|Home | About | Journals | Submit | Contact Us | Français|
Monozygotic twin and other epidemiologic studies indicate that epigenetic processes may play an important role in the pathogenesis of inflammatory bowel diseases that commonly affect the colonic mucosa. The peak onset of these disorders in young adulthood suggests that epigenetic changes normally occurring in the colonic mucosa shortly before adulthood could be important etiologic factors. We assessed developmental changes in colitis susceptibility during the physiologically relevant period of childhood in mice [postnatal day 30 (P30) to P90] and concurrent changes in DNA methylation and gene expression in murine colonic mucosa. Susceptibility to colitis was tested in C57BL/6J mice with the dextran sulfate sodium colitis model. Methylation specific amplification microarray (MSAM) was used to screen for changes in DNA methylation, with validation by bisulfite pyrosequencing. Gene expression changes were analyzed by microarray expression profiling and real time RT–PCR. Mice were more susceptible to chemically induced colitis at P90 than at P30. DNA methylation changes, however, were not extensive; of 23 743 genomic intervals interrogated, only 271 underwent significant methylation alteration during this developmental period. We found an excellent correlation between the MSAM and bisulfite pyrosequencing at 11 gene associated intervals validated (R2 = 0.89). Importantly, at the genes encoding galectin-1 (Lgals1), and mothers against decapentaplegic homolog 3 or Smad3, both previously implicated in murine colitis, developmental changes in DNA methylation from P30 to P90 were inversely correlated with expression. Colonic mucosal epigenetic maturation continues through early adulthood in the mouse, and may contribute to the age-associated increase in colitis susceptibility.
Transcript Profiling: Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/), accession numbers: GSE18031 (DNA methylation arrays), GSE19506 (gene expression arrays).
The high rate of discordance for inflammatory bowel diseases (IBD) within monozygotic twin pairs suggests that factors independent from the genetic code are important in their etiologies (1,2). This discordance rate is 50–80% for Crohn's disease (CD) and over 80% for ulcerative colitis (UC) (3). Other chief epidemiologic observations of IBD are their peak onset in late adulthood (4,5) and increasing prevalence with the adoption of the industrialized world's lifestyle (4,6). These observations indicate that environmentally responsive developmental processes occurring between late infancy and young adulthood are important pathogenic factors in IBD. Epigenetic mechanisms, which mediate mitotically heritable changes in gene expression that are not associated with DNA sequence changes (7) represent one such class of biological processes. The most stable epigenetic alteration is the methylation of cytosines at CpG dinucleotides. This covalent modification at gene promoters generally correlates with transcriptional down-regulation. DNA methylation, which is critical for development and differentiation (8), is catalyzed by DNA methyltransferases and ultimately dependent upon dietary substrates and co-factors (9). Accordingly, mouse models have shown that intrauterine (10,11) and even postnatal (12) dietary influences affect the establishment of epigenetic mechanisms at specific genomic loci, leading to permanent phenotypic changes. Environmentally induced epigenetic changes could likewise contribute to the development of human disease (9,13). Currently, however, our knowledge of epigenetic mechanisms in gastrointestinal development and diseases such as IBD is scarce (14).
IBD is thought to result from a complex interaction between the immune system and the intestinal microflora that is transmitted by the gut mucosa and modulated by genetic predisposition and possible epigenetic factors (1,15). Although the intestinal mucosa (comprised mostly of enterocytes) represents only one component of this intricate network (16), many of the genes thought to be involved in the development of IBD are expressed in the mucosal barrier epithelium (17).
Our hypothesis is that the late onset of IBD is explained in part by environmental and/or genetic influences on normal epigenetic maturation in the gut mucosa during childhood. Indeed, studies in FOXP3+ regulatory T cells (18,19) indicate that epigenetic mechanisms may play a role in IBD pathogenesis, and gene specific abnormalities in DNA methylation have recently been described in the colonic mucosa of IBD patients (20–23). Developmental changes in epigenetic regulation that could potentially explain the age dependence of IBD incidence, however, have not been assessed. For example, since FOXP3 expression is stable from childhood to adulthood both in the thymus and the peripheral blood (24), the age dependence of IBD incidence would not appear to be explained by developmental epigenetics at FOXP3.
