adenoviruses with tcf binding sites in multiple early promoters show

Gene Therapy (2002) 9, 270–281 2002 Nature Publishing Group All rights reserved 0969-7128/02 $25.00

www.nature.com/gt

RESEARCH ARTICLE

Adenoviruses with Tcf binding sites in multiple earlypromoters show enhanced selectivity for tumour cellswith constitutive activation of the wnt signallingpathway

C Fuerer and R IggoOncogene Group, Swiss Institute for Experimental Cancer Research (ISREC), Epalinges, Switzerland

Mutation of the adenomatous polyposis coli and �-cateningenes in colon cancer leads to constitutive activation of tran-scription from promoters containing binding sites for Tcf/LEFtranscription factors. We have constructed adenoviruseswith Tcf binding sites in the early promoters, in order to tar-get viral replication to colon tumours. Tcf regulation of theE1A promoter confers a 100-fold selectivity for cells withactivated wnt signalling in viral burst and cytopathic effect

Keywords: APC; beta-catenin; Tcf; E1A; colon cancer; oncolytic adenovirus

IntroductionMany oncogenic defects in human tumours target signaltransduction pathways resulting in pathological changesin the activity of transcription factors. These changes canbe targeted by DNA viruses containing promoters regu-lated by the defective pathway. Tumour-specific pro-moters have been used to regulate the expression of suic-ide genes, proapoptotic genes and essential viral genes.The therapeutic efficacy of non-replicating virusesexpressing suicide genes is limited by the inability clini-cally to infect a large fraction of tumour cells withinjected viruses. Expression of essential viral genes fromtumour-specific promoters results in tumour-specificviral replication.1–4 In principle, viruses of this type canreplicate and spread throughout the tumour mass untilall of the tumour cells are killed. Replicating adeno-viruses with defects in early genes which can be comp-lemented by p53 and Rb defects in tumour cells have alsobeen described.5–8 In practice, only a few cycles of reinfec-tion can occur before the immune system halts the infec-tion.9 Even a single cycle of infection should lead to amassive local increase in virus concentration within thetumour, making it possible to achieve the same level ofinfection of tumour cells after injecting much smalleramounts of replicating than non-replicating viruses. Sincethe toxicity of adenoviruses is closely linked to theamount of virus injected, the risk of immediate life-

Correspondence: R Iggo, ISREC, Ch des Boveresses 155, CH-1066 Epa-linges, SwitzerlandReceived 1 October 2001; accepted 11 December 2001

assays. p300 is a coactivator for �-catenin, and E1A inhibitsTcf-dependent transcription through sequestration of p300,but mutation of the p300 binding site in E1A leads to a 10-fold reduction in cytopathic effect of all of the Tcf-regulatedviruses. When Tcf sites are inserted in the E1A, E1B, E2and E4 promoters the viruses show up to 100 000-fold selec-tivity for cells with activated wnt signalling.Gene Therapy (2002) 9, 270–281. DOI: 10.1038/sj/gt/3301651

threatening reactions is potentially much lower withreplicating viruses.10

Recently, our group described adenoviruses that repli-cate in response to activation of the wnt signalling path-way.11 The rationale for the development of these viruseswas that wnt signalling is pathologically activated in vir-tually all colon tumours12 and this leads to transcriptionfrom promoters containing Tcf binding sites.13 The consti-tutive activation of the wnt pathway is caused bymutations in the APC, axin and �-catenin genes thatabrogate GSK-3� phosphorylation of �-catenin and itssubsequent degradation by the proteasome.12 Cytoplas-mic �-catenin enters the nucleus, where it can associatewith members of the Tcf/Lef family of transcription fac-tors and activate transcription of wnt target genes, suchas c-myc, cyclin D1, Tcf1 and matrilysin.14–17 Othergroups have used Tcf promoters to regulate theexpression of suicide genes.18,19 We placed Tcf bindingsites in the adenovirus E2 promoter, which regulatesexpression of the viral replication genes, becausemutations elsewhere in the virus or cell cannot bypassthe absolute requirement for E2 gene products in viralreplication. In order to achieve tight regulation of E2 tran-scription, the adjacent E3 enhancer was mutated. Tcf siteswere also placed in the E1B promoter, although the levelof regulation achieved did not affect viral replication invitro. These ‘Tcf’ viruses showed a 50- to 100-folddecrease in replication in nonpermissive cell lines,whereas their activity was comparable to wild-type Ad5in many, but not all, colon cancer cell lines.11 The remain-ing colon cell lines were semipermissive for the Tcfviruses.

We have tested two different approaches to improve

Wnt targeting replicating adenovirusesC Fuerer and R Iggo

271

Figure 1 Changes in the adenoviral early promoters. (a) Schematic diagram showing the mutagenesis of the E1A promoter (upper part) and E4 promoter(lower part). Both regions are shown from the ITRs to the beginning of the first open reading frame. The dark triangles represent the A motifs in thepackaging signal. (b) Schematic diagram showing mutant regions in the viruses used in this study (see Table 1 for details). To facilitate interpretationof the Figures, the viruses are given clone names (vCFs and vMBs) and a codename summarising their structure: A, B, 2, 4 = Tcf sites in the E1A,E1B, E2, and E4 promoters, respectively. 3 = silent mutations in the NF1, NF�B, AP1, and ATF sites in the E3 promoter. 3� = as 3, but without theATF site mutation. � = deletion of amino acids 2-11 in E1A that abolishes p300 binding.

our existing Tcf viruses and to render them active in abroader range of colon cell lines: mutation of the p300binding site in E1A, and insertion of Tcf sites in the E1Apromoter. Mutation of the p300 binding site in E1A par-tially relieved E1A-mediated repression of Tcf-dependenttranscription, but in the context of the virus this mutationdid not lead to increased transcription from the Tcf-E2promoter and actually reduced the activity of the virus.Similar attenuation by mutation of the aminoterminus ofE1A has been reported by the Onyx group.5 Mutation ofall early promoters resulted in highly selective viruses.The new virus containing only the Tcf changes in the E1Aand E4 promoters (vCF11) was selective for cells withactive wnt signalling and active in most of the coloncancer cells studied.

