Human Molecular Genetics Advance Access originally published online on September 29, 2005
Human Molecular Genetics 2005 14(21):3293-3308; doi:10.1093/hmg/ddi362
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
In vivo misregulation of genes involved in apoptosis, development and oxidative stress in mice lacking both functional Werner syndrome protein and poly(ADP-ribose) polymerase-1
Centre de Recherche en Cancérologie de l'Université Laval, Hôpital Hôtel-Dieu de Québec, Québec, Que., Canada
* To whom correspondence should be addressed at: Centre de Recherche en Cancérologie, Hôpital Hôtel-Dieu de Québec, 9 McMahon St, Québec, Que., Canada G1R 2J6. Tel: +1 4186915281; Fax: +1 4186915439; Email: michel.lebel{at}crhdq.ulaval.ca
Received August 8, 2005; Accepted September 22, 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Werner syndrome (WS) is a rare disorder characterized by the premature onset of a number of age-related diseases. The gene responsible for WS is believed to be involved in different aspects of transcription, replication and/or DNA repair. The poly(ADP-ribose) polymerase-1 (PARP-1) enzyme is also involved in DNA repair and is known to affect transcription of several genes. In this study, we examined the expression profile of cells lacking the normal function of either or both enzymes. All mutant cells exhibited altered expression of genes normally responding to oxidative stress. Interestingly, more than 58% of misregulated genes identified in double mutant cells were not altered in cells with either the Wrn or PARP-1 mutation alone. So, the impact on gene expression profile when both Wrn and PARP-1 are mutated was greater than a simple addition of individual mutant genotype. In addition, double mutant cultured cells showed major misregulation of genes involved in apoptosis, cell cycle control, embryonic development, metabolism and signal transduction. More importantly, in vivo analyses of double mutant mice have confirmed the increased apoptosis and the developmental defects in embryos as well as the major increase in intracellular phosphorylation and oxidative DNA damage in adult tissues. They also exhibited a progressive increase in oxidative stress with age. Thus, a major result of this study is that changes in expression of several genes and physiological functions identified in vitro were confirmed in mouse embryonic and adult tissues.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Werner syndrome (WS) is a rare disorder characterized by the premature onset of a number of processes associated with aging including malignancies (1
). The gene responsible for WS (WRN) was identified by positional cloning and its product contains a domain homologous to the RecQ-type DNA helicases (2). The protein also possesses a 3'–5' exonuclease activity in addition to its 3'–5' helicase activity (3–5). The WRN protein is considered a suppressor of illegitimate recombination, as skin fibroblasts and lymphoblastoid cell lines made from circulating lymphocytes of WS patients exhibit variegated chromosomal translocations and deletions (6,7). In addition, several reports have also indicated that human WS cells have abnormal telomere dynamics in vitro, which is likely to affect replicative senescence (8). Interestingly, transcription alterations in WS were also found to be similar to those in normal aging (9). Like human WS cells, mouse embryonic stem cells with a deletion of the helicase domain are sensitive to both camptothecin and etoposide which are type I and II topoisomerase inhibitors, respectively (10–12). Finally, crosses between Wrn mutant mice and animals with appropriate reporter genes in their genome have indicated a significant increase in illegitimate recombination in their tissues (13,14).
The enzyme poly(ADP-ribose) polymerase-1 (PARP-1) is also a protein that affects recombination in cells. Chemical or genetic abrogation of PARP-1 activity leads to an increase in the frequency of sister chromatid exchanges and genomic instability, especially after genotoxic stresses (15). A number of PARP-1 knockout mice have been generated by several groups (15). Although mice lacking a functional PARP-1 enzyme develop normally and are not cancer-prone, they are hypersensitive to DNA damage (16,17). Fibroblasts established from such mutant mice have a slower growth rate in culture compared with wild-type fibroblasts (18,19). In addition to a loss of proliferative capacity, PARP-1 null fibroblasts display increased telomere shortening compared with wild-type cells (20). However, an additional report has indicated that this increase in telomere shortening is only observed in late passage cells and not in primary embryonic tissue cultures (21). It is believed that the increased genomic instability (gain or loss of chromosomal DNA) contributes to the cell proliferation delay observed in PARP-1 mutant cells (19,21). Finally, microarray analysis on PARP-1 null fibroblasts has indicated that loss of PARP-1 enzyme results in deregulation of genes implicated in cancer initiation or progression and in normal or premature aging (22).
