INTRODUCTION

Recombinant expression systems may use regulatory elements derived from phylogenetically distinct lineages. A successful mammalian expression system that exploits the bacteriophage T7 RNA polymerase to transcribe T7 promoter-regulated target genes was described using vaccinia virus (VV)¹ as the vector (1). In this system, recombinant VV carries an integrated T7 polymerase gene regulated by a viral promoter. Originally, the T7 promoter-regulated template was carried on either a plasmid that was transfected into infected cells or on a second recombinant virus that was coinfected with the T7 polymerase-expressing virus (2, 3). The low translatability of the largely uncapped mRNAs was overcome by initiating T7 transcripts with the leader sequence of encephalomyocarditis virus (EMCV) (4). More recently, the VV/T7 hybrid system was rendered inducible by inclusion of the Escherichia coli lac operator/repressor system (5, 6), thereby allowing the construction of recombinant viruses containing both the T7 RNA polymerase and the T7 promoter-regulated gene. Because vaccinia virus infects a broad range of cells in tissue culture, this system has been used to express recombinant proteins in a variety of cells.

One notable exception to the usable cell lines is the Chinese hamster ovary (CHO) cell line. Despite the fact that CHO cells are one of a few cell lines approved for production of recombinant proteins for use in clinical trials, the VV/T7 technology has not yet been adapted to these cells. One major obstacle has been the restriction in growth and gene expression of VV in CHO cells (7). Insertion of the CHO hr gene of cowpox virus (CPV) into the genome of VV, however, enables VV to productively infect CHO cells (8). Replication of VV on CHO cells is blocked at the stage of viral intermediate protein synthesis (9). This defect is overcome by the CHO hr gene (9), which also delays the onset of apoptosis in VV-infected CHO cells (10). We constructed a new recombinant VV containing both the CHO hr and the bacteriophage T7 RNA polymerase genes, which allows comparable levels of expression in CHO and fully permissive cell lines. Expression was further enhanced by incorporating the EMCV untranslated leader sequence.

MATERIALS AND METHODS

BS-C-1 (kidney, African green monkey) and CV-1 (kidney, African green monkey) cells were grown in minimum essential medium supplemented with 5% fetal calf serum. HeLa S3 monolayer cultures were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. CHO cells were grown in monolayer culture in Ham's F-12 medium supplemented with 5% fetal calf serum. In suspension culture, CHO cells were maintained in an equal mixture of Ham's F-12 and CHO-S-SFM II (Life Technologies, Inc.) or UltraCHO (BioWhittaker) media. VV strain WR and recombinants thereof were propagated as described (11).

Recombinant viruses vTF7-3, vT7EMCAT, and vT7CAT have been described elsewhere (2, 3) and are depicted in Fig. 1. To construct the vT7CP recombinant virus, a 2.3-kilobase pair EcoRI/PstI fragment containing the entire open reading frame of the CPV hr gene CP77 was excised from pEA36 (8) and ligated to EcoRI/PstI cleaved pUC19. The resultant plasmid, pRECP77 (9), was used as a donor in homologous recombination with vTF7-3 to create vT7CP. In addition to the T7 polymerase gene regulated by a viral early/late promoter (P7.5) at the thymidine kinase locus, recombinant virus vT7CP has an intact copy of the CP77 open reading frame, instead of the disrupted WR copy (12), at the corresponding region of HindIII C. 

To examine viral protein synthesis, we seeded 5 × 10⁵ permissive or nonpermissive cells in minimum essential medium with 5% fetal calf serum. After 16-24 h, the cells were coinfected with recombinant viruses at a total multiplicity of 30 plaque-forming units (pfu)/cell, 15 pfu/cell for each virus. 10-20 min before each labeling period, cells were washed twice and incubated in minimum essential medium with 5% fetal calf serum without methionine. The cells were then incubated in the presence of 75 µCi of [³⁵S]methionine in 250 µl of methionine-free minimum essential medium with 5% fetal calf serum for 30 min. The labeling medium was removed and the cells were washed twice with ice-cold phosphate-buffered saline (PBS) and then incubated at 37 °C for 3-5 min with hypotonic lysis buffer (20 mM Tris-HCl (pH 8.0), 10 mM NaCl, and 0.5% Nonidet P-40). The lysates were collected and centrifuged for 2 min at 12,000 × g to sediment nuclei. The supernatants containing ³⁵S-labeled polypeptides were stored at 20 °C. A portion of each sample was mixed with an appropriate volume of 3 × or 5 × sodium dodecyl sulfate/-mercaptoethanol (SDS/2-mercaptoethanol, 5 to 3, Inc.) sample buffer and boiled for 5 min. The proteins were resolved by polyacrylamide gel electrophoresis (PAGE) in 10, 12, or 15% gels.

