Genetic and Functional Characterization of the N-Terminal Region of the Hepatitis C Virus NS2 Protein

Abstract

The hepatitis C virus (HCV) NS2 protein has dual roles within the HCV life cycle. While well characterized as an autoprotease that cleaves the NS2/NS3 junction, NS2, primarily via its N-terminal region, is also involved in virion morphogenesis. In order to map the determinants necessary for infectious virus production and gain further insight into the multiple points at which NS2 may impact this process, a detailed mutational analysis of residues spanning amino acids (aa) 1 to 92 was performed. Initial block mutagenesis (5 or 7 amino acid residues) in both bicistronic and monocistronic HCV cell culture-based (HCVcc) genomes revealed that all but two blocks had various levels of impaired infectious virus production. None of these mutations affected RNA replication, indicating that the N-terminal region of NS2 is not required for NS2-3 processing and replicase assembly. Fine mapping identified 29 critical residues that, when mutated, yielded at least a 1 log decrease in infectious virus titers. These mutants were characterized further with respect to release of extracellular HCV RNA and core, intracellular infectivity, thermal stability of virus particles, and NS2 interactions. While the most severely debilitated mutants were impaired early in the assembly process, which is in agreement with previous reports, others targeted later steps of virus production, most notably egress. Thus, in addition to participating in early steps in virion assembly, this comprehensive mutagenesis study suggests yet another role for NS2 in later steps in virus production.

INTRODUCTION

Hepatitis C virus (HCV) is a positive-strand RNA virus of 9.6 kb from the Flaviviridae family. An estimated 170 million individuals worldwide have been infected (1–3). Of these, 50% to 80% are chronic carriers and have an increased risk of developing hepatosteatosis, cirrhosis, hepatocellular carcinoma, and liver-related death. Phylogenetic analysis of various isolates has resulted in the classification of six main genotypes (gt) containing a number of subtypes (e.g., 1a, 1b, 1c, etc.) and the recent addition of a novel seventh genotype (1, 4). Due to the large sequence diversity of HCV and limited animal models (5), the development of broadly successful treatments such as with direct-acting antivirals (DAA) or prophylactic vaccine has been difficult. Multidrug combinations, similar to highly active antiretroviral therapy (HAART) for HIV infection, will likely be necessary to target different aspects of infection in order to reduce the possibility of viral resistance. The full scope of each viral protein's contribution to the HCV life cycle is still being determined, and such insight may provide additional targets to exploit.

With the establishment of an HCV cell culture-based infection system (HCVcc) dependent on the Japanese JFH-1 (gt 2a) strain, a more complete picture of the viral life cycle has begun to emerge, especially for processes such as virus entry and assembly (3, 6, 7). The HCV genome contains one large open reading frame (ORF) encoding structural and nonstructural proteins flanked on either end by 5′ and 3′ nontranslated regions (NTR) that aid in the control of translation and replication. Cellular and viral encoded proteases cleave the large polyprotein (approximately 3,000 amino acids [aa]) both co- and posttranslationally into 10 different proteins: core protein, envelope proteins E1 and E2, p7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B. The replicase complex, which produces both positive- and negative-stranded RNA molecules, is comprised of NS3, NS4A, NS4B, NS5A, and NS5B. Recent reports have highlighted several of these nonstructural proteins (i.e., NS3, NS4B, and NS5A) as factors contributing to virus production; however, the exact mechanisms are still unknown (8–11). Structural components of the HCV virion include the core protein, which encapsidates the RNA genome, and the glycoproteins E1 and E2, which are displayed in the endoplasmic reticulum (ER)-derived lipid bilayer that envelops the nucleocapsid and mediate early entry events (12, 13).

Studies examining p7 and NS2 in particular have also benefited from the HCVcc system. p7, an integral membrane protein, oligomerizes into higher-order structures (14, 15) and has been shown in vitro to have cation channel activity (16, 17). Genetic analysis has demonstrated that p7 is essential for assembly and release of infectious virus (13, 18, 19). This has led to the model that p7 functions as a viroporin similar to influenza A virus M2 and human immunodeficiency virus type 1 (HIV-1) vpu, which alter membrane permeability, resulting in the loss of vesicular compartment acidification and protection of virion progeny during egress/maturation (20).

NS2 was initially discovered to be a zinc-dependent autoprotease that acts in concert with NS3 for cleavage of the NS2/NS3 junction (21–23). However, later reports demonstrated that NS2 is also required for virus assembly (9, 13, 24–28). This role involves the full-length protein in its cleaved form as shown through the use of bicistronic constructs that express NS2 and NS3 independently of NS2-3 cleavage (13). The mature NS2 is a 23-kDa membrane-associated protein with perinuclear ER localization (22, 29). The N-terminal, membrane-anchoring domain consists of three putative transmembrane segments (TMS), and is followed by the C-terminal protease domain, which resides on the cytoplasmic face of the ER membrane (22, 30). A number of cell culture-adaptive mutations and compensatory mutations that enhance or restore virus production have been mapped to the NS2 N-terminal region (9, 28, 31–34). Likewise, mutagenesis of select residues within the N-terminal domain has implicated the involvement of NS2 in the early steps of the assembly process (9, 25, 28, 35, 36). Sequence alignment of several HCV genotypes demonstrates high conservation in the protease domain (aa 94 to 217), and the crystal structure of this domain indicates that NS2 forms a dimer containing a composite active site analogous to those of cysteine proteases (37). In addition to containing residues necessary for NS2-3 cleavage, this domain contains determinants necessary for virion morphogenesis as well (9, 26). Mutation within this domain has led to the proposal that NS2 may also act at a later step of particle assembly, possibly in maturation or egress (32). While the production of infectious virus requires both the N-terminal membrane association and protease domains of NS2, its proteolytic activity is not necessary (13, 25, 26).

In this study, we conducted a comprehensive mutagenesis screen of NS2 residues (aa 1 to 92 [1–92]) to gain further insight into the involvement of this region in virion morphogenesis. Characterization of mutations in 29 key residues suggests roles for NS2 in both virion assembly and egress.

MATERIALS AND METHODS

Plasmid construction.

J6/JFH NS2-IRES-nsGluc2AUbi and J6/JFH p7-nsGluc2A have been described previously (13). All mutations were introduced by overlap PCR using standard procedures. Plasmid and primer sequences for both block and single substitutions are available upon request. To generate single substitutions containing the H77 epitope (aa 197–208 within the NS2 protease domain), the J6 sequence (GREVLLGPADG; differences between gts are underlined) was swapped for the H77 sequence (GQEILLGPADG). A PAGE-purified primer, which encoded the H77 epitope and contained a stop codon and PmeI restriction site at the 3′ end of NS2, was used for subcloning into the J6/JFH NS2-IRES-nsGluc2AUbi mutated background. All constructs were confirmed by enzyme digests and sequencing.

