In vitro evaluation of cancer-specific NF-κB-CEA enhancer–promoter system for 5-fluorouracil prodrug gene therapy in colon cancer cell lines

Nuclear factor-kappa B (NF-κB) is a transcription factor with high transcriptional activity in cancer cells. In this study, we developed a novel enhancer–promoter system, κB4-CEA205, in which the basal carcinoembryonic antigen (CEA) promoter sequence (CEA205) was placed downstream of the four tandem-linked NF-κB DNA-binding sites (κB4). In combination with a κB4 enhancer, the transcriptional activity of the CEA promoter was significantly enhanced (three- to eight-fold) in cancer cell lines but not in normal cells. In cancer cell lines, the transcriptional activity of κB4-CEA205 was comparable with that of the SV40 promoter. We also constructed vectors in which the thymidine phosphorylase (TP) cDNA was under the control of CEA205, κB4, κB4-CEA205 and CMV promoters, respectively. TP protein and enzyme activity were detected at comparable levels in κB4-CEA205- and CMV-driven TP cDNA-transfected cancer cell lines (H630 and RKO). The κB4-TP and CEA205-TP-transfected cell lines, respectively, only demonstrated negligible and low levels of TP protein and enzyme activity. Both CMV- and κB4-CEA205-driven TP cDNA transiently transfected cells were 8- to 10-fold sensitised to 5-fluorouracil (5-FU) prodrug, 5′-deoxy-5-fluorouradine (5′-DFUR), in contrast to only 1.5- to 2-fold sensitised by the κB4- and CEA205-driven TP cDNA-transfected cells. The bystander killing effect of CMV- and κB4-CEA205-driven TP cDNA-transfected cells was comparable. This is the first report that indicates that the NF-κB DNA-binding site could be used as a novel cancer-specific enhancer to improve cancer-specific promoter activity in gene-directed enzyme prodrug therapy.

Lack of selectivity for malignant cells leading to dose-limiting toxicity is one of the major obstacles for the success of cancer chemotherapy. Gene-directed enzyme prodrug therapy (GDEPT) is potentially an elegant way to improve the therapeutic index of active anticancer drugs, but one of the major limitations for the clinical application of this approach is the lack of strong cancerspecific promoters (Aghi et al, 2000;Rooseboom et al, 2004).
Carcinoembryonic antigen (CEA) is not expressed in normal and overexpressed in epithelial cancer cells (Shively and Beatty, 1985). Four cis-acting DNA-binding elements mapped on the CEA promoter region within 300 bp upstream of CEA translation starting point, especially, the first three elements are essential for specific CEA transcription (Hauck and Stanners, 1995;Richards et al, 1995a). Utilisation of the CEA promoter in an adenovirus vector for cytosine deaminase (CD) gene expression improves the selectivity of 5 0 -deoxy-5-fluorocytidine (5-DFCR)/5-fluorouracil (5-FU) conversion in CEA-expressing tumours (Richards et al, 1995a;Lan et al, 1996a). However, the transcriptional activity of CEA in most cancer cell lines is 10-to 300-fold lower than that of CMV and RSV viral promoters (Lan et al, 1996b;Ueda et al, 2001).
Even when using adenovirus to deliver CEA promoter-driven suicide gene in vivo, few reports have demonstrated sufficient antitumour effects. Although a 2.1-kb CEA enhancer can increase the CEA promoter activity, it is still highly dependent on the intrinsic level of CEA expression (Richards et al, 1995b;Nyati et al, 2002). The use of this CEA enhancer-promoter-based prodrug approach in cells with low levels of CEA expression cannot induce sufficient CD activity to convert 5-DFCR to 5-FU other than at very low levels (Richards et al, 1995b). As with other cancer-specific promoters, the low promoter activity is a major obstacle for using CEA promoter-based prodrug gene therapy in vivo and in clinical trials.
