human immunodeficiency virus and other medically important viruses.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References
  8. Supporting Information

The authors thank Jachen Barblan, Rosa Castillo, Audrey Kennel, and Alexandra Potts for expert technical assistance, Lloyd Ruddock for helpful discussions and critical reading of the article, Pantxika Bellecave for contributions in the early phase of this project, as well as Ralf Bartenschlager, Angela Ciuffi, Johan Neyts, Charles M. Rice, and Didier Trono for reagents.

References

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
hep26671-sup-0001-suppinfo01.docx163KSupporting Information
hep26671-sup-0002-suppfig1.tif9944KSupporting Figure 1. Virtual electropherogram with quantitative values produced by “slice-SILAC”. A representative subset of the data from the screen #1 (without IFN-α stimulation) is shown, including GPx8 (Suppl. Table 2). Only proteins found in at least two gel slices with diverging heavy-over-light (H/L) ratios were prefiltered and are represented here. Only the region below a molecular weight of 50 kDa is shown for ease of representation. Bubble diameters are proportional to the number of quantified peptide matches, while the gradient color represents the H/L ratio, as indicated in the legend. GPx8 is highlighted in yellow. A representative gene name for every protein group is given at the bottom (standard abbreviations as used, e.g., in the Swiss-Prot database). Of note, histone proteins H3F3A and HIST2H2AA3 showed shifts to lower molecular weight upon NS3-4A expression. However, given their broad distribution and known high degree of modification these were not further investigated. Rab13 appeared as a potential candidate (although at less significant H/L ratios) but was not found to be a NS3-4A protease substrate in the subsequent experimental validation process.
hep26671-sup-0003-suppfig2.tif2463KSupporting Figure 2. Cell lysates subjected to protein separation and mass spectrometry. A portion of the lysates from the heavy (H) or light (L) isotope-labeled UNS3-4A-24 cells subjected to protein separation and mass spectrometry in the present study was analyzed by immunoblot using monoclonal antibody (mAb) 1B6 against NS3 (Wölk B et al. J Virol 2000;74:2293-2304), mAb Adri-1 against MAVS (Adipogen, Epalinges, Switzerland), polyclonal antibody E23 against Stat1 p84/p91 (Santa Cruz, Santa Cruz, CA) or mAb AC-15 against β-actin (Sigma-Aldrich, St. Louis, MO) (lanes 3-6). Lysates from naïve Huh-7.5 cells and Huh-7.5 cells harboring a subgenomic HCV Con1 (sg1b, genotype 1b) replicon were included as reference (lanes 1 and 2). A total of 10 μg protein was loaded in each lane. aa, amino acid; IFN-α, interferon-α; tet, tetracycline.
hep26671-sup-0004-suppfig3.tif1183KSupporting Figure 3. Basis for the three-dimensional model of the N-terminal cytosolic tip and the transmembrane segment of GPx8 illustrated in Figure 2C. Similarity of the GPx8 N-terminal sequence to the sequence of amino acids 155-171 of filamin A immunoglobulin-like repeat 10 (PDB entry 3RGH; Page RC et al. Acta Crystallogr Sect F Struct Biol Cryst Commun 2011;67:871-876) (top row) and to the transmembrane α-helix 20-41 of FXYD protein of the sodium-potassium ATPase (PDB entry 2ZXE, chain G; Shinoda T et al. Nature 2009;459:446-450) (middle row). The degree of amino acid physicochemical conservation at each position relatively to GPx8 sequence is indicated with the similarity index according to ClustalW convention (asterisk, invariant; colon, highly similar; dot, similar; Thompson JD et al. Nucleic Acids Res 1994;22:4673-4680). The GPx8 sequence displays 71% similarity (47% identity) within the 17 amino acid overlap for the 3RGH sequence, and 89% similarity (26% identity) with the 19 amino acid overlap for the 2ZXE sequence. “Struct.” represents the secondary structure of the different protein segments deduced from the corresponding three-dimensional structures: h, helix; s, bend; t, hydrogen-bonded turn. The beginning of the X-ray structure of GPx8 luminal domain is also indicated (PDB entry 3KIJ; Nguyen VD et al. J Mol Biol 2011;406:503-515; bottom row). #, GPx8 transmembrane segment predicted by algorithms PHDhtm (http://www.predictprotein.org/), TMHMM (http://www.cbs.dtu.dk/services/TMHMM), DAS (http://www.sbc.su.se/miklos/DAS/), TopPred2 (http://mobyle.pasteur.fr/cgi-bin/MobylePortal/portal.py?form_toppred), Tmpred (http://www.ch.embnet.org/software/TMPRED_form.html), HMMTOP (http://www.enzim.hu/hmmtop/), and SOSUI (http://bp.nuap.nagoya-u.ac.jp/sosui/).
hep26671-sup-0005-suppfig4.tif1875KSupporting Figure 4. siRNA-mediated silencing of GPx8. Huh-7.5 cells overexpressing GPx8 (Huh-7.5-GPx8) were transfected with GPx8 siRNAs #1 or #2 or a non-targeting control, as described in the Materials and Methods section. Cell lysates obtained 96 h later were analyzed by 15% SDS-PAGE and immunoblot using polyclonal antibody IN103 against GPx8. A total of 10 μg protein was loaded in each lane.
hep26671-sup-0006-suppfig5.tif16052KSupporting Figure 5. Subcellular localization of GPx8 and mutants. Huh-7.5 cells transduced with recombinant viruses expressing GPx8, GPx8C11A, GPx8S12 and GPx8C79T were infected with Jc1 HCVcc and stained 48 h later with polyclonal antibody IN103 against GPx8 and monoclonal antibody #337 against HCV NS3 (kindly provided by Thomas Pietschmann, Twincore, Hannover, Germany) as well as Alexa Fluor® 488- and 594-conjugated secondary antibodies (Life Technologies), respectively. Nuclei were stained with DAPI (AppliChem). Representative confocal laser scanning microscopy images are shown for each construct. GPx8 (green) and NS3 (red) are shown separately and as merged images including the DAPI staining (blue). An inset is shown at higher magnification for GPx8-transduced cells. GPx8 is mostly expressed at the ER membrane including the perinuclear membrane and partially colocalizes with HCV NS3. No appreciable difference is observed in the subcellular localization of GPx8 and the different GPx8 mutants.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.