The NS5A Protein of Bovine Viral Diarrhea Virus Contains an Essential Zinc-Binding Site Similar to That of the Hepatitis C Virus NS5A Protein

Cloning and in vitro mutagenesis.The EcoRI fragment of the BVDV Jiv⁻ replicon (pACNR/NADL Jiv-S-pac) (31), containing the NS5A coding region, was subcloned into the pBluescriptSK+ vector to generate the vector pBluescriptSK+ EcoRI. This vector was then used as a template for in vitro mutagenesis with the QuikChange mutagenesis kit in accordance with the manufacturer's (Stratagene) instructions. All mutations were confirmed by DNA sequence analysis. Introduced amino acid changes are referred to via the single-letter amino acid code of the parental sequence, the amino acid numbering of BVDV NS5A, and the single-letter amino acid code of the mutation; for example, C34G corresponds to a mutation of the cysteine residue at amino acid 34 of BVDV NS5A to a glycine. Conversion of the BVDV NS5A numbers used in this report to the amino acid numbering scheme of the BVDV strain NADL polyprotein can be accomplished by addition of 2773 to the amino acid positions in this report. Once generated, mutations were cloned back into the BVDV Jiv⁻ replicon (pACNR/NADL Jiv-S-pac) and the clones were confirmed by restriction endonuclease mapping and DNA sequence analysis. Clone 5′pac/NADL has been described previously (40). Generation of BVDV protein expression vector pET30ubiBVDV NS5A-Δh was performed by PCR amplification of the DNA fragment of pACNR/NADL Jiv-S-pac encoding amino acids 25 to 496 of the mature BVDV NS5A protein with primers that introduced SacII and KpnI restriction enzyme cleavage sites for direct cloning into previously described pET30ubiHCV NS5A-Δh, from which the HCV coding sequence had been removed (50). This expression vector generates an amino-terminal ubiquitin fusion to BVDV NS5A that is removed during expression in bacteria by coexpression of a yeast ubiquitin cleavage activity. The vector also adds a flexible carboxy-terminal linker of 32 amino acids and a hexahistidine tag (contained in the commercial pET30b vector). Mutations of the BVDV NS5A zinc-binding site residues were amplified from the pBluescriptSK+ EcoRI vector bearing these mutations and cloned into the expression vector by the same strategy. All expression clones were confirmed by DNA sequence analysis.

In vitro RNA transcription.Plasmid DNAs were linearized with SbfI (New England Biolabs), followed by removal of the 3′ overhang with excess deoxynucleoside triphosphates, Klenow, and T4 DNA polymerase and nucleotide removal. RNA transcripts were synthesized at 37°C for 3 h in a 100-μl reaction volume containing 40 mM Tris-HCl (pH 7.9), 10 mM NaCl, 12 mM MgCl₂, 2 mM spermidine, 10 mM dithiothreitol, 3 mM nucleoside triphosphate, 100 U of Superasin (Roche), 100 U of T7 RNA polymerase (Epicenter Technologies), and 5 to 10 μg of linearized DNA. At the end of the reaction, 10 U of DNase I (Roche) was added and reaction mixtures were incubated at 37°C for 20 min to remove the template DNA. DNase-treated RNAs were then purified with an RNA Easy kit (QIAGEN). RNAs were eluted in 50 μl of RNase-free distilled water. The integrity of the purified RNAs was confirmed by agarose gel electrophoresis, and RNA concentrations were determined by spectrophotometric analysis.

Cell culture.Madin-Darby bovine kidney (MDBK) epithelial cells were obtained from the American Type Culture Collection. Cells were used between passages 20 and 60. Cell monolayers were propagated in Dulbecco's modified Eagle medium (DMEM; Invitrogen) supplemented with penicillin (100 U/ml), streptomycin (100 μg/ml), and 10% heat-inactivated horse serum (HS). Cells were maintained at 37°C with 5% CO₂. Cells were passaged after treatment with 0.05% trypsin-0.02% EDTA.

Electroporation of BVDV replicons and selection.MDBK cells were trypsinized, collected, and washed three times in ice-cold, RNase-free phosphate-buffered saline (PBS). Cells were resuspended in PBS at 2.0 × 10⁷/ml. One microgram of the in vitro-transcribed RNA was mixed with 0.4 ml of the MDBK cell suspension, transferred to a 2-mm-gap cuvette (BTX), and pulsed with a BTX ElectroSquarePorator (0.99 kV; five pulses; 99-μs pulse duration, 1.1-s intervals). After 10 min at room temperature, the cells were transferred to 3.6 ml of DMEM-10% HS. To measure the efficiency of puromycin-resistant colony formation, the transfected cells were plated at a series of densities (10⁶, 10⁵, and 10⁴ cells). To maintain the total number of plated cells at 2 × 10⁶/100-mm-diameter dish, untransfected cells were used as a feeder layer. After 72 h, the medium was replaced with DMEM-10% HS supplemented with puromycin (5 μg/ml; Gibco Life Technologies). The medium was replaced after another 4 days. After 8 to 10 days, puromycin-resistant colonies were stained with 1% crystal violet in 50% ethanol. Colony counts from triplicate experiments were used to calculate the number of CFU per microgram of input RNA.