In the current study, we first employed dextran sulfate sodium (DSS) to confirm that developmental changes in colitis susceptibility occur during pre-adulthood in mice, as in humans. DSS administration in rodents mimics the clinical and pathological characteristics of human UC (25). Recent work has indicated that it may act by inducing hyperosmotic stress leading to NF-κB activation through the post-translational methylation of protein phosphatase 2A (26). Since NF-κB activation is an important element of IBD pathogenesis (27–29), DSS induced colitis appears to be a good molecular model of IBD. Indeed, at postnatal day 90 (P90) mice were more prone to colitis than at P30, demonstrating an age dependent increase in susceptibility to mucosal inflammation and underscoring the relevance of the mouse model.
Then, to test our hypothesis that the age dependence of IBD is associated with developmental epigenetics, we studied healthy, non-injured mice. Methylation-specific amplification and microarray hybridization (MSAM) (30) was employed to assess the extent of DNA methylation changes in colonic epithelium over the pediatric developmental period. A subset of array hits was validated with bisulfite pyrosequencing, and associated changes in gene expression were evaluated. Additionally, whole genomic expression microarrays were performed and analyzed in parallel with the MSAM results. We identified specific changes in epigenetic regulation during this late developmental period, indicating that epigenetic maturation in colonic mucosa continues beyond infancy in mice. These changes correlate with and may contribute to increased colitis susceptibility.
To test the functional relevance of our developmental mouse model, we compared susceptibility to chemically induced colitis in P90 and P30 mice. An earlier investigation in a murine germfree transfer model suggested that wild-type mice are more prone to develop transient colitis at P84 than at P21 (38). That study, however, used a different strain of mouse, and the nutritional/environmental change at P21 was complicated by simultaneous weaning and transitioning of the animals. Therefore, we set out to confirm that susceptibility of mice to colonic inflammation is age dependent. Colitis was induced by exposing mice to DSS in their drinking water. Compared with at P30, P90 mice lost more weight, had more severe shortening of their colons and higher histological scores of colitis (Supplementary Material, Fig. S1). These findings corroborate the earlier study (38) in indicating that young adult mice are more susceptible to colitis than infants, consistent with the incidence of IBD peaking in young adult humans.
We tested whether the functional differences in colitis susceptibility could perhaps be explained by developmental changes in DNA methylation in the colonic mucosa. First, the colons and mucosal scrapings were evaluated histologically. Consistent with earlier studies (39,40), we did not detect major differences between P30 and P90 colonic mucosal architecture, and no obvious differences in cell composition were observed (Supplementary Material, Fig. S2). At both ages, more than 90% of the cells in colonic mucosal scrapings were epithelial (colonocytes; Supplementary Material, Fig. S3). Similar findings have been reported in humans, in which even normal terminal ileal mucosa contains <5% intra-epithelial lymphocytes (41). Therefore, the overwhelming majority of the DNA in colonic mucosal scrapings (>90%) arises from colonic mucosal epithelial cells, and any age dependent DNA methylation or expression changes must reflect developmental changes within colonocytes rather than in the cellular composition of the mucosa.
We utilized a methylation-sensitive/insensitive restriction endonuclease isoschizomer (SmaI/XmaI) based, methylation specific amplification microarray approach (MSAM, see Materials and Methods). By analyzing 23 743 SmaI/XmaI intervals in the mouse genome, we identified 196 that gained methylation, and 75 that lost methylation from P30 to P90 (Supplementary Material, Table S2).
To validate the MSAM results, we used bisulfite pyrosequencing to measure site-specific CpG methylation at 11 of the SmaI/XmaI intervals. An excellent correlation (r = 0.94; P < 0.00001) was found between the two methods (Fig. 1). In a larger group of mice, 73% (8/11) of the intervals showed consistent P30 to P90 methylation differences of more than 10% (Fig. 2), confirming that in most cases our approach detected developmental changes. In the pyrosequencing assays where more than 1 CpG site was evaluated, the measured DNA methylation changes were regional, meaning that more than 1 CpG site was affected (Fig. 2). Although the observed changes were generally small, they were strictly regulated within the two age groups (Fig. 2), supporting very tight developmental epigenetic regulation within the colonic mucosa during murine childhood development. This was true for the further three validated genes (Cyp27b1, Smad3 and Thumpd2/AK012806) as well, where DNA methylation changes were very low, with values below 10% in both age groups at one of the flanking SmaI/XmaI sites tested (Supplementary Material, Figs S4 and S5).
