Optimizing Radiation-Responsive Gene Promoters for Radiogenetic Cancer Therapy

Gene Therapy (2002) 9, 1396–1402  2002 Nature Publishing Group All rights reserved 0969-7128/02 $25.00 www.nature.com/gt RESEARCH ARTICLE Optimizing radiation-responsive gene promoters for radiogenetic cancer therapy

SD Scott1,2, MC Joiner1,2 and B Marples1,2 1Department of Experimental Radiation Oncology, Gray Cancer Institute, Northwood, Middlesex, UK

We have been developing synthetic gene promoters respon- sequences following irradiation. Alteration of spacing (from sive to clinical doses of ionizing radiation (IR) for use in suic- six to 18 nucleotides) between elements had little effect, as ide gene therapy vectors. The crucial DNA sequences util- did the addition of an adjacent Sp1 binding site. Combining ized are units with the consensus motif CC(A/T)6GG, known the optimum number and sequence of CArG elements in an as CArG elements, derived from the IR-responsive Egr1 additional enhancer was found to produce the best IR induc- gene. In this study we have investigated the parameters tion levels. Furthermore, the improved enhancers also per- needed to enhance promoter activation to radiation. A series formed better than the previously reported prototype when of plasmid vectors containing different enhancer/promoters used in in vitro and in vivo experimental GDEPT. We envis- were constructed, transiently transfected into tumor cells age such enhancers will be used to drive suicide gene (MCF-7 breast adenocarcinoma and U-373MG expression from vectors delivered to a tumor within an glioblastoma) and expression of a downstream reporter irradiated field. The modest, but tight expression described assayed. Results revealed that increasing the number of in the present study could be amplified using a molecular CArG elements, up to a certain level, increased promoter ‘switch’ system as previously described using Cre/LoxP. In radiation-response; from a fold-induction of 1.95 ± 0.17 for combination with targeted delivery, this strategy has great four elements to 2.74 ± 0.17 for nine CArGs of the same potential for significantly improving the efficacy of cancer sequence (for MCF-7 cells). Specific alteration of the core treatment in the large number of cases where radiotherapy A/T sequences caused an even greater positive response, is currently employed. with fold-inductions of 1.71 ± 0.23 for six elements of proto- Gene Therapy (2002) 9, 1396–1402. doi:10.1038/sj.gt.3301822 type sequence compared with 2.96 ± 0.52 for one of the new

