This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental material
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowReprints and Permissions
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Bruins, W.
Right arrow Articles by Breit, T. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Bruins, W.
Right arrow Articles by Breit, T. M.

 Previous Article  |  Next Article 

Molecular and Cellular Biology, March 2008, p. 1974-1987, Vol. 28, No. 6
0270-7306/08/$08.00+0     doi:10.1128/MCB.01610-07
Copyright © 2008, American Society for Microbiology. All Rights Reserved.

The Absence of Ser389 Phosphorylation in p53 Affects the Basal Gene Expression Level of Many p53-Dependent Genes and Alters the Biphasic Response to UV Exposure in Mouse Embryonic Fibroblasts{triangledown} ,{dagger}

Wendy Bruins,1,# Oskar Bruning,2,# Martijs J. Jonker,2 Edwin Zwart,1 Tessa V. van der Hoeven,2 Jeroen L. A. Pennings,1 Han Rauwerda,2 Annemieke de Vries,1* and Timo M. Breit2

Laboratory of Toxicology, Pathology and Genetics (TOX), National Institute of Public Health and the Environment (RIVM), Bilthoven, The Netherlands,1 MicroArray Department & Integrative Bioinformatics Unit (MAD-IBU), Swammerdam Institute for Life Sciences, Faculty of Science, University of Amsterdam (UvA), Amsterdam, The Netherlands2

Received 31 August 2007/ Returned for modification 16 October 2007/ Accepted 2 January 2008


arrow
ABSTRACT
 
Phosphorylation is important in p53-mediated DNA damage responses. After UV irradiation, p53 is phosphorylated specifically at murine residue Ser389. Phosphorylation mutant p53.S389A cells and mice show reduced apoptosis and compromised tumor suppression after UV irradiation. We investigated the underlying cellular processes by time-series analysis of UV-induced gene expression responses in wild-type, p53.S389A, and p53–/– mouse embryonic fibroblasts. The absence of p53.S389 phosphorylation already causes small endogenous gene expression changes for 2,253, mostly p53-dependent, genes. These genes showed basal gene expression levels intermediate to the wild type and p53–/–, possibly to readjust the p53 network. Overall, the p53.S389A mutation lifts p53-dependent gene repression to a level similar to that of p53–/– but has lesser effect on p53-dependently induced genes. In the wild type, the response of 6,058 genes to UV irradiation was strictly biphasic. The early stress response, from 0 to 3 h, results in the activation of processes to prevent the accumulation of DNA damage in cells, whereas the late response, from 12 to 24 h, relates more to reentering the cell cycle. Although the p53.S389A UV gene response was only subtly changed, many cellular processes were significantly affected. The early response was affected the most, and many cellular processes were phase-specifically lost, gained, or altered, e.g., induction of apoptosis, cell division, and DNA repair, respectively. Altogether, p53.S389 phosphorylation seems essential for many p53 target genes and p53-dependent processes.


arrow
INTRODUCTION
 
UV irradiation initiates cellular stress responses that involve the induction of transcription factor p53, a DNA damage sensor that prevents the accumulation of genetic lesions and thus tumor development. To achieve this, the protein plays an active role in a variety of cellular processes, for instance, cell cycle arrest, DNA repair, apoptosis, and senescence (reviewed in 3 and 39). It functions predominantly by the transcriptional activation of its target genes. Upon UV exposure, p53 halts cell proliferation, allowing cells to repair their DNA damage. However, if a particular cell has an extensive, likely nonrepairable amount of DNA damage, p53 initiates apoptosis to prevent the damaged cell from dividing (35). If these p53-dependent, protective cellular responses are compromised or absent, the accumulation of mutations may lead to genomic instability and finally the development of cancerous lesions.

In nonstressed cells, p53 protein is kept at low levels by means of proteasome-mediated degradation, regulated by ubiquitination. Upon exposure to stress signals, the protein becomes stabilized and activated by posttranslational modifications (6). These p53 protein modifications are rather diverse, as p53 can be phosphorylated, acetylated, ubiquitinated, sumoylated, glycosylated, methylated, and neddylated (2). The most commonly occurring p53 posttranslational modification is phosphorylation.

Different stressors induce specific p53 modifications (1, 5, 33, 37). Most stressors activate more than one kinase, leading to phosphorylation of p53 at multiple sites. For example, in human cells, DNA damage induced by ionizing radiation or UV irradiation results in (de)phosphorylation of at least 14 different phosphorylation sites, i.e., serine residue 6 (Ser6), Ser9, Ser15, Ser20, Ser33, Ser37, and Ser46 plus threonine 18 (Thr18) and Thr81 in the amino-terminal region; Ser149, Thr150, and Thr155 in the central core; and Ser315 and Ser392 in the C-terminal domain. Interestingly, the most commonly used stressors, UV irradiation and gamma irradiation, lead to different modifications of p53. To illustrate, phosphorylation of human Ser392 (equivalent to mouse Ser389) is triggered specifically after UV irradiation but not after gamma irradiation (27, 38).

The role and significance of p53 phosphorylation were initially investigated using in vitro model systems. Although these experiments revealed important insights, results were highly contradictory. Later, mouse models with targeted germ line mutations were used to identify the significance of the specific phosphorylation events in vivo (recently reviewed in 22). Taken together, these studies showed that although alterations of amino acids that are involved in posttranslational modifications have a minor impact on p53 functioning compared to p53 mutations identified in human tumors, these sites are definitely necessary for fine-tuning the p53 stress response since, once they are altered, most mutant models showed an affected apoptotic or cell cycle arrest response after exposure to DNA damage.

To investigate the significance of the Ser389 phosphorylation site in the cellular responses to DNA damage, we generated mice with a single point mutation in the p53 gene that resulted in a substitution of a serine with an alanine, the p53.S389A mouse model (9). Cells isolated from p53.S389A mutant mice were partly compromised in their UV irradiation-induced, p53-regulated apoptosis, whereas gamma irradiation-induced responses were not affected (9). In addition, this mutant mouse model displayed increased sensitivity to UV-induced, skin- and 2-AAF-induced urinary bladder tumor development. This clearly demonstrates the importance of Ser389 phosphorylation for the tumor-suppressive function of p53 (9, 19). The impact of Ser389 phosphorylation on the role of p53 functioning as a transcription factor has not yet been established. For this, we have recently used microarray technology for genome-wide transcriptome analysis of the cellular processes underlying the 2-AAF-induced cancer-prone phenotype in urinary bladder tissue in vivo (8). We identified delayed gene activation after exposure to 2-AAF of a number of p53 target genes involved in apoptosis and cell cycle control. So, it was possible to detect the effects of absence of p53.S389 phosphorylation on gene activation in vivo.

In this study we used UV as a DNA-damaging agent to investigate the role of p53.S389 phosphorylation in stress responses. UV irradiation induces DNA damage to cells predominantly in the form of pyrimidine dimers and 6-4 photoproducts. These lesions are repaired by the nucleotide excision repair system (42, 51). The response to UV irradiation is complex and involves several pathways (34). More specifically, Fos/Jun and some growth factors are activated within a few minutes after exposure (46). Guo et al. analyzed the primary UV-induced stress responses in HeLa cells by cDNA microarray analysis (16). They identified an "immediate early" UV-C-induced stress response 30 to 60 min after exposure, with increased activation of (p53-independent) genes. Studies of murine embryonic stem cell exposure to DNA-damaging agents, such as UV irradiation, have already demonstrated rapid increases in p53 levels, including posttranslational events resulting in increased transcriptional activity (12, 29, 36). Some p53-dependent genes are regulated upon UV exposure, resulting in apoptosis (53). Thus far, however, the role of p53 phosphorylation in the broad transcriptome response to UV exposure in primary cells has not yet been elucidated. Here, genome-wide transcriptome analysis was performed on wild-type, p53.S389A, and p53–/– mouse embryonic fibroblasts (MEFs) before and after exposure to UV, using an extensive time course analysis. To unravel the role of p53.S389 phosphorylation in the complex UV response in MEFs, we analyzed (i) the effect of absence of p53.S389 phosphorylation on the basal gene expression levels of p53-dependent genes, (ii) the transcriptome response of wild-type MEFs to UV irradiation over time, and (iii) the effect of absence of p53.S389 phosphorylation on the UV response over time. Analysis of the responses at the transcriptome level of p53.S389A MEFs revealed that this p53.S389 phosphorylation site is involved in the regulation of basal expression levels of a large group of (p53-dependent) genes without any imposed exposure, as well as the altered expression levels of a large group of (p53-dependent) genes in response to UV exposure.


arrow
MATERIALS AND METHODS
 
Cell culture. Primary MEFs were isolated from embryonic day 13.5 embryos. For each genotype, the biological variance was spread by the use of five individual embryos obtained from three individual mothers, all in a C57BL/6 background (backcrossed for more than F8 generations). MEFs were cultured as described previously (7) in Dulbecco's modified Eagle medium (Invitrogen, Breda, The Netherlands) supplemented with 10% fetal bovine serum (Biocell, Rancho Dominguez, CA), 1% nonessential amino acids (Invitrogen, Breda, The Netherlands), penicillin (0.6 µg/ml), and streptomycin (1 µg/ml) at 37°C and 5% CO2. The experiment was performed with early-passage MEFs (prior to passage five).

