The Selectable Marker Neo Gene Down-Regulates Gene Expression from Retroviral Vectors Containing an Internal Ribosome Entry Site

Home , Reporter gene

Gene Therapy (1998) 5, 1441–1444  1998 Stockton Press All rights reserved 0969-7128/98 $12.00 http://www.stockton-press.co.uk/gt BRIEF COMMUNICATION The selectable marker neo gene down-regulates gene expression from retroviral vectors containing an internal ribosome entry site

J Byun1, J-M Kim1, PD Robbins2 and S Kim1 1Institute for Molecular Biology and Genetics, Seoul National University, Seoul, Korea; and 2Department of Molecular Genetics and Biochemistry, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA

The internal ribosome entry site (IRES) from the picorna- terial neo sequence has a negative effect on expression. virus family has frequently been used to express multiple Analysis of the steady-state RNA levels isolated from genes from a polycistronic message in retroviral vectors. transfected packaging cells showed that the neo-contain- While examining factors affecting levels of gene expression ing retroviral vectors produced signiﬁcantly lower levels of in IRES-containing retroviral vectors, it was found that RNA than those lacking this bacterial sequence, indicating retroviral vectors expressing the two genes linked by IRES, that neo interferes with expression of the neighboring gene the reporter gene and the selectable marker neo, produced at the level of RNA. Furthermore, the order of genes in the signiﬁcantly lower levels of protein than those containing a IRES-neo-containing vectors appeared to be more reporter gene alone. This observation has been made with important than in the vector lacking the neo sequence. Our various cDNA sequences. However, when the neo was results suggest that neo has to be used in the retroviral replaced with a different cDNA, the level of gene vector with care, especially when a high level gene expression was increased, often to the level achieved with expression is needed. a vector expressing a single gene, suggesting that the bac-

Keywords: IRES; neo; retrovirus; gene therapy

Retroviral vectors have been the vehicle of choice for unrelated genes to produce polycistronic mRNAs. many gene transfer experiments and a large number of Indeed, IRES elements have been demonstrated to work clinical protocols still employ ex vivo approaches which in the context of retroviral vectors to express two or more necessarily involve the use of selectable marker genes to genes.6–10 Despite the wide use of neo and IRES in retrovi- allow easy selection for the successfully transduced ral vector designs, factors affecting gene expression in cells.1,2 Of the various dominant selectable markers that IRES-containing retroviral vectors remain poorly under- are in use for mammalian cells, the neo gene has been the stood. Here we report that the presence of the neo gene in most extensively employed as a marker for retroviral the IRES-containing vectors significantly down-regulated vectors.3,4 expression of the second gene in the same vector, sug- There are two general ways of expressing neo along gesting a need for a different selectable marker when a with the gene of interest. In one type, an independent high level of gene expression is desired. promoter, either internal or the LTR itself, is used,5 while To test the effect of IRES on gene expression, vectors in the second type, the neo gene is expressed as part of containing only the reporter gene and those expressing a multicistronic message using an internal ribosome entry the two genes (the reporter gene and the neo selectable site (IRES).6–10 The internal promoter-based designs were marker) using IRES were first compared for levels of gene thought to suffer from the limitation of promoter expression by transduction assays (Figure 1). As a occlusion where the transcription from one promoter reporter gene, cDNA sequences for murine GM-CSF, suppresses transcription from another when the two pro- human IL-4 and human EPO were used because their lev- moters are arranged in the same orientations.11–13 IRES- els of gene expression could readily be measured from based designs avoid such problems by expressing mul- cell culture supernatant (Figure 1a). The retroviral vectors tiple genes in one message from a single promoter. MFG was chosen because IRES has already been used to The IRES sequences utilized to date are derived mostly drive gene expression efficiently in this vector.9 from members of the picornavirus family. They can be The amphotropic packaging line BING was transfected isolated from an original viral genome and linked to with respective retroviral vectors and cell-free viral supernatants were used to transduce NIH3T3 cells. Two days after transduction, culture supernatants were taken Correspondence: S Kim, IMBG, Bldg-105, Kwan-Ak-Gu, Seoul 151– to measure the concentration of each protein by the com- 742, Korea mercially available ELISA kit. In all three cases, retroviral Received 26 February 1998; accepted 5 June 1998 vectors lacking the IRES-neo sequence produced signifi- Negative effect of neo gene J Byun et al 1442

