Functional analysis of novel snoRNA-mRNA interactions

In humans and all other organisms, the genetic information is encoded in the sequence of extremely long molecules of DNA. Together, these molecules comprise the genome, which can be divided into functional units termed genes. However, the DNA is essentially a permanent storage medium, so for this information to be used it must be copied into a "working memory". This copying process is termed transcription and takes place in the cell nucleus. The copies take the form of long molecules that are closely related chemically to DNA, and are called RNA. It is essential for all organisms that the integrity of information encoded in the DNA be maintained, whereas information copied into the RNA is disposable. This greatly expands the range of uses to which it can be put, and there are many different functional classes of RNA. In the major information processing system of any cell, a class of RNAs called messenger RNAs (mRNAs) carry the genetic information from the cell nucleus to the cytoplasm, where it is used to programme the protein synthesis machinery. The function, regulation and maturation of the mRNAs have been major topics of research for many years. However, there are many other classes of RNAs that do not encode proteins and are collectively termed non-coding RNAs (ncRNAs). One class of very small ncRNAs, termed micro RNAs (miRNAs), has been shown to play important regulatory roles in humans. Misregulation of miRNA function is associated with many human diseases and this is being studied intensively. However, several other important groups of ncRNAs have key structural and functional roles during protein synthesis and other essential cellular activities. For example, proteins are synthesized in tiny, but intricate, machines called ribosomes, the cores of which are made up of ribosomal RNAs (rRNAs). The ribosomes themselves have a complex synthesis pathway that largely occurs in a specialized sub-nuclear compartment called the nucleolus. Within the nucleolus, the rRNA precursors (pre-rRNAs) undergo a multistep processing pathway and are chemically modified at >100 sites. Another class of ncRNAs, the small nucleolar RNAs (snoRNAs) function during rRNA maturation, either promoting pre-rRNA cleavage or acting as guides for site-specific nucleotide modification. The box C/D class snoRNAs (so called because of conserved sequence motifs) act by forming stable, extended base-paired interactions with targets RNAs. In addition to their vital roles in ribosome synthesis, there is evidence that box C/D snoRNAs can also interact with the nuclear precursors to mRNAs (pre-mRNA). In some cases this results in strong down-regulation of target mRNA levels, while other interactions impede "splicing" reactions needed to convert pre-mRNAs into functional mRNAs. Notably, loss of the snoRD115 family of box C/D snoRNAs in the genetic disease Prader-Willi syndrome impedes processing of the pre-mRNA encoding the serotonin 2C receptor. However, the extent of snoRNA-mRNA interactions remains unclear, as is the general impact of snoRNA binding on mRNA processing and/or stability. Strikingly, there are around 1000 different snoRNAs in human cells that do not have a known target RNA. Some of these have been conserved in evolution, showing that their unknown targets are functionally important, and therefore worth discovering. Using a newly developed technique we have generated preliminary data demonstrating that we can precisely identify hundreds of different sites at which snoRNAs bind to pre-mRNAs and mRNAs. We can do this in human cells, in yeast and in mouse stem cells that are differentiating into neurons. We are therefore in a position to characterise this potentially important, but previously largely unexplored, regulatory system in mechanistic detail.