UV treatment. MEFs (five replicates of the wild type, p53.S389A, and p53–/–) were expanded and plated at 1 x 106 cells per 10-cm plate. Twenty-four hours later (~80% confluence), cells were washed with phosphate-buffered saline and exposed to UV-C light (20 J/m2). Control samples were mock treated and immediately collected (0 h). At several time points after treatment (3, 6, 9, 12, and 24 h), MEFs were rinsed with phosphate-buffered saline and collected in 350 µl RLT buffer (enclosed in the RNeasy mini kit; see "RNA isolation").

Western blot analysis. Wild-type MEFs were expanded, plated, and exposed to UV-C light as described in "UV treatment." To detect p53, immunoprecipitation on 100 µg total protein followed by Western blot analyses was performed as described previously by Bruins et al. (9). Briefly, membranes were incubated for at least 12 h at 4°C with either anti-p53 mouse monoclonal antibody (Ab-1; Oncogene Research Products, San Diego, CA) or anti-phospho-p53 rabbit polyclonal antibodies (Ser392; Cell Signaling, Beverly, MA). Incubation for 1 h at room temperature with horseradish peroxidase (HRP)-linked antiactin affinity-purified goat polyclonal antibody (I-19-HRP; Santa Cruz Biotechnology, Santa Cruz, CA) was performed on the membrane with total cell extract. Primary antibodies were detected by incubating for 1 h at room temperature with HRP-linked sheep anti-mouse immunoglobulin G or HRP-linked donkey anti-rabbit immunoglobulin G (Amersham Pharmacia Biotech, Piscataway, NJ), and staining was done using ECL Plus reagent (Amersham Pharmacia Biotech, Piscataway, NJ). Membranes were scanned using a PhosphorImager/Storm 860 (Molecular Dynamics, Sunnyvale, CA), and ratios were determined by TotalLab version 2.00 by using the actin protein as a loading control (Nonlinear Dynamics, Durham, NC).

RNA isolation. Total RNA was isolated using the RNeasy mini kit (Qiagen, Valencia, CA) and then treated with the RNase-Free DNase set (Qiagen, Valencia, CA). RNA was assessed for quality with the Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA). Both the RNA integrity number and the presence of degradation products were checked.

Microarrays, labeling cDNA, hybridization, and validation. The mouse oligonucleotide libraries (catalog no. MOULIBST and MOULIB384B) were obtained from Sigma-Compugen, Inc. Technical support was supplied by LabOnWeb (http://www.labonweb.com/cgi-bin/chips/full_loader.cgi). The libraries represent in total 21,766 LEADS clusters plus 231 controls. The oligonucelotide library was printed with a Lucidea Spotter (Amersham Pharmacia Biosciences, Piscataway, NJ) on commercial UltraGAPS slides (amino-silane-coated slides, Corning 40017) and processed according to the manufacturer's instructions. The slides contained 65-mer oligonucleotides, and the batch was checked for the quality of spotting by hybridizing with SpotCheck Cy3-labeled nonamers (Genetix, New Milton, Hampshire, United Kingdom).

Total RNA samples were hybridized in randomized batches, according to a common reference design without dye swap, with embryonic mouse tissue taken as a common reference. From the total RNA samples with an RNA integrity number value of >7, 1.5 µg was amplified using the Amino Allyl MessageAmp amplified RNA kit (Ambion, Austin, Texas) and labeled with Cy3 (experimental samples) and Cy5 (common reference) reactive dye according to the manufacturer's instructions. The microarrays were hybridized overnight with 200 µl hybridization mixture, consisting of 50 µl Cy3- and Cy5-labeled amplified RNA (with 150 pmol Cy3 and 75 pmol Cy5), 100 µl formamide, and 50 µl 4x RPK0325 microarray hybridization buffer (Amersham Pharmacia Biosciences, Piscataway, NJ), at 37°C, washed in an automated slide processor (Amersham Pharmacia Biosciences, Piscataway, NJ), and subsequently scanned (Agilent DNA microarray scanner, Agilent Technologies, Palo Alto, CA).

To verify the microarray results, cDNA was generated from RNA using the high-capacity cDNA archive kit containing random hexamer primers (Applied Biosystems). The presence of mRNA was measured with TaqMan gene expression assays (Applied Biosystems) on a 7500 Fast real-time PCR system with a two-step PCR procedure, according to the manufacturer's protocol. The primers were as follows: Mdm2, primer forward (TGTGTGAGCTGAGGGAGATGT) and primer reversed (ATGCTCACTTACGCCATCGT); Reporter Fam (CTCGCATCAGGATCTTG); CcnB2, Mm00432351_m1; Caspase 8, Mm0080224_m1; and Pmaip1 (Noxa), Mm00451763_m1.

Data extraction and statistical procedure. Microarray spot intensities were quantified as artifact-removed densities, using Array Vision software (version 6.0). Further processing of the data was performed using R (version 2.2.1) and the Bioconductor MAANOVA package (version 0.98.8). All slides were subjected to a set of quality control checks, i.e., visual inspection of the scans, examining the consistency among the replicated samples by principal component analysis (PCA), testing against criteria for signal-to-noise ratios, testing for consistent performance of the labeling dyes, pen grid plots to check consistent pen performance, and visual inspection of pre- and postnormalized data with box and ratio intensity plots.

The data set concerned a two-factorial design, with the factors "time" (six levels: 0, 3, 6, 9, 12, and 24 h) and "genotype" (three levels: wild type, p53.S389A, and p53–/–). The design was completely balanced with five replicates each, so the experiment involved 90 observations per gene.

After log2 transformation, the data were normalized by a spatial Lowess smoothing procedure. The data were analyzed using a two-stage mixed analysis of variance (ANOVA) model (28). First, array, dye, and array-by-dye effects were modeled globally. Subsequently, the residuals from this first model were fed into the gene-specific model to fit treatment and spot effects on a gene-by-gene basis using a mixed-model ANOVA. These residuals can be considered normalized expression values and used in the graphs to depict gene expression profiles. All changes were calculated from the model coefficients. For hypothesis testing, a permutation-based F1 test, which allows relaxation of the assumption that the data are normally distributed, was used (1,500 permutations). The significance of the differences between factor level means was tested using contrasts. To account for multiple testing, all P values from the permutation procedure were adjusted to represent a false discovery rate of 5% (4).

Statistical tests. To quantify the effect of absence of p53.S389 phosphorylation on UV response over time, a gene-specific linear model was fitted on the wild-type and p53.S389 data that included coefficients for the effects of genotype (fixed), time (fixed), and array (random). The significance of the interaction between the factors genotype and time was tested to determine if gene-specific response profiles depended on the genotype. In addition, three different contrast analyses were performed to investigate the three research questions defined in the introduction.

(i) Test I. For the first research question, a gene-specific linear model that included coefficients for effects of genotype (fixed), time (fixed), and array (random) was fitted on the complete data set. The significance of each of the three pair-wise differences between the three genotype coefficients was tested using a contrast matrix. This test identified genes whose significant differences between the mean expression levels of the wild-type, p53.S389A, and p53–/– genotypes are similar for all time points. Due to the absence of significant interaction (see Results), these time profiles may be considered parallel. In this study, these differences across time are defined as the differences in basal gene expression levels between genotypes.

(ii) Test II. For the second research question, a gene-specific linear model that included coefficients for the effects of time (fixed) and array (random) was fitted on the wild-type data set containing six time points only. The genes were tested for a main effect among time points. The genes were also subjected to a test for differential gene expression between subsequent time points using a contrast matrix.

(iii) Test III. For the third research question, a gene-specific linear model that included coefficients for each genotype-time combination (fixed) and array (random) was fitted on the complete data set. The significance of differences in gene expression levels between subsequent time points for each genotype was tested separately using a contrast matrix. For each time contrast, genes were selected that showed a difference between time points in the wild-type MEFs and/or the p53.S389A mutant MEFs.

These three tests yielded three types of gene lists: (i) genes with different basal gene expression levels between the genotypes, (ii) genes for which expression levels changed over time, describing the wild-type response to UV irradiation, and (iii) genes with time-specific differences for both wild-type and p53.S389A MEFs. The immunoglobulin and T-cell receptor genes were deleted from the eventual gene lists, because the probes representing these composite genes were extremely overrepresented in the oligonucleotide libraries.

Additional data analyses. To compare the basal levels of gene expression in the p53.S389A MEFS with the basal levels in the wild-type and the p53–/– MEFS, the model coefficients from test I were subjected to

Formula
where {alpha}WT, {alpha}SA, and {alpha}KO are the model coefficients quantifying the wild-type, p53.S389A and p53–/– effects, respectively. Basically, if the basal level of gene expression of the p53.S389A mutants is higher than p53–/– and lower than the wild type, or lower than p53–/– and higher than the wild type, y = 1 by definition. This equation was used to screen for these intermediate responders.

In order to relate the differences in gene expression levels between the wild-type and p53.S389A MEFs to differences in functional biological processes, various analyses were performed. Lists of differentially expressed genes extracted from tests I, II, and III were all analyzed for overrepresentation of gene ontologies (GO) using Onto-Express (http://vortex.cs.wayne.edu/projects.htm) (10). GO terms with false discovery rate-corrected P values of ≤0.1 and at least five significantly differentially expressed genes from test I, test II, or test III were reported. The assemblies of the actual gene lists from test II were driven by biological considerations and based on the results and are therefore described in Results. The list of differentially expressed genes extracted from test I was also analyzed for the overrepresentation of pathways using Pathway-Express (http://vortex.cs.wayne.edu/projects.htm). Pathways with P values of ≤0.15 and at least five significantly differentially expressed genes were reported.