Cells were infected with 15 pfu/cell of each recombinant virus, labeled with [³⁵S]methionine as described above, and harvested at various times. Lysates were prepared in isotonic lysis buffer (50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.02% sodium azide, 0.1% SDS, 100 µg/ml phenylmethylsulfonyl fluoride, and 1% Nonidet P-40), centrifuged to remove nuclei, and then incubated with polyclonal antiserum to CAT at a 1:500 dilution at 4 °C overnight. An equal volume of 20% (w/v) protein A-Sepharose beads in PBS was added and incubation continued at room temperature for an additional 2-3 h or at 4 °C overnight. Immune complexes were washed twice in Triton buffer (300 mM NaCl, 50 mM Tris-HCl (pH 7.4), and 0.1% Triton X-100), denatured and solubilized in Laemmli buffer, and then resolved by SDS-PAGE in 15% gels.

For analysis of steady-state levels of proteins, Western blot analysis was performed. Infected cells (1 × 10⁶) were incubated with 0.5 ml of lysis buffer (20 mM Tris-HCl (pH 7.0), 0.5% Triton X-100 in PBS) for 5 min at 37 °C. Lysates were centrifuged at 14,000 × g for 5 min and the supernatants stored at 20 °C. A portion was mixed with 5 × SDS/2-mercaptoethanol sample buffer and boiled for 5 min, and proteins were resolved by SDS-PAGE in 10% gels. Proteins were electrophoretically transferred to nitrocellulose membranes and incubated with antisera (anti-CAT 1:500 and anti-T7 polymerase 1:1000) overnight at 4 °C and then with ¹²⁵I-protein A overnight at 4 °C.

RNA was isolated from cells that had been infected with two recombinant viruses at a multiplicity of 15 pfu/cell for each. At various times after infection, the cells were washed twice in ice-cold PBS, and then they were lysed and the RNA was extracted by the guanidinium thiocyanate/phenol/chloroform extraction method (13), using RNAzol B, as described by the manufacturer (Tel-Test).

To quantitate the steady-state levels of RNAs, samples of infected cell RNA were applied to a nylon membrane (Schleicher & Schüll nytran) using a Hoefer 24-well slot blot apparatus. Membranes were hybridized with a ³²P-labeled DNA probe complementary to CAT mRNA.

Cells were infected with recombinant viruses at a multiplicity of 30 pfu/cell. At various times after infection, the cells were washed twice with PBS and incubated for 15 min at 37 °C in 1 × reporter lysis buffer (Promega). The lysate was centrifuged and the supernatant was retained for further analysis. Protein content of each lysate was determined colorimetrically using the Pierce Coomassie Blue protein reagent. Either equal volumes of lysates or equal amounts of protein were used in CAT assays as described by the manufacturer (Promega). Standard curves of both protein content and enzyme activity were prepared for quantitative analysis of assay results.

RESULTS

We measured the impact of host range restriction of VV in CHO cells on expression of a target gene using the VV/T7 system. Permissive (BS-C-1) or nonpermissive (CHO) cells were coinfected with 15 pfu/cell of vTF7-3 (3), which encodes the T7 RNA polymerase gene regulated by the vaccinia virus early/late promoter (P7.5) and with vT7CAT, which has a T7 promoter-regulated CAT gene. In BS-C-1 cells, enzymatically active CAT accumulated rapidly between 4 and 24 h postinfection (hpi). In contrast, low amounts of CAT accumulated between 12 and 24 hpi in CHO cells (Fig. 2). To monitor CAT protein synthesis, we pulse labeled infected cells with [³⁵S]methionine and incubated aliquots of the lysates with polyclonal antiserum directed against the CAT protein. Immune complexes were dissociated with SDS and electrophoretically resolved (Fig. 3A). The results confirmed both the delayed and the low rate of CAT protein synthesis in CHO cells. Thus, there was at least 10-fold more CAT protein synthesis in BS-C-1 than CHO cells.

To determine if the differences in CAT expression reflected the steady-state levels of CAT mRNA, we performed slot blot analysis on total RNA from infected BS-C-1 and CHO cells with a CAT-specific DNA probe (Fig. 3B). Densitometric analysis of the autoradiogram revealed a 4-7-fold difference in the steady-state levels of CAT mRNA in BS-C-1 and CHO cells. This result suggests that the differences in the levels of CAT protein synthesis and enzymatic activity in the two cell lines could be due to the difference in the levels of CAT mRNA as well as in their translation.