Nuclear factor-kappa B (NF-kB) (Ephrussi et al, 1985) is a transcription factor, which is composed of five subunits (p50/p105, p52/p100, p65 (RelA), RelB and c-Rel) forming hetero-or homodimers. Nuclear factor-kappa B is normally retained in the cytoplasm as an inactive complex through the direct binding of the natural inhibitor of kB (IkB) (Karin, 1999). Nuclear factor-kappa B can be activated by cytokines, UV radiation, reactive oxygen species and anticancer drugs (Karin, 1999). All of these stimuli trigger phosphorylation and degradation of IkB and release NF-kB homo-or heterodimers, which are then translocated into nucleus and bound to DNA target site (kB site) to influence downstream gene expression. kB site is a strong cis-acting enhancer sequence possessing very weak transcriptional activity on its own. In cells with high NF-kB activity, kB enhancer specifically magnifies relevant promoter activity in a cis-acting manner (Ephrussi et al, 1985;Nelms et al, 1990). In a reporter vector, the enhancing activity of NF-kB can be further improved by multitandem copies of kB element located upstream of a weak promoter sequence. Compared with normal cells, most cancer cells possess high NF-kB activity, which is produced by the cancer cells or the surrounding inflammatory cells. The NF-kB activity in tumour tissues can be further induced by exposure to anticancer drugs, radiation or cytokines .
Capecitabine (CAP), a prodrug of 5-FU (Malet-Martino and Martino, 2002;Petty and Cassidy, 2004), has been approved as a first line therapeutic agent for colorectal cancer (CRC) chemotherapy. Capecitabine is converted into 5-FU in vivo by a threeenzyme-catalysed process. Capecitabine is converted to 5 0 -DFCR and then to 5 0 -deoxy-5-fluorouradine (5 0 -DFUR) by carboxylesterase and CD, respectively. Both conversions take place in liver and are not limiting steps for efficient conversion of CAP to 5-FU. The final and critical step is the conversion of 5 0 -DFUR to 5-FU by thymidine phosphorylase (TP). In comparison with normal cells, cancer cells express relatively higher levels of TP. The TP expression contrast between cancer and normal cells would potentially enable the enrichment of 5-FU in tumour cells, while reducing the exposure of normal tissues to 5-FU. However, TP is only detected in 24 -66% (Evans et al, 2002) of CRC. Most tumours only express moderate levels of TP with 5-FU-resistant tumours expressing at even lower levels (Boyer et al, 2004). Consequently, insufficient conversion of 5 0 -DFUR into 5-FU in tumour tissues becomes the major limitation of CAP-based chemotherapy. Strong promoters such as CMV have been used to increase CD and TP expression in cancer cells and demonstrate enhanced cytotoxicity of 5 0 -DFCR and 5 0 -DFUR in vitro and in vivo (Huber et al, 1994;Patterson et al, 1995;Evrard et al, 1999;Ciccolini et al, 2001;Manome et al, 2001). Yet, the lack of specificity of these promoters limits the clinical development of GDEPT strategies.
Based on our understanding of the molecular biology of cancer, in this study, we used a non-virus expression system to develop a strong and highly cancer-specific NF-kB-CEA enhancer -promoter system (kB4-CEA205), which demonstrated high cancer-specific transcriptional activity. Transfection of TP cDNA driven by the NF-kB-CEA enhancer -promoter cassette highly sensitised CRC cell lines to 5 0 -DFUR in vitro.

