Human tumor-derived p53 proteins exhibit binding site selectivity andtemperature sensitivity for transactivation in a yeast-based assay

Charles J Di Como and Carol Prives

Department of Biological Sciences, Columbia University, New York, New York 10027, USA

p53 is a sequence-speci®c transcriptional activator with anumber of known target genes which contain p53-responsive elements. Mutations in p53 have beenidenti®ed within its sequence-speci®c DNA bindingdomain in more than half of all human tumors, althougha subset of tumor-derived p53 mutants have retained theability to bind DNA and activate transcription undercertain conditions. In order to broaden our understandingof this transactivating ability, we examined the e�cacyby which p53 mutants bind to and activate reporters inan Saccharomyces cerevisiae-based assay. Analysis of 19human tumor-derived p53 mutants, spanning the DNAbinding domain of p53 and including the `hot-spot' class,revealed a broad array of transcriptional transactivationabilities at 248C, 308C and 378C, despite the fact thateach mutant had originally been identi®ed as beinginactive for transactivation in yeast against a single p53-responsive RGC site-containing reporter. One class ofmutants (P177L, R267W, C277Y and R283H) retainedwild-type or near wild-type activity that is binding site-selective, even at physiological temperature (378C).Another class of mutants (V143A, M160I/A161T,H193R, Y220C and I254F), all positioned for maintain-ing the b-sca�old of p53, also retained selective activity,but preferentially at sub-physiological temperatures (248and 308C). Strikingly, however, in contrast to the othertumor derived mutants, all of the previously identi®ed`hot-spot' mutants were completely inactive with all sitestested. Moreover, a double mutant, L22E/W23S, locatedwithin the activation region and previously shown to betranscriptionally inactive in ®broblasts, retained wild-type or near wild-type binding site-selective activity inyeast. Finally, we found that transcriptional activity invivo does not necessarily correlate with DNA binding invitro.

Keywords: p53; tumor-derived mutants; transactivation;Saccharomyces cerevisiae; FASAY

Introduction

The p53 tumor suppressor gene encodes a proteinwhich has been found to be mutated in over 50% of allhuman tumors (Friend, 1994; Hollstein et al., 1994). Inresponse to genotoxic stress resulting in genomicinstability, signals are transmitted to activate p53,e�ectively leading to cell cycle arrest and/or apoptosis

(reviewed in Gottlieb and Oren, 1996; Ko and Prives,1996; Levine, 1997; and references therein).P53 is a nuclear phosphoprotein which functions as

a transcription factor by binding DNA predominantlyas a tetramer (a dimer of a dimer) to two pairs of aconsensus sequence 5'-PuPuPuC(A/T)-3' arranged asinverted repeats (reviewed in Vogelstein and Kinzler,1992; Prives, 1994). Transactivation by p53 may bemediated through direct contacts with components ofthe general transcription machinery, as interactionsbetween the activation domain of p53 and p300/CBP(Avantaggiati et al., 1997; Gu and Roeder, 1997; Gu etal., 1997; Lill et al., 1997; Scolnick et al., 1997), TATA-binding protein (TBP) (Seto et al., 1992; Liu et al.,1993; Martin et al., 1993; Ragimov et al., 1993; Truantet al., 1993; Chang et al., 1995; Farmer et al., 1996),TFIIH components (p62, ERCC2 and ERCC3) (Xiaoet al., 1994; Wang et al., 1995; Leveillard et al., 1996),and a number of TAFs (Drosophila TAFII40 andTAFII60, human TAFII31) (Lu and Levine, 1995;Thut et al., 1995; Farmer et al., 1996) have beendemonstrated. Many of the target genes that containp53-responsive cis-acting elements have been impli-cated in the regulation of the cell cycle, DNA synthesisand apoptosis, such as; p21 (El-Deiry et al., 1993;Harper et al., 1993; Xiong et al., 1993; Noda et al.,1994), mdm2 (Wu et al., 1993), GADD45 (Kastan et al.,1992), Cyclin G (Okamoto and Beach, 1994; Zauber-man et al., 1995), Bax1 (Miyashita and Reed, 1995)and IGF-BP3 (Buckbinder et al., 1995).The p53 protein is modular and can be divided into

distinct domains: (1) a transcriptional activation regionat the amino terminus, comprising two subdomains(residues*1 ± 40 and*40 ± 70) (Fields and Jang, 1990;Raycroft et al., 1990; Unger et al., 1992; Chang et al.,1995; Candau et al., 1997); (2) a `PXXP' SH3-likedomain (residues *61 ± 94) (Walker and Levine, 1996);(3) a sequence-speci®c DNA binding domain (residues*102 ± 292) (Bargonetti et al., 1993; Halazonetis andKandil, 1993; Pavletich et al., 1993; Wang et al., 1993);and (4) a nonspeci®c- or damaged DNA-bindingcarboxyl terminal regulatory domain (residues *320 ±393): containing a tetramerization domain (residues*320 ± 360); a basic domain (residues *363 ± 393); andtwo nuclear localization signals (Shaulsky et al., 1991;Wang et al., 1993; Brain and Jenkins, 1994; Bayle et al.,1995; Lee et al., 1995; Reed et al., 1995). A survey of thegenomic mutations in p53 often reveals a deletion of oneallele coincident with a missense mutation of the otherallele. The vast majority of missense mutationsidenti®ed in over 6800 tumor-derived p53 alleles occurin the DNA binding domain (Hollstein et al., 1994;Levine et al., 1995; Hainaut et al., 1998). The solving ofthe crystal structure of the p53 DNA binding domainbound to its cognate site (Cho et al., 1994) has helped to

Correspondence: C PrivesReceived 22 January 1998; revised 6 April 1998; accepted 7 April1998

Oncogene (1998) 16, 2527 ± 2539 1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00

http://www.stockton-press.co.uk/onc

explain how mutations in this domain interfere withDNA binding and has reinforced the importance of theDNA binding domain in the normal function of p53.Among the mutations in the p53 DNA binding

domain are six amino acids residues (R175, G245,R248, R249, R273 and R282), often referred to as `hot-spots', which are mutated with an unusually highfrequency and together comprise about 40% of all p53missense mutants (Nigro et al., 1989; Hollstein et al.,1991; 1994). These residues have been divided into twoclasses, `conformational' (e.g.: R175, G245, R249 andR282) and `contact' (e.g.: R248 and R273), dependingon whether they contact DNA directly or aid inmaintaining the structural integrity of the DNAbinding domain (discussed in Cho et al., 1994). Manystudies have addressed the DNA binding propertiesand the transcription transactivation function oftumor-derived p53 mutants with the assumption thatmutations in the DNA binding domain would modifyor destroy the ability of p53 to bind to DNA andactivate transcription (reviewed in Gottlieb and Oren,1996; Ko and Prives, 1996; Levine, 1997). The resultsof these experiments have upheld the assumption thattumor-derived p53 mutants exhibit aberrant DNAbinding and transcriptional transactivation. Thesemutant proteins are also frequently expressed at highlevels in most tumors due to increased stability. Thatincreased levels of wild-type p53 can induce arrest orapoptosis in certain cell types has led to the strategy oftrying to identify therapeutic agents that can restoreDNA binding and transcriptional transactivationfunction to mutant p53 in vivo (Abarzua et al., 1995;1996; Selivanova et al., 1997).A number of studies have characterized tumor-

derived p53 mutant function with respect to transacti-vation and sequence-speci®c DNA binding. Examiningendogenous intact mutant p53 proteins in tumor celllines (Park et al., 1994; Niewolik et al., 1995),transiently transfected tumor-derived p53 mutants inp53 null cell lines (Chen et al., 1993a; Chumakov et al.,1993; Friedlander et al., 1996b) and mutant p53/GAL4DNA binding domain chimeras in CHO cells (Raycroftet al., 1991; Unger et al., 1993; Miller et al., 1993) hasprovided information that some mutants underrestricted conditions display DNA binding or transac-tivation. For example, Friedlander et al. (1996b) foundthat at the physiological temperature (378C), p53mutants are defective for sequence-speci®c DNAbinding, whereas, at sub-physiological temperatures(25 ± 338C), mutants are capable of binding to DNA.Wild-type p53 is also temperature sensitive (Picksley etal., 1992; Hainaut et al., 1995; Friedlander et al.,1996b), but markedly less so than the mutants(Friedlander et al., 1996b).Although naturally occurring missense mutations

have helped to elucidate and identify the central DNAbinding domain, experimentally derived mutations inother regions of p53 have been informative as well.Particularly relevant to the work in our study is adouble mutation L22E/W23S within the activationdomain of p53 (Lin et al., 1994). Mutation of p53 atthese two residues has been shown to dramaticallyreduce its ability to activate transcription in mamma-lian cells (Lin et al., 1994; Candau et al., 1997), as wellas to bind to TAFs (Lu and Levine, 1995; Thut et al.,1995) and p300/CBP (Gu et al., 1997).

