Limits...
Identification of thyroid hormone receptor binding sites in developing mouse cerebellum.

Gagne R, Green JR, Dong H, Wade MG, Yauk CL - BMC Genomics (2013)

Bottom Line: We found that while the occurrence of the TRE motif is strongly correlated with gene regulation by TH for some genes, other TH-regulated genes do not exhibit an increased density of TRE half-site motifs.Furthermore, we demonstrate that an increase in the rate of occurrence of the half-site motifs does not always indicate the specific location of the TRE within the promoter region.While we have identified 85 putative TREs within these regions, future work will study other mechanisms of action that may mediate the remaining observed TR-binding activity.

View Article: PubMed Central - HTML - PubMed

Affiliation: Environmental Health Science and Research Bureau, Healthy Environments and Consumer Safety Branch, Health Canada, Ottawa, ON K1A 0L2, Canada.

ABSTRACT

Background: Thyroid hormones play an essential role in early vertebrate development as well as other key processes. One of its modes of action is to bind to the thyroid hormone receptor (TR) which, in turn, binds to thyroid response elements (TREs) in promoter regions of target genes. The sequence motif for TREs remains largely undefined as does the precise chromosomal location of the TR binding sites. A chromatin immunoprecipitation on microarray (ChIP-chip) experiment was conducted using mouse cerebellum post natal day (PND) 4 and PND15 for the thyroid hormone receptor (TR) beta 1 to map its binding sites on over 5000 gene promoter regions. We have performed a detailed computational analysis of these data.

Results: By analysing a recent spike-in study, the optimal normalization and peak identification approaches were determined for our dataset. Application of these techniques led to the identification of 211 ChIP-chip peaks enriched for TR binding in cerebellum samples. ChIP-PCR validation of 25 peaks led to the identification of 16 true positive TREs. Following a detailed literature review to identify all known mouse TREs, a position weight matrix (PWM) was created representing the classic TRE sequence motif. Various classes of promoter regions were investigated for the presence of this PWM, including permuted sequences, randomly selected promoter sequences, and genes known to be regulated by TH. We found that while the occurrence of the TRE motif is strongly correlated with gene regulation by TH for some genes, other TH-regulated genes do not exhibit an increased density of TRE half-site motifs. Furthermore, we demonstrate that an increase in the rate of occurrence of the half-site motifs does not always indicate the specific location of the TRE within the promoter region. To account for the fact that TR often operates as a dimer, we introduce a novel dual-threshold PWM scanning approach for identifying TREs with a true positive rate of 0.73 and a false positive rate of 0.2. Application of this approach to ChIP-chip peak regions revealed the presence of 85 putative TREs suitable for further in vitro validation.

Conclusions: This study further elucidates TRβ gene regulation in mouse cerebellum, with 211 promoter regions identified to bind to TR. While we have identified 85 putative TREs within these regions, future work will study other mechanisms of action that may mediate the remaining observed TR-binding activity.

Show MeSH

Related in: MedlinePlus

Differences between peak width true values and splitter predicted values for the Whitehead dataset. a. Difference between the true and predicted start values for the true positives in the Whitehead dataset. b. Difference between the true and predicted end values for the true positives in the Whitehead dataset. On the y-axis, density is the empirical estimate of the underlying probability density function.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3716714&req=5

Figure 4: Differences between peak width true values and splitter predicted values for the Whitehead dataset. a. Difference between the true and predicted start values for the true positives in the Whitehead dataset. b. Difference between the true and predicted end values for the true positives in the Whitehead dataset. On the y-axis, density is the empirical estimate of the underlying probability density function.

Mentions: For each predicted peak, Splitter outputs a chromosome number, a starting location, and an ending location. Using the optimal threshold value of the Whitehead dataset, 68 out of 100 true peaks were successfully predicted by Splitter. For each predicted peak, the difference between the predicted and actual start and end of the genome location were calculated to check for systematic over- or under-estimation of peak width. Since a true positive was labelled when any overlap occurred between true and predicted peak regions (Figure 2), histograms were generated for the differences in predicted and true DNA segment extents. Figures 4a&b show the positional biases from the peak start and end prediction, calculated as (true start genome location - predicted start genome address) and (true end genome address - predicted end genome address) respectively. These histograms reveal a strong bias towards under-predicting peak extents, particularly for the estimated peak 3’ end. With these findings in mind, all peaks predicted by Splitter in our own dataset were corrected by adding 200 bp to the 3’ end, widening the peak.


