E2F Regulates DDB2: Consequences for DNA Repair in Rb-Deficient Cells

Oncogene (2007) 26, 3572–3581 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE E2F regulates DDB2: consequences for DNA repair in Rb-deﬁcient cells

S Prost, P Lu, H Caldwell and D Harrison

Queen’s Medical Research Institute, University of Edinburgh, Edinburgh, UK

DDB2, a gene mutatedin XPE patients, is involvedin which refers to repair in the overall genome. Mutation in global genomic repair especially the repair of cyclobutane various genes involved in either pathway produces pyrimidine dimers (CPDs), and is regulated by p53 in xeroderma pigmentosum (XP), rare autosomal recessive human cells. We show that DDB2 is expressedin mouse diseases characterized by a high incidence of UV- tissues anddemonstrate, using primary mouse epithelial induced cancers. Eight XP complementation groups cells, that mouse DDB2 is regulatedby E2F transcription have been defined and associated with mutation in factors. Retinoblastoma (Rb), a tumor suppressor critical specific proteins involved in NER (for general reviews for the control of cell cycle progression, regulates E2F on NER and XP diseases see Friedberg et al., 1995). activity. Using Cre-Lox technology to delete Rb in Mutations in the DDB2 gene are associated with the primary mouse hepatocytes, we show that DDB2 gene xeroderma pigmentosum group E (XPE) complementa- expression increases, leading to elevated DDB2 protein tion group, which is characterized by deficiency in GGR levels. Furthermore, we show that endogenous E2F1 and (Hwang et al., 1999). DDB2 is a small protein that E2F3 bindto DDB2 promoter andthat treatment with associates with DDB1 to form the UV-damaged DNA E2F1-antisense or E2F1-small interfering RNA (siRNA) binding (UVDDB) complex. This complex is critical as decreases DDB2 transcription, demonstrating that E2F1 its binding to CPDs creates a major distortion of the is a transcriptional regulator for DDB2. This has con- DNA helix leading to the recruitment of xeroderma sequences for global genomic repair: in Rb-null cells, pigmentosum group C (XPC) and subsequent repair of where E2F activity is elevated, global DNA repair is the damage. Mutation in DDB2 affects UVDDB activity increasedandremoval of CPDs is more efficient than in and compromises GGR, especially the repair of the CPDs. wild-type cells. Treatment with DDB2-siRNA decreases It has often been written that rodent cells do not DDB2 expression andabolishes the repair phenotype of express DDB2, have low or no UVDDB activity or are Rb-null cells. In summary, these results identify a new deficient in GGR. Although this has been clearly regulatory pathway for DDB2 by E2F, which does not demonstrated for Chinese hamster ovary cells (Hwang require but is potentiatedby p53, anddemonstrate that et al., 1998) and a few mouse fibroblast cell lines DDB2 is involvedin global repair in mouse epithelial cells. (Ishizaki et al., 1994; Tan and Chu, 2002), these data Oncogene (2007) 26, 3572–3581. doi:10.1038/sj.onc.1210151; have frequently been inappropriately extrapolated to all published online 18 December 2006 ‘rodent’ cells. In particular, there is some evidence that this is not the case in certain mouse cells: for example, Keywords: DDB2; retinoblastoma; E2F; transcription we have reported that the level of GGR, especially the factor; hepatocytes; DNA repair repair of CPDs, was similar in primary mouse hepatocytes and human fibroblasts (Prost et al., 1998b). Others have also detected UVDDB activity and DDB2 expression in some cell lines (Zolezzi and Linn, 2000; Tan and Chu, 2002). The question of DDB2 expression and its Introduction role in NER in mouse tissues had never been formally assessed and therefore needed clarification to better Nucleotide excision repair (NER) removes a wide understand and appropriately compare murine with variety of DNA lesions, in particular cyclobutane human repair responses. pyrimidine dimers (CPDs) and (6–4) photoproducts Human DDB2 expression has recently been shown to induced by ultraviolet light (UV). Two pathways have be positively regulated by p53, a transcription factor been identified: transcription-coupled repair (TCR), critical for cellular responses to DNA damage, including which preferentially repairs DNA damage within DNA repair. Indeed, p53-deficient cells have reduced transcribed DNA and global genomic repair (GGR), GGR, especially the repair of the CPDs, similar to DDB2 deficient cells (Prost et al., 1998a, b; Adimoolam and Ford, 2003 and references within). The demonstra- Correspondence: Dr S Prost, Queen’s Medical Research Institute, tion that in human cells DDB2 expression is dependent University of Edinburgh, 47 Little France Crescent, Edinburgh, on functional p53, in the presence or absence of DNA Scotland EH16 4TJ, UK. E-mail: [email protected] damage, provided an explanation for this phenotype. Received 5 December 2005; revised 18 October 2006; accepted 20 October However, the picture remains unclear for mouse cells, in 2006; published online 18 December 2006 which Tan and Chu (2002) recently showed that, unlike E2F, DDB2 and nucleotide excision repair S Prost et al 3573 in humans, mouse DDB2 was not regulated by p53 in expression was similar to that of wild-type (wt) cells mouse embryonic fibroblast and a murine liver cell line. (P ¼ 0.1312) but was significantly increased in retino- The present study describes a new regulatory pathway blastoma (Rb) null cells (P ¼ 0.0010). Protein could be for DDB2 in mouse hepatocytes. immunoprecipitated from cells of all genotypes (data not shown); however using Western blotting, the protein could be detected only in Rb null cells, and only from Results 72 h after adenovirus infection (Figure 2b), at which time the level of DDB2 transcription is correspondingly DDB2 is expressed in mouse tissues at least twofold higher than other genotypes and time We quantified DDB2 gene expression using real-time points tested (Figure 2a). polymerase chain reaction (PCR) and checked the level In human cells, DDB2 has been shown to be regulated of protein in nuclear extracts from various mouse tissues by p53 both at basal level and after DNA damage. We (Figure 1). Both mRNA (Figure 1a) and protein (1b) have reported that p53, which is stabilized and activated could be detected in all tissues tested, at variable levels. in hepatocytes conditionally deficient in Rb (Sheahan Expression was the highest in the lung, spleen and et al., 2004), is maximum 72 h after plating. We kidney, whereas nuclear levels of protein were greater in therefore investigated whether the increased DDB2 the liver and lung. Interestingly, a second band of lower expression in Rb null cells was due to the overexpression molecular weight was observed in the spleen (data not of p53. Interestingly, Rb loss in p53 null cells was still shown). able to increase DDB2 expression, although to a less substantial level than in cells with functional p53 (Figure 2c). This suggests that p53 is not strictly Retinoblastoma deficiency increases DDB2 expression in required for induction of DDB2 after the loss of Rb, a p53-independent manner but is necessary to achieve a maximum induction of DDB2 gene expression was detected in primary hepato- DDB2 expression in undamaged cells (Figure 2c). cytes regardless of genotype tested (Figure 2a, white In human cells, DDB2 has been shown to be activated bars). In p53 null cells, the level of DDB2 gene after UV-induced DNA damage in a p53-dependent manner. After UV however, DDB2 expression levels remained unchanged, in wt, p53 null or Rb null hepatocytes (Figure 2a, compare white and black bars). Taken together, these results show that in primary mouse hepatocytes, DDB2 expression is not upregulated after UV damage, in agreement with Tan and Chu (2002), but that acute loss of Rb affects the basal level of DDB2, to some extent through p53.

