|Home | About | Journals | Submit | Contact Us | Français|
Splicing is an essential eukaryotic process in which introns are excised from precursors to messenger RNAs and exons ligated together. This reaction is catalyzed by a multi-MegaDalton machine called the spliceosome, composed of 5 small nuclear RNAs (snRNAs) and a core set of ~100 proteins minimally required for activity. Due to the spliceosome’s size, its low abundance in cellular extracts, and its highly dynamic assembly pathway, analysis of the kinetics of splicing and the conformational rearrangements occurring during spliceosome assembly and disassembly has proven extraordinarily challenging. Here, we review recent progress in combining chemical biology methodologies with single molecule fluorescence techniques to provide a window into splicing in real time. These methods complement ensemble measurements of splicing in vivo and in vitro to facilitate kinetic dissection of pre-mRNA splicing.
Removal of introns from nascent RNA transcripts (precursors to messenger RNAs or pre-mRNAs) is an essential step in eukaryotic gene expression. The enzyme responsible for this process is the spliceosome, which carries out intron excision via two energy-neutral transesterification reactions: lariat intron formation and exon ligation. Despite the straightforward nature of the chemistry, the spliceosome itself is an extraordinarily complex, 2–3 MDa machine composed of 5 uridine-rich small nuclear RNAs (U1, U2, U4, U5 and U6 snRNAs) and anywhere from 90 to >300 proteins [1,2]. The snRNAs and a subset of the proteins form stable particles called small nuclear ribonucleoproteins (snRNPs) that constitute the largest building blocks of the spliceosome. Aided by a plethora of more loosely associated proteins (“splicing factors”), the snRNPs interact with one another and the pre-mRNA to complete each round of splicing via an extraordinarily dynamic process. The current model for spliceosome assembly involves step-wise association of first U1, then U2, followed by U4/U6.U5 tri-snRNP and the multi-protein Prp19 complex (NTC). Once all the major pieces are in place, additional structural rearrangements lead to U1 and U4 expulsion, catalytic activation, lariat formation, exon ligation, spliced product release and finally dissociation of the remaining components (Figure 1). Amazingly, this entire sequence is believed to occur anew on every intron in each pre-mRNA molecule, rendering each assembled and catalytically activated spliceosome a single-turnover enzyme.
As has been recently reviewed [3–6], our current understanding of spliceosome assembly is based largely on the procession of stable complexes that form upon addition of a simplified splicing substrate (i.e., two short exons separated by an efficiently spliced intron) to an in vitro splicing reaction. Spliceosomal components are most commonly provided as an S. cerevisiae whole cell extract (WCE) or as a mammalian nuclear cell extract (NCE). Stable complexes are resolved by native polyacrylamide gel electrophoresis (PAGE) or by density gradient centrifugation and can be purified by affinity chromatography. In some cases, such purified complexes retain the ability to carry out an individual step in the overall process (see, for example, [7,8]). Combined with more than two decades of intensive genetic dissection of the yeast spliceosome, these biochemical studies have yielded tremendous insight into the overall composition, structure and operation of the splicesosome. Nonetheless, what was missing until recently was any detailed kinetic information about the comings and goings of individual components and related structural transitions.
At any single moment in a splicing reaction, different spliceosomes are in different states and doing different things. Averaging of these states and behaviors across an entire population of molecules leads to significant loss of information about molecular dynamics. A means around this is to monitor the behavior of each spliceosome individually. Single molecule approaches can yield a wealth of information as to the stochastic and kinetic behaviors of biological systems, their static and dynamic heterogeneities, and the rates of individual steps in multi-step processes. They can also reveal rare and transient species along a reaction pathway that are difficult to detect in ensemble experiments [9,10]. This review focuses on recent advances in the application of single molecule fluorescence microscopy to the pre-mRNA splicing machinery and the novel insights derived from these analyses. Key to the success of these studies are chemical biology approaches for introducing bright fluorophores into nucleic acids and proteins.
Splicing of pre-mRNAs in vitro occurs over tens of minutes. Sustained microscope observation of individual molecules on this time scale requires their immobilization on a surface to prevent diffusion out of the field of view (Figure 2). The most common surface attachment method is a biotin:streptavidin:biotin sandwich, with one biotin molecule being linked to the glass surface (e.g., through a biotinylated polyethylene glycol chain) and the second biotin attached covalently to the pre-mRNA or to an oligonucleotide (oligo) to which the pre-mRNA is hybridized [11–13]. The shortest pre-mRNA routinely used for in vitro splicing assays is 135 nucleotides (nt) , well beyond the current limit for non-enzymatic synthesis of individual RNA molecules. Therefore, covalent attachment of a terminal biotin, as well as incorporation of the internal fluorophores discussed below, requires enzymatic ligation of multiple RNA fragments, one or more of which contains the modification(s) of interest.
