DNA Repair and Recombination Factor Rad51 Is Over-Expressed in Human Pancreatic Adenocarcinoma

Home , DNA repair, RAD51

Oncogene (2000) 19, 2791 ± 2795 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc SHORT REPORT DNA repair and recombination factor Rad51 is over-expressed in human pancreatic adenocarcinoma

H Maacke1, K Jost1, S Opitz1, S Miska1, Y Yuan1, L Hasselbach1,JLuÈ ttges2, H Kaltho3 and H-W StuÈ rzbecher*,1

1Institute for Human Genetics, Medical University, Ratzeburger Allee 160, D-23538 LuÈbeck, Germany; 2Institute for Pathology, Clinic for General and Thoracic Surgery, Christian-Albrechts-University, Michaelisstraûe 5, D-24105 Kiel, Germany; 3Molecular Oncology, Clinic for General and Thoracic Surgery, Christian-Albrechts-University, Michaelisstraûe 5, D-24105 Kiel, Germany

Molecular processes that could contribute to dierences (Yanagisawa et al., 1998; VispeÂ et al., 1998). Treatment in chemo- and radioresistance include variations in DNA of tumour cells in monolayer cultures with Rad51 repair mechanisms. In mammalian cells, the product of speci®c anti-sense oligonucleotides renders them radio- the rad51 gene mediates DNA repair via homologous sensitive (Ohnishi et al., 1998). recombination. We describe that in contrast to conven- During cell cycle progression only minor variations tional monolayer cell systems Rad51 protein accumulates in Rad51 protein level occur (Yamamoto et al., 1996; to high-levels in three-dimensional cell culture models as Chen et al., 1997). Immortalization of human ®bro- well as in orthotopic xeno-transplants of human blasts is accompanied by a 3 ± 4-fold increase in Rad51 pancreatic cancer cells. Strikingly, over-expression of mRNA expression levels (Xia et al., 1997). There is, wild-type Rad51 was also found in 66% of human however, no systematic description of the Rad51 pancreatic adenocarcinoma tissue specimens. Functional protein levels in dierent human tumour cell lines analysis revealed that Rad51 over-expression enhances derived from a single tumour entity. Therefore, we survival of cells after induction of DNA double strand compared the amount of Rad51 protein in a panel of breaks. These data suggest that perturbations of Rad51 thirteen well-characterized human pancreatic cancer expression contribute to the malignant phenotype of cell lines (Kaltho et al., 1993) by immunocytochem- pancreatic cancer. Oncogene (2000) 19, 2791 ± 2795. istry and Western blotting. Figure 1a shows represen- tative examples of staining intensity and subcellular Keywords: carcinogenesis; drug resistance; p53; genetic localization of Rad51 protein and by way of compar- instability; SOS-repair ison of p53 in 818-4 and BXPC3 cells. Both cell lines exhibited weak nuclear staining for Rad51 polypeptide similar to the signal intensity for p53 in 818-4 cells. By The capability of tumour cells to become resistant contrast, a very strong signal for over-expressed towards chemo- and/or radiotherapy is regarded one of mutant p53 was found in BXPC-3 cells. Examination the major problems that hinder the eciency of most at low order magni®cation revealed only minor established therapeutic regimes to treat advanced stages variations in staining intensities for both polypeptides of solid tumours. These critical aspects of the between individual cells of a given cell line. To malignant phenotype of cancer cells are mimicked establish that the dierences in staining intensity re¯ect more reliably by three-dimensional (3D) cell systems quantitative dierences in Rad51 and p53 content, than by classical monolayer cell cultures (for review see protein levels were determined by Western blotting Mueller-Klieser, 1997). 3D-growth enhances the level (Figure 1b). The amount of mutant p53 in BXPC-3 of radio- and chemo-resistance compared to the same cells is increased at least by a factor of 10 compared to cells grown as monolayers (Mueller-Klieser, 1997; p53 in 818-4. No such dierence between the two cell Olive and Durand, 1994). Since classical resistance lines was observed for Rad51. These results demon- genes are not alone responsible for the development of strate that pancreatic cancer cell lines display a very chemoresistance in 3D cultures (Desoize et al., 1998) it low level of Rad51 protein, hardly detectable by has been hypothesized that altered DNA repair immunocytochemical methods and comparable to the mechanisms might also contribute (Mueller-Klieser, amount of wild-type p53 in these cells. 1997). In order to test whether 3D growth might aect The product of the rad51 gene is one of the key- expression of Rad51, 818-4 and PancTU-I, cells were factors catalysing homologous recombination processes grown as spheroids. Figure 2a demonstrates that involved in the repair of DNA double strand breaks Rad51 accumulates to much higher levels in a sub- (Baumann and West, 1998). Elevated expression of population of tumour cell nuclei compared to mono- Rad51 enhances radioresistance of human tumour cells layer cultures. The proportion of cells over-expressing Rad51 varies between cell lines with about 5 ± 10% of cells for PancTU-I and more than 30% for 818-4. Western blot analysis of Rad51 in PancTU-I and 818-4 *Correspondence: H-W StuÈ rzbecher, Institute for Human Genetics, cells grown as monolayers or as spheroids con®rmed University Clinic LuÈ beck, Ratzeburger Allee 160, D-23538 LuÈ beck, the immunocytochemical data (Figure 2b). In contrast Germany Received 10 August 1999; revised 13 March 2000; accepted to Rad51, expression of p53 was not aected by cell 15 March 2000 culture conditions (data not shown). Quanti®cation of Over-expression of Rad51 in pancreatic cancer H Maacke et al 2792

