Isolation of Temperature-sensitive p53 Mutations from a Comprehensive Missense Mutation Library*

Temperature-sensitive (ts) mutations have been used as a genetic and molecular tool to study the functions of many gene products. Each ts mutant protein may contain a temperature-dependent intramolecular mechanism such as ts conformational change. To identify key ts structural elements controlling the protein function, we screened ts p53 mutants from a comprehensive mutation library consisting of 2,314 p53 missense mutations for their sequence-specific transactivity through p53-binding sequences in Saccharomyces cerevisiae. We isolated 142 ts p53 mutants, including 131 unreported ts mutants. These mutants clustered in β-strands in the DNA-binding domain, particularly in one of the two β-sheets of the protein, and 15 residues (Thr155, Arg158, Met160, Ala161, Val172, His214, Ser215, Pro223, Thr231, Thr253, Ile254, Thr256, Ser269, Glu271, and Glu285) were ts hot spots. Among the 142 mutants, 54 were examined further in human osteosarcoma Saos-2 cells, and it was confirmed that 89% of the mutants were also ts in mammalian cells. The ts mutants represented distinct ts transactivities for the p53 binding sequences and a distinct epitope expression pattern for conformation-specific anti-p53 antibodies. These results indicated that the intramolecular β-sheet in the core DNA-binding domain of p53 was a key structural element controlling the protein function and provided a clue for finding a molecular mechanism that enables the rescue of the mutant p53 function.

p53 tumor suppressor is a 393-amino acid transcription factor that activates the transcription of a number of downstream genes through p53 binding to two copies of the specific consensus DNA sequence 5Ј-RRRC(A/T)(T/A)GYYY-3Ј (in which R is a purine nucleoside and Y is a pyrimidine nucleoside) in their regulatory regions (1). These molecular switches are activated by post-translational modifications, including phosphorylation, acetylation, and prolyl isomerization (2)(3)(4)(5) of p53 in response to genotoxic or non-genotoxic stresses. The resulting biological effects are cell cycle arrest, apoptosis, DNA repair, and angiogenesis (6 -10). A growing number of p53 downstream genes have been isolated, and p53 has been structurally and func-tionally divided into three portions, namely the NH 2 -terminal portion containing the transactivation domain, the central core portion corresponding to the DNA-binding domain, and the COOH-terminal portion containing the oligomerization domain. The evolution of the DNA-binding domain is highly conserved in p53 orthologues (11) and also in the conserved human homologues p63 and p73 (12,13).
The structure of the DNA-binding domain (residues 94 -312) was resolved by x-ray crystallography (14). The domain consists of two ␣-helixes (H1 and H2) and 11 ␤-strands (S1, S2, S2Ј, and S3-S10) that were interconnected by loops (long L1-L3 loops and other short loops). Two anti-parallel ␤-sheets containing four (S1, S3, S5, and S8) and five (S4, S6, S7, S9, and S10) ␤-strands make up a large ␤-sandwich that serves as a scaffold for a loop-sheet-helix (LSH) motif (L1, S2, S2Ј, S10, and H2) and two large loops (L2 and L3). The loop-sheet-helix consists of two separate regions as follows: (i) the L1 loop (residues 113-123) and the S2-S2Ј ␤-hairpin (residues 124 -135) that correspond to evolutionary conserved region II (residues 117-142) (11); and (ii) the end of the S10 strand (residues 264 -274) and the H2 helix (residues 278 -286) that correspond to conserved region V (residues 270 -286). In the loop-sheethelix, the L1 loop and the H2 helix contact with a DNA major groove formed by the RRRC region of the consensus sequence. One of the large loops, the L2 (residues 164 -194), is interrupted by a short helix (H1) and contains conserved region III (residues 171-181). Another large loop, L3 (residues 237-250), coincides with conserved region IV (residues 234 -258) and makes contact with the DNA minor groove formed by the A/T rich region of the consensus sequence. The L2 loop stabilizes the L3 loop by packing through a side-chain interaction and a zinc atom tetrahedrally coordinated on residues Cys 176 , His 179 of the L2 loop and Cys 238 and Cys 242 of the L3 loop.