Keywords: radiation; promoters; CArG elements; SRF; GDEPT

Introduction to regulate expression of the herpes simplex virus thymidine kinase (HSVtk) suicide gene in experimental gene- Radiotherapy is one of the primary cancer treatment directed enzyme prodrug therapy (GDEPT) assays. Fol- modalities. It also provides a promising means of tar- lowing irradiation with a single 1–5 Gy dose, cell cultures geting the activation of anti-tumour gene therapy vectors. grown in the presence of the prodrug ganciclovir (GCV) Gene promoters activated by clinical doses of ionizing showed significantly greater growth inhibition compared radiation (IR) can provide suitable control over the with single-treatment controls.6 expression of therapeutic genes. Several groups have util- Previous studies have demonstrated that the binding ized enhancer/promoters of radiation-responsive genes of serum response factor (SRF) to CArG elements (known (eg early growth response 1 or Egr1 and p21/WAF1) to also as serum response elements or SREs) in the pro- drive the expression of heterologous genes, including moters of several genes, such as Egr1, c-fos and members 1–5 some specifically for anti-tumor therapy. To improve of the actin family, led to an increase in gene the activation response of cancer gene therapy vectors to expression.7,8 A number of accessory proteins (eg IR doses of clinical relevance, we constructed a synthetic p62/TCF, SAP-1 and Elk-1) have been shown to complex gene enhancer consisting of short DNA motifs of with SRF during this process.9–11 SRF-mediated gene sequence CC(A/T)6GG (CArG elements) derived from regulation occurs, for example, following mitogenic 6 the Egr1 gene. The new enhancer was shown to be more stimulation of cells by phorbol esters contained in serum. IR-responsive than the native Egr1 enhancer when tested Stimuli-driven activation of the Egr1 promoter has been in the same vector context, in MCF-7 breast adenocarcin- shown to require phoshorylation by mitogen-activated oma cells. This synthetic ‘CArG’ promoter was then used protein (MAP) kinases, which are often triggered by cellular stresses such as radiation.12,13 The above findings strongly suggest that SRF complexes play a central role Correspondence: SD Scott, Karmanos Cancer Institute, Room 823 Hud- in CArG-mediated, IR-induced activation of Egr1. son-Webber CRC, 4100 John R, Detroit, Michigan, 48201 USA 2Current address: Karmanos Cancer Institute, Hudson-Webber CRC, Despite some success in experimental GDEPT using 14–17 4100 John R, Detroit, Michigan 48201, USA the native Egr1 enhancer, there is scope to both Received 18 February 2002; accepted 23 May 2002 improve and tailor the activation responses of CArG pro- Radiation-responsive gene promoter SD Scott et al 1397 moters to clinical IR doses. Prior work using deletion mutants of the Egr1 gene promoter, demonstrated that multiple CArG elements gave greater inducibility by serum and specific growth factors than single units.8 Where more than one CArG unit is present in a native gene promoter, they are often of different sequence 18–20 though fitting the overall CC(A/T)6GG consensus. There is evidence that some of these CArG elements are more responsive than others to serum and cellular growth factors8 and also radiation.1 DNAase protection assays conducted on CArG-containing DNA fragments bound with SRF revealed that SRF/coactivator complex binding, although centered on the dyad, extended by up to 10 nucleotides beyond the CArG motif.21 Furthermore, self-regulation of SRF expression via CArG elements in the SRF gene promoter was enhanced by the presence of a binding site for the Sp1 transcription factor.22 In the present study, CArG numbers, core sequences, spacing and context were specifically altered in an attempt to increase enhancer sensitivity to low doses of ionizing radiation and define parameters important for future promoter development. Lastly, in vitro and in vivo Figure 1 Radiation-mediated activation of gene promoters. MCF-7 cells GDEPT assays were used to demonstrate the efficacy of were transiently-transfected (~20% efficiency) with plasmid constructs new generation suicide gene vectors containing the bearing the test promoters upstream of the coding region for the enhanced green fluorescent protein (GFP). Fold increase in GFP expression follow- improved promoters. ing a single 3 Gy dose was measured by FACS assay ~40 h after irradiation. The synthetic promoters tested were as follows: E4, E6, E9 Results and E12 contained four, six, nine and 12 directly repeated CArG elements of the prototype sequence CCTTATTTGG. E6, ns1 and ns2 also contained Influence of CArG element numbers on promoter six directly repeated CArG elements, but of different core A/T sequence. response E6, s6, s10 and s18 contained six CArG elements of the above prototype sequence, but with inter-element spacing of zero, six, 10 and 18 nucleo- To determine the effect that increasing CArG element tides, respectively. Sp1 is the E6 promoter with an adjacent Sp1 transcrip- number would have on the inducibility of synthetic pro- tion factor binding site. The pCIneo parent plasmid (pCI) and the CMV- moters to ionizing radiation (IR), our standard green regulated GFP plasmid (CMV) served as controls. fluorescent protein (GFP) reporter assay was employed.6 In addition to the previously described CArG promoter (containing four directly-repeated elements of identical duced (pE6ns1, pE6ns2), each containing enhancers with sequence; E4),6 new promoters (denoted E6, E9, E12) six directly-repeated CArG elements in which the A/T were tested, carrying six, nine and 12 tandem-repeat cop- core sequences differed from the prototype. These new ies of the prototype CArG sequence (CCTTATTTGG). sequence (ns) CArG motifs are also present in the native Reporter plasmids containing the novel promoter con- Egr1 promoter (ns1, CCATATTAGG; ns2, CCATATAAGG). structs were then transiently transfected into MCF-7 cells, The first new enhancer (E6-ns1) did not show a sig- and promoter activity assayed following a single 3 Gy nificantly different IR response than that containing the dose. Results showed that increasing the number of prototype CArG sequence (E6), with respective fold- elements from four to six and then nine produced a con- inductions of 1.71 (± 0.23) and 1.87 (± 0.20) in MCF-7 cells, current increase in IR-inducibility (Figure 1). However, following a single 3 Gy radiation dose (Figure 1). How- the E12 promoter was found to be less inducible than E9, ever, E6-ns2 gave a substantially higher induction level showing results similar to E4 and E6. This response trend (2.96 ± 0.51) under the same conditions. was seen in all individual experiments. When pooled the To examine how the spatial arrangement of CArG results revealed mean fold-induction rates of 1.95 (± 0.17), elements might affect IR-responsiveness, another series 2.29 (± 0.13), 2.74 (± 0.17) and 1.98 (± 0.12) for E4, E6, E9 of enhancers was constructed. In these vectors, the proto- and E12 respectively (Figure 1). As expected, the parental type CArG element sequences in the E6 enhancer (E6) vector pCIneo, showed no notable fluorescence before or were each separated by six, 10 or 18 nucleotide ‘spacer’ after irradiation (indicated by a fold-induction of 0.64 ± sequences (also derived from inter-CArG sequences in 0.19). Likewise, the strong constitutive cytomegalovirus the Egr1 enhancer), producing the plasmid constructs E6- immediate–early (CMV IE) gene promoter, although pro- s6, E6-s10 and E6-s18. Nevertheless, reporter assays in ducing high basal levels of GFP expression, showed a MCF-7 cells showed that the addition of these spacers negligible increase on irradiation (1.27 ± 0.12 fold- had negligible/little effect on IR-responsiveness. A single induction). Statistical analysis of the pooled results dem- 3 Gy dose produced fold-increases of GFP expression of onstrated a significant (P Ͻ 0.05, Student’s t test) 1.75 ± 0.22 (E6-s6), 1.61 ± 0.28 (E6-s10) and 1.65 ± 0.28 improvement of the E9 promoter over both the E4 proto- (E6-s18), compared with 1.71 ± 0.23 for the prototype E6 type and the other novel CArG promoters. enhancer (Figure 1). To test whether an Sp1 binding site adjacent to CArG Altering CArG element sequences and spacing elements would influence IR-activation, we inserted an To study the effect of CArG sequence alteration on the Sp1 site and intervening sequences derived from the SRF IR-induction response, two new test vectors were pro- promoter directly upstream of the E6 enhancer. However,