Recent evidences have indicated that WS cells exhibit deficiencies in the poly(ADP ribosyl)ation pathway after DNA damaging treatments due to the absence of physical interaction between WRN and PARP-1 enzymes (23,24). In a previous study, we crossed mice with a mutation in the helicase domain of the Wrn gene (Wrn
hel/hel mice) to PARP-1 null mice to determine if Wrn and PARP-1 enzymes act in concert during cell growth (25). Both Wrnhel/hel and PARP-1 null/Wrnhel/hel cohorts developed more tumors than wild-type animals. The tumor spectrum was the same between PARP-1 null/Wrnhel/hel mice and Wrn mutants. However, PARP-1 null/Wrnhel/hel mice developed tumors at a younger age (25). Mouse embryonic fibroblasts derived from PARP-1 null/Wrnhel/hel mice were distinguished by an increased frequency of chromatid breaks, chromosome rearrangements and fragmentation (25,26) compared with wild-type, Wrn mutant, and PARP-1 null fibroblasts. Yet such persistent breaks and DNA rearrangements appeared randomly in these cells. It was thus impossible to determine whether specific regions of the genome were more prone to DNA rearrangements in the absence of both enzymes. Wrnhel/hel, PARP-1 null and PARP-1 null/Wrnhel/hel cells also exhibited a decrease in their telomere length compared with wild-type cells (25). Shorter telomeres might have caused all mutant cells to acquire a slow-growth phenotype. However, the proportion of PARP-1 null/Wrnhel/hel chromosomes with very short telomeres is not greater than those from Wrnhel/hel or PARP-1 null cells, indicating other mechanisms accountable for the sudden cell proliferation arrest observed only in PARP-1 null/Wrnhel/hel cells. Recently, it has been suggested that several aspects of human WS are secondary consequences of aberrant gene expression (27). We thus hypothesized that, in addition to chromosomal rearrangements, the expression of a specific subset of genes might be affected by the absence of functional Wrn and PARP-1 proteins in these cells contributing to the observed phenotypes as well.
In this report, we explored the expression profile of mouse embryonic cells established from wild-type, PARP-1 null, Wrnhel/hel and PARP-1 null/Wrnhel/hel embryos with the microarray technology. Our results have indicated major changes in the expression of genes involved in the response to reactive oxygen species (ROS) in all embryonic mutant cells. Furthermore, we have found that PARP-1 null/Wrnhel/hel mice exhibited high levels of ROS and DNA oxidative damage in embryonic cells and vital organs (heart and liver) compared with wild-type animals. In addition, PARP-1 null/Wrnhel/hel embryos displayed a major increase in apoptosis during development and decreased survival in utero. These results indicate that a consequence of the absence of PARP-1 and Wrn proteins is the accumulation of ROS in vivo with age and genomic instability.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Experimental design for the microarray analyses
Previous data have indicated that Wrn
hel/hel mutant and PARP-1 null cell lines acquire a slower growth rate than wild-type cell lines with the number of passage in culture (25,26). Interestingly, PARP-1 null/Wrnhel/hel mouse embryonic cells already displayed a slow growth phenotype at the second passage (approximately 10 population doublings) in culture compared with cells of all other genotypes. They stopped growing by the third passage (after approximately 15–20 population doublings) in culture. At that point, such cultures could be maintained for several months without any evidence of cell division. Thus, in contrast to other mouse embryonic cells, which are still growing after the 15–20 population doublings in culture, PARP-1 null/Wrnhel/hel cells stop dividing abruptly. Hence, microarray analyses were performed on cell cultures at the second passage (10 population doublings). For each genotype, asynchronously dividing cells derived from three healthy 15.5-day embryos from one litter were pooled at the second passage and cytoplasmic RNA was extracted (see Fig. 1 for strategy). Cytoplasmic RNA was used in these experiments to avoid contamination with heterogeneous nuclear RNA and genomic DNA. (Note that DNase treatment was also applied to all samples.) This pool of RNA was labeled sample number 1 for each genotype. Pooling of embryonic cells was performed to minimize the effect of inter-individual biological differences. A second pool of embryonic cells was also created from a separate dam (second litter) for each genotype (called samples number 2). This strategy allowed obtaining samples in duplicate for each genotype. The cRNA from wild-type cells were synthesized with Cy-5-labeled nucleotides and cRNAs from Wrnhel/hel mutant, PARP-1 null and PARP-1 null/Wrnhel/hel cells were synthesized with Cy-3-labeled nucleotides. Hybridization was performed on Mouse Agilent 60-mer Oligo Microarray chips (containing 21 317 genes) by mixing wild-type-labeled cRNA (baseline expression levels) with either Wrnhel/hel mutant, PARP-1 null or PARP-1 null/Wrnhel/hel cRNA. Hybridization experiments were done in duplicates. For example, the wild-type RNA sample number 1 was mixed with the Wrnhel/hel mutant RNA sample number 1 and hybridized on one microarray slide. On a second microarray slide (which constitutes the duplicated experiment), hybridization was carried out with the wild-type RNA sample number 2 mixed with the Wrnhel/hel mutant RNA sample number 2. Thus, each RNA sample was used only once in the hybridization experiments (Fig. 1A). The complete cDNA microarray raw data (log ratios) can be found in Supplementary Material, Table S8 (Wrnhel/hel versus wild-type, PARP-1 null versus wild-type and PARP-1 null/Wrnhel/hel versus wild-type, respectively) published as supporting information at the journal's Web site.