Construction of luciferase and TP expression vectors
The human CEA promoter cDNA (pDRIVE-hCEA) purchased from InvivoGen (San Diego, CA, USA) was subcloned by polymerase chain reaction (PCR). The CEA basal promoter sequences in this paper are numbered relative to the middle nucleotide of the first ATG (Schrewe et al, 1990). Two basal CEA promoter sequences (CEA205: À245 to À41 and CEA421: À421 to þ 1) were used to construct luciferase reporter vectors in combination with or without NF-kB enhancer elements. As shown in Figure 1, both basal promoter sequences contain the essential CEA promoter region. CEA205 and CEA421 covered the first three and four cis-acting elements, respectively. The BglII and HindIII restriction sites were introduced into the forward and reverse primers, respectively. The primers used were as follows (the underlined: restriction sites). CEA205: forward 5 0 -GCGCAGATCT GAAAATAGAAGGGAAAAAAG-3 0 and reverse 5 0 -GCGCAAGCTT GAGTTCCAGGAACGTTT-3 0 ; CEA421: forward 5 0 -GCGCAGATCT AGAGCATGGGGAGACCCGGGA-3 0 and reverse 5 0 -GCG CAAGCTTGGTCTCTGCTGTCTGCTCTGTC-3 0 . The PCR fragments were subcloned downstream of four tandem NF-kB-binding sites (kB4; 5 0 -GGGAATTTCC-3 0 Â 4) replacing the TATA-like promoter fragment into a commercially available pNF-kB-Luc vector (BD Biosciences, San Jose, CA, USA). To produce the pkB4-Luc vector without the essential promoter sequence, the TATA-like promoter sequence was removed by restriction enzyme cutting and the backbone plasmid DNA was blunt-ended using DNA polymerase Klenow fragment (Promega, Southampton, UK) and religated. To produce expression vectors without the NF-kB enhancer, the kB-binding sites in pkB4-CEA205-Luc and pkB4-CEA421-Luc were removed and the backbones religated as described above. Thymidine phosphorylase cDNA (pCMV6-XL5/ TP) was purchased from Origene (Rockville, MD, USA). Thymidine phosphorylase cDNA was subcloned into pcDNA3 (Invitrogen, Paisley, UK) using EcoRI/XbaI to develop pcDNA3-TP in which TP cDNA is under the control of CMV promoter. To develop TP expression vectors for transient transfection, the luciferase cDNA in the above luciferase reporter gene expression vectors was replaced with TP cDNA derived from pcDNA3-TP by HindIII/XbaI cutting. For stable expression of TP in CRC cell lines, the CEA205-TP and kB4-CEA205-TP fragments were subcloned into a pFastBacDual vector (Invitrogen, Paisley, UK) from which the p10 and polyhedrin promoters have been removed (pFastBac-DualDpromoter). The integrity of the sequences of the constructs was verified by DNA sequencing.

TP-activity assay
The TP enzyme activity was assayed using the method published by Yoshimura et al (1990) with minor modification. Briefly, 10 ml of total protein extracted in RIPA buffer containing 25 mM Tris-HCl, 0.1% SDS, 1% Triton X-100, 0.15 M NaCl, 1 mM EDTA and 1 Â protease inhibitor mixture (Roche, East Sussex, UK) was mixed with 90 ml of reaction buffer consisting of 10 mM thymidine, 10 mM KH 2 PO 4 (pH 7.4). The reaction was incubated at 371C for 2 h and terminated by addition of 400 ml of 0.2 N NaOH. The concentration of thymine converted from thymidine was measured spectrophotometrically at 300 nm. The amount of thymine in the reaction was calculated from a standard curve plotted from known thymine concentrations. Thymidine phosphorylase activity was presented as nmol thymine/mg protein/hour.

Western blot
Total protein was extracted in RIPA buffer. The lysate was centrifuged for 5 min in a microfuge and the supernatants retained. The protein (200 mg/cell line) was electrophoresed through a 10% SDS-PAGE and transferred to a PVDF membrane (Millipore, Hertfordshire, UK) using a semi-dry electrophoretic transfer chamber (Millipore). Nonspecific binding was blocked by incubating the membranes for 1 h in TBS-T with 5% non-fat milk, which was also used to dilute primary (TP, Novocastra, Newcastle upon Tyne, UK, 1 : 250; CEA, Abcam, Cambridge, UK, 1 : 2000) and secondary (Amersham, Little Chalfont, UK; 1 : 5000) antibodies. The signal was detected using an ECL Western blotting detection kit (Amersham), and visualised by exposure to X-ray film.