Splicing of individual surface-immobilized pre-mRNA molecules has been visualized via incorporation of different fluorophores into different parts of the RNA. Crawford et al. utilized a model RP51A pre-mRNA containing a Cy3 fluorophore in the 5′ exon and an Alexa647 fluorophore in the intron near the 5′ splice site (SS) . In this configuration, splicing is detected in real time by loss of intronic Alexa647 fluorescence and retention of exonic Cy3 fluorescence (Figure 2A). An alternate strategy employed by Abelson et al. used a UBC4 pre-mRNA containing Cy3 in the 5′ exon and Cy5 in the 3′ exon . In this case, splicing was assessed by quenching the splicing reaction and then subjecting the surface-tethered RNAs to oligo-directed RNaseH cleavage . In the presence of a DNA oligo complementary to the intron, RNaseH cleaves the intron in unspliced pre-mRNA molecules to release the 5′ exon Cy3, but it will not cleave spliced mRNA molecules because they lack the intron target sequence (Figure 2B).
A drawback of the above approaches is that the fluorophore-labeled pre-mRNAs are difficult to synthesize, requiring ligation of multiple RNA fragments to form full length, 200–400 nt pre-mRNAs. An alternate strategy for detecting splicing is to hybridize fluorescent oligos to the intron. However, even with locked nucleic acid (LNA) oligos , spontaneous oligo dissociation can occur on the same time scale as splicing, complicating the analysis. Other strategies such as targeting peptide nucleic acids (PNAs)  to the intron or exon-exon junction may prove successful in the future and eliminate the need for costly and time-consuming synthesis of fluorescent splicing reporters.
One important finding of both Crawford et al.  and Abelson et al.  was that the extent and kinetics of splicing of surface-tethered pre-mRNA molecules were similar to that observed in traditional splicing reactions. Thus, results obtained from single molecule experiments are likely significant for ensemble reactions. With viable strategies in place for detecting intron loss from single pre-mRNA molecules, the stage was set for the experiments described below for real time observations of conformational rearrangements within the spliceosome and the comings and goings of individual components.
It has long been appreciated that pre-mRNA conformational changes are a pre-requisite for splicing: minimally the last nucleotide of the 5′ exon must first be juxtaposed with the branch site for lariat formation and subsequently with the 3′SS for exon ligation (Figure 1). While the full extent of conformational changes necessary for transition between the first and second catalytic steps of splicing is not known, there is ample evidence that the spliceosomal active site is flexible with different conformations favoring either lariat formation or exon ligation [16–18]. Single molecule fluorescence energy transfer (SM-FRET) provides a tool to probe these conformational changes during productive splicing.
In a pioneering set of experiments, Abelson et al. used SM-FRET to provide an initial picture of the range of pre-mRNA conformational changes during splicing . With the doubly-labeled UBC4 pre-mRNA described above (i.e., containing a 5′ exon Cy3 donor and a 3′ exon Cy5 acceptor), FRET is expected to increase when the 5′ and 3′ splice sites are brought close together in preparation for the chemical steps of splicing. This study identified a complex network of conformational transitions that included high FRET states consistent with exon proximity (Figure 3A). Importantly, all of the observed conformational transitions were reversible. Thus the pre-mRNA was at no point “locked” into a specific conformation by the spliceosome, but instead fluctuated among points on the pathway, possibly determined by the presence or absence of a given splicing factor. Consequently, the authors hypothesized that the primary role of the DExD/H-box ATPases associated with the spliceosome is not conversion of the spliceosome between a series of static complexes. Rather, these ATPases likely improve fidelity by providing pathways for rejection of incorrect substrates [19,20].
In addition to the insights obtained about pre-mRNA conformational changes during splicing, the Abelson et al. study is noteworthy for two other methodological developments that may prove broadly useful. First, they utilized a novel splicing substrate, UBC4 pre-mRNA, chosen both for its diminutive length (95 nt intron) and because it exhibited high in vivo commitment to mRNA formation subsequent to spliceosome assembly. The latter characteristic was divined by clever use of splicing-dependent microarrays to simultaneously assess genome-wide splicing efficiencies in vivo of many different pre-mRNAs . Secondly, the authors developed a new method to display and categorize their observed FRET transitions. POKIT plots (population-weighted and kinetically indexed transition density plots) visually represent the FRET transitions observed, the fraction of molecules in the sample undergoing the transition, and the lifetime of each initial FRET state prior to the transition (Figure 3A) . This approach greatly aids in interpretation of the voluminous, information-rich data sets produced in SM-FRET experiments.
