Limits...
DNA methylation and gene expression in Mimulus guttatus.

Colicchio JM, Miura F, Kelly JK, Ito T, Hileman LC - BMC Genomics (2015)

Bottom Line: Additionally, we find that DNA methylation is significantly depleted near gene transcriptional start sites, which may explain the recently discovered elevated rate of recombination in these same regions.Using a model-based approach, we demonstrate that methylation patterns are an important predictor of variation in gene expression.This model provides a novel approach for differential methylation analysis that generates distinct and testable hypotheses regarding gene expression.

View Article: PubMed Central - PubMed

Affiliation: Department of Ecology and Evolutionary Biology, University of Kansas, Lawrence, KS, 66045, USA. Colicchio@ku.edu.

ABSTRACT

Background: The presence of methyl groups on cytosine nucleotides across an organism's genome (methylation) is a major regulator of genome stability, crossing over, and gene regulation. The capacity for DNA methylation to be altered by environmental conditions, and potentially passed between generations, makes it a prime candidate for transgenerational epigenetic inheritance. Here we conduct the first analysis of the Mimulus guttatus methylome, with a focus on the relationship between DNA methylation and gene expression.

Results: We present a whole genome methylome for the inbred line Iron Mountain 62 (IM62). DNA methylation varies across chromosomes, genomic regions, and genes. We develop a model that predicts gene expression based on DNA methylation (R(2) = 0.2). Post hoc analysis of this model confirms prior relationships, and identifies novel relationships between methylation and gene expression. Additionally, we find that DNA methylation is significantly depleted near gene transcriptional start sites, which may explain the recently discovered elevated rate of recombination in these same regions.

Conclusions: The establishment here of a reference methylome will be a useful resource for the continued advancement of M. guttatus as a model system. Using a model-based approach, we demonstrate that methylation patterns are an important predictor of variation in gene expression. This model provides a novel approach for differential methylation analysis that generates distinct and testable hypotheses regarding gene expression.

No MeSH data available.


Related in: MedlinePlus

Correlations between DNA methylation and gene expression. A single star represents a significant linear correlation, two stars a significant second-order correlation, and three stars a third order correlation. The red dashed lines represent the means, the black line represents the regression line, and the blue line represents 95 % confidence intervals
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4492170&req=5

Fig5: Correlations between DNA methylation and gene expression. A single star represents a significant linear correlation, two stars a significant second-order correlation, and three stars a third order correlation. The red dashed lines represent the means, the black line represents the regression line, and the blue line represents 95 % confidence intervals

Mentions: A stepwise cubic polynomial model was selected to predict log(gene expression) based on minimum BIC. Out of a possible 454 parameters, the minimum BIC criterion selected a model with 29 factors that explained (R2) 20.1 % of the variation in log transformed expression values (SS Model: 1764, SS Error: 6981, F28,17042 = 153.6, p < 0.0001, Tables 2, 3 and 4, Fig. 5, Additional file 1: Figure S1). Including all 454 parameters increases R2 only marginally (to 23.3 %), and the minimum calculated R2 calculated in 3-fold cross-validation was 17.9 %. Generally, there is an excess of genes predicted to be expressed at log-transformed values between 1.5 and 2.5, that were actually expressed at levels less than 1.2, as well as genes expressed above 4, which this model never predicts (Additional file 1: Figure S1). It is clear that while gene methylation can modify gene expression, it cannot predict the complete repression, or extremely high expression of some genes. As all parameters were Z-transformed prior to modeling, the effect estimates are comparable across variables (Table 4). In order to maintain both statistical and molecular consistency throughout, both Z-transformed values and raw values are reported. The inclusion of both various forms of DNA methylation and gene architecture (number of exons, exon length, intron length) have not been included in a single model explicitly testing their ability to predict gene expression, but their independent effects have often been looked at in relation to gene expression. While it is hard to compare our integrative analysis on gene expression with prior studies, we generally find the same direction of effect in our data as was found in other plant systems [3]. Trends are thus not Mimulus specific, but likely more general effects of DNA methylation on gene expression in angiosperms. Finally, when discussing the role of various forms of methylation on gene expression we often designate a specific type of methylation as having a positive or negative effect on gene expression. In this context that indicates that there was significant predictive ability for a given type of methylation on gene expression. However, due to the nature of this experimental design we cannot definitively define the arrow of causation.Table 2


DNA methylation and gene expression in Mimulus guttatus.