Results

E1A promoter mutationsThe wild-type E1A enhancer contains two types of regu-latory element, termed I and II,20 which overlap the pack-

Gene Therapy

aging signal (Figure 1a). In addition to these elements,there are transcription factor binding sites in the invertedterminal repeat (ITR) and close to the E1A TATA box.To produce a tightly regulated E1A promoter respondingonly to wnt signals, half of the ITR, the E1A enhancerand the packaging signal were deleted and replaced withfour Tcf binding sites. The resulting E1A promoter con-tains four Tcf sites and a TATA box (Figure 1a). To main-tain the symmetry of the terminal repeats and preservethe ability of the two ITRs to anneal during viral DNAreplication, three Tcf sites were inserted in the right ITR.The packaging signal was also inserted at the right endof the genome to permit proper encapsidation of viralDNA. The changes in the ITR do not affect the minimalreplication origin.21 Adenoviral genomic DNA wasmutagenised in yeast and converted to virus in C7 cells22

expressing a stable �-catenin mutant. Primary virusstocks were plaque purified and expanded on SW480cells. The E1A/E4 mutant viruses grew readily on SW480cells, indicating that the ITR mutagenesis and exchangeof the packaging signal are compatible with the pro-


272

Gene Therapy

Table 1 Description of the recombinant adenoviruses used in this study

Virus Mutant Promoters ORFname regionsa E1A

E1A E1B E2 E3 E4

vCF11 A4 Tcfb wt wt wt mutc wtvCF42 A�4 Tcf wt wt wt mut �p300d

vMB31 B23� wt Tcf Tcf mut+Ae wt wtvCF22 AB23�4 Tcf Tcf Tcf mut+A mut wtvMB19 B23 wt Tcf Tcf mut-Af wt wtvCF81 �B23 wt Tcf Tcf mut-A wt �p300vCF62 A�B234 Tcf Tcf Tcf mut-A mut �p300

aAbbreviations used in this paper (see legend of Figure 1 for details).bReplacement of endogenous promoters by four Tcf binding sites.cInsertion of three Tcf binding sites and the packaging signal upstream of the endogenous promoter.dDeletion of amino acids 2-11 in E1A.eMutation of the NF1, NF�B, and AP1 sites in the E3 promoter.fMutation of the NF1, NF�B, AP1, and ATF sites in the E3 promoter.

duction of viable virus. The structure of the viruses usedin this study is summarised in Table 1 and shownschematically in Figure 1b.

Tcf-E1A promoter virusesTo determine whether the Tcf-E1A promoter responds toactivation of the wnt pathway, cMM1 cells were infectedwith vCF11, the virus with only the E1A/E4 promoterchanges. cMM1 cells are a clone of H1299 lung cancercells expressing �N-�-catenin from a tetracycline-regu-lated promoter. Wnt signalling was activated by removalof tetracycline from the medium (Figure 2, lanes 5–8, �N-�-catenin). This had no effect on E1A expression by wild-type Ad5, but induced expression of E1A by vCF11(Figure 2, compare lanes 3 and 7, E1A). Since DBP isexpressed from the normal E2 promoter in vCF11, theDBP level should rise following activation of wnt signal-ling, because the normal E2 promoter is activated by E1A.The promoter was weakly active in the absence of E1Ain H1299 cells, and showed a moderate and reproducibleincrease in activity following induction of �N-�-cateninexpression (Figure 2, lanes 3 and 7, DBP). We concludethat the mutant E1A promoter responds to activation of

Figure 2 The Tcf-E1A promoter responds to activation of wnt signalling.Western blot of cMM1 cells probed for E1A and DBP 24 h after infectionwith wild-type Ad5 and Tcf viruses. Tetracycline withdrawal leads toexpression of �N-�-catenin in these cells (‘-tet’).

the wnt pathway, and this feeds through to an effect onexpression of viral replication proteins.

The effect of the Tcf-E1A/E4 promoter substitutionswas then tested on a panel of colon cell lines with activewnt signalling: SW480, ISREC-01 and HT29 have mutantAPC; Hct116 has mutant �-catenin; and Co115 hasmicrosatellite instability, but the defect in wnt signallinghas not been defined.23 Three control cell lines with inac-tive wnt signalling were tested: H1299, HeLa and lowpassage normal human small airway epithelial cells(SAEC). E1A was detectable by Western blotting 24 hafter vCF11 infection of all of the colon cell lines, but notH1299, HeLa or SAEC (Figure 3, lane 3, E1A). Relativeto wild-type Ad5, the level of E1A expression was higherin SW480 and ISREC-01, the same in Co115 and lower inHT29 and Hct116 (Figure 3, compare lanes 2 and 3, E1A).The hierarchy of responsiveness of the Tcf-E1A promoterin the different cell lines was thus the same as with theTcf-E2 viruses,11 but the level of expression relative to thenormal promoter was higher for E1A than E2. Since theE1B and E2 enhancers are wild-type in vCF11, these tran-scription units should be inducible by E1A. The E4 pro-moter in vCF11 is potentially able to respond to both E1Aand Tcf. To test this, the blots were probed for E1B 55k,DBP and E4orf6. Consistent with the E1A results, allthree proteins were expressed normally in SW480,ISREC-01 and Co115, and undetectable in HeLa andSAEC (Figure 3, compare lanes 2 and 3). Despite theabsence of E1A expression, all three proteins wereexpressed weakly in H1299 cells, suggesting that thesecells contain an endogenous activity which can substitutefor E1A. Compared with wild-type infections, the levelof E1B 55k, DBP and E4orf6 was slightly reduced in HT29and more substantially reduced in Hct116 cells infectedwith vCF11 (Figure 3, compare lanes 2 and 3).

Viruses with Tcf sites in multiple early promotersTo test the effect of regulating E1A expression in the con-text of the previous generation of Tcf viruses, cells wereinfected with vMB31 (B23�) and vCF22 (AB23�4; Figure3, compare lanes 5 and 6). E1A and E4orf6 expressionwere well preserved in SW480, ISREC-01 and Co115infected with vCF22, but DBP expression was maintainedonly in SW480 and ISREC-01, and even there it wasslightly lower with vCF22 than wild-type Ad5 (Figure 3,


273

Figure 3 Western blot for E1A, E1B55k, DBP and E4orf6 24 h after infection of different cell lines with wild-type Ad5 and Tcf viruses. SW480 andISREC-01 are permissive colon cancer cell lines. Co115, Hct116 and HT29 are semi-permissive colon cancer cell lines. H1299, HeLa and SAEC arenon-permissive cell lines in which the wnt pathway is inactive. The SAEC blot is derived from two separate experiments giving similar wild-type Ad5activity. vMB31 was not tested on SAEC.

compare lanes 2 and 6, DBP). In the remaining cell lines,DBP expression was undetectable with vCF22. Insertionof Tcf sites in the E1A, E1B, E2 and E4 promoters invCF22 abolished the E1A-independent expression of E1B55K, DBP and E4orf6 seen in H1299 infected with vCF11(Figure 3, compare lanes 3 and 6, H1299). We concludethat insertion of Tcf sites into multiple early promotersproduces an extremely selective virus, but one withreduced activity in some colon cell lines.

Inhibition of Tcf-dependent transcription by E1AThe defect in early gene expression from the Tcf virusesin the semi-permissive cell lines is not restricted to a sin-gle promoter. Instead, there appears to be a general

Gene Therapy

defect in activation of viral Tcf promoters. This can bepartly explained by generally weaker Tcf activity. Thereason for this is unclear, but it does not reflect a lack ofwnt pathway activation per se, since the semi-permissivecell lines contain mutations in either APC or �-catenin,and the Tcf-E2 transcriptional activity measured byluciferase assay is not increased by transfection ofexogenous �N-�-catenin (Figure 4).