Increased DDB2 expression affects global DNA repair Human DDB2 is involved in GGR, especially the repair of CPDs. We assessed global DNA repair by quantify- ing unscheduled DNA synthesis (UDS) in wt and Rb null cells in which DDB2 expression is increased (Figure 3). The cells were UV-irradiated (10 J/m2)to induce DNA damage and cultured in the presence of tritiated thymidine. Repair synthesis was then quantified by counting the number of radioactive grains revealed on a photoemulsion (examples are shown in the photos of Figure 3A). The grain index reflects the incorporation of tritiated thymidine and therefore DNA repair synthesis. In untreated cells, the level of incorporation is low in more than 99% of the cells, showing that no DNA synthesis is occurring. One percent or less cells are Figure 1 DDB2 gene expression and protein level in mouse found in replicative DNA synthesis and have a grain tissues. (a) real-time PCR for DDB2 in mouse tissues. The figure index of 100. After UV-irradiation, the grain density shows a box plot representation of the relative expression of increased, regardless of genotype, reflecting DNA repair DDB2 corrected for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression in mouse tissues isolated from five wt mice synthesis (Figure 3A–C). The box plot representation of (three for thymus). (b) DDB2 protein levels in mouse tissues. grain indexes shows that unscheduled DNA synthesis DDB2 was immunoprecipitated from similar amounts of nuclear increases after Rb deletion (Figure 3B, Po0.0001). In extract prepared from various mouse tissues. Immunoprecipitation p53À/À cells, in which we have previously reported a was performed using an antibody recognizing the C-terminal part lower level of UDS than in wt cells (Prost et al., of the protein, whereas detection used an antibody recognizing the N-terminal of the protein. The negative control omits any cell 1998a, b), an increase in UDS is also consistently extract. observed after Rb deletion (Figure 3C); however it did

Oncogene E2F, DDB2 and nucleotide excision repair S Prost et al 3574 a wt p53-/- Rb-/- 0.6 0.6 2.5 no UV no UV no UV 2 2 2 0.5 10J/m 0.5 10J/m 2 10J/m 0.4 0.4 1.5 0.3 0.3 1 0.2 0.2

DDB2 expression 0.1 DDB2 expression 0.1 DDB2 expression 0.5

0 0 0 48 72 96 48 72 96 48 72 96 hrs after plating 24 48 72 24 48 72 24 48 72 hrs after UV

Rb -/- vs wt bcRb-/- p53-/-vs p53-/- 5

DL70 Cre 4

48 72 48 72 hrs after infection 3 DDB2 50 kDa 2 actin 40 kDa fold increase 1 in DDB2 expression

0 48 72 96 hrs after adeno infection

Figure 2 DDB2 expression in wt and Rb null hepatocytes. (a) DDB2 expression in cells of various genotypes: real-time PCR was performed at indicated times in wt, p53À/À and RbÀ/À hepatocytes treated or not with UV (10 J/m2, 6 h after plating). (b) Western blotting for DDB2 in primary hepatocytes treated with adenovirus control DL70 (wt cells) or adenovirus expressing Cre (RbÀ/À cells). Actin probing is shown as loading control. (c) DDB2 gene expression increases after Rb deletion. The increased DDB2 expression associated with Rb deletion is expressed as a ratio of the expression in control cells treated with Dl70 adenovirus (wt for Rb and either wt or null for p53). The quantiﬁcation was performed using real-time PCR, and the results are fold increased in Rb null cells7s.e.m. for 2 (for p53À/À and p53 RbÀ/À in white) to 6 (for wt and RbÀ/À in black) independent experiments.

not reach statistical significance (P ¼ 0.1814). This may The technique of UDS does not discriminate between indicate that the effect of Rb on unscheduled DNA the repair of (6–4) photoproducts and CPDs. In fact, as repair synthesis does not require functional p53. UDS quantifies the repair within the first 4 h after UV In order to confirm the role of DDB2 in the repair treatment and repair of the (6–4) photoproducts is phenotype of Rb null hepatocytes, we treated cells with described to be 5–10 times faster than the repair of the DDB2 small interfering RNA (siRNA). This treatment CPDs, in some systems UDS is likely to reflect mainly reduced significantly the level of DDB2 RNA in the the repair of the (6–4) photoproducts. However, as cells, to the level of untreated wt cells (Figure 3D) and CPDs are more abundant than (6–4) photoproducts (75 prevented the increases in UDS (Figure 3A, compare versus 25%) after UV treatment, and as in primary photos (e) and (f); Figure 3E) (Po0.0001 in both wt and hepatocytes, removal of CPDs and (6–4) photoproducts RbÀ/À cells treated with siRNA). appear to have closer kinetics of repair (40% of (6–4)

