Oncogene (2001) 20, 2318 ± 2324 ã 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00 www.nature.com/onc

The `wildtype' conformation of p53: epitope mapping using hybrid proteins

Peter L Wang*,1, Fiona Sait1 and Greg Winter1,2

1Centre for Protein Engineering, Hills Road, Cambridge CB2 2QH, UK; 2Laboratory of Molecular Biology, Hills Road, Cambridge CB2 2QH, UK

The function of p53 correlates with its `wildtype' p53. Conversely, PAb240 and several other antibodies conformation, speci®cally recognized by antibodies bind to p53 in the `mutant' conformation and also to PAb1620 and PAb246, and many cancer-associated denatured p53 (Gannon et al., 1990; Legros et al., mutations cause loss of this conformation. The epitopes 1994b; Vojtesek et al., 1995). Drugs restoring `wild- of these antibodies were identi®ed using hybrid p53 type' conformation and function to mutant p53 could proteins created by a new method. Plasmids carrying induce tumor growth arrest and apoptosis; lead homologous genes cut at appropriate sites recombined compounds have been identi®ed using a high- eciently when transformed into RecE+ E. coli. throughput screen for stabilization of the PAb1620 PAb1620 and PAb246 recognize mouse but not chicken epitope against thermal denaturation (Foster et al., p53; we created mouse ± chicken hybrids of the p53 core 1999). domain and tested antibody reactivity. PAb246 binding Full-length p53 is a tetramer, while the core domain mapped to residues 201 ± 212, while PAb1620 required is a monomer but binds as a tetramer to DNA both residues 145 ± 157 and 201 ± 212 (human p53 (Balagurumoorthy et al., 1995; Wang et al., 1995). numbering used throughout). An alanine-scan showed The existing crystal structure shows only one core that the key residues for PAb246 and PAb1620 are domain speci®cally bound to DNA (Cho et al., 1994), completely distinct: PAb246 recognizes residues 202 ± so we do not have a direct picture of these higher-order 204 (Tyr-Pro-Glu) while PAb1620 recognizes residues interactions. PAb1620 and PAb246 have e€ects on the Arg156, Leu206, Arg209, and Gln/Asn210, the last two p53-DNA interaction; these could be interpreted residues being essential. Both antibody epitopes are far structurally once their epitopes are known. from the p53 interface with DNA, but near the epitope of The PAb1620 and PAb246 epitopes could not be the `mutant' conformation antibody PAb240. These localized using sets of short synthetic peptides epitope locations may help in dissecting the interactions covering all of mouse and human p53 (Legros et of p53, including those with E6/E6-AP and in its DNA- al., 1994a; Lane et al., 1996). Other epitope mapping bound state. Oncogene (2001) 20, 2318 ± 2324. attempts (Wade-Evans and Jenkins, 1985; Ravera et al., 1998) also did not evaluate epitopes in the Keywords: p53 conformation; epitope mapping; recom- context of a folded core domain. Our strategy for bination; p53-DNA interaction mapping the epitopes took advantage of the species- speci®city of the antibodies: both PAb1620 and PAb246 bind mouse p53 but not chicken p53 (PAb1620 also binds human p53). We used a rapid p53 is a transcription factor that responds to cellular recombination method to generate inter-species hy- stresses, such as DNA damage, by promoting cell-cycle brids of the p53 core domain. Hybrid proteins should arrest and apoptosis (reviewed by Levine, 1997). maintain the same overall folding but display Mutations in p53 are the most common genetic di€erent surface residues, and correspondingly di€er alterations known in cancer (reviewed by Hainaut et in their antibody reactivity. After epitopes were al., 1998). The vast majority are missense mutations mapped to small regions of sequence, they were within the central core domain that mediates sequence- de®ned further by alanine-scanning mutagenesis. We speci®c binding to DNA. discuss the implications of the location of these p53 mutations often cause a change from the epitopes. `wildtype' conformation to a `mutant' conformation that can be detected by monoclonal antibodies Recombination of homologous genes (MAbs) binding to the core domain. PAb1620 and PAb246 are speci®c for the `wildtype' conformation A powerful way to generate hybrid genes is by (Cook and Milner, 1990) and do not bind denatured homologous recombination (reviewed by Wang, 2001). We have developed a simple method in E. coli that generates hybrids at high frequency and does not require any special plasmid construction; hybrids can *Correspondence: PL Wang Received 14 December 2000; revised 29 January 2001; accepted 30 easily be recombined again to make more complex January 2001 hybrids. Preliminary studies showed that two factors p53 `wildtype' conformation epitopes PL Wang et al 2319 were important for ecient recombination between co- EcoRI transformed plasmids: the plasmids must have double- GST p53 homolog 1 strand breaks, and the host bacteria must have an active RecE recombination pathway (Kolodner et al., GST p53 homolog 2 1994). Most colonies