An alternative explanation for the semi-permissivity ofsome cell lines is that E1A could be inhibiting the viralTcf promoters, for example by sequestering p300, whichis a coactivator of Tcf-dependent transcription.24,25 Todetermine whether E1A inhibits the viral Tcf promoters,we performed transcription assays using the Tcf-E1A and


274

Gene Therapy

Figure 4 �-catenin is not limiting in SW480 and Co115 colon cancer celllines. Luciferase assays in SW480 and Co115 using a Tcf-E2 reporter.

Tcf-E2 promoters coupled to the luciferase gene. InSW480, the Tcf-E2 promoter was more active than thewild-type E2 promoter in the absence of E1A (Figure 5b,lanes 1 and 6), and gave almost exactly wild-type activityin the presence of E1A (Figure 5b, lanes 2 and 7). Thisconvergence was due to increased wild-type E2 promoteractivity and decreased Tcf-E2 promoter activity in thepresence of E1A. Mutation of the E3 promoter is requiredto produce a tightly regulated Tcf-E2 promoter, becausethe E3 promoter is adjacent to the E2 promoter.11 E3mutation reduced the activity of the E2 promoter slightly

Figure 5 E1A inhibits Tcf-dependent transcription. (a) Schematic diagram of the E1A12S mutants. (b–d) Luciferase assays with a wild-type E2 reporterand Tcf-E2 reporters. The ‘Tcf-E2 mut E3’ reporter contains inactivating mutations in the E3 enhancer.11 Cells were transfected with luciferase reportersand plasmids expressing the E1A mutants shown in (a). (b) SW480, (c) Co115, (d) Hct116.

in SW480 cells transfected with E1A, but the activity wasstill close to that seen with the wild-type promoter(Figure 5b, lanes 2 and 12). The high activity of the Tcf-E2 promoter in SW480 probably explains why this cellline is permissive for all of the Tcf viruses. In contrast,the level of Tcf-E2 activity in the presence of E1A wassubstantially below the wild-type level in Co115 andHct116 cells (Figure 5c and d, lanes 2, 7 and 12).

To determine the mechanism of inhibition, we testeddifferent E1A mutants (Figure 5a). Mutation of the Rbbinding site in E1A impaired transactivation of the wild-type E2 promoter in SW480 and Co115 (Figure 5b and c,lane 3), but not in Hct116 cells (Figure 5d, lane 3),whereas mutation of the p300 or p400 binding sites hadlittle effect on transactivation of the wild-type promoterby E1A in all three cell lines (Figure 5b, c and d, lanes 4and 5). Reduced transactivation by an E1A mutant unableto bind Rb is expected, given the presence of E2F sites inthe E2 promoter. The Tcf sites replace the normalenhancer in the Tcf-E2 promoter.11 In all three cell linesthe Rb and p400 binding site mutations did not relieveinhibition of the Tcf promoters by E1A (Figure 5b, c andd, lanes 8, 10, 13 and 15). The only mutation to have aneffect was the p300 binding site mutation (E1A �2-11, lab-elled �p300N), and in SW480 and Co115 the maximumrecovery never exceeded 50% of the lost activity (Figure5b and c, lanes 9 and 14). Mutation of E1A amino acid 2to glycine (R2G), which also blocks p300 binding, hadthe same effect (data not shown). In Hct116, the �p300Nmutation completely restored activity of the Tcf-E2 pro-


275moter (Figure 5d, lane 9). Interestingly, we have noticedby Western blotting that p300 is mutant in this cell line(the protein has a mobility on SDS-PAGE of approxi-mately 240 kDa, data not shown). The effect of themutation in E1A is therefore surprising in this cell line,but could reflect E1A binding to CBP or to the residualp300 fragment.

Analysis of additional E1A mutantsTo explore possible explanations for the incompleterecovery of activity after mutation of the p300 bindingsite in E1A, additional luciferase assays were performedin H1299 cells (Figure 6). The Tcf-E2 promoter was acti-vated 10-fold by �N-�-catenin (Figure 6a, compare lanes1 and 2), and this was inhibited by E1A (Figure 6a, lane3). p300 binds to two sites in E1A and mutation of eithersite partially relieved the inhibition of Tcf-dependenttranscription (E1A �p300N and �p300C, Figure 6a, lanes4 and 5). The C-terminal p300 binding site lies within theconserved domain 1 (CR1), but deletion of the entiredomain did not restore activity (Figure 6a, lane 6). Thissuggests that there may be a positively acting factorwhich binds somewhere in CR1. To determine whetherthe E1A �p300N mutation only partially restored activitybecause it did not completely block p300 binding, wecotransfected increasing amounts of p300 with E1A(Figure 6b). Exogenous p300 reversed the inhibition ofpromoter activity to the same extent as mutation of thep300 binding site (Figure 6b, lanes 4 and 7), and theeffects of the �p300N mutation and p300 transfectionwere not additive (Figure 6b, lane 8). Large amounts ofexogenous p300 reduced promoter activity (Figure 6b,lanes 5, 6, 9 and 10), suggesting that a cofactor was beingtitrated. P/CAF is a candidate for this cofactor becauseit is a histone acetyltransferase (HAT) that binds to p300,and the coactivation of Tcf by p300 does not requireintrinsic p300 HAT activity.24 Since E1A inhibitsP/CAF,26 we tested whether mutation of the P/CAFbinding domain in E1A relieved inhibition of Tcf activityby E1A, but saw no effect (Figure 6a, lane 7). P/CAF wasnot limiting because cotransfection of P/CAF and wild-type or �P/CAF mutant E1A also failed to restoreactivity (Figure 6c, lanes 4 and 9). To test whether p300and P/CAF act together, an E1A gene with mutations inthe binding sites for both HATs was constructed (labelled�� in Figure 6), but this mutant also failed to relieve therepressive effect of E1A (Figure 6a, lane 8), as did cotrans-fection of P/CAF and E1A mutant in the p300 bindingsite (Figure 6c, lane 6), or cotransfection of p300 and E1Amutant in the P/CAF binding site (Figure 6c, lane 8).

As in colon cells (Figure 5), mutation of the Rb bindingsite in E1A had no effect on repression of Tcf-dependenttranscription (Figure 6a, lane 9). CtBP and TIP49 haveboth been implicated in transcription activation byTcf,27,28 but neither mutations in E1A which abolish CtBPbinding (�CtBP, �C52; Figure 6a, lanes 10 and 11) nortransfection of wild-type or dominant negative TIP49(Figure 6c, lanes 10 and 11) could overcome the repress-ive effect of E1A. In conclusion, the E1A mapping studiesshowed that mutation of the p300 binding domain couldrestore about half of the Tcf activity lost upon E1Aexpression, but the remaining repressive effect could notbe mapped to a known domain in E1A.