Previous studies (14) had shown that the mRNAs synthesized by T7 RNA polymerase are largely uncapped. The translation of the uncapped transcripts was enhanced by incorporating the EMCV leader at the 5-end of the mRNA (4). Therefore, we examined the ability of the EMCV leader to stimulate target gene expression in CHO cells. Total protein synthesis (Fig. 4) was analyzed in cells coinfected with vTF7-3 and vT7CAT or vTF7-3 and vT7EMCAT, which contains a cDNA copy of the EMCV leader preceding the CAT open reading frame. In BS-C-1 cells, decreased cellular protein synthesis was observed at 4 h and had virtually ceased by 12 hpi, as indicated by labeling of the 43-kDa actin protein. Synthesis of viral proteins occurred by 8 hpi and a major protein with the mobility of CAT was labeled between 12 and 24 hpi in the absence of the EMCV leader. In the presence of the EMCV leader, the predominant protein synthesized at 8-12 hpi corresponded to the CAT protein. In CHO cells, shutoff of both cellular and viral protein synthesis occurred by 4 hpi. Synthesis of a polypeptide with the mobility of CAT was observed in the presence of the EMCV leader at 4-24 hpi, but at greatly diminished levels compared to BS-C-1 cells. Quantitation of CAT protein synthesis in CHO cells by immunoprecipitation with anti-CAT antiserum (not shown) revealed some synthesis in the absence of the EMCV leader, but at less than 10% of the levels observed in the presence of the EMCV leader. The results indicated that in both BS-C-1 and CHO cells, the EMCV leader stimulated CAT synthesis approximately 10-fold, thereby maintaining the cell-dependent difference in gene expression. The stimulatory effect was attributed to cap-independent translation since the amount of CAT mRNA was independent of the presence of the EMCV leader (not shown).

To boost the level of gene expression in CHO cells, we prepared a recombinant VV, vT7CP, that expresses the CHO hr gene of CPV, as well as the T7 RNA polymerase gene (Fig. 1). Plasmid pRECP77, carrying sequences that include the promoter and coding sequences of the CPV CHO hr gene, was used for homologous recombination with vTF7-3. At 24 h after infection/transfection of CV-1 cells, lysates were harvested and passaged three times in CHO cells to select for recombinant viruses that contain the CHO hr gene. After the third passage, lysates were titered on BS-C-1 cells. The results showed that the population of viruses capable of growth in CHO cells had been enriched approximately 100-fold (data not shown). After plaque purification, we identified recombinant viruses that contained an intact copy of the CHO hr gene by Southern analysis of DNA from infected cells (data not shown) and selected one of these for further analysis.

We examined the ability of the CHO hr gene to enhance target gene expression in CHO cells. Permissive (BS-C-1 and HeLa S3) and nonpermissive (CHO) cells were coinfected with vT7EMCAT and vTF7-3 or vT7CP. At the indicated times, infected cells were metabolically labeled with [³⁵S]methionine for 30 min and labeled proteins were resolved by SDS-PAGE. A protein with apparent molecular mass of 25 kDa corresponding to CAT was synthesized at high levels at 4-24 hpi in BS-C-1 and HeLa S3 cells that were infected with recombinant viruses lacking (Fig. 5A) or containing (Fig. 5B) the CHO hr gene. In CHO cells, low viral and cellular protein synthesis occurred at 4-24 hpi, and the 25-kDa protein was barely detected (Fig. 5A) unless the CHO hr gene was present (Fig. 5B). Under the latter conditions, synthesis of viral proteins and the 25-kDa protein was comparable to that in permissive cells. We confirmed the identity of the 25-kDa band as CAT by immunoprecipitation (not shown).

To determine if enhanced expression of the target gene in CHO cells could be attributed to an increase in the level of T7 RNA polymerase, we analyzed total infected cell proteins by Western blotting (Fig. 6). Lysates prepared at 24 hpi were resolved by SDS-PAGE and then electrotransferred to nitrocellulose membranes that were subsequently exposed to antibody to T7 polymerase or CAT, followed by incubation with iodinated protein-A. The results revealed that in CHO cells infected with recombinant VV with the CHO hr gene, the increased level of CAT protein synthesis correlated with an increase in the level of T7 RNA polymerase. This result suggests that the absence of sustained viral protein and T7 RNA polymerase synthesis in CHO cells was the immediate cause of diminished target gene expression.