Cell culture.

Huh-7.5 (38) and Huh-7.5.1 (39) cells were propagated in Dulbecco's modified minimal essential medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 0.1 mM nonessential amino acids (NEAA). Prior to electroporation, penicillin (100 U/ml) and streptomycin (100 μg/ml) were added to growth media and kept throughout the course of the experiment. Huh-7.5 CD81 knockdown cells (Huh-7.5 CD81^lo) were grown in complete DMEM supplemented with 6 μg/ml blasticidin (40). Cells were grown at 37°C in a humidified 5% CO₂ atmosphere.

In vitro RNA synthesis.

Viral cDNAs were linearized with XbaI and purified using a MinElute PCR purification kit (Qiagen, Valencia, CA). Linearized constructs were eluted in EB buffer (Qiagen). RNAs were synthesized according to the manufacturer's directions using a T7 RiboMAX Express large-scale RNA production system (Promega, Madison, WI). RNAs were isolated with an RNeasy RNA isolation kit according to the manufacturer's directions with a second DNase I digestion (Qiagen, Valencia, CA). RNAs were eluted in 1 mM sodium citrate (RNA Storage Solution; Ambion, Austin, TX) (pH 6.4). The RNA integrity and concentration were determined by agarose gel electrophoresis and absorbance at 260 nm, respectively.

RNA transfection.

In vitro-transcribed RNAs were electroporated into cells utilizing a 4-mm-gap 25- or 96-well plate format (BTX ElectroSquare Porator ECM830 with Plate Handler; Harvard Apparatus, Holliston, MA). After trypsinization, cells were washed twice and resuspended in cold RNase-free Dulbecco's phosphate-buffered saline without Ca²⁺/Mg²⁺ (D-PBS; Gibco-Invitrogen) at 1.5 × 10⁷ cells/ml. An aliquot of the cell suspension (200 μl [3 × 10⁶ cells]) was mixed with 5 μg of RNA transcripts, placed into the cuvette, and electroporated (0.86 kV [25-well plates] or 0.80 kV [96-well plates]; 99 ms; 5 pulses). Pulsed cells were left to recover for 5 min at room temperature and then resuspended into complete DMEM. Cells were plated in either 96-well plates (8 × 10³ cells) for replication analysis or 6-well plates (6 × 10⁵ cells) for virus production and Western blot analysis. Initial block and single-substitution analyses were done in triplicate across five 96-well electroporation plates.

Replication and virus production analysis.

Replication kinetics were analyzed by harvesting cells at 4, 24, 48, 72, and 96 h postelectroporation (h.p.e.) in 1× Renilla luciferase assay lysis buffer (Promega, Madison, WI). Quantification of Gaussia luciferase activity was performed on a Centro LB960 luminometer (Berthold Technologies, Oak Ridge, TN) using Renilla luciferase substrate (Promega) following the manufacturer's instructions. Viral supernatants were collected from 6-well plates at 48 h.p.e., clarified by centrifugation (2100 × g, 5 min, 4°C), filtered (0.45 μm pore size), and stored at −80°C. HCV infectious titers were determined by a limiting dilution assay as described previously (41). The 50% tissue culture infectious dose (TCID₅₀) was calculated according to the method of Reed and Muench (42).

Isolation and quantification of released HCV RNA.

To quantify HCV genome release, total RNA was isolated from 140 μl clarified cell culture supernatant using a QiaAmp UltraSens kit (Qiagen). Quantitative reverse transcription-PCRs (qRT-PCRs) were assembled using a LightCycler amplification kit (Roche, Basel, Switzerland) with extracted total RNA, according to the manufacturer's instructions. Primers are directed against the viral 5′ NTR. qRT-PCRs were performed using a LightCycler 480 system (Roche).

Core protein ELISA.

Release of HCV core protein was quantified using the Ortho HCV antigen enzyme-linked immunosorbent assay (ELISA; Ortho Clinical Diagnostics, Raritan, NJ) according to the manufacturer's instructions. Briefly, 100 μl of filtered cell culture supernatants was applied to plates coated with a mixture of mouse anti-core monoclonal antibodies (MAb). Antigen was detected by addition of a second MAb cocktail conjugated to horseradish peroxidase (HRP). After washing and development, the absorbance at 492 nm was measured and core protein quantified by comparison to a standard curve.

Intracellular infectivity assay.

To measure cell-associated or intracellular infectious HCV (here termed intracellular infectious virus), 3.0 × 10⁶ electroporated cells were plated in 100-cm² plates. Media supernatants were filtered and stored at −80°C. Cells were trypsinized, washed twice in complete media, and then stored at −80°C until use. Pellets of cells were thawed and resuspended in 500 μl of complete medium. Cells were then lysed by four freeze-thaw cycles and centrifuged twice at 5,900 × g for 3 min to remove cellular debris. Titers of extra- and intracellular supernatants were determined on Huh-7.5 cells, as described above. After calculation of the total extra- and intracellular infectivity for the dish, the data were plotted as the percent intracellular infectivity of the total (i.e., of both extra- and intracellular supernatants).

Thermal stability of mutated HCV particles.

Infectious HCV stability was determined as previously described with minor alterations (34). Extracellular supernatants were harvested 48 h.p.e. Aliquots (500 μl) were incubated at 37°C for various times (0, 0.5, 1, 2, 3, 4, 6, 8, and 10 h) and then stored at −80°C until use. After collection, 200 μl was placed on naive Huh-7.5 cells seeded in a 96-well format. Cells were lysed at 48 h postinfection (h.p.i.), and relative levels of infectivity as measured by Gaussia luciferase activity were determined. The luciferase values of the mock controls were subtracted from all data points. All time points were then compared to the 0 h time point, which was normalized to 100%, and the data were plotted as percent infectivity versus incubation time. Curves generated from nonlinear regression data were fitted to a first-order decay equation by using the Prism 5 software program (GraphPad, La Jolla, CA) to determine the slope (−k) where the plateau was constrained to a value of 0. The half-live for each infectious HCV strain (t_1/2) was calculated in hour values by using ln 2/k (7).

Immunoprecipitation (IP) of NS3-associated factors.

Huh-7.5.1 cells electroporated with H77 epitope-containing genomes were harvested at 48 h.p.e. Cells were washed twice with ice-cold D-PBS and subjected to trypsinization. Pellets of cells were lysed for 30 min on ice in a modified radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 5 mM EDTA, 1 mM Na₃VO₄, 50 mM NaF, 1% Triton X-100, and 1 mini-Complete Protease inhibitor tablet [Roche, Indianapolis, IN]) and then centrifuged for 15 min at 4°C and 16,100 × g. Supernatants were mixed with glycerol (10% [wt/vol]). Protein concentrations were determined by bicinchoninic acid (BCA) assay (Pierce Thermo Fisher) and lysates frozen (−80°C) until immunoprecipitation.