Gene Therapy

Figure 6 Luciferase assays in the lung cancer cell line H1299 showinginhibition of Tcf-dependent transcription by mutant forms of E1A. (a)Cotransfection of a Tcf-E1A reporter with various E1A mutants and �N-�-catenin. (b) Cotransfection of increasing amounts of p300 plasmid (0.5,1, or 2 �g). (c) Effect of p300, P/CAF and Tip49 on Tcf-dependent tran-scription in the presence of wild-type and mutant forms of E1A. Thevalues represent the fold activation versus the E1A wild-type reporter inthe absence of E1A and �N-�-catenin. �N-�-catenin was cotransfectedin all the lanes shown in (b) and (c).

E1A �p300N mutant Tcf virusesTo test whether deletion of the p300 binding site in E1Awould increase the activity of the Tcf promoters in thecontext of the virus, the �p300N mutation was intro-duced into the Tcf-E1A, Tcf-E1B, Tcf-E2 and Tcf-E4viruses (Table 1 and Figure 1b). For the Tcf-E1A pro-moter, inhibition of p300 by E1A should inhibitexpression of E1A itself. This was tested by infecting thecMM1 cell line with vCF11 and vCF42, the �p300Nderivative of vCF11, in the presence and absence of tetra-cycline. Consistent with there being negative feedback by


276

Gene Therapy

E1A on its own expression, the level of E1A after acti-vation of wnt signalling was higher with vCF42 thanvCF11 (Figure 2, compare lanes 7 and 8, E1A). Despitethe increase in E1A expression, there was no differencein DBP expression, possibly because the �p300N mutantis defective in some other function required for activationof the wild-type E2 promoter (Figure 2, compare lanes 7and 8, DBP). The multiply-mutated viruses were thentested on a panel of cell lines (Figure 3). The effect of the�p300N mutation can best be appreciated by comparingmatched pairs of viruses: vCF11 versus vCF42 (Figure 3,lanes 3 and 4); vMB19 versus vCF81 (Figure 3, lanes 9 and8); and vCF22 versus vCF62 (Figure 3, lanes 6 and 7). Ineach case the latter is derived from the former by deletionof the p300 binding site in E1A (the only exception is thatthe E3 promoter ATF site is present in vCF22, but absentin vCF62). In almost every case the �p300N mutationactually reduced the level of expression of E1B 55K, DBPand E4orf6. The only promoter whose activity wasreasonably well maintained was the Tcf-E1A promoter(Figure 3, lanes 4 and 7, E1A). The wild-type E1A pro-moter was also little affected by the E1A �p300Nmutation (Figure 3, lane 8, E1A). After infection with themost comprehensively mutated virus (vCF62), viral pro-teins were undetectable in the control cell lines (Figure3, lane 7, H1299, HeLa and SAEC) and barely detectablein the semi-permissive colon cell lines (Co115, HT29 andHct116). The expression of DBP and E4orf6 was alsodecreased in the permissive cell lines (SW480 and ISREC-01), when multiple promoter changes were combinedwith the E1A �p300N mutation. In the absence of thismutation, the virus with multiple promoter changes(vCF22) showed wild-type expression of E1A, E1B55k,DBP and E4orf6 in the permissive cell lines (SW480 andISREC-01). The E1A �p300N mutation did not increaseE1B 55K or DBP expression in any of the viruses withTcf-E1B and Tcf-E2 promoters (Figure 3, compare lanes6 versus 7, and 9 versus 8). We conclude that in the contextof the virus, the E1A �p300N mutation does not rescuethe defect in Tcf promoter activity in the semi-permissivecell lines.

Since this result was unexpected, we also tested thenew viruses in cytopathic effect and burst assays. In themost permissive colon cell line, SW480, both vCF11 andvMB19 were at least 10-fold more active than wild-typeAd5 in CPE assays (Figure 7a, compare lane 1 with lanes2 and 6). For these viruses, the corresponding p300mutant viruses were about 10-fold less active (Figure 7a,compare lanes 2 versus 3, and 6 versus 7). Only for thevirus with Tcf sites in the E1A, E1B, E2 and E4 promoterswas the p300 mutant virus as active as the parent (Figure7a, compare lanes 4 versus 5), but these viruses were 100-fold less active than the virus with only the Tcf-E1A/E4changes (vCF11, Figure 7a, lane 2). vCF11 showed wild-type activity on Co115 (Figure 7b, compare lanes 1 versus2). This is 10-fold better than the previous best virus,vMB19 (Figure 7b, lane 6). In Hct116, the situation wasreversed: vMB19 was slightly better than vCF11, butwild-type was better than either Tcf virus (Figure 7c,lanes, 1, 2 and 6). In Co115, all of the p300 mutant viruseswere 10-fold less active than the corresponding viruseswith wild-type E1A (Figure 7b, compare lanes 2 versus 3,4 versus 5, and 6 versus 7). All of the Tcf viruses weresubstantially less active than wild-type Ad5 on HeLacells, which lack Tcf activity (Figure 7d). The most engi-

neered viruses failed to produce plaques on HeLa evenafter infection with 100 p.f.u./cell (Figure 7d, lanes 4 and5). The effect of mutation of the p300 binding site in E1Awas less obvious than on permissive cells. Overall, thebest virus was vCF11, which was 10-fold less active thanvMB19 and 1000-fold less active than wild-type Ad5 onHela cells (Figure 7d, lanes 1, 2 and 6). Since vCF11 is10-fold more active than wild-type Ad5 on SW480, itsoverall selectivity for the most permissive colon cells is10 000-fold relative to wild-type Ad5.

In burst assays, the effect of the p300 binding sitemutation was specific to the virus and the cell line. InSW480, the mutation reduced burst size 50-fold in the‘A4’ backbone (Figure 8, compare lanes 2 and 3), but hadalmost no effect in the ‘B23’ backbone (Figure 8, comparelanes 4 and 5). This difference may be due to the fact thatthe E2 promoter requires E1A function in vCF42, wherethe wild-type E2 enhancer is activated by ATF and E2F,but not in vCF81, where the E2 enhancer is replaced byTcf sites. The virus with Tcf sites in all the early pro-moters and the �p300 mutation in E1A (vCF62) was 100-fold less active than wild-type in SW480. vCF62 wasalmost as active as vCF42 (Figure 8, compare lanes 3 and6). There was a striking reduction in vCF62 burst size inthe non-permissive cells (107-fold in HeLa cells, 105-foldin SAEC; Figure 8, lanes 12 and 17). The remaining Tcfviruses showed 100- to 5000-fold reduced burst size inHeLa and SAEC. The �p300 mutation again reducedburst size in the virus with E2 driven by E1A (Figure 8,compare lanes 8 and 9), but actually increased burst size(albeit from a very low level) in SAEC when the E2 pro-moter was driven by Tcf (Figure 8, compare lanes 15and 16).