arising after such transformations were recombinants (data not shown). BamHI We tested recombination between the core domains of several homologs: human, mouse, and chicken p53, and human p73 (Kaghad et al., 1997). Human and mouse p53 core domains are the most closely related (83% nucleotide identity), while p73 is the most distant EcoRV from the others (56 ± 58% identity). The plasmids were Amp cut to give fragments with homologous ends (Figure 1). To regenerate a circular plasmid from these fragments, two crossovers are required, one of which must be Amp within the p53 homologs. PstI Cut plasmids were co-transformed into RecE+ bacteria in all combinations. Transformation with a single plasmid gave no or very few background colonies (Figure 1, `bu€er' row and column). As plasmid plasmid cut with EcoRI+PstI expected, recombinations between identical genes gave cut with mouse human chicken human the most colonies; however, even distant homologs like BamHI+EcoRV buffer p53 p53 p53 p73 human p53 and p73 also recombined. Many di€erent buffer 00000 crossovers occurred, as seen below. Inactivating mouse p53 0 721 69 40 5 mismatch repair (mutS) did not improve recombination human p53 4 460 1660 180 98 frequencies but did change the distribution of cross- chicken p53 58129730 169 overs (data not shown). human p73 10 85 100 83 2090 A variety of DNA fragments will recombine when co-transformed, so long as homologous ends are Figure 1 Recombination of p53 homologs. Above is a schematic of glutathione S-transferase (GST) fusion plasmids, identical present (data not shown). Recombination occurs except for the genes of interest (here p53 homologs). One is cut at eciently even if each plasmid has only a single cut EcoRI+PstI sites, the other at BamHI+EcoRV sites. Only if (e.g., one at BamHI and the other at EcoRI); this is fragments recombine both within the BamHI ± EcoRI interval and useful when the plasmids do not have suitable second the EcoRV ± PstI interval do they regenerate a circular plasmid (follow the thick lines). Irrelevant fragments are shown by dotted restriction sites. Many hybrids used below were derived lines. Below are the number of ampicillin-resistant colonies from such recombinations. As another example, co- produced after co-transformation using 60 ng of each cut plasmid transformation of three PCR-generated fragments gave or bu€er control; recombinations between identical genes are in recombinants resulting from three crossovers. italics. (The mouse p53 plasmid was cut at the nearby AhdI site instead of PstI, as it does not have a unique PstI site.) Methods: Core domains of mouse p53 (residues 88 ± 306), chicken p53 Mapping p53 epitopes using mouse ± chicken hybrids (residues 80 ± 293), and human p73 (residues 105 ± 333) were PCR- ampli®ed from cDNA and the p73 EcoRI site mutated; human We recombined mouse and chicken core domains to core domain DNA (residues 90 ± 312) was from Bullock et al., generate mouse/chicken and chicken/mouse hybrids 1997. Fragments were ligated into the BamHI+EcoRI sites of the vector pGEX-4T-1 (Pharmacia). For recombination, plasmids (Figure 2). Clones with distinct restriction ®ngerprints were cut with restriction enzymes and puri®ed with QIAquick were sequenced to determine crossover points; all columns (QIAGEN) or phenol-chloroform/precipitation. DNA hybrids had a single crossover event. No point (30 ± 1000 ng of each) was co-transformed (Chung et al., 1989) mutations were seen in over 30 hybrids sequenced. In into strain JC8679 (E. coli Genetic Stock Center; Gillen et al., the following, the residue numbering of human p53 is 1981). When plasmids were cut with only one enzyme, they were also phosphatased to reduce background from re-ligation. used for all sequences regardless of species. (Conver- Colonies were screened by restriction enzyme ®ngerprinting sions can be made using Figure 3b.) Crossovers are (typically with MspI) of PCR products, and most gave band identi®ed at the DNA level, but are numbered by the patterns di€erent from both parents most N-terminal residue in the crossover interval. Hybrid proteins were expressed directly from recombinants as GST fusions. Antibody reactivity (StuÈ rzbecher et al., 1992; Hansen et al., 1998). We was assayed by ELISA at low temperature, since the evaluated this using our ELISA. Puri®ed monomeric core domain is of marginal stability (Bullock et al., human core domain (Bullock et al., 1997; gift of P 1997). Under these conditions, all proteins maintained Nikolova) inhibited PAb1620 binding to immobilized the `wildtype' conformation (no binding by PAb240); GST-mouse core domain (50% inhibitory concentra- while heating revealed the PAb240 epitope. As tion=15 mM), implying that PAb1620 can recognize expected, PAb1620 and PAb246 bound mouse but monomeric p53 core domain. not chicken p53 core domain. It has been suggested In the mouse/chicken series of hybrids, both PAb246 that the PAb1620 epitope requires p53 oligomerization and PAb1620 bound hybrid m5 but not m6, implicat-

