Oncogene (2001) 20, 8276 ± 8280 ã 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00 www.nature.com/onc SHORT REPORTS The interacts and cooperates with in regulation of transcription and cell growth control

Igor V Garkavtsev*,1 Nikolai Kley2, Irina A Grigorian3 and Andrei V Gudkov*,1

1Department of Molecular Biology, Lerner Research Institute, The Cleveland Clinic Foundation, Cleveland, Ohio, OH 44195, USA; 2GPC Biotech Inc., 610 Lincoln Street, Waltham, Massachusetts, MA 02451, USA; 3Department of Pharmacokinetics and Pharmacodynamics, University of Illinois at Chicago, Chicago, Illinois, IL 60607, USA

Bloom syndrome is an autosomal recessive disorder types of human cancers (Wallace-Brodeur and Lowe, associated with mutations in BLM encoding 1999; Prives and Hall, 1999). Germ line p53 de®ciency protein that belongs to the family of DNA . is also associated with inherited disease (Li-Fraumeni It is characterized by predisposition to cancer, syndrome) that has many similarities with the Bloom immunode®ciency, high sensitivity to UV and genomic syndrome (German, 1993; Akashi and Koe‚er, 1998). instability of somatic cells. Here we show physical and Several di€erent mechanisms are involved in p53- functional cooperation between BLM and p53 . mediated control of genomic stability including induc- Ectopic expression of BLM causes anti-proliferative tion of p53-dependent growth arrest or apoptosis in e€ect in p53 wild type, but not in p53-de®cient cells; response to genotoxic stress, modulation of DNA p53-mediated transactivation is attenuated in primary repair machinery, control of accuracy of chromosomal ®broblasts from Bloom syndrome patients. BLM and segregation in mytosis, etc.; p53 exerts these functions p53 proteins physically interact in the cells as through physical and functional interaction with demonstrated in yeast and mammalian two-hybrid numerous cellular proteins (Prives and Hall, 1999). systems; interaction sites are mapped to 237 ± 272 aa Potential functional cooperation between BLM and residues of BML and 285 ± 340 aa of p53. Ectopic p53 has been suggested by Lu and Lane (1993), based expression of the fragment of wild type BML contain- on their observations of attenuated p53 response in the ing p53-interactive domain suppresses p53-mediated cells from Bloom syndrome patients. However, there transcription and interferes with p53-mediated growth has been no evidence of a mechanistic link between inhibition. These observations indicate that BLM might these two proteins. be an important component of p53 function and suggest This work was initiated by two independent that Bloom Syndrome phenotype may in part be the observations: (i) di€erence in growth suppressive result of the deregulation of the p53 tumor suppressor activity of ectopically expressed BLM in two sarcoma pathway. Oncogene (2001) 20, 8276 ± 8280. cell lines di€ering in their p53 status (Figure 1a), and (ii) strong interaction of BML and p53 in yeast two Keywords: Bloom syndrome; BLM protein; p53; inter- hybrid system con®rmed by co-precipitation of two action; two-hydrid system; transactivation proteins from mammalian cells (see below). These observations stimulated us to analyse p53-BML interaction in greater detail. BLM cDNA expressed in mammalian expression Bloom syndrome results from mutations inactivating vector pBK from CMV promoter was transfected into BLM gene that encodes a RecQ family DNA human osteosarcoma cell lines, Saos2 (p53-de®cient) (Mohaghegh et al., 2001). Patients with Bloom and U2OS (wild-type p53), along with the insert-free syndrome are hypersensitive to UV radiation, su€ering vector. There was a dramatic di€erence between two from sun-sensitive facial erythema, and show predis- cell lines in their sensitivity to ectopic expression of position to tumor development (German, 1993). BLM- BLM. While no di€erence was found in colony de®cient cells from such individuals are characterized formation between vector and BLM plasmids in p53- by a dramatic increase in sister chromatid exchange, negative Saos-2, BLM was clearly growth suppressive breaks and abnormally high mutation in p53-wild type U2OS cells (Figure 1a). These rate (Ellis et al., 1994). Genomic instability is also observations suggested that the growth-inhibitory e€ect associated with de®ciency of the p53 tumor suppressor of BLM required the presence of functional wild-type gene, a key mediator of cellular response to a variety of p53. stresses that is frequently mutated in many di€erent To con®rm whether BLM-mediated growth suppres- sion is, in fact, p53 dependent, we analysed the e€ect of p53 suppression on BLM-mediated growth inhibition by BLM-expressing plasmid in a more accurate *Correspondence: IV Garkavtsev and AV Gudkov Received 11 October 2001; revised 31 October 2001; accepted 31 experimental setup independent of the possible di€er- October 2001 ences in cell types. A variant of Balb 3T3-derived cell BLM-p53 interaction IV Garkavtsev et al 8277