Mutations in the TP53 gene are the most frequent genetic alterations in the various human tumors (15). According to the latest TP53 mutation databases (16,17), more than 15,000 somatic mutations have been reported to date. The mutations are clustered in the DNA-binding domain, and the majority (ϳ80%) are missense mutations. Among tumor-derived mutations, those at residues Arg 175 , Gly 245 , Arg 248 , Arg 249 , Arg 273 , and Arg 282 have frequently been reported, and all missense mutations were unable to bind the specific p53 binding sequences and the inactive transactivation for downstream genes. These are structurally important residues, because they directly involve DNA binding or stabilization of the L2 and L3 loops of the protein. However, the majority of remaining missense mutations have not yet been examined. Recently, we constructed 2,314 missense mutations that covered almost all of the tumor derived missense mutations, as well as a number of previously unreported missense mutations, and examined their ability to transactivate marker genes through distinct p53 binding sites when the mutants were expressed in yeast. We determined the functional effect of each mutant p53 and found that the p53 function correlated well with the structure and mutations (18).
Temperature-sensitive (ts) 1 p53 mutations have been reported and used as tools for conditional p53 expression in mammalian cells. We identified previously four distinct ts p53 mutations in eight of the 91 human tumor cell lines using a yeast-based transcription assay and predicted that 5-10% of the tumor-derived missense mutations should be ts mutations (19). To date, 61 p53 ts mutations have been isolated by using several different methods, including a yeast-based functional assay (Table I). Among these, the V272M ts mutant was reactivated by a small molecule, aminothiol WR1065 (20), at a non-permissive temperature, suggesting that ts mutants may be functionally rescued by small molecules.
The purpose of this study was the screening and isolation of a large number of ts mutations from a comprehensive missense mutation library, mapping them to the p53 structure, and considering the function-structure relationship through the ts mutants. To isolate a number of ts p53 mutations, we screened the p53 library containing 2,314 p53 missense mutations using a yeast-based p53 functional assay and found 142 ts p53 mutants, including previously unreported 131 mutants. We confirmed that most were also ts in p53-less mammalian cells. The ts mutants were preferentially mapped on one of the ␤-sheets, and there were hot spot sites for ts mutations. Because a fairly significant fraction of the p53 mutants in the TP53 mutation databases were ts mutants, these ts p53 mutant proteins may be novel molecular targets through the ts mechanism and structure-dependent restoration of p53 function.
Screening ts p53 Mutants Using a Yeast Assay-The 2,314 yeast clones expressing the mutant p53 were grown on 25 96-well formatted plates containing synthetic complete (SC) media lacking leucine and tryptophane (SC ϪLeu ϪTrp) in the case of the haploid strains, or SC media lacking leucine, tryptophane and histidine (SC ϪLeu ϪTrp ϪHis) in the case of the diploid strains.
Fluorescent Intensity-To evaluate the transactivity of each mutant p53 quantitatively, the yeast clones (haploid cells) were replicated on SC ϪLeu ϪTrp solid media using a 96-pin replicator and grown at 37 or 32°C for 2 days. The plates were then directly processed in a 96-well formatted fluorometer (Fluoroskan Ascent FL, Labsystems) to measure the fluorescent intensity (excitation, 485 nm; emission, 538 nm) of p53-dependent enhanced green fluorescent protein expression through a human p21 WAF1 -derived p53 binding sequence. The diploid cells, selected by mating reaction, were incubated on SC ϪLeu ϪTrp ϪHis plates at 37 or 30°C for 2 days, and the fluorescent intensity of Ds-Red was measured using the same fluorometer (excitation, 544 nm; emission, 590 nm) to evaluate the p53-dependent Ds-Red expression through other p53-binding sequences. At least two independent experiments were performed for each reporter, and the fluorescence intensities were averaged. The averaged values were standardized in each p53 binding sequence, clustered, and visualized using the CLUSTER and TREEVIEW programs. The standardized data were also spotted on a two-dimensional graph for 30 and 37°C. We defined the following criteria to select ts mutants from the p53 mutant library, namely M 30  indicate the fluorescent intensities of the p53 mutants at 30 and 37°C, respectively, and W 30 and W 37 indicate the fluorescent intensities of the wild-type p53 at 30 and 37°C, respectively.