Figure 1 Comparison of Rad51 and p53 expression levels in monolayer cell culture. (a) Immunocytochemistry of Rad51 and p53 in pancreatic cancer cell lines. Immunostaining of cell lines 818-4 and BXPC-3, respectively, was performed using monoclonal antibodies 1G8 (Buchhop et al., 1996) for Rad51 and PAb1801 (Dianova, Hamburg, Germany) for p53. Before staining, cells were ®xed in 3.7% neutral buered formalin and permeabilized with 0.2% Triton X-100. Rad51 and p53 proteins were visualized using diaminobenzidine tetrahydrochloride (DAB); counter-staining with Mayer's hemalum (Merck, Darmstadt, Germany). Magni®cation: 1000-fold. (b) Expression of full-length Rad51 and p53 proteins in pancreatic tumour cell lines. Western blot analysis of Rad51 and p53 proteins in 50 mg of total protein extracts from 818-4 and BXPC-3 pancreatic cell lines using monoclonals 1G8 for Rad51 and DO-1 (Dianova, Hamburg, Germany) for p53 detection. The Super Signal Substrate system (Pierce, Rockford, IL, USA) was used for chemiluminescence detection. Exposure time to detect Rad51 protein was about 20 min as compared to 45 s for p53

the Western blot reveals about a ®vefold dierence for specimens. Tumour specimens were scored positive Rad51 between monolayer and spheroid. Given that when more than 5% of tumour cell nuclei were stained Rad51 is over-expressed only in 20% of 818-4 cells, as intense as in spheroids or xeno-transplants. this argues that the level of Rad51 in over-expressing According to these criteria 27 (66%) out of 41 spheroid cells is increased 25-fold compared to pancreatic adenocarcinoma specimens expressed monolayers. Rad51 protein at high-levels. To further elucidate whether the observed accumula- Recent evidence suggests that expression of Rad51 is tion of Rad51 in spheroids might also occur under in regulated at the transcriptional level (Xia et al., 1997; vivo conditions, Rad51 expression was analysed in an Ohnishi et al., 1998). This prompted us to study the 5'- orthotopic xeno-transplantation model. 106 PancTU-I regulatory region of the human rad51 gene. An 8.1 kb cells were inoculated into the pancreas of SCID mice. DNA fragment of the 5'-region of this gene was Mice were sacri®ced on day 21 after inoculation. sequenced (GenBank accession number: AF203691). PancTU-I tumours were ®xed, paran embedded, and The 5'-UTR involves the ®rst exon and a 3.3 kb- prepared for immunohistochemistry. To rule out that nucleotide sequence encompasses the ®rst intron. The paran embedding would aect the outcome of the translation start codon is located immediately at the experiment, PancTU-I cells grown as monolayer were beginning of the second exon and the predicted processed under identical technical conditions. High- regulatory region has been pointed 5.4 kb upstream level Rad51 protein expression was detected only in of ATG. Nucleotide sequence analysis of this region PancTU-I tumours (Figure 3b) but not in cells grown revealed consensus sequences known to be involved in as monolayer (Figure 3a). These data show that high- RNA polymerase II mediated transcription. No level expression of Rad51 represents a unique feature TATA-like or initiator element sequences were found, of tumour cells grown as three-dimensional network in suggesting that the rad51 gene could belong to the vitro as well as in vivo. TATA-less GC-rich housekeeping gene family. In To further substantiate these ®ndings Rad51 expres- addition, 13 dierent pancreatic cancer cell lines and sion was investigated in paran embedded specimens 12 tumour specimens were screened for alterations in of human pancreatic adenocarcinoma. As shown in the Rad51 coding sequence using the highly sensitive Figure 3c, Rad51 protein accumulates to high levels in and speci®c non-isotopic RNase cleavage assay, but no tumour cell nuclei. Rad51 over-expression is restricted mutations were found (data not shown). Obviously, the to tumour cells and not found in nuclei of stroma cells. observed over-expressed Rad51 represents the wild- Intense staining was highly speci®c for Rad51 protein, type protein. since pre-incubation of the anti Rad51 monoclonal In order to create a model system in which the 1G8 with a peptide corresponding to the epitope biological consequences of modulating the Rad51 recognized by the antibody (Buchhop et al, 1996) content in cells can be easily monitored, cell clone completely blocked the staining reaction (data not UiRad51 was established. Treating these cells with shown). The percentage of Rad51 positive tumour cells muristerone A, an analogue of the insect steroid ranged from 5% to nearly 50% between dierent hormone ecdysone, induces ectopic expression of