Recent data from human colonic mucosa reveal that CpG methylation changes as far as 2 kb from the transcription start site (TSS) of a gene can influence gene expression (42). Since the interrogated SmaI/XmaI intervals are up to 2 kb in length, we declared a SmaI/XmaI interval ‘gene associated’ if its midpoint was within a gene or <3 kb upstream from the TSS. Approximately 70% of both the intervals that gained methylation (139/196) and of those that lost methylation (52/75) were gene associated by this designation (Supplementary Material, Table S3). Among a group of SmaI/XmaI intervals that did not change their methylation, gene association was 57% (2713/4745). Compared with those that did not change methylation, SmaI/XmaI intervals that changed methylation were more likely to be associated with a gene (gain of methylation versus unchanged odds ratio = 1.82, 95% CI = 1.33–2.5; loss of methylation versus unchanged odds ratio = 1.69, 95% CI = 1.03–2.8).
In studies of tissue-specific gene expression and tumorigenesis, epigenetic changes at enhancers, CCCTC-binding factor (CTCF) sites and DNAse hypersensitive regions have been shown to play an important role (43–45). A search for overlaps between all the genomic regions with methylation changes and enhancer elements specific for the gastrointestinal tract yielded one positive result within 2 kb of the chr2:74450600–74450846 SmaI/XmaI interval (enhancer: chr2:74451699–74453074). This degree of overlap was not significantly different from that of the control group of SmaI/XmaI intervals in which methylation did not change (12/4745). Of the 271 SmaI/XmaI intervals that underwent methylation changes, 55 were associated with DNAseI hypersensitive loci of Caco2 cells; a similar proportion was found in the control group (1148/4745). Interestingly, however, genomic regions that changed in methylation overlapped with eight different CTCF binding sites (Supplementary Material, Table S4), a highly significant enrichment compared with control regions (2/4745), odds ratio = 72 (95% CI = 15–341). A conserved tissue specific differentially methylated region in Casz1 (46) was also recognized. In this gene, two different SmaI/XmaI intervals changed in methylation during pediatric development.
DNA methylation changes in genomic regions not characterized as ‘gene associated’ by our criteria could still affect transcriptional regulation. Imprinting control elements, for example, can regulate allele specific gene expression over hundreds of kilobases (47). Nevertheless, we focused on seven gene associated genomic regions and used real-time RT–PCR to determine if the developmental changes in DNA methylation are correlated with gene expression in colonic epithelium. Expression of Ajap1, Pax8 and Spry1 was not detectable at either age, and that of Nrbp2 and Slc39a14 was similar at P30 and P90 (data not shown). We did, however, observe an inverse correlation between DNA methylation and expression of galectin-1 (Lgals1) with an average of 50% decrease in Lgals1 mRNA levels from P30 to P90 (Fig. 3). A similar, inverse correlation was observed at Smad3; methylation was decreased and expression increased at P90 compared with P30 (Supplementary Material, Fig. S5).
A gene ontology (GO) analysis (http://babelomics.bioinfo.cipf.es/) was performed to determine if gene regions undergoing methylation change during childhood and adolescence in the colonic mucosa are associated with particular biological processes, molecular functions or cellular components. The reference group consisted of 1690 gene-associated SmaI/XmaI intervals on the microarray, for which methylation was unchanged from P30 to P90. Interestingly, genes involved in urogenital/metanephros/kidney development (Wt1, Robo2, Spry1, Bdnf, Gdnf, Gli2, Pax8) were over-represented (adjusted P-value <0.01) among the genes that gained methylation; two of these (Bdnf, Gdnf) were associated with binding sites of the insulator protein CTCF (Supplementary Material, Table S4). We also investigated the number of colitis-associated genes in our gene lists through a PubMed search (http://www.ncbi.nlm.nih.gov/pubmed/). Compared with genes in the control group, however, those undergoing methylation changes were no more likely to be associated with colitis in previous literature.