The F1 statistics from test III were used for gene set enrichment analysis (GSEA) (43). All pathways (GenMAPP, KEGG, Biocarta, Sigma Aldrich) present in the c2 database of the Molecular Signatures Database (MSigDB 2.0; http://www.broad.mit.edu/gsea/msigdb/msigdb_index.html) were tested for significance using the geneSetTest function provided by the Limma package (version 2.7.3) in BioConductor. Pathways with P values of ≤0.1 and at least five significantly differentially expressed genes were reported. This analysis yielded pathways that are related to differences between time points for either wild-type or p53.S389A MEFs.

For an upstream motif discovery analysis, we used ~50 genes with the highest ratios that were identified for each of the following gene sets: wild type (WT) versus p53.S389A (SA) (2,253 genes), WT versus SA p53-induced (1,087 genes), WT versus SA p53-repressed (1,166 genes), WT UV-response (6,058 genes), WT unique UV-response (2,107 genes), the overlap in WT and SA UV-response (3,097 genes), and the SA unique UV-response (544 genes). These sets were loaded into the POXO (http://ekhidna.biocenter.helsinki.fi/poxo/) sequence retriever tool to retrieve 1-kb upstream sequences from the "mus musculus clean" data set. The retrieved sequences were then loaded into the POCO tool, and 40 patterns with a maximum of eight bases were requested per set. The resulting motifs with a P value of ≤E–05 were checked with the screener tool for known motifs from TRANSFAC.

For the analyses of direct p53 target genes, we used gene lists with upstream P53 binding motifs, present in the C3 database of the MSigDB (http://www.broad.mit.edu/gsea/msigdb): V$P53_02 and V$P53_DECAMER_Q2 (both derived from TRANSFAC). All generated lists of significant genes were checked for overrepresentation of each target gene list separately, their unions, and intersections by use of an adapted version of the hypergeometric test included in the GOstats package (version 1.4.0) from BioConductor. Sets with P values of ≤0.1 were taken as significant.


arrow
RESULTS
 
Wild-type, p53.S389A, and p53–/– MEFs were exposed to 20 J/m2 UV-C irradiation. Exposure to UV-C was previously shown to increase p53 phosphorylation at Ser389 (2, 9, 20, 27, 48). Furthermore, this specific UV dose was previously shown by us to result in reduced apoptotic responses in p53.S389A MEFs (9). Exposure of wild-type MEFs to 20 J/m2 UV-C irradiation resulted in the induction of total p53 protein levels and Ser389-phosphorylated p53 levels (see Fig. S1, top, in the supplemental material). To more quantitatively interpret the p53 protein levels, we measured the absolute intensities of the protein using actin as a loading control. The bottom of Fig. S1 in the supplemental material shows the induction of p53 protein levels and p53.S389 phosphorylation over time, both corrected for loading differences. An ~25-fold increase in total p53 levels was observed compared to untreated MEFs, as well as a 22-fold induction of the p53.S389 phosphorylated form.

For gene expression analysis, we analyzed the MEFs at different time points after UV exposure (for the experimental design, see Fig. 1, top). A first impression of the differences in gene expression levels obtained from a PCA is presented in Fig. 1 (bottom). This shows a clear separation of the three genotypes along the principal component 1 axis, explaining 32% of total variance. The control samples (i.e., time zero) did not cluster, indicating an endogenous difference in basal gene expression levels (i.e., without UV exposure). The principal component 2 axis, explaining 18% of total variance, shows clear separations between all time points. Markedly, the time course (including the control samples) after UV exposure of wild-type and p53.S389A MEFs shows the same trend along the principal component 2 axis. The 0- and 3-h time points representing gene expression in p53–/– MEFs also show the same coordinates on this axis. However, the 6-, 9-, 12-, and 24-h time points appear shifted compared to the wild-type and p53.S389A MEFs. Altogether, the gene expression response of p53.S389A MEFs lies between those of the wild-type and p53–/– MEFs.


Figure 1
View larger version (14K):
[in this window]
[in a new window]

  		FIG. 1. Experimental design and PCA of microarray data. (Top) The experimental design depicting the five replicates used at all six time points for the three genotypes, wild type (WT) (green), p53.S389A (SA) (blue), and p53–/– (KO) (red). (Bottom) PCA of all microarray data. The PCA shows segregation between the genotypes on the principal component 1 axis and segregation between the time points on the principal component 2 axis.

Test I: the effect of absence of p53.S389 phosphorylation on basal gene expression levels. To investigate the effect of the p53.S389A mutation in MEFs on basal gene expression levels, we tested for genotype differences. Although the gene expression changes in wild-type versus p53.S389A are rather small (see Fig. S2 in the supplemental material), no less than 2,253 genes were significantly affected by the absence of p53.S389 phosphorylation in MEFs (see Table 1 [column R; WTgvsSAg] posted at http://www.microarray.nl/mef-uv.html).

Genes affected by the p53.S389A mutation. To classify these 2,253 genes, we identified 78% (1,762) as p53-dependent genes that were also differentially expressed between wild-type and p53–/– MEFs (again after testing for genotype) (Fig. 2A and B). This category of genes needs a functional p53 to maintain basal gene expression levels, and Ser389 phosphorylation plays a role in this. For further classification, we identified 67% (1,499) as SA {approx} KO genes that were not differentially expressed between p53.S389A and p53–/– MEFs. For this category of genes, total absence of p53 or mutated p53.S389A induces a similar basal gene expression level. After grouping, four categories could be identified (Fig. 2A and B). The first and by far the largest category consisted of 1,128 genes (50%) that were affected in their basal gene expression by the mutation at the Ser389 site in a manner similar to a complete deletion of p53 (Fig. 2A, category A). The second category consisted of 634 genes (28%) for which, although affected by both the p53.S389A mutation and p53–/–, the absence of Ser389 phosphorylation had an effect other than a complete deletion of p53 (Fig. 2A, category B). The third category consisted of 120 genes (5%) that were unaffected by a complete deletion of p53, but for which phosphorylation of the Ser389 site is nevertheless important to maintain their basal expression level (Fig. 2A, category C). The fourth category consisted of 371 genes (17%) that were unaffected by a complete deletion of p53, and for which phosphorylation of the Ser389 site was of influence only in comparison with the wild-type MEFs (Fig. 2A, category D). (For all information, see Table 1 [column V, category WTgvsSAg] posted at http://www.microarray.nl/mef-uv.html.)


Figure 2
View larger version (41K):
[in this window]
[in a new window]

  		FIG. 2. Differences in basal gene expression levels between wild-type and p53.S389A MEFs. (A) Venn diagram of genes that showed differential basal expression levels in p53.S389A MEFs compared to the wild type (WTvsSA), classified into four categories by overlaps with the genes that gave differential basal expression levels between the wild-type and p53–/– genotypes (WTvsKO) and between the p53.S389A and p53–/– genotypes (SAvsKO). The four indicated categories should be read as follows: category A (p53-dependent genes), absence of Ser389 phosphorylation is similar to p53 loss; category B (p53-dependent genes), absence of Ser389 phosphorylation is dissimilar to p53 loss; category C (p53-independent genes), absence of Ser389 phosphorylation is dissimilar to p53 loss; and category D (p53-independent genes), absence of Ser389 phosphorylation is similar to p53 loss. (B) Percentages of genes, in these categories, with an intermediate basal gene expression level in p53.S389A compared to the wild type and p53–/– MEFs, or an assigned p53-repressed/induced trait. (C) The biological significance of genes with a different basal gene expression level between the wild type and p53.S389A, divided into four categories (for details, see text), is identified for overrepresentation of GO using Onto-Express and GSEA for pathways (see Materials and Methods for restrictions). Red, basal gene expression level of p53.S389A is higher than that in the wild type; green, basal gene expression level of p53.S389A is lower than that in the wild type. (D) Bar plot of normalized expression values from genes with significantly different basal gene expression levels, present in some example processes shown in Fig. 2C. Black bars, wild type; white bars, p53.S389A; and gray bars, p53–/–.

Basal expression levels of genes affected by the p53.S389A mutation. However informative it may be, gene classification does not reveal the relative gene expression levels of the genes involved. Because the PCA showed an overall intermediate response in p53.S389A MEFs compared to wild-type and p53–/– MEFs, we analyzed the relative basal gene expression levels. We defined an intermediate basal gene expression level, simplified, as wild type > p53.S389A > p53–/– or wild type < p53.S389A < p53–/–.

A total of 1,544 of the 2,253 genes (69%) affected by the p53.S389A mutation were found to have such an intermediate basal gene expression level in p53.S389A MEFs (Fig. 2B). Looking specifically at the p53-dependent genes (categories A and B), almost all genes showed an intermediate basal gene expression level (82% and 98%, respectively). The p53-independent genes (categories C and D) have by definition no intermediate expression levels (see Materials and Methods).

We further analyzed these genes with intermediate basal gene expression levels to discover potential relationships between p53.S389 phosphorylation and induction (wild type > p53.S389A) or repression (wild type < p53.S389A) of p53-dependent genes. The 2,253 genes are almost equally distributed in p53.S389 phosphorylation-dependent repressed (52%) and induced (48%) genes (Fig. 2B). However, 66% of genes in category A are p53.S389 phosphorylation-dependent, repressed genes, whereas 72% of category B are p53.S389 phosphorylation-dependent, induced genes. Category C with 65% is quite similar to category B, whereas for category D, almost equal percentages of repressed and induced genes were observed.