|
Verification of microarray data by relative RT–PCR
Expression levels of 10 randomly picked genes from the wild-type versus Wrn
hel/hel mutant expression profile were examined by RT–PCR. For such analyses, cytoplasmic RNA from embryonic cells derived from a third pool of 15.5-day embryos (one additional litter) was used. We analyzed cytoplasmic RNA samples different from the extracts used for the microarray hybridization experiments to minimize false data due to experimental variability. As shown in Figure 1, RT–PCR analyses indicated that in general the fold differences in gene expression between Wrnhel/hel and wild-type cells obtained with the microarray analyses were overestimated. For example, the gene Snrpb (small nuclear ribonucleoprotein B) showed a 2.6-fold difference between wild-type and Wrnhel/hel cells on the microarray, but only a 1.1-fold difference in expression was detected by RT–PCR between wild-type and Wrnhel/hel cells. Upon examination of the RT–PCR data, only genes whose expression was up- or down-regulated in the mutant cells by a factor of 2.7-fold on the microarrays were further considered in this study. This value corresponded to at least a 30% difference in expression between mutant and wild-type cells detectable by RT–PCR (see, e.g. the genes Igf2 and Cdkn2a in Fig. 1) and allowed us to identify subtle differences in gene expression between genotypes. This cut-off value of 2.7-fold was also applied for the comparisons between PARP-1 null/Wrnhel/hel versus wild-type and PARP-1 null versus wild-type cells. The final list of genes can be found in Supplementary Material, Tables S5–S7 (Wrnhel/hel versus wild-type, PARP-1 null versus wild-type and PARP-1 null/Wrnhel/hel versus wild-type, respectively) published as supporting information at the journal's Web site. These tables contain information on the accession number, the common name, description of the gene and their chromosome location.
Expression profiles in mutant cells compared with wild-type cells
On the basis of the selected criteria discussed earlier, we found that 119 genes (0.56% of the genes monitored) showed consistent changes in expression between Wrnhel/hel mutant and wild-type embryonic cells in the duplicated experiments. Table 1 lists the genes exhibiting more than 2.7-fold difference in expression between Wrnhel/hel mutant and wild-type embryonic cells and is given as an example in the text. Twenty-six of these genes were up-regulated and 93 genes were down-regulated in Wrnhel/hel cells (see Supplementary Material, Table S5 for information on description and map location of these genes). When we compared the expression profile of PARP-1 null and wild-type cells, we found that 285 genes (1.34% of the genes monitored) showed differential expression of more than 2.7-fold. Ninety-five and 190 genes were up- and down-regulated, respectively. The complete list of genes can be found in Supplementary Material, Table S6. Finally, when we compared the expression profile of PARP-1 null/Wrnhel/hel and wild-type cells, we found that 356 genes (1.67% of the genes monitored) showed differential expression of more than 2.7-fold. One hundred and twenty-three genes were up-regulated and 233 genes were down-regulated. The complete list of genes can be found in Supplementary Material, Table S7.
|
Near half of the genes exhibiting altered expression in Wrn
hel/hel cells when compared with wild-type cells were similarly altered in PARP-1 null or PARP-1 null/Wrnhel/hel cells. As shown in the Venn diagram of Figure 2A, 54 genes displayed similar expression changes in Wrnhel/hel, PARP-1 null and PARP-1/Wrnhel/hel cells. However, based on the microarray data, very few genes (only 10) displayed a frank additive effect in fold expression compared with wild-type culture when both enzymes were mutated in cells. Additional analyses on the expression of some of these genes (see results for endothelin-1 subsequently) did not confirm this additive effect of Wrn and PARP-1 mutations. Out of the 119 genes exhibiting changes in Wrnhel/hel cells, 28 (24%) were unique to this mutation (Table 1 and Fig. 2A). Of the 285 genes showing altered expression in PARP-1 null cells, 117 genes (41%) were unique to the PARP-1 mutation. Finally, of the 356 genes displaying expression difference in PARP-1 null/Wrnhel/hel when compared with wild-type cells (baseline expression), 205 genes (58%) were specific to these double mutant cells and were not changed in either Wrnhel/hel or PARP-1 null embryonic cells. Thus, each genotype displayed a unique set of genes with altered expression.