Oncogene p53 `wildtype' conformation epitopes PL Wang et al 2320

Figure 2 Antibody reactivity of p53 hybrids. Proteins are shown as rectangles, black or white indicating mouse or chicken p53 sequence, respectively. Numbering is shown by a ruler below each set; each crossover location is notated. Hybrid names are on the left and reactivity with PAb246 or PAb1620 (ELISA signal as a percentage of mouse core domain ELISA signal, normalized for fusion protein levels using the ELISA signal for GST) is on the right, using the symbols in the key. Methods: p53 hybrids were generated by recombination as in Figure 1. Proteins were expressed by inoculating saturated cultures 1 : 100 into 10 ml 26TY+100 mg/ml ampicillin, shaking at 378C until OD600 *1, then shaking at 208C overnight (soluble protein yields were better when IPTG induction was omitted). Bacteria were resuspended in 1 ml ice-cold bu€er (50 mM Tris pH 7.2, 5 mM dithiothreitol (DTT), 200 mM phenylmethylsulfonyl ¯uoride), freeze ± thawed, sonicated on ice, and insoluble material removed by centrifugation. ELISA plates were coated with 5 mg/ml goat anti-GST (Pharmacia) in phosphate-bu€ered saline (PBS) and blocked with MPBS (PBS, 3% nonfat dry milk (Marvel), 2% BSA). Reagents were diluted in MPBS+2 mM DTT. The following incubations were done on ice for 2 h : (1) bacterial extracts, diluted 1 : 5; (2) MAbs, diluted 1 : 100 ± 200 (PAb240 and PAb1620 from Alan Fersht's lab, PAb246 from Calbiochem); (3) peroxidase-conjugated anti-mouse IgG, diluted 1 : 2000 (Sigma #A2554). To compare fusion protein levels, incubations 2 and 3 were replaced by an incubation with peroxidase-conjugated anti-GST, diluted 1 : 1000 (Sigma #A7340); bound GST levels were similar for most of the hybrids, and 490% of the GST was present as full-length fusion by Western blot. Each incubation was followed by three washes with cold PBS+2 mM DTT; at the end an additional three washes with cold PBS were done to remove DTT. The plate was developed with tetramethylbenzidine