Processes involving genes with basal gene expression levels affected by the p53.S389A mutation. To get further insight into which cellular processes the genes with affected basal gene expression levels are involved in, GO analyses for overrepresentation of GO terms and GSEA for enriched pathways were performed (Fig. 2C). Thirteen significant GO terms were found for total wild type versus the p53.S389A genotype; 9 were found for category A, 14 for category B, 0 for category C, and just 1 for category D. Strikingly, analysis using the categories resulted in a loss of 6, but a gain of 15 GO terms, underlining the biological meaning of the defined categories. It appears that the more-general GO terms are replaced by more-specific GO terms, especially in category B, such as (induction of) apoptosis and protein amino acid phosphorylation. Moreover, there is only one GO term overlap between categories A and B.

To see whether GO terms were overall up- or downregulated, we calculated the percentage of significant up-/downregulated genes per GO term. If we consider 45 to 55% an indication of no clear direction, strikingly, almost all category A-specific GO terms are upregulated, whereas all category B-specific ones are downregulated (Fig. 2B) in p53.S389A MEFs. This means that the specific processes, represented by the GO terms found with category A genes, are mostly actively repressed via p53.S389 phosphorylation. Two examples are presented (Fig. 2D, top) for the frizzled-2 signaling pathway and cell-cell adhesion in which 100% and 67% of the respective genes showed an intermediate basal gene expression level in p53.S389A MEFs and 80% and 83% of the respective genes were expressed more in p53.S389A than in wild-type MEFs. Similarly, specific processes represented by the GO terms found with category B genes are all actively induced via p53.S389 phosphorylation. Two examples are presented (Fig. 2D, bottom) for the induction of apoptosis and regulation of cell growth, in which 100% of the respective genes showed an intermediate basal gene expression level in p53.S389A MEFs and 80% and 100% of the respective genes were expressed less in p53.S389A than in wild-type MEFs. Even the one overlapping GO term contained 71% upregulated category A genes and 88% downregulated category B genes.

The GSEA revealed seven significant pathways for total wild type versus the p53.S389A genotype, six for category A, eight for category B, none for category C, and just two for category D. Although applying the categories here did not result in a major gain of pathways, three out of the four p53-containing KEGG pathways were present (see Fig. S3 in the supplemental material). Here again, the trend was that all category A-specific pathways are upregulated, and all those of category B are downregulated, even for the two overlapping pathways.

The analyses of all genes with a differential basal gene expression level to identify upstream DNA sequence motifs resulted in three motifs, two of which were present in the TRANSFAC database (see Table 3 posted at http://www.microarray.nl/mef-uv.html). Overrepresentation analysis of genes with a p53 DNA binding sequence motif revealed only overrepresentation of V$P53_DECAMER_Q2-associated genes in category D, WT versus KO, and SA versus KO (see Table 4 posted at http://www.microarray.nl/mef-uv.html).

Test II: gene expression analysis of the response to UV exposure in wild-type MEFs. To analyze the role of p53.S389 phosphorylation in UV response, we started with a gene expression analysis of the UV response over time in wild-type MEFs. An ANOVA was performed, and 6,058 significantly differentially expressed genes were identified (see Table 1 [column S, WTt] posted at http://www.microarray.nl/mef-uv.html). In this set, a total of eight different clusters with common gene expression profiles were found after hierarchical clustering (Fig. 3A). These clusters comprised common gene expression profiles with predominantly early decrease (clusters 1 to 3 and 8), continuous decrease (5), late decrease (4), early increase (4 and 6), and late increase (1 to 3 and 7).


Figure 3
View larger version (36K):
[in this window]
[in a new window]

  		FIG. 3. Affected genes and processes in wild-type MEFs after exposure to UV. (A) Hierarchical clustering of the average log2 (z scores) of the 6,058 differentially expressed genes in wild-type MEFs over time after exposure to UV irradiation revealed eight clusters with a common gene expression profile. Each row represents an individual gene, and each column represents a time point after exposure to UV irradiation (untreated = 0). The degree of redness and greenness represents induction and repression, respectively. (For details, see Table 1 posted at http://www.microarray.nl/mef-uv.html.) (B) Clustering of 2,856 differentially expressed genes found by time-period-specific analysis of wild-type MEFs after exposure to UV. Each row represents an individual gene, and each column represents a time interval after UV exposure: 0 to 3 h, 3 to 6 h, 6 to 9 h, 9 to 12 h, and 12 to 24 h. A gene was either found (gray) or not found (black) differentially expressed in a specific time interval. From this, we defined four categories of responsive genes: early (0 to 3 h), late (12 to 24 h), early-late (0 to 3 h and 12 to 24 h), and miscellaneous. (C) Venn diagram illustrating the number of responsive genes found in defined phases I (0 to 3 h), II (3 to 6 h, 6 to 9 h, or 9 to 12 h), and III (12 to 24 h). (D) Significant GO terms (ranked with decreasing significance) for the four categories of responsive genes, plotted on the phases of the timeline.

Phase-specific genes involved in the response to UV exposure in wild-type MEFs. These common gene expression profiles were quite difficult to interpret but showed predominantly early and late effects. Therefore, we proceeded by analyzing the relative change within each phase of the timeline and found 2,856 genes differentially expressed in at least one of the five time intervals: 0 to 3 h, 3 to 6 h, 6 to 9 h, 9 to 12 h, and 12 to 24 h (Fig. 3B). This analysis revealed that the UV response takes place primarily during the three hours after exposure and 12 to 24 h after exposure, as most differentially expressed genes (1,427 and 1,756, respectively) (Fig. 3B and C) are found during these periods. We defined three phases, I (0 to 3 h), II (3 to 6 h, 6 to 9 h, or 9 to 12 h), and III (12 to 24 h), and four exclusive categories, early (923), middle (107 [identical to phase II]), late (1,257), and early-late (387) responsive genes. There is a rather high specificity of responsive genes with respect to the phases; the early responders compile 65% of the phase I genes, and the late responders compile 72% of phase III genes (Fig. 3C).The early-late responders compile almost the rest of the genes in phases I and III. Most of the genes found in phase I were downregulated (57%), whereas those found in phase III were mostly upregulated (61%).

Phase-specific processes involved in the response to UV exposure in wild-type MEFs. To identify the cellular processes involved, we subsequently analyzed these four categories of responsive genes using GO analysis. As was to be expected from the number of genes involved, we found 20 affected GO terms with early responsive genes, 3 with middle, 34 with late, and 9 with early-late (Fig. 3D). There is little overlap of the four categories among these GO terms.

We also found expected GO terms, such as "cell cycle," "DNA repair," "regulation of transcription from RNA polymerase II promoter," and "(induction of) apoptosis," as they were implicated previously with respect to treatment with a genotoxic agent, such as UV in a different cellular context (50). Furthermore, as somewhat expected, processes such as "response to (regulation of) transcription," "cell adhesion," and "DNA replication" are significantly present. Also, ubiquitin-related processes such as "ubiquitin cycle" were found to be significantly affected in response to UV irradiation. Interestingly, looking in more detail at the differences in processes found in the early, middle, late, and early-late responders, it can be observed that, for instance, "regulation of transcription, DNA-dependent" was found to be significantly affected in the early and middle responders, whereas the opposite process, "negative regulation of transcription, DNA-dependent," was found in late responders. It can be concluded that apoptosis-related and cell cycle regulation processes are involved early after UV exposure, whereas a variety of DNA replication and metabolism processes are involved later.

p53 target genes involved in the response to UV exposure in wild-type MEFs. Finally, we determined which of the 6,058 genes involved in the wild-type UV response had already been identified as p53 targets before. For this we used the p53 downstream model of Harris and Levine, comprising important p53 target genes and their function (17). Figure 5 shows an adapted version of this model (8) and provides an overview of UV-responsive genes in wild-type MEFs in response. The regulators of p53 stability and activity, Mdm2 and E2f1, were both involved. In almost all depicted downstream pathways, p53 target genes were involved: 70% of the cell cycle arrest pathway, 100% of the extrinsic-apoptotic pathway, 44% of the intrinsic-apoptotic pathway, one (of four) downstream of these apoptotic pathways, and even one (of four tested) in the angiogenesis and metastasis pathway. In summary, profiles of differential gene expression levels in wild-type MEFs after exposure to UV irradiation can be convincingly mapped to specific p53-dependent pathways.


Figure 5
View larger version (35K):
[in this window]
[in a new window]

  		FIG. 5. Affected p53 target genes in wild-type MEFs after exposure to UV. p53 target genes involved in apoptosis, cell cycle arrest, inhibition of angiogenesis and metastasis, and DNA repair processes are presented in a model adapted from references 8 and 17. Genes regulated in UV-exposed wild-type MEFs are depicted in yellow. Corresponding responses in p53.S389A MEFs are indicated with colored circles (see also the explanation of symbols in the figure).

The upstream DNA sequence motif analyses of UV irradiation-responsive genes resulted in no motifs (see Table 3 posted at http://www.microarray.nl/mef-uv.html). Analysis of genes with a p53 DNA binding sequence motif revealed almost always overrepresentation of both V$P53_O2 and V$P53_DECAMER_Q2-associated genes in wild-type UV irradiation-responsive genes over time and genes from phase II (see Table 4 posted at http://www.microarray.nl/mef-uv.html).

