A mammalianized synthetic nitroreductase gene for high-level expression

The nitroreductase gene (nfsB, ntr) was amplified from E. coli DH5α genomic DNA by PCR with the primers 5'-GTCTTTATGGATATCATTTCTGTCG-3' (forward) and 5'-AGAGAGAATTACACTTCGGTTAAGG-3' (reverse). The obtained 654 bp amplicon was subcloned into pGEM^®-T easy (Promega) and confirmed by sequencing. The codon-optimized ntro gene (Figure 1) was custom-synthesized by GeneScript (NJ, USA) and provided into the pUC57 subcloning vector. Codon usage tables were obtained from http://​www.​kazusa.​or.​jp[22]. To express nitroreductase with a N-terminal GFP-tag both genes were cut out with EcoR I from the respective source vectors and cloned into pEGFP-C1 (Clontech) using EcoR I restriction sites. The obtained constructs pGNTR and pGNTRo slightly differed in linker sequence (5 of 17 amino acids) and were used for all cell culture experiments. Amino- and carboxyterminally c-myc-tagged ntr constructs were generated by PCR-amplification from pGEM-NTR and pUC57-NTRo using either the primer pairs 5'-GGATCCATGGATATCATTTCTGTCG-3'/5'-AAGCTTTTACACTTCGGTTAAGGTG-3' (pMycNTR) and 5'-GGATCCATGGACATCATCAGCGTGG-3'/5'-AAGCTTTCACACC TCGGTCAGGG-3' (pMycNTRo) or 5'-GGATCCGCCGCCATGGATATCATTTCTG-3'/5'-AAGCTTCACTTCGGTTAAGGTGATGTTTTG-3' (pNTR-myc) and 5'-GGATCCGCCGC CATGGACATCATCAGCG-3'/5'-AAGCTTCACCTCGGTCAGGGTGATGTTC-3' (pNTRo-myc), respectively. The primers provided 5'-BamH I and 3'-Hind III restriction sites for subsequent cloning. All amplicons were subcloned into pGEM^®-T easy and confirmed by sequencing. To express N-terminally-tagged NTR, the amplicons were cloned into pCMV-Tag3B (Stratagene) using BamH I and Hind III restriction sites. For the expression of the C-terminally-tagged NTR, the corresponding amplicons were cloned into pcDNA3.1/myc-His (-) A (Invitrogen), using the same strategy.

Green monkey kidney (COS-7), human embryonic kidney (HEK-293), human neuroblastoma (SH-SY5Y), human small cell lung carcinoma (SHP-77), human osteosarcoma (U-2 OS), mouse fibroblast (3T3-L1), and mouse mastocytoma (P815) cells were maintained under standard conditions at 37°C and 5% CO₂ in DMEM or RPMI (SHP-77 and P815) supplemented with 10% fetal bovine serum and antibiotics. COS-7 cells were transfected with the DEAE-dextran pretreatment method. Transiently transfected COS-7 cells were either used for cytotoxicity assays 24 hours after transfection by cultivating them with different CB1954 concentrations or for Western blot analysis at indicated time points. To establish stable cell lines, cells were transfected with DreamFect (OZ Biosciences) according to the manufacturer's instructions and selected in presence of 500 μg/ml G418 for 2 weeks. After selection, G418-resistent cells were pooled and FACS-sorted to purity using GFP as a fluorescence marker.

Stable cells were seeded in quadruplicate on 24-well plates (2*10⁵ cells/well) in DMEM without Phenol Red supplemented with different CB1954 concentrations. MTT solution (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) was added to each well after 48 hours to a final concentration of 0.5 mg/ml and cells were further incubated for 2 hours. Cells were detached, washed with PBS and suspended in 1 ml MTT stop solution (0.04 M HCl in isopropanol) to solubilize formazan crystals. After centrifugation supernatants were transferred in duplicate to a 96-well plate and extinctions were measured at 570 nm.

Transiently transfected COS-7 cells were seeded in triplicate on a 96-well plate with the indicated concentrations of CB1954. The surviving green cells were counted after 48 hours of treatment within a field of view at 10× magnification using an Axiovert 40 microscope (Zeiss). A total of six randomly chosen fields were counted per experimental condition.

Transiently transfected COS-7 cells were seeded on glass cover slips and cultivated in presence of various CB1954 concentrations for 24 hours. Cells were then fixed with 3.7% formaldehyde in PBS for 10 min. Genomic DNA was stained with DAPI (1 μg/ml in PBS), washed with PBS and mounted in Mowiol. Specimens were analyzed with 10× and 63× planapochromat objectives on an Axiovert 200 M microscope (Zeiss) equipped with epifluorescence optics.

All data are presented as means ± SEM and two-tailed unpaired Student's t-test was used to compare data sets. Values of p < 0.05 were considered as statistically significant.

To test the efficacy of bacterial ntr gene expression in mammalian cell lines, we generated a gfp/ntr fusion construct (pGNTR) that facilitates the detection of NTR protein expression either by fluorescence microscopy or immunodetection (Figure 2A). Interestingly, we could detect perinuclear protein aggregates in the cytoplasm of approximately 10% of the pGNTR-transfected COS-7 cells 48 hours after transfection (Figure 2B). Thus, the question arose whether the divergent codon usage of prokaryotes and eukaryotes was responsible for the aggregates as approximately 60% of the bacterial ntr codons are underrepresented in mammalian genomes (e.g. Mus musculus; Figure 2A and Additional file 1) [22].