Figure 1 Growth suppression e€ect of BLM depends on the activity p53. (a) Human osteosarcoma cell lines (U2OS) and Saos2) di€ering in their p53 status (wild type and p53-de®cient, respectively) were transfected with the insert-free vector pBK (Stratagene) or pBK-BLM (full-length human BLM cDNA expressed from CMV promoter). Following 3 weeks of growth in the present of G418, plates were boxed amd stained with crystal violet. (b)105 ConA cells plated on 60-mm plates were transfected with 2 mgof retroviral vector plasmid, pBABE-phleo, (Morgenstern et al., 1990) expressing full-length cDNA for human BLM gene and bacterial bleo gene, confering resistance to phleomycin. BLM-expressing plasmid was mixed either with 3 mg of either empty pBABE-puro vector or with the pBABE-puro, expressing GSE56, a potent inhibitor of p53 function (Ossovskaya et al., 1996). Transfection eciency was estimated by adding GFP-expressing construct into transfection and calculating the proportion of ¯uorescent cells as described above. Fifty mg/ml of phleomycin were added 48 h after transfection and numbers of phleomycin resistant colonies were estimated after selection on phleomycin was completed. The results obtained demonstrate that GSE56 highly increased the number of phleomycin-resistant colonies, indicating that BLM-mediated cytotoxicity is indeed p53-dependent

Figure 2 Physical interaction between BLM and p53 proteins in yeast two hybrid system. (a) Mapping of p53-interacting domain in BLM protein. Di€erent portions of BLM gene (shown on top) were fused to the pJG4-5 plasmid were used as preys. Cdk4-pJG4- 5 was used as a negative and SV40-pJG4-5 as a positive control. Bait construct expressed p53 cDNA with truncated oligomerization domain in pEG202 vector (83 ± 393 aa). Interaction was detected by yeast growth on galactose leu ± plate and by activation of lacZ on the X-Gal plate. (b) Mapping of BLM-interacting domain in p53 protein. The experiment was done as described above (a). A series of p53-expressing `bait' constructs (shown on top) were tested for interaction with BLM p53-binding fragment (152 ± 500 aa) in `prey' plasmid. Max-pJG4-5 and SV40-pJG4-5 were shown as negative and positive controls, respectively

Oncogene BLM-p53 interaction IV Garkavtsev et al 8278 line, ConA, was used (Komarova et al., 1997). The results obtained (see Figure 1b) indicate that colony suppression by BLM-expressing construct was e€ec- tively abrogated by a dominant negative p53 mutant, GSE56, thus con®rming that the growth suppressive e€ect of BLM is p53-mediated. In parallel, we de®ned BLM protein as a p53 interactor in yeast two-hybrid selection (LexA system, Clontech Laboratories) by screening HeLa cDNA library using truncated human p53 cDNA as bait. A series of bait/prey constructs expressing truncated versions of p53 and BLM were generated (see Figure 2) to de®ne the interacting sites in these two proteins. Analysis of these results allowed us to map interacting domains to 237 ± 272 aa of BLM and 285 ± 340 aa of p53. Interaction of p53 and BLM detected in the yeast two-hybrid system suggested that these proteins might physically interact in the mammalian cell environment. In fact, p53 and BLM proteins can be co-immunopre- cipitated from human cells ectopically co-expressing both proteins (data not shown). However, none of these observations can serve as proof of functionality of p53-BLM interaction: p53 is known to be coprecipitated with numerous proteins in mammalian cell lysates while protein interaction in the two-hybrid system is a rather frequent yeast artifact. We therefore carried out the mammalian two-hybrid assay using Strategene Mammalian Two-Hybrid Assay Kit that allows veri®cation of protein ± protein interaction in the Figure 3 Interaction between Bloom protein and p53 monitored live cell environment. 1 ± 431 aa BLM DNA fragment by mammalian two hybrid system. BLM cDNA subcloned into that showed interaction with p53 in yeast fused with pCMV-AD vector and fused with NF-kB transcriptional the N-terminal portion of NF-kB tranactivation activation domain. p53 DNA fragment (72 ± 390 aa) was fused domain was used in these experiments in combination in frame to the carboxyl (C) terminus of the GAL4-DNA binding domain (pBD-53) and SV40 large T-antigen (84 ± 708 aa) was with the 72 ± 390 aa p53 fragment fused with the C- fused to the C-terminus of NF-kB p65 activation domain (pAD- terminus of NK-kB transactivation domain. Protein ± SV40T) and then these constructs were used as positive control protein interaction was determined by activation of for protein ± protein interaction assays. TRAF-2,13 (297 ± 503 aa) NF-kB-mediated transcription of a reporter construct which do not interact with p53 was used as negative control. Indicated combinations of plasmids (0.2 mg each) were transfected with luciferase. A series of constructs were used as together with the reporter plasmid pFR-Luc (0.25 mg) into 293 or controls of speci®city of the studied interaction. The H1299 cell lines and their luciferase activities were monitored in results of a representative experiment are shown in 48 h Figure 3 and demonstrates that the pair of BLM-p53 bait/prey plasmids induced activation of the reporter construct, indicative of physical interaction of these type p53 from CMV promoter and di€erent fragments two proteins in mammalian cells, resulting in restora- of BLM cDNA in sense or antisense orientation. Sense tion of a transactivator function of NK-kB 4 ± 5 times oriented BLM cDNA stimulated p53-mediated trans- lower than a positive control p53-SV40 T-ag pair. activation as judged by a 3 ± 4-fold increase in the None of the negative controls, including BLM levels of CAT activity in comparison to p53-expressing construct with a frameshift mutation (deletion of the plasmid alone; BLM-expressing plasmid did not 459 adenosine nucleotide) showed reporter activation activate the p21WAF/CAT reporter in the absence of higher than background. p53 (Figure 4a). Antisense BLM construct had an p53 is believed to exert its growth suppressive opposite e€ect, presumably by interfering with en- functions through modulation of transcription of a dogenous BLM gene expression. Vectors expressing number of p53-response , including p21WAF1/CIP1 truncated versions of BLM did not have the activity of (el Deiry et al., 1993). To directly test the ability of a full-length construct. Moreover, a BLM fragment BLM to modulate p53-mediated transcription, p53- capable of p53 binding had an e€ect similar to that of de®cient Saos2 cells were transfected with plasmid antisense construct, presumably through a dominant combinations, including a reporter plasmid expressing negative activity. BLM expression did not stimulate chloramphenicol acetyltransferase (CAT) gene from a transcription of E2F-responsive reporter construct, minimal promoter with p53 binding sites from human suggesting that BLM e€ects are speci®c for p53- p21 gene vector (el Diery et al., 1993) expressing wild- inducible promoters (data not shown).

