The Screening of the Second-Site Suppressor Mutations of The

Int. J. Cancer: 121, 559–566 (2007) ' 2007 Wiley-Liss, Inc.

The screening of the second-site suppressor mutations of the common p53 mutants Kazunori Otsuka, Shunsuke Kato, Yuichi Kakudo, Satsuki Mashiko, Hiroyuki Shibata and Chikashi Ishioka* Department of Clinical Oncology, Institute of Development, Aging and Cancer, and Tohoku University Hospital, Tohoku University, Sendai, Japan

Second-site suppressor (SSS) mutations in p53 found by random mutation library covers more than 95% of tumor-derived missense mutagenesis have shown to restore the inactivated function of mutations as well as previously unreported missense mutations. some tumor-derived p53. To screen novel SSS mutations against We have shown the interrelation among the p53 structure, the common mutant p53s, intragenic second-site (SS) mutations were function and tumor-derived mutations. We have also observed that introduced into mutant p53 cDNA in a comprehensive manner by the missense mutation library was useful to isolate a number of using a p53 missense mutation library. The resulting mutant p53s 16,17 with background and SS mutations were assayed for their ability temperature sensitive p53 mutants and super apoptotic p53s. to restore the p53 transactivation function in both yeast and Previous studies have shown that there are second-site (SS) mis- human cell systems. We identified 12 novel SSS mutations includ- sense mutations that recover the inactivated function of tumor- ing H178Y against a common mutation G245S. Surprisingly, the derived mutations, so-called intragenic second-site suppressor G245S phenotype is rescued when coexpressed with p53 bearing (SSS) mutations.18–23 Analysis of the molecular background of the H178Y mutation. This result indicated that there is a possibil- ity that intragenic suppressor mutations might restore the protein the SSS mutation should be interesting because the recovery of function in an intermolecular manner. The intermolecular mecha- mutant p53 function might be useful for cancer treatment, espe- cially the design of small molecules that restore mutant p53 func- nism may lead to novel strategies for restoring inactivated p53 24 function and tumor suppression in cancer treatment. tion. ' 2007 Wiley-Liss, Inc. The purpose of this study is the isolation of SSS mutations against the 10 most common tumor-derived p53 mutations, Key words: p53; second-site suppressor mutation; intragenic V157F, R175H, C176F, G245S, R248Q, R248W, R249S, R273C, suppressor; transactivation R273H and R282W using our comprehensive p53 mutation library (see above), and a yeast-based functional assay, and the investiga- tion of the underlying molecular mechanism. We isolated 12 Tumor-suppressor p53 protein is a 393 amino-acid nuclear pro- mutations against G245S and 1 mutation against R273H, and have tein that acts as a tetramer. It is divided into 3 domains, an NH2- shown that G245S phenotype is rescued when coexpressed with terminal domain containing a transactivation domain, a central p53 bearing the H178Y mutation. The underlying molecular core DNA-binding domain and a COOH-terminal domain contain- mechanism of the SSS mutation described here clearly differed ing a tetramerization domain.1–6 Among these, the core DNA- from previous knowledge. Further understanding of the inter-p53 binding domain is well conserved and retains high homology mechanism may lead to novel strategies for restoring inactivated between a variety of lower species and humans7 as well as human p53 function and tumor suppression in cancer therapeutics. p53 homologues.8,9 The structure of the core DNA-binding domain has been resolved by X-ray crystallographic analysis.2 The p53 forms a homotetramer through the tetramerization domain Material and methods and the p53 tetramer forms a sequence-specific DNA-binding Preparation of gap-repair vectors interface. The p53 protein is activated by a variety of cellular The pLSC53A25-based mutant p53 expression vectors with 1 of stresses including DNA damage and hypoxia, and is phosphoryl- the 10 most common tumor-derived p53 mutations, V157F, ated and acetylated after translation. The activated p53 binds to R175H, C176F, G245S, R248Q, R248W, R249S, R273C, R273H the specific DNA sequence in the regulatory region of downstream or R282W15 were digested by Bsu36I and StuI for the plasmids genes, resulting in cellular events including cell-cycle arrest and with V157F, R175H and C176F (Supplementary Fig. 1a)orby apoptosis. So far, a number of downstream genes involved in cell- NcoI and Bsu36I for the remaining 7 plasmids (Supplementary cycle, apoptosis, DNA-repair, angiogenesis and p53 stability have Fig. 1b). The linearized plasmids were separated by agarose-gel been identified. The loss of p53 function therefore fails to activate electrophoresis, purified with Gfx (Pharmacia) and used as gap- these genes after cellular stresses and is thought to be a critical repair vectors for each of the background mutations. cause of carcinogenesis and/or tumor progression. Although the frequency of TP53 mutations differs among tumor types, 50% of tumors contained the TP53 mutation.10–12 The Preparation of SS mutations published mutations have been summarized in the 2 major TP53 The p53 mutation library containing 2314 p53 missense muta- mutation databases that contain more than 20,000 mutations.13,14 tions was constructed through 96-well formatted site-directed mu- According to the databases, 74% of mutations are missense muta- tagenesis,15 and was used for PCR templates of SS mutations. For tions. So far, 1,200 distinct missense mutations have been reported background mutations, V157F, R175H and C176F, the p53 frag- and there are mutation hot spots at residues R175, G245, R248, ments covering codons from 192 to 354 were amplified from the R273 and R282. These residues reside in the L2 or L3 loop, or the yeast library using a set of primers, OKO-04 (50-CCCCACCAT- LSH motif and may be particularly critical because the stability of the structures is thought to depend on interactions among side- This article contains supplementary material available via the Internet at chains, or between side chains and the backbone structure. Amino http://www.interscience.wiley.com/jpages/0020-7136/suppmat. acid substitutions at these sites are therefore more likely to destroy Grant sponsors: Ministry of Education, Science, Sports and Culture, the DNA-binding interface than the core b-sandwich structure. Gonryo Medical Foundation. Recently, we have used a comprehensive site-directed mutagen- *Correspondence to: 4-1 Seiryo-machi, Aoba-ku, Sendai 980-8575, Japan. Fax: 181-22-717-8548. E-mail: [email protected] esis technique and a yeast-based functional assay to construct, ex- Received 9 August 2006; Accepted after revision 13 February 2007 pressed and evaluated 2,314 p53 mutants representing all possible DOI 10.1002/ijc.22724 amino-acid substitutions caused by a point mutation throughout Published online 6 April 2007 in Wiley InterScience (www.interscience. the protein (5.9 substitutions per residue).15 The TP53 missense wiley.com).