DiscussionThe goal of this study was to create adenoviruses whichreplicate efficiently in a wide range of colon cancer cells,but not in normal cells. We tested viruses with Tcf sites inmultiple viral early promoters and mutation of the p300binding site in E1A. Compared with the previous gener-ation of Tcf viruses,11 vCF11 is less toxic to cells lackingwnt activity and has broader activity in a panel of coloncancer cell lines in cytopathic effect assays. It has Tcf sitesin both ITRs, but only E1A transcription is tightly regu-lated by wnt signalling. This is partly explained by thefact that the Tcf sites are adjacent to the TATA box in theTcf-E1A promoter, but several hundred base pairsupstream of the E4 TATA box. To create an E1A pro-moter with the minimum possibility of interference fromextraneous signals, all of the normal regulatory elementswere deleted in vCF11. This contrasts with the approachused to produce prostate, hepatocellular cancer and bre-ast cancer targeting viruses, which retain the completeE1A enhancer, but place exogenous promoters betweenit and the E1A start site.1,3,4 To remove the E1A enhancerin vCF11, it was necessary to transfer the viral packagingsignal to the right ITR. In addition, half of the ITR wasreplaced by Tcf sites. This construction dictated the pos-ition of the Tcf sites relative to the E4 start site. Theendogenous E4 control elements were retained in vCF11because they confer repression of E4 transcription in nor-mal cells.29 The mutant E4 promoter thus contains thepart of the E1A enhancer contained in the packaging sig-nal, which could activate the promoter, flanked by Tcf


277

Figure 7 Cytopathic effect assays in different cell lines infected with 10-fold dilutions of wild-type Ad5 and Tcf viruses. (a) SW480 cells were infectedat a starting multiplicity of 10 p.f.u./cell and stained 6 days after infection. (b) Co115 and (c) Hct116 were infected at a starting multiplicity of 100p.f.u./cell and stained 7 days after infection. (d) HeLa were infected at a starting multiplicity of 100 p.f.u./cell and stained 8 days after infection.

and E4F sites, which should repress the promoter in nor-mal cells. The net result of these changes is reduced E4transcription measured by luciferase assay, regardless ofcell type (C Volorio and C Fuerer, unpublished data).

Replication of the previous generation of viruses isrestricted to cells with activated wnt signalling by the Tcfsites in the E2 promoter.11 vCF62 and vCF22, which haveTcf sites in multiple early promoters, were even moreseverely attenuated in cells lacking wnt activity. In cyto-pathic effect assays in HeLa cells, they were at least 104-fold less active than wild-type virus, and in burst assaysin HeLa and SAEC, vCF62 was 105- to 107-fold less activethan wild-type virus. The reduced activity of theseviruses in permissive cells might be due to deletion ofelement II in the E1A enhancer, which was previouslyreported to activate transcription of all of the early pro-moters in cis.20 Comparison of different viruses shows

Gene Therapy

that the Tcf-E1A and Tcf-E2 promoters display the samehierarchy of activity in a panel of colon cell lines, butrelative to the corresponding wild-type promoters, theTcf-E1A promoter is more active than the Tcf-E2 pro-moter. This probably explains why vCF11 is able toreplicate better than vMB19 in Co115 cells.

Notwithstanding minor differences between differentTcf promoters in the semi-permissive cell lines, there isa clear overall trend for Tcf promoters to be less activein some colon cell lines. Other groups have noted similardifferences in wnt activity in colon cancer cell lines.18,19

All of the cell lines tested (except Co115, where the defectis not known) and virtually all colon tumours possessAPC or �-catenin mutations resulting in activation of thewnt pathway. This indicates that wnt activation is a criti-cal step in colon cancer tumorigenesis. It is possible thatthe modest Tcf activity seen in some colon cell lines


278

Gene Therapy

Figure 8 Viral burst assays on permissive and non-permissive cell lines.SW480, HeLa and SAEC cells were infected with 300 viral particles/celland lysed 48 h after infection. The titre of viral particles present in thelysate was measured by plaque assay on SW480. Values were normalisedto the wild-type Ad5 titre on each cell line.

reflects a loss of activity during later tumour develop-ment or in vitro culture. In this case, there is no need toimprove the existing Tcf viruses. The wnt pathway isregulated at many levels, but the fact that exogenousmutant �-catenin does not increase Tcf activity allows usto rule out many upstream problems. Possible differenceswhich could explain the reduced Tcf activity in some celllines include increased expression of corepressors likegroucho and CtBP, decreased expression of coactivatorslike p300 and CBP, acetylation or phosphorylation of Tcf4preventing �-catenin binding or DNA binding, andincreased activity of the �N-Tcf1 negative feedbackloop.16,30

Luciferase reporter assays showed a systematic inhi-bition of Tcf-dependent transcription by E1A.Mutagenesis of E1A indicated that this effect was partlydue to inhibition of p300 by E1A, consistent with reportsthat p300 is a coactivator for �-catenin.24,31 Coexpressionof p300 together with E1A had the same effect on Tcf-dependent transcription as deletion of the p300 bindingsite in E1A, indicating that the remaining repression wasunlikely to be due to inhibition of p300. The residualrepressive effect of E1A could not be mapped to anyknown domain and merits further study. The negativeresults obtained with the �CR1 mutant are surprisingbecause deletion of the CR1 p300-binding subdomainalone did partially restore Tcf-dependent transcription.This could conceivably be explained by an artefactualelevation of transcription of the renilla luciferase controlby �CR1 E1A, but a more likely explanation is thatanother function of E1A is impaired by deletion of theentire CR1 domain.

The inhibition of Tcf-dependent transcription by E1Awas greatest in the semi-permissive cell lines like Co115,resulting in very low luciferase activity because the start-ing level of Tcf activity was also lower in these cells.Hence, we expected to see a substantial effect of the �2-11 E1A mutation in the context of the viruses. In practice,the mutation produced no increase in expression fromthe Tcf-E2 promoter in colon cell lines and reduced the

activity of the virus in cytopathic effect assays. There wasa small, but reproducible increase in E1A protein level incMM1 cells expressing mutant �-catenin infected withA�4 virus (vCF42) compared with A4 virus (vCF11),which would be consistent with decreased negative feed-back of E1A on its own expression through relief of p300inhibition, but the increase in E1A level could be due toprotein stabilisation. Consistent with the latter expla-nation, we have observed stabilisation of lower mobilityE1A isoforms in SW480 infected with �p300N virus (datanot shown). The mutation had complex and inconsistenteffects in burst assays: it appeared to reduce burst sizein permissive cells, when the E2 promoter was driven byE1A (ie wild-type), but increase burst size in some non-permissive cells, when the E2 promoter was driven byTcf. A general explanation is that any gain in Tcf activitydue to the E1A mutation was offset by a loss of otherE1A activities. Since we only tested 12S E1A, it is possiblethat these functions map to the other E1A isoformsexpressed during viral infection. In addition, there aresome basal promoter activities regulated by E1A whichmay be abrogated by the �2-11 mutation.32–35

In conclusion, we have shown that adenovirus repli-cation can be regulated by insertion of Tcf sites into theE1A or E2 promoters. Mutation of the p300 binding sitein E1A did not increase transcription from Tcf promotersin the context of the virus. Since the E1A �2-11 mutationconsistently reduced virus activity in cytopathic effectassays, it would be better to retain this domain intherapeutic viruses.