For NS3 immunoprecipitations, 10 μg of NS3 antibody (HCV-2E3 [28]; a generous gift from R. Bartenschlager) was incubated overnight with 1 mg of total protein lysates and captured with 150 μl of Dynabeads protein G for 1 h. Bound complexes were washed three times with 100 mM Tris HCl [pH 7.5], 150 mM NaCl, 1 mM EDTA, and 0.1% Triton X-100 and separated on NuPage 4% to 12% Bis-Tris Midi gels with a morpholinepropanesulfonic acid (MOPS) buffering system (Invitrogen, Carlsbad, CA).

Western blotting.

Cells were harvested at 48 h.p.e. by two washes with ice-cold D-PBS and lysing directly in 2× Laemmli buffer. The lysate was homogenized on a QIAshredder spin column (Qiagen, Valencia, CA). The homogenates were stored at −80°C until use. Lysates were boiled, and 1/10 of the total volume was loaded on a 4% to 20% Tris–glycine Midi gel. Proteins were transferred onto a 0.2-μm-pore-size polyvinylidene difluoride (PVDF) membrane (200 mA for 1 h and then 300 mA for 15 min). After blocking in 5% skim milk–D-PBS with 0.1% Tween 20, blots were incubated overnight at 4°C with primary antibodies against NS2 (6H6 [1:500] [26]), NS3 (1878 [Virostat, Portland, ME] [1:100] or HCV-2E3 [1:500]), NS5A (9E10 [1:2,000] [41]), core ([Austral Biologicals, San Ramon, CA], 1:1,000), E2 (3/11 [1:500] [43]), or β-actin (AC-15 [Sigma-Aldrich, St. Louis, MO], 1:2,000). The next day, blots were incubated with goat α-mouse IgG (Invitrogen) (1:1,000) or goat α-rat IgG (Invitrogen) (1:1,000) secondary antibody, each conjugated to Alexa Fluor 488 for 1 h at room temperature. For immunoprecipitations, HRP-conjugated goat α-mouse IgG (Jackson ImmunoResearch) (1:8,000) or donkey α-rat IgG (Jackson ImmunoResearch) (1:5,000) secondary antibody was used. Blots were washed and imaged on the Molecular Dynamics Typhoon 9400 Variable Imager (GE Healthcare Biosciences) or developed using a SuperSignal West Pico or Femto chemiluminescent substrate kit (Pierce Thermo Scientific).

RESULTS

Initial characterization of the N-terminal region of NS2 by block mutagenesis.

A number of studies have demonstrated the importance of the NS2 protein in HCV virion morphogenesis (9, 13, 24–28, 32, 35, 36, 44). The N-terminal region, in particular, has a number of determinants that, when mutated, can enhance or block virus assembly. Thus, with the goal of providing a more comprehensive picture of the contributions of the N-terminal membrane association domain of NS2 (aa 1–92), we employed an alanine-scanning mutagenesis approach. The bicistronic J6/JFH NS2-IRES-nsGluc2AUbi derivative, which contains a Gaussia luciferase (Gluc) initiating the second cistron after the encephalomyocarditis virus (EMCV) internal ribosome entry site (IRES), was used as the parental wild-type (WT) construct (13). This construct allows quantitative high-throughput analysis of NS2 mutations by measuring luciferase activity and was previously shown to replicate and produce levels of infectious virus similar to those produced using the nonreporter bicistronic derivative, J6/JFH NS2-IRES-NS3. By separation of NS2 from NS3, the role of NS2 in virus assembly can be assessed independently of its role in RNA replication (i.e., autoproteolytic cleavage). All nonalanine residues in the putative transmembrane segments (aa TMS1, 5–25; TMS2, 33–52; TMS3, 63–88 [based on prediction algorithms]) were mutated to alanine and alanine residues were mutated to leucine in blocks of five residues (Fig. 1A). Within the loops that connect the TMS, residues were mutated to disrupt charge or polarity.

Fig 1.

Block mutagenesis of the N terminus of J6 NS2. (A) Schematic depicting the block substitutions within the first 92 residues of NS2 in the parental bicistronic construct, J6/JFH NS2-EMCV IRES (I)-nsGluc2AUbi. All substitutions were generated in blocks of five residues, with the exception of the first loop. The three putative transmembrane segments (TMS) determined on the basis of a theoretical prediction algorithm are indicated. (B) Replication kinetics of the bicistronic genomes. HCV replication was analyzed for Gaussia luciferase activity in cell lysates at 4, 24, 48, 72, and 96 h.p.e. Stacked bars represent values for each time point, with standard deviations shown. J6/JFH NS2-I-nsGluc2AUbi residue blocks changed to alanine or leucine are identified. Block 170–175 contains residues Met 170 and Ile 175 previously shown to be impacted for virus production when mutated (26). WT, J6/JFH NS2-I-nsGluc2AUbi; GNN, J6/JFH NS2-I-nsGluc2AUbi/GNN; ΔE1E2, J6/JFH NS2-I-nsGluc2AUbi/ΔE1E2; RLU, relative light units. (C) Infectious virus production at 48 h.p.e. Titers determined by limiting dilution assay are plotted. The dashed gray horizontal bar indicates the lower limit of the TCID₅₀ assay. Means and standard deviations are plotted for the results determined for three replicates. (D) Intracellular protein expression. Lysates harvested 48 h.p.e. were analyzed for NS2, NS5A, and actin expression by Western blotting. Blots are representative of pooled extracts from two replicates.

RNA replication kinetics were quantified by measuring Gluc activity (Fig. 1B). As a negative control, we utilized a polymerase-defective genome (GNN), which reached background luciferase levels (approximately 4.5 × 10³ relative light units [RLUs]) at between 24 and 48 h.p.e. All mutant NS2 genomes exhibited replication levels similar to those seen with the parental genome at 24 h. For some mutants, later time points did show some (a roughly- 3 to 5-fold) decrease that was later shown to be due to decreased infectious virus production and spread (see below). Infectious titers were determined by limiting dilution assays (Fig. 1C), including negative controls with a partial in-frame deletion of the E1 and E2 coding region (ΔE1E2) or alanine substitutions for NS2 residues 170–175. This NS2 mutant includes alanine substitutions for Met 170 and Ile 175 and was previously shown to result in a severe decrease in virus production without affecting RNA replication in the bicistronic background (26). Only mutants with mutations in two blocks, 48–52 and 53–57, which are located at the end of TMS2 and the beginning of the second loop, respectively, showed titers similar to WT titers. All the remaining mutants exhibited either moderate (16–20, 58–62, 63–67, 68–72, 73–77, and 78–82) or severe (1–5, 6–10, 11–15, 21–25, 26–32, 33–37, 38–42, 43–47, 83–87, 88–92, and [control] 170–175) defects in infectious virus production. The level of NS2 protein at 48 h.p.e. varied; partial deletion of the E1E2 region (ΔE1E2) and mutagenesis of residues 43–47 led to increased levels of NS2, perhaps by increasing its stability, while the remaining mutants had NS2 levels similar to or lower than those of the WT (Fig. 1D).