Publication of the International Union Against Cancer 560 OTSUKA ET AL. GAGCGCTGCTCAGATAG-30) and LS6 (50-GCGAAGCTT- Mammalian p53 expression vectors 0 CAGTCTGAGTCAGGCCCTT-3 ) (Supplementary Fig. 1a). For Mutant p53-expression vectors, pCR259-p53MTs, with candi- G245S, R248Q, R248W, R249S, R273C, R273H and R282W, the date SSS mutations and background mutations in the p53 open p53 fragments covering codons from 110 to 225 were amplified reading frame were constructed. pCR259-p53WT and pCR259 0 from the yeast library using a set of primers, OKO-03 (5 - were control vectors expressing wild-type p53 or null p53, respec- 0 0 GCCCATGCAGGAACTGTTACACATG-3 ) and LS-5 (5 - tively.15 CGGGATCCATGGAGGAGCCGCAGTCA-30) (Supplementary Fig. 1b). Forty-eight p53 fragments were combined as a pool of Cell culture and transfection the PCR fragments. A TP53-deficient human osteosarcoma cell line, Saos-2, was Screening SSS mutations using the yeast assay cultured in RPMI1640 medium, supplemented with 10% heat- inactivated (56°C, 30 min) fetal calf serum (JRH Bioscience) in The linearized gap vector with the background mutation was the presence of 5% CO2 at 37°C. For luciferase assays, the cells cotransformed with the pooled PCR fragments (see above) into were grown to 60–90% confluence in 96-well tissue culture plates. the yeast haploid strain, YPH499 (MATa, his3D200, ade2-101, For immunoblotting, the cells were grown to the same confluence leu2D1, ura3-52, trp1-289, lys2-801) harboring p21WAF1 reporter 2 25 on 90 3 20 mm tissue culture plates. Transient transfections EGFP plasmid, pAS03G. The SS mutation was recombined into were performed using the Effectene transfection reagent (Qiagen). the vector on a solid medium-containing synthetic complete (SC) For luciferase assays, the cells were cotransfected with 12.5 ng of media lacking leucine and tryptophane (SC-leu-trp). The resulting one of the expression vectors (pCR259-p53WT, pCR259-p53MT yeast colonies expressed p53 protein with 2 amino acid substitu- or a p53-null pCR259 vector)15 or 6.25 ng of each of the 2 differ- tions derived from both the background and the SS mutation (Sup- ent expression vectors plus 85 ng of the p53-responsive luciferase plementary Figs. 1a and 1b). YPH500 (MATa, his3D200, ade2- plasmid (p21Ps-luc, pMDMPs-luc, pBAXPs-luc, pSIGMAPs- 101, leu2D1, ura3-52, trp1-289, lys2-801) harboring each of the luc p53R2Ps-luc or p53GADD45Ps-luc)15,25 and incubated for Ds-Red reporter plasmids (either pKS05R, pKS07R, pKS09R, 15 further 24 hr. For immunoblotting analysis, the cells were trans- pKS11R, pKS13R, pKS15R or pKS17R) was spread on the sur- fected with 2 lg of one of the expression vectors (a p53-null face of YPD plates. The p53-expressing YPH499 strains were pCR259, pCR259-p53WT, pCR259-p53MT, pCR259-Myc-H178Y then added to the YPH500-spread YPD plates with an autoclaved or pcDNA1.1-HA-G245S) or 1 lg each of the 2 different expres- velvet replicator and further incubated at 30°C for 12 hr. The mat- sion vectors and further incubated for 24 hr. ing diploid cells were selected by another replicator and inocu- lated into SC solid medium lacking leucine, tryptophan and histi- Luciferase assay dine (SC-leu-trp-his) (Supplementary Fig. 1c). The yeast diploid colonies were incubated at 37°C for 48 hr, then, more than 3,000 After 24-hr transfection, luciferin (Steady-Glo luciferase assay colonies for each background mutation were screened on p53-de- system, Promega), a substrate of luciferase, was added to the cul- pendent EGFP or Ds-Red fluorescence under fluorescence micros- ture media and further incubated for 20 min according to the man- copy (MZ8, Leica). The excitation and absorption wavelengths of ufacturer’s instructions. The fluorescent intensity was measured the filters were 480 and 510 nm for EGFP, and 546 and 590 nm using the Fluoroskan Ascent FL (see above). The relative fluores- for Ds-Red, respectively. Yeast colonies that clearly increased cent intensity to the wild-type control was calculated from 2 inde- EGFP or Ds-Red fluorescence were chosen with the