ing residues 180 ± 219. In the chicken/mouse series, binding, presumably by conferring additional direct PAb246 bound c6 but not c5, implicating a similar contacts or allowing better packing of the minimal region, residues 195 ± 220. PAb1620 bound c9 strongly, mouse regions. c6 ± c8 weakly, and c5 not at all; this suggests that The regions de®ned by the minimal hybrids are while residues 195 ± 220 confer some PAb1620 binding residues 195 ± 223 for PAb246, and residues 143 ± 157 (c5 vs c6), full reactivity also requires residues 143 ± 157 and 194 ± 220 for PAb1620. The crossovers were (c8 vs c9). de®ned by di€erences at the DNA level; at the amino We made more complex hybrids to determine if the acid level, some of the included residues are identical implicated regions are both necessary and sucient. between mouse and chicken. Taking this into account, These were made by recombining genes cut at new sites the epitopes can be narrowed to 201 ± 212 for PAb246, (hybrid x8 was made by crossing chicken p53 with and 145 ± 157 and 201 ± 212 for PAb1620 (underlined in mouse p53 cut internally by NcoI) and recombining black, Figure 3b). hybrids with each other (hybrid x1 was made by crossing c6 and m5). Alanine-scan of minimal epitope regions PAb246 did not bind hybrid x7, which is mouse except at 180 ± 241, indicating that those residues are We next made alanine mutants of polymorphic necessary for binding. Similarly, PAb1620 bound residues in the minimal regions and tested antibody neither x7 nor x8, indicating that residues 140 ± 180 reactivity (Figure 3a). Although both PAb246 and and 180 ± 241 are necessary. PAb246 bound hybrid x1, PAb1620 required residues 201 ± 212, their ®ne speci®- which only has mouse residues from 195 ± 223, so these cities were quite di€erent. PAb246 binding was are sucient to allow recognition; likewise, PAb1620 decreased by single mutations of Tyr202 or Pro203, bound hybrid x9, so residues 143 ± 157 plus 194 ± 220 and abolished by double mutations of residue 203 plus are sucient. These minimal mouse regions confer either 202 or 204. PAb1620 binding was decreased by binding to a chicken sca€old, but not full reactivity. mutation of Arg156 or Leu206, and was abolished by Larger stretches of mouse sequence can improve mutation of Arg209 or Gln210. Mutations that a€ected

Oncogene p53 `wildtype' conformation epitopes PL Wang et al 2321

Figure 3 Alanine-scan and epitope comparisons. (a) Antibody reactivity of alanine mutants. Reactivity with PAb1620 and PAb246 is shown above and below the mutated residue in the mouse p53 sequence, using the symbols in the key. Double alanine mutations are marked by brackets. (b) Epitope locations in p53 sequences. The aligned sequences of human, mouse, and chicken p53 core domains are shown (identity to the human residue shown by a dash, gaps by a space). Items concerning PAb1620 or PAb246 are notated on the human or mouse sequence, respectively: regions identi®ed by hybrid mapping are underlined in black, while regions identi®ed by others are underlined with gray brackets; essential and non-essential key residues from alanine-scan are highlighted with black and gray boxes, respectively. Human p53 numbering (used throughout) is marked along the top; for reference, species-speci®c numberings are given at the sides. Above the sequences is the secondary structure of human p53 core domain (Cho et al., 1994). Methods: Alanine mutations were introduced by PCR and veri®ed by DNA sequencing. Protein expression and ELISA were as in Figure 2

binding of PAb1620 did not a€ect binding of PAb246 Evans and Jenkins, 1985). The PAb246 epitope was and vice versa. assigned to residues 92 ± 113 (gray underline, mouse The key residues identi®ed by alanine-scan (boxed in sequence, Figure 3b) because PAb246 immunoprecipi- Figure 3b) are consistent with the species-speci®city of tated N-terminal deletions up to residue 91 but not to the antibodies. PAb246 binds mouse but not human 113. Deletion to 113 removes part of the core domain p53, and its key residues 202 ± 204 are di€erent in and abolishes DNA-binding activity (Halazonetis and mouse and human (Tyr-Pro-Glu vs Arg-Val-Glu). Kandil, 1993). This region of mouse sequence does not PAb1620 binds both mouse and human p53, and its confer PAb246 binding, and replacing with divergent key residues 156, 206, 209, and 210 are identical in chicken sequence does not decrease binding (m8 vs mouse and human except for a conservative di€erence chicken; c9 vs mouse, Figure 2). at 210 (Gln vs Asn). Neither antibody recognizes Phage display has been used to select peptides chicken p53, which di€ers from both human and binding to PAb1620 (Ravera et al., 1998); peptide mouse at all key residue positions. Hybrid mapping competition with p53 was not demonstrated. Using cannot exclude the possibility that residues conserved sequence comparison of the peptides with core between mouse and chicken may also be important; domain, the PAb1620 epitope was assigned to residues but any such residues would be contiguous with the 106 ± 113 and 146 ± 156 (gray underline, human epitope surfaces we have de®ned. sequence, Figure 3b). The second region is similar to one found by hybrid mapping; at its periphery is key residue 156. However, the essential residues 209 ± 210 Comparison to previous epitope mapping studies were not identi®ed, and no PAb1620 binding is seen Deletion mutants of mouse p53 have been used to map even if both regions are mouse sequence (hybrids m6 binding of PAb246 and several other MAbs (Wade- and x7, Figure 2).

