Overview

An overview of the lab's research activities.

Image
Illustration of functional link between transcription and splicing

We are studying pre-messenger RNA (pre-mRNA) splicing in the budding yeast Saccharomyces cerevisiae. Pre-mRNA splicing takes place in a very large RNA-protein complex, the spliceosome, within which there are subparticles, including the small ribonucleoprotein particles (snRNPs: U1, U2, U4, U5 and U6), each composed of a small nuclear RNA (snRNA) and a set of proteins. The spliceosome is a highly dynamic molecular machine. During spliceosome assembly and during the course of the splicing reactions many dynamic RNA-RNA and RNA-protein conformational changes occur that are regulated by proteins. Proteins also regulate the specificity, accuracy and efficiency of the splicing process. We have investigated the functions of a number of key splicing factors, characterising their molecular interactions in the spliceosome. In addition, we are using more quantitative systems biology approaches to study the flow of RNA through the various RNA processing pathways. This allowed the development of stochastic models for splicing and transcription. Recently, we optimised a metabolic labelling protocol that allows monitoring the splicing and turnover of nascent transcripts with high kinetic resolution. Currently, we are investigating functional links between different RNA metabolic pathways. In particular, we have identified a novel link between transcription and splicing, and we proposed the existence of splicing-dependent transcriptional checkpoints. More recently, we have been studying links between splicing and chromatin, using the auxin inducible degron system to knock-down splicing factors and investigate the effect on histone methylation. We use biochemical, cell biological and genetic approaches, including in vitro splicing assays, yeast two-hybrid screens, quantitative real-time RT-PCR, chromatin immunoprecipitation (ChIP), ChIP-seq and RNA-seq.

 

 

Image
 RNA polymerase II (POL II) pauses near the 3' ends of introns while splicing occurs.
Fig.1 RNA polymerase II (POL II) pauses near the 3' ends of introns while splicing occurs. We propose that this may be associated with a splicing-dependent transcriptional checkpoint. Completion of splicing signals satisfaction of the checkpoint. The amber and green Ps represent phosphorylation of serine 5 and serine 2 respectively in the carboxy-terminal domain (CTD) of the Pol II large subunit (Alexander et al., 2010).

A functional link between transcription and splicing

In eukaryotic cells there is evidence for functional coupling between transcription and processing of pre-mRNAs. An important goal in our lab is to identify factors involved in coupling transcription and splicing and to elucidate the mechanism of coupling and its potential role in promoting efficiency and fidelity in gene expression. To better understand this coupling we performed a high-resolution kinetic analysis of transcription and splicing in budding yeast. This revealed that shortly after induction of transcription, RNA polymerase accumulates transiently around the 3’ end of the intron on two reporter genes. This apparent transcriptional pause coincides with splicing factor recruitment to the gene and with the first detection of spliced mRNA, and is repeated periodically thereafter. Pausing requires productive splicing, as it is lost upon mutation of the intron and restored by suppressing the splicing defect in trans. The carboxy-terminal domain of the paused polymerase large subunit is hyper-phosphorylated on serine 5, and phosphorylation of serine 2 is first detected here. Phosphorylated polymerase also accumulates around the 3’ splice sites of constitutively expressed, endogenous yeast genes. We propose the existence of splicing-dependent transcriptional checkpoints (Figure 1) that are associated with Pol II pausing. (Figure 1; Alexander et al., Mol Cell, 2010).

Transcriptional pausing may allow more time for quality control to be carried out. If quality is unsatisfactory, the checkpoint is not satisfied and the RNA is degraded, thereby avoiding production of defective proteins. We discovered that certain defects in specific splicing factors can give rise to transcription defects, with Pol II accumulating over introns. The accumulated Pol II is hyper-phosphorylated on serine 5 (pSer5) of the CTD, a characteristic of paused or stalled transcription (Chathoth et al., 2014). The concept of splicing-dependent transcriptional checkpoints is novel, although it resembles the transcriptional checkpoint for capping mRNAs, in which Pol II with pSer5 pauses while the capping enzymes modify the 5’ end of the nascent RNA. We speculate that similar checkpoints might be associated with the acceptance/rejection of alternative splice sites in higher eukaryotes. This is compatible with evidence that the rate of transcription can affect alternative splice site usage.

Eight RNA-stimulated ATPases are involved in splicing. At least six of these (Prp2p, Prp5p, Prp16p, Prp22p, Prp28p, Prp43p; Figure 2) have quality control functions. They both promote conformational changes and monitor the accuracy of molecular interactions at different stages of the splicing process. We propose that transcriptional checkpoints are associated with the quality control activities of these ATPases.

Image
The spliceosome assembly/disassembly pathway, showing RNA stimulated ATPases known to have quality control functions.
Fig.2 The spliceosome assembly/disassembly pathway, showing RNA stimulated ATPases known to have quality control functions. Also, Cus2 and Prp45 that have properties consistent with roles in coupling transcription and splicing.

