Limits...
Correlation of microRNA levels during hypoxia with predicted target mRNAs through genome-wide microarray analysis.

Guimbellot JS, Erickson SW, Mehta T, Wen H, Page GP, Sorscher EJ, Hong JS - BMC Med Genomics (2009)

Bottom Line: To date, few studies have investigated an environmental perturbation for effects on genome-wide miRNA levels, or their consequent influence on mRNA output.Target prediction programs and expression profiling techniques do not yet adequately represent the complexity of miRNA-mediated gene repression, and new methods may be required to better elucidate these pathways.Our data suggest the physiologic impact of miRNAs on cellular transcription results from a multifaceted network of miRNA and mRNA relationships, working together in an interconnected system and in context of hundreds of RNA species.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Genetics, University of Alabama at Birmingham, Birmingham, AL 35294, USA. jsguimbellot@uasom.uab.edu

ABSTRACT

Background: Low levels of oxygen in tissues, seen in situations such as chronic lung disease, necrotic tumors, and high altitude exposures, initiate a signaling pathway that results in active transcription of genes possessing a hypoxia response element (HRE). The aim of this study was to investigate whether a change in miRNA expression following hypoxia could account for changes in the cellular transcriptome based on currently available miRNA target prediction tools.

Methods: To identify changes induced by hypoxia, we conducted mRNA- and miRNA-array-based experiments in HT29 cells, and performed comparative analysis of the resulting data sets based on multiple target prediction algorithms. To date, few studies have investigated an environmental perturbation for effects on genome-wide miRNA levels, or their consequent influence on mRNA output.

Results: Comparison of miRNAs with predicted mRNA targets indicated a lower level of concordance than expected. We did, however, find preliminary evidence of combinatorial regulation of mRNA expression by miRNA.

Conclusion: Target prediction programs and expression profiling techniques do not yet adequately represent the complexity of miRNA-mediated gene repression, and new methods may be required to better elucidate these pathways. Our data suggest the physiologic impact of miRNAs on cellular transcription results from a multifaceted network of miRNA and mRNA relationships, working together in an interconnected system and in context of hundreds of RNA species. The methods described here for comparative analysis of cellular miRNA and mRNA will be useful for understanding genome wide regulatory responsiveness and refining miRNA predictive algorithms.

No MeSH data available.


Related in: MedlinePlus

Correlation of miRNAs with at least one target site in miRanda-predicted target mRNA. In each panel, the reference distribution of t-statistics from all probes of the mRNA expression arrays is given by a dotted line. These are compared with the distribution of t-statistics for those probes that are predicted targets of given miRNAs, shown as a solid line. 'N' represents the number of transcripts in the reference sample (dotted line), and therefore is the same in each plot. Sample size (n) is the number of genes predicted to have target(s) of given microRNA(s) (solid line), and therefore changes from plot to plot. Note that n depends on the number of predicted target probes contained within the dataset, either combined among all miRNAs, or specific to an individual miRNA. The plotted distributions are Gaussian kernel density estimates (loosely, smoothed histograms), and the indicated bandwidth is in terms of the standard deviation of the smoothing kernel. The x-axis in each figure reflects the t-statistics for the comparison between groups (dotted line represents random distribution). The y-axis represents the density of observations at a given t-statistic value. a. Combined t-statistics for all significantly changed miRNAs with 1 site in the 3'UTR of predicted target miRNAs. b. Combined t-statistics for a set of non-significant miRNAs with 1 predicted target site. c. Representative significantly changed miRNA compared to all of its predicted targets. d. Representative non-significantly changed miRNA compared to all of its predicted targets.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2667434&req=5

Figure 3: Correlation of miRNAs with at least one target site in miRanda-predicted target mRNA. In each panel, the reference distribution of t-statistics from all probes of the mRNA expression arrays is given by a dotted line. These are compared with the distribution of t-statistics for those probes that are predicted targets of given miRNAs, shown as a solid line. 'N' represents the number of transcripts in the reference sample (dotted line), and therefore is the same in each plot. Sample size (n) is the number of genes predicted to have target(s) of given microRNA(s) (solid line), and therefore changes from plot to plot. Note that n depends on the number of predicted target probes contained within the dataset, either combined among all miRNAs, or specific to an individual miRNA. The plotted distributions are Gaussian kernel density estimates (loosely, smoothed histograms), and the indicated bandwidth is in terms of the standard deviation of the smoothing kernel. The x-axis in each figure reflects the t-statistics for the comparison between groups (dotted line represents random distribution). The y-axis represents the density of observations at a given t-statistic value. a. Combined t-statistics for all significantly changed miRNAs with 1 site in the 3'UTR of predicted target miRNAs. b. Combined t-statistics for a set of non-significant miRNAs with 1 predicted target site. c. Representative significantly changed miRNA compared to all of its predicted targets. d. Representative non-significantly changed miRNA compared to all of its predicted targets.