naked eye as pendent experimental data. When the value of the fluorescent clones expressing background and candidate SSS mutations. intensity of p53 double mutants was 1.5-fold higher than that of the corresponding background mutation, the SS mutation was Selection of SSS mutations defined as an SSS mutation. To quantitatively select the transactivity of p53 with the candi- FACS analysis date SSS mutations (above), the yeast clones were grown at 37°C for 2 days, and were directly processed in a fluorometer (Fluo- About 500 ng of pCR259 (p53-null), pCR259-p53WT, pCR259- roskan Ascent FL, Labsystems) to measure the fluorescent inten- p53 (G245S), pCR259-p53 (H178Y) or 250 ng each of pCR259- p53 (G245S) plus pCR259-p53 (H178Y) vectors was introduced sity (excitation: 485 nm, emission: 538 nm) of the p53-dependent 5 EGFP expression through a human p21WAF1-derived p53-binding into the Saos-2 cells (1.5 3 10 cells) by lipofection. After 72 hr incubation, the cells were analyzed for cell cycle in a FACS sequence, or the fluorescent intensity of Ds-Red was measured 17 using the same fluorometer (excitation: 544 nm, emission: 590 (Beckman Coulter) as described previously. The ability to nm) to evaluate the p53-dependent Ds-Red expression through induce apoptosis was estimated by DsubG1 value that was other p53-binding sequences. We defined the following criteria to obtained by subtracting the value of the subG1 fraction of mock select SSS mutants for all but the p53R2 reporter plasmid lipofection from the value of subG1 fraction of each p53 expression vector. (pKS17R): (Mb1s/W 0.3) and (Mb1s/Mb 5), Mb and Mb1s indicate the fluorescent intensities of a background mutant and a double mutant with background and SS mutations, respectively. W Immunoblotting analysis of p53 indicates the fluorescent intensity of the yeast clone expressing The yeast transformant expressing wild-type p53 from plasmid wild-type p53. In the p53R2-reporter plasmid, the fluorescent in- pLSC53A,25 pLSC53A12C or pLSC53A15C or p53-less 27 tensity of p53-G245S was retained relatively well (Mb/W 5 pLSX was cultured in YPD liquid media at 30°C. The plasmid, 0.323). We therefore defined the other criteria: Mb1s/Mb 1.3. To pLSC53A12C and pLSC53A15C were constructed by inserting confirm the SSS mutants, full-length p53 cDNA was amplified 2(50-CC-30) and 5 (50-CCCCC-30) nucleotides 50 flanking to an from a set of primers, LS-5 and LS-6, from the each yeast lysate ATG translation initiation codon of p53 cDNA in the pLSC53A. prepared with a Whole Cell Yeast PCR kit (BIO101). The PCR Equal amounts (OD600 5 0.35, 7 ml) of cells were centrifuged and products were purified using a MultiScreen-PCR plate (Millipore), vigorously agitated in 100 ll of lysis buffer containing a Y-PER and were sequenced with a DTCS-Quick Start kit and yeast protein extraction reagent (Pierce), 1 mM PMSF, 1 lM leu- CEQ2000XL DNA Analysis System (Beckman Coulter). The con- peptin, 0.5 mM benzamidine and 1 lM pepstatin. Saos-2 cell firmed p53 cDNA were reintroduced into the YPH499 harboring lysates were prepared in 200 ll of DOC buffer (50 mM Tris–HCl pAS03G with a gap vector, pSS16 (HindIII/StuI digest),26 and the (pH8.8), 150 mM NaCl, 5 mM EDTA, 1%NP-40, 0.5% sodium resulting transformants were further incubated with YPH500 har- deoxycholate). After removing cell debris by centrifugation, the boring the Ds-Red reporter plasmid. These transformants were supernatant lysates were separated by SDS-10% PAGE. For tetra- measured for their fluorescence intensities and the data from 3 merization analysis, 100 ll of the cell lysates were cross-linked by independent experiments were averaged. 0.1% glutaraldehyde, and then separated by SDS/5–20% gradient SECOND-SITE SUPPRESSOR MUTATIONS OF p53 561

FIGURE 1 – Transactivity of SSS mutations for 7 distinct p53-binding sequences in yeast. The transactivity was shown as a relative value to the activity of wild-type p53. Transactivity of mutant p53 (G245S) was rescued by SS mutations, H178Y and S183P. Columns, means of 3 independent experiments; bars, SD.