Efficient cleavage at the NS2/NS3 junction is required for formation of a functional HCV RNA replicase (45). To determine if mutations in the N-terminal region might affect NS2-3 processing and RNA replication, mutations were introduced into a monocistronic J6/JFH p7-nsGluc2A background (Fig. 2A [13]). All mutants displayed replication kinetics similar to that of the WT, and only the residue 170–175 control, previously shown to be impaired for polyprotein processing and RNA replication in the monocistronic background, had decreased levels similar to those of the GNN (Fig. 2B [26]). Most mutants demonstrated impaired infectious virus production, with the exception of those corresponding to residues 48–52, 53–57, and 68–72, for which production levels were comparable to or slightly elevated with respect to that of the WT (Fig. 2C). NS2 levels, compared to the levels of the monocistronic parent, were similar to those previously observed for the mutants in the bicistronic parental background, with the exception of the residue 6–10 mutant, where NS2 expression had previously been low but was now similar to WT expression (Fig. 2D). Thus, mutagenesis of the N-terminal membrane association domain of NS2 reinforces its primary role in virus production rather than in replicase formation and function. Consistent with results in the bicistronic background, specific residues in the 48 to 57 region are not required for virus production.

Fig 2.

Block mutagenesis of the N terminus of J6 NS2 within a monocistronic background. (A) Schematic depicting the block substitutions within NS2 in the parental monocistronic construct, J6/JFH p7-nsGluc2A. (B) Replication kinetics of the mutated monocistronic genomes. HCV replication was analyzed for Gaussia luciferase activity in cell lysates at 4, 24, 48, 72, and 96 h.p.e. WT, J6/JFH p7-nsGluc2A; GNN, J6/JFH p7-nsGluc2A/GNN; ΔE1E2, J6/JFH p7-nsGluc2A/ΔE1E2; RLU, relative light units. (C) Infectious virus production at 48 h.p.e. Titers determined by limiting dilution assay are plotted. The dashed gray horizontal bar indicates the lower limit of the TCID₅₀ assay. Means and standard deviations are plotted for the results determined for three replicates. (D) Intracellular protein expression. Lysates harvested 48 h.p.e. were analyzed for NS2, NS5A, and actin expression by Western blotting. Blots are representative of pooled extracts from two replicates.

Fine mapping identifies 29 critical residues.

To determine the contribution of individual residues, single substitutions were generated and analyzed in the bicistronic reporter background. Although a slight variation was observed at the 4-h time point, most mutants plateaued to replication levels comparable to WT levels by 48 h (Fig. 3A). A subset of mutants (D2A, Q8A, T23A, K27E, W35A, W36A, K81A, and A85L), which also had severely decreased infectious titers, exhibited slightly decreased Gluc levels at 48 h. To separate possible effects of these mutations on RNA replication versus virus production, we conducted a single-cycle assay by electroporation of Huh-7.5 CD81 knockdown cells (Huh-7.5 CD81^lo; Fig. 4A [40]). With these cells, which are permissive for HCV RNA replication and virus production but blocked for virus entry, all of the mutants exhibited WT RNA replication kinetics, again reinforcing the conclusion that this domain of NS2 is not required for replicase function.

Fig 3.

Single substitutions within the N terminus of NS2. (A) Replication kinetics of single substitutions at 4, 24, 48, 72, and 96 h.p.e. (B) Infectious virus production at 48 h.p.e. The dashed gray horizontal bar indicates the lower limit of the TCID₅₀ assay. Means and standard deviations are plotted for the results determined for three replicates for all mutated NS2 genomes, while results for controls WT, GNN, and ΔE1E2 were averaged across all electroporation sets (12 replicates).

Fig 4.

Replication and intracellular protein expression of mutated NS2 genomes in Huh-7.5 CD81^lo cells. (A) Replication kinetics of the mutated bicistronic genomes in electroporated Huh-7.5 CD81^lo cells. HCV replication of D2A, Q8A, I17A, T23A, K27E, W35A, W36A, F77A, K81A, W82A, A85L, and L92A NS2 genomes were analyzed for Gaussia luciferase activity in cell lysates at 4, 24, 48, 72, and 96 h.p.e. Means and standard deviations are plotted for the results determined for two replicates. (B) Intracellular protein expression. Lysates harvested 48 h.p.e. were analyzed for NS2, NS5A, and actin expression by Western blotting. Blots are representative of pooled extracts from two replicates.

There was great variation in infectious virus yield depending on the individual mutation (Fig. 3B). For simplicity, mutants were segregated into 5 categories based on comparison with WT titers (summarized in Table 1) as follows: (i) a single mutant with an increased titer (>0.5 log); (ii) mutants exhibiting no or a minimal effect (within one standard deviation [SD] of the WT titer); (iii, iv, and v) mutations with a slight (≤1 log), moderate (1 to 2 logs), or severe (>2 logs) reduction. While many substitutions fell into the no-effect or slight-effect groups, only one mutation (P73A) yielded increased titers. The remaining mutants with moderate (A3L, H6A, G7A, G10A, L13A, L22A, G25A, L30A, R32E, Y39A, L41A, G60A, R61E, D62R, A67L, F77A, W82A, G88A, and P89A) or severe (D2A, Q8A, I17A, T23A, K27E, W35A, W36A, K81A, A85L, and L92A) defects were examined further.

Table 1.

Comparison of single substitutions in NS2^a

Titer increase >0.5 log	No titer change	Titer decrease
		≤1 log	>1 to 2 log	>2 log
P73A	Y1A	S4A	A3L	D2A
	V5A	V15A	H6A	Q8A
	I9A	M16A	G7A	I17A
	A11L	T18A	G10A	T23A
	A12L	L19A	L13A	K27E
	L14A	T21A	L22A	W35A
	F20A	T28V	G25A	W36A
	P24A	L29A	L30A	K81A
	Y26F	L37A	R32E	A85L
	S31A	T42A	Y39A	L92A
	F33A	G44A	L41A
	L34A	G63A	G60A
	C38A	I64A	R61E
	L40A	I65A	D62R
	L43A	V68A	A67L
	E45A	A69L	F77A
	A46L	I70A	W82A
	M47A	F71A	G88A
	R58E	G74A	P89A
	G59A	D78A
	W66A	L83A
	Y72A	V86A
	V75A	L87A
	V76A	A90L
	I79A	Y91A
	T80A
	L84A

^aData represent average titers compared to the parental construct.