Total RNA isolation and RT -PCR
Cells (70% confluent) cultured in 25 cm flasks were harvested by trypsinisation. Total RNA was isolated using TRIzol reagent (Invitrogen, Paisley, UK) according to the manufacturer's protocols. The CEA mRNA expression levels in different cell lines were determined using the Access RT -PCR System (Promega, Southampton, UK) following the instruction of the supplier. The CEAspecific primers were designed from human CEA DNA sequence (GenBank ID: M17303). The forward (5 0 -CGC CAA AAT CAC GCC AAA TAA TAA-3 0 ) and reverse (5 0 -ACC CCA ACC AGC ACT CCA ATC AT-3 0 ) primers were used to amplify a 171 bp PCR product. The human housekeeping gene GAPDH (GenBank ID: XR018317) was used as the RNA loading control, which was amplified by the same RT -PCR system using the following primers, forward: 5 0 -CATGACAACTTTGGTATCGTG-3 0 ; reverse: 5 0 -GTGTCGCTGTTG AAGTCAGA-3 0 . The RT -PCR amplification conditions were as follows. One cycle at 481C for 45 min; 1 cycle at 941C for 2 min; 35 cycles at 941C for 30 s, 601C for 30 s, 721C for 1 min; 1 cycle at 721C for 5 min. The RT -PCR products were separated on a 1% agarose gel and the bands visualised and photographed under UV light.

MTT assay
For in vitro cytotoxicity analysis, the overnight-cultured cells (5000/well in 96-well flat-bottomed microtiter plates) were constantly exposed to 5 0 -DFUR for 72 h and then subjected to a standard MTT assay as previously described (Plumb et al, 1989).
To test the effect of transient expression of TP on cytotoxicity of 5 0 -DFUR, cells were transiently transfected with TP expression vectors for 48 h and subjected to MTT assay. To determine the bystander effect, the pcDNA3-TP-or pkB4-CEA205-TP-transfected cells were mixed with pcDNA3-transfected cells. The whole cell populations consisted of 0, 5, 10, 20, 40, 60 and 100% TP cDNA (pcDNA3-TP and pkB4-CEA205-TP)-transfected cells, respec-tively, were exposed to 5 0 -DFUR and subjected to MTT analysis. The IC 50 s of 5 0 -DFUR were calculated.

In vitro transfection
Stable transfection Colorectal cancer cell lines (1 Â 10 6 /well) were cultured in six-well plates overnight and transfected with pFastBacDualDpromoter/kB4-CEA205-TP using Lipofectamine 2000 (Invitrogen, Pasley, UK) following the manufacturer's instructions. The pFastBacDual empty vector-transfected cells were used as a negative control. The successfully transfected clones were selected in G418 (1 mg ml À1 ). After TP enzyme activity assay and Western blotting analyses, the positive clones were selected and cultured in G418-containing medium to enlarge the cell population. Two positive clones were subjected to further analysis.

Luciferase reporter gene assay
All transfections were performed using Lipofectamine transfection reagent according to the instruction of the manufacturer (Invitrogen, Paisley, UK). Briefly, 5 Â 10 4 /well cells were cultured in 24-well plates overnight. The luciferase reporter vectors (pkB4-Luc, pCEA205-Luc, pkB4-CEA205-Luc, pNF-kB-Tal-Luc and pGL3-Basic; 0.8 mg well À1 ) were co-transfected with 0.008 mg well À1 pSV40-Renilla DNA, an internal control for normalisation of the transcriptional activity of the reporter vectors. Forty-eight hours after transfection, the cells were lysed and luciferase activities were determined using Dual Luciferase Assay reagents (Promega, Southampton, UK) according to the manufacturer's instructions. The luciferase activity in each well was normalised to pSV40-Renilla using the formula of Ln ¼ L/R (Ln: normalised luciferase activity; L: luciferase activity reading and R: Renilla activity reading). The Ln was further standardised by the transcriptional activity of the pGL3-Basic using the formula of RLU ¼ Ln/pGL3-basic (RLU: relative luciferase unit). All transfections were performed in triplicate and all experiments were repeated at least twice.