Figure 3 Effect of the loss of Rb on UDS and CPDs removal. (A) Photos representative of UDS grain index for (a) wt untreated hepatocytes not undergoing replicative DNA synthesis (>99%), (b) UDS in wt hepatocytes treated with UV, (c) untreated hepatocytes undergoing DNA synthesis (o1%), d) RbÀ/À untreated hepatocytes, (e) UDS in RbÀ/À hepatocytes treated with UV, (f) UDS in RbÀ/À hepatocytes treated with siRNA DDB2 and UV. All photos were taken at Â 40 magnification. (B) UDS in wt and RbÀ/À cells. (C) UDS in p53À/À and p53À/ÀRbÀ/À cells. (B and C) Quantification was performed as described in Materials and methods. The graphs are box plot representations of the nuclear grain index of at least 100 nuclei in cells treated or not with 10 J/m2. Unirradiated cells show low incorporation of tritiated thymidine reflecting the absence of replicative DNA synthesis at the time of experiment. Few cells (zero of 113 wt, one of 146 RbÀ/À, zero of 101 p53À/À and two of 123 p53 RbÀ/À hepatocytes) were in S phase at the time of experiment. The grain index for these cells appears as outliners in the box plot and does not affect the results. Graphs are from one typical experiment performed three (for p53À/À genotypes ) to five times. (D–F) DDB2 siRNA decreases both GGR and removal of CPDs in RbÀ/À cells. (D) DDB2 expression is reduced after 24 h of siRNA DDB2 treatment. This typical experiment shows the reduction of DDB2 expression obtained in the wt and RbÀ/À cells treated with SiDDB2, of part (F) Controls are cells treated with siGLO. (E) Unscheduled DNA synthesis. White box: wt cells, black box: RbÀ/À cells. RbÀ/À siRNA: RbÀ/À cells treated with DDB2 siRNA for 24 h. (F) diagrammatic representation of the experiment. Time after isolation of the hepatocytes (‘p’ time for perfusion) and time after UV treatment (‘t’ time for treatment) are shown for UDS and CPDs immunodetection experiments. In all experiments the cells were treated for 24 h with SiRNA (shown as a red bar). p0, time of hepatocytes plating and adenovirus infection, t0 time of UV treatment. (G and H) Removal of CPDs. (G) Immunofluorescence for CPDs (green) was performed on untreated cells (No UV, top photos) and at indicated times after 10 J/m2, and the nuclei were counterstained using TO-PRO-3 (blue). Where indicated RbÀ/À cells were treated with DDB2 siRNA 6 h after UV-irradiation for 24 or 44 h. The photos are of one typical experiment repeated 2 (for siRNA) to 4 times. (H) Quantification of CPD immunofluorescence. The intensity of green fluorescence was quantified as described in Materials and methods. The graph shows the average green fluorescence7s.d. in wt, Rb null and Rb null cells treated with DDB2 siRNA.

Oncogene E2F, DDB2 and nucleotide excision repair S Prost et al 3575 versus 30% of CPDs repaired 4 h after UV (Prost et al., methods. All parameters were precisely set to be 1998a, b)), it is likely that the difference in UDS constant, allowing for quantification analysis. In wt observed is mainly due to the removal of CPDs. cells, the labelling of CPDs decreases slowly with time To refine the UDS result we then investigated the (Figure 3G and H) usually becoming detectable by eye ability of wt and RbÀ/À cells to specifically repair around 50 h after UV. By contrast in RbÀ/À cells, the CPDs. We labelled the CPDs by immunofluorescence intensity of CPDs labelling diminishes visibly as early as using a specific CPD antibody (a gift from T Mori) at 6 h after UV. This confirms that RbÀ/À cells are indicated times after UV irradiation (Figure 3G). The repairing CPDs faster than wt cells. Interestingly, images were captured under identical conditions in complete removal of the CPDs was not achieved even cells treated in parallel as described in Materials and in RbÀ/À cells, 96 h after UV (data not shown).

Oncogene E2F, DDB2 and nucleotide excision repair S Prost et al 3576 abcE2F activity DDB2 expression unscheduled DNAsynthesis 30 6 dl70 dl70 dl70 cre cre 48 cre 25 5 40 20 4 32 15 3 24

10 2 16 thymidine incorporation luciferase activity (a.u.) 5 1 3 8 H expression relative to GAPDH 0 0 0 AS control E2F1-AS control AS E2F1-AS Control AS Control AS E2F1-AS 10J/m2

Figure 4 E2F antisense decreases E2F activity, DDB2 expression and UDS in wt (hepatocytes infected by adenovirus DL70 – white bars) or RbÀ/À cells (hepatocytes infected with adenovirus Cre – black bars). (a) reporter assay for E2F activity. Hepatocytes were transfected with an E2F reporter plasmid and treated with either a control antisense (AS) or E2F1 antisense. The bars represent the reporter gene activity 72 h after infection, 30 h after transfection. (b) DDB2 gene expression quantiﬁed by real-time PCR. The results are expressed relative to GAPDH expression. Treatment with E2F1 antisense reduces the transcription of DDB2 in RbÀ/À cells. (c) Unscheduled DNA synthesis. Incorporation of tritiated thymidine was quantiﬁed as previously in wt (white bars) and RbÀ/À (black bars) hepatocytes treated with either a control antisense or E2F1 antisense. E2F1-AS decreases UDS.