Prior results have demonstrated that mammalianp53 can function as a transcription factor in yeast(Scharer and Iggo, 1992). To further study mutant p53,we have used an S. cerevisiae-based assay ®rstdescribed by Ishioka et al. (1993) into which eitherwild-type or mutant forms of p53 are introduced into aHis7 yeast strain carrying a reporter with a p53binding site (RGC) driving the HIS3 gene. Yeastgrown in the absence of histidine will only grow if thereporter is activated. While this assay was originallydevised as a diagnostic tool, it has other advantages aswell. It allows for rapid and quantitative assessment ofthe transcriptional activity of di�erent forms of p53 inan essentially isogenic background, avoiding possiblecomplications resulting from transient transfection ofmammalian cells such as overexpression of p53 at non-physiological levels. Furthermore, the process of DNAtransfection itself can activate or stabilize endogenousp53 (Siegel et al., 1995) suggesting that transfected p53might be similarly altered. Since Ishioka et al. (1993)found that all tumor-derived mutants tested wereinactive with the RGC-containing HIS3 reporter, itwas of interest to determine whether a number ofalternative p53 binding sites within this reporter wouldbe able to serve as response elements for wild-type andmutant forms of p53. Furthermore, the fact that yeastgrows well at temperatures ranging from 24 ± 378C hasallowed us to further study temperature sensitivetransactivation by p53. Our results indicate that wild-type and some mutant forms of p53 are transcription-ally active with a number of di�erent p53-responsivereporters. We discuss the possible consequences a cellmay face by maintaining a p53 mutant which retainsselective transcriptional activity.

Results

Human wild-type p53 in yeast activates p53-responsivereporters

Strains were constructed which contained the HIS3gene (see Materials and methods) under the control ofone of the following derived p53-responsive human ormurine target gene cis-acting elements: p21 (El-Deiry etal., 1993); mdm2 (Wu et al., 1993); GADD45 (Kastanet al., 1992); Cyclin G (Okamoto and Beach, 1994);Bax (Miyashita and Reed, 1995); IGF-BP3 Box A andBox B (Buckbinder et al., 1995); ribosomal gene cluster(termed RGC) (Kern et al., 1991b); and an arti®cialhigh a�nity binding-p53 consensus element (termedSCS) (Halazonetis et al., 1993) (see Figure 1 andMaterials and methods). Each of the above reporterstrains were transformed with plasmids expressingeither human wild-type or mutant p53 under controlof the constitutive alcohol dehydrogenase (ADH1 )minimal promoter (Ishioka et al., 1993). We chosethe constitutive ADH1 promoter to express wild-typeand mutant p53 proteins which does not expressextremely high levels of p53, since previous work hasdemonstrated that high level expression of wild typep53 and certain tumor-derived mutants imparts a slowgrowth phenotype in both S. cerevisiae (Nigro et al.,1992) and Schizosaccharomyces pombe (Wagner et al.,1991, 1993; Bischo� et al., 1992). The growth assayutilized for our phenotypic analysis relies on the fact

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2528

that the HIS3 gene is under the control of an inactiveGAL1 promoter (lacking an upstream activatingsequence or UAS). This promoter is activated andHIS3 expressed only when bound by a transcriptionalactivator, such as p53, at sites placed upstream of theminimal GAL1 promoter. We score transactivation asgrowth or lack thereof on histidine-de®cient media.Con®rming the results of Scharer and Iggo (1992),

we observed p53-dependent HIS3 transcription of theRGC-containing reporter as assayed by growth onhistidine-de®cient media (Figure 2b). Additionally,wild-type p53 transactivated to varying degrees allreporters, with the exception of those containing IGF-BP3 Box A and Box B (Figure 2b and not shown). Asa control, isogenic strains expressing either: (1) wild-type p53 and containing a HIS3 reporter with no p53-responsive cis-acting element; or (2) vector control andany one of the p53-responsive cis-acting elementreporters did not grow on histidine-de®cient media(Figure 2b, Tables 1, 2 and 3 (scored as `7'), and notshown), demonstrating the speci®city of the assay.Interestingly, the extent to which wild-type p53 wasable to transactivate each reporter was dependent onthe cis-acting element present upstream of the HIS3coding sequence. Whereas the RGC- and Bax-contain-ing reporter strains grew slowly on histidine-de®cientmedia (Figure 2b and Tables 1, 2 and 3, scored as `+'),the p21-, SCS-, mdm2-, GADD45- and Cyclin G-containing reporter strains grew with a growth ratescored as `+++' (hereinafter referred to as `wild-type') (Figure 2b and Tables 1, 2 and 3). While IGF-BP3 Box A and Box B have been shown to be p53-responsive cis-acting elements in mammalian cells

(Buckbinder et al., 1995; Friedlander et al., 1996b;Ludwig et al., 1996), we detected no such activation inyeast (Tables 1, 2 and 3 (scored as `7'), and notshown). Moreover, the relative activation of thesereporters by wild-type p53 was equal at all tempera-tures (248C, 308C and 378C), with two exceptions: (1)wild-type p53 activated the GADD45-containingreporter to a lesser extent than the p21-, SCS-,mdm2- and Cyclin G-containing reporters at 248C,but yet greater than the Bax- and RGC-containingreporters at the same temperature; and (2) wild-typep53 did not activate the Bax-containing reporter at248C (not shown).To determine if the extent to which wild-type p53

transactivated the p53-responsive reporters was due tothe expression level of p53 in the cell, each of thestrains shown in Figure 2a were grown to log phaseand total cell extracts were prepared. As detected byWestern blot analysis with a mixture of anti-p53antibodies, the protein levels of wild-type p53 in eachstrain were readily detected and were quite similar(Figure 3). Therefore, the growth rate di�erences arenot due to variations in the levels of p53 protein, butare due to the ability of wild-type p53 to transactivatethe cis-acting element present in each reporter.

The DNA binding properties of yeast-expressed humanwild-type p53 are similar to immunopuri®ed human wild-type p53

Using the electrophoretic mobility shift assay (EMSA),we compared DNA binding by either immunopuri®ed-or yeast-expressed human wild-type p53 proteins at

Figure 1 The p53-responsive cis-acting elements. Sequences are derived from the naturally occurring cis-acting elements found inp53-responsive target genes: p21 (El-Deiry et al., 1993); mdm2 (Wu et al., 1993); GADD45 (Kastan et al., 1992); Cyclin G (Okamotoand Beach, 1994); Bax1 (Miyashita and Reed, 1995); and IGF-BP3 (Buckbinder et al., 1995). The `H' refers to human origin, whilethe `M' refers to murine origin. The SCS high a�nity binding site was identi®ed by Halazonetis et al. (1993) and represents theperfect copy of the consensus deduced by El-Deiry et al. (1992)

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2529

248C to radiolabelled oligoduplex DNA containing thep53-responsive cis-acting element from p21 (CUO3/4,see Materials and methods). Human p53 expressed in

whole yeast cell extracts bound to the p21 probe (Figure4, lane 6). The monoclonal antibody PAb421 (Harlow etal., 1981), which interacts with an epitope (amino acids373 ± 381) within the C-terminus of p53 (Wade-Evansand Jenkins, 1985) and enhances the DNA bindingfunction of wild-type and certain mutant forms of p53(Hupp et al., 1992, 1993; Halazonetis and Kandil, 1993;Jayaraman and Prives, 1995) also stimulated binding byyeast-expressed p53 (see supershifted complex, Figure 4,lane 7). Additionally, the monoclonal antibodyPAb1801 (Banks et al., 1986), which interacts with anepitope (amino acids 46 ± 55) (Legros et al., 1994) withinthe N-terminus of p53, enhanced DNA binding byyeast-expressed p53 proteins, although to a lesser extentthan PAb421 (Figure 4, lane 9). Moreover, a peptidecorresponding to the PAb421 epitope of p53, shownpreviously to activate DNA binding by p53 (Hupp etal., 1995; Shaw et al., 1996), stimulated DNA bindingby yeast-expressed wild-type p53 (Figure 4, lane 8). Anisogenic whole cell extract containing a control vector(no human p53 expression) exhibited no detectableDNA binding to the p21 probe, demonstrating thespeci®city of the assay for human p53 expressed in yeast(Figure 4, lane 5). As expected, human wild-type p53,immunopuri®ed from baculovirus-infected insect cells,was also stimulated by PAb421 (Figure 4, lane 2), 421peptide (Figure 4, lane 3), and PAb1801 (Figure 4, lane4). A longer exposure of this autoradiogram revealedthe DNA binding complex in lane 1 of immunopuri®edp53 in the absence of peptide or antibodies (not shown).Note that approximately 10 ± 30 times more yeast-expressed p53 protein than immunopuri®ed p53protein was required in order to detect the p53/DNAcomplex. Whether this is because yeast-expressed p53binding occurs in the context of a crude whole cellextract is not yet established. We conclude from theabove results that the DNA binding properties, asmeasured by EMSA, of yeast-expressed human wild-type p53 to the p21 probe are similar to immunopuri®edhuman wild-type p53.

