Oncogene (2000) 19, 1473 ± 1476 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc SHORT REPORT Activity of MDM2, a ubiquitin ligase, toward p53 or itself is dependent on the RING ®nger domain of the ligase

R Honda1 and H Yasuda*,1

1School of Life Science, Tokyo University of Pharmacy and Life Science, 1432-1 Horinouchi, Hachioji, Tokyo 192-0392, Japan

We previously showed that oncoprotein MDM2 has phosphorylated at ser15 through ATM kinase activa- ubiquitin ligase activity toward tumor suppressor p53. In tion (Banin et al., 1999; Christine et al., 1999). The that paper, we showed very weak homology in the phosphorylated p53 seems not to be good substrate for carboxyl terminal portion between MDM2 and E6AP MDM2 (Honda et al., 1999) because of its defective (HECT domain). We mutated the cysteine residue binding to MDM2 (Shieh et al., 1997). As a result, the (C464) corresponding to the residue essential for the amount of p53 in the cell increases. ubiquitin ligase activity of E6AP and this mutation There are several types of E3 or ubiquitin ligase in diminished the ligase activity of MDM2. The cysteine mammals. One of them, E6AP, works as ubiquitin residue described above is also one of the cysteine ligase toward p53 in the presence of E6 protein of residues that form the RING ®nger domain of MDM2. papilloma virus 16 or 18. The catalytic site of the We tried to ®nd out whether the diminishing of the ubiquitin ligase is located in the carboxyl terminal activity by the mutation is attributable to the disruption portion of E6AP and this portion is named as HECT of the RING ®nger domain or not. When the ring ®nger domain. (Ciechanover, 1998). The MDM2 may also be domain of MDM2 was deleted, the truncation mutant the HECT-type E3, because in its carboxyl terminus, did not have the ubiquitin ligase activity. When we there is a weak homologous sequence with HECT mutated the seven cysteine residues of RING ®nger domain sequence (Figure 1a). When we aligned the domain of MDM2 in the carboxyl terminus, the sequence of the carboxyl terminal region of MDM2 disruption of each residue in the RING ®nger completely with the sequence of the HECT domain of E6AP, we diminished the ubiquitin ligase activity of MDM2 toward found that the C464 of MDM2 corresponded to the MDM2 itself and toward tumor suppressor p53. These catalytic cysteine of the HECT domain (Honda et al., data indicate that the RING ®nger domain in MDM2 is 1997). When the cysteine residue (C464) in the essential for its ubiquitin ligase activity toward p53 and carboxyl terminus of MDM2, which is a residue itself. Oncogene (2000) 19, 1473 ± 1476. essential for the ubiquitin ligase activity of HECT protein, was mutated to alanine, the ubiquitin ligase Keywords: MDM2; p53; ubiquitin ligase; E3; RING activity was not detected any more (Honda et al., ®nger domain 1997). On the other hand, this cysteine residue of MDM2 is also one of the cysteine residues in the RING domain of MDM2 (Boddy et al., 1994). Thus, The p53 tumor suppressor gene product induces the the mutation described also disrupts the RING ®nger expression of various genes (Gadd45, Waf1/p21/Cip1, domain. Recently, a novel component of SCF, ROC1/ cyclinG, Bax and MDM2) regulating cell-cycle arrest Rbx1, was isolated and shown to be the catalytic site of and apoptosis (Ko et al., 1996). The degradation of SCF (Kamura et al., 1999, Ohta et al., 1999, Skowyra p53 is regulated by the ubiquitin-proteasome system et al., 1999). Furthermore, the RING domain in its (Maki et al., 1996). In the ubiquitin conjugation carboxyl terminus was shown to be essential for this system, there are three enzymatic steps, i.e., E1, E2, catalytic activity (Ohta et al., 1999). Also, one of the and E3 (Ciechanover, 1998). The E3 enzyme, ubiquitin APC/C components, APC11, was shown to have ligase, recognizes the speci®c substrate and ubiquiti- similar RING ®nger domain in its carboxyl terminus nates the target protein in the presence of E1 and E2. (Kamura et al., 1999, Ohta et al., 1999). In cells infected with human papilloma virus, HPV16 Alignment of the sequence of the carboxyl terminus or 18, papilloma virus E6 and cellular E6AP (E6 of MDM2 with the sequences of ROC1 and APC11 associated protein) form complex and function as revealed very similar RING ®nger domains in them ubiquitin ligase, E3 toward p53 (Sche€ner et al., (Figure 1b). When the truncated form of MDM2 (1 ± 1993). In non infected cells, MDM2 works as ubiquitin 444), in which the RING ®nger domain was deleted, ligase, E3, in the presence of E1 and UbcH5 as E2 was checked for the ubiquitin ligase activity, the (Honda et al., 1997). The transfection of the cells with activity toward itself (auto-ubiquitination) and that the MDM2 gene accelerates the p53 degradation toward p53 was no longer detected (Figure 2). Another (Haupt et al., 1997; Kubbutat et al., 1997). When the truncated form of MDM2 (1 ± 489), in which the cells are exposed to genotoxic stress, the p53 is RING ®nger domain was intact but the phenylalanine residue (F490) corresponding to the residue essential for the ubiquitin ligase activity of E6AP, did not lose *Correspondence: H Yasuda the ubiquitin ligase activity (data not shown). This Received 19 October 99; revised 10 January 2000; accepted 14 ®nding showed that the RING ®nger domain may be January 2000 essential for the ubiquitin ligase activity. Then Figure RING domain of MDM2 is essential for its activity R Honda and H Yasuda 1474