Treatment of Rb null cells with DDB2 siRNA for 24 h abE2F1 expression DDB2 expression 10 8 delayed the repair of the CPDs in RbÀ/À hepatocytes wild type wild type Rb/-- (Figure 3G and H). Rb-/- 8 6

6 E2F activates transcription of DDB2 with consequences 4 for global DNA repair 4 As Rb deletion does not require functional p53 to affect 2 relative expression level DDB2 expression levels (Figure 2b), we analysed the 2 relative expression level mouse DDB2 promoter region (Ensembl Gene ID 0 0 ENSMUSG00000002109) and identified a putative siGLO siE2F1 siGLO siE2F1 E2F binding site (also mentioned in Nichols et al., 2003) located 143–136 nucleotides upstream of the Figure 5 E2F1 siRNA decreases DDB2 expression. Wt (white translation initiation site. We have previously shown bars) and RbÀ/À (black bars) hepatocytes were treated with siRNA against E2F1 for 48 h when real-time PCR for E2F1 (a) and that E2F transcriptional activity is elevated in Rb null DDB2 (b) were performed. Treatment with siE2F1 decreases hepatocytes (Sheahan et al., 2004). We therefore tested expression of E2F1 and DDB2 compared with treatment with the the hypothesis that in murine cells DDB2 expression is control siRNA siGLO. regulated by E2F. As several studies have reported a role for E2F1 in level (Supplementary Figure). We therefore repeated DNA damage response (Stevens and La Thangue, 2004 this experiment using a siRNA against E2F1.As and references therein) and as in hepatocytes, whether previously shown with the antisense, the siRNA against wt of Rb null, we found that E2F1 gene expression is E2F1 decreased efficiently E2F1, RNA level, but not about fourfold higher than either E2F2 or E2F3 (data E2F2 or E2F3 (Supplementary Figure), and to a lesser not shown), we concentrated on the effect of E2F1 extent DDB2 (Figure 5), suggesting that E2F1 may inhibition. regulate DDB2 expression. Using an antisense against E2F1, we reduced E2F To confirm this, we then performed chromatin activity, together with DDB2 gene transcription (Figure immunoprecipitation (ChIP) assay. Hepatocytes in 4a and b). This was accompanied by a reduction of culture were lysed and proteins bound to the DNA unscheduled DNA synthesis (Figure 4c) (Po0.000 for were crosslinked as described in Materials and methods. wt and P ¼ 0.001 for RbÀ/À cells), suggesting that E2F1 After appropriate shearing, the DNA–protein complex is responsible for DDB2 activation and increased DNA was immunoprecipitated with two different antibodies repair. against E2F1. Both antibodies pulled down DNA However, the decrease in E2F activity after treatment fragments that were amplified with primers specific for with E2F1 antisense was somewhat stronger than the 210 bp sequence surrounding the putative E2F site expected. Indeed, the majority of E2F target promoters (data not shown and Figure 6a), showing that E2F1 are regulated by several E2Fs (Ishida et al., 2001; does bind to this sequence in murine hepatocytes. Attwooll et al., 2004 and references therein), and E2F2 Although some specificity has been described (Takahashi and E2F3, which are also regulated by Rb, should et al., 2000) the majority of E2F target promoters therefore contribute to this reporter activity. It was seem to be regulated by several E2Fs (Ishida et al., suspected that E2F1 antisense may also inhibit E2F2 2001; Attwooll et al., 2004 and references therein). We and E2F3 activity, despite no decrease in the mRNA therefore repeated the ChIP analysis for E2F2 and

Oncogene E2F, DDB2 and nucleotide excision repair S Prost et al 3577 UVDDB stimulate the excision of (6–4) photoproducts by twofold (Wakasugi et al., 2002). We also show for the first time that loss of Rb upregulates DDB2 and GGR by means of increased E2F activity. We have shown that E2F1-antisense or siRNA decreased DDB2 expression and DNA repair. Moreover, we show that both E2F1 and E2F3 are able to bind the DDB2 promoter. The recent finding from Berton et al. (2005) showing that E2F1-deficient mice have impaired DNA repair supports our finding that E2F affects DNA repair by means of regulation of repair proteins including DDB2. Intriguingly both DDB2 and E2F1–3 are normally cell cycle regulated. Figure 6 ChIP assay for E2F binding on DDB2 promoter. (a) diagram showing a representation of the putative E2F site in mouse In human cell lines in culture, through a combination of DDB2 50UTR together with the position of the amplification post-translational modification and increased gene primers chosen. The nucleotides are numbered relative to the first expression (Nag et al., 2001), DDB2 protein increases nucleotide of the initiation of translation (Met). (b) ChIP assay for during G1/S and peaks in S phase. Similarly, we E2F1–3. Total: total lysate before immunoprecipitation; No-Ab, observed in wt mouse primary hepatocytes that DDB2 non-antibody control; E2F1, E2F2, E2F3 using anti-E2F1–3 antibodies -VE, negative PCR control. E2F1 and E2F3 antibodies mRNA and protein levels were maximum in S phase immunoprecipitate a DNA fragment that is amplified with the (Figure 2a and data not shown). The basis of this cell primers specific for a putative binding site in DDB2 50 untranslated cycle regulation has never been investigated, but our region at the expected size. demonstration that E2F1 and E2F3 are able to bind to a consensus site situated in the 50 untranslated region of E2F3 and found that E2F3 but not E2F2 was able to the gene and that a decrease in E2F1 mRNA levels by bind the DDB2 promoter. Interestingly, E2F3 binding siRNA or antisense is associated with decreased DDB2 was increased in Rb null cells (Figure 6b). E2F3 expression proves that E2F regulates the expression of expression is unchanged in Rb null cells (data not mouse DDB2, at least in hepatocytes. This finding shown) but the transactivation ability of a particular provides for the first time, a plausible mechanism for the E2F may be conferred by the binding to protein partners cell cycle regulation of expression of DDB2 as upon Rb (Attwooll et al., 2004) and does not simply depend on its phosphorylation in G1/S, E2F activity also increases to level of expression (Muller and Helin, 2000). This may peak in S phase (for reviews, see Dyson, 1998; Muller explain the higher binding of E2F3 in Rb null and Helin, 2000). Whether E2F transcriptionally regu- hepatocytes and also indicates that, in some circum- lates DDB2 in other species remains to be confirmed, stances E2F2 could contribute to DDB2 expression, but the identification of an E2F site downstream of the even if no binding was observed here. initiation of transcription site in human DDB2 suggests that this may also be the case in human cells (Nichols et al., 2003). Discussion