Mentions: While it is well established that miRNAs play an important role governing gene expression, the correlation of mRNA and miRNA profiles across the entire genome and in response to an acute cellular perturbation (such as hypoxia) have not been thoroughly assessed. Accordingly, we next evaluated the extent to which the profile of miRNAs observed during hypoxia was consonant with mRNA expression (i.e., the transcriptome response). The distribution of t-statistics for all probes on mRNA expression arrays was used as a reference distribution (dotted line), as shown in Figure 3. In comparison to the reference distribution, the thirteen most significant differentially expressed miRNAs (Tables 4 and 5) and twenty randomly selected non-significant miRNAs were analyzed. The distribution of t-statistics for mRNA probes that are predicted targets of the given miRNAs were generated (solid lines, Figures 3a and 3b). Figure 3 depicts the distribution of mRNAs with one predicted target site for a given miRNA (microrna.org); a deviation from the reference curve (dotted lines) would be interpreted as a significant relationship between the mRNA and miRNA arrays, as mediated via the target predictions. This methodological approach to investigating 'cause and effect' allows a visualization of mRNA/miRNA correlations. No consistent deviations, however, were observed in the combined t-statistics. As a further test of this finding, evaluations of individual miRNAs were performed (representative examples in Figures 3c and 3d). No consistent deviations were noted in t-statistics for any single miRNA in either the significant or the non-significant miRNA group. Figure 3 illustrates the distribution for miRanda-predicted targets, while Figure 4 through Figure 6 contain the results for PicTar, TargetScanS, and miRBase, respectively. When all miRNAs that changed significantly during oxygen restriction were compared against mRNAs containing one target prediction site in the 3'UTR, none of the target prediction programs indicated a significant relationship between data from the mRNA and miRNA expression arrays. When the findings were evaluated using the most recent TargetScanS predictive method (i.e. a more stringent threshold [40] that includes only the highest ranked targets with a context score percentile of 85%), no consistent deviations were observed in the combined t-statistics (Figure 5).


Correlation of microRNA levels during hypoxia with predicted target mRNAs through genome-wide microarray analysis.

Guimbellot JS, Erickson SW, Mehta T, Wen H, Page GP, Sorscher EJ, Hong JS - BMC Med Genomics (2009)

Correlation of miRNAs with at least one target site in miRanda-predicted target mRNA. In each panel, the reference distribution of t-statistics from all probes of the mRNA expression arrays is given by a dotted line. These are compared with the distribution of t-statistics for those probes that are predicted targets of given miRNAs, shown as a solid line. 'N' represents the number of transcripts in the reference sample (dotted line), and therefore is the same in each plot. Sample size (n) is the number of genes predicted to have target(s) of given microRNA(s) (solid line), and therefore changes from plot to plot. Note that n depends on the number of predicted target probes contained within the dataset, either combined among all miRNAs, or specific to an individual miRNA. The plotted distributions are Gaussian kernel density estimates (loosely, smoothed histograms), and the indicated bandwidth is in terms of the standard deviation of the smoothing kernel. The x-axis in each figure reflects the t-statistics for the comparison between groups (dotted line represents random distribution). The y-axis represents the density of observations at a given t-statistic value. a. Combined t-statistics for all significantly changed miRNAs with 1 site in the 3'UTR of predicted target miRNAs. b. Combined t-statistics for a set of non-significant miRNAs with 1 predicted target site. c. Representative significantly changed miRNA compared to all of its predicted targets. d. Representative non-significantly changed miRNA compared to all of its predicted targets.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2667434&req=5