Test III: effect of absence of p53.S389 phosphorylation on UV-induced gene expression. We continued with the analysis of UV-responsive genes and mechanisms in p53.S389A MEFs, where 4,166 significantly differentially expressed genes were identified (see Table 1 [column T; SAt] posted at http://www.microarray.nl/mef-uv.html) This is substantially less than the 6,058 genes found in wild-type MEFs. The ANOVA did not show any genes with a significant difference in gene expression over time between wild-type and p53.S389A MEFs after UV exposure (interaction term of genotype by time). Although common in ANOVAs, this result means that any potential difference in response is likely to be quite subtle, which forced us to use alternative approaches to analyze the gene expression data.

Genes involved in the UV response of p53.S389A and wild-type MEFs. We integrated all previous analyses at the gene level by a mutual comparison of the 4,166 p53.S389A UV-responsive genes with the 6,058 wild-type UV-responsive genes and with the 2,253 genes with a changed basal gene expression level by the absence of p53.S389 phosphorylation (Fig. 4A; also see Table 1 [column R, WTgvsSAg; column S, WTt; column T, SAt] posted at http://www.microarray.nl/mef-uv.html). A total of 918 genes (41%) with a changed basal gene expression level in p53.S389A MEFs are involved in the response to UV exposure in either wild-type or p53.S389A MEFs. Conversely, 2,107 genes (35%) were found solely in wild-type MEFs in response to UV exposure, indicating that phosphorylation of p53.S389 is somehow a prerequisite for involvement of these genes in the normal UV response. Also, 544 genes (13%) were found solely in the p53.S389A UV response, indicating that the absence of phosphorylation of p53.S389 causes the involvement of these genes in the UV response. Finally, 3,558 genes were found to be differentially expressed in both wild-type (59%) and p53.S389A (85%) MEFs, which indicates that phosphorylation of p53.S389 is not exclusively necessary for the involvement of these genes in the normal UV response.


Figure 4
View larger version (34K):
[in this window]
[in a new window]

  		FIG. 4. Phase-specific genes and processes in wild-type and p53.S389A MEFs after exposure to UV irradiation. (A) Venn diagram combining the gene lists of three analyses: (i) WT versus SA (WTvsSA), genes with differences in basal expression levels between the wild type and p53.S389A; (ii) WT in time, genes with changing expression levels over time in wild-type MEFs after UV exposure; and (iii) SA in time, genes with changing expression levels over time in p53.S389A MEFs after UV exposure. (B) Venn diagrams per phase of genes involved in wild-type and p53.S389A UV response. For phase definitions, see the legend for Fig. 3B. (C) Phase- and wild-type-specific GO terms and GSEA pathways determined on the basis of phase- and wild-type-specific genes. (D) Phase- and p53.S389A-specific GO terms and GSEA pathways determined on the basis of phase- and p53.S389A-specific genes. (E) GO terms determined as described for panels C and D but only genotype specific for a certain phase. (F) Genes involved in showing the wild-type and p53.S389A UV response for the GO terms shown in panel E. Bold genes are those present in both genotypes. Red, expression level of genes in the specific pathway increases over time; green, expression level of genes in the specific pathway decreases over time.

Phase-specific genes involved in the UV response of p53.S389A and wild-type MEFs. From earlier studies (8), we suspected that time-related UV responses, such as delayed gene activation, were specific to the p53.S389A MEFs. To test this, we identified differentially expressed genes in wild-type and p53.S389A MEFs in response to UV irradiation, applying the phases defined previously (Fig. 3B). Combining the results revealed for each phase, wild-type-specific, shared, and p53.S389A-specific UV-responsive genes (Fig. 4B; also see Table 2 posted at http://www.microarray.nl/mef-uv.html) with the same implications as those for the role p53.S389 phosphorylation represents for these genes are explained in the previous paragraph. Judging from the large fractions of phase I wild-type-specific (67%) and phase III shared (58%) UV-responsive genes, p53.S389 phosphorylation seems required mostly in the early phase of normal UV response.

Phase-specific processes involved in the UV response of p53.S389A and wild-type MEFs. To determine the effects on a process level, we performed an integrated GO and GSEA analysis on these phase-specific genes involved in the wild-type and p53.S389A UV response. We distinguished phase-specific GO terms and pathways that were also genotype-specific for wild-type UV response (requiring p53.S389 phosphorylation [Fig. 4C]), genotype-specific for p53.S389A UV response (the result of the absence of p53.S389 phosphorylation [Fig. 4D]), and present in both wild-type and p53.S389A UV responses but in a different phase (the p53.S389A mutation has a different effect in a different phase [Fig. 4E]). By far the majority of identified genotype-specific GO terms and pathways again occurred in phases I and III. Also, they were extremely specific, as there was no genotype-specific GO term or pathway present in either phase. Especially in the GO terms, it is clear that most phase I-related processes are reduced, in general, whereas most phase III-related processes were induced. There were only five phase-specific, genotype-nonspecific GO terms, of which three had reduced presence in wild-type phase I and induced presence in p53.S389A phase II (Fig. 4E). Although this may hint toward a delayed response, comparison of the individual genes showed that only a few genes of these GO terms were wild type as well as p53.S389A specific (Fig. 4F). The consequent findings of overall reduced and induced GO terms and pathways may indicate that the cell controls cellular processes and pathways via the regulation of specific genes.

p53 target genes involved in the UV response and affected by the p53.S389A mutation. Finally, we mapped the results regarding the role of p53.S389 phosphorylation to the previously introduced p53 downstream model (17). Figure 5 gives an overview of the effects of absence of p53.S389 phosphorylation on the wild-type UV response: 6 wild-type genes were unchanged in p53.S389A MEFs, 13 had a lower level of gene expression in p53.S389A, and 1 had a higher level of gene expression. This last observation of the Cdc2 gene fits with the reduced expression level of its (indirect) negative regulators, i.e., Reprimo and Gadd45. More importantly, the apoptotic pathways showed mainly reduced activity, whereas the cell cycle arrest pathways seem either off (G1-S) or induced (G2-M) in UV-exposed p53.S389A MEFs.

The upstream DNA sequence motif analyses on UV irradiation-responsive genes in p53.S389A MEFs resulted in four motifs, none of which were present in the TRANSFAC database (see Table 3 posted at http://www.microarray.nl/mef-uv.html). An analysis of genes with a p53 DNA binding sequence motif revealed almost always the expected overrepresentation of both V$P53_O2 and V$P53_DECAMER_Q2-associated genes in UV-responsive genes in p53.S389A MEFs (see Table 4 posted at http://www.microarray.nl/mef-uv.html).


arrow
DISCUSSION
 
Posttranslational modifications of p53 are important in regulating p53 stability and activity (2, 5). We have shown previously that the p53.S389 phosphorylation site is partially responsible for the tumor suppression of UV-induced skin tumors and 2-AAF-induced urinary bladder tumors (8, 9, 19). Mutant p53.S389A cells showed an affected apoptosis response after UV exposure (9), and several p53 target genes involved in apoptosis and cell cycle control showed a delayed response in p53.S389A urinary bladders after 2-AAF exposure. Here, we used transcriptome analysis on primary MEFs before and after exposure to UV in order to analyze the effect of absence of p53.S389 phosphorylation on apoptosis and other p53-dependent pathways.

Test I: the effect of absence of p53.S389 phosphorylation on basal gene expression levels. Phosphorylation of p53.S389 occurs specifically after exposure to DNA-damaging agents (23, 49), especially UV (27, 38). Given that the Ser389-phosphorylated p53 level in untreated cells is extremely low (9), cells lacking this specific phosphorylation capacity would supposedly be affected only in their response to DNA-damaging agents such as UV. However, when considering only the genotype (without exposure), we found 2,253 genes slightly differentially expressed in p53.S389A MEFs, i.e., p53.S389 phosphorylation-dependent genes. This seems rather high for a single p53 point mutation, as 7,567 genes were differentially expressed in p53–/– MEFs, i.e., p53-dependent genes (results not shown). The overlap was 23% of all p53-dependent genes also found in p53.S389A and 78% of the p53.S389 phosphorylation-dependent genes also found in p53–/– MEFs. Although in line with similar observations where the p53.S389A genotype (8) or complete deletion of p53 (55) resulted in altered gene expression prior to any exposure, the number of genes with adjusted basal gene expression levels seems rather high. This phenomenon could be caused by so-called spontaneous DNA damages, such as those induced by reactive oxygen species (ROS) or depurination (reviewed in 11). There is a relationship between ROS, p53 protein levels, and oxidation-inducing DNA-damaging agents (40, 52). Another possible explanation could be the in vitro culture conditions, since, for instance, the exposure of cells to 20% O2 and 6% CO2 clearly imposes (genotoxic) stress on the cells. Alternatively, the whole system might be readjusted as a network to respond to the effect of the introduced p53.S389A mutation. If so, this means that the genes involved, though only slightly affected, are somehow related to normal p53.S389 functioning, and their analysis would be extremely informative.

To interpret these p53.S389A-affected genes, we categorized them using basal gene expression levels in p53–/– MEFs. As such, we were able to identify whether these genes are p53-dependent (SA versus WT = KO versus WT), show a similar change compared to p53–/– (SA {approx} KO), have a basal gene expression level intermediate to wild-type and p53–/– (WT > SA > KO or WT < SA < KO), and are repressed (WT < SA) or induced (WT > SA) by intact p53.S389 phosphorylation. This turned out to be quite a successful approach. We were able to identify the p53-independent genes (22%), and from the almost complete lack of results from the GO and pathway analyses, we assumed that these genes, although affected, do not play an important role in the context of p53.S389 phosphorylation. This left us with the p53-dependent genes of which almost all (88%) showed a basal gene expression level between those of the wild type and p53–/–, meaning that the lack of p53.S389 phosphorylation results follows largely the direction of the p53–/– adjustment. Moreover, (p53-dependent) genes that are normally p53.S389 phosphorylation-dependently repressed generally (81%) showed a similar adjustment compared to p53–/–, whereas genes that are normally p53.S389 phosphorylation dependently induced showed no bias. Since this applies to 23% of all p53-dependent genes, it might be a general effect that p53-dependent gene expression repression can be lifted by just a small p53 modification mutation in a fashion similar to the complete absence of p53. Likewise but inversely, for p53-dependent gene expression induction, the effect of absence of p53 cannot be mimicked as easily, probably due to redundancy in activation mechanisms.