Gene Therapy Radiation-responsive gene promoter SD Scott et al 1398 this ‘Sp1-E6’ enhancer proved to be no more responsive than the E6 prototype, with fold IR-induction rates of 1.52 ± 0.12 and 1.71 ± 0.23, respectively (Figure 1). Once CArG number and sequence were defined as important factors in IR-response, an enhancer combining both these features was made. The new construct (E9- ns2) had nine CArG elements of the sequence CCATA- TAAGG. The IR-induction of E9-ns2 was compared with both the prototype E4 and the native Egr1 (N) enhancers, with the CMV IE promoter again as a control. After a single 3 Gy dose, the fold-increase of GFP expression in MCF-7 cells, mediated by wild-type enhancer N was 1.44 ± 0.08, significantly higher at 2.24 ± 0.15 for E4 and great- est at 2.94 ± 0.10 for E9-ns2 (Figure 2). As expected, the CMV promoter construct produced no increase in GFP expression on irradiation (1.14 ± 0.13-fold induction). Finally, the basal (ie unirradiated) expression of the constructs was examined. The number of transfected cells classified as GFP positive by fluorescence was 23.2% (± 0.06) for the CMV construct, 4.5% (± 0.25) for the native Egr1 enhancer (N), 2.9% (± 0.42) for E4 and lowest for E9-ns2 at 2.2% (± 0.02). Thus the synthetic enhancers were notably less ‘leaky’ than the native promoter and the more IR-sensitive E9-ns2 CArG enhancer actually exhibited the least non-specific reporter expression. In vitro GDEPT assays Having determined that both CArG element number and sequence significantly altered IR-induction responses by GFP assay, the most responsive test promoters (E9 and E9ns2) were then used for experimental GDEPT in MCF- 7 and U373-MG tumor cells. The HSVtk/GCV suicide gene system was employed, but using the more effective mutant 30 gene variant, which has been shown to be more effective in GDEPT then the wild-type due to greater GCV affinity.23 The prototype E4 promoter was also included for comparative purposes.6 Cell survival following irradiation and GCV exposure was measured by MTS growth delay assay. Untreated, unmodified (parent) cells were defined as 100% survival. Exposure of unmodified and plasmid-transfected MCF-7 cell samples to GCV did not significantly reduce cell survival (Figure 3). Thus basal (not IR-induced) HSVtk expression from the test promoters (leading to GCV-mediated cell death) was undetectable. As expected, a single 3 Gy irradiation dose reduced cell survival. For MCF-7 cells this was to 82.3 ± 2.0%, and 88.2 ± 2.4% for U373-MG, with similar values for transfected cell populations. The addition of GCV to the irradiation treatment had no significant effect ± Figure 2 Radiation-mediated activation and basal expression of synthetic on unmodified cells, with survivals of 80.7 2.4% and and native promoters in MCF-7 cells. (a) Fold induction of test promoters 91.7 ± 1.1% for MCF-7 and U373-MG, respectively. How- containing enhancer regions from the CMV immediate–early gene ever, for plasmid-transfected cells survival was reduced (CMV), native Egr1 gene (N), four prototype sequence CArG elements to 71.7 ± 2.2% for the E4 enhancer, 65.7 ± 1.2% for E9 and (E4) or nine CArGs of new sequence (E9ns2). (b) Basal expression from 61.9 ± 1.7% for E9-ns2. The same trend was seen for U373- un-irradiated test promoters. Percentage numbers of ‘fluorescence positive’ cells measured (see Ref. 6). MG cells, though with generally lower levels of cell kill (Figure 3). Plasmid transfection efficiencies were ~20% in MCF-7 and 10–15% for U373 cells. mouse U87-MG glioblastoma xenograft model was Thus these GDEPT experiments demonstrate that the employed. Preliminary, qualitative assessments were nine CArG-element promoters were more effective at conducted using GFP reporter constructs. Xenografts driving IR-responsive gene expression than both the were established using stable cell lines containing the prototype E4 and native Egr1 promoter, with E9-ns2 prototype pE4GFP plasmid. When the tumors had showing the best response. reached ~3 mm in diameter, they were irradiated with 3 Promoter function in vivo × 3 Gy (24 h intervals). After a further 48 h, the tumors In order to test whether synthetic CArG promoters were were excised and examined by fluorescent microscopy. IR-inducible in an in vivo context, an established nude GFP fluorescence was clearly visible in irradiated