polyacrylamide gel electrophoresis. The expressed p53 was Results detected by immunoblot analysis using an HRP-conjugated anti- Introduction and screening of SS mutations identified human p53 antibody, p53 (FL-393) HRP, (Santa Cruz Biotechnol- as SSS mutations ogy), a HRP-conjugated goat polyclonal anti-HA antibody (Bethyl We used yeast-based transcriptional assays to evaluate mutant Laboratories) or an anti-HRP-conjugated goat polyclonal anti- p53 function by yeast HIS3-based growth reporter assay26 or by Myc antibody (Bethyl Laboratories). The endogenous actin 25 expression was also detected by using an HRP-conjugated rabbit GFP-based fluorescent reporter assay. The latter reporter assay polyclonal antiactin antibody (Sigma). The p53 proteins were was useful because it is quantitative and straightforward (Supple- visualized and quantitatively analyzed using an ECL plus West- mentary Fig. 2). We selected 10 tumor-derived common TP53 ern-blotting detection system (Amersham Biosciences) and a mutations, V157F, R175H, C176F, G245S, R248Q, R248W, lumino image analyzer (LAS1000, Fuji Film). R249S, R273C, R273H and R282W, as background mutations. 562 OTSUKA ET AL. These mutations were functionally inactive and had no or signifi- cantly reduced transactivity for different p53-binding sequences.15 To prepare SS mutations against the background mutations, TP53 cDNA fragments corresponding to codons 110–225 and codons 192–354 were PCR-amplified from our comprehensive missense mutation library.15 These fragments theoretically contain 576 and 960 mutations, respectively, and 562 (97.6%) and 888 (92.5%) mutations were subsequently amplified, respectively. The former PCR fragments were introduced into the p53 expression vector with G245S, R248Q, R248W, R249S, R273C, R273H or R282W. The latter PCR fragments were introduced into the p53 expression vector with V157F, R175H or C176F. These processes were achieved in vivo in yeast cells by a gap-repair assay (see Materials and methods section). The total number of background mutations and SS mutations was theoretically 6,598. The resulting yeast transformants expressing p53 with one of the background mutations and a SS mutation in a p53 molecule were screened for fluorescent intensity, a marker of transactivation ability. We first identified 25 transformants expressing candidate SSS mutations under the observation of fluorescent microscopy with the naked eye. These 25 transformants were chosen with the naked eye, so some of these could be the false positive transformants. Then, we carried out the quantitative measurement using the fluorometer. Among these mutations, 10 transformants, A159T-G245S, H178Y-G245S, C182F-G245S, S183P-G245S, D184A-G245S, D184N-G245S, D186N-G245S, G187S-G245S, L188R-G245S and H178Y-R273H, fulfilled the selection criteria of SSS mutations in yeast (see Materials and methods section) (Fig. 1 and Sup- plementary Fig. 3a). The p53 (S183P-G245S) reactivated the G245S transactivity in all 8 p53-binding sequences. The p53 (H178Y-G245S) reactivated the transactivity of G245S in 5 p53- binding sequences (p21WAF1, p53AIP1, GADD45, Noxa, p53R2). The remaining 8 mutations fulfilled the criteria only for the p53R2-derived p53-binding sequence.

SSS mutations isolated in yeast were also SSS in human cells To conﬁrm whether the 25 candidate suppressor mutations isolated by the yeast are also suppressor mutations in human cells, we constructed 24 mammalian p53 expression vectors with both background and candidate suppressor mutations (we also had constructed the remaining one vector, but it had an additional mutations, so we excluded this one) and examined their ability to transactivate a p53-responsive luciferase gene. Twelve mutations, A159T-G245S, H178Y-G245S, C182F-G245S, C182Y-G245S, S183P-G245S, D184A-G245S, D184G-G245S, D184N-G245S, D184V-G245S, D186N-G245S, L188R-G245S and H178Y- R273H, fulﬁlled the selection criteria of SSS mutations in human cells (see Materials and methods section) (Fig. 2 and Supplemen- tary Fig. 3b). The mutant p53 (H178Y-G245S) reactivated the transactivity of G245S in all six p53-responsive promoters (p21WAF1, MDM2, BAX, 14-3-3r, GADD45 and p53R2). The mutant p53 (C182F-G245S) reactivated the transactivity of G245S in 4 p53-binding sequences (p21WAF1, MDM2, GADD45 and p53R2). The remaining 10 mutations reactivated G245S in only 1 (p53R2)or2(p21WAF1and p53R2) promoters.

Coexpression of p53 with H178Y rescued the inactivated FIGURE 2 – Transactivity of SSS mutations for 5 distinct p53-bind- transactivation function of p53 with G245S ing sequences in Saos-2 cells. The transactivity was shown as a rela- To verify whether the p53 (G245S) phenotype is rescued when tive value to the activity of wild-type p53. Transactivity of mutant p53 coexpressed with p53 (H178Y), we coexpressed p53 (H178Y) and (G245S) was rescued by SS mutations, H178Y, C182F, S183P, p53 (G245S), and compared their ability to transactivate luciferase D184A, D184G and D186N for 5 p53-binding sequences. Columns, reporters with that of p53 (H178Y-G245S) in Saos-2 cells. After values of 2 independent experiments. 24 hr of transfection, the amount of expressed p53 proteins was almost identical to wild-type p53 (Fig. 3a). To exclude some pos- moters (Fig. 3b). These results indicated that the p53 (G245S) sibility that increased transactivation can be explained by the phenotype is rescued when coexpressed with p53 (H178Y). increased expression level, we adjusted the transactivation levels To see whether wild-type p53 can restore the G245S mutation by the net intensities of the protein expression levels (Fig. 3b). when they are coexpressioned, we carried out the additional coex- Consistent with the result shown in Figure 2, the transactivity of presssion experiment (Supplementary Fig. 4). In the case of coex- p53 (H178Y-G245S) surpassed that of p53 (G245S) for all 6 pro- pression of wild-type p53 and G245S mutant, the transactivity is SECOND-SITE SUPPRESSOR MUTATIONS OF p53 563