Materials and methods

Adenovirus mutagenesisAn Ad5 E1A fragment (nucleotides nt 1 to 952) wasamplified by PCR from ATCC VR5 adenovirus 5 genomicDNA with primers CGGAATTCAAGCTTAATTAACATCATCAATAATATACC (G76) and GGGTGGAAAGCCAGCCTCGTG (oCF1), cut with PacI, and cloned into theBamHI/PacI sites in pMB1 to give pCF4. pMB1 containsthe left end of Ad5 cloned into the EcoRI/SmaI sites ofpFL39.11,36 The endogenous adenoviral sequence from themiddle of the ITR to the E1A TATA box was replacedwith four Tcf binding sites by inverse PCR with primerstccAGATCAAAGGGattaAGATCAAAGGGccaccacctcattat(oCF3) and tCCCTTTGATCTccaaCCCTTTGATCTagtcctatttatacccggtga (oCF4) to give pCF25 (the Tcf sites inthe primers are shown in capitals). The final sequenceof the mutant ITR and E1A promoter is catcatcaataatataccttattttggattgaagccaatatgataatgaggTggtggCCCTTTGATCTTAATCCCTTTGATCTGGATCCCTTTGATCTCCAACCCTTTGATCTAGTCCtatttata, where the wt Ad5sequence is in lowercase and the E1A TATA box isunderlined. A G to T mutation was introduced just beforethe first Tcf binding site to mutate the Sp1 binding site.37

The Ad5 E4 fragment (nt 35369 to 35938) was amplifiedby PCR from VR5 DNA with primers G76 andACCCGCAGGCGTAGAGACAAC (oCF2), cut with PacIand cloned into the BamHI/PacI sites in pMB1 to givepCF6. To compensate for the mutations introduced in theleft ITR, three Tcf binding sites were introduced, and theendogenous sequence (nt 35805 to 35887) was simul-taneously deleted by inverse PCR with primers oCF3 andtCCCTTTGATCTccactagtgtgaattgtagttttcttaaaatg (oCF5)


279to give pCF16 (the Tcf site is shown in capitals and theSpeI site is underlined). The packaging signal was ampli-fied by PCR from pCF6 with primers GAACTAGTAGTAAATTTGGGCGTAACC (oCF6) and ACGCTAGCAAAACACCTGGGCGAGT (oCF7), cut with SpeI/NheIand cloned into the SpeI site in pCF6 to give pCF34. Thepackaging signal has the same end-to-centre orientationas at the left end of the adenoviral genome.

The �2-11 mutation was introduced in two steps. First,plasmids pCF4 (wild-type E1A promoter) and pCF25(Tcf-E1A mutant) were cut by SnaBI/SphI following byself ligation to give pRDI-283 and pRDI-284, respectively.Second, the 2-11 region in pRDI-283 and pRDI-284 wasdeleted by inverse PCR with primers CATTTTCAGTCCCGGTGTCG (oCF8) and ACCGAAGAAATGGCCGCCAG(oCF9) to give pCF61 and pCF56, respectively.

The YAC/BAC vector pMB1938 was cut with PacI fol-lowed by self ligation to give pCF1, a YAC/BAC vectorharbouring a unique PacI site.

In order to produce the gap repair vectors, combi-nations of left and right adenoviral ends were firstassembled and then transferred to the YAC/BAC vectoritself. During the first step, pCF34 was cut withEcoRI/SalI and cloned into the Pst/SalI sites of pCF25 togive pRDI-285. Similarly, pCF56 was cut withHindIII/SalI and cloned into the PstI/SalI sites of pCF34to give pCF46. Finally pCF61 was cut with HindIII/SalIand cloned into the PstI/SalI sites of pCF16 to givepCF52. pRDI-285, pCF46 and pCF52 all contain a cassettewith the left and right ends of the genome separated bya unique SalI site. These cassettes were isolated by PacIdigestion and cloned into the PacI site of pCF1 to givepCF78, pCF79 and pCF81, respectively. pCF78 hasmutant E1A and E4 promoters, pCF79 has mutant E1Aand E4 promoters plus the �2-11 mutation, and pCF81has wild-type E1A and E4 promoters plus the �2-11mutation.

vCF11 and vCF22 were constructed by gap repair38 ofpCF78 with VR5 (ATCC) and vMB3111 DNA, respect-ively. vCF42 and vCF62 were constructed by gap repairof pCF79 with VR5 and vMB19 DNA,11 respectively.vCF81 was constructed by gap repair of pCF81 withvMB31 DNA. The viral DNA was cut with ClaI beforegap repair to target the recombination event to a siteinternal to the mutations at the left end of the genome.

Viral genomic DNA was converted into virus by trans-fection of PacI digested YAC/BAC DNA into cR1 cells.The viruses were then plaque purified on SW480 cells,expanded on SW480, purified by CsCl banding, bufferexchanged using NAP25 columns into 1 M NaCl, 100 mMTris-HCl pH 8.0, 10% glycerol and stored frozen at�70°C. The identity of each batch was checked by restric-tion digestion and automated fluorescent sequencing ona Licor (Lincoln, NE, USA) 4200L sequencer in the E1A(nt 1–1050), E1B (nt 1300–2300), E2/E3 (nt 26700–27950)and E4 (nt 35250–35938) regions using primers IR213(E1A antisense: CAGGTCCTCATATAGCAAAGC),IR190 (E1B sense: TGTCTGAACCTGAGCCTGAG),IR110 (E2/E3 sense: CATCTCTACAGCCCATAC), IF171(E2/E3 antisense: AGTTGCTCTGCCTCTCCAC) andIR215 (E4 sense: CGTGATTAAAAAGCACCACC). Apartfrom the desired mutations, no differences were foundbetween the sequence of VR5 and the Tcf viruses. Particlecounts were based on the OD260 of virus in 0.1% SDSusing the formula 1 OD260 = 1012 particles/ml.