Gene Therapy Radiation-responsive gene promoter SD Scott et al 1399 by GCV treatments (8.4 ± 1 day with no GCV and 9.6 ± 3 days with GCV). However, compared with the results obtaining using U87-MG xenografts, GCV caused a sig- niﬁcant tumor regrowth delay for irradiated U87- MG/pE9HSVtk tumors. Mean regrowth times for irradiated tumors containing pE9HSVtk were 33.4 ± 5 days without GCV and 44.3 ± 4 days when combined with GCV treatment (statistical signiﬁcance P Ͻ 0.01, Wil- coxon rank sum test). Thus, only when the IR-responsive suicide gene construct was present in the tumor was GCV able to enhance radiation treatment. These results demonstrate that IR-responsive CArG promoters function in vivo as well as in vitro.

Discussion

The promoter sequence motif CC(A/T)6GG, known as the CArG element, has been shown to have a pivotal role in the radiation-induced expression of the Egr1 gene.1 This response is mediated, at least in part, by reactive oxygen species produced by ionizing radiation and oxi- dative chemicals such as hydrogen peroxide.24 The precise mechanism of this response, and the transcription factors interacting with CArG elements, has yet to be elucidated. However, complexes of phosphorylated transcription factors and accessory proteins, including serum response factor (SRF), are known to promote gene expression by binding to CArG motifs (also known as serum response elements or SREs). This has been noted in the promoters of the Egr1 (previously known as zif 268) and c-fos genes, both of which are IR-responsive.8,25 Egr1 enhancer deletion studies also showed that multiple CArG elements conferred greater promoter inducibility than single elements to both SRF and IR.1,8 In the current study, involving the assay of a series of synthetic promoter constructs, we demonstrate that increasing the numbers of tandemly repeated CArG elements in the test enhancer from four to six and then Figure 3 IR-mediated experimental GDEPT using synthetic CArG pro- to nine improves the induction response to low doses of moters. MCF-7 and U373MG (parent) tumor cells were transiently trans- radiation. This may be by simply providing more target fected with plasmid constructs containing the HSVtk coding region under sequences, presented in a suitably accessible configur- control of enhancers containing four (E4) and nine (E9) prototype sequence CArG elements or nine CArGs of new sequence (E9ns2). For ation, for the efficient binding of specific transcription each transfectant, four subsequent treatments are shown; sham- factor complexes, thus driving gene expression. How- irradiation/no GCV, sham-irradiation/GCV, irradiation/no GCV and sin- ever, when the 12 CArG element enhancer was used, the gle 3 Gy irradiation/GCV. Growth inhibition (MTS assay) is shown as IR-response was no more than for four and six units. % regrowth (growth fraction) of treated cultures. Thus, the continual addition of CArG elements may not necessarily increase enhancer IR-response. X-ray crystal- lographic studies have shown that the binding of SRF to samples, with none apparent in unirradiated control CArG sequences causes ‘bending’ of the DNA helix as it tumors (data not shown). To determine promoter utility wraps around the protein.26 Cumulative changes in sec- in GDEPT, the same model was adopted, using the E9 ondary structure caused by occupancy of multiple bind- promoter to drive HSVtk expression. Tumors were per- ing sites, although initially increasing corresponding mitted to grow to a diameter of ~3 mm then irradiated gene expression, might eventually lead to down-regu- (3 × 5 Gy) or sham-treated, with or without systemic GCV lation. In this case, sufficient spacing between elements administration (see Materials and methods for experi- may alleviate this restriction. Interestingly, the influence mental details). For the parental U87-MG tumors, GCV of responsive element number has also been reported in inoculation did not affect the regrowth of sham- studies using DNA sequences responsive to hypoxia irradiated tumors. The mean time taken to reach a tumor (hypoxia responsive elements; HREs). For example, volume of 500 mm3 following sham-irradiation was 17.3 although no significant difference in induction response ± 4 days compared with 15.4 ± 6 days when GCV was was observed when the number of HREs (from the also given. As expected, irradiation increased this erythropoietin gene) was increased from three to six cop- regrowth time to a mean of 47.6 ± 7 days, and GCV ies, an increase to nine copies had a marked positive administration did not significantly affect this (44.5 ± 9 effect.27 Conversely, a saturation effect was seen when days). For sham-irradiated U87-MG/pE9HSVtk xeno- more than five copies of the vascular endothelial growth grafts, mean tumor regrowth time was again unaffected factor HRE were employed in test promoters.28 Further-

