Abstract
Sufficient information exists within a set of tumor samples to detect endogenous correlations between miRNA and mRNA levels. Based on the validations, the causal arrow for these correlations is likely to be directed from the miRNAs to the mRNAs. From these data sets, we inferred and validated a tumor suppression pathway linked to miR-181c.
Introduction
MicroRNAs (miRNAs) are small non-coding RNAs that play a critical role in post-transcriptional gene regulation, primarily by binding to messenger RNAs (mRNAs) to induce degradation or inhibit translation. Traditionally, our understanding of miRNA function has relied heavily on "gain-of-function" or "loss-of-function" experiments. While these methods—which involve over-expressing or silencing a specific miRNA—are powerful, they often utilize concentrations far beyond physiological levels. This raises questions about whether the observed effects accurately reflect natural biological processes in vivo.
This study investigates whether the subtle, endogenous fluctuations of miRNA and mRNA levels across human tissue samples contain a detectable regulatory signal. By analyzing primary brain tumor biopsies, the researchers aimed to map functional regulatory networks without the artifacts introduced by artificial manipulation.
Methodology
The researchers analyzed 12 human primary brain tumor biopsies, encompassing a range of glioma grades. The experimental design integrated two high-throughput profiling techniques:
- mRNA Expression: Measured using microarrays covering approximately 14,500 genes.
- miRNA Profiling: Conducted via multiplex RT-PCR for 330 distinct miRNAs.
To identify potential regulatory pairs, Pearson correlation coefficients ($r$) were calculated for every possible miRNA-mRNA combination, totaling over 650,000 pairs. To distinguish biological signal from stochastic noise, the observed correlation distribution was compared against a mathematically derived "random case" and 100 random permutations of the data.
Validation of the "causal arrow" was performed in U251 glioblastoma cell lines. High-correlation pairs were selected for experimental testing, where miRNAs were either transfected or suppressed to observe corresponding changes in mRNA levels.
Results and Discussion
The Statistical Signal of Regulation
The primary finding was the presence of "fat tails" in the correlation distribution. In a purely random system, high correlations ($|r| > 0.9$) would be extremely rare. However, the real tissue data yielded 35 pairs with high negative correlations and 37 with high positive correlations. The probability of this occurring by chance was calculated at $P < 5 \times 10^{-21}$, indicating a robust biological signal.
Nature of the Correlations
The study revealed distinct patterns based on the direction of correlation:
- Negative Correlations ($r < -0.9$): These were significantly enriched for predicted miRNA targets (using TargetScan), suggesting direct mRNA degradation.
- Positive Correlations ($r > 0.9$): These often identified intronic miRNAs. These miRNAs are located within the introns of "host genes" and are co-transcribed with them, leading to naturally synchronized expression levels.
- Indirect Effects: Interestingly, the majority of high-correlation pairs were not direct targets. This suggests that the strongest signals in in vivo data often represent downstream "ripples" of regulation, where a miRNA influences a key regulator that subsequently shifts a broad transcriptional profile.
miR-181c and the p53 Pathway
A significant discovery was the role of miR-181c as a potential tumor suppressor. In the biopsy data, miR-181c levels positively correlated with the histone acetyltransferase PCAF and the tumor suppressor p53. Experimental validation showed that transfecting miR-181c into cancer cells led to a 30% decrease in cell viability and a 12% increase in apoptosis. The data suggests a pathway where miR-181c indirectly stabilizes p53 by increasing PCAF protein levels, offering a novel therapeutic target for glioblastoma.
Conclusion and Limitations
This research demonstrates that endogenous miRNA-mRNA variations in human tissue are not merely noise but contain structured information about regulatory networks. However, the study's small sample size ($n=12$) limits the ability to detect more subtle interactions. Furthermore, while the authors provide evidence for a miRNA-to-mRNA causal direction, complex biological systems likely involve feedback loops that require further investigation. Future work with larger cohorts and diverse tissue types will be essential to fully map the human "regulome."
Summary updated