FIGURE 3 – Intertermolecular rescue of mutant p53 (G245S) function by another coexpressed mutant p53 (H178Y). The transactivity was shown as a relative value to the activity of wild-type p53. (a) p53 expression with wild type (WT), G245S, H178Y, H178Y-G245S and the coexpression of H178Y and G245S (H178Y 1 G245S) in Saos-2 cells. The total amount of plasmid DNA used for transfection was the same in each lane. The net intensities of each lanes were calculated by the 1D Image Analysis Software (Kodak Digital Science) and indicated as a relative value of wild-type p53. (b) Rescue of G245S transactivity by coexpressed H178Y was shown in all but the p53R2 promoter. We adjusted the transactibvation levels by the protein expression levels indicated by the net intensities. Columns, means of 3 independent experiments; bars, SD. not far beyond the sum of the transctivities of wild-type p53 and SSS mutations restore the background mutation are similar to each that of G245S mutant. So, we consider that it could not be said other (Supplementary Fig. 5). that wild-type p53 can restore the G245S mutation intermolecu- larly so as the case of wild-type p53 and H178Y mutant. We also tested other 5 p53 mutants, C182F, S183P, D184A, Rescue of apoptotic function of mutant p53 (G245S) D184G, D186N, if they also act as SSS when coexpressed with by coexpressed mutant p53 (H178Y) p53 (G245S). Interestingly, these 5 mutants also rescued the To examine whether the coexpression of the mutant H178Y G245S phenotype though it is not clear that the mechanism these with G245S recovers the inactivated ability of the mutant G245S 564 OTSUKA ET AL. Functional rescue of p53 with G245S by p53 with H178Y required tetramer formation The p53 protein forms a homotetramer through its tetramerization domain in the COOH-terminal domain when it binds to spe- ciﬁc DNA sequences.3 Therefore, the rescue of the mutant p53 (G245S) phenotype when coexpressed with p53 (H178Y) may require tetramerization of the p53 proteins. To conﬁrm this possi- bility, we introduced a missense mutation L344P into the tetramerization domain. The L344P is a loss of function mutation27 and has been shown to disrupt both the dimer and tetramer formation of wild-type p53.28,29 The L344P is predicted to destroy the center of the hydrophobic core of the 4 a-helix bundles of the domain.3,15 p53 expressed in Saos-2 cells was cross-linked by glutaraldehyde and separated by SDS-PAGE (Supplementary Fig. 7a). Similar to wild-type p53, mutant p53s with an intact tetramerization domain form a tetramer in Saos-2 cells. In contrast, mutant p53s with L344P failed to form a dimer and tetramer as expected. A mutant p53 expression vector and another mutant p53 vector or p53-null control vector were cotransfected into Saos-2 cells and the induced luciferase activities were measured (Supplementary Fig. 7b). The introduction of L344P into p53 (H178Y), p53 (G245S) and p53 (H178Y-G245S) inactivated their original transactivity (lanes 5, 8 and 10). L344P also inactivated the functional rescue of the coexpression of G245S by H178Y (lanes 12, 13 and 14). We therefore concluded that the functional rescue of p53 (G245S) by p53 (H178Y) requires tetramer formation.

Computer-based structural simulation predicted the inter-p53 interaction of H178Y with G245S To analyze the possible structural mechanism of H178Y modifi- cation of the inactivated G245S mutant structure, we introduced G245S and H178Y into 1TUP (PDB), a known p53 structure resolved by X-ray crystallography,2 using the MUTATE function of the Swiss-PDB Viewer.30 When H178Y was introduced into the 1TUP ‘‘A’’ molecule of p53 (G245S) and the lowest score (indicates the most stable) of the rotamer of the tyrosine side chain was selected, the H178Y (A) side chain was not directed toward the Zn finger and was located far from the G245S residue (Supple- FIGURE 4 – Rescue of apoptotic function of mutant p53 (G245S) by coexpressed mutant p53 (H178Y). (a–e) Representative DNA histo- mentary Fig. 8a). Therefore, the tyrosine substitution at the H178 gram of the Saos-2 cells expressing null, wild-type, mutant p53 in the same p53 molecule may not provide a striking effect in the (G245S), mutant p53 (H178Y) or both mutant p53 (G245S) and local p53 (G245S) structure. In contrast, when H178Y was intro- mutant p53 (H178Y). (f) Apoptotic fraction of the transfection experi- duced into the ‘‘B’’ molecule of the p53 1TUP and the lowest ments was estimated as DsubG1 fraction (see Materials and methods rotamer score of tyrosine was defined, it was clear that the H178Y section) by FACS analysis. Columns, values of 2 independent experi- (B) was very close to the G245S (A) and that the H178Y makes a ments. novel hydrogen-bond with the G245S (A) and/or the R249 (A). Direct interaction between the G245S (A) and the H178Y (B) was to induce apoptosis, FACS analysis of DNA histogram was per- observed when we used alternative software, ICM 3.0 (Supple- mentary Figs. 8b and 8c). The results of the computer-based simu- formed and average DsubG1 was calculated after 2 independent FACS analyses (see Materials and methods section). As shown in lation predicted that the mutant p53 (G245S) would be reactivated by another mutant p53 (H178Y) in an intermolecular manner Figure 4, DsubG1 values of the 2 mutants were lower than that of wild-type p53. When the 2 mutant p53s were coexpressed, the rather than an intramolecular mechanism. DsubG1 value was enhanced and comparable to or higher than that of wild-type p53. These results indicated that impaired apoptotic function of the 2 mutant p53s was rescued by each other. Discussion According to our search of previous publications, 78 distinct combinations of intragenic suppressor mutations of the p53 The p53 tetramer contained both p53 (G245S) and p53 (H178Y) mutants have been reported.18–23 Those SSS mutations were iden- To confirm whether the p53 tetramer contained both p53 tified from randomly mutagenized cDNA and, therefore, the exact (G245S) and p53 (H178Y) when they were expressed simultane- percentage of the SS mutations was not evaluated. We theoreti- ously, HA-tagged p53 (G245S) and Myc-tagged p53 (H178Y) cally screened 6,598 combinations of SS mutations and the top 10 were coexpressed in Saos-2 cells (Supplementary Fig. 6a). The tumor-derived mutations, and finally 12 (0.2%) SS mutations (9 HA-tagged p53 (G245S) and the Myc-tagged p53 (H178Y) were and 11 SS mutations in yeast and mammalian cells, respectively) immunoprecipitated by both an anti-HA antibody and an anti-Myc were confirmed as SSS mutations. This frequency is an underesti- antibody, indicating that these 2 mutant p53s interact each other mate and does not represent all SSS mutations because our study (Supplementary Fig. 6b). The cross-linked p53 tetramer precipi- screened only SS mutations relatively distant from background tated by the anti-HA antibody and the anti-Myc antibody con- mutations (16–197 residues) on TP53 cDNA and did not screen tained both the HA-tagged p53 (G245S) and the Myc-tagged p53 those relatively close to the background mutations. This was at (H178Y) (Supplementary Fig. 6b), indicating that the p53 tetramer least one reason why some of the mutations reported by others contained both the mutant p53s. could not be detected in our assay. SECOND-SITE SUPPRESSOR MUTATIONS OF p53 565 In this study, all of the SSS mutations but one had a specific tary Fig. 3). Recent NMR studies in solution revealed that H1 he- background mutation, G245S. There are various possible explana- lix (residues 173–182) is localized on the surface of the p53 core tions for this observation: First, G245S is functionally inactive but domain and forms the core:core domain interface in the p53 DNA retains