Background CEA and TP gene expression and the transcriptional activity of NF-jB in CRC and normal cells
To evaluate the efficacy of NF-kB-CEA enhancer -promoter system in 5 0 -DFUR prodrug based GDEPT, it is essential to determine the background status of TP (protein and enzyme activity), CEA (mRNA and protein) and NF-kB (transcriptional activity) in cancer and normal cells. In a panel of eight CRC cell lines, only two (LoVo and HCT116) expressed the 55 kDa TP protein at moderate levels (25%). No TP protein expression was detected in the remaining six CRC and normal endothelial and epithelial cells (Figure 2A). The TP enzyme activity levels in the CRC cell lines were consistent with the TP protein expression levels ( Figure 2C). The 180 kDa CEA protein was detected in four CRC cell lines (50%). Although the CEA mRNA was detected in most of the CRC cell lines (7/8, Figure 2B), it was only expressed at relatively high levels in three cell lines (LoVo, Caco2 and HT29) and at very low levels in the other four cell lines (DLD-1, HCT116, H630 and RKO). In contrast with cancer cell lines, no CEA protein and mRNA were identified in normal cells ( Figure  2A and B). The transcriptional activity of NF-kB in CRC and normal cells was also examined. High NF-kB activity was detected in all eight CRC cell lines by luciferase assay. In contrast, the normal human cells only demonstrated negligible background levels of NF-kB activity ( Figure 2D).

The transcriptional activity and specificity derived from different combinations of CEA basal promoters and NF-jB enhancer
To improve the transcriptional activity of CEA promoter and maintaining its cancer-targeting specificity, we took advantage of the high NF-kB activity in CRC cell lines and designed a novel NF-kB-CEA chimeric enhancer -promoter system (Figure 1). In this system, the basal CEA promoter sequences were placed downstream of the four directly linked NF-kB DNA-binding sites (GGGAATTTCC) in tandem (kB4). There are four cis-acting DNAbinding sites located within À303 to À143 base pairs of the CEA promoter region. The transcription factors Sp1, Sp1-like and USF bind to the first three DNA-binding sites and the transcription factors for the fourth element are still not clear (Hauck and Stanners, 1995). To develop an optimal enhancer -promoter system, we firstly compared the transcriptional activity and specificity of the luciferase reporter vectors driven by the basal CEA promoter regions covering the first three (CEA205) or four (CEA421) DNA-binding elements in combination with or without kB4 enhancer (Figure 1). The transcriptional activities of the kB4 enhancer, CEA basal promoters (CEA205 and CEA421) and chimeric enhancer -promoter elements (kB4-CEA205 and kB4-CEA421) were examined in CRC and normal human endothelial and epithelial cells. As showed in Figure 3A, both CEA205 and CEA421 basal promoters demonstrated selective transcriptional activity in all CRC cell lines but not in normal endothelial and epithelial cells. In combination with kB4, the transcriptional activity of both CEA205 and CEA421 in cancer cell lines was enhanced. The kB4 enhancer demonstrated a significantly greater enhancing effect on the CEA205 promoter than on the CEA421 promoter ( Figure 3A). The kB4 enhancer alone showed very low transcriptional activity in all tested cancer cell lines (Table 1). Only background transcriptional activity of CEA promoters, kB4 enhancer and the chimeric kB4-CEA enhancer -promoter cassette was detected in normal cells ( Figure 3A and Table 1). Furthermore, we compared the transcriptional activity of kB4-CEA205 with a strong SV40 virus promoter. To avoid the transcriptional interference between the SV40 promoter in pRL3-SV40 and  pGL3-SV40, the CRC cell lines were singly transfected with pkB4-CEA205 and pGL3-SV40. Equal amount of protein (10 mg) from each transfection was subjected to luciferase assay. The transcriptional activity of kB4-CEA205 enhancer -promoter system was highly comparable to that of the strong SV40 virus promoter in all CRC cell lines ( Figure 3B). The SV40 promoter also demonstrated high transcriptional activity in a panel of seven normal human cell lines and primary cell cultures. In contrast, the transcriptional activity of pkB4-CEA205 in normal cells was negligible ( Figure 3C).