Luciferase Assay-After 24 h of transfection, luciferin (Steady-Glo luciferase assay system, Promega), a substrate of luciferase, was added to the culture media and further incubated for 60 -120 min according to the manufacturer's instructions. The fluorescent intensity was measured using the Fluoroskan Ascent FL (see above). The relative fluorescent intensity to the wild-type control was calculated from three sets of independent experimental data at 32 and 37°C. The value differences at the two temperatures were statistically evaluated by t test. The ts mutants were defined when the p value was Ͻ0.001.
Drawing p53 Peptide Structures-To map the ts p53 mutants on the p53 core domain, the NCBI structure file, 1TUP, was customized for our purpose and visualized using Cn3D 4.0 software (22).

Clustering of 2,314 Mutations on Transactivities at Two Distinct
Temperatures-An unsupervised, hierarchical one-dimensional cluster analysis allowed us to cluster the 2,314 p53 mutants on the basis of similar measured transactivities for eight distinct p53 binding sequences (p53 binding sites) at 30 and 37°C (Fig. 1A). The mutants are divided into two major clusters. In one of these clusters the mutants retain transactivities; in the other they lose activity, and these clusters are mostly temperature-independent. Notably, there is one temperature-dependent sub-cluster within the latter cluster (Fig. 1B). The cluster consists of 64 p53 mutants, and the transactivities of the mutants are inactive on almost all p53 binding sites at 37°C but active on some p53 binding sites at 30°C, indicating that a large number of mutants are ts for transactivation in yeast cells.
Isolation of ts p53 Mutants in Yeast-Although the cluster analysis found the typical ts mutants that represent temperature sensitivity for most p53 binding sites, there are mutants that show temperature sensitivity on limited types of p53 binding sites and, therefore, are not clustered. To also isolate such clones, the transactivities of the 2,314 mutant clones at 30 and 37°C were standardized and overviewed by a scatter plot for each p53 binding site (Fig. 2). Among the 18,512 data points (8 ϫ 2,314 clones), the majority had similar transcriptional activity (either active or inactive) at both 30 and 37°C, indicating that they were not ts. Obviously, there were significant numbers of p53 mutant clones that represented higher transactivity at 30°C than at 37°C, showing ts mutants for the transactivation function (circled spots in Fig. 2). On the other hand, only a limited number of clones represented higher transactivity at 37°C than at 30°C, showing cold-sensitive mutants. As there is no clear boundary between ts and non-ts mutants, we defined the borders for convenience as described under "Experimental Procedures." According to the definition, 142 p53 mutants were selected as ts for yeast transactivation assay (Fig. 3A), indicating that 6.1% (142 of 2,314) of the p53 mutants were ts for at least one of the p53 binding sites. The 142 mutants, including 131 previously unreported ts mutants, a The meaning of the asterisk symbols used in this column is as follows: *, ts mutants not constructed in this study; **, ts mutants also isolated in this study; ***, distinct substitution(s) at the same residue were ts mutants in this study.
b The meaning of the numbers used in this column is as follows: 1, yeast system; 2, mammalian cell system; 3, cell-free system. c All but three of the gene names used in this column refer to those used in the Online Mendelian Inheritance in Man (OMIM) site (www.ncbi.nlm.nih.gov/entrez/query.fcgi?dbϭOMIM). The three exceptions are: CON, p53-binding consensus sequence; GAL4, yeast GAL4binding sequence; and RGC, human ribosomal gene cluster sequence. The study on the GAL4 was performed by GAL4-binding domain and p53 fusion protein.