Since previous reports have indicated that NS2 is a relatively short-lived protein (t_1/2 of approximately 3 h [35, 46]), mutations destabilizing NS2 could lead to diminished levels and loss of function and explain the observed decreased virus titers. While the D2A, Q8A, W35A, W36A, F77A, K81A, W82A, and L92A mutants did display lower viral protein levels (Fig. 5, top panel) in Huh-7.5 cells, when analyzed in the CD81^lo background only, D2A showed diminished NS2 levels relative to the other viral proteins (Fig. 4B). Interestingly, a change in charge from an acidic to a basic residue (i.e., D62R) altered the mobility of this NS2 mutant, which has been observed before with an alteration of this residue to an alanine (9). Given the proximity of the D2A mutation to the p7/NS2 cleavage site and the presence of a higher-migrating band noted upon longer exposure of the NS2 immunoblots (data not shown), this was probably due, at least in part, to inefficient processing. Expression of other HCV factors known to associate with NS2, including core, E2, NS3, and NS5A, did not change drastically with respect to the NS2 mutations (Fig. 5). Thus, with the possible exception of the D2A mutant, decreased infectious titers could not be attributed to diminished levels of NS2 or other HCV proteins implicated in virus assembly.

Fig 5.

Intracellular protein expression of assembly factors. Lysates harvested 48 h.p.e. were analyzed by Western blotting for NS2, core, E2, NS3, NS5A, and actin expression. Blots are representative of pooled extracts from two replicates.

Further characterization of highly impaired NS2 mutants.

We first examined the effect of these mutations on particle release and specific infectivity (Fig. 6). Parental genomes with in-frame deletions of NS2 (ΔNS2) or core (ΔC) were included as additional controls. In general, there was a strong correlation of extracellular RNA and core levels with infectious titers, i.e., those mutants with mutations reducing infectious titers by 2 logs or more (D2A, Q8A, G10A, I17A, T23A, K27E, W35A, W36A, K81A, and L92A) had decreased RNA and core quantities (Fig. 6B and C; Table 2). These results were similar to those seen with the assembly-defective genomes, ΔC, ΔE1E2, and ΔNS2, indicating a defect upstream of virus particle release. The only exception to this was the A85L mutant, which also had severely impaired infectious virus production and lower RNA release but core protein amounts comparable to WT levels. This mixed phenotype may suggest a decoupling of RNA packaging from nucleocapsid formation. Other mutants with moderately reduced titers that also demonstrated a disparity between RNA and core release were A3L, G7A, L13A, G25A, Y39A, R61E, F77A, and P89A. The remaining substitutions had amounts of core and RNA either equal to or larger than the WT amounts. Furthermore, all mutants displayed diminished specific infectivity, with D2A, Q8A, I17A, T23A, Y39A, R61E, K81A, A85L, P89A, and L92A having at least a 10-fold difference relative to the WT (Fig. 6D). Thus, all these mutations disrupted virion assembly or secretion to various degrees.

Fig 6.

Characterization of extracellular virus. (A) Infectious virus production at 48 h.p.e. ΔNS2, J6/JFH NS2-I-nsGluc2AUbi/ΔNS2; ΔC, J6/JFH NS2-I-nsGluc2AUbi/ΔC. The dashed gray horizontal bar indicates the lower limit of the TCID₅₀ assay. (B and C) Release of total viral RNA (B) or core protein (C) into culture supernatants at 48 h.p.e. Isolated RNA was analyzed by qRT-PCR to determine the copy number of HCV genomes per milliliter of supernatant. Core ELISA was performed to determine amount of core released per milliliter of supernatant. (D) The specific infectivity, calculated as infectivity per HCV RNA copy, was plotted. Values for those mutations and controls whose titers were below the limit of detection (K27E, W35A, W36A, GNN, ΔE1E2, ΔNS2, and ΔC) were not determined. Means and standard deviations are plotted for the results determined for two replicates.

Table 2.

Comparison of extracellular RNA and core protein levels^a

Levels relative to WT	Core protein levels for indicated mutation
	<Core		=Core		>Core
<RNA	D2A	K27E	A3L	F77A
	Q8A	W35A	G7A	A85L
	G10A	W36A	L13A	P89A
	I17A	K81A	G25A
	T23A	L92A
=RNA	Y39A		H6A	D62R
	R61E		R32E	W82A
			L41A	G88A
>RNA			A67L		L22A
					L30A
					G60A

^aMutants with infectious virus titers <2 logs are highlighted with italics and boldface. A cutoff of ±25% was used to denote similarity to the WT.

Comparison of intracellular infectivity and extracellular infectivity.

In order to differentiate effects on assembly versus secretion, we evaluated the levels of infectious intra- and extracellular virus at 48 h.p.e. (Fig. 7A). All of the mutants, with the exception of K27E, W35A, and W36A, had detectable intracellular infectious virus, indicating an early and complete defect in infectious virus assembly. To determine the ratio between intracellular virus production and extracellular virus production, intracellular infectivity was determined as a percentage of the total infectivity for each individual mutation (Fig. 7B). The WT demonstrated little (approximately 5%) retention of infectious intracellular particles; however, most mutants fell within a range of 8% to 25% intracellular infectivity, with several having higher relative amounts of cell-associated infectious virus. The strongest phenotype was exhibited by the G25A and K81A mutants, with levels of 48% and 38%, respectively.

Fig 7.

Intracellular infectious virus produced by mutated NS2 genomes. Extra- and intracellular supernatants were harvested 48 h.p.e. and used to infect naïve Huh-7.5 cells. (A) Intra- and extracellular infectivity per transfection as determined by limiting dilution assay are plotted. (B) The total infectivity per transfection was determined for both intra- and extracellular supernatants. The percentage of intracellular infectivity was expressed relative to the sum of the amounts of infectivity for the dish. Dashed line represents 5-fold increase in intracellular infectivity relative to that of the WT. Intra, intracellular; Extra, extracellular. Means and standard deviations are plotted for the results determined for four replicates.