p53 `hot-spot' mutants are transcriptionally inactive

We examined the transcription-activating potential ofa panel of human tumor-derived p53 mutants. All of

a

b

Figure 2 Expression of human wild-type p53 in yeast activatesp53-responsive reporters. (a) Strains expressing wild-type p53(pADH1 : p53 on a LEU2/CEN plasmid) and containing one ofthe following p53-responsive reporters (SCS :HIS3, RGC :HIS3,p21 :HIS3, mdm2 :HIS3, GADD45 :HIS3, Cyclin G :HIS3 andBax :HIS3 on a TRP1/CEN plasmid) or control reporter(DUASpGAL1 :HIS3 on a TRP1/CEN plasmid) were streakedout for single colonies onto SC minus leucine minus tryptophanplates and grown for 2 days at 308C. (b) Colonies in (a) werestreaked out for singles onto SC minus leucine minus tryptophanminus histidine plates and grown for 2 days at 308C

Table 1 Activation of reporters by `hot-spot' mutants

Cis-acting element Cont WT R175H . G245D . R248W* R249S . R273H* R282A .

noneRGC

SCS

p21

mdm2

Cyclin G

GADD45

Bax

IGF (A)IGF (B)

308308308378308378308378308378308378308378308308

7777777777777777

7+

++++++++++++++++++++++++++++++++77

7777777777777777

7777777777777777

7777777777777777

7777777777777777

7777777777777777

7777777777777777

`7'=no growth; `+'=growth; `++'=moderate growth; `+++'=wild-type growth; `*'=contact mutant;`.'=structural mutant

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2530

these mutants, with the exception of V143A, R175H,R248W, R249S and R282A were isolated by FASAY(functional analysis of separated alleles in yeast),which allows the detection of p53 mutations fromtumors or germline p53 mutations from patients'lymphocytes (Ishioka et al., 1993). Those p53mutants isolated by FASAY by Ishioka et al. (1993)were identi®ed as such by being defective fortransactivation of the identical RGC-containingreporter utilized in this survey (Tables 1, 2 and 3,scored as `7'). We constructed the remaining ®vealleles from human tumor-derived p53 cDNA (seeMaterials and methods).We determined whether the p53 mutants tested

above were capable of activating transcription fromreporters containing p53-responsive cis-acting elementsother than RGC. This analysis was carried out atvarious temperatures: 248C (sub-physiological); 308C(optimal growth condition for S. cerevisiae); and 378C(physiological temperature of mammalian cells); and bydi�erent growth-measuring techniques (replica platingand streak-out from histidine-containing media tohistidine-de®cient media). The results of this survey

Table 2 Activation of reporters at 378C

Cis-acting element Cont WT

L22E

W23S P177L R267W C277Y* R283H*

noneRGC

SCS

p21

mdm2

Cyclin G

GADD45

Bax

IGF (A)IGF (B)

308308308378308378308378308378308378308378308308

7777777777777777

7+

++++++++++++++++++++++++++++++++77

77

++++++++++++77

++++++++++7777

77

+++++++++++7777777777

77

++++++++++++7777777777

7777

++++++7777777777

77

++++++++++++++

+++++++7777

`7'=no growth; `+'=growth; `++'=moderate growth; `+++'=wild-type growth; `*'=contact mutant

Table 3 Activation at sub-physiological temperatures

Cis-acting element Cont WT V143A

M1601

A161T b H193R Y220C b 1254F b

noneRGC

SCS

p21

mdm2

Cyclin G

GADD45

Bax

IGF (A)IGF (B)

248248248308248308248308248308248308248308248248

7777777777777777

7+

+++++++++++++++++++++++++++++7+77

77

+++++++++++7777777777

77

+++++++++++++7

+++++++77777

77+7+77777777777

77

+++++++++++++7+7+77777

77

+++++++++++++7+7+77777

`7'=no growth; `+'=growth; `++'=moderate growth; `+++'=wild-type growth; `b '=b-sca�oldmaintenance

Figure 3 The expression levels of human wild-type p53 in yeast.Human wild-type p53-expressing yeast strains denoted in Figure2a were grown to log phase, extracts prepared, and subjected toWestern blot analysis, where 50 mg of total cell extract wasloaded. The Western blot was probed with a mixture of anti-p53antibodies at a 1/3000 dilution (upper panel) and an anti-b-tubulin antibody at a 1/3000 dilution (lower panel). As a controlfor anti-p53 antibody speci®city, a strain containing a controlvector (pADH1 on a LEU2/CEN plasmid) and the p21 :HIS3reporter was used (lane 9)

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2531

are described in detail below and summarized in Tables1, 2 and 3.With respect to the `hot-spot' alleles, we found that

the two contact mutants (R248W and R273H) andfour conformational mutants (R175H, G245D, R249S,and R282A) were incapable of transactivating each ofthe reporters at all temperatures, including 248C (Table1). It is worth reiterating that these six residues arecritical for p53/DNA interactions (Cho et al., 1994)and are among the most frequently mutated residuesfound in human tumors (Nigro et al., 1989; Hollsteinet al., 1991; 1994). It is noteworthy that these `hot-spot'p53 mutants were the only ones that were completelydefective in transactivation.

A class of p53 mutants which retain wild-type activityand binding-site selectivity at 378C

A second class identi®ed in our survey are thosemutants which retained transcriptional activity at alltemperatures, including 378C. These alleles are: P177L(located on the L2 loop, a region where a majority ofthe p53 mutations occur); R267W (located in the S10strand of the b-sandwich); C277Y (which contactsDNA on the pyrimidine-rich strand); and R283H(which contacts the phosphate backbone on thepurine-rich strand) (Cho et al., 1994). All of thesemutants transactivated the SCS- and p21-containingreporter plasmids and no other reporter constructs,with the exception of C277Y, which did nottransactivate the SCS-containing reporter, andR283H, which transactivated all reporters, but tovarying degrees (Table 2, scored as `+' to `+++'),excluding Bax and IGF-BP3 Box A and Box B, which

were either weak or inactive p53 cis-acting elements,respectively, with wild-type p53 (Tables 1, 2 and 3). Inaddition, we detected weak binding by the p53P177Land p53R267W proteins to the p21 probe by EMSA.In contrast, two independently isolated mutants(H179L and H179R), located on the H1 helix ofLoop 2 and which aid in coordinating a zinc atom,were completely inactive (not shown).The C277Y mutant, ®rst identi®ed in a Ewing

tumor, as well as in breast and colon cancer (Kovaret al., 1993; Hainaut et al., 1997), has recently beendescribed as being a true speci®city mutant, in that it iscapable of binding to an altered p53 consensus bindingsite (5'-TTTCATGAAA-3') better than to a wild-typesite (5'-AGGCATGTCT-3') (Gagnebin et al., 1998).We found, by replica plating, that the C277Y mutantalso transactivated the p21-containing reporter as wellas wild-type p53, but no other (Table 2, scored as`+++'). However, under more stringent conditions,that is, by streak-out for singles on histidine-de®cientmedia, the ability of the C277Y mutant to transactivatewas greatly reduced (scored as `+', not shown).Interestingly, C277Y was the only mutant surveyedcapable of transactivating the p21-containing reporter,but not the SCS-containing reporter. The SCSsequence was identi®ed by Halazonetis et al. (1993)as a sequence which binds better to wild-type p53 thanother consensus sites. All other mutants which werecapable of activating the p21-containing reporterdemonstrated at least comparable activity with theSCS-containing reporter.The R283H allele was the most wild-type-like

mutant tested, in that it retained transcriptionalactivity towards many of the 53-responsive reportersat the physiological temperature (Table 2). This is theonly tumor-derived mutant surveyed to retain this levelof activity. The crystal structure of p53 bound to itscognate site (Cho et al., 1994) reveals that R283contacts the phosphate backbone of DNA at a distancefrom the primary DNA contacts made by R175 andR273. This might explain why we did not detect adramatic diminution in transactivating function.Interestingly, the p53R283H protein bound moree�ciently (®vefold) than wild-type p53 to both p21and SCS probes by EMSA.

A class of p53 mutants which are binding-site selective atsub-physiological temperature

A third class identi®ed in our survey did nottransactivate or displayed very weak activation(V143A) at 378C. However, these mutants retainedwild-type or near-wild-type activity towards the SCS-and p21-containing reporters at 308C. Additionally,they displayed extreme temperature sensitivity towardsthe remaining reporter constructs, such that activationwas only demonstrated when the temperature wasreduced even further to 248C. These alleles are: V143A,(located in the S3 strand of the b-sandwich); the doublemutant M160I/A161T, (located on the S4 strand);H193R (located on the L2 loop near the S5 strand);Y220C (located on the short loop between the S7 andS8 strand); and I254F (located on the S9 strand) (Choet al., 1994). A feature shared by these residues is thatthey are positioned to maintain the structural integrityof the b-sca�old, although they are at a distance from

Figure 4 Electrophoretic mobility shift assays using immunopur-i®ed and yeast-expressed human wild-type p53. DNA binding byimmunopuri®ed human wild-type p53 (30 ng) to a 32P-labeled p21probe (6 ng) at 248C (lanes 1 ± 4) and by yeast-expressed wild-typep53 (160 mg whole cell extract) to a 32P-labeled p21 probe (6 ng)at 248C (lanes 6 ± 9) in the absence of antibody (lanes 1 and 6) orin the presence of PAb421 (200 ng, lanes 2 and 7) or of PAb1801(200 ng, lanes 4 and 9) or of PAb421 epitope peptide (200 ng,lanes 3 and 8). Lane 5 (C) contains 160 mg of whole yeast cellextract containing a control vector (pADH1 on a LEU2/CENplasmid) expressing no p53. `Flu-p53' denotes immunopurifedhuman wild-type p53; `yeast-p53' denotes yeast-expressed humanwild-type p53; `C' denotes the control whole yeast cell extract. Itshould be noted that *160 mg of yeast extract containsapproximately 800 ng of p53