Many regulatory, non-protein coding RNAs (ncRNAs) function via RNA-RNA base-pairing, but these interactions have been difficult to identify experimentally or using bioinformatics. To address this, we developed a robust technique for the identification of in vivo RNA-RNA interactions, termed crosslinking, ligation and sequencing of hybrids (CLASH). Applying CLASH to human microRNAs (miRNAs) precisely mapped 18,000 mRNA target sites. Here we propose to apply this approach to box C/D small nucleolar RNAs (snoRNAs). In human cells around 60 box C/D snoRNAs direct site-specific rRNA and snRNA modification, but around 1,000 other snoRNA-like species are termed "orphans" since they lack known targets. This number is comparable to the number of known human miRNAs. Our initial CLASH data indicate that box C/D snoRNAs bind many pre-mRNAs in human, mouse and yeast cells. In human cells, the targets are predominately pre-mRNA introns of genes that show alternative splicing. This suggests the testable hypothesis that snoRNA binding alters alternative splicing patterns. We will use bioinformatics, biochemistry and reporter constructs to determine the mechanism of snoRNA-induced change in pre-mRNA splicing regulation and/or mRNA stability. The largest number of human snoRNA-mRNA interactions were found for the U3 snoRNA, which is 20-50 fold more abundant than other snoRNAs and we hypothesise that this difference in abundance reflects its very numerous non-rRNA targets. In yeast, the largest numbers of mRNA hits were found for snR190, which is conserved to humans but has no known function in yeast ribosome synthesis. We hypothesise that snoRNAs link pre-mRNA processing to the ribosome synthesis rate, which is a key sensor of cellular growth state. Dynamic changes in newly identified snoRNA interactions will be followed during alterations in ribosome synthesis, cell growth, stress, and during neuronal differentiation, with the aim of determining their roles in these key processes.

We anticipate that the work proposed here will lead to the development of major new lines of research with medical relevance. Loss of snoRD115 is linked to the genetic disease Prader-Willi syndrome, while loss of the U32a/U33/U35a cluster confers increased resistance to lipotoxic and oxidative stress. We know that snoRNAs can alter mRNA stability and pre-mRNA splicing, and it seems very likely that that medically relevant interactions will be uncovered among the large numbers of snoRNA targets that we will identify. In particular, the reports that many orphan snoRNAs, with no currently known targets, are specifically expressed in the human brain, strongly indicates the existence of numerous brain-specific regulatory interactions. We expect the targets of neuronal-specific snoRNAs to be uncovered in the mouse cell analyses during this project. This should pave the way for future work in human neurons.
The identification of miRNAs as important regulators has led to the establishment of several companies that specifically exploit their therapeutic potential. We are optimistic that the targets for snoRNAs will also be of significant medical relevance, offering starting point for similar therapeutic intervention. It is my understanding that the use of synthetic snoRNAs for mRNA depletion has been patented by the University of Dundee, underlining the robust silencing of gene expression that can be conferred by snoRNA-mRNA interactions. Better understanding of the authentic targets for endogenous snoRNAs would greatly enhance the relevance of such technology.
For in vivo UV crosslinking we make use of two, custom-built devices, which were designed to irradiate either 0.7l or 2.2l culture volumes. These were developed in collaboration with UVO3 Ltd. (St. Ives, Cams.), who constructed the prototypes and have since commercialized these units. In recent detailed discussions with the Managing Director of UVO3, Peter Wadsworth, we have agreed on a new design that should allow shorter irradiation times, better suited for rapid kinetic analyses. The prototype of this model is currently being fabricated by UVO3, and will be supplied to us, free of charge, for testing and evaluation. If this performs as expected, it will also be commercialised by UVO3.

Tool development
The pipeline that we have developed for the identification of chimeric sequences, which represent RNA-RNA hybrids, will be of use in future analyses by many groups. We have found that running these analyses on our previous CRAC analyses, which were performed solely to identify protein-RNA interactions, frequently also identifies novel RNA-RNA interactions. Notably, this is not the case for published CLIP datasets, due to technical differences in the ligation steps during the two protocols. An initial report of the CLASH bioinformatics pipeline has been submitted for publication (Travis, A.J. Moody, J. Helwak, A. Tollervey, D. and Kudla, G. (submitted) Hyb: a bioinformatics pipeline for the analysis of CLASH (crosslinking, ligation and sequencing of hybrids) data) and we will make this software generally available. However, it currently operates in a command line mode, whereas its ease of use would be greatly enhanced by adapting the software to run on a Galaxy server. This will be one of the tasks of the Bioinformatician and should be of benefit to many research groups with an interest in RNA based regulation, including miRNAs and lncRNAs, in addition to those following up the snoRNA analyses proposed here.