Our targeted gene expression studies identified a low proportion of genes at which developmental changes in DNA methylation were associated with expression changes (2/7, ~30% of the genes tested). We therefore conducted a broader screen using whole genomic gene expression microarrays. A total of 266 genes with an accession number in RefGen showed increased expression (Supplementary Material, Table S5), and 379 showed decreased expression in P90 versus P30 colonic mucosa (Supplementary Material, Table S6). The increased expression of Smad3 found in our RT–PCR studies (Supplementary Material, Fig. S5), was confirmed by the expression microarray. Among the genes with decreased expression we were particularly interested in Stat1 since its expression has been shown to be increased in the colonic mucosa of IBD patients (50,51). Subsequent real time RT–PCR measurements confirmed the ~50% decline in the mRNA level of Stat1 from P30 to P90 on the microarray (Fig. 4).
GO analyses revealed that the genes with increased expression were less involved in sensory perception, but more involved in phosphorous metabolism, biopolymer processing and negative regulation of biological processes than the rest of the genome. Genes with decreased expression were less associated with sensory perception and signal transduction, but more involved in nucleotide/nucleic acid, biopolymer metabolic processes and cell cycle, as well as response to DNA damage (GO biological process at level 4; adjusted P-value <0.01). A significant differential association with several transcription factors (Pax6, Major T-antigen, Cdx-2, Srebp-1, USF, Evi-1, Ppar direct repeat 1, Pit-1) was also identified among the genes with modified expression, implicating the rearrangement of the colonic mucosal transcriptional machinery during pediatric development.
We identified all genes (upstream and downstream) nearest to the SmaI/XmaI intervals that underwent methylation change from P30 to P90, and examined their overlap with the genes that concurrently changed expression. Of 271 genomic intervals with a change in methylation, only 25 (9%) were associated with an expression change at a neighboring transcript (Supplementary Material, Table S7). A major reason for this limited overlap is that only ~50% of the genes that changed expression were associated with a potentially informative SmaI/XmaI interval. An additional possible explanation is that while these methods are robust, they do have inherent limitations in sensitivity and specificity. For example, the decreased expression of Lgals1, identified by the candidate gene approach, was not detected by the gene expression microarray approach used in the study.
DNA methylation in colonic mucosa has been extensively studied in the context of colon cancer, in some cases with implications for IBD. An age dependent increase in the methylation of CpG dense genomic regions (commonly referred to as CpG islands) has been observed at genes involved in colorectal oncogenesis. Even normal appearing colonic mucosa from colorectal cancer patients displays significant epigenetic aberrations, suggesting that those precede and potentially contribute to tumor development (52,53). Similar findings have been observed in the inflamed and cancerous mucosa of IBD (54,55). These appear to be non-IBD specific, however, and can even be less common than in sporadic colorectal tumors (56).
In spite of the extensive epigenetic studies in colon cancer, there is limited current knowledge on normal colonic mucosal epigenetic development and how it may relate to other gastrointestinal disorders such as IBD. Moreover, the timing of epigenetic differentiation and maturation is of great relevance to the developmental origins hypothesis, which postulates that environmental influences during critical periods of development cause permanent changes in organismic structure and function that can influence adult metabolism and disease risk (9). Mammalian epigenetic mechanisms appear to be most sensitive to environment when undergoing developmental establishment or maturation (10,57). Epigenetic changes during infancy and adolescence may therefore reveal important, environmentally labile pathogenic processes and help explain the peak incidence of IBD in young adulthood. Our study is the first to address genome wide epigenetic developmental changes during normal postnatal gastrointestinal development in mammals.
