|Home | About | Journals | Submit | Contact Us | Français|
We developed a rapid method to generate recombinant vaccinia viruses (rVVs) based upon a bicistronic cassette encoding the gene for green fluorescent protein (GFP) and a foreign gene of interest separated by an internal ribosome entry site (IRES). As proof-of-concept, we inserted a mutant env gene of human immunodeficiency virus (HIV) into the cassette, which was cloned into the vaccinia virus (VV) insertion vector pSC59 under the control of the early-late VV synthetic promoter and flanked by disrupted tk gene sequences. To generate rVVs, 293T cells were inoculated with wild-type (wt) VV, followed by transfection of the modified pSC59 vector containing the bicistronic cassette, which allows expression of GFP and the protein of interest. Next, GFP-positive cells were isolated by flow cytometry or by picking under a fluorescent microscope. Thymidine kinase–deficient (Tk−) 143B cells were then exposed to lysates of GFP-positive 293T cells and cultured in the presence of bromodeoxyuridine. This selection allows only Tk− rVV to remain viable. We demonstrated the success of this GFP selection strategy by expressing high levels of mutant HIV Env. Our approach shortens the time needed to generate rVVs and represents a practical approach to generate recombinant proteins.
Protein expression via recombinant vaccinia viruses (rVVs) is widely used to express foreign genes in mammalian cells, where posttranslational modifications, such as glycosylation, can occur and lead to correct folding of foreign proteins (Moss and Earl, 2001; Bleckwenn et al., 2003). The rVV technology has been in use since 1984 (Mackett et al., 1984) and has proven to be a reliable approach to express foreign proteins (Moss, 1996; Terajima et al., 2002). Various rVV strains have also been employed as live vaccines in animal model studies (Hu et al., 1992; Doria-Rose et al., 2003; Edghill-Smith et al., 2005; Zhan et al., 2005; Li et al., 2008).
Important aspects involving the usage of rVVs include generating the viral clones and maintaining the purity of viral stocks derived from them (Mackett et al., 1984). The standard experimental protocols to generate and purify rVVs are labor intensive and time consuming (Lorenzo et al., 2004). Further, the yield of recombinant gene products expressed under the control of conventional vaccinia virus (VV) promoters is relatively low (Mackett et al., 1984; Davison and Moss, 1989a, 1989b). To shorten the time required to generate and clone rVVs and to increase the level of expression of the gene of interest, we employed a bicistronic cassette under the control of a single promoter. This strategy is based on the usage of the internal ribosomal entry site (IRES) sequence that permits both the gene of interest and the selection marker to be translated from a single bicistronic mRNA (Borman et al., 1997; Gaines and Wojchowski, 1999).
In our study, we took advantage of green fluorescence protein (GFP) as a selection marker to isolate and purify rVVs. As a test gene of interest, we chose a mutant version of env derived originally from a primary strain of human immunodeficiency virus (HIV) clade C, termed envC (Song et al., 2006). The envC and GFP genes were separated by IRES and positioned under the control of the VV early-late synthetic promoter in the VV insertion vector pSC59 (Chakrabarti et al., 1997). This new approach allowed us to generate the desired rVVs in less than 10 days. The newly created rVV was able to express the foreign gene of interest at high levels.
The insertion vector pCS59 (Chakrabarti et al., 1997) was kindly provided by Dr. B. Moss (NIAID, NIH, Bethesda, MD). An expression vector containing IRES and the GFP gene, pIRES2-EGFP, was purchased from Clontech Laboratories (Mountain View, CA). A recombinant plasmid encoding a mutant HIV envC was created by subcloning SHIV-1157ip env (Song et al., 2006) into vector pJW4303 and subsequent site-directed mutagenesis to remove the variable loops V1–V3; the plasmid was propagated in Escherichia coli strain DH5α (Invitrogen, Carlsbad, CA).
The thymidine kinase–deficient (Tk−) human osteosarcoma cell line 143B Tk− (ATCC, Manassas, VA) and the human embryonic kidney cell line 293T (ATCC CRL-11268) were propagated in DMEM supplemented with 10% fetal calf serum (FCS) (Sigma, St. Louis, MO), 2mM of glutamine (Invitrogen), and 100units/mL penicillin and 100μg/mL streptomycin mix (Invitrogen) at 37°C in the presence of 5% CO2.