0 -DFUR cytotoxicity in CRC cell lines transiently transfected with a TP expression vector
Owing to the strong and cancer-specific transcriptional activity, the CEA205 promoter and kB4-CEA205 enhancer -promoter cassette were subjected to further study. In the TP expression vectors, the TP cDNA was placed under the control of the CMV, CEA205, kB4-CEA205 or kB4 promoter. Two CRC cell lines (H630 and RKO) with low TP expression were transiently transfected with pkB4-TP, pCEA205-TP, pkB4-CEA205-TP, pcDNA3-TP or pcDNA3 empty vector. After 48 h transfection, the cells were harvested and subjected to different analysis. High levels of TP protein were detected in both pcDNA3-TP-and pkB4-CEA205-TPtransfected cells. Only very low levels of TP protein were detected in the pCEA205-TP-transfected cells and no TP protein was detected by Western blotting analysis in the pcDNA3-and pkB4-TP-transfected cell lines ( Figure 4A). In line with protein expression levels, the pcDNA3-TP-and pkB4-CEA205-TP-transfected cells also possessed high TP enzyme activity ( Figure 4B). The TP activity and protein expression patterns were highly consistent one another in the transfected cells. The transiently transfected cells were then subjected to MTT analysis. In comparison with the control (pcDNA3 transfected), the pkB4-CEA205-TP-transfected RKO and H630 cells were 7.6 and 10 times, respectively, more sensitive to the 5 0 -DFUR-induced cytotoxicity. The cytotoxicity of 5 0 -DFUR to the pkB4-CEA205-TP-and pcDNA3-TP-transfected cell lines was comparable. In comparison   with pcDNA3-TP-and pkB4-CEA205-TP-transfected cells, the pkB4-TP and pCEA205-TP transfection only slightly increased the cytotoxicity of 5 0 -DFUR to the CRC cell lines ( Figure 4C and D and Table 2).

The bystander effect of pFastBacDualDpromoter/jB4-CEA205-TP-transfected CRC cells on the neighbouring cells
To produce cell lines for the bystander effect analysis, the H630 and RKO cell lines were stably transfected with pFastBacDualD promoter/kB4-CEA205-TP (pFBD/KCT) or pcDNA3-TP. Two positively transfected clones were selected from each cell line for further study. The TP protein expression and TP enzyme activity in the pFBD/KCT-transfected cell lines were high and comparable with that in the pcDNA3-TP-transfected cells ( Figure 5A and B). Furthermore, the cytotoxicity of 5 0 -DFUR to the pcDNA3-TP and pFBD/KCT stably transfected cell lines was determined by MTT analysis. The pFBD/KCT stable transfection significantly enhanced the cytotoxicity of 5 0 -DFUR to both RKO and H630 cell lines (over 28-fold). The IC 50 s of 5 0 -DFUR to the pFBD/KCT-transfected cells were similar to those of pcDNA3-TP-transfected cells (Table 3). Low transfection rate is still one of the major barriers for the success of cancer GDEPT in the clinic. It has been reported that the small portion of CMV promoter-driven TP-transfected cells can effectively convert 5 0 -DFUR into 5-FU, which can be released and performed cytotoxic effect onto the neighbouring cells (bystander effect) (Patterson et al, 1995). Figure 5C and D show that when pcDNA3-TP-or pFBD/KCT-transfected RKO or H630 cells composed only 5% of the whole-cell population, the sensitivity of H630 and RKO cell lines to 5 0 -DFUR was significantly enhanced. If the percentage of the transfected cells was further increased, the IC 50 s of 5 0 -DFUR in these cells were further reduced but the curves significantly levelled off. The bystander effect of pFBD/KCT-and pcDNA3-TP-transfected cells was comparable.