Approximately 1% of the interrogated genomic regions showed DNA methylation changes from late infancy to young adulthood, indicating that epigenetic development in the colonic mucosa proceeds beyond infancy in mice. Importantly, both a germfree transfer model (38) and our current data on DSS exposure indicate that susceptibility to colitis increases during this period. Among genes that gained methylation we found an overrepresentation of genes involved in kidney development, but the expression of those tested was already undetectable at P30. Nevertheless, these findings are consistent with increased methylation at these genomic regions serving to perpetuate the transcriptional silencing of associated genes. PAX genes, for example, are important during embryogenesis, but their constitutive expression promotes tissue hyperplasia. In particular, reactivation of PAX8 expression has been observed in several cancers, including colon cancer (59). Increased methylation at Pax8 may therefore contribute to its persistent transcriptional silencing, stabilize the differentiated state and protect against malignant transformation. Our recent findings in a mouse model favor this reinforcing, stabilizing role of DNA methylation changes during development (30). In further support of this paradigm, two of the six genes involved in kidney development (Bdbf, Gdnf ) are associated with CTCF insulator sites, which have been shown to be important regulators of tumor development (44). Indeed, both Bdnf (60) and Gdnf (61) participate in the regulation of cancer. Additionally, the genomic regions with change in methylation during pediatric development were significantly more likely to associate with CTCF insulator sites in general, compared with the control SmaI/XmaI intervals. Our data support earlier observations suggesting functional links between developmental epigenetics and cancer epigenetics (42), and indicate that relevant developmental processes in mammalian colonic mucosa proceed well beyond infancy.
Our results, both with the targeted assessment of candidate genes and with the gene expression microarray analysis, indicated that most of the DNA methylation changes we identified were not associated with expression changes at nearby genes. One potential explanation is that DNA methylation changes may in fact regulate expression of genes located quite distally. Histone modifications at enhancer elements as far as 200 kb from TSSs correlate with cell type gene expression (43), and as mentioned earlier, imprinting control elements can regulate allele specific gene expression over hundreds of kilobases (47). Cooperative interactions between epigenetic modifications have been shown to influence even chromosomal/chromatin rearrangements (62,63). Hence, DNA methylation changes not directly related to the expression of nearby genes may have significant, yet less easily identifiable, physiologic consequences.
We did discover coordinated developmental changes in DNA methylation and expression of galectin-1 (Lgals1) and Smad3, underscoring the potential pathophysiological relevance of developmental epigenetics in the colonic mucosa. Lgals1 expression has been shown to decrease following chemical induction of colitis in mice, and recombinant human GAL-1 protein protected against colitis in that model system by inducing activated T-cell apoptosis (48). In murine models, decreased expression of Smad3 has been shown to support epithelial recovery and inhibit colorectal fibrosis following chemical induction of colitis (49,58). Although the DNA methylation loss associated with Smad3 was low (~3%), it was very tightly regulated between P30 and P90 and correlated with an increased expression of the gene (~1.4-fold). Similar observations have been made in human colonic mucosal T cells, in which a ~5% decrease in DNA methylation was associated with a 3-fold increase in IFNG expression (23). Therefore, the decreased expression of galectin-1 (decreased protection) and increased expression of Smad3 (decreased regenerative capability following mucosal injury) may contribute to the greater colitis susceptibility of P90 mice, exemplifying the potential physiologic relevance of the epigenetic changes we have identified.
Although most of the identified changes in gene expression were not correlated with methylation changes, they may nonetheless yield important insights into developmentally regulated GI pathogenesis. For example, our results show that STAT1 expression, which is elevated in the inflamed colonic mucosa of CD patients (50,51), decreases during murine post-infant development.
As discussed in the introduction, IBD is likely to result from an exaggerated immune response against the intestinal microflora that is transmitted by the intestinal epithelium. This pathologic response appears to be modulated by genetic and epigenetic factors and highlights the physiologic importance of a flexible but well regulated crosstalk between the immune system and the intestinal microbiome. Therefore, it is very plausible that a dynamic and complex interaction between the immune system, the gut epithelium and the luminal flora shapes the age dependent epigenetic changes of colonocytes. Our findings herein reveal potential elements of this intricate network.
In conclusion, this work provides a compendium of genomic regions that undergo pediatric epigenetic maturation and expression changes in mammalian colonic mucosa, which may be pathophysiologically relevant in different intestinal disorders including IBD and cancer. Our findings indicate a dynamic and age dependent epigenetic programming in the genome of colonic epithelial cells. These changes proceed far beyond infancy, parallel an increased predisposition to colitis and may reinforce colonocyte differentiation. Future studies will be required to address the full degree of the physiologic correlates of this epigenetic maturation, and the potential susceptibility of these DNA methylation and gene expression changes to environmental influences such as nutrition. Such studies will advance our understanding of the epigenetic basis of gastrointestinal development and disease.