As for the processes related to the defined p53.S389 phosphorylation-dependent gene categories, several GO terms and pathways were found for the p53-dependent genes, such as frizzled-2 signaling pathway, cell adhesion, (induction of) apoptosis, and regulation of cell growth. There appears to be a bias that signal transduction and cellular interaction processes (i.e., environmental information processing) are normally repressed by p53 (requiring Ser389 phosphorylation), whereas cellular processes seem to be induced by p53.S389 phosphorylation. Specifically for cell growth, cultured p53–/– MEFs grow faster than wild-type MEFs (18), which might also be true for p53.S389A MEFs, but since this process was found for p53-induced genes with changes closer to those of the wild type, it is not that clear. The pathways "cell adhesion" and "metabolism" were previously also found to be affected by a p53 codon 237 mutation in human lymphoblastoid cells (55). Furthermore, several Wnt genes that are able to activate the important Wnt-signaling pathway, associated with a broad panel of developmental and physiological processes, such as embryogenesis and cancer development (44), showed an increased basal gene expression level. Depletion of Wnt/β-catenin made cells more sensitive to apoptosis (21). Thus, the upregulation of the four Wnt genes in p53.S389A mutant MEFs might explain the reduced apoptotic response observed previously (9).

These observations in adjusted basal gene expression levels might relate to altered responses in p53.S389A MEFs to DNA-damaging compounds, such as UV irradiation. One can envision that certain basal levels are preferable when a direct response to DNA damage (here UV exposure) is required, and cells with certain genotypes lacking these basal levels thus might have delayed or reduced capacities to initiate the proper response efficiently after DNA damage. This hypothesis is supported by the fact that no less than 41% of the genes with adjusted basal expression in p53.S389A are also found after UV exposure in either wild-type MEFs (17%), p53.S389A MEFs (3%), or both (20%). Obviously, this leaves an intriguing 59% (1, 335) of genes that are involved in p53.S389 phosphorylation-dependent processes other than those used in UV response.

Test II: analysis of differentially expressed genes in wild-type MEFs after UV exposure. The majority of in vitro studies with UV as a challenging agent were carried out with immortalized or cancer cell lines, analyzing gene expression differences using a limited amount of time points (19, 40-46). These limitations drove our current experiment design to study transcriptional responses to DNA damaging agents in primary cells (MEFs) after UV-C exposure with an extensive time course. The consequence, of course, is quite a complex bioinformatics analysis.

Before analysis of the p53 mutant UV response, we firstly needed to understand the wild-type UV response. It turned out that this response is highly biphasic, which had, to various degrees, also been found by others (25, 26). Many genes (1,427) change in the first three hours, hardly any (289) between 3 and 12 h, and again many (1,756) from 12 to 24 h. In total, 2,856 genes were involved, which showed a remarkable specificity (80%) for being used in only one specific phase. The defined UV-responsive categories with uniquely used genes were: early (35%), middle (4%), late (47%), and the biphasic category, early-late (14%). Most of the early responsive genes were repressed, which is in line with other studies (13, 15, 47). Many early-late responsive genes showed an opposite gene expression response in the early versus late phase.

These genes led to the identification of many phase-specific cellular processes (i.e., GO terms). This clearly biphasic UV response showed as early processes transcription, apoptosis, cell growth, and cell cycle. The late processes were replication, cell proliferation, transport, adhesion, and several metabolism processes. These early and late UV responses were reported earlier (13, 14, 47, 54). One study with an extensive time course, 0.5, 3, 6, 12, and 24 h after UV exposure, also defined (five) different response phases (13). Although in that study a UV-B response in human keratinocytes was analyzed, many of the processes involved were comparable with our findings in the UV-C response in murine fibroblasts. So, the specificities of the UV and the cell type have a minor influence on the cellular response. It seems that the early UV response focuses on direct activation of processes that avoid sustained DNA damage in cells such as apoptosis, transcriptional regulation of DNA damage response genes, and cell cycle-related processes. The late responses are related more to reentering the cell cycle, e.g., DNA replication, nucleic acid metabolism, and ATP synthesis. In the early-late responsive group, the GO term DNA repair, initiated by exposure to UV irradiation (16), was present. This DNA repair response might, in the early response, aim at the immediate removal of DNA damage from actively transcribed DNA essential for the cell to survive, whereas in the late response it might aim to eliminate DNA damage from the overall genome. Another interesting finding is that several ubiquitin-related processes, required for marking (old, damaged, or misfolded) proteins for destruction, were found in all phases, indicating that these processes play a prominent role throughout the UV response.

Test III: effect of p53.S389 phosphorylation on UV-induced gene expression in the specific phases. Whereas the ANOVA revealed many genes for the term "genotype" and even more for the term "time," the interaction term "genotype by time" showed no genes. This means that the ANOVA method is not powerful enough to identify the subtle changes in gene expression, which is a common characteristic of ANOVA interaction terms. Aligning the analysis of the UV responses in wild-type and p53.S389A MEFs resulted, as expected, in a major overlap (59% and 85%, respectively). But also, a considerable number of genes (3,108) were either not found (80%) or newly found (20%) after UV exposure in p53.S389A compared to wild-type MEFs. This was expected, since phosphorylation of p53.S389 has been predominantly observed after UV exposure and, as such, will likely have an impact on p53 functioning as a transcription factor when cells are exposed to UV (9). Of all genes involved in UV response, 14% showed an adjusted basal gene expression level before exposure to UV irradiation. Together, these findings point toward a system where a significant part of the p53-dependent gene network is readjusted in response to the p53.S389A mutation, so that the response to UV exposure results only in minor differential expression responses of the involved p53-dependent genes and a weak differential phenotype response.

Extensive analysis of the affected biphasic UV response in p53.S389A MEFs revealed that phase I appears affected mostly by the absence of Ser389 phosphorylation, showing the absence of some processes dealing with cell cycle and apoptotic responses. The latter is in line with the reduced apoptotic responses detected in p53.S389A MEFs following the same UV dose applied here (9). It is tempting to speculate that when the optimal transcriptional activation of genes and consequently the functioning of proteins in these processes are affected in p53.389A cells, these cells will sustain more persistent DNA damage, and that this damage might be fixed into gene mutations or other genetic alterations. As a defense mechanism, the p53.389A cells could increase levels or activities of processes such as DNA repair or responses to DNA damage stimuli. Indeed, these two processes were found specifically upregulated in p53.S389A cells in phase III, showing an affected defense response in p53.S389A MEFs.

Our analysis revealed an interesting detail: the process of induction of apoptosis, which was absent in p53.S389A phase I, contained the recently identified UV-related apoptotic p53 target gene in MEFs, Siva (24). Apparently, p53.S389 phosphorylation is needed for the optimal response of this apoptotic target gene. Also, the GO term "protein ubiquitination," which was absent in p53.S389A phase I, contains six relevant genes: Fbxw, the ubiquitin ligase implicated in the control of chromosome stability and further identified as a p53-dependent tumor suppressor gene (41); Kcmf1, identified as a potential metastasis suppressor (30); p53-target gene Mdm2, functioning as a primary regulator of p53 (31); Vhlh, playing a role in tumor suppression by participating as a component of the p53 transactivation complex during DNA damage response (45); and Wwp1, recently identified as a Mdm2-independent regulator of p53 activity which, analogous to Mdm2, also showed a possible feedback loop mechanism (32). All these genes display a role in DNA damage response pathways or are related to tumorigenic processes, all but one (Kcmf1) are clearly related to p53, and p53.S389 phosphorylation plays a role in their functioning.

Our initial analyses to identify reoccurring motifs in the upstream DNA sequence of involved genes identified a few motifs, of which two are present as transcription factor binding sites in the TRANSFAC database. The p53 DNA binding motifs were, however, not identified, so we also applied an overrepresentation approach using two known p53 DNA binding motifs. Although 78% of the genes affected by the p53.S389A mutation were identified as p53-dependent genes, genes with these two p53 DNA binding motifs were not significantly overrepresented in genes with an affected p53.S389A basal gene expression. An explanation for this could be that regulation of these genes is indirect and does not involve direct p53 binding. This might be expected to some extent, since the S389A mutation is not present in the DNA binding domain of the p53 protein. In contrast, genes with these p53 DNA binding motifs appeared overrepresented in UV-responsive genes of wild-type, p53.S389A, and p53–/– MEFs.

Comparing this study with the p53.S389A bladders exposed to 2-AAF (8), we noticed several distinct differences. In both studies, genes with adjusted basal gene expression levels were observed, though many more were found in MEFs than in bladder cells. This might relate to the in vitro-versus-in vivo setup, as we also experienced this difference in other studies. Both studies showed corresponding affected processes, such as the p53-related pathways, cell cycle arrest and apoptosis. However, in the 2-AAF-exposed p53.S389A bladders, we found delayed gene expression profiles, whereas in the UV-exposed p53.S389A MEFs, an overall reduced expression profile was found. Likely, the important differences in setup, in vivo versus in vitro, different compounds (2-AAF versus UV), and different time scales (weeks versus hours), determine these outcomes.