DISCUSSION
As with other cancer-specific promoter systems, low transcriptional activity limits the clinical use of the CEA promoter in GDEPT. It has been reported that the transcriptional activity of CEA promoter is significantly improved by multimerisation of the essential CEA promoter region and/or fusing the CEA promoter with a CEA enhancer sequence. Although the transcriptional activity of CEA promoter is significantly improved by these modifications, all of these approaches are highly cell line and CEA activity dependent. The transcriptional activity of these modified promoter systems remains at very low levels in the low CEAexpressing cancer cells (Richards et al, 1995a;Nyati et al, 2002). High NF-kB activity is a very common feature of cancers (Rayet and Gelinas, 1999). In this study, four linked NF-kB-binding sites in tandem (kB4) were exploited as a cancer-specific enhancer to improve the transcriptional activity of the CEA promoter. The enhancing effect of kB4 on the transcriptional activity of two basal CEA promoters covering the first three (CEA205) and four (CEA421) DNA-binding sites on the CEA promoter region was examined. Both promoters demonstrated comparable transcrip-tional activity, which was further enhanced by kB4 in CRC cell lines. In comparison with that on CEA421, the kB4 enhancer demonstrated stronger enhancing effect on CEA205 promoter ( Figure 3A). The luciferase assay showed that the transcriptional activity of kB4-CEA205 was comparable with that of SV40 promoter ( Figure 3B). Similar to other cancer-related genes, although CEA and NF-kB are highly expressed in cancer cells, the low levels of both are still detectable in some normal tissues (Hammarstrom, 1999). Therefore, we also tested the potential transcriptional leakage of the NF-kB-CEA enhancer -promoter system. In agreement with a previous report (Perkins et al, 1993), the NF-kB-binding sites by itself, as an enhancer element, only possessed very low transcriptional activity. The transcriptional activity in cancer cell lines was highly increased by the combination of kB4 and CEA promoter and meanwhile the cancer-targeting specificity was still remained ( Figure 3A and C and Table 1). Our results demonstrated that kB4 enhancer efficiently improved the transcriptional activity of CEA basal promoter without impairing its cancer specificity. The fidelity of the kB4-CEA205 enhancer -promoter system was therefore improved by the increased transcriptional contrast between the cancer and normal cells.
In this study, TP/5 0 -DFUR was used as a model to test the efficacy of the kB4-CEA205 enhancer -promoter system. Thymidine phosphorylase is a key enzyme for CAP, a 5-FU prodrug (Malet-Martino and Martino, 2002). As the TP activity in various cancers is highly heterogeneous, the clinical response of cancer patients to CAP remains relatively low (Malet-Martino and Martino, 2002). In this study, the TP protein expression was only detected in two CRC cell lines. Transient transfection of pkB4-CEA205-TP and pcDNA3-TP resulted in comparable expression levels of TP protein in human CRC cell lines. In contrast, the pCEA205-TP-transfected cells only expressed TP protein at very low levels. The TP protein expression in these cell lines was reflected by the TP activity assay results. These results indicated that the transcriptional efficacy of kB4-CEA205 was comparable with that of CMV strong virus promoter. In line with the TP protein and enzyme activity results, the MTT assay demonstrated that the cytotoxicity of 5 0 -DFUR to H630 and RKO cell lines was highly sensitised by the transfection of pcDNA3/TP or pkB4-CEA205-TP (7.6 to 14.5-fold, Figure 4C and D and Table 2).