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2532

the DNA binding surface. While the mutations maycause local deformations in structure, they appear notto have that dramatic an e�ect on p53's DNA bindingsurface, as all still retain some activity, albeit at non-physiological temperatures (Table 3, scored as `+' and`+++' at 248C and 308C). We were able to detectbinding by the p53Y220C protein to both p21 andSCS, whereas the p53H193R and p53I254F proteinsonly bound weakly to p21 by EMSA.Prior results have demonstrated that the V143A

mutant activates transcription of p21-, mdm2-,GADD45-, and Cyclin G-promoter-containing repor-ters in a temperature-sensitive manner (at 328C, butnot 378C) in transiently transfected H1299 humantumor cells (Friedlander et al., 1996a). In ourexamination of this same mutation, we observedtransactivation of the SCS-containing reporter at248C and 308C, but not at 378C (Table 3, scored as`+++', and not shown). Moreover, the V143Amutant transactivated the p21-containing reporter tonear wild-type levels at 248C, and to a lesser extent at308C and 378C (Table 3, scored as `+++' to `+', andnot shown). However, we observed no such activationof the mdm2-, GADD45-, or Cyclin G-containingreporters by this mutant. To test whether this lack oftransactivation ability was due to low protein levels, weoverexpressed the V143A allele from the GAL1promoter and found no di�erences in transactivationof p21- and SCS-containing reporters from the ADH1expressed constructs (not shown), and neither mdm2,GADD45-, nor Cyclin G-containing reporters weretransactivated. Note, however, that the mdm2- andCyclin G-containing reporters utilized by Friedlanderet al. (1996a) contained large genomic segments ofeither intronic or promoter DNA, respectively. We cannot explain the lack of transactivation of theGADD45-containing reporter, as our intronicGADD45 element is nearly `equivalent' to that usedby Friedlander et al. (1996a).

Mutations which disrupt the p53 DNA binding domainby deletion or intron retention are inactive

Two additional tumor-derived mutations surveyed arethe S261R allele (which also retains intron 7 due to amutation at the donor splicing site) and the Intron 7allele (which retains intron 7 due to a six base deletionaround the intron 7 splice donor site). Thesemutations, by including intron 7, result in truncatedproteins terminating in intron 8 (as determined bysequence analysis and Western blot analysis, notshown) and excluding exons 9 and 10. While exons 9and 10 do not encode residues crucial to DNA binding,they do encode the tetramerization domain (residues*320 ± 360), the basic domain (residues *363 ± 393);and one of the two nuclear localization signals (thesignal located downstream of the tetramerizationdomain). Both mutants failed to transactivate everyp53-responsive reporter at all temperatures (notshown).

The double mutant L22E/W23S retains transcriptionalactivity in yeast towards p53-responsive reporters

The hydrophobic residues at positions L22 and W23 ofhuman p53 were originally identi®ed by Lin et al.

(1994) as being critical for transcriptional transactiva-tion, as well as for the binding of Hdm2 (the humanhomolog of mdm2) to the p53 amino terminus.Moreover, as described above, this transactivation byp53 appears to be mediated through direct contacts ofthe amino terminus of p53 with components of thegeneral transcription machinery. While the doublemutant L22E/W23S fails to activate transcription inmammalian cells, it still retains the ability to bind top53-responsive cis-acting elements as well as wild-typep53 proteins (Lin et al., 1994; CJD and CP,unpublished). Unexpectedly, when we tested this samemutation, we observed transactivation of the SCS-,p21-, Cyclin G-, and GADD45-containing reporters(Table 2, scored as `++' to `+++'). The L22E/W23S mutant was as active as wild-type p53 towardsthe SCS- and p21-containing reporters at 378C, butdisplayed lesser activity with Cyclin G and GADD45by streak-out (Table 2, scored as `+' to `++' and notshown). Interestingly, L22E/W23S displayed extremelyweak transactivation of the mdm2-containing reporter,and was virtually incapable of activating the Bax-containing reporter at all temperatures (Table 2, scoredas `7'). Additionally, we were able to detect DNAbinding by the p53L22E/W23S protein to a p21 probein the absence and presence of p53-speci®c monoclonalantibodies by EMSA.

Transcriptional activation by mutant p53 proteins inyeast does not always correlate with DNA binding byEMSA

Most tumor-derived mutations are found within theDNA binding domain of p53. Our analysis hasdemonstrated that many tumor-derived mutants retaintranscriptional activity that is binding site selective andtemperature sensitive (Tables 1, 2 and 3). We weretherefore interested in comparing DNA binding byeither wild-type or those mutant forms of p53, whichhad retained transcriptional activity towards the p21

Table 4 Activation versus DNA binding

EMSA with p21 probep53 Txn Bind Bind+421 Bind+1801

Controlwild-typeL22E/W23SV143AM1601/A161TR175HP177LH179LH179RH193RY220CG245DR248WR249S1254FS261RR267WR273HC277YR282AR283HIntron 7

7++++7+77++777+7+7+7+7

7++77777ND77777777777++ND

7++777+7ND7+7777777+7++ND

7++777+7ND++777+7+7+7++ND

`7': lack of transactivation and binding to p21; `+': ability totransactivate and bind to p21; `N.D.': not determined

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2533

cis-acting element-containing reporter, in yeast wholecell extracts. Unexpectedly, we found that transcrip-tional activity in vivo did not necessarily correlate withDNA binding in vitro (Table 4).As seen in Figure 4, using EMSA, human wild-type

p53 expressed in whole yeast cell extracts binds to a p21probe, in the presence and absence of p53-speci®cantibodies or PAb421 peptide. Similarly, whole cellextracts were prepared from p53 mutant-expressingstrains (V143A, M160I/A161T, I254F and R267W)which were found to transactivate the p21 :HIS3reporter at wild-type levels. However, we were not ableto detect binding by any of these mutants to a p21 probewith the same sequence as in theHIS3 reporter construct(Tables 2, 3 and 4). Moreover, no DNA binding wasdetectable even in the presence of the activatingmonoclonal antibody PAb421. Since it is possible thatthe epitope for this antibody, located in the C-terminusof p53, is masked in these mutants, we repeated theEMSA in the presence of PAb1801, whose epitope islocated in the N-terminus of p53 (residues 46 ± 55). Inthis case, while no detectable binding to a p21 probe byV143A and M160I/A161T was observed, weak bindingwas detected by H193R, I254F, and R267W mutantswith PAb1801 (Table 4). The results with C277Y werealso unexpected: this mutant bound (with antibody)comparably to RGC, SCS, and p21 probes in vitro andyet was only capable of transactivating the p21-containing reporter. With the notable exception ofC277Y, our studies suggest that transcriptional activityin vivo requires a minimal level of wild-type or mutantp53 to be bound to it's cognate binding sites, which drivetranscription of the DUASpGAL1 :HIS3 reporters.

Discussion

In this report we examined the ability of human wild-type and mutant p53 proteins to activate a variety ofp53-responsive binding site-containing reporters in S.cerevisiae. This assay allowed us: (1) to survey of abroad panel of tumor-derived and non-tumor-derivedp53 mutants; (2) to measure the transactivation ofnaturally occurring (RGC, p21, mdm2, GADD45,Cyclin G, Bax, IGF-BP3 Box A and IGF-BP3 BoxB) and synthetic (SCS) p53-responsive elements bywild-type and mutant p53; and (3) to determine thee�cacy of this ability at a variety of temperatures(248C, 308C and 378C).

Yeast-expressed human p53 `hot-spot' mutants aretranscriptionally inactive, even at reduced temperatures

It is remarkable that of all mutants that were tested,only the six `hot-spot' mutants were completelydefective with all reporters and at all temperatures.From these data, we might speculate that the highfrequency of mutation at these codons in p53 re¯ects amore profoundly defective phenotype. P53 has anumber of di�erent transcriptional targets in cellswhose functions regulate cell cycle and cell death.Di�erent types of cells may require unique combina-tions of such targets being activated as part of thetumor suppression regimen of p53. A mutant whichretains partial activity would be more likely to conservesome tumor suppressor activity under certain condi-

tions, perhaps in a cell type speci®c manner. Mutantswhich have lost all transcriptional activity would thenbe more likely to be selected during tumor progression,and thus would appear more frequently in the databaseof p53 mutations. Arguing against this simple hypoth-esis, however, are the observations from a number ofgroups showing that some `hot-spot' mutations displaytranscriptional activity in transiently transfected mam-malian cells, albeit with only a subset of p53-responsivereporters (Chen et al., 1993a; Chumakov et al., 1993;Friedlander et al., 1996b; Ludwig et al., 1996). Whetherthe yeast-based assay in fact re¯ects more accurately thetrue potential (or lack thereof) of various p53 mutantsthus remains to be determined. It should be noted thatPientenpol et al. (1994) observed that R273H displayedpartial transactivation of reporters containing two (butnot single) copies of high a�nity arti®cial binding sites(El-Deiry et al., 1992; Funk et al., 1992). Additionally,those authors used a galactose-inducible promoterwhich yields approximately 25 ± 30-fold more p53protein than the ADH1 promoter used in our studies(not shown, CJD and CP, unpublished). Thus, atextremely high levels of protein, the R273H mutantshows very restricted transactivation function. In oursurvey of the literature related to mutant p53 functionin mammalian cells (Chen et al., 1993a; Chumakov etal., 1993; Miller et al., 1993; Park et al., 1994; Niewoliket al., 1995; Friedlander et al., 1996b), R273H appearsto be the mutant most commonly associated with sometranscriptional activity.