Figure 3: Correlation of miRNAs with at least one target site in miRanda-predicted target mRNA. In each panel, the reference distribution of t-statistics from all probes of the mRNA expression arrays is given by a dotted line. These are compared with the distribution of t-statistics for those probes that are predicted targets of given miRNAs, shown as a solid line. 'N' represents the number of transcripts in the reference sample (dotted line), and therefore is the same in each plot. Sample size (n) is the number of genes predicted to have target(s) of given microRNA(s) (solid line), and therefore changes from plot to plot. Note that n depends on the number of predicted target probes contained within the dataset, either combined among all miRNAs, or specific to an individual miRNA. The plotted distributions are Gaussian kernel density estimates (loosely, smoothed histograms), and the indicated bandwidth is in terms of the standard deviation of the smoothing kernel. The x-axis in each figure reflects the t-statistics for the comparison between groups (dotted line represents random distribution). The y-axis represents the density of observations at a given t-statistic value. a. Combined t-statistics for all significantly changed miRNAs with 1 site in the 3'UTR of predicted target miRNAs. b. Combined t-statistics for a set of non-significant miRNAs with 1 predicted target site. c. Representative significantly changed miRNA compared to all of its predicted targets. d. Representative non-significantly changed miRNA compared to all of its predicted targets.
Mentions: While it is well established that miRNAs play an important role governing gene expression, the correlation of mRNA and miRNA profiles across the entire genome and in response to an acute cellular perturbation (such as hypoxia) have not been thoroughly assessed. Accordingly, we next evaluated the extent to which the profile of miRNAs observed during hypoxia was consonant with mRNA expression (i.e., the transcriptome response). The distribution of t-statistics for all probes on mRNA expression arrays was used as a reference distribution (dotted line), as shown in Figure 3. In comparison to the reference distribution, the thirteen most significant differentially expressed miRNAs (Tables 4 and 5) and twenty randomly selected non-significant miRNAs were analyzed. The distribution of t-statistics for mRNA probes that are predicted targets of the given miRNAs were generated (solid lines, Figures 3a and 3b). Figure 3 depicts the distribution of mRNAs with one predicted target site for a given miRNA (microrna.org); a deviation from the reference curve (dotted lines) would be interpreted as a significant relationship between the mRNA and miRNA arrays, as mediated via the target predictions. This methodological approach to investigating 'cause and effect' allows a visualization of mRNA/miRNA correlations. No consistent deviations, however, were observed in the combined t-statistics. As a further test of this finding, evaluations of individual miRNAs were performed (representative examples in Figures 3c and 3d). No consistent deviations were noted in t-statistics for any single miRNA in either the significant or the non-significant miRNA group. Figure 3 illustrates the distribution for miRanda-predicted targets, while Figure 4 through Figure 6 contain the results for PicTar, TargetScanS, and miRBase, respectively. When all miRNAs that changed significantly during oxygen restriction were compared against mRNAs containing one target prediction site in the 3'UTR, none of the target prediction programs indicated a significant relationship between data from the mRNA and miRNA expression arrays. When the findings were evaluated using the most recent TargetScanS predictive method (i.e. a more stringent threshold [40] that includes only the highest ranked targets with a context score percentile of 85%), no consistent deviations were observed in the combined t-statistics (Figure 5).

Bottom Line: To date, few studies have investigated an environmental perturbation for effects on genome-wide miRNA levels, or their consequent influence on mRNA output.Target prediction programs and expression profiling techniques do not yet adequately represent the complexity of miRNA-mediated gene repression, and new methods may be required to better elucidate these pathways.Our data suggest the physiologic impact of miRNAs on cellular transcription results from a multifaceted network of miRNA and mRNA relationships, working together in an interconnected system and in context of hundreds of RNA species.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Genetics, University of Alabama at Birmingham, Birmingham, AL 35294, USA. jsguimbellot@uasom.uab.edu

ABSTRACT

Background: Low levels of oxygen in tissues, seen in situations such as chronic lung disease, necrotic tumors, and high altitude exposures, initiate a signaling pathway that results in active transcription of genes possessing a hypoxia response element (HRE). The aim of this study was to investigate whether a change in miRNA expression following hypoxia could account for changes in the cellular transcriptome based on currently available miRNA target prediction tools.

Methods: To identify changes induced by hypoxia, we conducted mRNA- and miRNA-array-based experiments in HT29 cells, and performed comparative analysis of the resulting data sets based on multiple target prediction algorithms. To date, few studies have investigated an environmental perturbation for effects on genome-wide miRNA levels, or their consequent influence on mRNA output.

Results: Comparison of miRNAs with predicted mRNA targets indicated a lower level of concordance than expected. We did, however, find preliminary evidence of combinatorial regulation of mRNA expression by miRNA.

Conclusion: Target prediction programs and expression profiling techniques do not yet adequately represent the complexity of miRNA-mediated gene repression, and new methods may be required to better elucidate these pathways. Our data suggest the physiologic impact of miRNAs on cellular transcription results from a multifaceted network of miRNA and mRNA relationships, working together in an interconnected system and in context of hundreds of RNA species. The methods described here for comparative analysis of cellular miRNA and mRNA will be useful for understanding genome wide regulatory responsiveness and refining miRNA predictive algorithms.

No MeSH data available.


Related in: MedlinePlus