Correlation between NS5A dimerization and hepatitis C virus replication

Abstract

Hepatitis C virus (HCV) is the main agent of acute and chronic liver diseases leading to cirrhosis and hepatocellular carcinoma. The current standard therapy has limited efficacy and serious side effects. Thus, the development of alternate therapies is of tremendous importance. HCV NS5A (nonstructural 5A protein) is a pleiotropic protein with key roles in HCV replication and cellular signaling pathways. Here we demonstrate that NS5A dimerization occurs through Domain I (amino acids 1-240). This interaction is not mediated by nucleic acids because benzonase, RNase, and DNase treatments do not prevent NS5A-NS5A interactions. Importantly, DTT abrogates NS5A-NS5A interactions but does not affect NS5A-cyclophilin A interactions. Other reducing agents such as tris(2-carboxyethyl)phosphine and 2-mercaptoethanol also abrogate NS5A-NS5A interactions, implying that disulfide bridges may play a role in this interaction. Cyclophilin inhibitors, cyclosporine A, and alisporivir and NS5A inhibitor BMS-790052 do not block NS5A dimerization, suggesting that their antiviral effects do not involve the disruption of NS5A-NS5A interactions. Four cysteines, Cys-39, Cys-57, Cys-59, and Cys-80, are critical for dimerization. Interestingly, the four cysteines have been proposed to form a zinc-binding motif. Supporting this notion, NS5A dimerization is greatly facilitated by Zn(2+) but not by Mg(2+) or Mn(2+). Importantly, the four cysteines are vital not only for viral replication but also critical for NS5A binding to RNA, revealing a correlation between NS5A dimerization, RNA binding, and HCV replication. Altogether our data suggest that NS5A-NS5A dimerization and/or multimerization could represent a novel target for the development of HCV therapies.

FIGURE 1.

NS5A dimerization occurs directly through Domain I. A and B, GST or GST-Domain I NS5A containing the linker was used as bait to pull down full-length NS5A-His (A) or Domain I NS5A-His (B). Captured proteins were analyzed by Western blotting using anti-His and anti-GST antibodies. C, deletion mutants of Domain I NS5A-His were generated and used as prey in pulldown assays with GST or GST-Domain I NS5A as bait. Captured proteins were analyzed by Western blotting using anti-His and anti-GST antibodies. D, recombinant proteins were treated with benzonase, RNase, or DNase to remove contaminating nucleic acids before pulldown assays. GST-CypA/NS5A-His were used as controls because this interaction has been shown to be direct. Captured proteins were analyzed by Western blotting using anti-His and anti-GST antibodies.

FIGURE 2.

Cyclophilin inhibitors and NS5A inhibitors do not affect NS5A dimerization. Pulldown assays using GST or GST-Domain I NS5A as bait to capture full-length NS5A-His or Domain I NS5A-His were performed in the presence of 2 μm CsA (A), alisporivir (B), or BMS-790052 (C). Captured proteins were analyzed by Western blotting using anti-His and anti-GST antibodies.

FIGURE 3.

Reducing agents and EDTA reduce NS5A dimerization, whereas Zn²⁺, but not Mg²⁺ or Mn²⁺, increases dimerization. A, pulldown assays using GST or GST-Domain I NS5A as bait to capture full-length NS5A-His or Domain I NS5A-His were performed in the presence of DTT (5 μm). B, pulldown assays using GST or GST-CypA as bait to capture full-length NS5A-His were performed in the presence of DTT (5 μm). C and D, pulldown assays using GST or GST-Domain I NS5A as bait to capture full-length NS5A-His were performed in the presence of TCEP (C, 5 μm) and 2-mercaptoethanol (D, 50 μm). E, pulldown assays were performed using GST or GST-Domain 1 NS5A to capture full-length NS5A-His in the presence of EDTA (20 mm), EDTA (5 mm) + Zn²⁺(10 mm), EDTA (5 mm) + Mg²⁺ (10 mm), or EDTA (5 mm) + Mn²⁺ (10 mm). Captured proteins were analyzed by Western blotting using anti-His and anti-GST antibodies. In three independent experiments, we measured the relative intensity of specific bands using a GS-800 densitometer and performed statistical analysis using the Student's t test. A p value <0.05 (*) denotes a statistically significant difference.

FIGURE 4.

Cysteines at positions 39, 57, 59, and 80 of Domain I are critical for NS5A dimerization. A, full-length NS5A-His with single mutations at each cysteine residue in Domain I were generated and used as prey in pulldown assays with GST or GST-Domain I NS5A as bait. These data are representative of three independent experiments using three different batches of recombinant NS5A proteins. B, cysteine mutants in A were tested for their ability to bind cyclophilin A with or without the presence of CsA (2 μm).

FIGURE 5.

Cysteines critical for dimerization are also important for RNA-binding and replication. A, NS5A (0–2000 nm) and 0.1 nm rU15-FI were gently mixed in binding reaction buffer (20 mm HEPES, pH 7.5, 100 mm NaCI, 5 mm MgCl₂, and 0.5 mm TCEP) and incubated for 30 s at 25 °C. Binding of NS5A was measured by the change in polarization (mP). The data were fit to a hyperbola using KaleidaGraph (Synergy Software). B, full-length NS5A-His with alanine substitutions at Cys-39, Cys-57, Cys-59, and Cys-80 were tested for their ability to bind GST-NS5BΔ21 relative to wild-type NS5A. C, Huh7.5.1 cells were electroporated with luciferase reporter subgenomic RNA transcripts containing cysteine mutations in NS5A. The cells were lysed at the given time points and post-electroporation, and the replication fitness was measured using a luciferase assay. The results (triplicates) are representative of three independent experiments. D, Huh7-Con1 cells were transfected with wild-type or cysteine NS5A-His plasmids for 3 days. The cell lysates were incubated with Ni⁺ beads for 1 h, bound material was eluted, RNA was purified, and quantitative real time PCR was executed as described (18). The amount of HCV RNA precipitated by wild-type NS5A-His was arbitrary fixed at 100. The results (triplicates) are representative of two independent experiments.

FIGURE 6.

Model for Domain I-Domain I contacts displaying Cys-39, Cys-57, Cys-59, and Cys-80 required for NS5A dimerization. The model was prepared using Chem3D Pro software with the Tellinghuisen et al. (29) and Love et al. (31) crystal structures of NS5A Domain I (Protein Data Bank codes 1ZH1 and 3FQM, respectively) (29, 31). The two monomers are cyan and purple. The zinc atoms are represented as white balls. Cys-39, Cys-57, Cys-59, and Cys-80 are highlighted in yellow on both monomers with amino acid side chains displayed. The Tyr-93 residue in Domain I responsible for the NS5A inhibitor BMS 790052 resistance is highlighted in dark pink. For orientation purposes, the two N-terminal amino acids are colored green in each model: Phe-36/Phe-37 for the Tellinghuisen et al. structure and Leu-32/Gly-33 for the Love et al. structure. Three different views of the dimers are presented.