Postnatal age 21 days (P21) and P82 C57BL/6J male mice (Jackson Laboratories, Bar Harbor, ME, USA) were provided free access to standard rodent diet (2920X, Harlan-Teklad, Madison, WI, USA) within the same room of our animal facility for 8–9 days to minimize potential environmental, microbial and nutritional differences between the two age groups. At P30 and P90, mice were killed by CO2 asphyxiation between 9:00 and 11:00 AM without any previous food restriction. Colons were removed and placed on ice. Colonic mucosa (all mucosa distal to the cecum) was collected rapidly (31), flash frozen on dry ice and stored at −80°C. For histological evaluations, mouse colons and scrapings were placed in 10% formaldehyde for more than 24 h before processing. All applicable institutional and governmental regulations concerning the ethical use of animals were followed. The protocol was approved by the Institutional Animal Care and Use Committee of Baylor College of Medicine.
At both P30 and P90, male mice (10 per group) received 3% (wt/vol) DSS (32) (MW = 36 000–50 000, MP Biomedicals, LLC, Solon, OH, USA) dissolved in their drinking water ad libitum for 5 days, followed by 5 days of regular drinking water. DSS of this molecular weight induces diffuse colitis from cecum to distal large bowel (33). Control mice received regular drinking water throughout. Mice were weighed daily during and after the DSS exposure. Five mice from each group were killed at day 5, and the others at day 10 of the experiment. Colonic lengths were measured and the colons were evaluated by histology.
Formaldehyde fixed tissues were paraffin embedded, sectioned and hematoxylin-eosin (HE) stained according to standard pathological procedures. HE and cytokeratin staining was carried out on cell-blocks from fixed mucosal scrapings by standard procedures (34). Histological scoring of coded samples in our DSS colitis studies was performed by two blinded pathologists (RS, NT) who consulted and agreed in the scores according to Albert et al. (32). This grading system incorporates inflammatory cell infiltration (0–3), tissue damage (0–5) and edema (0–2).
Genomic DNA was isolated by standard proteinase-k digestion and phenol-chloroform extraction and bisulfite converted as described previously (10). Total RNA was isolated with RNA Stat-60 (Tel-Test, Inc., Friendswood, TX, USA) and stored at −80°C (10).
MSAM was carried out exactly as previously described (30); two independent P90 versus P30 co-hybridizations were performed (http://www.ncbi.nlm.nih.gov/geo/, accession number: GSE18031). Because MSAM is based upon serial digestion of genomic DNA with the methylation sensitive/insensitive isoschizomers SmaI and XmaI, followed by ligation-mediated PCR, we designed a custom microarray (Agilent Technologies, Santa Clara, CA, USA). The array includes 90 535 probes covering 77% of the 31 019 SmaI/XmaI intervals between 200 bp and 2 kb in the mouse genome (average 3.8 probes per interval). The average signal intensity and median P-value from the multiple probes within each SmaI/XmaI interval were calculated. An interval was considered a ‘hit’ if it showed a >1.6-fold change and a median P-value log ratio <0.003 in both P90 versus P30 co-hybridizations.
Bisulfite pyrosequencing (35) was performed at 11 gene-associated MSAM hits (defined as 3000 bp upstream from TSS or within the gene). The genomic regions of interest were amplified prior to pyrosequencing utilizing a universal biotinylated primer approach (36). The primer sequences are listed in Supplementary Material, Table S1. Methylation measurements at both of the flanking SmaI/XmaI sites of the gene associated intervals (AK012806, Cyp27b1, Epha6, Nrbp2, Lgals1, Pax8, Pigl, Slc39a14, Smad3, Spry1) were obtained, except for Ajap1, where DNA methylation was measured in a stretch of CpG sites within the interval. Initial pyrosequencing studies focused on the four samples used in the MSAMs, to determine the MSAM-pyrosequencing correlation. The measurements were subsequently extended to a larger number of p30 and p90 mice. Confirmation of complete bisulfite conversion was routinely integrated into the pyrosequencing assays. Whereas MSAM is allele-specific (both SmaI/XmaI sites on a specific allele must be methylated for the SmaI/XmaI interval to be amplified) bisulfite pyrosequencing provides no allele-specific information. Therefore, at SmaI/XmaI intervals where both informative CpGs changed in methylation, an algorithm was used to compare methylation ratios obtained by bisulfite pyrosequencing with hybridization ratios obtained by MSAM (30).
