Regulatory RNAs: Have mRNA Untranslated Regions Joined the Party

Thomas Hughes and Pamela Jones discuss the implications of a new study on the 5' non-coding region of vascular endothelial growth factor mRNA and its influence on gene expression.

I n the last decade or so, we have become familiar with the discovery of new classes of nuclearencoded regulators (see Glossary) as fundamental controllers of gene expression. Each new discovery has again highlighted the incomplete nature of our understanding of the genome and its regulation. First, small interfering RNAs (siRNAs) demonstrated that RNA molecules are not merely components of the cellular machinery (such as tRNAs and rRNAs) or gene-expression intermediates (mRNAs), but can function as potent trans-acting regulators of specifi c genes. MicroRNAs (miRNAs) continued this theme and now attract much attention among basic scientists and clinicians alike, both as potential regulators of most human genes and as potential diagnostic tools. In this issue of PLoS Medicine, a research article by Shigetada Teshima-Kondo and colleagues supports the suggestion that another class of regulatory RNAs exists [1]. Furthermore, this class could be the largest to date, since potential members are contained within every mRNA.

A New Study on the 5' Non-Coding Region of VEGF mRNA
The authors show that the 5' untranslated region (UTR) of the vascular endothelial growth factor (VEGF) mRNA infl uences expression of other genes, and thereby cell function, independently of the VEGF reading frame. Many UTRs have long been known to function as cis-acting elements on the expression of their own mRNAs [2,3]-but trans-acting regulatory functions are an exciting possibility, which, if commonplace, would be every bit as infl uential as siRNAs or miRNAs.
VEGF is a key regulator of cancer development, acting to promote survival of tumour cells and to stimulate angiogenesis; as such, VEGF is well established as a target for cancer therapies [4]. The starting point for this new study was the observation that some uses of therapeutics targeting VEGF protein were less effective than might have been hoped [5]. The authors optimistically (but seemingly correctly) interpreted this observation as a suggestion that VEGF transcripts might have cancer-related functions separable from the functions of the VEGF protein, and set about further analyses.
Working initially in tissue culture, the authors showed that VEGF promoted resistance to apoptosis (programmed cell death) induced by the chemotherapeutic 5-fl uoro-uracil. This was not surprising given VEGF's familiar anti-apoptotic function [6]. However, in a thorough and detailed set of experiments, the authors also showed that expression of VEGF protein alone was insuffi cient for the full anti-apoptotic effect seen, and furthermore, that expression of the VEGF 5'UTRs in isolation from the VEGF reading frame promoted signifi cant resistance.
These experiments are especially complex since the VEGF 5'UTR is known to have a variety of cis-acting effects on translation of its reading frame [7]. Notably and unusually, translation can initiate from within the 5'UTR at a number of CUG codons [8]. The authors have established that translation of sequences from within the UTR is not required to specify this apoptotic resistance, and thereby that the activity is truly associated with the RNA itself. Directing their attentions more towards a role for the VEGF 5'UTR in carcinogenesis, the authors showed that expression of the UTR, again without its reading frame, induced increased colony-forming activity in vitro, and tumour-forming ability in a mouse xenograft model. Finally, the authors established a link between expression of VEGF 5'UTR and a dramatic change in response to interferon alpha using both in vitro and in vivo models. The conclusion is that VEGF 5'UTR has independent tumourpromoting activity, acting via mediators  Shigetada Teshima-Kondo and colleagues fi nd that cancer cells have a survival system that is regulated by vegf mRNA and that vegf mRNA and its protein may synergistically promote the malignancy of tumor cells.
of interferon alpha signalling, and therefore that the VEGF mRNA itself is oncogenic.