One of the major limitations for the clinical application of GDEPT is the low transfection rate in vivo, which can be overcome by bystander effect. Using strong virus promoters, it has been reported that the 5-FU derived from 5 0 -DFUR can diffuse from the transfected cancer cells to exert bystander killing effect on their neighbour cells in a gap junction-independent manner (Patterson et al, 1995;Kato et al, 1997). In this study, we also compared the bystander effect of pFBD/KCT-and pcDNA3/TP-transfected cancer cell lines. Figure 5C and D shows that the cytotoxicity of 5 0 -DFUR to RKO and H630 cell lines was significantly enhanced by the mixture of only 5% of pcDNA3/TP-or pFBD/KCT-transfected cells. The bystander effect was levelled off when further increasing the TP-transfected cells in the mixture. As well, the bystander effect of pcDNA3/TP and pFBD/KCT transfection was very similar in both cell lines.
The molecular mechanisms of enhancing effect of kB4 on CEA promoter activity in this study were still not clear. Although by definition, enhancer is a DNA element, which efficiently regulates promoter activity distantly and orientation-independently (Banerji et al, 1983;Sassone-Corsi et al, 1983), the relationship between enhancer and promoter elements is far more complex. A cooperative interaction between NF-kB and Sp1 has been demonstrated in human immunodeficiency virus long terminal repeat (HIV-1 LTR) promoter region (Nabel and Baltimore, 1987;Perkins et al, 1993). The cooperation between NF-kB and Sp1 in the HIV-1 LTR is highly space and orientation-dependent. The promoter activity is significantly impaired by the insertion of DNA The data represent the mean IC 50 (mm) of 5 0 -DFUR from three replicates7s.d.
jB4-CEA205 enhancer -promoter system improves GDEPT X Guo et al fragments between the kBand Sp1-binding sites. Mutation of the adjacent kB and Sp1 DNA-binding sites completely abolishes the phorbol myristate acetate inducible transcriptional activity (Perkins et al, 1993). It was also reported that an Sp1-dependent promoter is distantly enhanced by an NF-kB enhancer (Teferedegne et al, 2006). The cooperation between NF-kB and Sp1 needs direct contact of both proteins and other transcription modulators (Perkins et al, 1993;Teferedegne et al, 2006). Chromatin modulation is also involved. Two Sp1 and Sp1-like DNA-binding sites are located on the CEA promoter region (FP2 and FP3) (Chen et al, 1995;Hauck and Stanners, 1995). Therefore, the interaction between NF-kB and Sp1 may also play a role in our kB4-CEA205 enhancer -promoter system. The kB4 enhancer demonstrated higher enhancing effect on CEA205 than CEA421 in most CRC cell lines ( Figure 3A). The distance between the kB4 and the 5 0 Sp1 in kB4-CEA205 and kB4-CEA421 fragments was 29 and 250 nt, respectively. The spatial rule may also partially apply to the enhancer -promoter systems in this study. Sp1 is a common element in cancer-specific promoters, for example CEA, survivin and hTERT (Chen et al, 1995;Hauck and Stanners, 1995; The TP protein (A) and enzyme activity (B) levels in the TP cDNA stably transfected cancer cell lines. pFBD/KCT-1 and pFBD/KCT-2: pFastBacDual/kB4-CEA205-TP clone 1 and clone 2. Vinculin: protein loading control. The bystander killing effect of CMV-or kB4-CEA205-driven TP cDNA transfection was compared in H630 (C) and RKO (D) cells. The curves were derived from three replicates and the bars represent s.d. The data represent the mean IC 50 (mm) of 5 0 -DFUR from three replicates7s.d. C1 and C2: clones 1 and 2.
jB4-CEA205 enhancer -promoter system improves GDEPT X Guo et al Li and Altieri, 1999;Takakura et al, 1999). Our unpublished data showed that kB4 also improved the transcriptional activity of hTERT promoter. Therefore, investigation of the relationship between kB4 and Sp1 might provide evidence for using kB4 to improve the transcriptional activity of other cancer-specific promoters.
The lack of strong cancer-specific promoter system remains one of the major obstacles for the clinical success of GDEPT. This is the first in vitro study using NF-kB DNA-binding site as a cancerspecific enhancer element to improve the cancer-specific transcriptional activity of CEA promoter. The in vivo investigation needs to be carried out to use this enhancer system for GDEPT approach.