Oncogene BLM-p53 interaction IV Garkavtsev et al 8279

Figure 4 BLM activates p53-responsive promoters and modulates growth inhibition by p53. (a) Expression of BLM activates the transcription of p53 target gene p21/WAF1. CAT ELISA kit for the quantitative determination of chloramphenicol acetyltransferase in transfected cells was used. CAT assays of Saos2 cells transfected with p21/WAF CAT reported plasmid together with p53 and di€erent BLM constructs were done. (b) Levels of p21 protein in MCF7 cells after transfection with 10 mgof the indicated constructs. Cell lysates were harvested in 48 h after transfection and 20 mg of protein per lane was analysed by Western blotting using an anti-p21 antibody. The arrow indicates p21 protein. (c) p53-de®cient variant of Balb 3T3 cells, line 10(1), growing on 60-mm plates were co-transfected with 1.5 mg of retroviral vector plasmid pPS-p53-hygro (Garkavtsev et al., 1998) together with 3.5 mg of either pBLM1 (154 ± 500 aa) or pBLM2 (501 ± 1418 aa) and 1 mg of pL(GFP)SN vector expressing enhanced green ¯uorescent protein to monitor the ecacy of transfection. Two hundred mg/ml of hygromycin were added 48 h after transfection and the numbers of growing colonies were estimated 14 days of hygromycin selection conferred by p53-expressing vector. Parallel plates were treated similarly, but instead of the construct expressing the fragments of BLM, were used either `empty' pBK vector (vec2) or retroviral vector pL(GSE56)SN, expressing GSE56. Transfection eciencies were estimated by the number of ¯uorescent GFP-expressing cells 48 h after transfection and the results of colony assay were normalized accordingly. Colony numbers are presented in comparison with the plates transfected with a mixture of two `empty' vectors (vec1+vec2). (d) Primary cultures of BLM-de®cient and wild type ®broblasts were co-transfected with 6 mg of p21/WAF CAT reported plasmid together with 2 mg pQB125 vector expressing enhanced green ¯uorescent protein to monitor the ecacy of transfection and CAT activity was estimated 24 h post-transfection

p53-modulating e€ects of full-length and truncated in the two-hybrid system, behaves similarly to the p53-binding BLM fragments were con®rmed on dominant negative mutant of p53 in co-transfection endogenous p21 and p53 proteins (Figure 4b): p53- experiments. interacting BLM fragment and antisense BLM The results of the above transfection experiments are construct interfered with p21 activation while full- consistent with the attenuated p53 transactivation length BLM cDNA expression had a stimulatory typical for the cells from Bloom syndrome patients as e€ect. demonstrated by comparison of p21-CAT reporter Biological signi®cance of this modulating e€ect was construct transfected into primary ®broblasts di€ering demonstrated by colony assay for growth-suppressing in their BLM status (Figure 4d). activity of ectopically expressed p53 (Figure 4c). The In conclusion, we have demonstrated that BLM results obtained indicate that the construct expressing protein physically and functionally interacts with p53 the fragment of BLM protein, de®ned as p53 interactor in cell growth control, presumably by modulating