Tumor-derived p53 mutants, once thought to betranscriptionally inactive, retain activity that is bindingsite selective and temperature dependent

Many of the tumor-derived p53 mutants we surveyedwere originally identi®ed as being inactive fortransactivation of an RGC-containing HIS3 reporterat 308C (Ishioka et al., 1993). Surprisingly, when wetested these tumor-derived mutants against otherphysiologically relevant target gene binding site-containing reporters, we found that many retainedactivity. We divided the mutants into two classes: (1)those that retain activity which is binding site selectiveat sub-physiological temperatures (V143A, M160I/A161T, H193R, Y220C and I254F); and (2) thosethat retain activity which is binding site selective, evenat physiological temperatures (L22E/W23S, P177L,R267W, C277Y and R283H). The class of mutantswhich retain activity at sub-physiological temperaturesare all poised to maintain the integrity of the b-sca�old(Cho et al., 1994). While these mutations a�ect residueslocated at a distance from the DNA binding surface,their detrimental e�ect on transactivation reinforces theimportance of the b-sca�old structure in supporting theDNA binding surface of p53.The ®ve mutants which retain activity at physiolo-

gical temperature merit further discussion. Presumably,the tumor-derived mutants (P177L, R267W, C277Yand R283H) were all selected for at 378C. Why wouldthese residues be selected for in tumors, given that theyretain partial wild-type activity? Do these allelespossess a novel function and might these proteinsregulate gene(s) involved in the fate of a cell?Friedlander et al. (1996a) and Ludwig et al. (1996)have shown that V143A and R175P, respectively, are

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2534

defective for inducing apoptosis, yet each can activatetranscription from a subset of p53-responsive reporters.It would be interesting to examine the apoptotic andtransactivation spectrum of P177L, R267W, C277Yand R283H.

The L22E/W23S mutant reveals a fundamentaldi�erence between yeast and mammalian cells

The L22E/W23S double mutant, when expressed inyeast, retained almost full wild-type activity towardsmany of the p53-responsive reporters. This mutant,however, di�ers signi®cantly from wild-type p53 in thatit only very weakly transactivates the mdm2-containingreporter at all temperatures tested, and can not activatethe Bax-containing reporter at all. Recently, Candau etal. (1997), in examining chimeras containing the p53amino terminus and the GAL4 DNA binding domainin yeast, identi®ed a new transactivation subdomain inp53 located between amino acids 40 ± 83. Moreover,residues W53 and F54 were found to be critical forfunction in both yeast and mammalian cells. That theL22E/W23S double mutant retains almost full wild-type activity in a binding site-selective manner, lendssupport to the ®ndings of Candau et al. (1997) andfurther suggests that the subdomain between 40 ± 83may be the primary region in yeast that supportstransactivation, even in the context of the full lengthp53 protein. It has been demonstrated that residuesL22 and W23 are crucial for p53's ability tocommunicate with the general transcriptional machin-ery in mammalian cells. Clearly, however, therequirement for these residues is relaxed in yeast.

DNA binding properties of p53 in yeast

We were able to show that yeast-expressed human wild-type p53 is capable of binding to sequences which arecloned into those reporters. Not only did human p53 inyeast whole cell extracts bind to a p21 probe by EMSA,these complexes were supershifted upon addition ofp53-speci®c monoclonal antibodies and DNA bindingwas enhanced by a p53-speci®c peptide, similar toimmunopuri®ed p53. Although these results suggestthat the transactivation measured in our growth assaysis mostly due to p53 binding to and activating thebinding site-containing reporters, there were discrepan-cies noted upon examination of mutant p53 binding.The ®rst discrepancy was that, in contrast to wild-typep53, a number of mutants were capable of transactiva-tion of a subset of reporters but either did not binddetectably to p21 DNA by EMSA (V143A and M160I/A161T) or bound weakly only in the presence ofantibody (P177L, H193R, I254F and R267W).Addressing the lack of detectable DNA binding by

certain p53 mutants with EMSA, we put forth aspossible explanations the following: (1) despite theirapproximate equivalence with wild-type p53 proteinlevels, the quantities of certain p53 mutants in theextracts might have been limiting; (2) the epitopes forthe antibodies we utilized to detect supershifts might bemasked; (3) the DNA binding activity of some p53mutants might be susceptible to the conditionsemployed in preparing the whole cell extracts; or (4)certain p53 mutants might require a de®ned chromatinstructure when binding to the reporter plasmids in vivo,

and this structure would be absent in the oligoduplexprobes utilized in the electrophoretic mobility shiftassays in vitro. Indeed, recent studies examining thein¯uence of DNA conformation on sequence-speci®cDNA binding by p53 showed that wild-type p53 bindsto certain DNA structures, such as Holliday junctions,in a non-speci®c manner (Lee et al., 1997), as well asnon-B-DNA, in a sequence-speci®c manner (Kim et al.,1997). It is noteworthy that Chen et al. (1993b) foundthat TFIID can stabilize the ability of p53 to bind to itscognate site in DNA. Furthermore, Friedlander et al.(1996b) showed that antibodies to the amino terminusof p53 can stabilize mutant p53 binding to DNA. It isattractive to speculate that certain factors, such ascomponents of the general transcription machinery oras yet unidenti®ed cellular proteins, stabilize or enhanceDNA binding by p53 in vivo. Such interactions withmutant p53 might be disrupted by lysis of yeast cells. A®fth possibility is that sequence-speci®c DNA bindingby certain p53 mutants may be abrogated by the excessdouble-stranded poly(d(I-C)) present in the EMSAreaction bu�er, in addition to the cellular DNA/RNApresent in the yeast cell extracts. A recent studypresented evidence that binding of large, but notsmall, DNAs by the C-terminus of p53 interfered withsequence-speci®c DNA binding by p53 (Anderson et al.,1997). It should be noted that we tested a variety ofEMSA conditions (including varying the ratio ofpoly(d(I-C)) to probe, the amount of spermidine and/or BSA, and the concentration of yeast cell extract) inorder to optimize binding by wild-type p53. Note that itis the binding (or lack thereof) by the tumor-derivedmutants relative to wild-type p53 that we are assaying.Finally, it may be that full transactivation (scored as`+++' in our assay) requires a threshold of binding byp53. Wild-type p53 may be capable of DNA bindingthat is markedly higher than the required threshold, andthe rate limiting factor(s) might then be one or moreinteractions with components of the general transcrip-tion machinery.Another incongruity was the surprising fact that

C277Y cannot activate SCS (or RGC) and yet bindsdetectably to both SCS and RGC by EMSA (albeitonly in the presence of p53 antibodies). The C277Ymutant, recently identi®ed as an altered speci®citymutant (Gagnebin et al., 1998), is clearly worthy offurther attention in order to understand how it bindsto DNA and activates transcription.Finally, that more tumor-derived p53 mutants than

once thought have the intrinsic ability to bind DNAand activate transcription and that antibodies andsmall chemically modi®ed peptides can restore thisDNA binding activity to certain mutants, in vitro aswell as in vivo (Halazonetis and Kandil, 1993; Hupp etal., 1995; Abarzua et al., 1995, 1996; Friedlander et al.,1996b; Selivanova et al., 1997), lends support to thestrategies of identifying molecules that can mimic andextend the stabilizing e�ects of antibodies and peptides.

Materials and methods

Yeast strains and media

All yeast strains were isogenic with S288C, except that theywere wild-type at GAL2. Prior to the introduction of thewild-type or mutant p53 constructs, CUY5 (trp1D1 ura3-52

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2535

his3D200 leu2-3, -112 lys2-801) was transformed with oneof the HIS3 reporter plasmids or control plasmid. Rich(YP) and synthetic complete (SC) media were constitutedas described elsewhere (Rose et al., 1990), except that theYP media also contained 0.1 g of tryptophan per liter and0.1 g of adenine per liter and the SC media contained 0.2 gof leucine, 0.1 g of all other amino acids, 0.1 g of uracil,and 0.1 g of adenine per liter. All strains were grown inglucose to a ®nal concentration of 2%.