 

A Cus2p dependent checkpoint links pre-spliceosome formation with transcription

The Prp5p ATPase controls interaction of U2 snRNA with the intron’s branchpoint sequence as the pre-spliceosome forms (Figure 3). Our checkpoint hypothesis is supported by the finding that a temperature-sensitive prp5-1 mutation, that blocks pre-spliceosome formation, causes a transcription defect at 37oC, with pSer5 Pol II accumulating on introns. The U2-associated Cus2p, which should be displaced from the U2 snRNP by the ATPase activity of Prp5p, remains in a defective splicing complex. Significantly, deletion of CUS2 suppresses the transcription defect but not the splicing defect caused by prp5-1. Therefore, the transcription defect is not simply a direct consequence of a splicing defect but is Cus2p-dependent. We propose that Cus2p is a checkpoint factor that signals the status of pre-spliceosome formation to the transcription machinery. Tat-SF1, the human orthologue of Cus2p, is also involved with both transcription and splicing.  

Image
Cartoon showing a possible intermediate complex in pre-spliceosome formation.
Fig.3 Cartoon showing a possible intermediate complex in pre-spliceosome formation. Cus2 and the U2 snRNP join the U1 snRNP on the intron. Prp5 causes a conformational change in U2 snRNA, which interacts with the branchpoint (BP) sequence to form the pre-spliceosome. Prp5 hydrolyses ATP, displacing Cus2. See Chathoth et al., 2014.

 

Image
UCSC genome browser screen shots show the change in distribution of reads at different labelling times (y-axis) for RPL28, RPL39
Figure 4. UCSC genome browser screen shots show the change in distribution of reads at different labelling times (y-axis) for RPL28, RPL39 and RPS13, with annotation below in blue. Exons are represented by blue boxes and intron indicated by a blue line. Steady-state data were generated by sequencing total RNA. (Adapted from Barrass et al., 2015, Figure 6)

Quantitative analyses of transcription and splicing

Increasingly, we are applying more systematic approaches, including functional genomic approaches using deep sequencing for genome-wide analyses of transcription, co-transcriptional splicing and chromatin modification (ChIP-seq), RNA-protein interactions (CRAC), and more quantitative and kinetic analyses of RNA processing, using real-time quantitative RT-PCR and RNA-seq. Related to this, Jean coordinated an EC-funded Framework Programme 6 (FP6) project, "RiboSys", in which we, and our collaborators, used systems biology approaches to model pre-messenger RNA metabolism in Saccharomyces cerevisiae. Jean was also a partner in the EC-funded FP6 Network of Excellence in Alternative Splicing, EURASNET, and in the EC-funded FP7 project "UNICELLSYS".

To facilitate more quantitative analyses, David Barrass in our lab optimised a metabolic labelling method that allows the measurement of nascent RNA metabolism in yeast with unprecedented kinetic resolution. Incorporation of 4-thiouracil into nascent RNA for short times, biotinylation and affinity selection of the thiolated, newly synthesised RNA, followed by RNA-seq permits the kinetics of RNA processing to be monitored. Data analysis is performed in collaboration with Dr Sander Granneman (Synthsys) and Dr Guido Sanguinetti (Informatics Forum, Edinburgh). In this way, we identified features in RNA that affect pre-mRNA splicing kinetics and turnover of non-coding RNAs (Barrass et al., 2015).                                                

We have compared the speed of splicing of different pre-mRNAs obtained by this method, with independent next-generation sequencing methods from other laboratories, that jointly quantify transcription and splicing in budding yeast, in particular, with efficiency of co-transcriptional splicing, derived from nascent RNA-seq analysis (Harlen et al., 2016) and the distance (nt) of Pol II from 3’ splice sites when splicing is 90% complete (Carrillo Oesterreich at al., 2016). We noted a striking point of agreement between these datasets regarding the splicing of ribosomal protein transcripts. Ribosomal protein transcripts splice faster, more co-transcriptionally and with higher fidelity than other intron-containing transcripts, (Figure 5).

 

 

 

Image
Ribosomal protein transcripts (blue) tend to be spliced faster and more co-transcriptionally than other transcripts (red).
Figure 5. Ribosomal protein transcripts (blue) tend to be spliced faster and more co-transcriptionally than other transcripts (red). Distance in nts of Pol II from 3’splice sites when 90% of transcripts are spliced (Carrillo Oesterreich et al., 2016) plotted versus fraction spliced co-transcriptionally (Harlen et al., 2016) (A), or (B) versus speed of splicing (AUC) by 4tU-seq (Barrass et al 2015). (C) Fraction spliced versus AUC. (from Wallace & Beggs, 2017)

 

Functional interactions between splicing transcription and chromatin

Image
Illustration of Functional interactions between splicing transcription and chromatin

Bioinformatics studies found that tri-methylation of lysine residue 36 in histone H3 (H3K36me3) is enriched at the boundaries between introns and exons in higher eukaryotic genes, potentially marking the positions of introns. There was also some evidence that histone modifications can influence splicing and alternative splicing decisions (reviewed in Kornblihtt et al, 2009, NSMB, 16:902-903). On the other hand, inhibition of splicing in human cells was found to impair recruitment of the H3K36 methyltransferase, Setd2, reducing the H3K36me3 mark on intron-containing genes (de Almeida et al., 2011, NSMB) and/or shifting the distribution of H3K36me3 towards the 3’ ends of genes (Kim et al., 2011, PNAS). Currently, there is no mechanistic information about how or why splicing affects chromatin modifications nor whether these effects are mediated by direct interaction between chromatin and the splicing machinery.

We are investigating potential links between splicing and chromatin modification in budding yeast. Using auxin-induced degrons to knock-down splicing factors, we find that splicing defects reproducibly reduce trimethylation of H3K36 and Set2 recruitment at endogenuos genes. We aim to characterise the molecular interactions that mediate these effects.