Figure 1 Effect of IRES-neo cassette on gene expression. (a) Schematic diagram of retroviral vectors. SD, splice donor; SA, splice acceptor; X, gene of interest; mGM-CSF, murine GM-CSF; hIL-4, human IL-4; hEPO, human erythropoietin; IRES, internal ribosome entry site from EMCV (encephalomyocarditis virus). (b) Effect of IRES-neo cassette on gene expression from transduced cells. The cell-free retroviral supernatants harvested from transfected BING packaging cells were used to transduce NIH3T3 cells plated the previous day at 1 × 106 cells in 60-mm plate. The levels of mGM-CSF, hIL-4 and hEPO were determined by ELISA (R&D Systems, MN, USA) 2 days after transduction. − and + indicate the absence and the presence of IRES-neo cassette, respectively. Values are mean ± s.d. (n = 4 for mGM-CSF, n = 4 for hIL-4, n = 3 for hEPO). (c) Effect of IRES-neo cassette on gene expression from transiently transfected BING cells. The level of each cytokine was measured 48 h after transfection. All transfections were done on 60-mm plate where 2 × 106 cells were plated the night before transfection.

Figure 2 The negative effect of neo on gene expression. The neo sequence was replaced with either bacterial CAT or murine GM-CSF and compared for the level of hIL-4 expression. The vectors differ only in their genes downstream from IRES. These retroviral constructs were transiently transfected to 293T cells (2 × 106 per 60-mm plate) and the levels of hIL-4 were measured 48 h after transfection by ELISA (R&D). Values are mean ± s.d. (n = 3). 4, MFG-hIL-4; 4-N, MFG-hIL-4-IRES-neo; 4-C, MFG-hIL-4-IRES-CAT; 4-G, MFG-hIL-4-IRES-mGM-CSF.

cantly higher levels of these reporter proteins (Figure 1b). gene expression in transfected packaging cells where This result suggested that the presence of IRES-neo gene expression would be mediated by the provirus in sequence in the retroviral vector somehow down-regu- the plasmid instead of from the integrated retroviral gen- lated gene expression in the context of MFG. ome. As shown in Figure 1c, the magnitude of difference Next, we compared the same vectors for their levels of in the level of cytokines between the transfected vectors Negative effect of neo gene J Byun et al 1443 containing and lacking the IRES-neo sequence was com- of neo produced higher levels of GM-CSF than those con- parable with the results obtained from the transduction taining the selectable marker. assays, indicating that decreased levels of reporter pro- The retroviral vectors used in this study produced two teins occurred probably at the level of gene expression. RNA species, a genomic transcript and a spliced RNA To identify the sequence responsible for the decreased (Figure 3b). The level of RNAs that specifically level of gene expression, neo was replaced with different hybridized with the virus-specific probe was much sequences whose product could readily be assayed. First, higher with the retroviral vectors expressing GM-CSF in the retroviral vector expressing human IL-4, neo was and CAT than those containing neo and CAT. This result replaced with the bacterial CAT or murine GM-CSF suggested that neo interfered with gene expression from sequences. As shown in Figure 2, replacement of neo with IRES-containing retroviral vectors, probably at the level different sequences increased the level of IL-4 three- to of RNA. five-fold, close to the level achieved with MFG expressing The above result also suggested that the order of genes IL-4 alone. A similar observation was made in the retrovi- in the IRES-containing retroviral vector might have a dif- ral vector expressing GM-CSF. This result indicated that ferent effect on gene expression depending on sequences the neo sequence when used in conjunction with IRES has in the vector. It was reproducibly found that GM-CSF a negative effect(s) on gene expression, at least in the con- and CAT could be placed either downstream or upstream text of MFG. from IRES without significantly affecting the level of gene To understand at what level of gene expression the expression. However, the position of the neo sequence negative effect of neo occurs, we analyzed the quality and appeared to have a small, but reproducible effect in that quantity of viral RNAs produced in transfected 293T the level of GM-CSF was slightly higher when the neo cells. Retroviral constructs expressing GM-CSF, neo and sequence was placed upstream from IRES. CAT in various positions were transfected to the 293T To our knowledge, in all approved gene therapy clini- cells and total RNAs were prepared followed by North- cal protocols employing retroviral vectors, neo is used ern blot hybridization analysis using the GM-CSF cDNA when selection of transduced cells is needed. Our results as a probe (Figure 3). First, the levels of GM-CSF in cul- suggest that the neo sequence down-regulates gene ture supernatant of transfected cells were also measured expression from the IRES-containing retroviral vector, (Figure 3a). Similar to results obtained with the IL- probably at the level of RNA. The magnitude of a nega- 4/CAT vector, retroviral vectors containing CAT in place tive effect appears to vary depending on the genes