In addition to the pre-mRNA, the spliceosomal snRNAs are also expected to undergo multiple conformational changes during splicing, and SM-FRET is giving insight into these as well. Guo et al. used a protein-free U2/U6 snRNA model system to characterize a set of conformational changes believed critical for spliceosomal active site formation . This study provided evidence for dynamic docking and undocking of a U2/U6 helix containing the highly conserved ACAGAGA sequence onto the U6 internal stem loop (ISL). This work also demonstrated interconversion between two alternate U2/U6 snRNA duplex structures at high and low Mg2+ concentrations, consistent with previously postulated mechanisms for dynamic rearrangements of the U2/U6 duplex during splicing [22,23]. By bridging results obtained from yeast genetics [24,25], in vivo assays in mammalian cells , and in vitro NMR studies , Guo et al. demonstrated the power of single molecule methods to directly test mechanistic models for U2/U6 snRNA duplex dynamics derived from a wide-range of ensemble sources. Based in part on the work of Guo et al., the U2/U6 snRNA duplex likely undergoes reversible conformational changes during splicing as opposed to being a static structure. The next challenge facing single molecule studies of snRNA dynamics is to confirm these results in the context of an intact and active spliceosome.
In addition to understanding conformation changes in the RNAs, it is necessary to define the pathway(s) by which spliceosome composition changes over the course of the splicing process (Figure 1). Protein inventories of isolated splicing complexes have revealed that dozens of proteins may specifically associate with or dissociate from the spliceosome during transitions between stable intermediates . While ensemble analyses had suggested that spliceosome assembly occurs by stepwise addition and removal of components [26–28], it was not previously known to what extent these reactions are ordered on individual pre-mRNA molecules or if other pathways (e.g. pre-association of the snRNPs ) contribute to spliceosome formation on a subset of molecules. A recent study combining chemical biology approaches for fluorescently labeling spliceosomal subcomplexes with multiwavelength total internal reflectance fluorescence (TIRF) microscopy shed new light on the order and dynamics of spliceosome assembly .
Hoskins et al.  employed homologous recombination to create haploid yeast strains containing C-terminal protein tags on a variety of essential spliceosomal proteins. The protein tags were comprised of a short glycine/serine linker followed by either the E. coli dihydrofolate reductase (DHFR) enzyme  or the SNAP tag, a variant of human alkylguanine S-transferase . After preparation of yeast WCE, DHFR tags were labeled by addition of Cy3- or Cy5-trimethoprim (TMP) analogs to the WCE. Although TMP binds prokaryotic DHFRs non-covalently, the interaction is extremely tight (KD < 1 nM) . The SNAP tag was labeled by incubation of WCE with fluorescent benzyl-guanine derivatives [e.g., Snap Surface 549™ (DY549), New England Biolabs] that covalently modify the active-site cysteine of the SNAP tag; excess dye was then removed by gel filtration. Using both the DHFR and SNAP tags in a single yeast strain enabled the authors to label two spliceosomal subcomplexes with spectrally distinguishable fluorophores in the same WCE (e.g., a U1-DHFR/Cy5-TMP-labeled, U2-SNAP/DY549-labeled WCE). Colocalization of the fluorescent subcomplexes on surface-tethered pre-mRNAs was monitored by Colocalization Single Molecule Spectroscopy (CoSMoS). These experiments were enabled by a novel TIRF optical system that efficiently collects fluorescence from multiple fluorophores simultaneously .
Using CoSMoS to simultaneously monitor association of pairs of spliceosomal subcomplexes (U1/U2, U2/U5, or U5/NTC) with pre-mRNAs in real-time (Figure 3B), Hoskins et al. found that spliceosome assembly occurred stepwise (U1->U2->U5->NTC) on >90% of the pre-mRNA molecules examined. Further, comparison of the rates of spliceosome subcomplex association revealed that no kinetic bottleneck was present for any single subcomplex association. Thus, spliceosome assembly is a kinetically efficient process in vitro.