To quantitate the amount of CAT protein synthesized in the vaccinia virus/T7 system adapted to CHO cells and compare it with that obtained in other cell types, we performed CAT assays on samples from three cell types grown in monolayer culture (Fig. 7) and quantitative Western analysis on samples from CHO cells grown in suspension culture (not shown). When the CHO hr gene was present, the amounts of CAT activity recovered per cell at 24 hpi were comparable in BS-C-1, HeLa S3, and CHO cells. We adapted CHO cells to growth in suspension culture in either of two media and coinfected with vT7CP and vT7EMCAT. Quantitation of the results revealed that by 24 hpi, 2-5 mg of CAT protein were produced per 10⁸ suspension CHO cells.

DISCUSSION

The present studies were initiated to learn more about the mechanism of vaccinia virus host restriction in CHO cells and to adapt the widely used vaccinia virus/T7 hybrid expression system to these cells. Vaccinia virus gene expression is regulated by a cascade mechanism: early, intermediate, and late mRNAs are synthesized in succession (15). In CHO cells, however, only early and intermediate mRNAs are synthesized (9). Translation of intermediate mRNAs is severely inhibited, preventing the synthesis of late stage transcription factors and subsequent transcription of late genes. Since the translational defect pertains to mRNAs transcribed from both authentic intermediate stage genes and reporter genes regulated by intermediate promoters, sequence-specific features of intermediate mRNAs are probably not involved. Instead, translation may be generally blocked at the time of intermediate gene transcription. We were curious to determine whether we could overcome the expression block by using T7 RNA polymerase to transcribe reporter genes and by using the leader sequence of EMCV to achieve cap-independent translation. In recombinant VV vTF7-3, the T7 RNA polymerase gene is regulated by an early/late promoter; therefore, synthesis of some T7 RNA polymerase should occur in CHO cells, followed by transcription of the T7 promoter-regulated reporter gene. We found that T7-regulated reporter gene expression was delayed and severely reduced in CHO cells relative to that in permissive BS-C-1 cells. Moreover, the EMCV leader stimulated expression to a similar extent in CHO and BS-C-1 cells, suggesting that the translational block was not related to a specific cap-dependent step in translation. Further analysis indicated the presence of low amounts of T7 transcripts in CHO cells relative to BS-C-1 cells, regardless of whether the EMCV leader was present.

An alternative approach was taken to adapt the VV/T7 hybrid expression system to CHO cells. Spehner et al. (8) had originally shown, and we confirmed (9), that the CP77 hr gene of CPV allows VV to replicate in CHO cells. Sequence studies had shown that the open reading frame of the VV homolog of CP77 was interrupted in the WR strain of VV (12). We therefore used homologous recombination to repair the VV hr gene of a recombinant VV containing the T7 RNA polymerase gene. The resulting recombinant vT7CP replicated in CHO cells and produced larger amounts of viral proteins and T7 RNA polymerase. Importantly, vT7CP could be used as an efficient expression vector either with transfected plasmids containing a T7 promoter-regulated gene (data not shown) or by co-infecting CHO cells with a second VV containing an integrated copy of a T7 promoter-regulated gene. As the CP77 gene is dominant, the second recombinant VV need not have an intact copy. That is a great convenience, since recombinant VV previously constructed for use in permissive cells can be used with vT7CP in CHO cells. With this system, we estimated that approximately 2-5 mg of CAT protein was produced in 10⁸ CHO cells in 24 h. This value could probably be doubled by adapting the inducible VV/T7 system (6) to CHO cells by inserting the CP77 gene into the vector vT7lacOI.

Extension of the VV/T7 expression system to CHO cells was important for several reasons. First, CHO cells are frequently used for overexpression of proteins (16, 17, 18, 19, 20) and can be grown in large scale suspension or hollow fiber reactor cultures in defined serum-free medium. CHO cells are among the very few continuous cells lines approved for expression of recombinant proteins for use in humans. For example, recombinant DNase I, for treatment of cystic fibrosis, has been produced in CHO cells (21). Furthermore, because of the genetic stability of CHO cells, mutants lines deficient in metabolic steps such as glycosylation (22) have been made. In addition, both monochromosomal and polychromosomal human/CHO somatic cell hybrids are available (Coriell Institute for Medical Research). In our laboratory, we have used glycosylation-deficient CHO cells to study the processing of the HIV-1 envelope protein synthesized with the newly modified VVCP/T7system.²