Oncogene Over-expression of Rad51 in pancreatic cancer H Maacke et al 2793 Rad51 (Miska et al., submitted for publication). To numbers of non-induced versus induced controls not test whether high-level expression of Rad51 might be treated with calicheamicin g1. advantageous for cell survival after the induction of Variations of Rad51 expression have been observed DNA double strand breaks, cells were treated with during cell cycle progression (Yamamoto et al., 1996; various concentrations of calicheamicin g1 for 16 h Chen et al., 1997) and during immortalization of followed by ®xation of the cells after 72 h of recovery. primary human ®broblasts (Xia et al., 1997). In our Massive cell death occurs in response to increasing hands, dierences in Rad51 protein levels in cancer concentrations of calicheamicin g1 (Figure 4). However, cells are hardly detectable using antibody based over-expression of Rad51 signi®cantly potentiated the methods. By contrast, over-expressed Rad51 in tumour rate of survival compared to cells expressing basal specimens and 3D cultures can readily be observed. levels. Over-expression of Rad51 induces cell cycle Taking into account that only about 20% of the cells arrest in G1 and G2/M (Miska et al., submitted for over-express Rad51, the level in this sub-population publication). This explains the dierences in cell should be elevated by a factor of at least 25. This dimension of over-expression by far exceeds the undulations of Rad51 described for monolayer cultures, raising the question on the mechanisms behind A this phenomenon. While cancer cells predominantly express high levels of mutant rather than wild-type p53, over-expression of Rad51 protein is not associated with alterations in the coding region. Up-regulation of rad51 expression has been reported on the transcriptional level during immortalization of primary human ®broblasts (Xia et al., 1997). From our analysis of the 5'-regulatory region, the rad51 gene appears to contain a TATA-less, GC-rich promoter known from housekeeping genes. All evidence suggests that a combina- tion of 3-dimensional cell contact and aspects of a malignant phenotype are required. Current investiga- tion concentrates on the identi®cation of involved regulatory elements. In a DNA repair pathway, recombinational processes may act to maintain genetic stability, but if deregulated or increased, genomic instability and malignant transformation can result. There is a striking similarity between Rad51 over-expression in human tumours and the over-expression of its homologue RecA during SOS response in E. coli. It has been argued that the normal mutation rate of mammalian cells is insucient to generate the changes required in a single cell and that a condition of genetic instability is a necessary concomitant of tumorigenesis (for review, see Loeb, 1998; Strauss, 1998; Israel, 1996). Although classical functions of RecA during SOS response B appear to be conserved in mammalian Rad51, there are fundamental dierences concerning regulatory aspects. Over-expression of RecA is predominantly induced by DNA damage. This stimulus does not lead to an increase in Rad51 protein levels (Haaf et al., 1995 and data not shown). It is obvious that there is no analogy in bacteria to the interaction of Rad51 with dierent tumour suppressor proteins (StuÈ rzbecher et al., 1996; Scully et al., 1997; Sharan et al., 1997). BRCA1, BRCA2 and p53 are intimately involved in the regulation of cell survival, cell proliferation and DNA repair, processes that might be modulated by over-expressed Rad51. It will be of great interest to learn how over-expression of Rad51 in tumours relates to functional aspects of these tumour suppressors. Figure 2 Accumulation of Rad51 protein in human pancreatic The capability of tumour cells to become resistant cancer cell lines grown as spheroids. (a) Immunohistochemistry. PancTU-I and 818-4 cells were grown as spheroids and analysed against radio- and/or chemotherapy represents a major by immunohistochemistry. Spheroids were harvested and treated obstacle for successful treatment of patients suering as described in Figure 3. Rad51 was detected using monoclonal from advanced stages of cancer. Our data imply that antibody 1G8, counter-staining with Mayer's hemalum. The order only 3D systems reliably re¯ect the expression status of of magni®cation is indicated. (b) Western-blot analysis. Western blot analysis of Rad51 protein in 50 mg of total lysates from Rad51 in human pancreatic cancer. Dysregulation of PancTU-I and 818-4 pancreatic cell lines grown as monolayer (a) DNA repair pathways has been discussed to contribute or as spheroid (b) using monoclonals 1G8 for Rad51 detection to an enhanced resistance against chemo- and radio-