DDB2 is expressed in mouse tissues and affects DNA DDB2 is not activated by p53 repair We (Prost et al., 1998a, b) and many other groups The present results establish that DDB2 is expressed in (reviewed in Adimoolam and Ford, 2003) have shown mouse tissues, including in the liver where the nuclear that p53 is involved in NER in both human and mouse protein level is high. The differences between expression cells and that loss of p53 function affects GGR, but not and quantity of protein observed in the liver and lung TCR (Ford and Hanawalt, 1995, 1997; Prost et al., suggest that mouse DDB2 protein is likely to be 1998b; Zhu et al., 2000). This effect is more specific to regulated by both post-translational modifications and the removal of CPDs than another UV-induced damage regulation of gene expression, as described in human (6–4 photoproducts). These characteristics are similar to cells (Nag et al., 2001). DDB2 gene expression and those of XPE cells bearing mutation in the DDB2 gene. protein were also detected in primary hepatocytes. With Human DDB2 expression has been shown to be these primary cells, using siRNA technology we were regulated by p53 at both basal level and after DNA able to demonstrate that mouse DDB2 does have a role damage (Hwang et al., 1999) when p53 binds to a in global genomic repair, and specifically in the repair of consensus site in the 50untranslated region of DDB2 CPDs. Our experiments do not rule out the possibility of gene and stimulates DDB2 transcription (Tan and Chu, DDB2 also having an effect on the repair of (6–4) 2002). photoproducts. Indeed, although expression of human In a recent paper Wang et al. (2004) demonstrated in DDB2 has no effect on the repair of (6–4) photopro- human cells that DDB2 is a downstream effector of p53, ducts in hamster cells (in which DDB2 is not normally which recognizes CPDs and can stimulate XPC recruit- expressed), it does have an effect in human cells: the ment and binding at the site of damage (Fitch et al., repair of (6–4) is mildly defective in XPE patients (Tang 2003a). This provides a clear explanation for the defect and Chu, 2002) and references therein), and low levels of in global repair in p53-deficient human cells: lack of

Oncogene E2F, DDB2 and nucleotide excision repair S Prost et al 3578 DDB2 jeopardizes the recruitment of XPC to the site of function that is important to mouse and perhaps human DNA damage and thus repair (Wang et al., 2004). cells. Basal level of expression of DDB2 was similar in wt and p53 null cells (Figure 2a). Furthermore, DDB2 expression did not significantly increase after UV Materials andmethods irradiation in primary mouse hepatocytes (Figure 2a) despite stabilization and activation of p53 (Sheahan Hepatocyte isolation, culture and adenovirus infection et al., 2004). This suggests that murine p53 does not The mice used were produced by crossing p53À/À (Purdie directly regulate DDB2 gene expression, unlike in et al., 1994) with Rb-floxed mice that are homozygous for the human cells, and is supported by a recent study showing Rb-floxed allele (exon 19 of the Rb gene is flanked by LoxP that murine p53 is unable to bind the sequence sequences) (Vooijs et al., 2002). Wt specimens were littermates corresponding to the p53-binding site identified in of the genetically modified animals. Primary hepatocytes, from adult mice (male, 6–12 weeks old), were isolated by two-step humans (Tan and Chu, 2002). Interestingly however, retrograde perfusion as described previously (Sheahan et al., although Rb deletion resulted in increased DDB2 gene 2004). RbÀ/À cells are obtained by deletion of the Rb-floxed expression regardless of p53 genotype (Figure 2c), the alleles in vitro by infection with an adenovirus expressing Cre- effect was maximum in the presence of functional p53. recombinase (as we previously described (Prost et al., 2001)). The precise mechanism for this potentiation of the effect Deletion of Rb is completed within 16–24 h after infection by p53 is unknown, but physical interaction of p53 with (Prost et al., 2001). Upon Rb deletion, RbÀ/À hepatocytes E2F1 (O’Connor et al., 1995) or other transcription exhibit a high level of p53 that is active for transcriptional factors (Ragimov et al., 1993) could be involved. activation (Sheahan et al., 2004).