Gene Therapy

E1A, p300, P/CAF, Tip49 and �-catenin plasmidsWild-type 12S E1A (pCF9) and E1A mutants �pRb(124A,135A), �p300N (�2-11), �p300C (�64-68), �p400(�26-35), �P/CAF (E55), �CtBP (LDLA4), and �C52 weredescribed by Alevizopoulos et al39,40 and Reid et al.26 Allthe mutants were provided in a pcDNA3 backbone(Invitrogen, Carlsbad, CA, USA) except the �p300N and�p300C mutants that were isolated with BamHI/EcoRIand cloned into the BamHI/EcoRI sites of pcDNA3. The�CR1 mutant (�38-68) was made by inverse PCR of pCF9with primers TCTGTAATGTTGGCGGTGCAGGAAG(oCF10) and ATGGCTAGGAGGTGGAAGAT (oCF12) togive pCF45. The �� p300-P/CAF double mutant wasconstructed by three-way ligation of BstXI fragmentsfrom the single mutants. The E1A mutants are shown inFigure 5a. The �N-�-catenin plasmid was described byvan de Wetering et al.41 The p300 vector contains HA-tagged p300 expressed from the CMV promoter. TheP/CAF expression vector was described by Blanco et al.42

The Tip49 and Tip49DN vectors were described by Woodet al.43

Cell linesISREC-01,44 SW480 (ATCC CCL-228) and Co11523 weresupplied by Dr B Sordat (ISREC). HCT116 (CCL-247),HT29 (HTB-38), 293T were supplied by ATCC. HeLa(CCL-2) were supplied by ICRF. H1299 were supplied byDr C Prives (Columbia University).45 The cMM1 cell lineexpressing myc-tagged �N-�-catenin41 from the tet-offpromoter will be described elsewhere (M Malerba and RIggo, manuscript in preparation). C7 cells were suppliedby Dr J Chamberlain (University of Michigan).22 To createthe cR1 packaging cells, C7 cells were infected with a len-tivirus expressing myc-tagged �N-�-catenin.11,41 Clonet-ics small airway epithelial cells (SAEC) and SAGMmedium were supplied by Cambrex (East Rutherford,USA). All the other cell lines were grown in Dulbecco�smodified Eagle�s medium with 10% foetal calf serum(Invitrogen).

Luciferase assaysThe E2 reporters were described by Brunori et al.11 Toconstruct E1A reporters, wild-type and mutant E1A pro-moters were amplified by PCR from pCF4 and pCF25,respectively, with primers G76 and GTGTCGGAGCGGCTCGGAGG (oCF13), cut with HindIII, andcloned into the NcoI/HindIII sites of pGL3-Basic(Promega, Madison, WI, USA). Cells were seeded at 2.5× 105 cells per 35-mm well 24 h before transfection. 4.5�l of Lipofectamine (Invitrogen) was mixed for 30 minwith 100 ng of reporter plasmid, 1 ng of control Renillaluciferase plasmid (Promega) and 500 ng of vectorsexpressing E1A, P/CAF, p300 or TIP49. pcDNA3 emptyvector was added to equalise the total amount of DNA.In Figure 6b, 0.5, 1 and 2 �g of p300 vector were used.Cells were harvested 48 h after transfection and dualluciferase reporter assays performed according to themanufacturer�s instructions (Promega) using a Bio-counter (Lumac, Landgraaf, The Netherlands). Eachvalue is the mean of one to nine independent experi-ments done in triplicate, and transfection efficiency isnormalised to the activity of the Renilla control.

Western blottingCells were infected with 1000 viral particles per cell. Twohours after infection, the medium was replaced. Cells


280

Gene Therapy

were harvested 24 h later in SDS-PAGE sample buffer.E1A, E1B55K, DBP and E4orf6 were detected with theM73 (Santa Cruz Biotechnology, Santa Cruz, CA, USA),2A6,46 B647 and RSA348 monoclonal antibodies, respect-ively. Myc-tagged �-catenin was detected with the 9E10monoclonal antibody.49

Cytopathic effect assayCells in six-well plates were infected with 10-fold logdilutions of virus. Two hours after infection, the mediumwas replaced. After 6 to 8 days (Figure 7), the cells werefixed with paraformaldehyde and stained with crystalviolet.

Virus replication assayCells in six-well plates were infected with 300 viral par-ticles per cell. Two hours after infection, the medium wasreplaced. Cells were harvested 48 h later and lysed bythree cycles of freeze–thawing. The supernatant wastested for virus production by counting plaques formedon SW480 cells after 10 days under 1% Bacto agar inDMEM 10% FCS. Each bar in the Figure represents themean ± s.d. of triplicate plaque assays.

AcknowledgementsWe thank E Lurati and A Ducraux for expert technicalassistance. We thank M Malerba for providing the cMM1cell line. We thank K Alevizopoulos, J Chamberlain, HClevers, MD Cole, T Dobner, AJ Levine, C Prives, B Sor-dat and D Trono for supplying reagents. We thank MPeter and P Beard for critical reading of the manuscriptand M Brunori for helpful discussions. We thank theSwiss National Science Foundation, Swiss CancerLeague, Roche Research Foundation and ISREC forfinancial support.

References1 Rodriguez R et al. Prostate attenuated replication competent

adenovirus (ARCA) CN706: a selective cytotoxic for prostate-specific antigen-positive prostate cancer cells. Cancer Res 1997;57: 2559–2563.

2 Yu DC, Sakamoto GT, Henderson DR. Identification of the tran-scriptional regulatory sequences of human kallikrein 2 and theiruse in the construction of calydon virus 764, an attenuated repli-cation competent adenovirus for prostate cancer therapy. CancerRes 1999; 59: 1498–1504.

3 Hallenbeck PL et al. A novel tumor-specific replication-restrictedadenoviral vector for gene therapy of hepatocellular carcinoma.Hum Gene Ther 1999; 10: 1721–1733.

4 Hernandez-Alcoceba R, Pihalja M, Wicha MS, Clarke MF. Anovel, conditionally replicative adenovirus for the treatment ofbreast cancer that allows controlled replication of E1a-deletedadenoviral vectors. Hum Gene Ther 2000; 11: 2009–2024.

5 Heise C et al. An adenovirus E1A mutant that demonstratespotent and selective systemic anti-tumoral efficacy. Nature Med2000; 6: 1134–1139.

6 Khuri F et al. A controlled trial of intratumoral ONYX-015, aselectively-replicating adenovirus, in combination with cisplatinand 5-fluorouracil in patients with recurrent head and neck can-cer. Nature Med 2000; 6: 879–885.

7 Heise C et al. ONYX-015, an E1B gene-attenuated adenovirus,causes tumor-specific cytolysis and antitumoral efficacy that canbe augmented by standard chemotherapeutic agents. NatureMed 1997; 3: 639–645.

8 Bischoff JR et al. An adenovirus mutant that replicates selec-

tively in p53-deficient human tumor cells. Science 1996; 274:373–376.

9 Chen Y, Yu DC, Charlton D, Henderson DR. Pre-existent aden-ovirus antibody inhibits systemic toxicity and antitumor activityof CN706 in the nude mouse LNCaP xenograft model: impli-cations and proposals for human therapy. Hum Gene Ther 2000;11: 1553–1567.

10 Teichler ZD. US gene therapy in crisis. Trends Genet 2000; 16:272–275.

11 Brunori M, Malerba M, Kashiwazaki H, Iggo R. Replicatingadenoviruses that target tumors with constitutive activation ofthe wnt signaling pathway. J Virol 2001; 75: 2857–2865.

12 Polakis P. Wnt signaling and cancer. Genes Dev 2000; 14:1837–1851.

13 Korinek V et al. Constitutive transcriptional activation by a beta-catenin-Tcf complex in APC-/- colon carcinoma. Science 1997;275: 1784–1787.