Evaluation of the ts p53 Mutants in Mammalian Cells-To evaluate whether the isolated p53 mutants in yeast were also ts for sequence-specific transactivation in mammalian cells, we randomly chose 54 p53 mutants from the 142 ts mutant p53 cDNA clones (Fig. 3B), and constructed expression vectors for mammalian cell experiments. Each mutant p53 was expressed in a p53-deficient human osteosarcoma cell line, Saos-2, and examined for the sequence-specific transactivation at both 32 and 37°C by luciferase assay. When the values of the three independent experiments relative to the wild-type p53 at 32°C were significantly (p Ͻ 0.001; t test) different from those at 37°C in at least one of the six promoters (p21 WAF1 , MDM2, BAX, 14-3-3, p53R2, and GADD45), the mutant clone was defined as a ts mutant in mammalian cells. Among the 54 mutants, 48 (89%) were ts mutants in at least one of the six promoters. The results indicated that most ts mutants isolated in the yeast assay are also ts mutants in mammalian cells, suggesting that many of the remaining 88 clones may also be ts mutants in mammalian cells. Among the 48 clones, 16 were ts in all 6 promoters, whereas 32 clones were ts in a limited number of promoters, although many retained weak ts phenotypes for other promoters (data not shown).
Epitope Analyses of the p53 Protein Expressed in Saos-2 Cells Using Conformation-sensitive Antibodies-To examine whether the ts mutants display ts changes in their epitopes against conformation-sensitive antibodies, PAb1620 for wild-type-like conformation and PAb240 for denatured mutant conformation, six randomly selected ts mutants, M160R, H193Y, T211A, P219S, T253I, and V274A, were expressed in Saos-2 cells at both 32 and 37°C. The cell lysates were immunoprecipitated using the two antibodies, detected by Western blot analysis using an HRP-conjugated anti-p53 antibody, and quantitatively analyzed using a lumino-image analyzer. In the case of wild-type p53, the PAb1620 epitope was exclusive, and only a trace of the PAb240 epitope was detected (Fig. 5A). Similar to wild-type p53, the PAb1620 epitope was dominant in R273H, although the PAb240 epitope was also detected. On the other hand, the PAb240 epitope was dominant, and the PAb1620 epitope was less abundant in R175H. The ratios of the epitope expressions of PAb1620 to PAb240 are shown in Fig. 5B. R175H and R273H were not ts because there were no significant differences in the ratios between 32 and 37°C. Among the ts mutants, P219S and T253I showed an obvious ts increase in ratio. The remaining ts mutants showed no change or only a slight change in ratio.

DISCUSSION
Comparing 142 ts p53 Mutants with the Previously Reported p53 Mutant-Among the 142 ts mutants, 131 were previously unreported mutants. In our survey of previous papers, including our own, 61 human ts p53 mutants have been reported (Table I). These obviously include ts mutants not isolated in our system. We speculate that there are two reasons for the discrepancy. First, they were isolated using experimental systems different from those in our study, including a reporter assay for sequence-specific transactivation in mammalian cells, similar yeast assays with different p53 binding sites, an electrophoretic mobility shift assay (EMSA) in a cell-free system, and monitoring changes in structure-sensitive antibody reactivity. Therefore, it is possible that there are many potential ts mutants not isolated by the method adopted in this study. For example, a known ts mutant, V143A, did not appear as ts in the yeast cells because the ts phenotype may be mediated by ts interaction with human ASPP2 (p53BP2), a positive modulator of p53 transactivation (23,24) that does not exist in yeast cells. Obviously, there may be mechanisms not directly affecting p53 binding to DNA. We are now planning to screen such novel ts mutants by using protein-protein interactions that may modify FIG. 3. Panels of ts p53 mutants. A, the ts p53 mutants isolated by yeast-based functional assay. 142 mutants were listed from the NH 2 terminus to the COOH terminus of p53. Filled boxes represent temperature sensitivity that satisfied the defined criteria (see "Experimental Procedures"). B, 54 of 142 ts mutants were examined for ts transactivities in Saos-2 cells. When, according to t test, the luciferase activities at 32 and 37°C were statistically different with a p value Ͻ0.001 or Ͻ0.0001, the corresponding boxes in the panel were colored gray or black, respectively. p53 structure by their post-translational mechanisms. Second, as shown in Fig. 2, there is distinct strength in ts transactivation, and some reported ts mutants have been eliminated from our criteria because of a weak ts phenotype. In fact, several mutants clustered in Fig. 1B were not selected in our defined criteria. We also note that many previously identified ts mutants had a weak ts phenotype in our yeast screening (data not shown).