DNA repair DDB2 and transcription For DNA repair studies, cells were UV-irradiated using a E2F activity is critical for the G1/S transition and is spectro-linker XL-1500 (Spectronics Corporation, Westbury, therefore tightly regulated. It has previously been shown NY 11590, USA) fitted with UV-C bulbs (254 nm). that DDB2 interacts with E2F1 and together with Unscheduled DNA synthesis was quantified as published previously (Prost et al., 1998b). Briefly, cells were cultured in DDB1, stimulates E2F1-activated transcription (Hayes the presence of tritiated thymidine for 3 h after 10 J/m2. After et al., 1998; Shiyanov et al., 1999, also reviewed by Tang 1 h chase, the slides were fixed, stained and covered with an and Chu, 2002). Our results support the hypothesis that autoradiographic emulsion (LM-1 – Amersham plc, Little in untransformed primary mouse epithelial cells, activa- Chalfont, UK). Nuclear grain counting was performed using tion of DDB2 by E2F could be critical for the regulation image analysis software (Kontron KS400 version 3.4 - Zeiss, of the G1/S transition. In undamaged cells, DDB2 is Carl Zeiss Ltd., Welwyn Garden City, UK). At the time of the activated by E2F, and in turn cooperates with DDB1 experiments, less than 1% of the hepatocytes are in S phase, and E2F1 to transcriptionally activate genes required and replicative DNA synthesis is easily detectable with a grain for DNA replication and G1/S transition. However, index above 80. Box plot representation allows all the data to after DNA damage, DDB2’s main function as a be plotted; a cell undergoing replicative DNA synthesis appears as an outliner and does not affect the overall result. transcription cofactor switches to global repair, and DNA repair of CPDs was quantified by immunofluore- the pool of protein, whose expression is not further scence immediately after or at various times after UV activated in primary mouse hepatocytes, is likely to irradiation (10 J/m2). Cells were treated with ice-cold detergent relocate to the site of damage as shown by Fitch et al. (0.5% Triton, 0.2 mg/ml ethylenediaminetetraacetic acid, 1% (2003a, b) using DDB2 overexpression systems. This bovine serum albumin (BSA)) and fixed in acetone/methanol decrease in available DDB2 would reduce the number of (1:1 v/v). Immunodetection of CPDs was performed as UVDDB-E2F1 complexes so reducing E2F1 transcrip- described previously (Nishiwaki et al., 2004) using TDM-2 tional activity and promoting cell cycle arrest, as antibody (Mori et al., 1991) 1/5000 (a kind gift of T Mori), and observed when the paramyxovirus simian virus 5v revealed by an Alexa 488 rabbit anti-mouse secondary protein sequesters DDB2 protein (Lin et al., 2000). antibody. Nuclear counterstain is TO-PRO-3. Images were taken using a Zeiss LSM5–10 inverted confocal microscope In conclusion, we show here that all mouse tissues using multitracking. Comparisons were made between cells tested express DDB2. Using mouse primary hepatocytes that were treated, fixed and stained and imaged on parallel on of various genotypes together with siRNA and antisense the same days. The conditions used for the confocal analysis technology, we show, in an authentic, non-immortalized are very precisely set to be consistent (the laser intensity, gain, system, that DDB2 is regulated by E2F but not p53, and pinhole settings and time for exposure are kept constant for is involved in DNA repair, especially the repair of the the whole series of images). An argon/krypton laser was used CPDs. As the cells are not immortalized or selected for a to excite fluorescein isothiocyanate at 488 nm at 6.1 A. The deficiency, as no overexpression system is used, and as confocal aperture pinhole was set digitally at 17 mm. The image the phenotype is observed rapidly after Rb deletion, it is was visualized with a Zeiss Plan-Neofluar 40 Â /1.3 DIC-oil thought to reflect an authentic regulation rather than an immersion objective. The contrast/brightness setting was selected so that the image for bright cells would not exceed artefact of cell culture or adaptation. Therefore, being saturated (o255 for pixel intensity). All of the settings although mouse and human cells share a requirement were kept constant throughout the analysis to yield unbiased for DDB2 in normal NER, they differ in requirement measurements for each set of comparisons. Focus was done for p53 to regulate this process, but we have identified on randomly selected fields while ‘fast scanning’ for green a new downstream regulatory role for E2F in DDB2 immunofluorescence (CPD). The adjacent field was then