Oncogene p53 `wildtype' conformation epitopes PL Wang et al 2322 Antibody epitopes and the three-dimensional antigen used, SV40 virus-immortalized cells (Ball et al., 1984; Yewdell et al., 1986); SV40 large T antigen core domain structure stabilizes p53 by binding to the core domain and masks Human p53 core domain is a sandwich of two twisted its DNA-binding surface. This stabilization may have b-sheets, from which extend additional elements that facilitated the generation of `wildtype' conformation contact DNA and coordinate a zinc ion (Figure 4). The antibodies, as immunization with puri®ed p53 gave PAb1620 and PAb246 epitopes are both in the vicinity only core domain antibodies against the denatured/ of strand S6; the two antibodies can compete for `mutant' conformation (Legros et al., 1994b; Vojtesek binding (Milner et al., 1987). Both epitopes are distant et al., 1995). from the DNA-binding interface. The locations of the The PAb1620 key residues 156, 206, 209, and 210 two epitopes may have in part been dictated by the de®ne an elongate surface region; the two most

Figure 4 Three-dimensional locations of p53 epitopes. The main-chain ribbon and side-chains of key residues are shown. The PAb1620, PAb246, and PAb240 epitopes are blue, red, and violet, respectively; core domain N- and C-terminal residues (96 ± 97 and 288 ± 289) are green and cyan, respectively; DNA is blue-gray; and zinc is a dark gray ball. Methods: PDB entry 1TSR was drawn using MOLSCRIPT (Kraulis, 1991) and RASTER3D (Merritt and Murphy, 1994)