Gene Therapy Radiation-responsive gene promoter SD Scott et al 1400 more, in our recent studies involving hypoxia-induction rounding CArG motifs appears to have little importance of synthetic HRE enhancers derived from phosphoglycer- in determining activity status. In native promoters, the ate kinase 1 promoter sequences, the use of 10 direct- sequence around active units varies considerably, and repeat units was found to be less inducible than five.29 even studies with synthetic targets for SRF showed that it Interestingly, our results also reveal that increasing the was the CArG consensus rather than adjacent sequences number of CArG elements in the synthetic promoters which were critical to binding.31 does not lead to a corresponding increase in non-specific The expression of the SRF gene itself is also partially gene expression. Conversely, synthetic enhancers with self-regulated via CArG elements situated in its own pro- more IR-responsive units showed progressively less moter.19 However, maximal serum responsiveness ‘leakiness’ under non-irradiated conditions, all of which requires the presence of the Sp1 binding site immediately were less than for the native Egr1 enhancer. Indeed, the adjacent to one of the three CArG elements.22 The Egr1 in vivo data reveal no detectable leakiness (no increased promoter contains several potential Sp1 binding sites, one GCV-mediated cell kill for U87-MG/E9HSVtk compared of which is located between two IR-responsive CArG with parental U87-MG xenografts). In a clinical setting, elements. Nevertheless, the placing of an Sp1 binding site this important finding could considerably enhance differ- in the E6 synthetic promoter (using the identical spacing ential expression of a suicide gene delivered to the and sequence context seen in the SRF promoter) did not irradiated tumor mass compared with any vector DNA yield an increase in IR-response. delivered to the surrounding unirradiated tissue. Finally, GDEPT experiments with the synthetic pro- Our data on the influence of CArG sequence on IR- moters agreed with the results of the GFP reporter assays. response reveal that different elements, although fitting The in vitro studies also showed that both the number to the CC(A/T)6GG consensus motif, exhibit diverse lev- of CArG elements and their core A/T sequences were els of inducibility. Previous studies on the affinity of SRF important factors in determining promoter response to binding to oligonucleotides bearing various CArG low IR doses. As anticipated, the animal model data elements, showed distinct differences depending on the reveal that synthetic CArG promoters also function in precise A/T core sequences.8,30 Specifically, the ATATAA vivo, and are able to significantly contribute to tumor core sequence (equivalent to our ns2) was found to have growth control through IR-mediated regulation of a more SRF affinity than the ATATTA core sequence (ns1 GDEPT system. Although in vitro and in vivo CArG pro- in this study, also known to be IR-responsive in the c- moter effects are quantitatively modest (though appar- fos SRE) in concordance with the IR-responses presented ently with little basal expression), if combined with an here. Datta et al1 also showed that the three proximal appropriate expression amplification and maintenance CArG elements in the native Egr1 promoter did not con- system such as described in Scott et al,32 the resultant tribute to the IR response, in contrast to the three distal level of therapeutic product expression could be suf- elements. Similarly, both our prototype and ns2 CArG ficient to elicit high-level tumor cytotoxicity. By this sys- sequences are identical in sequence to distal Egr1 CArGs. tem, following activation of the vector via the IR-respon- Nevertheless, the ns1 element, which we also found to sive promoter, the role of regulating therapeutic gene be IR-responsive, has the same sequence as one of the expression is ‘transferred’ to another promoter. In our proximal, reportedly non-responsive units in the native previous study,32 the strong, constitutive CMV IE pro- Egr1 promoter.1 However, the ns1 CArG sequence was moter drove suicide gene expression, however additional shown to be the SRF-responsive in an earlier study.8 Thus levels of control could be achieved by using alternative it is possible that sequence context and spatial pos- promoters (eg cell type specific). Consequently, even fol- itioning may be important in dictating CArG responses. lowing a systemic vector delivery, only vector molecules With a view to reducing potential steric hindrance delivered to the correct target cell type, which also fall between transcription factor complexes binding to CArG within the irradiated field would be activated during elements, we introduced ‘spacer’ sequences between the radiotherapy. We believe a new generation of ‘radiogene- units to determine if this could enhance IR-response. It tic’ suicide gene vectors based upon these principles has is noteworthy that DNAse protection experiments carried great potential in the future of radiotherapeutic treat- out on the c-fos SRE demonstrated that SRF binds asym- ments for a wide range of cancer conditions. metrically, engaging the DNA over a region, which extends approximately 10 bases upstream and 4 bases downstream of the CArG motif.7,21 However, in this Materials and methods study the introduction of up to 18 nucleotides between units in our synthetic promoter did not improve IR- Cell culture and transfection inducibility. It is possible that even 18 bp may not be Human MCF-7 mammary adenocarcinoma, U87-MG and sufficient inter-element spacing for transcription factor U373-MG glioblastoma cells (European Collection of Cell complexes to bind to adjacent CArG units, particularly if Cultures, Salisbury, UK) were maintained in RPMI binding causes bending of the DNA helix, as is the case medium (Life Technologies, Paisley, UK) supplemented for SRF.26 For native enhancers (eg c-fos) multiple CArG with 10% fetal calf serum, 2 mM L-glutamine (Life elements may be unnecessary for efficient function. Those Technologies), and incubated in a humidified incubator ° genes that possess several units (eg Egr1) display notable at 37 C in 5%O2:5%CO2:90%N2. Cells were routinely inter-element spacing, with no more than two CArG tested and found free of mycoplasma infection (MycoTect motifs in close proximity (eg 6 bp18,20). Further work will kit, Life Technologies). Transient cell transfectants for all be needed to determine if IR-responses for synthetic plasmid constructs were produced by using Lipofectam- CArG enhancers could be further improved with an ine (Life Technologies) by the method described in Ref. increase in element spacing in line with that seen in the 6, when cells were at ~60% confluence. In each experi- native promoters (eg 28 bp in Egr1). The sequence sur- ment, a plasmid in which the CMV IE gene promoter