Wild-type (wt) VV (strain WR) was obtained from Dr. B. Moss (NIAID, NIH). The control rVVs expressing GFP (rVV–GFP) was generated by us (Sergei Popov, unpublished results). Crude stocks of both strains were prepared on 143B Tk− cells.
To obtain the sequence of the mutated envC, we designed primers flanking the env gene of HIV-1. To clone env into the IRES-containing vector, we added tails containing an XhoI site to the sense primer, while the antisense primer incorporated a BamHI site (sense-primer: 5′ GGGGCTCGAGATGCGCGTGAAGGAGAAGTA 3′; antisense-primer: 5′ GCGCGGATCCTCACAGCAGGATGCGCTC 3′). Thermostable high fidelity DNA polymerase and dNTPs were obtained from Roche Diagnostics (Mannheim, Germany). Polymerase chain reaction (PCR) was carried out in 50μL thin-wall polypropylene tubes (Axygen Scientific, Union City, CA) using a thermocycler (PerkinElmer, Shelton, CT) under the following conditions: denaturation at 95°C for 5min followed by 30 cycles of denaturation at 95°C for 45s, annealing at 52°C for 45s, elongation at 72°C for 2min, and a final extension at 72°C for 7min. PCR products were visualized in 1% agarose gel with ethidium bromide staining, cut out, and purified using a Qiagen gel extraction kit (Qiagen, Valencia, CA). Next, the fragment was digested with XhoI and BamHI, purified in a 1% agarose gel, and cloned into vector pIRES2-EGFP, which had been linearized with XhoI and BamHI (New England Biolabs, Ipswich, MA). The resulting construct was used as a template to obtain the expression cassette GFP–IRES–envC by PCR. To amplify this cassette, the primer containing the XhoI site described above, and the primer with the EcoRI site (5′ CCCCGAATTCTTACTTGTACAGCTCGTC 3′) were used under the same PCR conditions. The cassette was cloned into the insertion vector pSC59 (Fig. 1A) giving rise to the final construct, termed pSC–GFP–envC, and used to generate rVVs.
Transfection experiments were set up in six-well plates (Becton Dickinson Labware, Franklin Lakes, NJ) using subconfluent monolayers of 293T cells cultured for 12h. First, cells were infected with 3μL of a crude stock of wt VV (strain WR) (107 plaque-forming units [pfu]/mL). Infected cells were incubated for 2h at 37°C in a 5% CO2 incubator. Then, cells were transfected by standard calcium phosphate precipitation with the pSC–GFP–envC construct (5μg of DNA/250μL of transfection mix), incubated for 6h, and washed with DMEM with 5% of FCS, and incubated for another 12h at 37°C in the presence of 5% CO2. As negative control, 293T cells infected with wt VV were transfected with plasmid pSC59. When GFP-positive cells appeared, cells were sorted by flow cytometry as described (Popov et al., 2005).
To generate large numbers of cells (109 cells) expressing the foreign gene of interest, HIV envC, we inoculated 143B Tk− cells in 800mL tissue culture flasks (Becton Dickinson Labware) with crude stock of rVV–envC at a high multiplicity of infection (MOI; 10pfu/cell; 50μL/flask); 10 such flasks were used. The cells were maintained in DMEM supplemented with 10% FCS and bromodeoxyuridine (BrdU; 50μg/mL) to suppress replication of wt VV and monitored for infection under a fluorescent microscope. When 75% of the infected cells expressed GFP, cells were harvested and the pellet was stored at −80°C.
To examine the expression of HIV clade C Env (HIV EnvC) in infected cells, 105 cells were resuspended in 100μL of phosphate-buffered saline (PBS) mixed with 2× Laemmli buffer (BioRad Laboratories, Hercules, CA), boiled for 4min, and mildly sonicated to degrade chromosomal DNA. A 50μL aliquot was subjected to gradient PAGE (4–12%) (PerkinElmer, Woodbridge, Ontario, Canada) and transferred to a PVDF membrane (BioRad Laboratories). To detect EnvC, polyclonal IgG from serum of an HIV-positive individual was used (1:2000) followed by staining with anti-human IgG labeled with horseradish peroxidase (1:2500). Specific bands were detected by chemoluminescence (PerkinElmer Life Sciences, Boston, MA).