|
Functional categorization of the altered genes was performed to determine which pathways in each mutant cells were more affected. Genes were grouped into 14 functional classes based on the function-related gene ontology annotations. Of the 119 genes altered in Wrn
hel/hel cells, 87 genes were annotated a known function. Similarly, 207 in PARP-1 null and 276 altered genes in PARP-1 null/Wrnhel/hel cells were annotated a known function. As indicated in Figure 2B, essentially all classes of functions were affected in Wrnhel/hel cells. However, a greater number of genes involved in metabolism and signal transduction (including growth receptors) were affected by the Wrn helicase mutation (see Table 1 and Supplementary Material, Table S5 for more details on these genes). PARP-1 null cells also displayed changes in expression of genes in every categorized class (Fig. 2B). On average, there was a 2–3-fold increase in the number of altered genes over baseline levels (taking wild-type cells as reference) in PARP-1 null cells when compared with Wrnhel/hel cells in most functional categories. The only exceptions were genes involved in apoptosis, cell cycle control and development of the central nervous system. In the latter categories, the number of altered genes was similar between PARP-1 null and Wrnhel/hel cells. Finally, the functional groups that showed the greatest number of altered genes in PARP-1 null cells were metabolism, protein processing (including ubiquitination and degradation of proteins), signal transduction and transport of biomolecules across membranes of any kind including electron or ion channels (Fig. 2B) (see Supplementary Material, Table S6 for details on each gene).
PARP-1 null/Wrnhel/hel cells exhibited changes in every functional category as well (Fig. 2B). The functional classes that displayed the greatest changes in PARP-1 null/Wrnhel/hel cells compared with PARP-1 null cells (almost twice the number of genes) were apoptosis, cell cycle control, embryonic development including development of the central nervous system and DNA/RNA processing (Fig. 2B) (see Supplementary Material, Table S7 for details on each gene).
Chromosomal location of altered expressed genes
To investigate a possible relationship between genotype-specific altered expression and gene location, chromosomal positions for each altered gene were retrieved from the mouse MapViewer's Web site. Tables S5–S7 of Supplementary Material give the chromosomal band location for each gene. A chromosomal gene cluster was defined as two or more genes located within a specific stretch of linear DNA. The incidence of clusters was compared with values generated from randomly sampled lists of the same size using the bootstrapping algorithm described earlier (28). By using 100 kb as the distance defining cluster boundaries, no cluster of altered expressed genes was detected in Wrnhel/hel cells. Four clusters of altered expressed genes were observed in PARP-1 null cells and nine clusters were observed in PARP-1 null/Wrnhel/hel cells. Table 2 gives the list of clustered genes with their position. Calculation of the randomly expected values for the average number of false-positive clusters based on 285 randomly picked genes (calculated value of six random clusters for PARP-1 null cells) or 356 randomly picked genes (calculated value of 10 random clusters for PARP-1 null/Wrnhel/hel cells) indicated that the observed and calculated incidence of clusters is due to random chance alone. Thus, there is no evidence that genes displaying altered expression in each genotype are localized exclusively to specific regions of the mouse genome.
|
Increased oxidative stress in mutant embryonic cells
Analysis of the genes exhibiting changes in their expression has indicated that the metabolic state of all mutant cells is dramatically altered. The expression of several genes involved in glycolysis, lipid catabolism or energy expenditure is modified in mutant cells. Moreover, the list of genes for each genotype (see Supplementary Material, Tables S5–S7) indicates changes in the overall redox state of these cells. Upon closer examination of the number of genes known to be affected by the redox state of the mutant cells, it was hypothesized that these cells suffer from a marked oxidative stress. Approximately 38% of the genes annotated with a function in the Wrn
hel/hel (35 genes), PARP-1 null (79 genes) and PARP-1 null/Wrnhel/hel (104 genes) lists are known to be influenced by oxidative stress or are known to modify levels of cellular ROS (Table 3). (See Supplementary Material, Tables S5–S7 for more details on the genes). Thus, intracellular ROS levels were examined directly in mouse embryonic cells established from all cohorts with the dye 2'–7' dichlorofluorescein diacetate. This dye is highly fluorescent upon oxidation. As it was impossible to detect a significant difference in the basal levels of intracellular ROS between cohorts by flow cytometry, cells were lysed and the extent of fluorescence released from cells was measured with a fluorescence spectrophotometer as described previously (29). As shown in Figure 3A, ROS levels were 10% higher in Wrnhel/hel cells (t-test; P<0.05), 25% higher in PARP-1 null (t-test; P<0.05) and 220% higher in PARP-1 null/Wrnhel/hel cells (t-test; P<0.05) than in wild-type cells. As cells were lysed to quantify the amount of ROS, additional evidence for increased ROS was provided by examining the extent of oxidative DNA damage in these cells. Oxidative DNA lesions create abasic sites in cells which can be detected with the K-Assay Oxidative DNA damage Kit (Kamiya Biomedical Co., Seattle, WA). As shown in Figure 3B, there was a 28 and 70% increase in the number of abasic sites in genomic DNA derived from Wrnhel/hel and PARP-1 null cells, compared with wild-type samples, respectively. However, there was a lot of variation in the number of abasic sites in cells from one sample to another within each genotype. In contrast, there was a significant 2-fold increase in the number of abasic sites in genomic DNA from PARP-1 null/Wrnhel/hel cells (t-test; P<0.05) compared with wild-type samples. These results indicate that mutant cells exhibit oxidative stress which leads to DNA damage in vitro and the PARP-1 null/Wrnhel/hel mutant cells display the greatest number of oxidative DNA lesions.