Gene Therapy Radiation-responsive gene promoter SD Scott et al 1401 controlled GFP reporter gene expression was used to GATCT(CCATATAAGG)9GCGAT-3’ and 5’-CGC(CCTT determine transfection efficiency. The U87- ATATGG) 9A-3’. Finally, the Sp1E6 enhancer was cloned MG/pE9HSVtk cell line used in tumor xenografts was in a single step using the following ODN pair in the established by initial plasmid transfection followed by unique BglII site immediately upstream of E6; 5’-GA enrichment in neomycin (0.5 mg/ml) to ensure each cell TCTCCAATGGGGCGGGGGCGCTGGGGCTCG-3’ and in the population carried the construct. 5’-GATCCGAGCCCCAGCGCCCCCGCCCCATTGGA-3’. Therefore, in the resultant clone the CCAAT box is Vector construction present and the sequence between the adjacent Sp1 site The pCineo plasmid (Promega, Southampton, UK) was (GGGCGG) and the downstream CArG element is used as the basis for all new constructs. All restriction exactly that seen in the native SRF promoter.19 Cloning and modifying enzymes were supplied by either Pro- of the native Egr1 enhancer (N) is described in Ref. 6. In mega or Life Technologies and used according to the all cases where ODNs were ligated into vectors via manufacturer’s instructions. All DNA isolation and puri- unique restriction sites, vector DNA was pre-treated with fication was carried out using appropriate kits from calf intestinal alkaline phosphatase. The sequence integ- Qiagen (Crawley, UK). Cloning methodologies for the rity of all plasmids was confirmed using a Thermo GFP and mutant 30 HSVtk,23 a kind gift of Dr Margaret Sequenase Cycle Sequencing Kit (Amersham Pharmacia Black, University of Washington) coding regions, and Biotech, Amersham, UK) and a Gene Readir DNA Ana- custom-made oligodeoxyribonucleotides (ODNs; from lyzer (LI-COR, Lincoln, USA). MWG-Biotech, Milton Keynes, UK) enhancers are described in Ref. 6. The six CArG element enhancer (E6) Irradiation of cells was cloned using the complementary ODN pair: 5’- Twenty-four hours after transfection, cells were ° GATCT(CCTTATTTGG)6GCGAT-3’ and 5’-CGC(CC irradiated at 37 C using a 240 kVp X-ray source at a dose AAATAAGG)6A-3’. Some enhancers had to be cloned in rate of 0.44 Gy/min. In all experiments, mock-irradiated stages, due to CArG rearrangement problems, revealed controls were also used. Following radiation treatment, by sequencing, when attempts were made to introduce cells were reincubated and gene expression assessed ~36 too many units in a single cloning step. For example, to h later. clone E9, the following ODN pair was cloned immediately upstream of E6 at the unique BglII site; 5’- Reporter gene and tumor cell growth assays

GAT(CCTTATTTGG)3A-3’ and 5’-GATCTC(CCAAA Detailed methodologies for both assays are given in Ref. TAAGG)2CCAAATAAG-3’. To produce E12, three stages 6. Both MCF-7 and U373MG cells were treated similarly. were needed after the failure of alternative strategies. Data given for reporter assays is from between three and ± First, the 5’-GATCT(CCTTATTTGG)3CGAT-3’ and six independent experiments ( s.d.), each containing two GATCT(CCAAATAAGG)3CGC ODN pair was cloned to three replicates. Modiﬁcations to the published GDEPT into pCIneo using BglII and SgfI. The next pair, cloned at protocol are as follows. GCV (Cymevene) was obtained