Promoter Specificity of the ts p53 Mutants-We have shown that several p53 mutants differ in transactivity spectrum in different p53 binding sites (18). Similarly, ts mutants differed in the ts transactivity spectra in different p53 binding sites (Fig. 3). We speculate that there are subtle differences in structural alterations caused by specific mutations and temperatures and that such alterations are responsible for the partial inactivation or reactivation of p53-binding to the distinct DNA sequences. In fact, there are similarities in the transactivity spectra among mutants in the same or contiguous residues (Fig. 3A), suggesting similar structural alterations. In particular, some showed ts in only one or two promoters, suggesting the possible application of such mutants in the conditional transactivation of specific promoters to study p53 downstream gene functions. Various ts transactivity spectra on different p53-responsive promoters were also observed in mammalian cells (Fig. 3B). The promoter selectivity of wild-type p53 by Ser 46 phosphorylation has been shown as the mechanism of p53AIP1 transactivation (25). Overall, from the results of this study and our previous observations (18), we propose that there may be other unknown potential mechanisms determining the promoter selectivity of wild-type p53 on p53 downstream promoters other than the p53AIP1 gene. The ts transactivity against different promoters was similar in part but significantly different between human and yeast cells (data not shown). We speculate that there are several reasons for this discrepancy. First, the p53 binding elements, other than p21 WAF1 and MDM2 used in the yeast study, were three copies of the specific p53-binding elements and differed from the genomic sequences used in the mammalian cell study. Second, the temperature for the identification of ts mutants in yeast was 30°C, whereas it was 32°C in mammalian cells. Third, post-translational modification and the interaction of other proteins may differ in yeast cells and mammalian cells. Finally, FIG. 4. Location of the ts p53 mutants in the core DNA-binding domain. A, precise map of the constructed p53 mutants and ts mutants (white characters in black boxes) in codons from 101 to 300. Italicized characters, original residues observed in wild-type p53; open boxes, four hot spots for ts mutants; asterisk, 15 hot spot residues for ts mutants; bold lines, secondary structures (11 ␤-strands and 2 ␣-helices). B, schematic representation of the core DNA-binding domain (14) and the fraction of the ts mutants in the two ␤-sheet structures. Percentages, fraction of the ts mutants within the constructed mutants in the indicated secondary structures; small numbers, codon numbers showing the NH 2 -terminal and COOH-terminal ends of the indicated secondary structures. C, three-dimensional structure of p53 core DNA-binding domain with double-strand DNA oligonucleotides. Positions of the 10 tumor-derived hot spot residues (left panel) were compared with 15 representative ts hot spot residues (right panel). p53 and the interacting DNA structure were derived from Protein Data Bank file 1TUP (14), and views from the longitudinal axis of DNA are shown using CD3n 4.0 software (22). the criteria to define ts mutants were strict and differed be-ber of the mutations examined showed ts changes in the expression of epitopes. We speculate that most ts mutants partially recovered their structural alteration, but their structure and transactivation function were not completely restored. It will be interesting to examine whether such partial restoration of p53 function is sufficient to suppress tumor formation and/or progression when expressed under physiological conditions.
Frequency of ts Mutants in TP53 Mutation Databases-According to the latest International Agency for Research on Cancer (IARC) data base for tumor-derived somatic mutations (17), 1,135 distinct missense mutations, including 1,066 missense mutations with a single nucleotide substitution, are registered. These mutations have been reported 12,032 times in total. Among them, 10.3% (110 of 1,066) of mutants were thought to be ts mutants, and such ts mutations comprised 10.4% (1,254 of 12,032) of the total number of mutations. Therefore, we conclude that ts p53 mutation is not as rare as it was previously thought to be (19), and it may be a molecular target for the pharmacological rescue of p53 protein.