In terms of thinking about levels of intracellular versus extracellular infectious virus, we offer the following conceptual model. In an infected cell, as the first infectious virions are assembled, the ratio of intracellular infectivity to released virus is expected to be maximal. As virus production increases and reaches the steady state, the level of intracellular infectivity should also ramp up and then reach a plateau (assuming no changes in secretion kinetics). As released extracellular infectious virus accumulates over time, the ratio of intracellular to total infectivity would decrease. Assumptions (and caveats) with regard to this simple model include (among others) (i) constant secretion kinetics once intracellular infectious virus is formed, (ii) constant virion half-life and specific infectivity over the course of virus production, and (iii) no virus-induced effects on cell viability or production capacity, particularly at later time points. While the WT data generally fit this simple model, the G25A and K81A mutants differed. Rather than reaching a plateau, a peak of intracellular infectivity was observed at 48 h.p.e. which declined by 72 h.p.e. Extracellular virus levels plateaued (G25A) or declined (K81A; Fig. 8A and B). For both the WT strain and the K81A mutant, the portions of intracellular infectivity (relative to the total) decreased over time, albeit at disproportionate levels (Fig. 8B). In contrast, for the G25A mutant, while the ratio of intracellular infectivity to total infectivity decreased at 72 h.p.e., this was due to a decrease in intracellular infectivity rather than accumulation of extracellular infectious virus, the level of which remained constant. When examining these mutations in the other cell backgrounds where spread was inhibited (Huh-7.5 CD81^lo) or virus production enhanced (Huh-7.5.1 [39]), a similar trend in the ratio of intracellular to total infectivity was observed (WT<G25A<K81A; Fig. 8C and D). Although the data from the 48 h.p.e. time point for the G25A mutant, where intracellular infectivity is the same as WT infectivity but extracellular infectivity is greatly diminished, might suggest a secretion defect, the 72 h.p.e. data point with decreased intracellular infectivity could also point to a defect in assembly of infectious virus that becomes more pronounced as infection progresses.

Fig 8.

Kinetics of intracellular virus for G25A and K81A NS2 mutants. (A and C) Extra- and intracellular supernatants were harvested at either (A) 24, 48, and 72 h.p.e. from Huh-7.5 cells or (C) 48 h.p.e. from Huh-7.5 CD81^lo and Huh-7.5.1 cells. Infectivity as determined by limiting dilution assay is plotted. (B and D) The total infectivity per transfection for each was determined for both intra- and extracellular supernatants at either (B) 24, 48, and 72 h.p.e. from Huh-7.5 cells or (D) 48 h.p.e. from Huh-7.5 CD81^lo and Huh-7.5.1 cells. The percentage of intracellular infectivity was expressed relative to the sum of the amounts of infectivity for the dish. Means and standard deviations are plotted for the results determined for at least three replicates.

Thermal stability of extracellular virus particles.

A number of mutants had diminished infectious virus production but levels of extracellular core and RNA at or above the WT level (summarized in Table 2). This could be due to release of noninfectious or unstable particles. To examine this further, we determined the thermal stability of particles produced from mutants H6A, L22A, L30A, R32E, L41A, G60A, D62R, A67L, W82A, and G88A (Fig. 9A and B). Extracellular supernatants were harvested at 48 h.p.e. and replicate aliquots incubated at 37°C for 0.5, 1, 2, 3, 4, 6, 8, and 10 h. Infectivity for each sample was measured and plotted against the value for the zero time point. The half-life of all mutant particles was comparable to that of the WT (average WT t_1/2, 2.4 h), indicating that the lower specific infectivity previously observed (Fig. 6D) did not result from instability at 37°C. Thus, it seems likely that these mutant defects result in the production of noninfectious particles, which might occur because of diminished incorporation of HCV glycoproteins or critical host components such as ApoE (47–49) or, alternatively, by compromising p7's role in preventing acidification and virus inactivation during egress (20).

Fig 9.

The thermal stability of extracellular virus for select NS2 mutations is similar to that of the wild type. The infectious stability of WT and NS2 mutants was determined by incubating supernatants harvested 48 h.p.e. at 37°C for 0.5, 1, 2, 3, 4, 6, 8, or 10 h. Naïve Huh-7.5 cells were incubated with supernatants, and Gaussia luciferase activity was determined 48 h p.i. Levels of infectivity at various times for WT and NS2 mutants (in panel A, L22A, L30A, R32E, W82A, and G88A; in panel B, H6A, L41A, G60A, D62R, and A67L) are indicated. Infectious stability (t_1/2) values were calculated as described in Materials and Methods. The dark solid line is the curve generated from the nonlinear regression analysis of WT, while dashed lines represent the different NS2 mutations. Means and standard deviations are plotted for the results determined for at least two replicates.

Impact of NS2 mutations on NS3 interactions with viral assembly partners.

Genetic and biochemical analyses have shown that NS2 associates with both structural and nonstructural HCV proteins that are required for virus production (28, 36, 44). Thus, we examined whether NS2 mutations might disrupt binding to these proteins, resulting in the impaired virus production observed. Immunoprecipitation of J6 NS2 protein under nondenaturing conditions, however, proved inefficient. This monoclonal antibody (6H6) was raised against genotype 1a NS2 (H77) and was capable of efficient immunoprecipitation of this protein under similar conditions (data not shown [26]). Concerned about the effects of an N-terminal epitope (C-terminal tags are nonfunctional [26]), especially for mutants located within TMS1, we examined the antibody epitope. Since only two amino acids (residues R199Q and V201I) differ between J6 and H77 in the 6H6 epitope located in the NS2 protease domain, we introduced these substitutions into our parental mono- and bicistronic genomes. After observing that RNA replication and virus production were unchanged and NS2 could be readily immunoprecipitated (data not shown), we engineered the 6H6 H77 epitope into our mutant panel. For the majority of these reengineered mutants, virus production (Fig. 10A) and NS2 levels (data not shown) paralleled those observed in our initial analyses. However, for T23A, L30A, R32E, Y39A, and R61E, virus production was impaired by 0.5 to 2 logs, despite RNA replication kinetics comparable to WT kinetics (data not shown). Thus, either direct or indirect interactions between the first loop/TMS2 of NS2 and the protease domain (harboring the altered epitope sequence) may influence virus production. Finally, we also examined these in Huh-7.5.1 cells (Fig. 10A [39]). For most mutants, infectious titers in Huh-7.5.1 cells were increased approximately 0.5 to 1 log except for a small group (G25A, A67L, K81A, W82A, and A85L) that exhibited no or modest changes. Interestingly, the mutants which had lower titers with the epitope modification (T23A, L30A, R32E, Y39A, and R61E) were restored to their original levels; suggesting that NS2 interacts with one or more host factors to perform its function or functions in virus assembly, which differ between the two cell backgrounds.

Fig 10.