|
|
We next examined the levels of expression of each individual gene (from Table 3) known to be altered by the redox state of cells, to determine which genes correlated with the levels of ROS and/or oxidative DNA damage in cells. Of the 104 genes known to be altered by oxidative stress in PARP-1 null/Wrn
hel/hel cells (Table 3), 44 genes were also misregulated in PARP-1 null and/or Wrnhel/hel cells. However, as indicated in Figure 3C, only 18 genes showed changes in expression correlating with the levels of oxidative stress in each mutant cell (i.e. PARP-1 null/Wrnhel/hel>PARP-1 null>Wrnhel/hel>wild-type). This indicates that the expression of several genes known to be altered by the redox state of the cells are not only influenced by the levels of oxidative stress, but also by the presence or absence of functional PARP-1 and/or Wrn enzymes as well.
Increased levels of intracellular protein phosphorylation in mutant mice
On the basis of microarray data (Fig. 2B), the greatest number of kinases and phosphatases exhibiting changes in expression were found in PARP-1 null/Wrnhel/hel mutant cells (categorized as signal transduction proteins). To examine the overall phosphorylation state in embryonic cells, total lysates from cultured cells were analyzed with anti-phosphoserine antibodies. No phosphatase inhibitor was used in the extraction buffer. As shown in Figure 3D, the greatest change in phosphorylation was detected in PARP-1 null/Wrnhel/hel mutant cells. Intermediate levels of phosphorylation were found in PARP-1 null and Wrnhel/hel cells compared with wild-type and PARP-1 null/Wrnhel/hel mutant cells. Although no phosphorylated protein above 185 kDa was detected in our cell lysates, the anti-phosphoserine blot correlated with the microarray data and indicated an increase in protein phosphorylation in mutant cells.
As PARP-1 null/Wrnhel/hel mutant cultured cells displayed the greatest increase in protein phosphorylation, adult tissues were also examined. As indicated in Figure 3D, protein extracts from liver tissues of PARP-1 null/Wrnhel/hel mutant mice indicated an increase in serine phosphorylation of several proteins compared with wild-type tissues. Similar results were also obtained with cardiac tissues of PARP-1 null/Wrnhel/hel mutant animals (data not shown). These results indicate that changes in phosphorylation status in cultured cells are also observed in tissues. The same conclusion was obtained with an antibody against tyrosine phosphorylation and tissue extracts from double mutant and wild-type mice (data not shown). The exact identity of these phosphorylated proteins remains to be confirmed.
Impaired embryonic development in mutant embryos correlates with the expression profile of misregulated genes
Preliminary data on the number of pups per litter indicated embryonic lethality in mutant embryos. The average litter size for wild-type animals was 9.0 pups per litter (20 litters) and 6.6 pups per litter (20 litters) for the Wrnhel/hel mice (see Table 4 for a summary of the data). In contrast, the average litter size for the PARP-1 null mice was 4.9 pups per litter (32 litters). Finally, the average litter size for PARP-1 null/Wrnhel/hel animals was 4.0 pups per litter (30 litters) which is significantly different from Wrnhel/hel mice. To correlate the observed changes in expression profile between genotype and embryonic development, embryos were examined at 9.5, 10.5 and 11.5 days post-coitum. By counting the number of necrotic decidua and abnormal embryos, it was found that 40% of PARP-1 null/Wrnhel/hel embryos were dying by the 12.5 days embryonic stage. Approximately, 14% of PARP-1 null embryos were dead by that stage. This is in contrast with wild-type and Wrnhel/hel pregnant dams in which only 4.5% of their embryos were dead at 12.5 days post-coitum.
|
Severe developmental defects were regularly observed in PARP-1 null/Wrn
hel/hel litters at every embryonic stage. These defects ranged from mild developmental delay (Fig. 4E) to severe malformations mainly characterized by an open neural plate or exencephaly as well as atrophy of branchial arches, limbs and tail (Fig. 4D). These observations correlated with the expression data obtained from the microarrays. There were twice as many genes categorized as affecting embryonic development or central nervous development (Fig. 2B) that displayed changes in expression in PARP-1 null/Wrnhel/hel compared with animals of all the other genotypes. Thus, changes in expression of several genes normally involved in embryonic development did impact on the survival of PARP-1 null/Wrnhel/hel animals early in life. Finally, even though there was a decrease in PARP-1 null pups per litter, no abnormality was detected in 10.5 days old (or older) embryos but there were several sites of embryonic resorption (or necrotic deciduas). This indicates that several PARP-1 null embryos died before 10.5 days in utero. This finding also correlated with the higher number of developmental genes (excluding central nervous system specific genes; Fig. 2B) that displayed misregulation compared with either Wrnhel/hel or wild-type animals.