weak DNA-binding activity under specific conditions such complex.32,33 H178 is one of the 3 solvent-exposed residues (resi- as 20°C and the presence of heparin,31 and is therefore pre- dues 177, 178, 181), which were predicted to participate in pro- ferentially restorable. In fact, 8 SSS mutations (F113L, tein–protein interactions.34 On the other hand, G245 is the neigh- L114V1T123P1V172I1A189V, T123P, T123P1A189V, S227P1 boring residue of G244. G244 region is predicted as a possible N239Y, N239Y, S240N and N268D) have been reported so far. dimerization interface.33 So, we speculate the conformational dis- Second, among the 10 background mutations, 7 mutations were tortion of dimerization interface of p53 (G245S) monomer is ‘‘from arginine’’ mutations. In p53, arginine residues at hot-spot restored by p53 (H178Y) monomer when they form dimer each mutations directly bind to DNA or contribute to the surrounding other, and it enables mutant p53 to recover its function. This spec- structures through van der Waals interactions, hydrogen bonds or ulation was biochemically confirmed by the fact the inactivated cross-links by their guanidinium moieties. Disruption of the com- p53 (G245S) transactivation and apoptotic function was restored plex structures may be difficult to restore. In contrast, a glycine to by p53 (H178Y) expressed from the independent plasmid. We serine substitution at residue 245 may have minimal effect in the showed that the p53 tetramer contained both p53 (G245S) and p53 local structure because these 2 residues are small. All SSS muta- (H178Y) (Supplementary Fig. 6b), so we consider core–core inter- tions but A159T (S4 strand) are located in the L2 loop. As both L2 action could be occurred actually. However, under the existence and L3 loops provide 4 zinc-finger residues, C176 and H179 from of L344P, mutant p53s with G245S and/or H178Y mutation failed the L2 loop, and C238 and C242 from the L3 loop, one possible to neither form a dimer and tetramer nor recover their transactiva- explanation is that G245S may partially destroy the zinc finger tion function (Supplementary Fig. 7). SSS mutations have also structure by local structural torsion in the L3 loop, and SSS muta- been identified in many proteins other than p53, such as staphylo- tions in the L2 loop can restore the local structure and reorganize coccal nuclease,35 E. coli ribonuclease HI,36 metal-tetracyclin/H1 the zinc-finger structure. antiporter,37 the photosynthetic reaction center of Rhodobacter capsulatus,38 and the lactose carrier of E. coli,39 E. coli ATP syn- The p53 function recovered by the SSS mutations clearly 40 depended on differences in p53BS. This was not surprising thase. The molecular mechanisms of SSS were considered only because we have shown that some of the p53 mutants (635 of in an intragenic and intramolecular manner. Another type of SSS 2,314 mutants: 27.5%) differed in transactivity spectrum in differ- mutation is extragenic/intermolecular SSS found in a hetero- ent p53BSs,15 and many ts mutants differed in the ts transactivity oligomer molecule, e.g., the enzyme activity of E. coli ATP syn- 16 thase destroyed by G213N in the (i) subunit was restored by D61E spectra in different p53BSs. We again speculate that there are 41 subtle differences in structural alterations caused by specific muta- in the (ii) subunit. When a protein function acts as a homo- tions themselves, distinct temperatures and SSS mutations, and oligomer, we should consider that an intragenic SSS mutation that such alterations are responsible for the partial inactivation or theoretically restores the function not only by intramolecular reactivation of p53 binding to the distinct DNA sequences. Most mechanism but also by intermolecular mechanism. H178Y against of the SSS mutations against G245S were identified through the G245S in the p53 homo-oligomer described here is the first exam- p53BS of the p53R2 gene. This was mainly because the transactiv- ple. Our computer-based simulation model may provide one of the ity of G245S for the p53BS of the p53R2 gene was relatively speculations. It showed that H178Y of one p53 molecule in the higher and distinct criteria on SSS isolation were adopted. There- homotetramer directly interacted with G245S of one of the other fore, it may be assumed that the result for the p53R2 gene was p53 molecules including a hydrogen bonding, and that the mutant overestimated. However, we confirmed that most mutations were p53 (G245S) could be reactivated by another mutant p53 (H178Y) also SSS in human cells, indicating that the criteria for the p53R2 in an intermolecular mechanism rather than an intramolecular were rational and the structural effect of SSS mutations is similar manner (Supplementary Fig. 8). in the 2 cell systems. So, we consider that the yeast system is fea- Our results suggested that intermolecular interaction not only sible for screening SSS mutations. through the tetramerization domain but also through the DNA- Because the relative fluorescent intensity of G245S to that of binding domain was important for the functional assembly of wild-type wais higher in yeast cells than in human cells we applied the p53 homotetramer and disruption of the latter type of inter- a different criteria for SSS selection, (Mb1s/Mb > 5) for yeast action by a specific mutation (such as G245S) might be a p53- cells and (Mb1s/Mb > 1.5) for human cells. The reasons the strin- inactivating mechanism in some tumors potentially restored in gency for SSS selection in yeast seems higher is both the reporter an intermolecular manner. The results indicated here may pro- systems used differ and the frequency of false positives high in the vide novel strategies for restoring inactivated p53 function and yeast screen. In addition, at the first yeast screening, we picked up tumor suppression and offer a clue in the search for small mol- the yeast candidate clones by our naked eyes. But if their fluores- ecules that restore the inactivated function by common p53 cence intensities did not reach at about more than 0.3 times as that mutations. of wild-type, we could not detect it as the SSS candidates. So, we Acknowledgements apply a criteria (Mb1s/W > 0.3) for SSS selection in yeast. To predict the structural mechanism underlying the functional We thank Shuang-Yin Han and Wen Liu for their contribution restoration by SSS mutations, we selected the H178Y mutation as on the mutant p53 library. We also thank Yuka Fujimaki for tech- a representative SSS mutation because H178Y had the most potent nical assistance. This study was supported in part by Ministry of suppressor activity against G245S among the isolated SSS muta- Education, Science, Sports and Culture (C. Ishiokca, S. Kato and tions in both yeast and human cells (Figs. 1 and 2, and Supplemen- K. Otska and the Gonryo Medical Foundation (C. Ishioka)).