Finally, it is obvious that the analysis presented here, though already extensive, is just a starting point for such a complex transcriptomics experiment as described here. We have identified many processes involved in several p53 genotypes before and after UV exposure, each of which deserved to be further microdissected so as to determine precisely what its role is in (p53-dependent) DNA-damaging responses and how it is affected by the absence of p53.S389 phosphorylation. As a preview of the complexity, we showed in Fig. 4F the many (different) genes that are affected in some key processes in UV response, such as DNA repair and response to DNA damaging stimulus. To comprehend the overall interactions, we need a broad model of the p53 network. As such, the updated model from Harris and Levine (17) (Fig. 5) is a perfect starting point for this. We were already able to map our initial findings described above, which immediately showed the significance of p53.S389 phosphorylation in the major p53-induced pathways.

In summary, we identified the absence of a number of processes needed to negatively regulate tumor promoting processes and further the gain of a number of processes positively regulating tumorigenic processes in our p53.S389A mutant MEFs after exposure to UV irradiation compared to wild-type MEFs. As a consequence, the absence or decreased efficiency of p53.S389 phosphorylation may result in more initiated cells followed by increased incidences of (skin) tumors after exposure to UV, indeed, a phenomenon observed when mice lacking this phosphorylation event are chronically exposed to UV irradiation (9). Whether affected p53.S389 phosphorylation is also observed in humans (i.e., human p53.S392) and as such accounts for increased sensitivity for sunlight-induced skin cancers is an interesting question that needs to be addressed.


arrow
ACKNOWLEDGMENTS
 
We are grateful to M. M. Schaap for her assistance with the RT-PCR analyses, and R. Stad for fruitful discussions.

This work was supported by the Dutch Cancer Society (KWF) grant 2000-2352, NIH/NIEHS (Comparative Mouse Genomics Centers Consortium) grant 1UO1 ES11044-02, and a BSIK grant through The Netherlands Genomics Initiative in the context of the BioRange program of The Netherlands Bioinformatics Centre.


arrow
FOOTNOTES
 
* Corresponding author. Mailing address: National Institute of Public Health and the Environment, Laboratory of Toxicology, Pathology, and Genetics, P.O. Box 1, 3720 BA Bilthoven, The Netherlands. Phone: 31 30 274 3483. Fax: 31 30 274 4446. E-mail: Annemieke.de.Vries{at}rivm.nl Back

{triangledown} Published ahead of print on 14 January 2008. Back

{dagger} Supplemental material for this article may be found at http://mcb.asm.org/. Back