the unique SgfI site, was 5’-CGC(CCTTATTTGG)6 from Roche (Welwyn Garden City, UK) and used at 50 ␮ GCGAT-3’ and 5’-CGC(CCAAATAAGG)6GCGAT-3’. M in media. Cells were irradiated 3 h after plating, with The ﬁnal pair, cloned at the BglII site, was 5’- cell growth determined after a further 3 days. Forty ␮l GATCT(CCTTATTTGG)3A-3’ and GATCT(CCAAA of a 1.8 mg/ml solution of MTS [3-(4,5-dimethylthiazol- TAAGG)3A. The new sequence (ns) CArG element clones 2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulphophenyl)-2H- were made in single steps using the following ODNs; 5’- tetrazolium] solution (containing phenazine metho- GATCT(CCATATTAGG)6GCGAT-3’ and 5’-CGC(CCT sulphate) in phosphate-buffered saline (PBS; Promega) AATATGG)6A-3’ (for ns1) and GATCT(CCATAT was added to the culture medium and re-incubated for AAGG)6GCGAT-3’ and 5’-CGC(CCTTATATGG)6A-3’ 4 h. Plates were scanned on a multiplate reader (Flow) (for ns2). Enhancers containing spacers between each at 490 nm. Mean plating efﬁciency (PE) values for each CArG elements were made by cloning the following untreated cell line was used to calculate measures of ODN pairs into pCIneo using BglII and SgfI; 5’- tumor cell killing and growth inhibition (represented as GATCT(CCTTATTTGGGCAGCA)5CCTTATTTGGCGAT- growth fraction on the survival plot; Figure 3) for each 3’ and 5’-CG(CCAAATAAGGTGCTGC)5CCAAATAA treated sample. A mean value of growth fraction was GGA-3’ for E6s6, and 5’-GATCT(CCTTATTTGGGCA then calculated from at least two individual experiments,

GCAGCAG)5CCTTATTTGGCGAT-3’ and 5’-CG(CCA each containing a minimum of two replicates. AATAAGGCTGCTGCTGC)5CCAAATAAGGA-3’ for E6s10. The six-nucleotide spacer sequence (GCAGCA) Experimental animal models was derived from a spacer found between CArG U87-MG and U87-MG/pE9HSVtk cells for implantation elements in the native Egr-1 promoter.18 The other spa- were washed twice in Hank’s buffered salt solution by cers were multiples of this unit. For E6s18, two cloning centrifugation and resuspended to a ﬁnal concentration stages were needed. First, the 5’-GATCTGCA of 107 cells/ml. For each group of 10 female BALB/c GCA(CCTTATTTGGGCAGCAGCAGCAGCAGCA)2CC nude mice, whole body irradiation (5 Gy) was performed TTATTTGGCGAT-3’ and 5’-CG(CCAAATAAGGTGCTG 24 h before implantation to improve tumor take rate.

CTGCTGCTGCTGC)2CCAAATAAGGTGCTGCA-3’ were Xenografts were established in nude mice by a single cloned into pCIneo via the BglII and SgfI sites. Then a subcutaneous inoculation of 100 ␮l cell suspension into second ODN pair was cloned into the BglII site; 5’- the ﬂank of the animal using our established protocol.33 GATCT (CCTTATTTGGGCAGCAGCAGCAGCAGCA)2 When tumors reached a mean diameter of ~3 mm, they CCTTATTTGGGCAGCAG-3’ and 5’-GATCCTGCTGC were assigned to one of three GCV dose groups (non- (CCAAATAAGGTGCTGCTGCTGCTGCTGC)2CCAAAT dosed, vehicle-dosed and GCV-dosed). GCV (20 mg/kg AAGGA-3’. E9ns2 was cloned using one ODN pair; in sterile PBS17 was administered by 100 ␮l intraperi-