References 1. Arrowsmith CH, Morin P. New insights into p53 function from structure of the oligomerization domain of p53: correction. Science tural studies. Oncogene 1996;12:1379–85. 1995;267:1515–16. 2. Cho Y, Gorina S, Jeffrey PD, Pavletich NP. Crystal structure of a p53 4. Fields S, Jang SK. Presence of a potent transcription activating tumor suppressor-DNA complex: understanding tumorigenic muta- sequence in the p53 protein. Science 1990;249:1046–9. tions. Science 1994;265:346–55. 5. Raycroft L, Wu HY, Lozano G. Transcriptional activation by wild- 3. Clore GM, Omichinski JG, Sakaguchi K, Zambrano N, Sakamoto H, type but not transforming mutants of the p53 anti-oncogene. Science Appella E, Gronenborn AM. Interhelical angles in the solution struc- 1990;249:1049–51. 566 OTSUKA ET AL.

6. Jeffrey PD, Gorina S, Pavletich NP. Crystal structure of the tetrameri- 25. Shimada A, Kato S, Enjo K, Osada M, Ikawa Y, Kohno K, Obinata zation domain of the p53 tumor suppressor at 1.7 angstroms. Science M, Kanamaru R, Ikawa S, Ishioka C. The transcriptional activities of 1995;267:1498–502. p53 and its homologue p51/p63: similarities and differences. Cancer 7. Soussi T, Caron de Fromentel C, May P. Structural aspects of the p53 Res 1999;59:2781–6. protein in relation to gene evolution. Oncogene 1990;5:945–52. 26. Ishioka C, Frebourg T, Yan YX, Vidal M, Friend SH, Schmidt S, Iggo 8. Kaghad M, Bonnet H, Yang A, Creancier L, Biscan JC, Valent A, R. Screening patients for heterozygous p53 mutations using a func- Minty A, Chalon P, Lelias JM, Dumont X, Ferrara P, McKeon F, tional assay in yeast. Nat Genet 1993;5:124–9. et al. Monoallelically expressed gene related to p53 at 1p36, a region 27. Ishioka C, Englert C, Winge P, Yan YX, Engelstein M, Friend SH. frequently deleted in neuroblastoma and other human cancers. Cell Mutational analysis of the carboxy-terminal portion of p53 using both 1997;90:809–19. yeast and mammalian cell assays in vivo. Oncogene 1995;10:1485– 9. Osada M, Ohba M, Kawahara C, Ishioka C, Kanamaru R, Katoh I, 92. Ikawa Y, Nimura Y, Nakagawara A, Obinata M, Ikawa S. Cloning 28. Ishioka C, Shimodaira H, Englert C, Shimada A, Osada M, Jia LQ, and functional analysis of human p51, which structurally and func- Suzuki T, Gamo M, Kanamaru R. Oligomerization is not essential for tionally resembles p53. Nat Med 1998;4:839–43. growth suppression by p53 in p53-deficient osteosarcoma Saos-2 10. Greenblatt MS, Bennett WP, Hollstein M, Harris CC. Mutations in cells. Biochem Biophys Res Commun 1997;232:54–60. the p53 tumor suppressor gene: clues to cancer etiology and molecular 29. Kawaguchi T, Kato S, Otsuka K, Watanabe G, Kumabe T, Tominaga pathogenesis. Cancer Res 1994;54:4855–78. T, Yoshimoto T, Ishioka C. The relationship among p53 oligomer for- 11. Hollstein M, Rice K, Greenblatt MS, Soussi T, Fuchs R, Sorlie T, mation, structure and transcriptional activity using a comprehensive Hovig E, Smith-Sorensen B, Montesano R, Harris CC. Database of missense mutation library. Oncogene 2005;24:6976–81. p53 gene somatic mutations in human tumors and cell lines. Nucleic 30. Guex N, Peitsch MC. SWISS-MODEL and the Swiss-Pdb Viewer: an Acids Res 1994;22:3551–5. environment for comparative protein modeling. Electrophoresis 12. Martin AC, Facchiano AM, Cuff AL, Hernandez-Boussard T, Olivier 1997;18:2714–23. M, Hainaut P, Thornton JM. Integrating mutation data and structural 31. Bullock AN, Henckel J, Fersht AR. Quantitative analysis of residual analysis of the TP53 tumor-suppressor protein. Hum Mutat 2002; folding and DNA binding in mutant p53 core domain: definition of 19:149–64. mutant states for rescue in cancer therapy. Oncogene 2000;19:1245– 13. Olivier M, Eeles R, Hollstein M, Khan MA, Harris CC, Hainaut P. 56. The IARC TP53 database: new online mutation analysis and recom- 32. Klein C, Planker E, Diercks T, Kessler H, Kunkele KP, Lang K, Han- mendations to users. Hum Mutat 2002;19:607–14. sen S, Schwaiger M. NMR spectroscopy reveals the solution dimeri- 14. Soussi T, Dehouche K, Beroud C. p53 website and analysis of p53 zation interface of p53 core domains bound to their consensus DNA. gene mutations in human cancer: forging a link between epidemiol- J Biol Chem 2001;276:49020–7. ogy and carcinogenesis. Hum Mutat 2000;15:105–13. 