To isolate HIV EnvC, frozen pellets of 109 infected 143B Tk− cells were resuspended in hypotonic buffer (10mM Tris pH 8.0; 15mM NaCl; 0.5mg/L Leupeptin/0.7mg/L Pepstatin/0.01% [w/v] aprotinin; 0.02% sodium azide) and homogenized with a Dounce homogenizer. Cell membranes pelleted by centrifugation were resuspended in lysis buffer (20mM Tris pH 7.4; 500mM NaCl; 0.5mg/L Leupeptin/0.7mg/L Pepstatin/0.01% [w/v] aprotinin; 0.02% sodium azide; 1.0% [v/v] Empigen) and sonicated on ice. Clarified cell lysate was adjusted to 0.25% Empigen and fractionated by lentil lectin affinity and size exclusion chromatography as described (Earl et al., 1994). Postchromatography fractions were subjected to Western blot analysis; positive fractions were combined and dialyzed against PBS. The EnvC concentration was determined using a Bradford assay kit (BioRad Laboratories).
Conventional methods to generate rVVs generally require at least 3 weeks for obtaining a pure stock of rVVs, as several plaque purification steps are involved (Lorenzo et al., 2004). We devised a novel experimental strategy that significantly streamlines the procedure and contains practical short cuts (Fig. 1B). Our key reagent is a new insertion vector, a derivate of plasmid pSC59, which was modified to include a bicistronic cassette encoding eGFP and a foreign gene of interest separated by an IRES. This cassette was placed under the control of the early-late VV synthetic promoter (Fig. 1A). The final construct was designated as pSC–GFP–envC.
The experimental steps in our new strategy include (1) generating the initial rVV stock after transfection of the insertion vector into cells previously inoculated with wt VV; (2) visual monitoring of infected cultures by fluorescent microscopy to assess the level of rVV infection and to determine the optimal time to harvest cells; and (3) isolation of the foreign protein. Our experimental system allowed efficient expression of correctly processed recombinant polypeptides at a relatively large scale.
When pSC–GFP–envC was transfected into 293T cells that had been previously exposed to wt VV, spots of brightly green cells appeared within 12h (Fig. 2A, B). This result implied that VV RNA polymerase driven by the synthetic promoter was able to read the GFP sequence, and that the resulting mRNA was successfully translated.
Next, we proceeded to the rapid purification and cloning of rVVs expressing GFP by taking advantage of flow cytometry (Fig. 2C, D). After sorting, GFP-positive 293T cells were placed into 24-well plates at 1, 10, and 100 cells/well and expanded (for 24–48h). After three cycles of freezing–thawing of GFP-positive 293T cells, their lysates were used to infect 143B Tk− cells (5×104 cells) in the presence of BrdU, which blocks the replication of wt VV, thus allowing the propagation of rVVs in a short period of time. The following day, abundant GFP expression was seen in the 143B Tk− cell monolayers. Lysate obtained from one sorted GFP-positive 293T cell was sufficient to yield GFP-positive 143B Tk−cells, and the yield of GFP-positive 143B Tk−cells increased proportionally to the numbers of sorted GFP-positive cells used to prepare the lysates. The reason for using three different numbers of GFP-positive starting cells is to find a culture in which sparse, individual green plaques can be identified. As an alternative to FACS sorting, rVVs can also be isolated by picking of green 293 T cell plaques under fluorescent microscopy (Fig. 1B); either way represents a rapid method to generate rVVs for preparative protein expression that substantially shortens conventional protocols.