To test our hypothesis, we designed a synthetic version of the ntr gene, in which the rare codons for mammalian expression of the wild type E. coli gene were replaced by codons preferentially used in mammalian genomes (Figure 1). A total of 144 base substitutions in 124 codons were introduced without changing its amino acid sequence. The resulting codon-optimized ntr version (ntro) was fused to gfp and the corresponding pGNTRo construct (Figure 2A) was used to transfect COS-7 cells to directly compare gntro expression with gntr. Indeed the cytoplasmic aggregates evident for gntr-expressing cells could no longer be detected in cells expressing gntro (Figure 2B). Furthermore, the GNTR expression was remarkably improved after codon usage optimization to ~3-fold higher protein levels 24 hours after transfection (Figures 3A and 3B). The elevated gntro expression was stable throughout 48 to 72 hours with an average expression level 2-fold higher than of the wild type gene. In order to exclude any artefactual expression difference caused by the aminoterminal GFP-tag, amino- and carboxyterminal c-myc constructs were generated and tested by immunoblotting (Figures 3C and 3D). All codon-optimized expression constructs proved to lead to higher NTR levels as compared with the wild type cDNAs, using several independent batches of plasmid DNA.

To test the enzymatic activity of GFP-tagged nitroreductase, we performed cytotoxicity assays to determine the sensitivity of both GNTR-expressing cell lines to the prodrug CB1954 (Figure 4A). Transfected COS-7 cells were treated with different concentrations of CB1954 for 24 hours. Fluorescence microscopy revealed efficient ablation of gntr-expressing cells by CB1954 at 50 μM. However, similar results were observed already at the 10 times lower prodrug concentration of 5 μM in cells expressing the optimized gntro gene. Control cells expressing gfp were not affected by 5 μM CB1954, but interestingly an unspecific reduction of cell proliferation was observed at 50 μM.

To identify the lowest effective concentration, transfected COS-7 cells were exposed to CB1954 in the range of 0 to 150 μM (Figure 4B). Cells expressing gntro already responded to 1 μM CB1954 with only 41% of living cells 48 hours after prodrug treatment. In contrast, no effect could be detected in gntr-expressing cells and significant signs of apoptosis became first evident at 10 μM and 25 μM CB1954, where 39% and 67% of the cells were killed, respectively. However, at these concentrations only about 5% of the gntro-expressing cells survived. Comparable ablation for gntr-expressing cells were obtained only at 100 μM and 150 μM CB1954, but the number of control cells was also decreased at these concentrations, most likely due to inhibition of cell proliferation as revealed by microscopic analysis (data not shown).

COS-7 cells represent an artificial system for ectopic gene expression [23]. Thus, stable cell lines from diverse tissues were required to confirm the improvement of the NTR/CB1954 system by codon usage optimization under more native conditions. Therefore, we generated human embryonic kidney (HEK-293), human neuroblastoma (SH-SY5Y), human small cell lung carcinoma (SHP-77), human osteosarcoma (U-2 OS), mouse fibroblast (3T3-L1), and mastocytoma (P815) cell lines stably expressing gntr, gntro and gfp (control) and treated them with different concentrations of CB1954 (Figure 5 and Additional file 2).

Remarkable differences in survival rates were detected in a concentration dependent manner after 48 hours of treatment (Figure 5A, B and 5C). All gntro-expressing cell lines showed a dramatic decrease in cell number already at 5 μM CB1954 with only 17%, 30% and 16% of cells still alive as compared with control cells, respectively. Almost no cells survived at 50 μM CB1954. These results were in line with the data obtained from COS-7 cells transiently transfected with gntro (Figure 4B). CB1954 also decreased the cell number of gntr-expressing HEK-293 and SHP-77 cells dose-dependently, but the sensitivity was clearly lower at all concentrations compared with gntro-expressing cells. Indeed, 58% and 64% of gntr-expressing HEK-293 and SHP-77 cells were still alive at 5 μM and higher ablation rate of 72% and 94% were only reached at 100 μM, respectively (Figure 5A and 5C). These results were consistent with other stable human (U-2 OS) and mouse (P815, 3T3-L1) cell lines tested (Additional file 2).

Surprisingly, gntr expression in SH-SY5Y cells had no effect on their sensitivity to CB1954 (Figure 5B). The observed decrease in cell number of approximately 40% in these cells was not different from gfp-expressing control cells and was most likely caused by a CB1954-mediated inhibition of cell proliferation, as revealed by microscopic analysis. These results were consistent with another neuronal cell line (NG-108) tested (data not shown), most likely due to differential tissue-specific codon usage [24, 25].

The presence of full-length GNTR protein was analyzed in all cell lines through immunoblotting of whole cellular lysates using an anti-GFP antibody (Figure 5D). The expression of gntr was less efficient to that of gntro in all three cell lines and correlates with the decreased sensitivity of these cells to CB1954 (Figure 5A, 5B and 5C).

The bystander effect is an important aspect of tumor ablation using GDEPT, which is often desired to overcome limitations of gene delivery to allow effective tumor regression with a low number of suicide gene-expressing cells [26‐28]. In order to assess for this effect, various ratios of stably transfected ntr-positive (ntr⁺) and wild type HEK-293 cells were mixed and kept at the indicated concentrations of CB1954 (Figure 6). Cell survival and viability was then quantified using a MTT assay after 48 hours of treatment. At 5% ntr⁺ cells, only gntro-expressing cells showed a detectable bystander effect, even at a CB1954 concentration of 5 μM (Figure 6A). A similar effect was detected for gntr-expressing cells only at 25% ntr⁺ cells and 25 μM prodrug (Figure 6A). SHP-77 cells, although responding only at 25% ntr⁺ and 25 μM (gntro), also showed a bystander effect, while gntr-expressing cells did not (Figure 6B).