Oncogene Over-expression of Rad51 in pancreatic cancer H Maacke et al 2794

Figure 3 Rad51 expression in pancreatic monolayer cells, in xeno-transplants thereof and in pancreatic adenocarcinoma specimens. Rad51 expression was determined by immunohistochemistry in: (a) PancTU-I monolayer cells, (b) PancTU-I cells growing as tumours in SCID mice after orthotopic transplantation or (c) in dierent specimens of human pancreatic adenocarcinoma. For immunohistochemical staining, tissue from 41 surgical resections (38 Whipple specimens and three left pancreatectomies) from patients with ductal adenocarcinoma of the pancreas were classi®ed according to the criteria of the WHO (KloÈ ppel et al., 1996). The mean age of the patients was 59.7 years (45 ± 75). The tumour stages were pT2 (n=6), pT3 (n=29) and pT4 (n=6). Histologically, six cases were grade 1, 24 grade 2 and 11 grade 3. Specimens were ®xed in neutral buered formalin (4%) and subsequently embedded in paran. Consecutive 5 mm thick sections were placed on slide glasses, dewaxed in xylene, rehydrated by passage through alkohol and washed in PBS. As antigen retrieval treatment the sections were immersed in citrate buer (100 mM Na-citrate, pH 7.2) and boiled for 3 min under pressure in a pressure cooker. Following this treatment, endogenous peroxidase activity was blocked by incubation in 0.3% hydrogen peroxide in PBS for 10 min. Sections were incubated with monoclonal 1G8 for 1 h at room temperature in a wet chamber. After washing in PBS, a biotinylated anti-mouse antibody (Vector Laboratories, Burlingame, USA) was added for 30 min at room temperature. Sections were incubated with ABC complexes (Vector Laboratories) and subsequently stained using DAB. Counter-staining was performed using Mayer's hemalum. Order of magni®cation is indicated

Figure 4 Over-expression of Rad51 confers resistance to DNA double strand breaks. UiRad51 cells were plated at identical cell numbers and allowed to adhere for 24 h. Induction of ectopic protein production was induced with 1 mM muristerone A, non- induced cells received 1% ethanol. Twenty-four hours after induction, cells were treated with calicheamicin g1 at the concentrations indicated for 16 h, washed in complete medium and allowed to recover for 72 h. Subsequently, cells were stained with crystal violet. Cell survival was quanti®ed using a GS700 densitometer (Biorad, Munich, Germany)