# Both authors contributed equally to this work. Back


arrow
REFERENCES
 
    1
  1. Appella, E., and C. W. Anderson. 2000. Signaling to p53: breaking the posttranslational modification code. Pathol. Biol. (Paris) 48:227-245.[Medline]
    2. 2
  2. Appella, E., and C. W. Anderson. 2001. Post-translational modifications and activation of p53 by genotoxic stresses. Eur. J. Biochem. 268:2764-2772.[Medline]
    3. 3
  3. Ashcroft, M., and K. H. Vousden. 1999. Regulation of p53 stability. Oncogene 18:7637-7643.[CrossRef][Medline]
    4. 4
  4. Benjamini, Y., and Y. Hochberg. 1995. Controlling the false discovery rate—a practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B 57:289-300.
    5. 5
  5. Bode, A. M., and Z. Dong. 2004. Post-translational modification of p53 in tumorigenesis. Nat. Rev. Cancer 4:793-805.[CrossRef][Medline]
    6. 6
  6. Brooks, C. L., and W. Gu. 2006. p53 ubiquitination: Mdm2 and beyond. Mol. Cell 21:307-315.[CrossRef][Medline]
    7. 7
  7. Brugarolas, J., C. Chandrasekaran, J. I. Gordon, D. Beach, T. Jacks, and G. J. Hannon. 1995. Radiation-induced cell cycle arrest compromised by p21 deficiency. Nature 377:552-557.[CrossRef][Medline]
    8. 8
  8. Bruins, W., M. J. Jonker, O. Bruning, J. L. Pennings, M. M. Schaap, E. M. Hoogervorst, H. van Steeg, T. M. Breit, and A. de Vries. 2007. Delayed expression of apoptotic and cell cycle control genes in carcinogen-exposed bladders of mice lacking p53.S389 phosphorylation. Carcinogenesis 28:1814-1823.[CrossRef]
    9. 9
  9. Bruins, W., E. Zwart, L. D. Attardi, T. Iwakuma, E. M. Hoogervorst, R. B. Beems, B. Miranda, C. T. M. van Oostrom, J. van den Berg, G. J. van den Aardweg, G. Lozano, H. van Steeg, T. Jacks, and A. de Vries. 2004. Increased sensitivity to UV radiation in mice with a p53 point mutation at Ser389. Mol. Cell. Biol. 24:8884-8894.[Abstract/Free Full Text]
    10. 10
  10. Cadwell, C., and G. P. Zambetti. 2001. The effects of wild-type p53 tumor suppressor activity and mutant p53 gain-of-function on cell growth. Gene 277:15-30.[CrossRef][Medline]
    11. 11
  11. Coates, P. J., S. A. Lorimore, and E. G. Wright. 2005. Cell and tissue responses to genotoxic stress. J. Pathol. 205:221-235.[CrossRef][Medline]
    12. 12
  12. Corbet, S. W., A. R. Clarke, S. Gledhill, and A. H. Wyllie. 1999. P53-dependent and -independent links between DNA-damage, apoptosis and mutation frequency in ES cells. Oncogene 18:1537-1544.[CrossRef][Medline]
    13. 13
  13. Dazard, J. E., H. Gal, N. Amariglio, G. Rechavi, E. Domany, and D. Givol. 2003. Genome-wide comparison of human keratinocyte and squamous cell carcinoma responses to UVB irradiation: implications for skin and epithelial cancer. Oncogene 22:2993-3006.[CrossRef][Medline]
    14. 14
  14. Enk, C. D., J. Jacob-Hirsch, H. Gal, I. Verbovetski, N. Amariglio, D. Mevorach, A. Ingber, D. Givol, G. Rechavi, and M. Hochberg. 2006. The UVB-induced gene expression profile of human epidermis in vivo is different from that of cultured keratinocytes. Oncogene 25:2601-2614.[CrossRef][Medline]
    15. 15
  15. Gentile, M., L. Latonen, and M. Laiho. 2003. Cell cycle arrest and apoptosis provoked by UV radiation-induced DNA damage are transcriptionally highly divergent responses. Nucleic Acids Res. 31:4779-4790.[Abstract/Free Full Text]
    16. 16
  16. Guo, Y. L., H. C. Chang, J. H. Tsai, J. C. Huang, C. Li, K. C. Young, L. W. Wu, M. D. Lai, H. S. Liu, and W. Huang. 2002. Two UVC-induced stress response pathways in HeLa cells identified by cDNA microarray. Environ. Mol. Mutagen. 40:122-128.[CrossRef][Medline]
    17. 17
  17. Harris, S. L., and A. J. Levine. 2005. The p53 pathway: positive and negative feedback loops. Oncogene 24:2899-2908.[CrossRef][Medline]
    18. 18
  18. Harvey, M., A. T. Sands, R. S. Weiss, M. E. Hegi, R. W. Wiseman, P. Pantazis, B. C. Giovanella, M. A. Tainsky, A. Bradley, and L. A. Donehower. 1993. In vitro growth characteristics of embryo fibroblasts isolated from p53-deficient mice. Oncogene 8:2457-2467.[Medline]
    19. 19
  19. Hoogervorst, E. M., W. Bruins, E. Zwart, C. T. van Oostrom, G. J. van den Aardweg, R. B. Beems, J. van den Berg, T. Jacks, H. van Steeg, and A. de Vries. 2005. Lack of p53 Ser389 phosphorylation predisposes mice to develop 2-acetylaminofluorene-induced bladder tumors but not ionizing radiation-induced lymphomas. Cancer Res. 65:3610-3616.[Abstract/Free Full Text]
    20. 20
  20. Huang, C., W. Y. Ma, A. Maxiner, Y. Sun, and Z. Dong. 1999. p38 kinase mediates UV-induced phosphorylation of p53 protein at serine 389. J. Biol. Chem. 274:12229-12235.[Abstract/Free Full Text]
    21. 21
  21. Huang, M., Y. Wang, D. Sun, H. Zhu, Y. Yin, W. Zhang, S. Yang, L. Quan, J. Bai, S. Wang, Q. Chen, S. Li, and N. Xu. 2006. Identification of genes regulated by Wnt/beta-catenin pathway and involved in apoptosis via microarray analysis. BMC Cancer 6:221.[CrossRef][Medline]
    22. 22
  22. Iwakuma, T., and G. Lozano. 2007. Crippling p53 activities via knock-in mutations in mouse models. Oncogene 26:2177-2184.[CrossRef][Medline]
    23. 23
  23. Jackson, M. W., M. K. Agarwal, M. L. Agarwal, A. Agarwal, P. Stanhope-Baker, B. R. Williams, and G. R. Stark. 2004. Limited role of N-terminal phosphoserine residues in the activation of transcription by p53. Oncogene 23:4477-4487.[CrossRef][Medline]
    24. 24
  24. Jacobs, S. B., S. Basak, J. I. Murray, N. Pathak, and L. D. Attardi. 2007. Siva is an apoptosis-selective p53 target gene important for neuronal cell death. Cell Death Differ. 14:1374-1385.[CrossRef][Medline]
    25. 25
  25. Kannan, K., N. Amariglio, G. Rechavi, J. Jakob-Hirsch, I. Kela, N. Kaminski, G. Getz, E. Domany, and D. Givol. 2001. DNA microarrays identification of primary and secondary target genes regulated by p53. Oncogene 20:2225-2234.[CrossRef][Medline]
    26. 26
  26. Kannan, K., N. Kaminski, G. Rechavi, J. Jakob-Hirsch, N. Amariglio, and D. Givol. 2001. DNA microarray analysis of genes involved in p53 mediated apoptosis: activation of Apaf-1. Oncogene 20:3449-3455.[CrossRef][Medline]
    27. 27
  27. Kapoor, M., and G. Lozano. 1998. Functional activation of p53 via phosphorylation following DNA damage by UV but not gamma radiation. Proc. Natl. Acad. Sci. USA 95:2834-2837.[Abstract/Free Full Text]
    28. 28
  28. Kerr, M. K., M. Martin, and G. A. Churchill. 2000. Analysis of variance for gene expression microarray data. J. Comput. Biol. 7:819-837.[CrossRef][Medline]
    29. 29
  29. Ko, L. J., and C. Prives. 1996. p53: puzzle and paradigm. Genes Dev. 10:1054-1072.[Free Full Text]
    30. 30
  30. Kreppel, M., D. N. Aryee, K. L. Schaefer, G. Amann, R. Kofler, C. Poremba, and H. Kovar. 2006. Suppression of KCMF1 by constitutive high CD99 expression is involved in the migratory ability of Ewing's sarcoma cells. Oncogene 25:2795-2800.[CrossRef][Medline]
    31. 31
  31. Kubbutat, M. H., S. N. Jones, and K. H. Vousden. 1997. Regulation of p53 stability by Mdm2. Nature 387:299-303.[CrossRef][Medline]
    32. 32
  32. Laine, A., and Z. Ronai. 2007. Regulation of p53 localization and transcription by the HECT domain E3 ligase WWP1. Oncogene 26:1477-1483.[CrossRef][Medline]
    33. 33
  33. Lakin, N. D., and S. P. Jackson. 1999. Regulation of p53 in response to DNA damage. Oncogene 18:7644-7655.[CrossRef][Medline]
    34. 34
  34. Latonen, L., and M. Laiho. 2005. Cellular UV damage responses—functions of tumor suppressor p53. Biochim. Biophys. Acta 1755:71-89.[Medline]
    35. 35
  35. Latonen, L., Y. Taya, and M. Laiho. 2001. UV-radiation induces dose-dependent regulation of p53 response and modulates p53-HDM2 interaction in human fibroblasts. Oncogene 20:6784-6793.[CrossRef][Medline]
    36. 36
  36. Levine, A. J. 1997. p53, the cellular gatekeeper for growth and division. Cell 88:323-331.[CrossRef][Medline]
    37. 37
  37. Liu, G., and X. Chen. 2006. Regulation of the p53 transcriptional activity. J. Cell. Biochem. 97:448-458.[CrossRef][Medline]
    38. 38
  38. Lu, H., Y. Taya, M. Ikeda, and A. J. Levine. 1998. Ultraviolet radiation, but not gamma radiation or etoposide-induced DNA damage, results in the phosphorylation of the murine p53 protein at serine-389. Proc. Natl. Acad. Sci. USA 95:6399-6402.[Abstract/Free Full Text]
    39. 39
  39. Lu, X. 2005. p53: a heavily dictated dictator of life and death. Curr. Opin. Genet. Dev. 15:27-33.[CrossRef][Medline]
    40. 40
  40. Macip, S., M. Igarashi, P. Berggren, J. Yu, S. W. Lee, and S. A. Aaronson. 2003. Influence of induced reactive oxygen species in p53-mediated cell fate decisions. Mol. Cell. Biol. 23:8576-8585.[Abstract/Free Full Text]
    41. 41
  41. Mao, J. H., J. Perez-Losada, D. Wu, R. Delrosario, R. Tsunematsu, K. I. Nakayama, K. Brown, S. Bryson, and A. Balmain. 2004. Fbxw7/Cdc4 is a p53-dependent, haploinsufficient tumour suppressor gene. Nature 432:775-779.[CrossRef][Medline]
    42. 42
  42. Mitchell, D. L., C. A. Haipek, and J. M. Clarkson. 1985. (6-4) Photoproducts are removed from the DNA of UV-irradiated mammalian cells more efficiently than cyclobutane pyrimidine dimers. Mutat. Res. 143:109-112.[CrossRef][Medline]
    43. 43
  43. Mootha, V. K., C. M. Lindgren, K. F. Eriksson, A. Subramanian, S. Sihag, J. Lehar, P. Puigserver, E. Carlsson, M. Ridderstrale, E. Laurila, N. Houstis, M. J. Daly, N. Patterson, J. P. Mesirov, T. R. Golub, P. Tamayo, B. Spiegelman, E. S. Lander, J. N. Hirschhorn, D. Altshuler, and L. C. Groop. 2003. PGC-1{alpha}-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat. Genet. 34:267-273.[CrossRef][Medline]
    44. 44
  44. Polakis, P. 2007. The many ways of Wnt in cancer. Curr. Opin. Genet. Dev. 17:45-51.[CrossRef][Medline]
    45. 45
  45. Roe, J. S., and H. D. Youn. 2006. The positive regulation of p53 by the tumor suppressor VHL. Cell Cycle 5:2054-2056.[Medline]
    46. 46
  46. Sachsenmaier, C., A. Radler-Pohl, R. Zinck, A. Nordheim, P. Herrlich, and H. J. Rahmsdorf. 1994. Involvement of growth factor receptors in the mammalian UVC response. Cell 78:963-972.[CrossRef][Medline]
    47. 47
  47. Sesto, A., M. Navarro, F. Burslem, and J. L. Jorcano. 2002. Analysis of the ultraviolet B response in primary human keratinocytes using oligonucleotide microarrays. Proc. Natl. Acad. Sci. USA 99:2965-2970.[Abstract/Free Full Text]
    48. 48
  48. Takekawa, M., M. Adachi, A. Nakahata, I. Nakayama, F. Itoh, H. Tsukuda, Y. Taya, and K. Imai. 2000. p53-inducible wip1 phosphatase mediates a negative feedback regulation of p38 MAPK-p53 signaling in response to UV radiation. EMBO J. 19:6517-6526.[CrossRef][Medline]
    49. 49
  49. Thompson, T., C. Tovar, H. Yang, D. Carvajal, B. T. Vu, Q. Xu, G. M. Wahl, D. C. Heimbrook, and L. T. Vassilev. 2004. Phosphorylation of p53 on key serines is dispensable for transcriptional activation and apoptosis. J. Biol. Chem. 279:53015-53022.[Abstract/Free Full Text]
    50. 50
  50. van Delft, J. H., E. van Agen, S. G. van Breda, M. H. Herwijnen, Y. C. Staal, and J. C. Kleinjans. 2004. Discrimination of genotoxic from non-genotoxic carcinogens by gene expression profiling. Carcinogenesis 25:1265-1276.[Abstract/Free Full Text]
    51. 51
  51. van Zeeland, A. A., M. P. Vreeswijk, F. R. de Gruijl, H. J. van Kranen, H. Vrieling, and L. F. Mullenders. 2005. Transcription-coupled repair: impact on UV-induced mutagenesis in cultured rodent cells and mouse skin tumors. Mutat. Res. 577:170-178.[Medline]
    52. 52
  52. Xie, S., Q. Wang, H. Wu, J. Cogswell, L. Lu, M. Jhanwar-Uniyal, and W. Dai. 2001. Reactive oxygen species-induced phosphorylation of p53 on serine 20 is mediated in part by polo-like kinase-3. J. Biol. Chem. 276:36194-36199.[Abstract/Free Full Text]
    53. 53
  53. Xu, D., T. J. Wilson, D. Chan, E. De Luca, J. Zhou, P. J. Hertzog, and I. Kola. 2002. Ets1 is required for p53 transcriptional activity in UV-induced apoptosis in embryonic stem cells. EMBO J. 21:4081-4093.[CrossRef][Medline]
    54. 54
  54. Zhao, R., K. Gish, M. Murphy, Y. Yin, D. Notterman, W. H. Hoffman, E. Tom, D. H. Mack, and A. J. Levine. 2000. Analysis of p53-regulated gene expression patterns using oligonucleotide arrays. Genes Dev. 14:981-993.[Abstract/Free Full Text]
    55. 55
  55. Zschenker, O., K. Borgmann, T. Streichert, I. Meier, A. Wrona, and E. Dikomey. 2006. Lymphoblastoid cell lines differing in p53 status show clear differences in basal gene expression with minor changes after irradiation. Radiother. Oncol. 80:236-249.[CrossRef][Medline]

Molecular and Cellular Biology, March 2008, p. 1974-1987, Vol. 28, No. 6
0270-7306/08/$08.00+0     doi:10.1128/MCB.01610-07
Copyright © 2008, American Society for Microbiology. All Rights Reserved.




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental material
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowReprints and Permissions
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Bruins, W.
Right arrow Articles by Breit, T. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Bruins, W.
Right arrow Articles by Breit, T. M.

Copyright © 2009 by the American Society for Microbiology.   For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»