Immunofluorescence microscopy analysis.MDBK cells cultured on two-well culture slides (Becton Dickinson, Franklin Lakes, NJ) were washed once in PBS and fixed and permeabilized with an ice-cold methanol-acetate solution for 20 min at room temperature as described previously (1). After being washed three times in PBS, cells were incubated with anti-BVDV NS3 monoclonal antibody 184 (19) diluted 1:500 in PBS and 0.5% goat serum for 1 h at room temperature. Following three additional washes in PBS, an Alexa 594-conjugated goat anti-mouse immunoglobulin G secondary antibody (Molecular Probes, Eugene, Oreg.) diluted 1:1,000 in PBS and 0.5% goat serum was added to the cells and the mixture was incubated for 30 min at 4°C. Hoechst 333421 (1 μg/ml) was added, and cells were incubated for an additional 10 min at 4°C. Cells were washed three times in PBS, and coverslips were mounted. Slides were viewed on a Nikon Eclipse TE 300 microscope.

Blocking of BVDV replication with a polymerase inhibitor.The BVDV polymerase inhibitor VP32947 has been described previously and was kindly provided by Marc Collett (ViroPharma, Exton, PA) (13). Electroporated cells were seeded for immunofluorescence assay (see above), and cells were left untreated or treated with VP32947 at a final concentration of 250 nM at the time of seeding.

BVDV NS5A-Δh expression and purification.Expression and purification of BVDV NS5A-Δh were performed as described previously for HCV NS5A-Δh, with minor modifications (50). Briefly, 1 liter of LB medium supplemented with 30 μg/liter kanamycin and 25 μg/liter chloramphenicol was inoculated from an overnight culture of Escherichia coli BL21(DE3) containing pET30ubiBVDV NS5A-Δh and the ubiquitin-specific protease pCG1 plasmids such that the cell optical density at 600 nm was approximately 0.05. Cells were grown at 37°C and 250 rpm until an optical density at 600 nm of 0.6 was reached. Cells were then chilled at 4°C for 30 min, induced with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG), and incubated for 5 h at 20°C and 250 rpm. Cells were then collected by centrifugation at 6,000 × g for 10 min and resuspended in 20 ml of buffer A (100 mM Tris-HCl [pH 8.0], 150 mM NaCl, 50 mM imidazole, 10 mM 2-mercaptoethanol) per liter of the original culture volume. Cells were lysed by three passes through a cold Avestin air emulsifier at 15,000 lb/in². Following lysis, cell extracts were clarified at 25,000 × g for 30 min at 4°C. Clarified extracts were then loaded on a 5-ml bed volume HiTrap IMAC column (Amersham/Pharmacia) charged with nickel and equilibrated with buffer A at a flow rate of 2.5 ml/min. Following extensive washing with buffer A and buffer A supplemented with 1 M NaCl, NS5A-Δh was eluted with a 25-ml linear gradient of 0 to 500 mM imidazole. Fractions containing NS5A-Δh were pooled and exchanged into buffer B (25 mM HEPES [pH 7.4], 25 mM NaCl) with a HiPrep 26/10 desalting column at a flow rate of 8 ml/min (Amersham/Pharmacia). Desalted protein was then concentrated via Amicon ultracentrifugal concentrators (Millipore). The concentration of partially purified NS5A-Δh was determined by the Bio-Rad protein assay (Bio-Rad). Protein yields were typically between 1 and 3 mg/liter of bacterial culture. Mutant proteins were of lower solubility in the initial cell lysates, and therefore more input bacterial lysate was used to produce the mutant proteins in final yields comparable to that of wild-type NS5A-Δh. Purity was estimated to be approximately 80% on the basis of visual analysis of Coomassie R-250 staining after sodium dodecyl sulfate-polyacrylamide gel electrophoresis. A sample of BVDV NS5A prepared by this method is shown in Fig. 1C.