|
To gain insight into the developmental consequences of the molecular misregulations observed in the Wrn
hel/hel, PARP-1 null or double mutant embryos, progenies of each genotype were harvested at 9.5, 10.5 and 11.5 days post-coitum and global apoptosis was assessed by the Nile Blue staining method. This analysis was conducted primarily on embryos that did not exhibit malformation. No obvious macroscopic differences among the four genotypes were noted in embryonic tissues known to contain high levels of apoptotic cells such as in the developing somites, branchial arches or apical ectodermal ridges (Fig. 4A–C and E). However, closer histological examination of embryonic sections by TUNEL assays revealed a major increase in overall cell death in PARP-1 null/Wrnhel/hel embryos compared with the embryos of all other genotypes. The number of apoptotic cells was estimated on equivalent serial sagittal sections of three embryos for each genotype (Fig. 4F–I). This number was nearly 10-fold higher in PARP-1 null/Wrnhel/hel embryos than in wild-type, Wrnhel/hel, and PARP-1 null animals (Fig. 4J). Interestingly, cell death was homogeneously distributed across all embryonic tissues in the PARP-1 null/Wrnhel/hel individuals (Fig. 4I). Little apoptosis was detected in similar sections of Wrnhel/hel and PARP-1 null embryos (Fig. 4G and H). This result correlated positively with the number of altered genes involved in apoptosis in PARP-1 null/Wrnhel/hel embryonic cells. There were twice as many altered genes involved in apoptotic regulation in PARP-1 null/Wrnhel/hel cells compared with Wrnhel/hel and PARP-1 null cells (Fig. 2B).
Increased oxidative stress in adult mutant mice
As an increase in oxidative stress was observed in mutant cells in culture, we next examined ROS levels in whole 10.5 days embryos. As shown in Figure 5A, there was a 26 and 20% increase in ROS levels in PARP-1 null and PARP-1 null/Wrnhel/hel embryos compared with wild-type embryos, respectively. However, there was a lot of variation from one embryo to another within each litter in all genotypes such that the differences were not statistically significant. Similarly, oxidative DNA damage was not significantly higher in mutant than in wild-type whole embryos (data not shown). It was impossible to determine whether embryos exhibiting the highest levels of ROS would have survived the final days of embryonic development. We then examined ROS levels in the serum of 4-month-old animals. As shown in Figure 5B, levels of hydrogen peroxide were significantly higher in serum of all mutant mice compared with wild-type animals. It is known that ROS can modify lipids by a peroxidation reaction. Additional evidence for increased ROS was thus provided by measuring the amount of stable lipid peroxides in serum of animals. As shown in Figure 5C, PARP-1 null and PARP-1 null/Wrnhel/hel mice had significantly more serum lipid peroxides compared with wild-type animals.
|
ROS levels were also analyzed in heart and liver homogenates with the dye 2'–7' dichlorofluorescein diacetate as indicated in Materials and Methods. Figure 5D shows that levels of ROS were significantly higher in the heart of mutant animals compared with wild-type mice at 6 months of age. Similarly, PARP-1 null and PARP-1 null/Wrn
hel/hel mice had more ROS in their liver compared with wild-type animals (Fig. 5E). Wrnhel/hel mice did not have significantly more ROS than wild-type animals. Additional evidence for increased ROS was provided by measuring the amount of oxidative DNA damage in the heart of animals. As shown in Figure 5F, PARP-1 null and PARP-1 null/Wrnhel/hel mice had more abasic sites in the DNA of heart tissues compared with wild-type animals at 6 months of age. We then investigated younger animals. In contrast to older animals, and similar to embryonic tissues, we did not detect any significant difference in tissue ROS levels between all mutant and wild-type animals at either 2 weeks or 4 months of age (data not shown). This indicates a progressive deterioration of the redox balance in mutant mice.
