The Unique C Termini of Orthopoxvirus Gamma Interferon Binding Proteins Are Essential for Ligand Binding^{▿ †}

Cell and virus culture.

African green monkey kidney cell lines BS-C-1 (ATCC CCL-26) and CV-1 (ATCC CCL-70) were maintained in Dulbecco's modified Eagle's medium containing 50 units/ml penicillin, 50 μg/ml streptomycin, 2 mM l-glutamine, and 10% FetalClone II (HyClone Laboratories, Inc., Logan, UT) at 37°C in a 5% CO₂ atmosphere. The Western Reserve strain of VACV and a derivative virus expressing the T7 RNA polymerase were kindly provided by Bernard Moss (15). VACV-t7,b8r⁻ (hereafter VACV-t7) virus was constructed by deleting the VACV IFN-γbp (b8r) open reading frame using the gpt transient-dominant selection system (14). Virus stocks were prepared in HELA-S3 (ATCC CCL-2.2) and stored at −70°C until use. Virus infectivity was measured on BS-C-1 monolayers as described previously (10).

Plasmids and mutagenesis.

The ECTV ifn-γbp gene was cloned downstream of the T7 promoter of the pTM1 vector (15) and expressed as previously described (6). Single-base mutations were carried out using the Gene Tailor site-directed mutagenesis system (Invitrogen, Carlsbad, CA) and a two-step PCR protocol (17). All mutant ifn-γbp genes contained native IFN-γBP signal peptide, and genotypes were confirmed by sequencing.

Transfection.

We have previously expressed and purified ECTV IFN-γBP that retains its biological activity using the VACV-t7 expression system (5, 15). The expression of IFN-γBP in this system has been optimized by comparative studies that varied the multiplicity of infection, time of harvest, and chemically defined medium to a yield of approximately 0.5 μg/10⁶ CV-1 cells. In brief, CV-1 cells were grown in six-well tissue culture plates to ∼80% confluence and were infected with VACV-t7 at a multiplicity of infection of 10 PFU/cell for 1 h at 37°C. The cells were then washed in Opti-MEM and transfected with pTM1 (2 μg; vector control) alone or pTM1 containing wild-type (WT) or mutant ifn-γbp⁺ genes (2 μg) using Lipofectamine 2000 reagent in Opti-MEM according to the manufacturer's protocol (Invitrogen, Carlsbad, CA). At 24 h posttransfection, supernatant was collected, clarified, and filtered to remove virus using 0.1-μm centrifugal filters (Millipore, Bedford, MA). Transfection supernatants were assayed and quantified via Western blot densitometry using an IFN-γBP standard purified through ion exchange (Q Sepharose) and gel filtration (Superdex 200) chromatography. No further purification of the transfection supernatants was necessary prior to surface plasmon resonance (SPR) or bioassay experiments.

Surface plasmon resonance (Biacore) with mIFN-γ.

Real-time interaction of IFN-γBP with mIFN-γ was measured by SPR on a Biacore 2000 (Biacore Inc., Piscataway, NJ). For mIFN-γ, flow cells of a carboxymethylated dextran (CM5) sensor chip were activated using 50 mM N-hydroxysuccinimide and 200 mM N-ethyl-N′-(dimethylaminopropyl)carbodiimide for 7 min at a flow rate of 5 μl/min. mIFN-γ (R&D Systems, Minneapolis, MN), diluted in 10 mM sodium acetate pH 5.0, was immobilized at a flow rate of 5 μl/min for 5 min. The surface was treated using 1 M ethanolamine hydrochloride at pH 8.5 for 7 min with a flow rate of 5 μl/min to deactivate excess reactive esters and remove noncovalently bound ligand. The mIFN-γ surface was stable following repeated rounds of regeneration and remained so for several weeks. IFN-γBP supernatants from the VACV-t7 system were diluted in Opti-MEM to concentrations from 50 to 0.5 nM and were injected in random order over the mIFN-γ sensor chips. Dissociation occurred in HBS-EP running buffer (10 mM HEPES pH 7.4, 150 mM NaCl, 3 mM EDTA, 0.005% surfactant P20; Biacore Inc.) for 20 min following injection. To correct for refractive index changes, responses generated with the control (activated and blocked) surface were subtracted from the responses with the mIFN-γ surface. To correct for nonspecific binding, responses generated during the injection of pTM1 control transfection supernatant were subtracted from the data set. The doubly referenced data were analyzed and fit globally to the 1:1 binding model using BIAevaluation 4.1 software.

Human IFN-γ expression, refolding, and surface plasmon resonance (Biacore).