Oncogene Over-expression of Rad51 in pancreatic cancer H Maacke et al 2795 therapy of 3D cultures (Mueller-Klieser, 1997). Indeed, Acknowledgements elevated levels of Rad51 have been correlated with We thank Maike Pacena for her technical assistance in enhanced cellular resistance against radiation (Yanagi- preparing the tumour slides. mRNA samples were kindly sawa et al., 1998; VispeÂ et al, 1998) and confer higher provided by PD Dr Hartmut Juhl and Dr Yuan Gao, recombination rates to mammalian cells (VispeÂ et al., orthotopic tumours of SCID mice were provided by Dr Alexander Fiedler, Clinic of General Surgery and Thoracic 1998, Xia et al., 1997). Our data show that engineered Surgery, Christian-Albrechts-University, Kiel. Calicheami- over-expression of Rad51 confers enhanced resistance cin g1 was a gift by Wyeth-Ayerst Research, Pearl River, towards DNA double strand breaks induced by NY, USA. This work was supported by Deutsche Kreb- calicheamicin g1 and that in vivo models and 3D shilfe/Dr Mildred Scheel Stiftung (10-0944-St I, II) and culture systems for in vitro studies are more suitable to Wilhelm-Sander Stiftung (97.020.01). H Kaltho was uncover the role of Rad51 in tumour biology than supported by a grant given by the Medical Faculty of the standard monolayer culture. Using these systems will University of Kiel (IZKF). The work is part of the MD be an important step towards understanding the theses of K Jost and S Opitz. mechanisms that generate multiple mutations in cancer cells and to test whether Rad51 might be a suitable target for prevention and treatment of cancer.

References

Baumann P and West SC. (1998). Trends Biochem. Sci., 23, Ohnishi T, Taki T, Hiraga S, Arita N and Morita T. (1998). 247 ± 251. Biochem. Biophys. Res. Commun., 245, 319 ± 324. Buchhop S, Donzelmann B, Nastainczyk W and StuÈ rzbecher Olive PL and Durand RE. (1994). Cancer Metastasis Rev., H-W. (1996). Hybridoma, 15, 205 ± 210. 13, 121 ± 138. Chen F, Nastasi A, Shen Z, Brennemann M, Crissman H and Scully R, Chen J, Plug A, Xiao Y, Weaver D, Feunteun J, Chen DJ. (1997). Mutat. Res., 384, 205 ± 211. Ashley T and Livingston DM. (1997). Cell, 88, 265 ± 275. Desoize B, Gimonet D and Jardiller JC. (1998). Anticancer Sharan SK, Morimatsu M, Albrecht U, Lim DS, Regel E, Res., 18, 4147 ± 4158. Dinh C, Sands A, Eichele G, Hasty P and Bradley A. Haaf T, Golub EI, Reddy G, Radding CM and Ward DC. (1997). Nature, 386, 804 ± 810. (1995). Proc. Natl. Acad. Sci. USA, 92, 2298 ± 2302. Strauss BS. (1998). Genetics, 148, 1619 ± 1626. Israel L. (1996). J. Theor. Biol., 178, 375 ± 380. StuÈ rzbecher H-W, Donzelmann B, Henning W, Knippschild Kaltho H, Schmiegel W, Roeder C, Kasche D, Schmidt A, U and Buchhop S. (1996). EMBO J., 15, 1992 ± 2002. Lauer G, Thiele HG, Honold G, Pantel K, RiethmuÈ ller G, VispeÂ S, Cazaux C, Lesca C and Defais M. (1998). Nucl. Scherer E, Maurer J, Maacke H and Deppert W. (1993). Acids Res., 26, 2859 ± 2864. Oncogene, 8, 289 ± 298. Xia SJ, Shammas MA and Reis RJ. (1997). Mol. Cell Biol., KloÈ ppel G, Solcia E, Longnecker DS, Capella C and Sobin 17, 7151 ± 7158. LH. (1996) Histological typing of tumours of the exocrine Yamamoto A, Taki T, Yagi H, Habu T, Yoshida K, pancreas. Second edition. WHO International histological Yoshimura Y, Yamamoto K, Matsushiro A, Nishimune classi®cation of tumours, Springer Verlag, Berlin. Y and Morita T. (1996). Mol. Gen. Genet., 251, 1±12. Loeb LA. (1998). Adv. Cancer Res., 72, 25 ± 56. Yanagisawa T, Urade M, Yamamoto Y and Furuyama J. Mueller-Klieser W. (1997). Am.J.Physiol.,273, C1109 ± (1998). Oral Oncol., 34, 524 ± 528. C1123.

Oncogene