Colicchio JM, Miura F, Kelly JK, Ito T, Hileman LC - BMC Genomics (2015)

Correlations between DNA methylation and gene expression. A single star represents a significant linear correlation, two stars a significant second-order correlation, and three stars a third order correlation. The red dashed lines represent the means, the black line represents the regression line, and the blue line represents 95 % confidence intervals
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4492170&req=5

Fig5: Correlations between DNA methylation and gene expression. A single star represents a significant linear correlation, two stars a significant second-order correlation, and three stars a third order correlation. The red dashed lines represent the means, the black line represents the regression line, and the blue line represents 95 % confidence intervals
Mentions: A stepwise cubic polynomial model was selected to predict log(gene expression) based on minimum BIC. Out of a possible 454 parameters, the minimum BIC criterion selected a model with 29 factors that explained (R2) 20.1 % of the variation in log transformed expression values (SS Model: 1764, SS Error: 6981, F28,17042 = 153.6, p < 0.0001, Tables 2, 3 and 4, Fig. 5, Additional file 1: Figure S1). Including all 454 parameters increases R2 only marginally (to 23.3 %), and the minimum calculated R2 calculated in 3-fold cross-validation was 17.9 %. Generally, there is an excess of genes predicted to be expressed at log-transformed values between 1.5 and 2.5, that were actually expressed at levels less than 1.2, as well as genes expressed above 4, which this model never predicts (Additional file 1: Figure S1). It is clear that while gene methylation can modify gene expression, it cannot predict the complete repression, or extremely high expression of some genes. As all parameters were Z-transformed prior to modeling, the effect estimates are comparable across variables (Table 4). In order to maintain both statistical and molecular consistency throughout, both Z-transformed values and raw values are reported. The inclusion of both various forms of DNA methylation and gene architecture (number of exons, exon length, intron length) have not been included in a single model explicitly testing their ability to predict gene expression, but their independent effects have often been looked at in relation to gene expression. While it is hard to compare our integrative analysis on gene expression with prior studies, we generally find the same direction of effect in our data as was found in other plant systems [3]. Trends are thus not Mimulus specific, but likely more general effects of DNA methylation on gene expression in angiosperms. Finally, when discussing the role of various forms of methylation on gene expression we often designate a specific type of methylation as having a positive or negative effect on gene expression. In this context that indicates that there was significant predictive ability for a given type of methylation on gene expression. However, due to the nature of this experimental design we cannot definitively define the arrow of causation.Table 2

Bottom Line: Additionally, we find that DNA methylation is significantly depleted near gene transcriptional start sites, which may explain the recently discovered elevated rate of recombination in these same regions.Using a model-based approach, we demonstrate that methylation patterns are an important predictor of variation in gene expression.This model provides a novel approach for differential methylation analysis that generates distinct and testable hypotheses regarding gene expression.

View Article: PubMed Central - PubMed

Affiliation: Department of Ecology and Evolutionary Biology, University of Kansas, Lawrence, KS, 66045, USA. Colicchio@ku.edu.

ABSTRACT

Background: The presence of methyl groups on cytosine nucleotides across an organism's genome (methylation) is a major regulator of genome stability, crossing over, and gene regulation. The capacity for DNA methylation to be altered by environmental conditions, and potentially passed between generations, makes it a prime candidate for transgenerational epigenetic inheritance. Here we conduct the first analysis of the Mimulus guttatus methylome, with a focus on the relationship between DNA methylation and gene expression.

Results: We present a whole genome methylome for the inbred line Iron Mountain 62 (IM62). DNA methylation varies across chromosomes, genomic regions, and genes. We develop a model that predicts gene expression based on DNA methylation (R(2) = 0.2). Post hoc analysis of this model confirms prior relationships, and identifies novel relationships between methylation and gene expression. Additionally, we find that DNA methylation is significantly depleted near gene transcriptional start sites, which may explain the recently discovered elevated rate of recombination in these same regions.

Conclusions: The establishment here of a reference methylome will be a useful resource for the continued advancement of M. guttatus as a model system. Using a model-based approach, we demonstrate that methylation patterns are an important predictor of variation in gene expression. This model provides a novel approach for differential methylation analysis that generates distinct and testable hypotheses regarding gene expression.

No MeSH data available.


Related in: MedlinePlus