Construction and characterization of pentacistronic retrovirus vectors

Footnotes

The three pentacistronic retrovirus vectors of increasing lengths (Fig. 1) were made by standard genetic-engineering techniques and were maintained as retrovirus plasmids. The autocleavable FMDV Δ1D2A sequence and three different IRESs were used to link various selection genes (pac, neo, hygro, tk) and/or reporter genes (egfp, luc, plap). Penta-7.9 derives from Penta-8.5 and was engineered last. Penta-8.5 and Penta-9.3 derive from Tetra-1 and Tetra-2, described previously (de Felipe & Izquierdo, 2000), introducing a BglII FMDV IRES-luc cassette at the unique BglII site between the neo gene and the EMCV IRES. The FMDV IRES-luc cassette came from plasmid pBiC (Martínez-Salas et al., 1993), filling the BfmI single-stranded ends with Klenow and ligating it to the also filled BglII site of pSXLC-TK (Sugimoto et al., 1994). The construction of Penta-7.9 was achieved by deleting the luc gene from Penta-8.5 and replacing it with the hygro gene via the BglII site downstream of the FMDV IRES. The hygro gene comes from pBabeHygro (Morgenstern & Land, 1990) and was amplified by PCR incorporating BamHI and BglII sites at both ends of the gene. The hygro gene and the junction between FMDV IRES and hygro were verified by standard sequencing procedures.

Plasmid retrovirus vectors were used to transiently transfect ecotropic packaging ψCRE mouse cells and their supernatants to infect amphotropic ψCRIP cells, as described by Miller & Rosman (1989). Cells were selected with 2 µg puromycin ml^-1 for 6 weeks to obtain a population of resistant cells. Titres were estimated by infecting NIH 3T3 cells as described previously (Izquierdo et al., 1997). In all experiments, a plate of uninfected NIH 3T3 cells was placed under the same selection conditions as the infected plates to ensure that a lethal dose of antibiotic selection was used. Cells in the ψCRIP-infected plates used to harvest the supernatant were counted after collection and the titres expressed as transducing units (TU) ml^-1 per cell. The luciferase activity assay was performed as described by Martín et al. (2000) with minor modifications (1·3x10⁵ cells were used and 10 µl of the total 100 µl of cell extract was taken to measure luciferase activity). To estimate plap expression, cells were fixed and stained with BCIP/NBT, as described previously (de Felipe & Izquierdo, 2000).

Supernatants from ψCRIP populations showed titres of about 2x10⁵ TU ml^-1 for Penta-7.9 (equivalent to 0·02 TU ml^-1 per cell) and somewhat surprisingly, titres were higher in the larger vectors (0·36 and 0·15 TU ml^-1 per cell for Penta-8.5 and Penta-9.3, respectively). Expression for all of the reporter genes was positive in cell lawns from Penta-7.9 but low or negative for the last two genes in lawns from Penta-8.5. Expression was also very weak or negative for all genes in Penta-9.3, except for the selection gene pac (data not shown). These results suggested the presence of rearrangements in the larger vectors. No replication-competent retroviruses were detected (de Felipe & Izquierdo, 2000) using the same supernatants from producer ψCRIP cells.

As lawns may represent a mixture of different integration events, it is always preferable to work with individual clones descendant from a single cell. Retrovirus-containing supernatants from producer ψCRIP cells were used to infect NIH 3T3 fibroblasts. These cells were selected in medium containing 2 µg puromycin ml^-1 and surviving colonies were picked and expanded for gene expression analysis. Four NIH 3T3 cell clones harbouring the vector Penta-7.9 were obtained and studied. Expression of egfp was monitored by flow cytometry (Fig. 2a) under the conditions described previously (de Felipe & Izquierdo, 2000).

Individual cell clones containing vectors Penta-8.5 and Penta-9.3 were also isolated by resistance to 2 µg puromycin ml^-1 and the expression of the first, third and fifth genes was assayed. These results are summarized in Fig. 1 by positive or negative symbols under each gene and corresponding clone number. The three NIH 3T3 clones isolated after infection with Penta-8.5 displayed the same pattern of gene expression as that with the ψCRIP population (lawn): resistance to puromycin and G418 (and the combination of both drugs), but absolute lack of GCV sensitivity. In clone number 2, full analysis was carried out and expression of egfp was detected, while no luciferase activity was monitored. A final group of three clones infected with Penta-9.3 showed only resistance to puromycin. The expression of plap was not determined in these clones as the lawn estimation for the gene was negative.

To verify the structural integrity of the proviruses, a long PCR was carried out on genomic DNA extracted from the 10 clones isolated (Fig. 3). The extraction of genomic DNA and the long PCR procedures were carried out as described previously (de Felipe & Izquierdo, 2000), except that elongation times were extended to 20 min. The amplified region begins at the centre of the pac gene and finishes at the 3' LTR. The experiment enables us to correlate deficiencies in gene expression with major rearrangement in the vectors (involving changes in the size of the bands amplified by PCR). Clones 1, 2 and 3 from Penta-7.9 show a band that matches with the expected size fragment of 6·0 kb, but clone 4 seemed to have suffered a large deletion and produced a small band of about 1·6 kb. The only functional gene coded in this deleted provirus is pac, as the expression of all other genes is not detected. More difficult to explain is clone 2, in which a deletion is not apparent (within the detection limits of the electrophoretic analysis), but the second gene, egfp, is not expressed. Sequencing revealed a run of nine thymidine residues between the FMDV 2A gene and the ATG codon of egfp, a situation known to create a hot spot for frameshift mutations in retroviruses (Pathak & Temin, 1990; Burns & Temin, 1994). Two different PCR bands amplified from the genome of clone 2 showed 10 thymidine residues instead of the original nine, thereby creating a frameshift mutation.

Despite the aforementioned restrictions, we have demonstrated that is possible to introduce five foreign cistrons into a retrovirus. Naturally, as the number of cistrons increases, the genes cloned have to become smaller. In our vectors, all the genes carried are smaller than 2 kb; therefore, it may be difficult to acquire substantial additional space. To our knowledge, these are the first functional pentacistronic retrovirus vectors constructed to date that allow the expression of five independent proteins from a single transcription unit. There are reports of pentacistronic vectors using herpes simplex viruses (Krisky et al., 1998) or vaccinia virus (Carroll et al., 1998). However, these vectors, unlike the ones reported here, use several promoters along their large viral genomes. In the present study, we have constructed complex pentacistronic vectors able to co-express five independent genes provided that the size of the wild-type genome is maintained. These results increase our knowledge about the capability and limitations of retroviruses as polycistronic gene vehicles.