Gene Therapy Radiation-responsive gene promoter SD Scott et al 1402 toneal injections twice daily for 10 days. The tumors in 12 Lim CP, Jain N, Cao X. Stress-induced immediate–early gene, the vehicle-control group were similarly injected with egr-1, involves activation of p38/JNK1. Oncogene 1998; 16: PBS alone. Animals from each GCV dosing group were 2915–2926. then randomly divided in radiation or sham-irradiation 13 Meyer RG, Kupper JH, Kandolf R, Rodemann HP. Early growth response-1 gene (Egr-1 ) promoter induction by ionizing radi- treatment cohorts. On days 3, 4 and 5 of GCV/vehicle ation in U87 malignant glioma cells in vitro. Eur J Biochem 2002; dosing, the animals were placed in a lead restraining jig 269: 337–346. apparatus (to shield the animal) and the tumor volume 14 Hallahan DE et al. Spatial and temporal control of gene therapy irradiated with 5 Gy of 240 kVp X-rays or ‘sham- using ionizing radiation. Nat Med 1995; 1: 786–791. irradiated’ (jigged, placed on the X-ray set for the stan- 15 Seung LP et al. Genetic radiotherapy overcomes tumor resist- dard period, but not irradiated). Tumor size was determ- ance to cytotoxic agents. Cancer Res 1995; 55: 5561–5565. ined in three planes using calipers, twice weekly 16 Staba MJ et al. Adenoviral TNF-alpha gene therapy and radi- throughout the experiment.33 Individual tumor growth ation damage tumor vasculature in a human malignant glioma rates and volumes were then mathematically determined xenograft. Gene Therapy 1998; 5: 293–300. by polynomial fitting. 17 Kawashita Y et al. Regression of hepatocellular carcinoma in vitro and in vivo by radiosensitizing suicide gene therapy under the inducible and spatial control of radiation. Hum Gene Ther 1999; 10: 1509–1519. Acknowledgements 18 Sakamoto KM. 5’ upstream sequence and genomic structure of The work described in this study was initiated at the the human primary response gene, EGR-1/TIS8. Oncogene 1991; Gray Cancer Institute, Northwood, Middlesex, UK, while 6: 867–871. 19 Spencer JA, Misra RP. Expression of the serum response factor funded by the Gray Laboratory Cancer Research Trust gene is regulated by serum response factor binding sites. J Biol and the Cancer Research Campaign (CRC project grants Chem 1996; 271: 16535–16543. SP2490/0101 and SP2500/0101). The authors would also 20 Schwachtgen JL, Campbell CJ, Braddock M. Full promoter like to thank Mrs Sara Bourne, Ms Amanda Walker, Mr sequence of human early growth response factor-1 (Egr-1): dem- Peter Boulton and Mrs Barbara Joiner for their excellent onstration of a fifth functional serum response element. DNA technical support. Seq 2000; 10: 429–432. 21 Prywes R, Roeder RG. Purification of the c-fos enhancer-binding protein. Mol Cell Biol 1987; 7: 3482–3489. References 22 Spencer JA, Misra RP. Expression of the SRF gene occurs through a Ras/Sp/SRF-mediated- mechanism in response to 1 Datta R et al. Ionizing radiation activates transcription of the serum growth signals. Oncogene 1999; 18: 7319–7327. EGR1 gene via CArG elements. Proc Natl Acad Sci USA 1992; 89: 23 Kokoris MS, Sabo P, Adman ET, Black ME. Enhancement of 10149–10153. tumor ablation by a selected HSV-1 thymidine kinase mutant. 2 Weichselbaum RR et al. Gene therapy targeted by radiation pref- Gene Therapy 1999; 6: 1415–1426. erentially radiosensitizes tumor cells. Cancer Res 1994; 54: 24 Datta R et al. Reactive oxygen intermediates target CC(A/T)6GG 4266–4269. sequences to mediate activation of the early growth response 1 3 Joki T, Nakamura M, Ohno T. Activation of the radiosensitive transcription factor gene by ionizing radiation. Proc Natl Acad EGR-1 promoter induces expression of the herpes simplex virus Sci USA 1993; 90: 2419–2422. thymidine kinase gene and sensitivity of human glioma cells to 25 Treisman R. Identification of a protein-binding site that ganciclovir. Hum Gene Ther 1995; 6: 1507–1513. mediates transcriptional response of the c-fos gene to serum fac- 4 Worthington J et al. Modification of vascular tone using iNOS tors. Cell 1986; 46: 567–574. under the control of a radiation-inducible promoter. Gene Ther- 26 Pellegrini L, Tan S, Richmond TJ. Structure of serum response apy 2000; 7: 1126–1131. factor core bound to DNA. Nature 1995; 376: 490–498. 5 Nuyts S et al. Radio-responsive recA promoter significantly 27 Ruan H et al. A hypoxia-regulated adeno-associated virus vector increases TNFalpha production in recombinant clostridia after for cancer-specific gene therapy. Neoplasia 2001; 3: 255–263. 2 Gy irradiation. Gene Therapy 2001; 8: 1197–1201. 28 Shibata T et al. Enhancement of gene expression under hypoxic 6 Marples B et al. Development of synthetic promoters for radi- conditions using fragments of the human vascular endothelial ation-mediated gene therapy. Gene Therapy 2000; 7: 511–517. growth factor and the erythropoietin genes. Int J Radiat Oncol 7 Treisman R. The SRE: a growth factor responsive transcriptional Biol Phys 1998; 42: 913–916. regulator. Semin Cancer Biol 1990; 1:47–58. 29 Greco O et al. Novel chimeric gene promoters responsive to 8 Christy B, Nathans D. Functional serum response elements hypoxia and ionizing radiation. Gene Therapy 2002 (in press). upstream of the growth factor-inducible gene zif268. Mol Cell 30 Pollock R, Treisman R. A sensitive method for the determination Biol 1989; 9: 4889–4895. of protein-DNA binding specificities. Nucleic Acids Res 1990; 18: 9 Shaw PE, Schroter H, Nordheim A. The ability of a ternary com- 6197–6204. plex to form over the serum response element correlates with 31 Treisman R, Marais R, Wynne J. Spatial flexibility in ternary serum inducibility of the human c-fos promoter. Cell 1989; 56: complexes between SRF and its accessory proteins. EMBO J 563–572. 1992; 11: 4631–4640. 10 Dalton S, Treisman R. Characterization of SAP-1, a protein 32 Scott SD et al. A radiation-controlled molecular switch for use recruited by serum response factor to the c-fos serum response in gene therapy of cancer. Gene Therapy 2000; 7: 1121–1125. element. Cell 1992; 68: 597–612. 33 Kjellen E et al. A therapeutic benefit from combining normobaric 11 Hill CS et al. Functional analysis of a growth factor-responsive carbogen or oxygen with nicotinamide in fractionated X-ray transcription factor complex. Cell 1993; 73: 395–406. treatments. Radiother Oncol 1991; 22:81–91.

Gene Therapy