Finally, we examined levels of endothelin-1 and serum amyloid A in blood of 4-month-old mice because these two proteins are found in the lists of genes exhibiting altered expression (Supplementary Material, Tables S5–S7) and they are both up-regulated during oxidative stress (30,31). As shown in Figure 6A, levels of serum endothelin-1 were significantly increased in PARP-1 null and PARP-1 null/Wrnhel/hel mice. Although the microarray analysis indicated that endothelin-1 was up-regulated in Wrnhel/hel mouse embryonic cells, adult mice did not exhibit an increase in blood endothelin-1. Microarray analyses also indicated that the gene SAA3 (serum amyloid A3) was up-regulated in both PARP-1 null and PARP-1 null/Wrnhel/hel mouse embryonic cells but not in Wrnhel/hel cells. An ELISA assay recognizing several types of serum amyloid A was used in this study. As shown in Figure 6B, there was a 45% increase in serum amyloid A levels in PARP-1 null mice compared with wild-type animals. There was a 3-fold increase in serum amyloid A in PARP-1 null/Wrnhel/hel mice compared with wild-type animals and a 2-fold increase compared with PARP-1 null mice. As expected, serum amyloid A levels were not increased in Wrnhel/hel animals compared with wild-type mice. The fold differences detected in mutant adult mice compared with wild-type were different from the values determined by microarray analyses for cultured cells. Nevertheless, these results indicate that several genes up-regulated in embryonic cells were also up-regulated in adult PARP-1 null and PARP-1 null/Wrnhel/hel mice.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Previously, we have observed that embryonic cells lacking functional Wrn, PARP-1 or both enzymes exhibited a slow growth phenotype (25
). Hence, the major intent of this study was to establish the pattern of gene expression in such cells. Most importantly, we wanted to confirm or correlate the in vitro findings with the in vivo phenotype. For this purpose, cells were established from 15.5-day-old embryos and cytoplasmic RNA was extracted after 10 population doublings. It is worth mentioning that cells were established from healthy embryos and not from deformed or necrotic embryos. The purpose of employing cultured cells instead of tissues for oligo-arrays hybridization was to estimate the reliability of using such in vitro systems in general to extrapolate and infer on actual biological events in vivo. This is an important issue because most human microarray analyses reported in the literature were performed with cell lines or primary cells in culture. A major result of this study is that changes in expression of several genes and physiological functions identified in vitro were confirmed in mouse embryonic and adult tissues.
Examination of genes displaying altered expression in mutant cells (taking wild-type cells as baseline expression) revealed that there are three times as many genes misregulated in PARP-1 null/Wrnhel/hel cells, and twice as many in PARP-1 null cultures when compared with Wrnhel/hel cells. This correlates with the frequency of embryonic malformations and lethality observed in mutant mice. Indeed, PARP-1 null/Wrnhel/hel litters showed the greatest number of sick embryos at 10.5 days post-coitum. This correlation is emphasized by the number of genes involved in embryonic development (including central nervous system) that are altered in PARP-1 null/Wrnhel/hel cells. There are two and three times as many altered developmental genes in PARP-1 null/Wrnhel/hel cells being altered compared with PARP-1 and Wrnhel/hel null cells, respectively. Accordingly, the greatest number of deformed embryos was observed among PARP-1 null/Wrnhel/hel animals. Moreover, several deformities in the PARP-1 null/Wrnhel/hel embryos were located in the neural plate. Additional microarray analyses have indicated altered expression in more genes involved in cell cycle control and apoptosis in PARP-1 null/Wrnhel/hel cells compared with other genotypes. These results correlated well with the sudden slow growth phenotype observed in PARP-1 null/Wrnhel/hel cell culture compared with PARP-1 and Wrnhel/hel null cells (25). Furthermore, TUNEL assays on embryonic tissues have confirmed the major increase in apoptotic figures in PARP-1 null/Wrnhel/hel embryos compared with embryos of other genotypes. Apoptosis was homogeneously distributed in these embryos, indicating that abnormal cells appeared randomly in these tissues. An increase in apoptosis and a delay in cell proliferation could lead to developmental defects and lethality as observed in several PARP-1 null/Wrnhel/hel embryos. Finally, as the apoptosis levels in the PARP-1 null/Wrnhel/hel embryos is much greater than the sum of the levels seen in the single mutant backgrounds, it is likely that a synergistic phenomenon between the Wrnhel/hel and the PARP-1 null mutations affects the embryonic apoptosis mechanisms.
The microarray data provide information on the exact genes that, when de-regulated in our cells, lead to a specific phenotype. However, when examining the list carefully, it was noticed that several misregulated genes seem to have opposite effects in cells. This can be exemplified by the list of altered genes categorized as apoptotic regulators in PARP-1 null/Wrnhel/hel cells (see Supplementary Material, Table S5). An increase in clusterin expression, as seen in PARP-1 null/Wrnhel/hel cells, is known to protect cells against apoptosis (30). In contrast, Bok protein is known to be pro-apoptotic when up-regulated (31) as observed in PARP-1 null/Wrnhel/hel cells. In contrast, a decrease in Bbc3 expression would suppress apoptosis (32) in PARP-1 null/Wrnhel/hel cells. However, the phenotype observed in PARP-1 null/Wrnhel/hel embryos is increased apoptosis. This suggests that it is the number of altered genes in a specific functional category which determines the extent of physiological aberrations in that category. In PARP-1 null/Wrnhel/hel cells, the altered balance between several apoptotic regulators leads to increased apoptosis in embryos. Fewer pro- or anti-apoptotic regulators were affected in either PARP-1 null or Wrnhel/hel cells. Accordingly, much fewer apoptotic cells were detected in the corresponding mutant embryos.