NS2 mutants display differential levels of binding between structural and nonstructural component proteins. (A) Infectious virus production comparison between virus with parental mutations and virus with mutations reengineered with the H77 epitope. Values for parental mutations in Huh-7.5 cells are from the data described for Fig. 6A. Mutations reengineered with the H77 epitope were electroporated into Huh-7.5 cells (H77 epitope) or Huh-7.5.1 cells (H77 epitope - 7.5.1). Supernatants were harvested at 48 h.p.e. Titers determined by limiting dilution assay are plotted. The dashed gray horizontal bar indicates the lower limit of the TCID₅₀ assay. Means and standard deviations are plotted for the results determined for two replicates. (B) NS3-associated complexes isolated from Huh-7.5.1 cells electroporated with mutated NS2, WT, ΔNS2, ΔC or ΔE1E2 genomes harvested 48 h.p.e. Isolated complexes were separated by gel electrophoresis as described above and immunoblotted for NS2, NS3, E2, and core proteins. WT extracts (100 μg) were loaded as an input control. The control lane corresponds to isotype IgG control immunoprecipitation with WT extracts. Asterisks indicate bands from heavy and light chains within the NS2, NS3, and core blots.

Based on these results, Huh-7.5.1 cells were used to determine possible effects of the NS2 mutations on interactions between HCV factors involved in assembly, specifically, NS3, NS5A, and E2, by coimmunoprecipitation (co-IP). While the NS2 mutants displayed differences in the level of NS2-associated factors, these viral factors could be precipitated in the absence of NS2 and additional attempts under higher-stringency wash conditions failed to diminish this background (data not shown). To overcome these potentially confounding background problems, we performed immunoprecipitations with NS3 antibody (Fig. 10B). Since NS2 is thought to regulate interaction between nonstructural proteins and the virion components for productive assembly, this provided an independent means of assessing the effect of NS2 mutations on the interaction of NS3 with NS2, E2, and core. We did not examine NS5A using the anti-NS3 antibody, given that it would be difficult to distinguish between NS5A involved in RNA replication versus virion morphogenesis. In this analysis, the downstream portions of the N-terminal region of NS2 were required for interaction (direct or indirect) with NS3. The greatest decrease in association was seen for mutations that targeted TMS2 (W35A, W36A, Y39A, L41A) and part of the TMS3 (K81A, W82A, A85L), in addition to P89A and L92A. This effect of NS2 mutations clustered within these transmembrane segments on NS3:NS2 interaction is in agreement with other reports (28, 36).

We then examined the interaction of NS3 with the E2 glycoprotein, several species of which could be coimmunoprecipitated and resolved (Fig. 10B). Previous reports have revealed the presence of multiple forms of E2, including E2-p7 and E2-p7-NS2 (50–54). Supporting the likelihood that the higher E2 species was an E2-p7-NS2 species, an NS2-reactive band running at the same size could be detected in WT lysates, and with NS2 immunoprecipitation, this E2-reactive band disappeared in the absence of NS2 (ΔNS2; data not shown). In general, mutations of NS2 mutants with reduced NS3-associated E2 were located throughout the N-terminal region (L13A, I17A, L22A, T23A, G25A, R61E, D62R), with the greatest reduction observed with mutations within TMS2 and TMS3 (W35A, W36A, Y39A, F77A, K81A, W82A, A85L, G88A) as well as P89A and L92A. Furthermore, the importance of NS2 for E2 association is highlighted by the dramatic decrease of NS3-bound E2 in its absence (Fig. 10B, ΔNS2 lane).

Interaction of NS3 with core was also investigated (Fig. 10B). Surprisingly, while complete loss of NS2 (ΔNS2) did not affect the level of associated core, NS2 mutations did. Specifically, mutations within the TMS1/first cytoplasmic loop (G25A, K27E, L30A, R32E), the TMS2/second lumenal loop (W36A, Y39A, L41A, R61E, D62R), and the latter part of the N-terminal region (K81A, W82A, A85L, G88A, P89A, L92A) resulted in partial or severely decreased levels of associated core. Interestingly, with the exception of L30A and R32E, mutants with diminished core corresponded to those with reduced NS3-associated E2 and NS2. However, maintenance of a NS3:core interaction did not always correlate with binding to the other viral factors. For example, the W35A and F77A mutants had distinctive patterns. For W35A, NS3 association with both E2 and NS2 was severely reduced, whereas the F77A mutant had loss of only E2. Collectively, these data suggest that additional factors, representing either viral (e.g., p7 or E1) or host proteins, are likely involved in the interaction between NS3, core protein, and NS2.

DISCUSSION

Analysis of the block mutations in the context of both bi- and monocistronic genomes (Fig. 1 and 2) indicates that processing at the NS2/NS3 junction and hence RNA replication are largely unaffected by mutagenesis of the N-terminal region of NS2. Since formation of an NS2 dimer is required for NS2-NS3 cleavage (21, 30, 37) in the monocistronic background, our results also imply that this region does not impact functional dimer formation. Whether dimer formation and stability are required for virus production still remains unknown. However, interactions between the N-terminal region and the protease domain appear to occur and are important in virion morphogenesis. This conclusion stems from the surprising observation that engineering an H77 epitope (two amino acid substitutions) in the J6 NS2 protease domain selectively impaired virus production for certain mutants by as much as 2 logs (L30A, R32E, and Y39A; Fig. 10A). This may point to critical intra- or intermolecular homotypic NS2 interactions or associations with host or other viral components that influence virus assembly.

Additional catalytic function(s) beyond its autoprotease activity have not been reported for NS2, and this activity is dispensable for infectious virus production (13, 25, 26). Thus, NS2 is hypothesized to regulate this process through modulating interactions between structural and nonstructural viral components. Infectious virus production is a multifaceted process involving early (nucleocapsid formation) and late (envelopment, budding) assembly events, egress/maturation, and release. Early events involving the formation of the nucleocapsid most likely occur at the ER in close association with lipid droplets (LDs) where mature core accumulates (12, 55). Reports from studies of related pestiviruses (bovine viral diarrhea virus) and flaviviruses (Kunjin and yellow fever viruses) have shown that uncleaved NS2-3 and NS2A, respectively, play important roles in assembly of infectious virus (56–58). While they have been linked by a number of genetic studies (31, 59), a biochemical interaction between HCV NS2 and core has been shown only recently in the context of a high-titer cell culture-adapted virus (60). Furthermore, live-cell imaging to monitor tagged intracellular core trafficking in studies by two groups (61, 62) demonstrated that NS2 is necessary for core migration and requires interaction with NS3. The postcleavage form of NS2 can interact with NS3 and NS5A, both of which are components of the replicase complex that bind RNA and are also implicated in virus production (9–11, 28, 36, 63–65) (Fig. 10B). Thus, NS2 may participate in RNA encapsidation through its interaction with NS3, NS5A, and possibly core.