The C-terminal AviTag (7) sequence (Avidity, Denver, CO) was inserted at the C terminus of the hIFN-γ sequence (31) using standard techniques to generate pET21a hifn-γb. BL21(DE3) cells (Invitrogen, Carlsbad, CA) were transformed with pET21a hifn-γb and pACYC-184 containing an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible biotin protein ligase (BirA; Avidity, Denver, CO). Cells were grown in the presence of ampicillin (100 μg/ml) and chloramphenicol (10 μg/ml) to an optical density at 600 nm of 0.6 to 0.8 followed by induction with 0.1 mM IPTG plus 10 μM d-biotin for 3 h. Cells were harvested by centrifugation, and inclusion bodies were isolated using BugBuster reagent (EMD Biosciences, San Diego, CA). Inclusion bodies were solubilized in 100 mM Tris pH 8.0, 2.5 mM EDTA, 6 M guanidine HCl and refolded by drop-wise dilution into 100 mM Tris pH 8.0, 2.5 mM EDTA, 500 mM l-arginine. Refolded protein was dialyzed into 20 mM Tris pH 8.0, 100 mM urea prior to ion-exchange purification on S-Sepharose (Sigma, St. Louis, MO). Gel filtration with Superdex 75 (GE Biosciences, Piscataway, NJ) was used to enrich for properly folded hIFN-γB homodimers and exclude unfolded monomers and protein aggregates. Streptavidin (SA)-coated sensor chips (Biacore, Piscataway, NJ) were used to capture refolded hIFN-γB. The IFN-γBP samples were injected over the hIFN-γB surface for 2 min at a flow rate of 5 μl/min in HBS-EP buffer. To correct for refractive index changes, responses generated with the control surface (SA blocked with 10 μM d-biotin) were subtracted from the responses of the hIFN-γB surfaces. hIFN-γB surfaces were regenerated using a 1-min injection of 100 mM glycine pH 2.8.

Interferon protection (VSV) assay.

ECTV IFN-γBP and mutants were tested for their ability to neutralize the antiviral activity of mIFN-γ on L929 cells. WT and mutant IFN-γBPs in virus-free (0.1-μm-filtered) culture medium from the VACV-t7 transfection system were serially diluted in Opti-MEM (Invitrogen, Carlsbad, CA). Recombinant mIFN-γ (R&D Systems, Minneapolis, MN) was added to each diluted sample to a final concentration of 25 U/ml. Mixtures were incubated at 37°C for 1 h. Incubated mixtures (100 μl/sample) were transferred to 96-well plates previously seeded with 2 × 10⁴ L929 cells/well. After 24 h, cells were challenged for 2 days with ∼50 PFU of vesicular stomatitis virus (VSV) followed by staining with crystal violet solution.

Minimal truncation of IFN-γBP results in loss of ligand binding activity.

The orthopoxvirus IFN-γBPs have ∼20% identity to the extracellular portion of the host IFN-γ receptors; however, this homology is restricted to the first ∼200 amino acids of the IFN-γBPs. The remaining C-terminal domain of the IFN-γBPs is not homologous to the transmembrane or signaling domains of the receptors but is very well conserved within the orthopoxvirus IFN-γBP family. We previously identified C216 in the ECTV sequence as being important for covalent dimerization of the secreted protein; however, the function of the remaining 60 amino acids still remained unclear. To investigate the functional importance of this region, we constructed a series of truncation mutants in both the ECTV and VACV sequences, beginning at the C terminus and progressing towards the N terminus (Fig. 1). The truncated IFN-γBPs were expressed in the VACV-t7 system as described previously (6) and were expressed at levels comparable to wild-type IFN-γBP (see Fig. S1 in the supplemental material). Expressed proteins were screened for both mIFN-γ (Fig. 2) and hIFN-γ (Fig. 3) binding activity by surface plasmon resonance. Examination of the sensorgrams demonstrated that truncation of the ECTV sequence by two amino acids (Δ2) was well tolerated and this protein retained its ability to bind both mIFN-γ and hIFN-γ, while truncation by an additional amino acid (Δ3) eliminated the ability to bind IFN-γ of either species. Further truncations up to residue 211 resulted in a similar loss of activity (data not shown). Equivalent truncation of the VACV IFN-γBP (identified in Fig. 1) demonstrated a similar loss of binding to hIFN-γ. No binding of VACV IFN-γBP to mIFN-γ could be detected in these experiments.

Mutagenesis scanning of the terminal nine amino acids.

Following the observation that no IFN-γ binding activity was detectable in constructs truncated by more than two amino acids, we constructed a series of alanine scanning mutants from positions 258 to 266 (ECTV) to examine the potential importance of each of the nine C-terminal amino acid side chains on the activity of the protein. Each mutant was expressed, and the kinetics of binding to mIFN-γ were analyzed using SPR (Fig. 4). Examination of the resultant kinetics profiles revealed that no single residue in the C terminus appeared to be responsible for conservation of activity. Alanine substitution at any position resulted in a similar overall affinity (K_D); however, in all cases there was a shift in the rate constants to both a faster on-rate (k_a) and equivalently faster off-rate (k_d).