Oncogene BLM-p53 interaction IV Garkavtsev et al 8280 (stimulating) p53-mediated transactivation. BLM is a p53 since neither premature aging (Werner syndrome) member of the RecQ subfamily of DexH box-contain- nor high frequency of sister chromatid exchange ing RNA and DNA helicases (Ellis et al., 1995). p53 (Bloom syndrome) are the characteristics of p53 has already been shown to interact with another family de®ciency. member, Werner syndrome protein (WRN), that binds Although WRN and BLM proteins share a high p53 through its C-terminus and is involved in induction degree of homology in DEAN helicase domains of p53-dependent apoptosis (Spillare et al., 1999). (Morozov et al., 1997), their p53-interacting interaction Furthermore, when this work was in preparation, proteins have no detectable similarities. The mechanics Wang et al. (2001) reported about p53-BLM interac- of p53 interaction with these two important helicases tion without mapping interacting domains. Werner and and how this interaction is translated into deregulated Bloom syndromes are inherited diseases characterized control of genomic stability remains to be understood. by cancer predisposition and genomic instability, both also known to be associated with p53 de®ciency (Li- Fraumeni syndrome). Deregulation of p53 response in the absence of WRN and BLM p53-cooperating proteins may be the underlying cause of the above Acknowledgments phenotypic abnormalities of Werner and Bloom We thank NA Ellis for providing full-length BLM cDNA. patients. At the same time, not all Bloom and Werner This work was supported by NIH grant CA75179 to AV syndrome-associated symptoms could be attributed to Gudkov.

References

Akashi M and Koe‚er HP. (1998). Clin. Obstet. Gynecol., Morozov V, Mushegian AR, Koonin EV and Bork P. (1997). 41, 172 ± 199. Trends Biochem. Sci., 22, 417 ± 418. El Deiry WS, Tokino T, Velculescu VE, Levy DB, Parsons R, Morgenstern JP and Land H. (1990). Nucleic Acids Res., 18, Trent JM, Lin D, Mercer WE, Kinzler KW and Vogelstein 3587 ± 3596. B. (1993). Cell, 75, 817 ± 825. Ossovskaya VS, Mazo IA, Chernov MV, Chernova OB, Ellis NA, Groden J, Ye TZ, Straughen J, Lennon DJ, Ciocci Strezoska Z, Kondratov R, Stark GR, Chumakov PM and S, Proytcheva M and German J. (1995). Cell, 83, 655 ± 666. Gudkov AV. (1996). Proc. Natl. Acad. Sci. USA, 93, Ellis NA, Roe AM, Kozloski J, Proytcheva M, Falk C and 10309 ± 10314. German J. (1994). Am.J.Hum.Genet.,55, 453 ± 460. Prives C and Hall PA. (1999). J. Pathol., 187, 112 ± 126. Garkavtsev IV, Grigorian IA, Chernov MV, Ossovskaya VS, Spillare EA, Robles AI, Wang XW, Shen JC, Yu CE, Chumakov PM and Gudkov AV. (1998). Nature, 391, Schellenberg GD and Harris CC. (1999). Genes Dev., 13, 295 ± 298. 1355 ± 1360. German J. (1993). Medicine (Baltimore), 72, 393 ± 406. Wallace-Brodeur RR and Lowe SW. (1999). Cell. Mol. Life KomarovaEA,ChernovMV,FranksR,WangK,ArminG, Sci., 55, 64 ± 75. Zelnick CR, Chin DM, Bacus SS, Start GR and Gudkov Wang XW, Tseng A, Ellis NA, Spillare EA, Linke SP, Robles AV. (1997). EMBO J., 16, 1391 ± 1400. AI, Seker H, Yang Q, Hu P, Beresten S, Bemmels NA, Lu X and Lane DP. (1993). Cell, 75, 765 ± 778. Gar®eld S and Harris CC. (2001). J. Biol. Chem., 276, MohagheghP,KarowJK,BroshJrRM,BohrVAand 32948 ± 32955. Hickson ID. (2001). Nucleic Acids Res., 29, 2843 ± 2849.

Oncogene