Are microRNAs located in genomic regions associated with cancer?

Philippe Lamy, Claus L. Andersen, Lars Dyrskjøt, Niels Tørring, Torben Ørntoft, Carsten WiufBritish Journal of Cancer 2006 · DOI

TL;DR

This study investigates whether microRNA (miRNA) genes are preferentially located in genomic regions that commonly undergo copy number changes in cancer. While prostate and colon cancers show miRNAs enriched in amplified regions and depleted in deleted regions, bladder cancer exhibits a reversed pattern. Surprisingly, miRNA gene location didn't correlate with their expression levels, suggesting distinct regulatory mechanisms compared to mRNA.

What Problem Does This Paper Address?

MicroRNAs (miRNAs) are small non-coding RNAs that play crucial roles in regulating gene expression, and accumulating evidence suggests their involvement in cancer. Previous research indicated that human miRNA genes are often found near "fragile sites" and in genomic regions associated with cancer. However, these studies were not specific to particular cancer types, leaving open the question of whether this association holds true across different cancers and if it directly relates to miRNA expression levels. This paper aims to address this by examining the genomic location of miRNAs in relation to copy number alterations in three specific cancer types: prostate, bladder, and colon, and then correlating these findings with miRNA expression data.

Methods

The researchers utilized Affymetrix SNP arrays (10k and 50k) to analyze DNA copy number changes in samples from prostate, bladder, and colon tumors, as well as normal tissue. They employed the dChip software for intensity extraction and normalization. Copy number alterations (gain, loss, or normal) were determined based on standard deviations from the normal baseline. Genomic regions were then defined by consecutive SNPs exhibiting these copy number states. Finally, they mapped the locations of 283 known human miRNAs onto these altered genomic regions to assess for over- or under-representation. To investigate the link between copy number and expression, they compared their copy number data with previously published miRNA expression profiles from other studies.

Datasets

The study used two main types of datasets:

  • Genomic DNA copy number data: Generated from 128 samples using Affymetrix GeneChip Mapping 10k Early Access Arrays and 143 samples using Affymetrix GeneChip Mapping 50K arrays. These included normal tissues and samples from prostate, bladder, and colon tumors.
  • miRNA expression data: Publicly available data from previous studies (Volinia et al., 2006; Lu et al., 2005) were used for comparison.

Key Insights and Ideas

  • Cancer-type specific localization: The relationship between miRNA gene location and copy number alterations is not uniform across all cancers. Prostate and colon tumors show a consistent pattern, while bladder cancer presents a contrasting scenario.
  • Copy number vs. expression disconnect: A surprising finding is the lack of a statistically significant correlation between miRNA gene copy number status and their expression levels. This challenges the intuitive assumption that copy number changes directly drive miRNA expression, at least in the context of these cancers.
  • Alternative regulatory mechanisms: The observed disconnect suggests that miRNA expression is likely regulated by mechanisms other than, or in addition to, simple gene copy number changes. This could involve post-transcriptional regulation, epigenetic modifications, or other complex pathways.

Key Results

  • In prostate and colon tumors, miRNAs were found to be significantly over-represented in regions with copy number gain and under-represented in regions with copy number loss (P<0.00001 for colon, P<0.00001 for prostate).
  • In bladder tumors, a slight under-representation of miRNAs in gain regions was observed (P=0.0089 for all bladder tumors, P=0.0085 for stage T1 tumors).
  • When comparing miRNA copy number data with published expression data, no statistically significant relationship was found between differentially expressed miRNAs and the copy number alterations in their genomic locations for either prostate or colon tumors (P-values > 0.18).

Limitations and Future Work

  • Sample Discrepancies: The study acknowledges that the copy number data and the miRNA expression data were obtained from different sample sets. While the authors argue their samples are representative, this difference could potentially contribute to the observed discrepancies.
  • Data Validity: Although the methods for copy number determination using SNP arrays are considered reliable, the authors implicitly raise the question of data validity by discussing potential explanations for the unexpected results.
  • Future Directions: The paper strongly suggests that miRNA expression is regulated by mechanisms distinct from those governing mRNA expression. Future research should focus on elucidating these alternative regulatory pathways, exploring epigenetic modifications, and investigating the role of miRNA targets and their genomic locations. Further studies with matched copy number and expression data from the same samples would also be valuable.

Summary updated

Abstract

We report on the location of 283 miRNAs in the human genome in relation to copy number changes in three distinct types of tumours: prostate, bladder and colon. In prostate and colon tumours, we find miRNAs over-represented in regions with copy number gain and under-represented in regions with copy number loss. Surprisingly this pattern appears to be reversed in bladder cancer. We compared our miRNA copy number data to published miRNA expression data; unexpectedly, we did not find a statistically significant relationship between miRNA copy number and expression level. This suggests that miRNA expression is regulated through different mechanisms than mRNA expression.
BibTeX
@article{lamy2006are,
  author = {Philippe Lamy and Claus L. Andersen and Lars Dyrskjøt and Niels Tørring and Torben Ørntoft and Carsten Wiuf},
  title = {{Are microRNAs located in genomic regions associated with cancer?}},
  journal = {British Journal of Cancer},
  year = {2006},
  doi = {10.1038/sj.bjc.6603381},
}
Figure