Figure 3 Protein and RNA analysis in 293T cells transiently transfected with various retroviral constructs. (a) Analysis of the protein level. The four retroviral constructs expressing two genes in different positions with differing combinations of cDNAs were transfected to 293T cells and the levelsof GM-CSF in the cell culture supernatants were measured by ELISA (R&D). C-G, MFG-CAT-IRES-mGM-CSF; G-C, MFG-mGM-CSF-IRES-CAT; N- G, MFG-neo-IRES-mGM-CSF; G-N, MFG-mGM-CSF-IRES-neo. (b) Northern blot analysis of 293T cells transiently transfected with the four retroviral constructs. Lane 1, C-G; lane 2, G-C; Lane 3, N-G; lane 4, G-N. Total RNAs were prepared 48 h after transfection. The 425 bp NcoI/BamHI fragment from MFG-GM-CSF was used as a probe. G, genomic RNA; SG, subgenomic RNA. T denotes tubulin RNA. Negative effect of neo gene J Byun et al 1444 expressed together. However, its negative effect spans a 4 Byun J et al. A simple and rapid method for the determination broad spectrum of sequences. It is not yet clear how the of recombinant retrovirus titer by G418 selection. Gene Therapy neo sequence negatively affects the level of mRNA. One 1996; 3: 1018–1020. possibility would be the presence of an inhibitory 5 Miller AD, Rosman GJ. Improved retroviral vectors for gene transfer and expression. Biotechniques 1989; 7: 980–990. sequence in the neo coding region that lowered the rate 6 Adam MA, Ramesh N, Miller AD, Osborne WRA. Internal of RNA elongation or the stability of RNA. Whatever the initiation of translation in retroviral vectors carrying picorna- mechanism, it is recommended that when the neo has to virus 5′ nontranslated regions. J Virol 1991; 65: 4985–4990. be used, the choice of vectors and conﬁguration of genes 7 Morgan RA et al. Retroviral vectors containing putative internal in a given vector be carefully determined to achieve opti- ribosome entry site: development of a polycistronic gene trans- mum level of gene expression. fer system and applications to human gene therapy. Nucleic Acids Res 1992; 20: 1293–1299. 8 McLachlin JR et al. Factors affecting retroviral vector function Acknowledgements and structural integrity. Virology 1993; 195: 1–5. 9 Zitvogel L et al. Construction and characterization of retroviral This work was supported in part by research grants from vectors expressing biologically active human interleukin 12. the Korea Ministry of Science and Technology (SK) and Hum Gene Ther 1994; 5: 1493–1506. the Ministry of Education (SK) and by a public health 10 Ramesh N et al. High-titer bicistronic retroviral vectors service grant DK44935 (PDR). employing foot-and-mouth disease virus internal ribosome entry site. Nucleic Acids Res 1996; 24: 2697–2700. 11 Emerman M, Temin HM. Genes with promoters in retrovirus References vectors can be independently suppressed by an epigenetic mechanism. Cell 1984; 39: 459–467. 1 Ross G et al. Gene therapy in the United States: a 5-year status 12 Emerman M, Temin HM. Quantitative analysis of gene sup- report. Hum Gene Ther 1996; 7: 1781–1790. pression in integrated retrovirus vectors. Mol Cell Biol 1986; 6: 2 Marcel T, Grausz JD. The TMC worldwide gene therapy 792–800. enrollment report, end 1996. Hum Gene Ther 1997; 8: 775–800. 13 Olsen JC et al. Retrovirus-mediated gene transfer to cystic 3 Miller AD, Miller DG, Garcia JV, Lynch CM. Use of retroviral ﬁbrosis airway epithelial cells: effect of selectable marker vectors for gene transfer and expression. Meth Enzymol 1993; sequences on long-term expression. Nucleic Acids Res 1993; 21: 217: 581–599. 663–669.