Using the PredictProtein algorithm (24), we were able to predict that the C terminus of the IFN-γBP likely adopts an α-helical secondary structure once folded. To test the importance of this local secondary structure on the function of ECTV IFN-γBP, we inserted a proline (F261P) into the middle of this region to disrupt the secondary structure. This alteration resulted in a complete loss of mIFN-γ binding activity as measured by SPR, suggesting an importance of C-terminal secondary structure on the activity of the protein. Conversion of all of the last nine amino acids to alanines (9xAla) also resulted in no mIFN-γ binding activity, even though this sequence is predicted to share a similar α-helical secondary structure (26). Similarly, randomization of the nine terminal amino acids (MLLSTITDF, LTSTLIDFM) resulted in no binding activity, despite a similar predicted secondary structure and secretion into the culture medium (data not shown). To test the effects of extending the C terminus, we constructed ECTV IFN-γBPs containing the naturally occurring VACV C-terminal extension (ECTV+VACV3′) as well as a C-terminal hemagglutinin (HA) tag. The shorter VACV extension (five residues) was not only tolerated but resulted in a slightly slower off-rate than for wild-type ECTV IFN-γBP. Extension with the C-terminal HA tag (nine residues) was not tolerated and resulted in a complete loss of detectable activity. The mIFN-γ binding sensorgrams and kinetics for each construct are presented in Fig. 4 and Table 1, respectively.

The C terminus and secondary structure of the IFN-γBP are necessary to neutralize the antiviral effects of IFN-γ.

To demonstrate that the effect of IFN-γBP truncation on activity was not confined to detection by SPR, we examined the effect of IFN-γBP truncation on its ability to inhibit mIFN-γ in a bioassay (Fig. 5). The murine fibroblast cell line L929 is responsive to the antiviral effects of interferons. Pretreatment of L929 monolayers with mIFN-γ induces a cellular antiviral state, and the cells are no longer permissive for VSV infection. Inclusion of IFN-γBP in the pretreatment conditions results in sequestration of mIFN-γ and prevents induction of the antiviral state. VSV can then successfully infect cells in the monolayer, and the extent of this infection can be monitored by staining with crystal violet. To examine the role of the C-terminal amino acids in inhibiting the antiviral effects of IFN-γ, equal amounts of WT and mutant ECTV IFN-γBPs were tested for their ability to inhibit the biological activity of mIFN-γ. All of the alanine scanning mutants tested were able to block the antiviral effects of mIFN-γ on L929 cells, consistent with their ability to bind IFN-γ with high affinity in SPR experiments. Neutralizing activity could not be detected from the truncated (Δ8) or F261P constructs, both of which failed to bind mIFN-γ in SPR experiments. Additional experiments demonstrated mIFN-γ inhibition from the Δ2 mutant identical to WT IFN-γBP. No inhibition of mIFN-γ activity could be detected from the Δ3 mutant, consistent with its failure to bind IFN-γ in SPR experiments (data not shown).

Active IFN-γBP forms higher-ordered structures.

It is clear that truncation of the ECTV IFN-γBP results in a loss of activity in both SPR and a VSV assay. Truncation mutants are expressed at levels similar to wild type, although only ∼40% of the inactive truncation constructs (Δ3 and Δ8) form disulfide-linked dimers by nonreducing SDS-PAGE (see Fig. S1 in the supplemental material). To address the possibility that truncation results in misfolding of the IFN-γBP, we performed gel filtration analyses on supernatants from the VACV-t7 system (Fig. 6). All IFN-γBP constructs that demonstrated activity by SPR and the VSV assay eluted in the range of 163 to 382 kDa with a median of 250 kDa, indicating that the active protein assembles into higher-ordered complexes. Those mutations that resulted in no detectable IFN-γ binding activity eluted in the range of 52 to 122 kDa with a median of 80 kDa, corresponding to the presence of covalently linked dimers. The notable exception to this trend was with VACV IFN-γBP, whose lack of mIFN-γ binding is likely the result of amino acid differences between it and the ECTV IFN-γBP and not the result of differences in the CTD. Assembly of the higher-ordered complexes appears to be independent of the intrachain disulfide bond, as the C216A mutant migrated identically to the WT protein and pretreatment of the IFN-γBP with dithiothreitol (DTT) failed to alter the mobility in the presence of the full-length CTD. Pretreatment of the Δ3 mutant with DTT resulted in a migration pattern consistent with a monomeric form of the IFN-γBP, consistent with failure of the truncated CTD to facilitate oligomerization beyond the covalent homodimer. Gel filtration analysis of purified IFN-γBP demonstrated the same fractionation pattern as the unpurified material from the VACV-t7 system and a single band could be identified by silver staining, indicating that the complexes are assembled from the IFN-γBP alone (see Fig. S2 in the supplemental material).