Previous reports indicate that mutations that target the first cytoplasmic loop and TMS2 and TMS3 of NS2 (K27A, W35F, W36A, Y39A, F77A) have decreased NS3 binding (28, 36). In our study, we did observe a severe reduction in the amount of NS3-bound NS2 with mutations clustered in these segments (summarized in Table 3). These TMS mutations also influenced NS3:core binding, although the complete loss of the NS2 protein did not. While NS2 is clearly important, it is likely that we are not observing the full complement of factors involved in early assembly events. Other players include p7, which has been shown influence NS2 interaction with NS3 and E2 (36, 44) and factors affecting core distribution between the ER and LDs (59, 62). NS2 mutants Y39A, F77A, A85L, and P89A within these transmembrane segments displayed differences with respect to the release of extracellular RNA and core relative to WT results (Table 2), further supporting the hypothesis that interactions involving the downstream portion of the N-terminal region of NS2 are important for efficient nucleocapsid assembly.

Table 3.

Comparison of biochemical phenotypes with NS2 mutants^a

^aWestern blot bands were quantified by densitometry in Fiji software and then compared to the parental WT construct. Box shading represents percentage of WT as follows: black, 100% to 75%; dark gray, 75% to 50%; light gray, 50% to 25%; white, 25% to 0%. Mutants with titers < 2 logs and residing in TMSs are highlighted with italics and boldface.

NS2 may also coordinate late assembly events, i.e., the envelopment and budding of the nucleocapsid into the ER lumen, via interactions with HCV glycoproteins. Compensatory mutations that restore infectious titers of mutated NS2 or intergenotypic chimeric genomes have been mapped to E1 and E2 (9, 25, 26, 28). While we envision that NS3:NS2 interaction is important for RNA encapsidation, the ability of NS3 to coimmunoprecipitate E2 suggests either that NS3 may have additional functions in envelopment/budding or that early and late assembly events occur concomitantly rather than sequentially. Similar to other reported studies (28, 36, 44), we observed a reduction in the level of the NS3-bound E2 for TMS2 and TMS3 NS2 mutants and a second cluster of residues that span the end of TMS1 (Fig. 10B and Table 3). Moreover, native, high-molecular-weight complexes that contain all three proteins (NS2, NS3, and E2) have been shown to comigrate, and interaction between NS3 and E2 was abrogated in the absence of NS2 (36) (Fig. 10B). Due to lack of available tools, we were unable to examine whether these mutants had deficiencies in binding E1 as well. Regardless, these results strengthen the argument that NS2 is a major regulator of NS3 and glycoprotein interactions.

It was interesting that, in addition to E2 and E2-p7, significant quantities of a larger species, likely E2-p7-NS2, were immunoprecipitated with the NS3 antibody (36, 44) (Fig. 10B). This suggests that the association of key factors ultimately involved in the assembly process and in virions may occur early, even before polyprotein processing in this region is complete. TMS1 (A3L, L13A, I17A, L22A, T23A, G25A) and second lumenal loop (R61E, D62R) NS2 mutants had higher levels of NS3-associated E2-p7/E2, while D2A and first cytoplasmic loop (K27E, L30A, and R32E) mutants had more NS3-associated E2-p7-NS2. Although earlier studies using bicistronic constructs have demonstrated that E2-p7-NS2 is not absolutely necessary for virus production in cell culture (13), such constructs are nonviable in chimpanzees (18). These observations and the results from our study may indicate a role for this intermediate in achieving optimal virus assembly.

There are several caveats that make clear correlations between the virologic properties of these mutations and the coimmunoprecipitation results difficult. NS2-containing complexes are involved in multiple steps in virus assembly, with likely differences in composition and function (36). Although we utilized NS3 for capture, we are still sampling a dynamic mixture rather than a homogeneous complex associated with a single step. Even when examining NS2-specific partners, we observed that, despite using saturating antibody, there is a fraction of NS2, approximately 25% to 35%, which we were unable to isolate (data not shown). Thus, the sampling of NS2-associated complexes is incomplete. Thus far, the NS3 interaction profiles do not allow us to distinguish between NS2 mutations that hit a specific motif required for a critical protein interaction and mutations that disrupt proper NS2 folding.

We observed a wide spectrum of phenotypes with this large panel of NS2 mutants. Compared to the results of other NS2 mutagenesis studies, some mutations appear to be genotype specific. The G10A and P73A mutants have each been previously characterized as highly impaired (9, 25, 28). Yet, in our screen, titers were moderately decreased and increased, respectively. While some of these mutations may prove to be highly specific in the genotypic background tested, they nonetheless have allowed identification of several areas of interest for future studies. HCV virions are infectious soon after budding into the ER lumen; however, it is not until they pass through the secretory pathway that they mature into acid-resistant lower-density infectious particles (66–68). We found that G25A at the C-terminal end of TMS1 and K81A in TMS3 have a substantial increase in relative intracellular infectivity (Fig. 7B). The K81A mutant may additionally influence an early step in assembly, as shown by reduced levels of intracellular infectivity compared to those seen with the WT (Fig. 7A). Conversely, the G25A mutant was impaired only in egress, and the thermal stability of released particles did not differ from that seen with the WT (data not shown). A possible explanation for the impaired secretion could be the disruption of viral proteins involved in virus particle maturation, such as p7 (19, 20). However, on the basis of the results observed in the different cell backgrounds (Fig. 8C), we speculate that NS2-host interactions that are required for efficient secretion are disrupted. In support of this hypothesis, and not surprisingly, knockdown of host factors involved in the constitutive secretory pathway increases the amount of cell-associated infectivity (61). Finally, several NS2 mutants that are moderately impaired in infectious virus production still secreted and accumulated core and RNA at levels similar to or slightly elevated above WT levels (summarized in Table 2). After examining the thermal stability of particles produced from these mutants, it was concluded that their lower specific infectivity (Fig. 6D) did not result from a general instability. Rather, we hypothesize that a proportion of the particles produced are rendered noninfectious by deficiencies in postassembly modifications or incorporation of host factors necessary for infectivity, such as ApoE (47–49).

In conclusion, our report reinforces the hypothesis put forth by other groups that NS2 acts as a central player in virus assembly, promoting key interactions between viral nonstructural and structural proteins. We now have evidence that NS2 can also influence particle secretion and infectivity. NS2 may mediate these later steps by regulating critical virus-host interactions. Further mechanistic dissection of these unique mutant phenotypes is now needed to tease apart the wide complement of NS2 interactions that are required for infectious HCV production.