Transformation of yeast

Yeast were transformed by a modi®ed version of thelithium acetate (LiOAc) method (Guthrie and Fink, 1991).An overnight culture was diluted in 100 ml of media andgrown for two doublings. The cells were collected bycentrifugation at 5000 r.p.m. for 5 min and washed oncewith sterile water and once with 0.1 M LiOAc. The cellpellet was resuspended in 1/100 of the initial culturevolume in 0.1 M LiOAc. For each transformation, 100 ml ofcells were mixed with plasmid DNA, carrier DNA (2 ml of10 mg/ml boiled, sonicated salmon sperm (Sigma)), andincubated in 800 ml of a 40% polyethylene glycol/0.1 M

LiOAc solution for 30 min. The samples were heat shockedat 428C for 5 min and 10 ml of sterile water was added.The cells were centrifuged at 5000 r.p.m. for 5 min, thepellets resuspended, and the cells spread on selective plates.Plates were incubated at 308C until colonies appeared.

Plasmids

Expression of p53 in yeast pCUB7 (pLS76) expresses fulllength p53 cDNA from the ADH1 promoter with the CYC1terminator downstream of the p53 cDNA, CEN6 andARSH4 for stable, low copy number replication, and theLEU2 gene for plasmid maintenance (Ishioka et al., 1993).The plasmids expressing full length p53 mutants wereisolated by FASAY (Ishioka et al., 1993), except forL22E/W23S, V143A, R175H, R248W, R249S and R282A.In order to obtain pADH1p53L22E/W23S, a 1.6 kb HindIII/XbaI cDNA fragment in pUC119 was subcloned into theHindIII/SacI site of pCB1153 (pADH1 on a LEU2/2mplasmid, (Di Como et al., 1995). In order to obtainpADH1p53V143A, a PshAI/NcoI fragment spanning co-dons 60 ± 159 was swapped into pCUB7 from a cDNA clone(pCUB2: pRS314-CX3, Pientenpol et al., 1994) harboringthe missense mutation. In order to obtain pADH1p53R175Hand pADH1p53R248W, a NcoI/StuI fragment spanningcodons 160 ± 346 was swapped into pCUB7 from cDNAclones (pCUB12: pC53-CX22 (Kern et al., 1991a) andpCUB14: pC53-248 (Kern et al., 1991a), respectively)harboring the missense mutation. In order to obtainpADH1p53R249S and pADH1p53R282A, an Eco47IIIfragment spanning codons 181 ± 336 was swapped intopCUB7 from cDNA clones (pCUB11: pC53-249S (Kern etal., 1992) and pCUB15: pGEMhp53-282A (Waterman et al.,1995), respectively) harboring the missense mutation. Allconstructs were sequenced to con®rm that the p53 cDNAcontained the missense mutation(s).

Reporter plasmids p53-responsive reporter plasmids wereconstructed as follows: a duplex oligonucleotide encodingone of the p53-responsive cis-acting elements (see below)was cloned upstream of the inactive GAL1 promoter(DUASpGAL1) at a unique BamHI site in pCUB79 (aPstI/SacI fragment in pUC119) and the resulting plasmidswere sequenced for the orientation and insert number ofthe oligoduplex. From these constructs, an AatII/SacIfragment was swapped into pCUB53 (pBM2389, Ishioka etal., 1993) containing the HIS3 coding sequence on a TRP1/CEN plasmid. The synthesized oligonucleotides (OperonTechnologies, Inc.) encoding the p53-responsive cis-actingelements (shown in bold) were as follows: CUO3

(5'-GATCCTCGAGGAACATGTCCCAACATGTTGCTC-GAG -3') and CUO4 (5'-GATCCTCGAGCAACATGTT-GGGACATGTTCCTCGAG-3') for p21; CUO11 (5'-GAT-CCTCGAGGGTCAAGTTGGGACACGTCCCTCGAG-3')and CUO12 (5'-GATCCTCGAGGGACGTGTCCCAAC-TTGACCCTCGAG-3') for mdm2; CUO13 (5'-GATCCTC-GAGGAACATGTCTAAGCATGCTGCTCGAG - 3') andCUO14 (5'-GATCCTCGAGCAGCATGCTTAGACATG-TTCCTCGAG-3') for GADD45; CUO15 (5'-GATCCTC-GAGAGACCTGCCCGGGCAAGCCTCTCGAG - 3') andCUO16 (5'-GATCCTCGAGAGGCTTGCCCGGGCAGG-TCTCTCGAG-3') for Cyclin G; CUO17 (5'-GATCCTC-GAGTCACAAGTTAGAGACAAGCCTGGGCGTGGGC-TATATTCTCGAG-3') and CUO18 (5'-GATCCTCGA-GAATATAGCCCACGCCCAGGCTTGTCTCTAACTTG-TGACTCGAG-3') for Bax; CUO24 (5'-GATCCTCGAGG-GGCAAGACCTGCCAAGCCTCTCGAG-3') and CUO25(5' - GATCCTCGAGAGGCTTGGCAGGTCTTGCCCCT-CGAG-3') for IGF-BP3 (Box B); CUO26 (5'-GATCCTC-GAGAAACAAGCCACCAACATGCTTCTCGAG - 3') andCUO27 (5' - GATCCTCGAGAAGCATGTTGGTGGCTT-GTTTCTCGAG-3') for IGF-BP3 (Box A); and CUO28 (5'-GATCCTCGAGGGGCATGTCCGGGCATGTCCCT-CG-AG-3') and CUO29 (5'-GATCCTCGAGGGACATGCC-CGGACATGCCCCTCGAG-3') for SCS. The RGC-con-taining p53-responsive reporter plasmid (pSS1) is asdescribed (Ishioka et al., 1993), where a duplex oligonu-cleotide 5'-TCGACCTTGCCTGGACTTGCCTGGCCTT-GCCTTTTTCGA-3' with XhoI ends was cloned into theBamHI site of pCUB53 (pBM2389) after half-®lling theends with Klenow.

Preparation of protein extracts and detection of p53 protein byWestern immunoblot analysis

Cellular extracts and Western immunoblots were preparedas described, except for minor changes (Sutton et al., 1991;Di Como and Arndt, 1996). Brie¯y, exponentially growingcells were harvested by centrifugation and washed with icecold lysis bu�er (100 mM Tris-HCl (pH 7.5), 200 mM NaCl,1 mM EDTA, 5% glycerol, 0.5 mM dithiothreitol). Cellswere resuspended in 100 ml of lysis bu�er containing PI(1 mM phenylmethylsulfonyl ¯uoride and 1.2 mg each ofleupeptin, antipain, chymostatin, and pepstatin per ml(Sigma)). The cells were lysed by vortexing four times for15 s in the presence of glass beads. An additional 50 ml oflysis bu�er containing PI was added, the cells vortexed oncefor 10 s and the free liquid removed. This step was repeatedand the free liquids (extracts) were combined andcentrifuged at 16 000 g for 20 min to remove cell debris.Protein concentrations were determined using the Bio-Radassay using bovine globulin as a standard (Bio-Rad Lab,CA). An equal volume of 26gel sample bu�er (Silhavy etal., 1984) was added to the extracts. Samples were heated to958C for 5 min, centrifuged for 3 min at 16 000 g, andelectrophoresed through a 10% SDS-polyacrylamide gel.Protein gels were transferred to polyvinylidene ¯uoride(PVDF) membranes (Millipore). For p53 detection, amixture of puri®ed p53 monoclonal antibodies was usedcontaining PAb421, PAb1801, PAb240, and DO-1, each at a1/3000 dilution of a 50 ng/ml stock; for b-tubulin detection,the primary monoclonal antibody was B1BE2 at a 1/3000dilution. Proteins were visualized with an enhancedchemiluminescence detection system (Amersham).

Electrophoretic mobility shift assays with immunopuri®ed p53and whole yeast cell extracts expressing p53

Electrophoretic mobility shift assays were performed asdescribed, except for minor changes (Freeman et al., 1994;Friedlander et al., 1996b). Brie¯y, for whole yeast cellextracts, exponentially growing cells were harvested by

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2536

centrifugation and washed with ice cold lysis Bu�er A(20 mM Tris-HCl (pH 7.2). 100 mM NaCl, 5 mM EDTA,10% glycerol, 0.1% Nonidet P-40, 2 mM dithiothreitol).Cells were resuspended in 100 ml of lysis Bu�er Acontaining PI (1 mM phenylmethylsulfonyl ¯uoride and1.2 mg each of leupeptin, antipain, chymostatin andpepstatin per ml (Sigma)). The cells were lysed byvortexing four times for 15 s in the presence of glassbeads. An additional 50 ml of lysis Bu�er A containing PIwas added, the cells vortexed once for 10 s and the freeliquid removed. This step was repeated and the free liquids(extracts) were combined and centrifuged at 16 000 g for20 min to remove cell debris. Protein concentrations weredetermined using the Bio-Rad assay using bovine globulinas a standard (Bio-Rad Lab, CA). When necessary, totalcell extracts were frozen in a dry ice/alcohol bath andstored at 7808C.