Oncogene p53 `wildtype' conformation epitopes PL Wang et al 2323 important residues, 209 and 210, are highly exposed on e€ects on interactions between subunits rather than on a tight turn between strands S6 and S7. The other two the interaction with DNA. residues lie in a groove along the top of the core Another interesting case is the lack of recognition by domain. Since PAb246 is mouse-speci®c, the human PAb1620 of full-length human p53-DNA complexes structure can only give us the general setting of its (Hupp et al., 1992; Hall and Milner, 1995; Cohen et epitope: residues 202 and 203 are part of a turn al., 1999), even though PAb1620 recognizes free full- between strands S5 and S6, and residue 204 is the length human p53 (Cook and Milner, 1990) and beginning of strand S6. Both `wildtype' epitopes disrupts human core-DNA complexes (Hansen et al., involve loops attached to one edge of the larger b- 1998). The crystal structures of the core domain sheet (strands S6-S7-S4-S9-S10), and may be quite whether free, bound to DNA, or bound to another sensitive to b-sheet perturbations. The epitope of the protein (Cho et al., 1994; Gorina and Pavletich, 1996) `mutant' conformation antibody PAb240 lies close by do not show signi®cant conformational di€erences; in strand S7 (residues 213-217; Stephen and Lane, therefore the simplest explanation for the loss of the 1992). This region is hidden by strand S6 and PAb1620 epitope is masking by other parts of p53. The neighboring loops, so that exposure of the `mutant' PAb1620 epitope is near the core domain N-terminus epitope and `wildtype' epitopes is mutually exclusive. and distant from its C-terminus (Figure 4), suggesting that the N-terminal region can mask the PAb1620 epitope in the full-length complex. Consistent with this, HPV E6 and the PAb246 epitope human residues 80 ± 93 (just N-terminal to the core Degradation of p53 is induced by the E6 protein of domain) are a `negative regulatory region' that may human papillomaviruses associated with cervical cancer change interactions upon activation of DNA binding (`high-risk' HPV-16 and HPV-18). E6 alone binds (MuÈ ller-Tiemann et al., 1998). Changes in N-terminal weakly to p53 but strongly when the cellular protein region con®guration could modulate transactivation E6-AP is present (Li and Cono, 1996a); the E6/E6- and interactions with other proteins. AP complex ubiquitinates p53 and targets it for The PAb1620 epitope is present on mouse p53-DNA degradation by proteosomes (Sche€ner et al., 1990; complexes (Hall and Milner, 1995), in contrast to Huibregtse et al., 1991). Degradation requires the human complexes; also, the mouse N-terminal region `wildtype' conformation; `mutant' conformation p53 does not have negative regulatory function (Hansen et is not degraded (Medcalf and Milner, 1993). PAb246 al., 1998). This suggests functional di€erences between (but not PAb1620) inhibits E6-mediated degradation of human and mouse N-terminal regions, the most mouse p53 (Medcalf and Milner, 1993; Li and Cono, evolutionarily divergent part of p53 (reviewed by 1996b). However PAb246 does not block E6 binding to Soussi and May, 1996), and may allow these p53 (Molinari and Milner, 1995), suggesting that the phenomena to be dissected using p53 hybrids involving epitope de®ned here may be a region of contact with the N-terminal region. the E6/E6-AP complex.

The `wildtype' conformation and p53 DNA binding PAb1620 and PAb246 have been used previously to Acknowledgements We thank Alex Bullock, Penka Nikolova, and Alan Fersht probe p53-DNA complexes; the results are complex for puri®ed human p53 core domain protein, DNA, and and depend on many factors. In some cases, they cause MAbs;andCarolMidgley,SylvainArnould,andPenka disruption of p53-DNA complexes (Halazonetis et al., Nikolova for critical reading of the manuscript. PL Wang 1993; Hall and Milner, 1995; Wolkowicz et al., 1995; was supported by a Hitchings-Elion fellowship from the McLure and Lee, 1996); since the antibodies bind far Burroughs-Wellcome Fund and by the Medical Research from the DNA interface, disruption must be due to Council.

References

Balagurumoorthy P, Sakamoto H, Lewis MS, Zambrano N, Cohen PA, Hupp TR, Lane DP and Daniels DA. (1999). Clore GM, Gronenborn AM, Appella E and Harrington FEBS Lett., 463, 179 ± 184. RE. (1995). Proc. Natl. Acad. Sci. USA, 92, 8591 ± 8595. Cook A and Milner J. (1990). Br. J. Cancer, 61, 548 ± 552. Ball RK, Siegl B, Quellhorst S, Brandner G and Braun DG. Foster BA, Co€ey HA, Morin MJ and Rastinejad F. (1999). (1984). EMBO J., 3, 1485 ± 1491. Science, 286, 2507 ± 2510. Bullock AN, Henckel J, DeDecker BS, Johnson CM, Gannon JV, Greaves R, Iggo R and Lane DP. (1990). EMBO Nikolova PV, Proctor MR, Lane DP and Fersht AR. J., 9, 1595 ± 1602. (1997). Proc. Natl. Acad. Sci. USA, 94, 14338 ± 14342. Gillen JR, Willis DK and Clark AJ. (1981). J. Bacteriol., 145, Cho Y, Gorina S, Je€rey PD and Pavletich NP. (1994). 521 ± 532. Science, 265, 346 ± 355. Gorina S and Pavletich NP. (1996). Science, 274, 1001 ± 1005. Chung CT, Niemela SL and Miller RH. (1989). Proc. Natl. Acad. Sci. USA, 86, 2172 ± 2175.