Figure 1 (a) Alignment of carboxyl terminal portion of MDM2 (amino acids 438-491; Z12020), and HECT domain of E6AP (amino acids 794-852; Q050869). Identical amino acids were shown by asterisks. A cysteine residue (C464) essential for the ubiquitin ligase activity of E6AP was underlined. (b) Alignment of RING ®nger domains of MDM2 (amino acids 438-479; Z12020), ROC1/ Rbx1 (amino acids 110-138; AF142059), and APC11 (amino acids 115-140; Z74056). Their consensus sequence that form the RING ®nger domain contains seven cysteine residues (C) and one histidine residue (H)

AB

Figure 2 Deletion of carboxyl terminal region of MDM2 diminished the ubiquitin ligase activity of MDM2. Wildtype MDM2 (1 ± 491) and truncated MDM2 (1 ± 444) were expressed in the baculovirus protein expression system. (a) Ubiquitin ligase activity of MDM2 toward MDM2 itself. Left panel; Ubiquitination of MDM2 detected by labeled ubiquitin. Wildtype MDM2 was ubiquitinated but the truncated mutant was not. Right panel; the amount of MDM2 proteins added to reaction was detected by anti-MDM2 antibody. (b) Ubiquitin ligase activity of wildtype or truncated MDM2 toward p53. Left panel; the ubiquitinated protein ladder of p53 was detected by immunoblotting using anti-p53 antibody. Right panel; the amount of MDM2 proteins added to reaction was detected by anti-MDM2 antibody. In vitro ubiquitination assay was performed as described previously (Honda et al., 1999). Brie¯y the standard reaction mixture (50 ml) contained Sf-9 cell extract expressing mouse E1 (500 ng), UbcH5 (500 ng), biotinylated ubiquitin (15 mg) and GST-MDM2 (500 ng) or truncated MDM2, 50 mM Tris-HCl (pH 7.4) 5 mM MgCl2,2mM ATP and 2 mM dithiothreitol (DTT). After incubation at 258C for 30 min, GST-MDM2 or truncated MDM2 was bound to glutathione Sepharose 4B resin (Pharmacia) and resolved by SDS-7.5% PAGE. The proteins in the gel were transferred to a PVDF membrane (Millipore). The PVDF membrane was incubated with ExtrAvidin (Sigma). Ubiquitinated proteins were detected by use of ECL (Amersham). When the ubiquitination of p53 was checked, p53 was added to the reaction mixture and detected by anti-p53 antibody (DO-1)

4c shows that all of the mutants could bind to p53 but mutants could still bind to p53 (Figure 4c), the they did not have ubiquitin ligase activity toward p53, function of RING ®nger domain does not seem to be although the mono, di-ubiquitinated p53 was observed substrate recognition. Since ROC1 in SCF seems to be in the presence of E2 (Figure 4a). The C464 is the necessary for E2 binding, the RING domain may have residue which is correspondent to the residue essential important role for interaction with E2. for the ubiquitin ligase of E6AP (Figure 1a). The Many proteins have been shown to have RING activity of the mutant C464A was diminished com- ®nger domains. Thus, some of these proteins may pletely. Furthermore, when the cysteine residue, C449 function as E3 or ubiquitin ligase at least toward in CXXH motif was mutated to alanine the mutant, themselves. Actually, recent reports suggests that C449A did not have ubiquitin ligase activity toward RING ®nger proteins have ubiquitin ligase activity MDM2 itself nor toward p53 (Figure 4d). These data toward themselves (Lorick et al., 1999). More recently, show that the RING ®nger domain of MDM2 is the RING ®nger domain of c-Cbl, an adapter protein essential for the ubiquitin ligase activity toward for receptor protein-tyrosine kinases (Joazeiro et al., MDM2 itself and toward p53. Since the RING ®nger 1999) or Ubr1p, E3 component of the N-end rule

Oncogene RING domain of MDM2 is essential for its activity R Honda and H Yasuda 1475