Another key finding of the Hoskins et al. study was the reversibility of every major assembly step. Inspection of hundreds of single molecule traces revealed that every spliceosomal subcomplex bound dynamically. These results have important implications for the concept of splicing commitment. That is, at what stage of the assembly process does the splicing machinery commit to the utilization of a particular pair of splice sites? Prior to the Hoskins et al. study, it was generally believed that commitment occurs at the earliest stage of assembly – U1 addition [35,36]. However, by comparing intron loss events between early (U1) and late (NTC) binding subcomplexes, Hoskins et al. showed that commitment of individual pre-mRNAs is not an all-or-nothing event at the beginning. Because U1 binding is highly dynamic, a given interaction with U1 does not strongly commit a pre-mRNA to intron excision. Rather, the degree of commitment increases as the spliceosome assembles. In both early and late stages of assembly, reversible subcomplex binding allows spliceosomes to disassemble from the pre-mRNA without splicing. These results may have significant implications for our understanding of the regulation of alternative splicing. If spliceosomes formed around a particular set of splice sites can disassemble at any stage prior to intron removal, regulation of splice site choice could occur at any point in the pathway rather than just at the beginning. Thus regulation of alternative splicing is likely to be even more complicated than currently envisioned.
Through the combined efforts of several laboratories, single molecule splicers now have the tools in hand to monitor conformational changes in pre-mRNAs or snRNAs during splicing by SM-FRET and the association and dissociation of spliceosome components by CoSMoS. The combination of these approaches (FRET-CoSMoS) to correlate specific structural transitions during splicing with the presence or absence of a given spliceosomal component is likely to represent the next major advance in the single molecule studies of splicing. Indeed the fundamentals of the FRET-CoSMoS approach have already been applied to studies of translation  and nucleosome remodeling . The utilization of mutant yeast strains with known splicing phenotypes in single molecule assays may also prove informative for interpreting single molecule splicing data. One exciting prospect for future development is the combination of strains containing temperature-sensitive splicing phenotypes (e.g., stalling of splicing a low temperatures followed by release at the higher, permissive temperature) with recently developed methods for precise temperature control of single molecule experiments [39,40]. The ability to start or stop single splicing reactions by raising or lowering the temperature could prove extraordinarily useful and may be critical for understanding both how active spliceosomes assemble and how fidelity processes can reject incorrectly assembled complexes.
The results obtained so far with the yeast spliceosome on a limited set of substrates must be extended to other pre-mRNAs and other systems. Notably, a genome-wide study of in vivo splicing efficiencies revealed that different yeast pre-mRNAs respond in unique ways to mutations in core spliceosome proteins . These results suggest that pre-mRNA sequence and structure profoundly influence the pathway of spliceosome assembly and/or catalysis. Consequently, it is not yet clear which kinetic features uncovered by single molecule methods are specific to the pre-mRNAs studied and which represent generalities of spliceosome mechanism. One of the powers of the single molecule approach is that it can readily distinguish when two molecules have spliced by different pathways. This ability will be important for extending these results to metazoans where the same pre-mRNA is frequently alternatively spliced into different mRNAs . Thus single molecule methods are particularly well suited for shedding mechanistic insight into this central process for encoding genetic diversity in eukaryotes. Encouragingly, several single molecule studies of the human alternative splicing factor polypyrimidine tract binding protein (PTB) have already been carried out [43,44].
Finally, one aspiration of many biochemists is to study biological processes in their native environments. For splicing, this would require imaging spliceosome dynamics in eukaryotic cell nuclei. Recent results from several laboratories indicate that this goal is not far off. Work from the Singer laboratory has demonstrated that imaging transcription of single mRNAs in yeast nuclei is possible , and it is likely these methods can be extended to imaging intron excision as well. In contrast, imaging of single snRNPs will be complicated due to the high concentration of these components in the nucleus (HeLa cell nuclei contain ~106 copies of the U1 snRNP ). However, several groups have used fluorescence recovery after photobleaching (FRAP) and fluorescence correlation spectroscopy (FCS) to analyze spliceosomal dynamics in living cells [47,48]. These studies suggest that the dynamic behaviors and step-wise assembly mechanism observed in vitro are conserved in vivo. In the future, recent developments in photoactivated localization microscopy (PALM)  may prove useful for resolving individual snRNPs and studying their dynamics. The past few years have seen remarkably rapid progress in the studies of spliceosomes by fluorescence microscopy and many more unique insights into this mega-machine will undoubtedly emerge. Just keep watching!
This work was supported by NIH National Research Service Award fellowship GM079971 (A.A.H.), K99/R00 GM086471 (A.A.H.), and RO1s GM043369 (J.G.), GM81648 (J.G.), and GM053007 (M.J.M). M.J.M. is a Howard Hughes Medical Institute investigator.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.