Oncogene p53 `wildtype' conformation epitopes PL Wang et al 2324 Hainaut P, Hernandez T, Robinson A, Rodriguez-Tome P, Medcalf EA and Milner J. (1993). Oncogene, 8, 2847 ± 2851. Flores T, Hollstein M, Harris CC and Montesano R. Merritt EA and Murphy MEP. (1994). Acta Crystallogra- (1998). Nucleic Acids Res., 26, 205 ± 213. phica Section D, 50, 869 ± 873. Halazonetis TD, Davis LJ and Kandil AN. (1993). EMBO J., Milner J, Cook A and Sheldon M. (1987). Oncogene, 1, 453 ± 12, 1021 ± 1028. 455. Halazonetis TD and Kandil AN. (1993). EMBO J., 12, Molinari M and Milner J. (1995). Oncogene, 10, 1849 ± 1854. 5057 ± 5064. MuÈ ller-Tiemann BF, Halazonetis TD and Elting JJ. (1998). Hall AR and Milner J. (1995). Oncogene, 10, 561 ± 567. Proc. Natl. Acad. Sci. USA, 95, 6079 ± 6084. Hansen S, Lane DP and Midgley CA. (1998). J. Mol. Biol., Ravera MW, Carcamo J, Brissette R, Alam-Moghe A, 275, 575 ± 588. DedovaO,ChengW,HsiaoKC,KlebanovD,ShenH, Huibregtse JM, Sche€ner M and Howley PM. (1991). EMBO Tang P, Blume A and Mandecki W. (1998). Oncogene, 16, J., 10, 4129 ± 4135. 1993 ± 1999. Hupp TR, Meek DW, Midgley CA and Lane DP. (1992). Sche€ner M, Werness BA, Huibregtse JM, Levine AJ and Cell, 71, 875 ± 886. Howley PM. (1990). Cell, 63, 1129 ± 1136. Kaghad M, Bonnet H, Yang A, Creancier L, Biscan JC, Soussi T and May P. (1996). J. Mol. Biol., 260, 623 ± 637. Valent A, Minty A, Chalon P, Lelias JM, Dumont X, Stephen CW and Lane DP. (1992). J. Mol. Biol., 225, 577 ± Ferrara P, McKeon F and Caput D. (1997). Cell, 90, 809 ± 583. 819. StuÈ rzbecher H-W, Brain R, Addison C, Rudge K, Remm M, Kolodner R, Hall SD and Luisi-DeLuca C. (1994). Mol. Grimaldi M, Keenan E and Jenkins JR. (1992). Oncogene, Microbiol., 11, 23 ± 30. 7, 1513 ± 1523. Kraulis PJ. (1991). J. Applied Crystallog., 24, 946 ± 950. Vojtesek B, Dolezalova H, Lauerova L, Svitakova M, Havlis LaneDP,StephenCW,MidgleyCA,SparksA,HuppTR, P, Kovarik J, Midgley CA and Lane DP. (1995). DanielsDA,GreavesR,ReidA,VojtesekBandPicksley Oncogene, 10, 389 ± 393. SM. (1996). Oncogene, 12, 2461 ± 2466. Wade-Evans A and Jenkins JR. (1985). EMBO J., 4, 699 ± Legros Y, Lafon C and Soussi T. (1994a). Oncogene, 9, 706. 2071 ± 2076. Wang PL. (2001). Disease Markers, 16. Legros Y, Meyer A, Ory K and Soussi T. (1994b). Oncogene, Wang Y, Schwedes JF, Parks D, Mann K and Tegtmeyer P. 9, 3689 ± 3694. (1995). Mol. Cell. Biol., 15, 2157 ± 2165. Levine AJ. (1997). Cell, 88, 323 ± 331. Wolkowicz R, Elkind NB, Ronen D and Rotter V. (1995). Li X and Cono P. (1996a). J. Virol., 70, 4509 ± 4516. Oncogene, 10, 1167 ± 1174. Li X and Cono P. (1996b). J. Biol. Chem., 271, 4447 ± 4451. Yewdell JW, Gannon JV and Lane DP. (1986). J. Virol., 59, McLure KG and Lee PW. (1996). Oncogene, 13, 1297 ± 1303. 444 ± 452.

Oncogene