Atomic absorption analysis.Atomic absorption experiments were performed by the Trace Metal Core Facility, Department of Environmental Sciences, Mailman School of Public Health, Columbia University, New York, NY. All experiments were performed according to the standard operating procedures of the core facility. Measurement of the background zinc level of the protein analysis buffer was performed, and this value was subtracted from all values determined for protein samples. All of the values reported are averages of three independent measurements of two independently prepared aliquots of each sample. Zinc values determined were based on comparison with a zinc standard calibration curve. All of the glassware and plasticware used for the preparation of BVDV NS5A-Δh was washed overnight in 50% nitric acid and then washed with copious amounts of ultrapure deionized water. All buffers and reagents were prepared with the highest-quality grade of chemicals commercially available. The concentrations of BVDV NS5A-Δh used for calculation of protein-to-zinc molar ratios were determined by Bio-Rad protein assay.

Identification of an HCV NS5A-like zinc-binding site in BVDV NS5A.In our previous work on HCV NS5A, we attempted to identify the most conserved and, consequently, functionally most important amino acid residues in the NS5A protein. We performed comparative sequence analysis of the NS5A proteins of a number of HCV isolates, as well as the closely related GB viruses and two members of the Pestivirus genus (50). This analysis showed that the sequence conservation among these diverse NS5A sequences was limited to approximately 90 amino acid residues proximal to the N terminus. A striking feature of these alignments was the presence of four conserved cysteine residues centered around a cysteine-proline-cysteine tripeptide sequence. The presence of four conserved cysteine residues in close proximity in an intracellular protein is often associated with a structural tetrahedral metal (zinc)-binding site similar to that identified for HCV NS5A. On the basis of the results of this analysis, we proposed that BVDV NS5A might contain a zinc-binding site composed of residues C34, C57, C59, and C84. The relative positions of these residues in the BVDV and HCV NS5A sequences are indicated in Fig. 1A. We extended this analysis to include a number of diverse pestivirus NS5A sequences, all of which contain conserved cysteine residues at these positions fitting the general template Cx²²CxCx²⁴C (Fig. 1B). The four cysteines previously shown to be responsible for zinc coordination in HCV NS5A are shown for comparison. Although direct interaction with zinc was only demonstrated for the Con1 1b isolate of HCV, sequences of the infectious clone 1a H77 strain and the related virus GB virus B are also shown to highlight the conservation of this sequence element in the hepaciviruses and related viruses. In our previous analyses, the large number of cysteine residues (approximately 14, depending on the HCV isolate) in the HCV NS5A protein required an extensive mapping study to identify residues responsible for metal ion binding. In the case of BVDV NS5A, the case is much simpler, with only six cysteines present in the entire protein, four of which align well with the previously validated HCV NS5A zinc-binding site. As this comparative sequence analysis suggested that the BVDV NS5A protein coordinated a single zinc atom via contacts with four cysteine residues arranged similarly to those of the HCV NS5A protein, we decided to investigate the zinc-binding properties of BVDV NS5A and determine if metal ion coordination is also important for RNA replication in the pestiviruses.

Analysis of the zinc content of BVDV NS5A-Δh and zinc-binding site mapping.A system for the generation of milligram quantities of partially purified BVDV NS5A protein from E. coli was generated on the basis of our previous work with HCV NS5A and the work originally described by Huang et al. (26, 50). In this system, BVDV NS5A from the NADL isolate lacking the first 24 amino acids (NS5A-Δh), corresponding to the bulk of the hydrophobic membrane anchor, was produced as an N-terminal ubiquitin fusion with a C-terminal flexible linker and a polyhistidine tag to facilitate purification. Expression of NS5A in bacterial cells harboring an expression vector for a ubiquitin-specific protease produced soluble NS5A-Δh with the N-terminal ubiquitin tag removed in vivo. With this expression system and metal chelate chromatography, 1 to 3 mg of BVDV NS5A-Δh could be produced per liter of bacterial culture at an approximate purity of 80%. With a system to produce BVDV NS5A-Δh in hand, the zinc content of the protein and the residues responsible for zinc coordination could be determined.

By in vitro mutagenesis, the four candidate cysteine residues believed to constitute the BVDV NS5A zinc-binding site (cysteines 34, 57, 59, and 84) were mutated to glycine residues. Each mutation was generated by altering at least two nucleotide changes to limit the potential of same-site reversion of the introduced mutations in future replicon experiments (see below). The cysteine-to-glycine mutations were subcloned into the BVDV NS5A expression system, and the proteins were expressed and partially purified as described for the wild-type BVDV NS5A-Δh protein. Each of the zinc-binding site mutations generated NS5A-Δh that was of lower solubility in the bacterial lysates, suggesting that these point mutations might affect proper protein folding. Consequently, larger volumes of initial cell lysates were required to generate quantities of the mutant proteins sufficient for subsequent analyses.