14 He TC et al. Identification of c-MYC as a target of the APC path-way. Science 1998; 281: 1509–1512.

15 Tetsu O, McCormick F. Beta-catenin regulates expression ofcyclin D1 in colon carcinoma cells. Nature 1999; 398: 422–426.

16 Roose J et al. Synergy between tumor suppressor APC and thebeta-catenin-Tcf4 target Tcf1. Science 1999; 285: 1923–1926.

17 Crawford HC et al. The metalloproteinase matrilysin is a targetof beta-catenin transactivation in intestinal tumors. Oncogene1999; 18: 2883–2891.

18 Lipinski KS et al. High-level, beta-catenin/TCF-dependent trans-gene expression in secondary colorectal cancer tissue. Molec Ther2001; 4: 365–371.

19 Chen RH, McCormick F. Selective targeting to the hyperactivebeta-catenin/T-cell factor pathway in colon cancer cells. CancerRes 2001; 61: 4445–4449.

20 Hearing P, Shenk T. The adenovirus type 5 E1A enhancer con-tains two functionally distinct domains: one is specific for E1Aand the other modulates all early units in cis. Cell 1986; 45:229–236.

21 Challberg MD, Kelly TJ. Animal virus DNA replication. AnnuRev Biochem 1989; 58: 671–717.

22 Amalfitano A, Chamberlain JS. Isolation and characterization ofpackaging cell lines that coexpress the adenovirus E1, DNApolymerase, and preterminal proteins: implications for genetherapy. Gene Therapy 1997; 4: 258–263.

23 Cottu PH et al. Inverse correlation between RER+ status and p53mutation in colorectal cancer cell lines. Oncogene 1996; 13:2727–2730.

24 Hecht A. The p300/CBP acetyltransferases function as transcrip-tional coactivators of beta-catenin in vertebrates. EMBO J 2000;19: 1839–1850.

25 Takemaru KI, Moon RT. The transcriptional coactivator CBPinteracts with beta-catenin to activate gene expression. J Cell Biol2000; 149: 249–254.

26 Reid JL et al. E1A directly binds and regulates the P/CAF acetyl-transferase. EMBO J 1998; 17: 4469–4477.

27 Brannon M et al. XCtBP is a XTcf-3 co-repressor with rolesthroughout Xenopus development. Development 1999; 126:3159–3170.

28 Bauer A et al. Pontin52 and Reptin52 function as antagonisticregulators of beta-catenin signalling activity. EMBO J 2000; 19:6121–6130.

29 Fajas L et al. pRB binds to and modulates the transrepressingactivity of the E1A-regulated transcription factor p120E4F. ProcNatl Acad Sci USA 2000; 97: 7738–7743.

30 Roose J, Clevers H. TCF transcription factors: molecularswitches in carcinogenesis. Biochim Biophys Acta 1999; 1424:M23–M37.

31 Sun Y et al. Regulation of beta-catenin transformation by thep300 transcriptional coactivator. Proc Natl Acad Sci USA 2000;97: 12613–12618.

32 Lipinski KS, Esche H, Brockmann D. Amino acids 1-29 of theadenovirus serotypes 12 and 2 E1A proteins interact with rap30(TF(II)F) and TBP in vitro. Virus Res 1998; 54: 99–106.


28133 Lewis BA et al. Adenovirus E1A proteins interact with the cellu-lar YY1 transcription factor. J Virol 1995; 69: 1628–1636.

34 Lee JS et al. Relief of YY1 transcriptional repression by aden-ovirus E1A is mediated by E1A-associated protein p300. GenesDev 1995; 9: 1188–1198.

35 Kraus VB et al. Interaction of the Dr1 inhibitory factor with theTATA binding protein is disrupted by adenovirus E1A. ProcNatl Acad Sci USA 1994; 91: 6279–6282.

36 Bonneaud N et al. A family of low and high copy replicativeintegrative and single-stranded S. cerevisiae/E. coli shuttle vec-tors. Yeast 1991; 7: 609–615.

37 Leza MA, Hearing P. Cellular transcription factor binds to aden-ovirus early region promoters and to a cyclic AMP responseelement. J Virol 1988; 62: 3003–3013.

38 Gagnebin J et al. A photosensitising adenovirus for photodyn-amic therapy. Gene Therapy 1999; 6: 1742–1750.

39 Alevizopoulos K, Catarin B, Vlach J, Amati B. A novel functionof adenovirus E1A is required to overcome growth arrest by theCDK2 inhibitor p27(Kip1). EMBO J 1998; 17: 5987–5997.

40 Alevizopoulos K, Sanchez B, Amati B. Conserved region 2 ofadenovirus E1A has a function distinct from pRb bindingrequired to prevent cell cycle arrest by p16INK4a or p27Kip1.Oncogene 2000; 19: 2067–2074.

41 van de Wetering M et al. Armadillo coactivates transcriptiondriven by the product of the Drosophila segment polarity genedTCF. Cell 1997; 88: 789–799.

42 Blanco JC et al. The histone acetylase PCAF is a nuclear receptorcoactivator. Genes Dev 1998; 12: 1638–1651.

Gene Therapy

43 Wood MA, McMahon SB, Cole MD. An ATPase/helicase com-plex is an essential cofactor for oncogenic transformation by c-Myc. Mol Cell 2000; 5: 321–330.

44 Cajot JF, Sordat I, Silvestre T, Sordat B. Differential display clon-ing identifies motility-related protein (MRP1/CD9) as highlyexpressed in primary compared to metastatic human colon car-cinoma cells. Cancer Res 1997; 57: 2593–2597.

45 Chen X, Ko LJ, Jayaraman L, Prives C. p53 levels, functionaldomains, and DNA damage determine the extent of the apop-totic response of tumor cells. Genes Dev 1996; 10: 2438–2451.

46 Sarnow P, Sullivan CA, Levine AJ. A monoclonal antibodydetecting the adenovirus type 5-E1b-58Kd tumor antigen:characterization of the E1b-58Kd tumor antigen in adenovirus-infected and -transformed cells. Virology 1982; 120: 510–517.

47 Reich NC, Sarnow P, Duprey E, Levine AJ. Monoclonal anti-bodies which recognize native and denatured forms of the aden-ovirus DNA-binding protein. Virology 1983; 128: 480–484.

48 Marton MJ, Baim SB, Ornelles DA, Shenk T. The adenovirus E417-kilodalton protein complexes with the cellular transcriptionfactor E2F, altering its DNA-binding properties and stimulatingE1A-independent accumulation of E2 mRNA. J Virol 1990; 64:2345–2359.

49 Evan GI, Lewis GK, Ramsay G, Bishop JM. Isolation of mono-clonal antibodies specific for human c-myc proto-oncogene pro-duct. Mol Cell Biol 1985; 5: 3610–3616.

adenoviruses with tcf binding sites in multiple early promoters show

Documents