Recombinant baculovirus expressing wild-type p53 hasbeen described (Friedman et al., 1990; Bargonetti et al.,1992). Extracts of infected sf9 insect cells were prepared andp53 was puri®ed from lysates by immunoa�nity procedures(Wang et al., 1989). Puri®ed p53 protein was prepared withprotein A-Sepharose columns cross-linked with the p53-speci®c monoclonal antibody PAb421 (Harlow et al., 1981).The proteins were eluted with a molar excess of PAb421epitope-containing peptide (KKGQSTSRHKK-OH) (Wade-Evans and Jenkins, 1985) or with 50% ethylene glycol. Theprotein was dialyzed into Dialysis Bu�er (10 mM HEPES(pH 7.5), 5 mM NaCl, 0.1 mM EDTA, 1 mM dithiothreitoland 50% glycerol).

Probes were made as follows: synthesized oligonucleotidesencoding the p53-responsive elements (40-mers) were self-annealed and extended with Klenow in the presence of[a-32P]dATP.

Binding reaction mixtures contained either 160 mg ofwhole yeast cell extract or 30 ng of immunopuri®ed wild-type p53 in Gel Shift Bu�er (20 mM HEPES (pH 7.9), 25 mM

KCl, 0.1 mM EDTA, 2 mM MgCl2, 0.5 mM dithiothreitol,0.025% Nonidet P-40, 2 mM spermidine (Sigma), 10%glycerol, 0.1 mg acetylated bovine serum albumin (NEB)and double-stranded poly(d(I-C)); 120 ng (BoehringherMannheim) to a ®nal volume of 20 ml. In all cases, volumeswere equalized with Dialysis Bu�er for immunopuri®ed p53or Bu�er A for whole yeast cell extract-containing p53.Reactions were incubated for 10 min at 248C, after which6 ng of 32P-labeled oligonucleotide was added and theincubation was continued at 248C for 15 min. Samples wererun on 4.5% native polyacrylamide gels at 180 ± 200 V (notexceeding 40 mA current) in 0.56TBE bu�er at 248C for 2 h.Gels were dried under vacuum at 808C and exposed toKodak XAR ®lm overnight at 7808C.

AcknowledgementsWe gratefully thank R Iggo for the human tumor-derivedp53 expression plasmids, for the pSS1 and pBM2389reporter constructs, and for his insightful discussions andcomments. We also thank G Petsko for the S288C yeaststrain (CUY5) and helpful comments. P Friedlander, LJayaraman, L Ko, K Okamoto, and S-Y Shieh are thankedfor additional plasmids. Finally, we thank E Freulich forthe preparation of the anti-p53 antibodies and B Futcherfor the anti-b-tubulin antibody. Critical comments anddiscussion on the manuscript were made by J Ahn, CGaiddon and K Okamato. CJD is supported by the CancerResearch Fund of the Damon Runyon-Walter WinchellFoundation Fellowship, DRG-1427. CP is supported by agrant from the US Army, (DAMD17-94)4275.

References

Abarzua P, LoSardo JE, Gubler ML, Spathis R, Lu Y-A,Felix A and Neri A. (1996). Oncogene, 13, 2477 ± 2482.

Abarzua P, LoSardo JE, Gubler ML and Neri A. (1995).Cancer Res., 55, 3490 ± 3494.

Anderson ME, Woelker B, Reed M, Wang P and TegtmeyerP. (1997). Mol. Cell. Biol., 17, 6255 ± 6264.

Avantaggiati ML, Ogryzko V, Gardner K, Giordano A,Levine AS and Kelly K. (1997). Cell, 89, 1175 ± 1184.

Banks L, Matlashhewski G and Crawford L. (1986). Eur. J.Biochem., 159, 529 ± 534.

Bargonetti J, Manfredi JJ, Chen X, Marshak DR and PrivesC. (1993). Genes & Dev., 7, 2565 ± 2574.

Bargonetti J, Reynisdottir I, Friedman PN and Prives C.(1992). Genes & Dev., 6, 1886 ± 1898.

Bayle JH, Elenbaas B and Levine AJ. (1995). Proc. Natl.Acad. Sci. USA, 92, 5729 ± 5733.

Bischo� JR, Casso D and Beach D. (1992). Mol. Cell. Biol.,12, 1405 ± 1411.

Brain R and Jenkins JR. (1994). Oncogene, 9, 1775 ± 1780.Buckbinder L, Talbott R, Velasco-Miguel S, Takenaka I,Faha B, Seizinger BR and Kley N. (1995). Nature, 377,646 ± 649.

Candau R, Scolnick DM, Darpino P, Ying CY, HalazonetisTD and Berger SL. (1997). Oncogene, 15, 807 ± 816.

Chang J, Kim D-H, Lee SW, Choi KY and Sung YC. (1995).J. Biol. Chem., 270, 25014 ± 25019.

Chen J-Y, Funk WD, Wright WE, Shay JW and Minna JD.(1993a). Oncogene, 8, 2159 ± 2166.

Chen X, Farmer G, Zhu H, Prywes R and Prives C. (1993b).Genes & Dev., 7, 1837 ± 1849.

Cho Y, Gorina S, Je�rey PD and Pavletich NP. (1994).Science, 265, 346 ± 355.

Chumakov AM, Miller CW, Chen DL and Koe�er HP.(1993). Oncogene, 8, 3005 ± 3011.

Di Como CJ, Chang H and Arndt KT. (1995). Mol. Cell.Biol., 15, 1835 ± 1846.

Di Como CJ and Arndt KT. (1996). Genes & Dev., 10, 1904 ±1916.

El-Deiry WS, Kern SE, Pientenpol JA, Kinzler KW andVogelstein B. (1992). Nat. Genet., 1, 45 ± 49.

El-Deiry WS, Tokino T, Velculescu VE, Levy DB, ParsonsR, Trent JM, Lin D, Mercer WE, Kinzler KW andVogelstein B. (1993). Cell, 75, 817 ± 825.

Farmer G, Colgan J, Nakatani Y, Manley JL and Prives C.(1996). Mol. Cell. Biol., 16, 4295 ± 4304.

Fields S and Jang SK. (1990). Science, 249, 1046 ± 1049.Freeman J, Schmidt S, Scharer E and Iggo R. (1994). EMBO

J., 13, 5393 ± 5400.Friedlander P, Haupt Y, Prives C and Oren M (1996a).Mol.

Cell. Biol., 16, 4961 ± 4971Friedlander P, Legros Y, Soussi T and Prives C. (1996b). J.

Biol. Chem., 271, 25468 ± 25478.Friedman PN, Kern SE, Vogelstein B and Prives C. (1990).

Proc. Natl. Acad. Sci. USA, 87, 9275 ± 9279.Friend S. (1994). Science, 265, 334 ± 335.Funk WD, Pak DT, Karas RH, Wright WE and Shay JW.(1992). Mol. Cell. Biol., 12, 2866 ± 2871.

Gagnebin J, Kovar H, Kajava AV, Estreicher A, Jug G,Monnier P and Iggo R. (1998). Oncogene, 16, 685 ± 690.

Gottlieb T and Oren M. (1996). Biochim. Biophys. Acta,1287, 77 ± 102.

Gu W and Roeder RG. (1997). Cell, 90, 595 ± 606.Gu W, Shi X-L and Roeder RG. (1997). Nature, 387, 819 ±823.

Guthrie JA and Fink GR. (1991). Methods in Enzymology.Volume 194, San Diego, California: Academic Press, Inc.

Hainaut P, Butcher S and Milner J. (1995). Br. J. Cancer, 71,227 ± 231.

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2537

Hainaut P, Soussi T, Shomer B, Hollstein M, Greenblatt M,Hovig E, Harris CC and Montesano R. (1997).Nucl. AcidsRes., 25, 151 ± 157.

Hainaut P, Hernandez T, Robinson A, Rodriguez-Tome P,Flores T, Hollstein M, Harris CC and Montesano R.(1998). Nucl. Acids Res., 26, 207 ± 215.

Halazonetis TD, Davis LJ and Kandil AN. (1993). EMBO J.,12, 1021 ± 1028.

Halazonetis TD and Kandil AN. (1993). EMBO J., 12,5057 ± 5064.

Harlow E, Crawford LV, Pim DC and Williamson NM.(1981). J. Virol., 39, 861 ± 869.

Harper JW, Adami GR, Wei N, Keyomarsi K and ElledgeSJ. (1993). Cell, 75, 805 ± 816.

Hollstein M, Sidransky D, Vogelstein B and Harris CC.(1991). Science, 253, 49 ± 53.

Hollstein M, Rice K, Greenblatt MS, Soussi T, Fuchs R,Sorlie T, Hovig E, Smith-Sorensen B, Montesano R andHarris CC. (1994). Nucl. Acids Res., 22, 3551 ± 3555.

Hupp TR, Meek DW, Midgley CA and Lane DP. (1992).Cell, 71, 875 ± 886.

Hupp TR, Meek DW, Midgley CA and Lane DP. (1993).Nucl. Acids Res., 21, 3167 ± 3174.

Hupp TR, Sparks A and Lane DP. (1995). Cell, 83, 237 ± 245.Ishioka C, Frebourg T, Yan Y-X, Vidal M, Friend SH,Schmidt S and Iggo R. (1993). Nat. Genet., 5, 124 ± 129.