Identification of thyroid hormone receptor binding sites in developing mouse cerebellum.

Gagne R, Green JR, Dong H, Wade MG, Yauk CL - BMC Genomics (2013)

Differences between peak width true values and splitter predicted values for the Whitehead dataset. a. Difference between the true and predicted start values for the true positives in the Whitehead dataset. b. Difference between the true and predicted end values for the true positives in the Whitehead dataset. On the y-axis, density is the empirical estimate of the underlying probability density function.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3716714&req=5

Figure 4: Differences between peak width true values and splitter predicted values for the Whitehead dataset. a. Difference between the true and predicted start values for the true positives in the Whitehead dataset. b. Difference between the true and predicted end values for the true positives in the Whitehead dataset. On the y-axis, density is the empirical estimate of the underlying probability density function.
Mentions: For each predicted peak, Splitter outputs a chromosome number, a starting location, and an ending location. Using the optimal threshold value of the Whitehead dataset, 68 out of 100 true peaks were successfully predicted by Splitter. For each predicted peak, the difference between the predicted and actual start and end of the genome location were calculated to check for systematic over- or under-estimation of peak width. Since a true positive was labelled when any overlap occurred between true and predicted peak regions (Figure 2), histograms were generated for the differences in predicted and true DNA segment extents. Figures 4a&b show the positional biases from the peak start and end prediction, calculated as (true start genome location - predicted start genome address) and (true end genome address - predicted end genome address) respectively. These histograms reveal a strong bias towards under-predicting peak extents, particularly for the estimated peak 3’ end. With these findings in mind, all peaks predicted by Splitter in our own dataset were corrected by adding 200 bp to the 3’ end, widening the peak.

Bottom Line: We found that while the occurrence of the TRE motif is strongly correlated with gene regulation by TH for some genes, other TH-regulated genes do not exhibit an increased density of TRE half-site motifs.Furthermore, we demonstrate that an increase in the rate of occurrence of the half-site motifs does not always indicate the specific location of the TRE within the promoter region.While we have identified 85 putative TREs within these regions, future work will study other mechanisms of action that may mediate the remaining observed TR-binding activity.

View Article: PubMed Central - HTML - PubMed

Affiliation: Environmental Health Science and Research Bureau, Healthy Environments and Consumer Safety Branch, Health Canada, Ottawa, ON K1A 0L2, Canada.

ABSTRACT

Background: Thyroid hormones play an essential role in early vertebrate development as well as other key processes. One of its modes of action is to bind to the thyroid hormone receptor (TR) which, in turn, binds to thyroid response elements (TREs) in promoter regions of target genes. The sequence motif for TREs remains largely undefined as does the precise chromosomal location of the TR binding sites. A chromatin immunoprecipitation on microarray (ChIP-chip) experiment was conducted using mouse cerebellum post natal day (PND) 4 and PND15 for the thyroid hormone receptor (TR) beta 1 to map its binding sites on over 5000 gene promoter regions. We have performed a detailed computational analysis of these data.

Results: By analysing a recent spike-in study, the optimal normalization and peak identification approaches were determined for our dataset. Application of these techniques led to the identification of 211 ChIP-chip peaks enriched for TR binding in cerebellum samples. ChIP-PCR validation of 25 peaks led to the identification of 16 true positive TREs. Following a detailed literature review to identify all known mouse TREs, a position weight matrix (PWM) was created representing the classic TRE sequence motif. Various classes of promoter regions were investigated for the presence of this PWM, including permuted sequences, randomly selected promoter sequences, and genes known to be regulated by TH. We found that while the occurrence of the TRE motif is strongly correlated with gene regulation by TH for some genes, other TH-regulated genes do not exhibit an increased density of TRE half-site motifs. Furthermore, we demonstrate that an increase in the rate of occurrence of the half-site motifs does not always indicate the specific location of the TRE within the promoter region. To account for the fact that TR often operates as a dimer, we introduce a novel dual-threshold PWM scanning approach for identifying TREs with a true positive rate of 0.73 and a false positive rate of 0.2. Application of this approach to ChIP-chip peak regions revealed the presence of 85 putative TREs suitable for further in vitro validation.

Conclusions: This study further elucidates TRβ gene regulation in mouse cerebellum, with 211 promoter regions identified to bind to TR. While we have identified 85 putative TREs within these regions, future work will study other mechanisms of action that may mediate the remaining observed TR-binding activity.

Show MeSH
Related in: MedlinePlus