Total RNA was reverse transcribed using random priming (Multiscribe reverse transcriptase, Applied Biosystems, Branchburg, NJ, USA). Quantitative real-time PCR was performed using the following TaqMan gene expression assays (Applied Biosystems), according to the manufacturer's instructions: Ajap1 (Mm01263768_m1), B-actin (Mm00607939_s1), Lgals1 (Mm00839408_g1), Nrbp2 (Mm00522922_m1), Pax8 (Mm00440623_m1), Slc39a14 (Mm00522697_m1), Smad3 (Mm00489638_m1), Spry1 (Mm01285700_m1) and Stat1 (Mm00439531_m1). Expression changes were quantified relative to beta-actin as an endogenous control, using the 2−ΔΔCt method. 2−ΔΔCt values differing from the group mean by more than 3 standard deviations were omitted as outliers.
Four different P90/P30 whole genomic expression microarray comparisons (Agilent Technologies, Santa Clara, CA, USA: 4 × 44k Whole Mouse Genome microarray, Quick-Amp two-color labeling kit) were performed with 0.4 µg total RNA/sample according to the manufacturer's recommendations (for details see: http://www.ncbi.nlm.nih.gov/geo/, accession number: GSE19506). The average probe signal intensities and mean P90/P30 signal intensity ratios were calculated for every gene on the four microarrays. At transcripts where mean probe intensities were <150, the intensities were increased to 150 to eliminate false positives due to noise. Thereafter, paired two-tailed t-tests were performed on the P90/P30 transcript average probe intensities. Genes with P90/P30 ratio means >1.2 or <0.8 and with P < 0.05 were considered.
Unpaired, two tailed t-tests and odds ratio calculations were utilized in the group comparisons. Statistical significance was declared at P < 0.05. Error bars represent standard error of the mean (SEM). Fatigo functional enrichment analysis (http://babelomics.bioinfo.cipf.es/) (37) was used to identify GO classifications significantly over- or underrepresented among the gene lists; cited P-values were adjusted for multiple testing. Searches for enhancer and CTCF site overlaps were performed using the VISTA Enhancer Browser (http://enhancer.lbl.gov/frnt_page.shtml) and CTCFDSBD (http://insulatordb.utmem.edu/), respectively. The search for overlaps with DNAse hypersensitive sites utilized the ENCODE Univ. of Washington DNaseI Hypersensitivity by Digital DnaseI browser (http://genome.ucsc.edu/cgi-bin/hgTrackUi?db=hg18&g=wgEncodeUwDnaseSeq; J.A. Stamatoyannopoulos, unpublished). The data obtained from Caco2 cells (colorectal adenocarcinoma) with HotSpot algorithm in two replicates (1 and 2) were used. The presence of enhancers, CTCF sites and DNAse hypersensitive sites was assessed in genomic intervals encompassing the SmaI/XmaI intervals that changed methylation from P30 to P90, and 2 kb upstream and downstream. All reported genomic coordinates are based on mm8 (RefGen; http://genome.ucsc.edu).
R.K. was supported by funding from the Broad Medical Research Program, the Broad Foundation (IBD-0252) and a young investigator joint award from the Crohn's and Colitis Foundation of America-Children's Digestive Health and Nutrition Foundation/North American Society of Pediatric Gastroenterology Hepatology and Nutrition (CCFA Ref #2426). This work was also supported by NIH grant R01-DK081557, research grant #1-FY08-392 from the March of Dimes Foundation, and USDA CRIS #6250-51000-055 to R.A.W. DNaseI mapping data were funded through NHGRI ENCODE grant U54HG004592 to John Stamatoyannopoulos, University of Washington.
We would like to thank Adam Gillum for his valuable help with figure preparation.
Conflict of Interest statement. None declared.