The WRN protein is known to interact with PARP-1 enzymes (24,25,33). We suspected that mutation in either of these genes would alter similar biochemical pathways (or genes) in cells. In addition, a mutation in both proteins would results in an additive composite of PARP-1 null and Wrnhel/hel lists. Surprisingly, more than half (58%) of the genes modified in PARP-1 null/Wrnhel/hel cells were not altered in either PARP-1 null or Wrnhel/hel cells. Only 54 genes were common between all mutant cells and only 10 of them showed an additive effect of the Wrn and PARP-1 mutations on expression based on the microarray data. Thus, the observed expression profile in cells when both Wrn and PARP-1 are mutated does not reflect a simple addition of expression patterns obtained when each individual enzyme is mutated. Each genotype affects a specific subset of genes.
It was hypothesized that rearrangements in the chromatin structure of a specific region of the genome would coordinately alter the expression of genes localized within this region. This is not the case as the chromosomal positions of the misregulated genes in cells of each genotype were not all clustered to specific regions of the genome exhibiting gross recurrent rearrangements (26). However, the resolution of the cytogenetic studies performed up to now with these cells is not below 10 Mb and smaller mutation affecting gene expression would have been missed (26). Higher resolution analyses are required to determine whether mutations in some of these genes are the causes for their misregulation. Nonetheless, several connections between genes localized to different chromosomes can still be made from our lists. For example, up-regulation of Il-6 is known to induce expression of serum amyloid A (34,35) as seen in our mutant mice (see Supplementary Material, Tables S6 and S7). Oxidative stress will increase expression of TGFß genes which in turn will affect expression of laminin or fibronectin-like proteins (36,37) (see Supplementary Material, Tables S6 and S7). The expression of several genes of the immune response is known to be altered during oxidative stress as well as proteins of the proteasomes which are normally known to degrade abnormal oxidized proteins (38) as can be seen in the lists of our mutant mice.
Little comparison can be made between our results and microarray analyses that have been conducted on PARP-1 null mouse embryonic fibroblasts (22) or human WS cell lines (9,27). The lists of genes are quite different from our data because we used a different RNA extraction method, a different microarray platform (Agilent platform in this study), and more importantly, a different experimental design. Nevertheless, every analysis points to the activation of an adaptive stress response consistent with increased ROS levels. A conclusion that was also reached with the microarray data from the aging brain and skeletal muscles of C57Bl/6 mice (39). Approximately 38% of the genes annotated with a function in the Wrnhel/hel, PARP-1 null and PARP-1 null/Wrnhel/hel lists are known to be influenced by oxidative stress or are known to modify levels of cellular ROS. These numbers are an underestimation as several identified expression sequence tags from the microarrays code for proteins with an unknown function and the effect of ROS on the expression of several genes in each list has not been reported in the literature as of today. ROS levels and oxidative damage correlated with the number of genes identified in each cell culture (PARP-1 null/Wrnhel/hel>PARP-1 null>Wrnhel/hel>wild-type). On the other hand, the difference in oxidative stress between mutant embryos was less significant in tissues than in established mouse embryonic cells. This is not surprising, as cells in culture are in an abnormal hence more stressful environment (40). We could not determine whether there were local ROS increases in embryos leading to apoptosis. However, an increase in ROS levels was obviously detected in adult mutant mice with age. Additional evidences for oxidative stress in mutant mice came with the presence of increased lipid peroxidation in serum and increased oxidative DNA damage at least in cardiac tissues of mutant mice. Oxidative stress was cumulative as mutant babies and juvenile animals did not show a significant increase in ROS levels compared to age matched wild-type animals. Thus, mutant mice exhibited a progressive deterioration of their redox balance. It is well established now that increased ROS is involved in a number of diseases associated with aging including heart failure and cancers (41–43). A major DNA lesion generated by excessive ROS is the 8-hydroxydeoxyguanosine. If unrepaired, such lesions can ultimately lead to single-strand or double-strand breaks and DNA rearrangements (28). It has been proposed that WRN protein may be required for the repair of such lesions (44). PARP-1 enzyme has also been associated with the repair of oxidative DNA damage (45). Persistent oxidative DNA lesions would certainly exacerbates the phenotype observed in mutant mice. Indeed, increased oxidative damage was detected in PARP-1 null/Wrnhel/hel cultured cells and cardiac tissues. Accordingly, the frequency of DNA rearrangements was much higher in PARP-1 null/Wrnhel/hel cells (26). The next step will be to determine the effects of anti-oxidants on the stability of the genome, the proliferation of cultured cells and the phenotype of our PARP-1 null/Wrnhel/hel mice.
PARP-1 null mice exhibited high ROS levels and lipid peroxidation in their serum compared with all other phenotypes. Paradoxically, these mice did not age more rapidly than Wrnhel/hel animals (25). In fact, it is known that inactivation of PARP-1 enzyme will decelerate some of the diseases associated with aging such as heart fibrosis or any ischemia/reperfusion injuries (46–48). Accordingly, PARP-1 null mice never developed cardiac fibrosis in our colony. Approximately, 65% of Wrnhel/hel animals developed cardiac fibrosis but only 36% of mice developed such fibrosis when PARP-1 gene