Jayaraman L and Prives C. (1995). Cell, 81, 1021 ± 1029.Kastan MB, Zhan Q, El-Deiry WS, Carrier F, Jacks T,Walsh WV, Plunkett BS, Vogelstein B and Fornace AJ.(1992). Cell, 71, 587 ± 597.

Kern SE, Kinzler KW, Baker SJ, Nigro JM, Rotter V, LevineAJ, Friedman P, Prives C and Vogelstein B. (1991a).Oncogene, 6, 131 ± 136.

Kern SE, Kinzler KW, Bruskin A, Jarosz D, Friedman P,Prives C and Vogelstein B. (1991b). Science, 252, 1708 ±1711.

Kern SE, Pientenpol JA, Thiagalingam S, Seymour A,Kinzler KW and Vogelstein B. (1992). Science, 256,827 ± 830.

Kim E, Albrechtsen N and Deppert W. (1997). Oncogene, 15,857 ± 869.

Ko LJ and Prives C. (1996). Genes & Dev., 10, 1054 ± 1072.Kovar H, Auinger A, Jug G, Aryee D, Zoubek A, Salzer KMand Gadner H. (1993). Oncogene, 8, 2683 ± 2690.

Lee S, Elenbaas B, Levine A and Gri�th J. (1995). Cell, 81,1013 ± 1020.

Lee S, Cavallo L and Gri�th J. (1997). J. Biol. Chem., 272,7532 ± 7539.

Legros Y, Lafon C and Soussi T. (1994). Oncogene, 9, 2071 ±2076.

Leveillard T, Andera L, Bissonnette N, Schae�er L, BraccoL, Egly J-M and Wasylyk B. (1996). EMBO J., 15, 1615 ±1624.

Levine AJ, Wu MC, Chang A, Silver A, Attiyeh EF, Lin Jand Epstein CB. (1995). Ann. NY Acad. Sci., 768, 111 ±128.

Levine AJ. (1997). Cell, 88, 323 ± 331.Lill NL, Grossman SR, Ginsberg D, DeCaprio J andLivingston DM. (1997). Nature, 387, 823 ± 827.

Lin J, Chen J, Elenbaas B and Levine AJ. (1994). Genes &Dev., 8, 1235 ± 1246.

Liu X, Miller CW, Koe�er PH and Berk AJ. (1993). Mol.Cell. Biol., 13, 3291 ± 3300.

Lu H and Levine AJ. (1995). Proc. Natl. Acad. Sci. USA, 92,5154 ± 5158.

Ludwig RL, Bates S and Vousden KH. (1996). Mol. Cell.Biol., 16, 4952 ± 4960.

Martin DW, Munoz RM, Subler MA and Deb S. (1993). J.Biol. Chem., 268, 13062 ± 13067.

Miller CW, Chumakov A, Said J, Chen DL, Aslo A andKoe�er HP. (1993). Oncogene, 8, 1815 ± 1824.

Miyashita T and Reed JC. (1995). Cell, 80, 293 ± 299.

Niewolik D, Vojtesek B and Kovarik J. (1995). Oncogene, 10,881 ± 890.

Nigro JM, Baker SJ, Preisinger AC, Jessup JM, Hostetter R,Cleary K, Bigner SH, Davidson N, Baylin SJ, Devilee P,Glover T, Collins FS, Weston A, Modali R, Harris CCand Vogelstein B. (1989). Nature, 342, 705 ± 708.

Nigro JM, Sikorski R, Reed SI and Vogelstein B. (1992).Mol. Cell. Biol., 12, 1357 ± 1365.

Noda A, Ning Y, Venable SF, Pereira-Smith OM and SmithJR. (1994). Exp. Cell Res., 211, 90 ± 98.

Okamoto K and Beach D. (1994). EMBO J., 13, 4816 ± 4822.Park DJ, Nakamura H, Chumakov AM, Said JW, MillerCW, Chen DL and Koe�er HP. (1994). Oncogene, 9,1899 ± 1906.

Pavletich NP, Chambers KA and Pabo CO. (1993). Genes &Dev., 7, 2556 ± 2564.

Picksley SM, Meek DW and Lane DP. (1992). Oncogene, 7,1649 ± 1651.

Pientenpol JA, Tokino T, Thiagalingam S, El-Deiry WS,Kinzler KW and Vogelstein B. (1994). Proc. Natl. Acad.Sci. USA, 91, 1998 ± 2002.

Prives C. (1994). Cell, 78, 543 ± 546.Ragimov N, Krauskopf A, Navot N, Rotter V, Oren M andAloni Y. (1993). Oncogene, 8, 1183 ± 1193.

Raycroft L, Wu H and Lozano G. (1990). Science, 249,1049 ± 1051.

Raycroft L, Schmidt JR, Yoas K, Hao M and Lozano G.(1991). Mol. Cell. Biol., 11, 6067 ± 6074.

Reed M, Woelker B, Wang P, Wang Y, Anderson ME andTegtmeyer P. (1995). Proc. Natl. Acad. Sci. USA, 92,9455 ± 9459.

Rose MD, Winston F and Heiter P. (1990). LaboratoryCourse Manual for Methods in Yeast Genetics. Cold SpringHarbor Laboratory Press: Cold Spring Harbor, NewYork.

Scharer E and Iggo R. (1992). Nucl. Acids Res., 20, 1539 ±1545.

Scolnick DM, Chehab NH, Stavridi ES, Lien MC, Caruso L,Moran E, Berger SL and Halazonetis TD. (1997). CancerRes., 57, 3693 ± 3696.

Selivanova G, Iotsova V, Okan I, Fritsche M, Strom M,Groner B, Grafstrom RC and Wiman KG. (1997). Nat.Med., 3, 632 ± 638.

Seto E, Usheva A, Zambetti GP, Momand J, Horikoshi N,Weinmann R, Levine AJ and Shenk T. (1992). Proc. Natl.Acad. Sci. USA, 89, 12028 ± 12032.

Shaulsky G, Gold®nger N, Tosky MS, Levine AJ and RotterV. (1991). Oncogene, 6, 2055 ± 2065.

Shaw P, Freeman J, Bovey R and Iggo R. (1996). Oncogene,12, 921 ± 930.

Siegel J, Fritsche M, Mai S, Brandner G and Hess RD.(1995). Oncogene, 11, 1363 ± 1370.

Silhavy TJ, Berman ML and Enquist LW. (1984). Experi-ments with Gene Fusions. Cold Spring Harbor LaboratoryPress: Cold Spring Harbor, New York.

Sutton A, Immanuel D and Arndt KT. (1991). Mol. Cell.Biol., 11, 2133 ± 2148.

Thut C, Chen JL, Klemm R and Tjian R. (1995). Science,267, 100 ± 104.

Truant R, Xiao H, Ingles CJ and Greenblatt J. (1993). J.Biol. Chem., 268, 2284 ± 2287.

Unger T, Nau MM, Segal S and Minna JD. (1992). EMBOJ., 11, 1383 ± 1390.

Unger T, Mietz JA, Sche�ner M, Yee CL and Howley PM.(1993). Mol. Cell. Biol., 13, 5186 ± 5194.

Vogelstein B and Kinzler KW. (1992). Cell, 70, 523 ± 526.Wade-Evans A and Jenkins J. (1985). EMBO J., 4, 699 ± 706.Wagner P, Simanis V, Maimets T, Keenan E, Addison C,Brain R, Grimaldi M, Sturzbecher HW and Jenkins JR.(1991). Oncogene, 6, 1539 ± 1547.

Wagner P, Grimaldi M and Jenkins JR. (1993). Eur. J.Biochem., 217, 731 ± 736.

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2538

Walker KK and Levine AJ. (1996). Proc. Natl. Acad. Sci.USA, 93, 15335 ± 15340.

Wang EH, Friedman PN and Prives C. (1989). Cell, 57, 379 ±392.

Wang Y, Reed M, Wang P, Stenger JE, Mayr G, AndersonME, Schwedes JF and Tegtmeyer P. (1993). Genes & Dev.,7, 2575 ± 2586.

Wang XW, Yeh H, Schae�er L, Roy R, Moncollin V, Egly J-M, Wang Z, Friedberg EC, Evans MK, Ta�e BG, BohrVA, Weeda G, Hoeijmakers JHJ, Forrester K and HarrisCC. (1995). Nat. Genet., 10, 188 ± 193.

Waterman JLF, Shenk JL and Halazonetis TD. (1995).EMBO J., 14, 512 ± 519.

Wu X, Bayle JH, Olson D and Levine AJ. (1993). Genes &Dev., 7, 1126 ± 1132.

Xiao H, Pearson A, Coulombe B, Truant R, Zhang S, RegierJL, Triezenberg SJ, Reinberg D, Flores O, Ingles CJ andGreenblatt J. (1994). Mol. Cell. Biol., 14, 7013 ± 7024.

Xiong Y, Hannon GJ, Zhang H, Casso D, Kobayashi R andBeach D. (1993). Nature, 366, 701 ± 704.

Zauberman A, Lupo A and Oren M. (1995). Oncogene, 10,2361-2366.

Target gene selectivity by mutant p53 proteinsCJ DiComo and C Prives

2539