33. Rippin TM, Freund SM, Veprintsev DB, Fersht AR. Recognition of 15. Kato S, Han SY, Liu W, Otsuka K, Shibata H, Kanamaru R, Ishioka DNA by p53 core domain and location of intermolecular contacts of C. Understanding the function-structure and function-mutation rela- cooperative binding. J Mol Biol 2002;319:351–8. tionships of p53 tumor suppressor protein by high-resolution missense 34. Walker DR, Bond JP, Tarone RE, Harris CC, Makalowski W, mutation analysis. Proc Natl Acad Sci USA 2003;100:8424–9. Boguski MS, Greenblatt MS. Evolutionary conservation and somatic 16. Shiraishi K, Kato S, Han SY, Liu W, Otsuka K, Sakayori M, Ishida T, mutation hotspot maps of p53: correlation with p53 protein structural Takeda M, Kanamaru R, Ohuchi N, Ishioka C. Isolation of tempera- and functional features. Oncogene 1999;18:211–18. ture-sensitive p53 mutations from a comprehensive missense mutation 35. Shortle D, Lin B. Genetic analysis of staphylococcal nuclease: identi- library. J Biol Chem 2004;279:348–55. fication of three intragenic ‘‘global’’ suppressors of nuclease-minus 17. Kakudo Y, Shibata H, Otsuka K, Kato S, Ishioka C. Lack of correla- mutations. Genetics 1985;110:539–55. tion between p53-dependent transcriptional activity and the ability to 36. Haruki M, Noguchi E, Akasako A, Oobatake M, Itaya M, Kanaya S. induce apoptosis among 179 mutant p53s. Cancer Res 2005;65:2108– A novel strategy for stabilization of Escherichia coli ribonuclease HI 14. involving a screen for an intragenic suppressor of carboxyl-terminal 18. Brachmann RK, Yu K, Eby Y, Pavletich NP, Boeke JD. Genetic deletions. J Biol Chem 1994;269:26904–11. selection of intragenic suppressor mutations that reverse the effect of 37. Yamaguchi A, O’Yauchi R, Someya Y, Akasaka T, Sawai T. Second- common p53 cancer mutations. EMBO J 1998;17:1847–59. site mutation of Ala-220 to Glu or Asp suppresses the mutation of 19. Nikolova PV, Wong KB, DeDecker B, Henckel J, Fersht AR. Mecha- Asp-285 to Asn in the transposon Tn10-encoded metal-tetracycline/ nism of rescue of common p53 cancer mutations by second-site sup- H1 antiporter of Escherichia coli. J Biol Chem 1993;268:26990–5. pressor mutations. EMBO J 2000;19:370–8. 38. Hanson DK, Baciou L, Tiede DM, Nance SL, Schiffer M, Sebban P. 20. Wieczorek AM, Waterman JL, Waterman MJ, Halazonetis TD. Struc- In bacterial reaction centers protons can diffuse to the secondary ture-based rescue of common tumor-derived p53 mutants. Nat Med quinone by alternative pathways. Biochim Biophys Acta 1992;1102: 1996;2:1143–6. 260–5. 21. Wright JD, Noskov SY, Lim C. Factors governing loss and rescue of 39. King SC, Hansen CL, Wilson TH. The interaction between aspartic DNA binding upon single and double mutations in the p53 core do- acid 237 and lysine 358 in the lactose carrier of Escherichia coli. Bio- main. Nucleic Acids Res 2002;30:1563–74. chim Biophys Acta 1991;1062:177–86. 22. Baroni TE, Wang T, Qian H, Dearth LR, Truong LN, Zeng J, Denes 40. Iwamoto A, Omote H, Hanada H, Tomioka N, Itai A, Maeda M, Futai AE, Chen SW, Brachmann RK. A global suppressor motif for p53 M. Mutations in Ser174 and the glycine-rich sequence (Gly149, cancer mutants. Proc Natl Acad Sci USA 2004;101:4930–5. Gly150, and Thr156) in the beta subunit of Escherichia coli H(1)- 23. Joerger AC, Ang HC, Veprintsev DB, Blair CM, Fersht AR. Struc- ATPase. J Biol Chem 1991;266:16350–5. tures of p53 cancer mutants and mechanism of rescue by second-site 41. Kuo PH, Nakamoto RK. Intragenic and intergenic suppression of the suppressor mutations. J Biol Chem 2005;280:16030–7. Escherichia coli ATP synthase subunit a mutation of Gly-213 to Asn: 24. Bullock AN, Fersht AR. Rescuing the function of mutant p53. Nat functional interactions between residues in the proton transport site. Rev Cancer 2001;1:68–76. Biochem J 2000;347:797–805.