The availability of relatively large quantities of BVDV NS5A-Δh protein made atomic absorption spectroscopy the method of choice for evaluating NS5A zinc content. The zinc contents of three independent preparations of wild-type NS5A-Δh and NS5A-Δh bearing each of the four cysteine mutations were determined by triplicate measurements of each sample. The zinc content of each sample (495 μg of total protein per measurement) was determined and, following subtraction of background metal contamination from a buffer-only blank, values were converted to a ratio of moles of zinc to moles of protein. Typical raw zinc content data were quite consistent, with variations of less than 1.5% between independent preparations of the same protein. As an example, for wild-type BVDV strain NADL NS5A-Δh, the raw zinc contents of approximately 495 μg of protein were 661, 664, and 664 μg/liter for preparation 1; 654, 658, and 657 μg/liter for preparation 2; and 654, 657, and 664 μg/liter for preparation 3. The ratios of moles of zinc to moles of protein determined for wild-type NS5A-Δh, NS5A-Δh C34G, NS5A-Δh 57G, NS5A-Δh C59G, and NS5A-Δh C84G are presented in Table 1. The values shown were calculated from the average zinc content from the nine measurements collected for each protein. The ratio of moles of zinc to moles of protein determined was 0.802 for wild-type NS5A-Δh, indicating that this protein coordinates one zinc atom per protein molecule. The value measured for this protein, 0.802, was close to the ideal value of 1 mol of zinc per mol of protein, given the purity of the NS5A-Δh preparations (approximately 80%; Fig. 1C.). The percentage of metal-binding-incompetent protein present in our preparations is unknown, with the potential for improperly folded, aggregated, or otherwise metal-coordinating-incompetent protein potentially present and leading to a reduction in the ratio of moles of zinc to moles of protein. Nonetheless, the value determined is in good agreement with our previous results obtained with HCV NS5A and strongly suggests that BVDV coordinates a single zinc atom. The four mutant NS5A-Δh proteins, bearing point mutations of the four putative metal-coordinating cysteine residues, had levels of zinc coordination at the level of the background buffer blank (typically undetectable levels, although the C84 mutant had levels of zinc approximately 10 μg/liter above the background), indicating that the observed metal ion coordination of NS5A-Δh was dependent on cysteine residues 34, 57, 59, and 84. Collectively, these data indicate that BVDV NS5A is a zinc metalloprotein coordinating a single metal ion per protein molecule.

Zinc-binding site mutations abolish BVDV RNA replication.Our atomic absorption results demonstrated the coordination of a single zinc atom by the BVDV NS5A protein. Furthermore, mutagenesis of each the four cysteine residues of the predicted zinc-binding site residues generated mutant NS5A proteins that were incapable of zinc coordination. Although the data convincingly demonstrated zinc binding, it was not clear if this metal ion coordination was required for the activity of NS5A in BVDV RNA replication. To assess the importance of zinc binding in BVDV RNA replication, mutations of the NS5A zinc coordination residues (cysteines 34, 57, 59, and 84 changed to glycine) were subcloned into noncytopathic BVDV subgenomic replicon pACNR/NADL Jiv-S-pac. This replicon was then introduced into MDBK cells, and following drug selection for the expression of a puromycin reporter gene, the replication efficiency of the input RNAs was determined and expressed as the number of puromycin-resistant CFU per microgram of input RNA (Table 2). The parental NADL Jiv⁻ replicon RNA had a transduction efficiency of 5.66 × 10⁴ CFU/μg, a value typical for this replicon in MDBK cells. Replicon RNAs bearing mutations of any one of the four cysteines responsible for NS5A zinc-binding activity failed to produce puromycin-resistant colonies. Attempts to select revertant or pseudorevertant RNA replicons restoring replication were not successful (data not shown). Since mutations in noncytopathic BVDV RNA and replicons that render these RNAs cytopathic to cells have been described previously (29), RNA replication of these cysteine mutation bearing RNA replicons was examined in transient assays by immunofluorescent (IF) staining of transfected cells with an anti-NS3 antibody. As shown in Fig. 2, the lack of an IF signal for NS3 indicated that no RNA replication occurred in the mutant RNAs, data that were comparable to the introduction of wild-type RNAs in the presence of a BVDV polymerase inhibitor. A cytopathic BVDV subgenomic replicon (5′pac/NADL) that replicates efficiently in MDBK cells and produces a strong NS3 IF signal was used as a positive control in these experiments. These data indicate that the four cysteine residues of the BVDV NS5A zinc-binding site are required for RNA replication.