Implications of the Study
While the idea that mRNAs might have functions that are independent of their translation may be surprising, there are published precedents for related trans-acting functions of UTRs. Such examples remain few and far between (Teshima-Kondo et al. cite four, and we have found a further two [9,10]) and have not had a signifi cant impact on the general consciousness of researchers. This new work represents an important advance, in that the function takes place in the context of a key player in carcinogenesis that is already used as a cancer therapy target, and therefore the study highlights this class of regulation. What this new study shares with its precedents, however, is the lack of a defi ned mechanism of action. As such it is diffi cult to determine the generality of trans-acting functions of UTRs, and it is too early to start considering mRNAs as another class of fundamental gene regulators.

How Do UTRs Exert Trans-Acting Infl uences?
It is possible to postulate a variety of mechanisms by which UTRs might exert trans-acting infl uences. First, the RNA-activated protein kinase (PKR) may be activated by binding to doublestranded regions of UTRs. Activated PKR is well known to cause changes in the translational effi ciency of certain mRNAs as part of the cellular response to viral infection, but more recently has been implicated in other pathways, including cancer-related mitogenactivated protein kinase signalling [11]. PKR has been shown to bind to one published trans-acting UTR [12], and Teshima-Kondo et al. report preliminary data (although these are not shown) indicating that PKR may bind to the VEGF mRNA.
Alternatively, rather than activating a binding protein, trans-acting UTRs may sequester binding proteins away from other targets. In experiments where the UTR sequence has been over-expressed, as is the case for almost all these studies, this over-expression could allow artifactual effects, which is a concern for the physiological relevance of the observations. Some authors have speculated about UTR sequences making direct RNA:RNA interactions with target transcripts and thereby infl uencing their stability or translation. This mechanism is possible, but whether separate mRNA molecules, each of which would form its own mRNP structure, would have the opportunity to base-pair with each other with any signifi cant frequency seems debatable. Finally, for some 5'UTRs the trans-acting function appears to be dependent on translation of short upstream open reading frames (uORFs) within the UTR. The resultant peptides repress translation of other mRNAs, although seemingly only of those mRNAs arising from the same gene [9,10]. Whether this peptide-mediated transrepression is truly this gene-specifi c remains to be determined. This fi nal proposed mechanism could provide an explanation for the trans-repression seen from the c-myc P0 5'UTR [13], which contains a uORF. However, it could not account for the infl uence of VEGF 5'UTRs, since Teshima-Kondo et al. demonstrate, at least in vitro, that their effect is maintained in the absence of signifi cant translation.

Conclusion
This study has a clear implication for VEGF-targeted therapies, in that strategies to target the mRNA may be required to inhibit the full oncogenic infl uence of the gene. More generally, the study highlights a little-studied gene regulatory pathway that may yet prove to be far more commonplace than currently appreciated. c-myc P0: c-myc is a proto-oncogene that is very frequently over-expressed in cancers. The gene is transcribed from a number of different promoters, one of which is termed P (for promoter) 0. miRNA: microRNAs are a class of short (~21 nucleotide) single-stranded RNAs that can reduce the expression of target mRNAs to which they are partially complementary.
mRNA: messenger RNAs are transcribed copies of protein-encoding genes that are exported from the nucleus and translated into proteins by ribosomes in the cytoplasm. mRNP: mRNPs are mRNAs that are packaged into messenger ribonucleoprotein particles during nuclear export.

Nuclear-encoded regulators:
regulators of gene expression that are encoded by the nucleus, the most obvious example being the transcription factors. Non-nuclear-encoded regulators of gene expression include hormones and drugs.
Protein kinases: enzymes capable of phosphorylating specifi c target proteins, leading to changes in their function.
rRNA: ribosomal RNAs are structural components of ribosomes.
siRNA: short interfering RNAs are a class of short (~21 nucleotide) double-stranded RNAs that can reduce the expression of target mRNAs to which they are complementary.
Trans-acting: trans-acting regulatory molecules are those that infl uence the activity of other physically separate molecules.
tRNA: transfer RNAs are essential components of the translational machinery. They bind to individual amino acids and recruit them to extending peptide chains at ribosomes.
Upstream open reading frames: sequences of nucleotides within 5'UTRs that could potentially be translated.