Oncogene E2F, DDB2 and nucleotide excision repair S Prost et al 3579 scanned for green and blue fluorescence using multitracking Antisense treatment scanning with four lines average and a digital image was Mouse-specific E2F1 antisense and control antisense were acquired at a resolution of 1024 Â 1024. The intensity of purchased from Biognostik, Germany (cat. no. 11855-0203). A nuclear green fluorescence in each cell was quantified using mixture of three E2F1-specific or three control antisenses Image-pro plus software. An average of 40 cells were (100 mM) was added into the culture medium (50 ml of the mix quantified per condition. per 6 cm plates) from 10 h after isolation. Medium was changed and more antisense was added every 24 h. Immunodetection of DDB2 protein For primary hepatocytes: Total cell proteins were separated on SiRNA treatment a 10% sodium dodecyl sulphate–polyacrylamide gel electro- siGENOME SMARTpool siRNA mouse E2F1 (cat. no.M- phoresis, transferred to nitrocellulose, probed with anti- 044993-01) and mouse DDB2 (M-044959-00) were purchased DDB2N antibody (directed against the N-terminal part of from Dharmacon (Lafayette, CO, USA). These siRNA the protein) 1:50 for 3 h at room temperature (Zymed products are pools of four SMARTselection designed siRNA Laboratories, Invitrogen Ltd., Paisley, UK, ZMD06 cat. no. duplexes targeting mouse E2F1 and DDB2 mRNA transcripts 34-2400) and revealed by the appropriate horseradish perox- respectively. SiGLO cyclophilin B siRNA was used as a positive idase secondary antibody (1:2000) (Dako UK Ltd., Ely, UK) control of transfection efficiency. Hepatocytes were plated in a six-well plates and treated with an adenovirus expressing and enhanced chemiluminescence þ . To confirm loading, the blot was probed using an anti-actin antibody (AB8227, Cre-recombinase (CRE) or DL70 adenovirus. At indicated times, AbCam plc, Cambridge, UK). the culture medium was replaced by medium containing 100 nM For tissues: Nuclear extract was prepared from murine siRNA and 0.8% of DharmaFECT 1.1 (T-2001-02) transfection tissues as previously described (Nag et al., 2001). Immunopre- reagents, leading to a transfection efficiency above 70%. cipitation was performed from equal amounts of protein SiRNA was added 24–48 h before fixing for analysis or (400 mg) using TrueBlott (eBioscience) according to the harvesting cells for real-time PCR experiments. manufacturer’s instructions using an antibody directed against Quantitative real-time PCR the C-terminal part of the protein (DDB2C, Zymed Labora- Total RNA was harvested at indicated times after virus tories ZMD10, cat. no. 33-2800). The samples were run on a infection, using Qiagen RNeasy mini-column (cat.no.74104; gel, transferred and detected using the different anti-DDB2N QIAGEN Ltd., Crawley, UK). Total RNA was eluted from the antibody to the N-terminal part of the protein as described column in nuclease-free water before being re-concentrated by above, and the secondary antibody provided in the immuno- ethanol precipitation. Normally about 1–3 mgtotalRNAwas precipitation kit. Hence DDB2 protein was precipitated and obtained from the cells in a single well of a six-well plate. A260/ immunodetected using independent antibodies to different A280 ratios were measured by spectrophotometer Biomate 3 and parts of the protein. were between 1.8 and 2.0. The RNA quality was also examined on RNA 6000 Nano LabChIP by an Agilend Bioanalyzer ChIP for DDB2 (Agilent Technologies, Santa Clara, CA, USA). cDNA was The DDB2 mouse gene (Ensembl Gene ID ENSMUS- synthesized using MMLV (Qiagen) reverse transcriptase and an G00000002109) was analysed and a putative E2F binding site, oligo-dT primer. RNA templates were removed at the end of also reported by others (Nichols et al., 2003), was identified reaction by RNase H. The efficiency of the reverse transcription located 143–136 nucleotides upstream of the translation was examined by PCR using a pair of b-actin primers. initiation site. We used Applied Biosystems’s Taqman assays products and Protein/DNA crosslinking was performed 96 h after cell the reactions were run on an ABI Prism 7900 Sequence plating by treatment with 1% formaldehyde at 371C for Detection System (AME Bioscience Ltd., Sharnbrook, UK). 10 min. Cells were then lysed and DNA was sheared by The genes studied were Gapd (MM99999915_q1), E2F1 sonication, resulting in mainly 500–750 bp long DNA pieces. (MM00432939_m1) and DDB2 (MM00651374_m1). Immunoprecipitation was carried out using two different antibodies against E2F1 (Biomeda CA, US, and NeoMarkers, Statistical analysis CA, USA), E2F2 (Santa Cruz Biotechnology, Inc., Santa Analyses were carried out with Minitab for Windows (version Cruz, CA, USA, SC633, clone C-20) or E2F3 (Santa Cruz 13.0). For unscheduled DNA synthesis, the nuclear index was Biotechnology, Inc., SC878, clone C-18) and protein-A arcsin-transformed, and analysis of variance (ANOVA) was agarose. Immunoprecipitated DNA was purified and subjected performed using a general linear model, with, when appropriate, to PCR using primers that cover DDB2 regulatory element (50- comparison with control using Bonferroni test. For expression CAAGACAGAAAATACTGTG-30 and 50-CGAAGCAAG levels, differences between means were evaluated with ANOVA. GAGTGGAAAAG-30) where the putative site for E2F was Satisfactory homogeneity of variances was determined with identified. The expected PCR product is 210 bp long. Bartlett’s test. Differences were taken to be significant if Po0.05.