Figure 3 E€ect of destruction of RING structure on ubiquitin ligase activity of MDM2 toward MDM2 itself. Left panel shows ubiquitin ligase activity of mutated MDM2 toward MDM2 itself. Six cysteine residues (C438, C441, C461, C464, C475, C478) in RING domain of MDM2, shown in Figure 1, were mutated to alanine residues by PCR method using primer(s) containing mutation base(s). In vitro ubiquitination assay was performed as described in Figure 2. Right panel shows the amount of wildtype or mutated MDM2 added to the reaction, which was detected by immunoblotting

A C

B D

Figure 4 Diminishing of the ubiquitin ligase activity of MDM2 toward p53 by the destruction of the RING ®nger domain. (a) Ubiquitination of p53 was detected by anti-p53 antibody. The ubiquitination assay was performed with or without UbcH5 (E2). (b) The amount of MDM2 proteins added to the reaction, shown in (a). (c) The binding of mutated MDM2 to p53 was checked. A Sf-9 cell extract expressing GST-MDM2 was incubated with an extract expressing p53, followed by precipitating GST-MDM2 with glutathione Sepharose 4B resin. Wildtype MDM2, mutated MDM2, and p53 bound to MDM2 were detected by immunoblotting. (d) Ubiquitination of p53 was detected by anti-p53 antibody. Ubiquitination of MDM2 itself was detected by use of biotinylated ubiquitin as described in Figure 2 pathway in Saccharomyces cerevisiae (Xie et al., 1999) Acknowledgments was also shown to be required for the ubiquitin ligase We would like to express thanks to Ms Akiko Hotta for activity. The ROC1/Rbx1 or APC11 possibly has her technical assistance. This work was supported in part catalytic ubiquitin ligase activity in SCF or APC/C by a grant-in-aid from the Ministry of Education, Science, but does not recognize substrate by itself. The MDM2, Culture and Sports in Japan. however, has the catalytic activity of ubiquitin ligase in the RING ®nger domain and recognizes the substrate, p53, by its amino terminal portion (Chen et al., 1993).

References

Banin S, Moyal L, Shieh S-Y, Taya Y, Anderson CW, Canman CE, Lim D, Cimprich KA, Taya Y, Tamai K, Chessa L, Smorodinsky NI, Prives C, Reiss Y, Shiloh Y Sakaguchi K, Appella E, Kastan MB and Siliciano JD. and Ziv Y. (1998). Science, 281, 1674 ± 1677. (1998). Science, 281, 1677 ± 1679. Boddy MN, Freemont PS and Borden KL. (1994). Trends Chen J, Marechal V and Levine AJ. (1993). Mol. Cell. Biol., Biochem. Sci., 19, 198 ± 199. 13, 4107 ± 4114.

Oncogene RING domain of MDM2 is essential for its activity R Honda and H Yasuda 1476 Ciechanover A. (1998). EMBO J., 17, 7151 ± 7160. Lorick K, Jensen JP, Fang S, Ong AM, Hatakeyama S and Haupt Y, Maya R, Kazaz A and Oren M. (1997). Nature, Weissman AM. (1999). Proc. Natl. Acad. Sci. USA, 96, 387, 296 ± 299. 11364 ± 11369. Honda R and Yasuda H. (1999). EMBO J., 18, 22 ± 27. Maki CG, Huibregtse JM and Howley PM. (1996). Cancer Honda R, Tanaka H and Yasuda H. (1997). FEBS Lett., 420, Res., 56, 2649 ± 2654. 25 ± 27. Ohta T, Michel JJ, Schottelius AJ and Xiong Y. (1999). Mol. Joazeiro CA, Wing SS, Huang H, Leverson JD, Hunter T Cell, 3, 535 ± 541. and Lui YC. (1999). Science, 286, 309 ± 312. Sche€ner M, Huibregtse JM, Vierstra RD and Howley RD. Kamura T, Koepp DM, Conrad MN, Skowyra RJ, More- (1993). Cell, 75, 495 ± 505. land RJ, Iliopoulos O, Lane WS, Kaelin Jr WG, Elledge Shieh SY, Ikeda M, Taya Y and Prives C. (1997). Cell, 91, SJ, Conway RC, Harper JW and Conway JW. (1999). 325 ± 334. Science, 284, 657 ± 661. Skowyra D, Koepp DM, Kamura T, Conrad MN, Conway Kastan MB and Siliciano JD. (1998). Science, 281, 1677 ± RC, Conway JW, Elledge SJ and Harper JW. (1999). 1679. Science, 284, 662 ± 665. Ko LJ and Prives C. (1996). Genes Dev., 10, 1054 ± 1072. Xie Y and Varshavsky A. (1999). EMBO J., 18, 6832 ± 6844. Kubbutat MHG, Jones SN and Vousden SN. (1997). Nature, 387, 299 ± 303.

Oncogene