Next, we demonstrated EnvC expression in 143B Tk− cell lysates by Western blot analysis using a polyclonal HIV-positive human serum. A specific band with a molecular weight of 140kDa (gp140), corresponding to the expected size of the truncated HIV clade C envelope glycoproteins, was detected, while no signal was seen in 143B Tk− cells infected with control virus expressing GFP only (Fig. 3A). Further, no bands were detected on the membrane containing lysates of 143B Tk− cells infected with rVV–GFP–envC or rVV–GFP, when serum from a healthy donor was used to probe the membrane (Fig. 3B). These data demonstrate that rVV–GFP–envC was able to express two foreign genes under the control of a single promoter. To demonstrate that this strategy is reproducible, we constructed another insertion vector by introducing a codon-optimized env gene derived from a pediatric HIV clade C isolate 1084i (Grisson et al., 2004; Rasmussen et al., 2006) into the pSC–GFP plasmid. Similarly, after transfection of 293T cells that had been previously exposed to wt VV, spots of brightly green cells appeared within 12h (Supplemental Fig. 1, available online at www.liebertonline.com). Western blot analysis of cell extracts showed that the resulting rVV expressed HIV1084i envelope glycoproteins (Supplemental Fig. 1).
To generate recombinant envelope glycoprotein on a preparative scale, we exposed 109 of 143B Tk− cells to rVV–GFP–envC at a high MOI (10pfu/cell) and monitored the infection by fluorescent microscope. When 70% of the cells were GFP positive, they were harvested, and lysates were subjected to a two-step purification procedure using lectin and size exclusion chromatography methods described previously (Earl et al., 1994). Polypeptides of the expected size (140kDa) were observed after purification (Fig. 3C, D). After the appropriate fractions were pooled, the yield of EnvC was 1mg. We further characterized the purified protein using Western blot analysis with anti-HIV human serum (Supplemental Fig. 2A, available online at www.liebertonline.com) and anti-HIV gp120 C-terminal monoclonal antibody (Supplemental Fig. 2B, available online at www.liebertonline.com). Both antibodies could detect the purified EnvC (gp140) (lane 3 in Supplemental Fig. 2A, B) and a control, full-length HIV gp160 (lane 2) at the expected size, suggesting that the rVV–GFP–envC-infected cells produced correctly processed EnvC. Further, the purified EnvC was able to induce specific antibody responses in animals (data not shown).
We have established an expeditious strategy to generate rVV strains. The novelty includes the following: (1) use of a bicistronic cassette in the insertion vector with an IRES positioned between the reporter gene GFP and gene of interest and (2) use of GFP as a selection marker to isolate cells harboring recombinant virus by flow cytometry. These modifications substantially reduced the time to obtain cloned rVVs that express the gene of interest. Moreover, GFP expression in infected cells provides a reliable, straightforward methodology to maintain viral stocks free from contamination with spontaneous VV revertants by rapid selection of GFP-positive cells.
We believe that our results are of interest not only because of the technical advance. Since VV RNA polymerase was able to generate functional, bicistronic mRNA in the cytoplasm and this mRNA yielded the expected functional protein products, a single rVV may potentially be engineered to express several genes of interest. Given that the VV genome can accept inserts of >20kB of foreign genetic material (Moss and Earl, 2001), designing rVVs that express up to four foreign genes using our IRES methodology may be feasible. If so, the possibility to design polyvalent vaccines based upon this vector may be considered. The rVV system has advantages over DNA plasmid vaccines and other viral vectors because the use of rVVs eliminates the risk of integrating foreign genetic material into the host genome, although potential side effects of rVVs in immunocompromised individuals may raise safety concerns.
Finally, we believe that our novel GFP-based vector construct may represent a practical tool to study VV infection and pathogenesis in animal models. Previous studies have used rVV-mediated GFP expression to monitor infection in vitro (Dominguez et al., 1998) or in vivo (Norbury et al., 2002); these studies did not use bicistronic cassettes. Our bicistronic cassette gives the added advantage of precise identification of cells expressing a gene of interest; GFP localization will thus give the opportunity to study host–virus interactions at the single-cell level in vivo. Such studies may have added significance given the renewed interest in vaccines against small pox (Qutaishat and Olson, 2003; Stanford and McFadden, 2005).
To conclude, we designed a new, expeditious approach to generate and purify rVVs based upon the use of a bicistronic insertion vector. Our method substantially reduces the time required to obtain constructs expressing the gene of interest.
We thank Dr. Bernard Moss (Laboratory of Viral Diseases, NIAID, NIH) for the gift of VV (strain WR) and pSC59 vector, and Susan Sharp for assistance in the preparation of the manuscript. This work was supported by NIH Grant P01 AI048240 to R.M.R.
No competing financial interests exist.