E2F activity Acknowledgements Hepatocytes in culture for 48 h were transfected using TFx-50 (Promega Biotech, Southampton, UK) reagent (ratio 1/5 w/w We thank Chris Bellamy for critical comments on the DNA/lipid) as described previously (Prost et al., 1998b). To manuscript. The Rb-floxed (Rblox/lox) mice were a kind gift control for transfection efficiency, cells were transfected with from Anton Berns (Netherlands Cancer Institute, Amsterdam) a mixture of pCMVb, constitutively expressing Lac Z and to whom we are very grateful. Many thanks to Toshio Mori pE2F-TA-Luc (Mercury pathway profiling system, Clontech for kindly providing us with the TDM-2 antibody. We are very Laboratories, Inc., Mountain View, CA, USA) (ratio 1/10). grateful to S Sheahan for providing primers for E2F1 and Twenty-four hours after transfection, cells were scraped, and E2F2, and antibodies against E2F2 and E2F3, without which luciferase and bGal activities were quantified using Luciferase we would not have been able to complete this study. This work and bGalactosidase Assay reagents according to the manu- was supported by a grant from the Melville Trust for the Care facturer (Promega). and Cure of Cancer to SP.

Oncogene E2F, DDB2 and nucleotide excision repair S Prost et al 3580 References

Adimoolam S, Ford JM. (2003). p53 and DNA damage Nichols AF, Itoh T, Zolezzi F, Hutsell S, Linn S. (2003). Basal inducible expression of the xeroderma pigmentosum group transcriptional regulation of human damage-specific DNA- C gene. DNA Repair (Amsterdam) 2: 947–954. binding protein genes DDB1 and DDB2 by Sp1, E2F, Attwooll C, Lazzerini DE, Helin K. (2004). The E2F family: N-myc and NF1 elements. Nucleic Acids Res 31: 562–569. specific functions and overlapping interests. EMBO J 23: Nishiwaki Y, Kobayashi N, Imoto K, Iwamoto TA, Yama- 4709–4716. moto A, Katsumi S et al. (2004). Trichothiodystrophy Berton TR, Mitchell DL, Guo R, Johnson DG. (2005). fibroblasts are deficient in the repair of ultraviolet-induced Regulation of epidermal apoptosis and DNA repair by E2F1 cyclobutane pyrimidine dimers and (6–4)photoproducts. in response to ultraviolet B radiation. Oncogene 24: J Invest Dermatol 122: 526–532. 2449–2460. O’Connor DJ, Lam EW, Griffin S, Zhong S, Leighton LC, Dyson N. (1998). The regulation of E2F by pRB-family Burbidge SA et al. (1995). Physical and functional interac- proteins. Genes Dev 12: 2245–2262. tions between p53 and cell cycle co-operating transcription Fitch ME, Cross IV, Ford JM. (2003a). p53 responsive factors, E2F1 and DP1. EMBO J 14: 6184–6192. nucleotide excision repair gene products p48 and XPC, but Prost S, Bellamy CO, Cunningham DS, Harrison DJ. (1998a). not p53, localize to sites of UV-irradiation-induced DNA Altered DNA repair and dysregulation of p53 in IRF-1 null damage, in vivo. Carcinogenesis 24: 843–850. hepatocytes. FASEB J 12: 181–188. Fitch ME, Nakajima S, Yasui A, Ford JM. (2003b). In vivo Prost S, Ford JM, Taylor C, Doig J, Harrison DJ. (1998b). recruitment of XPC to UV-induced cyclobutane pyrimidine Hepatitis B x protein inhibits p53-dependent DNA repair in dimers by the DDB2 gene product. J Biol Chem 278: primary mouse hepatocytes. J Biol Chem 273: 33327–33332. 46906–46910. Prost S, Sheahan S, Rannie D, Harrison DJ. (2001). Ford JM, Hanawalt PC. (1995). Li-Fraumeni syndrome Adenovirus-mediated Cre deletion of floxed sequences in fibroblasts homozygous for p53 mutations are deficient in primary mouse cells is an efficient alternative for studies of global DNA repair but exhibit normal transcription-coupled gene deletion. Nucleic Acids Res 29: E80. repair and enhanced UV resistance. Proc Natl Acad Sci USA Purdie CA, Harrison DJ, Peter A, Dobbie L, White S, 92: 8876–8880. Howie SE et al. (1994). Tumour incidence, spectrum and Ford JM, Hanawalt PC. (1997). Expression of wild-type p53 is ploidy in mice with a large deletion in the p53 gene. required for efficient global genomic nucleotide excision Oncogene 9: 603–609. repair in UV-irradiated human fibroblasts. J Biol Chem 272: Ragimov N, Krauskopf A, Navot N, Rotter V, Oren M, 28073–28080. Aloni Y. (1993). Wild-type but not mutant p53 can repress Friedberg EC, Walker GC, Siede W. (1995). DNA Repair and transcription initiation in vitro by interfering with the Mutagenesis. ASM Press: Washington, pp 283–366. binding of basal transcription factors to the TATA motif. Hayes S, Shiyanov P, Chen X, Raychaudhuri P. (1998). DDB, Oncogene 8: 1183–1193. a putative DNA repair protein, can function as a transcrip- Sheahan S, Bellamy CO, Treanor L, Harrison DJ, Prost S. tional partner of E2F1. Mol Cell Biol 18: 240–249. (2004). Additive effect of p53, p21 and Rb deletion in triple Hwang BJ, Ford JM, Hanawalt PC, Chu G. (1999). knockout primary hepatocytes. Oncogene 23: 1489–1497. Expression of the p48 xeroderma pigmentosum gene is Shiyanov P, Hayes SA, Donepudi M, Nichols AF, Linn S, p53-dependent and is involved in global genomic repair. Slagle BL et al. (1999). The naturally occurring mutants of Proc Natl Acad Sci USA 96: 424–428. DDB are impaired in stimulating nuclear import of the p125 Hwang BJ, Toering S, Francke U, Chu G. (1998). p48 subunit and E2F1-activated transcription. Mol Cell Biol 19: Activates a UV-damaged-DNA binding factor and is 4935–4943. defective in xeroderma pigmentosum group E cells that lack Stevens C, La Thangue NB. (2004). The emerging role of E2F- binding activity. Mol Cell Biol 18: 4391–4399. 1 in the DNA damage response and checkpoint control. Ishida S, Huang E, Zuzan H, Spang R, Leone G, West M et al. DNA Repair (Amsterdam) 3: 1071–1079. (2001). Role for E2F in control of both DNA replication and Takahashi Y, Rayman JB, Dynlacht BD. (2000). Analysis of mitotic functions as revealed from DNA microarray promoter binding by the E2F and pRB families in vivo: analysis. Mol Cell Biol 21: 4684–4699. distinct E2F proteins mediate activation and repression. Ishizaki K, Ejima Y, Matsunaga T, Hara R, Sakamoto A, Genes Dev 14: 804–816. Ikenaga M et al. (1994). Increased UV-induced SCEs but Tan T, Chu G. (2002). p53 Binds and activates the xeroderma normal repair of DNA damage in p53-deficient mouse cells. pigmentosum DDB2 gene in humans but not mice. Mol Cell Int J Cancer 58: 254–257. Biol 22: 3247–3254. Lin GY, Paterson RG, Richardson CD, Lamb RA. (2000). Tang J, Chu G. (2002). Xeroderma pigmentosum complemen- The V protein of paramyxovirus SV5 interacts with damage- tation group E and UV-damaged DNA-binding protein. specific DNA binding protein. Virology 249: 189–200. DNA Repair (Amsterdam) 1: 601–616. Mori T, Nakane M, Hattori TKN, Matsunaga T, Ihara M, Vooijs M, te RH, van d V, Berns A. (2002). Tumor formation Nikaido O. (1991). Selective regulation of T cell IL-5 in mice with somatic inactivation of the retinoblastoma gene synthesis by OM-01, JTE-711 and p38 MAP kinase in interphotoreceptor retinol binding protein-expressing inhibitor: independent control of Th2 cytokines, IL-4 and cells. Oncogene 21: 4635–4645. IL-5. Photochem Photobiol 54: 225–232. Wakasugi M, Kawashima A, Morioka H, Linn S, Sancar A, Muller H, Helin K. (2000). The E2F transcription factors: key Mori T et al. (2002). DDB accumulates at DNA damage regulators of cell proliferation. Biochim Biophys Acta 1470: sites immediately after UV irradiation and directly stimu- M1–M12. lates nucleotide excision repair. J Biol Chem 277: 1637–1640. Nag A, Bondar T, Shiv S, Raychaudhuri P. (2001). The Wang QE, Zhu Q, Wani G, Chen J, Wani AA. (2004). UV xeroderma pigmentosum group E gene product DDB2 is a radiation-induced XPC translocation within chromatin specific target of cullin 4A in mammalian cells. Mol Cell Biol is mediated by damaged-DNA binding protein, DDB2. 21: 6738–6747. Carcinogenesis 25: 1033–1043.

Oncogene E2F, DDB2 and nucleotide excision repair S Prost et al 3581 Zhu Q, Wani MA, El Mahdy M, Wani AA. (2000). slow repair sites in transcribed strand. J Biol Chem 275: Decreased DNA repair efﬁciency by loss or disruption 11492–11497. of p53 function preferentially affects removal of cyclobutane Zolezzi F, Linn S. (2000). Studies of the murine DDB1 and pyrimidine dimers from non-transcribed strand and DDB2 genes. Gene 245: 151–159.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene