|Home | About | Journals | Submit | Contact Us | Français|
SR proteins have been studied extensively as a family of RNA binding proteins that participate in both constitutive and regulated pre-mRNA splicing in mammalian cells. However, SR proteins were first discovered as factors that interact with transcriptionally active chromatin. Recent studies have now uncovered properties that connect these once apparently disparate functions, showing that a subset of SR proteins seem to bind directly to the histone 3 tail, play an active role in transcriptional elongation, and co-localize with genes that are engaged in specific intra- and inter-chromosome interactions for coordinated regulation of gene expression in the nucleus. These transcription-related activities are also coupled with a further expansion of putative functions of specific SR protein family members in RNA metabolism downstream of mRNA splicing, from RNA export to stability control to translation. These findings therefore highlight the broader roles of SR proteins in vertical integration of gene expression and provide mechanistic insights into their contributions to genome stability and proper cell cycle progression in higher eukaryotic cells.
About 20 years ago, a monoclonal antibody mAb104 raised against nuclear proteins in the germinal vesicle (GV) of amphibian oocytes was shown to intensively stain lampbrush chromosome loops that correspond to actively transcribed genes (Roth et al., 1990). At about the same time, another monoclonal antibody called B52 was found to decorate induced Hsp70 loci on Drosophila polytene chromosomes, bracketing Pol II in a symmetric fashion on the heat shock gene (Fig. 1a), and by UV-crosslinking, the B52 antigen was shown to bind directly to both RNA and DNA (Champlin et al., 1991). These findings suggest a potential role for the nuclear antigens detected by mAb104 and B52 in transcription. However, it was soon recognized that B52 is closely related to then the newly defined splicing factor SF2/ASF essential for constitutive splicing and capable of modulating splice site selection (Ge and Manley, 1990; Ge et al., 1991; Krainer et al., 1990; Krainer et al., 1991). These findings, together with electron microscopic evidence (Beyer and Osheim, 1991), established the concept of co-transcriptional mRNA processing, but the contribution of splicing factors and RNA processing to gene transcription was only realized a decade later (Fong and Zhou, 2001) and its biological importance has since become increasingly appreciated (Li and Manley, 2006; Maniatis and Reed, 2002; Moore and Proudfoot, 2009; Pandit et al., 2008).
We now know that a key link between transcription and RNA splicing is provided by a family of highly conserved RNA binding splicing initiators known as SR proteins based on extensive studies on its founding members SF2/ASF and SC35, the latter of which was initially detected by a monoclonal antibody against a spliceosome component (Fu and Maniatis, 1990; Fu and Maniatis, 1992). mAb104 defines the core group of SR proteins by recognizing a common phosphoepitope in the C-terminal domain of individual SR proteins that are enriched with serines and arginines, hence the same of SR proteins (Fig. 1b), and B52 corresponds to the Drosophila ortholog of SRp55, a specific member of the SR family (Zahler et al., 1992). Besides “classic”, mAb104 reactive SR proteins, many splicing regulators share some key structural features with SR proteins, particularly the domain enriched with serine/arginine or arginine/serine dipeptide repeats (called the RS domain) (Fu. 1995), and a genome survey revealed a large number of such SR-related proteins through eukaryotes, including yeast (Boucher et al., 2001). Importantly, SRp20 and several SR-related proteins, including the splicing co-activator SRm160, are involved in 3′ end formation (Blencowe et al., 1999; Lou et al., 1998), while others appear to play roles in a variety of cellular functions from transcription to cell cycle control, raising the possibility that SR proteins and SR-like factors may function to coordinate many different steps in RNA metabolism (Rosonina and Blencowe, 2002).
Given abundant evidence for co-transcriptional pre-mRNA splicing on most eukaryotic genes, it has become widely accepted that SR proteins promote spliceosome assembly on nascent RNAs to catalyze intron removal before the completion of transcription, thus explaining the association of SR proteins with active genes in vivo (see the latest in (Bjork et al., 2009). However, SR protein binding to nascent RNA does not seem to discriminate intronless from intron-containing genes, begging the question about why and how SR proteins are recruited to intronless genes where their presumed function in splicing is not needed. The recent burst of discoveries provides clues to this puzzle, which suggest a much broader role for SR proteins in gene expression. As the function of SR proteins and SR-related proteins in constitutive and regulated splicing has been extensively reviewed (Fu, 1995; Graveley, 2000; Lin and Fu, 2007; Long and Caceres, 2009; Manley and Tacke, 1996), here we mainly focus on the function of SR proteins beyond splicing, with particular emphasis on their activities in facilitating transcription, maintaining genome stability, organizing gene networks in interphase nuclei, and perhaps coordinating events for proper cell cycle progression in the M phase.
While collectively essential for splicing, different SR protein family members show a degree of functional redundancy as they can often substitute for one another in splicing assays with cell-free extracts. A degree of functional redundancy in vivo was also observed in C. elegans treated with RNAi against individual SR proteins (Longman et al., 2000). However, different SR family members are clearly functionally distinct as demonstrated in numerous studies in cell-free systems, in transfected cells, and in intact animals. Genetic inactivation of a typical SR protein causes cell lethality in model cell lines, which cannot be complemented by overexpressing a related SR protein (Lin et al., 2005; Wang et al., 1998). The unique functional requirement for B52/SRp55 was also demonstrated in the fly (Ring et al., 1994). We therefore refer to SR proteins when describing some functional properties that likely apply to multiple SR proteins or to specific members of the SR family when discussing some unique functions. Surprisingly, however, total RNA at the steady state appears little altered in SR protein-depleted cells (Ding et al., 2004; Lemaire et al., 2002). Although changes in alternative splicing have been detected in SR protein-depleted cells and certain specific alternative splicing events attributed directly to specific cellular functions (Ghigna et al., 2005; Li et al., 2005; Xu et al., 2005), it has been largely unclear whether composite splicing defects are behind cellular mortality.
In mammalian cells, aberrant splicing often results in the inclusion of a premature stop codon(s) in spliced transcripts, which are quickly eliminated by the nonsense-mediated RNA decay (NMD) pathway. This RNA surveillance pathway is triggered by an exon-exon junction complex (EJC), deposited after a stop codon, that cannot be removed by a scanning ribosome during “pioneer” translation (Lejeune and Maquat, 2005). Several RS domain factors, including RNPS1 and SRm160, have been identified to be core components of EJC (Le Hir et al., 2000). Interestingly, some classic SR proteins, particularly SF2/ASF, have been found to enhance NMD, likely by stimulating and/or stabilizing EJC deposition, and this process requires a functional RS domain, but not the ability of SR proteins to shuttle to the cytoplasm (Zhang and Krainer, 2004). This finding emphasizes the contribution of SR proteins to other RNA metabolism pathways within the nucleus.
Most SR proteins with the exception of SC35 seem to be able to shuttle rapidly between the nucleus and the cytoplasm (Caceres et al., 1998), it is logical to consider a post-splicing role of shuttling SR proteins in RNA export (Huang and Steitz, 2001; Huang and Steitz, 2005). This function of shuttling SR proteins has been linked to their ability to interact directly with a key RNA export factor Nxf1(TAP), which prefers SR proteins in a hypo-phosphorylated state (Huang et al., 2003; Huang et al., 2004; Lai and Tarn, 2004). Because SR protein phosphorylation is known to mediate spliceosome assembly, but their dephosphorylation is essential for progression of spliceosome to catalysis, this implies that SR proteins may play a role in integrating splicing with mRNA export in a concerted fashion. Thus, the rearrangement of the spliceosome during the splicing reaction may be functionally linked to nuclear export, which is also supported by the observation that mRNA derived from splicing is exported more efficiently than the identical RNA expressed from a cDNA (Valencia et al., 2008). However, it seems unlikely that the ability of a specific SR protein to shuttle and export mRNA is tied to its requirement for cell viability, because mutations in the shuttling SR protein SF2/ASF prevented its ability to shuttle, but did not impair cell viability as demonstrated in functional rescue experiments (Lin et al., 2005). It is presently unclear whether the non-essential function of SR proteins in mRNA export is due to a degree of functional redundancy among multiple shuttling SR proteins or merely reflects an auxiliary role of SR proteins in mRNA transport.
Shuttling SR proteins are also capable to stimulating protein synthesis by promoting mRNA entrance to polysomes (Sanford et al., 2005; Sanford et al., 2004). A recent study further revealed a plausible mechanism by which the SR protein SF/ASF stimulates mTOR (a kinase for multiple translation regulators) and/or inhibits the protein phosphatase PP2A (Michlewski et al., 2008). As a result, 4E-BP1 becomes hyperphosphorylated, which releases the cap binding protein eIF4E from the inhibitory eIF4E/4E-BP1 complex, thereby enhancing cap recognition and translational initiation. While the stimulation of the mTOR pathway is consistent with the oncogenic potential of SF2/ASF (Karni et al., 2007; Karni et al., 2008), such translational enhancement in the cytoplasm is unlikely to explain the requirement of this SR protein for cell survival as fusion of a nuclear retention signal to this shuttling SR protein abrogated its ability to translocate to the cytoplasm, but had little effect on cell viability (Lin et al., 2005).
Collectively, these findings suggest a broad role of SR proteins in multiple pathways of RNA metabolism. Because RNA transport, surveillance, and translation are all downstream of mRNA splicing, it is conceptually appealing to suggest that these downstream events may result from the function of SR proteins that remain on spliced mRNA during the splicing reaction, given the increasing evidence for SR proteins in coupling between transcription and splicing. However, it is important to emphasize that SR proteins were initially found to facilitate nuclear export of the intronless histone mRNA (Huang and Steitz, 20001). In addition, all experiments that demonstrate the involvement of SF2/ASF in translation have been carried out on intronless reporters (Sanford et al., 2005; Sanford et al., 2004). These observations therefore raise the question on whether some of these activities might also reflect certain splicing-independent functions of SR proteins.
A potential splicing-independent function of SR proteins would also be consistent with the observation made with SR protein-deficient cells, where in vivo depletion of the SR protein SC35 and SF2/ASF dramatically attenuated the production of nascent RNA (Lin et al., 2008). Further studies based on the nuclear run-on assay provide evidence for an active role of the SR protein SC35 in transcriptional elongation (Lin et al., 2008). A similar role of the budding yeast SR-like protein Npl3p in transcriptional elongation has also been reported (Dermody et al., 2008). Together, these recent advances suggest that an SR protein journey may start at the very beginning of gene expression. Interestingly, a recent study indicates that SC35 appears distinct from other SR proteins in interaction with DNA, implying that SC35 may be more extensively involved in transcription than other SR proteins in the nucleus (Fededa and Kornblihtt, 2008; Sapra et al., 2009).
SR proteins have been reported to directly or indirectly associate the phosphorylated C-terminal domain (CTD) of the largest subunit of Pol II (Misteli and Spector, 1999). More recently, proteomic analysis of purified Pol II complexes revealed a key role of SR proteins in linking the splicing machinery to the transcription apparatus, and consistent with the splicing commitment function of SR proteins reported earlier (Fu, 1993), SR proteins were found to stimulate U1 snRNP recruitment to nascent RNA for co-transcriptional splicing (Das et al., 2007). However, despite strong influence of promoter choice on alternative splicing (Cramer et al., 1999), it seems unlikely that SR proteins are pre-assembled into the transcription initiation complex because SR proteins prefer phosphorylated CTD, but CTD phosphorylation is known to take place during the elongation phase of transcription (Bentley, 2005; Phatnani and Greenleaf, 2006; Saunders et al., 2006). A direct evidence for this came from the recent observation that SR proteins are not stably associated with Pol II at the promoter of the FOS gene before transcriptional induction and the interaction of SR proteins with the gene after the induction gene appears to be mediated by nascent RNA (Sapra et al., 2009). Therefore, SR proteins are most likely to be dynamically recruited to the elongating Pol II complex to facilitate transcriptional elongation and concurrently link transcription to initiation of RNA splicing.
Interestingly, in vivo depletion of SC35 was also found to diminish the association of Pol II with pTEFb, a kinase responsible for CTD phosphorylation at Ser2 positions, which is critical for transcription elongation (Lin et al., 2008). This observation indicates that SR proteins recruited to Pol II may in turn help stabilize critical transcriptional elongation factors, such as pTEFb, providing further evidence that splicing may be reversely coupled with transcription as first suggested by Fong and Zhou (2001). Therefore, SR proteins may play an integral part in the process of transcriptional elongation, rather than just riding with the elongating Pol II complex to pick up emerging splice sites.
If SR proteins play an active role in transcriptional elongation ahead of their roles in RNA processing and/or during “reverse coupling”, they may help overcome certain elongation barriers (i.e. some natural pausing sites). This would be consistent with the original observation that B52 was deposited onto specific locations on heat shock gene puffs (see Fig. 1), and this SR protein could be crosslinked to DNA by UV, which requires direct physical contacts between the SR protein and the transcribed DNA (Champlin et al., 1991). These observations strongly suggest that SR proteins may become involved in gene expression before engaging in splice site selection, providing one tangible explanation for the functional interaction of SR proteins with both intron-containing and intronless genes.
The interaction of SR proteins with DNA seems likely to be mediated by nascent RNA, but how SR proteins gain such privilege to see nascent RNA in competition with a large number of other highly abundant RNA binding proteins expressed in mammalian cells needs to be explained. An intriguing clue came from a recent finding that the SR protein SRp20 and SF2/ASF seem to be capable of binding directly to the H3 tail (Loomis et al., 2009). Because H3 binding is mediated by the RS domain, there is no reason to believe that this activity is unique to the two SR proteins, even though other SR proteins remain to be tested experimentally. Since the RS domain is also required for SR proteins to associate (albeit indirectly) with the phosphorylated CTD, the open questions are whether SR proteins bind to Pol II and the histone 3 tail independently or sequentially and whether one binding event might influence the other during transcription.
It is well established that the H3 tail is extensively modified post-transcriptionally and specific modification events are generally associated with gene activation (e.g. methylation at lysine 4, acetylation at lysine 9, etc.) or repression (e.g. methylation at lysine 27) (Margueron et al., 2005). SR proteins do not seem to prefer specific modifications on the H3 tail with the exception of Ser10 phosphorylation, which is responsible for releasing SR proteins from the chromatin during the M phase of the cell cycle (see below). These observations imply that SR proteins may bind to the exposed H3 tail as a consequence of nucleosome rearrangement during transcriptional elongation. Intriguingly, a recent genome-wide location analysis in C. elegans revealed that H3 methylation at lysine 36 (H3K36), a modification that has been tightly linked to transcriptional elongation, appears to associate preferentially with exons (Kolasinska-Zwierz et al., 2009). This raises the intriguing possibility that H3K36 may potentiate the recognition of exonic sequences by SR proteins during co-transcriptional RNA processing.
Considering all evidence together, we envision a working model, which is largely speculative at this point, for the participation of SR proteins in transcription and co-transcriptional mRNA splicing as illustrated in Fig. 2. In our current understanding, a series of events take place during the transition from transcription initiation to elongation: Shortly after transcription initiation or re-initiation, Pol II becomes phosphorylated at Ser5 positions by TFIIH, which helps recruit the capping enzymes; Pol II escapes/clears the promoter and progresses to the promoter proximal pausing site about 20 to 50nt from the transcription start; the pTEFb kinase phosphorylates negative elongation factors associated with paused Pol II; and finally, the pTEFb kinase phosphorylates Pol II at Ser2 positions, allowing the Pol II complex to enter the productive elongation phase (Saunders et al., 2006). This step of Pol II phosphorylation at Ser2 positions may concur with dynamic recruitment of SR proteins to nascent RNA, which may bridge, at least in part, the interaction between SR proteins and Pol II (Sapra et al., 2009), because inhibition of Pol II Ser2 phosphorylation by DRB abolished the association of Pol II with SR proteins (Misteli and Spector, 1999).
When Pol II-associated SR proteins encounter an emerging binding site on elongating RNA, they may be unloaded from the polymerase onto the nascent transcript. This may allow SR proteins to be among the first to see nascent transcripts coming out from the elongating Pol II complex. Considering an apparent mutual dependence of SR proteins and the elongation factor pTEFb, this SR protein unloading event may also cause temporary destabilization of pTEFb and therefore transient pausing of the elongating Pol II complex. This may afford the Pol II complex more time on transcribed DNA, especially at genomic regions that are wrapped around nucleosomes, providing a potential mechanism for elevated H3K36 tri-methylation on exons relative to introns, which may in turn modulate Pol II processitivity during subsequent rounds of transcription at exon-containing regions in the genome (Kolasinska-Zwierz et al., 2009). Indeed, the level of H3K36 tri-methylation has been linked to splicing because alternative exons appear to be less marked by H3K36 methylation than constitutive exons (Kolasinska-Zwierz et al., 2009). Therefore, Pol II might travel across constitutive exons at a lower speed than across alternative exons. As a result, the splice sites associated with constitutive exons may be more efficiently recognized by the splicing machinery than those with alternative exons, which is consistent with the observed inverse correlation between the processitivity of Pol II and the efficiency of splice site selection (de la Mata et al., 2003; Howe et al., 2003).
Given the observation that certain SR proteins are capable of binding directly to the H3 tail (Loomis et al., 2009), we may further speculate that the alternation of nucleosome structure at exon-containing regions might in turn help recruit additional SR proteins to Pol II using H3 tail as a stepping stone, which may help ensure the continuation of transcriptional elongation (Fig. 2). The interaction of SR proteins with the exposed histone H3 tail may also promote and/or stabilize the opening of nucleosomes to facilitate the next round of transcriptional elongation. As SR proteins are responsible for initiating spliceosome assembly on pre-mRNA (Fu, 1993), SR proteins may be poised to commit pre-mRNA to splicing by first binding to the exposed H3 tail immediately flanking genomic regions that will give rise to authentic splice sites upon transcription. This working model therefore implies that nucleosome positioning may influence the definition of splice sites even before the RNA is made.
The intimate involvement of SR proteins in transcriptional elongation provides a plausible mechanism for the dramatic effect of SR protein deficiency on genome instability. In vivo depletion of the prototype SR protein SF2/ASF in chicken DT40 cells induced double-stranded DNA breaks (DSBs) and gross DNA recombination (Li and Manley, 2005). Similarly, removal of SC35 in mouse embryo fibroblasts (MEFs) was also found to trigger overwhelming DSBs, which induced the S-phase checkpoint as a result of ATM activation, leading to transcriptional induction of inhibitors to cyclin-dependent kinases (Xiao et al., 2007). A key contribution to this pathway is p53 phosphorylation by ATM, because siRNA-mediated knockdown of p53 could partially rescue the ability of SC35-depleted cells to enter S-phase (Xiao et al., 2007). Remarkably, both SF2/ASF and SC35 appear to be dispensable for viability of non-dividing cardiomyocytes in the heart (Ding et al., 2004; Xu et al., 2005). This phenotype may be explained by the absence of S-phase checkpoint and/or the lack of DNA replication to convert initial ssDNA breaks to DSBs, therefore rendering induced DNA damage undetectable in non-dividing cells. This makes the heart an ideal biological model system to study the cellular function of SR proteins in the absence of complications associated with cell cycle arrest and death.
How does SR protein depletion induce DNA breaks to cause genome instability? This has to do with a DNA/RNA configuration commonly referred to as the R-loop, a transient transcription bubble that forms within the elongating Pol II complex, in which nascent RNA remains attached to the template DNA and displaces the non-template strand. In the normal process of transcription, such a bubble is quickly resolved as the Pol II complex travels along the DNA. However, as illustrated in Fig. 3, blockage of transcriptional elongation may result in a prolonged R-loop behind the stalled Pol II complex, and the exposed ssDNA become attackable by nucleases or by other DNA modification enzymes, such as the cystidine deaminase AID, whose catalysis of C-to-U conversions may lead to ssDNA breaks by the DNA glycosylase UNG and subsequent gross DNA arrangement (Longerich et al., 2006). The ssDNA breaks may be further converted to DSBs, which will nucleate the formation of the MRN (Mre11/Rad50/Nbs1) complex (MRX in budding yeast) to recruit and activate ATM to initiate DNA repair (Harrison and Haber, 2006). Due to the exonuclease activity of the MRN complex during DNA replication in the S-phase, DSBs may be further processed to create an extensive 3′ overhang, which attracts the single strand DNA binding RPA, which then recruits and activates ATR (Harrison and Haber, 2006). These events quickly induce cell cycle arrest for DNA repair or commit cells to apoptosis if the damage is excessive.
It is well documented that mRNP formation (mRNA in complex with RNA binding proteins) plays a vital role in preventing R-loop formation across all organisms from bacteria to mammals (Li and Manley, 2006; Luna et al., 2005). SR proteins seem to participate actively in preventing R-loops as demonstrated in SF2/ASF-depleted chicken DT40 cells as well as in reconstituted nuclear extracts (Li and Manley, 2005), even though similar studies have not yet been extended on other SR protein-depleted cells. Interestingly, the R-loop phenotype can be detected on intronless genes, re-enforcing a splicing-independent role of SR proteins in mRNP formation as discussed above. The question here is how SR proteins might help prevent R-loop formation during transcriptional elongation. A commonly assumed mechanism is that mRNPs may hinder re-annealing of nascent RNA back to the template DNA behind the elongating Pol II complex (Aguilera and Gomez-Gonzalez, 2008). However, since strand invasion normally requires a free end, it is difficult to image how an RNA segment within a continuous strand could be re-inserted into duplex DNA once the RNA segment has been released from the template DNA. Here we may consider the possibility that SR proteins associated with the elongating Pol II complex may help the displacement of nascent RNA from the template DNA (Fig. 3). This function of SR proteins may be particularly important when the elongating Pol II complex encounters natural pausing sites.
There is another interesting twist with respect to the interplay between SR proteins, R-loop formation, and transcriptional elongation. It is well known that transcriptional elongation would induce positive and negative DNA supercoiling in the front and back of a R-loop, respectively, which would in turn retard the elongation of the Pol II complex, and a key enzyme involved in removing DNA supercoiling is Topoisomerase I (Champoux, 2001). Interestingly, Topo I has been shown to interact with and phosphorylate SF2/ASF (Rossi et al., 1996). The kinase activity of Topo I towards the SR protein is inhibited when Topo I is bound to DNA, and conversely, the DNA unwinding activity of Topo I is suppressed when Topo I is associated with hypophosphorylated SF2/ASF (Andersen et al., 2002). Based on these observations and on the assumption that Topo I has a broad activity on SR proteins in this transcriptional elongation related aspect, we may envision the following scenario: If an SR protein were to become hypophosphorylated in the elongating Pol II complex, it would switch affinity from Pol II to Topo I, which could induce transcriptional pausing by impairing the DNA unwinding activity of Topo I. Re-phosphorylation of the SR protein by Topo I would facilitate a switch of the SR protein back to Pol II or even to nascent RNA, allowing transcriptional elongation to continue. In this way, Topo I may help reduce the interference of hypophosphorylated SR proteins with transcription elongation and reinforce proper SR deposition from Pol II to nascent RNA during transcriptional elongation.
It is important to note that R-loop independent mechanisms may also contribute to cell death in SR protein-depleted cells. Cell death continued to take place in SR protein-depleted chicken DT40 cells even when DSBs were experimentally inhibited by degrading RNA in R-loops with overexpressed RNase H (Li and Manley, 2005). DT40 cells can proliferate rapidly (doubling in 8h) and lack p53 (and are thus checkpoint deficient) (Takao et al., 1999). As a result, apoptosis may be quickly induced and, interestingly, some specific components of the apoptosis pathway (including DNA fragmentation) have been linked to SF2/ASF-regulated splicing (Li et al., 2005). However, apoptosis was not detected in SR protein-depleted MEFs (Xiao et al., 2007), suggesting that these normal p53+/+ cells may die of multiple defects in transcriptional and post-transcriptional gene expression pathways. In this regard, it is astonishing that such a fundamental requirement for SR proteins in proliferating cells becomes non-essential for the specialized gene expression program in non-dividing cardiomyocytes, although it remains to be determined whether SR proteins are generally dispensable for non-dividing cells, such as differentiated neurons.
All SR proteins are unevenly distributed in the nucleus, concentrating on the nuclear domains commonly known as nuclear speckles or the SC35 domain at steady state in interphase nuclei (Fig. 4). In fact, most factors involved in pre-mRNA splicing exhibit a similar distribution pattern in the nucleus. Under EM, these speckled nuclear domains correspond to interchromatin granule clusters (Spector, 1993). Although nuclear speckles have become diagnostic of proteins involved in mRNA splicing, the functional meaning of such nuclear domains continues to be the subject of debate. A popular view emphasizes these speckled nuclear domains as a consequence of transcription and co-transcriptional RNA processing, rather than obligatory nuclear “organelles” for gene expression activities (Singer and Green, 1997). Clearly, transcription can take place throughout the nucleus without apparent “compartmentalization” and factors accumulated in nuclear speckles can be rapidly recruited to nearby active genes (Misteli et al., 1997), which gives rise to the view of nuclear speckles as “storage” sites for splicing factors. However, the demonstration that some, though not all, genes become associated with nuclear speckles upon activation argues for a more active role of speckles in gene expression (reviewed by Hall et al., 2006).
The center of the debate is on the origin of nuclear speckles and the location of the domains relative to actively transcribed genes in the nucleus. One model is that, after transcription is terminated, a fraction of released mRNPs may travel some distance in the nucleus in search for some common locations for post-transcriptional processing and subsequent factor recycling, thus giving rise to the uneven distribution of splicing factors in the nucleus. In an alternative model, which does not have to be mutually exclusive from the first one, active genes may be unevenly localized in the nucleus in the first place and both nascent RNA-associated mRNPs and released mRNPs may accumulate near or around some gene clusters. The second model would be consistent with an earlier observation that interchromatin fibrils (which correspond to nascent RNPs) are frequently detected around interchromatin granules (Fakan, 1994), and with a more recent observation that active genes and gene-rich R-bands tend to associate the SC35 domain more frequently than silent genes and gene-poor G-bands (Shopland et al., 2003). Therefore, while transcription can take place throughout the nucleus, certain clustered activities of gene expression may directly contribute to the initial formation of nuclear speckles (Hall et al., 2006; Lamond and Spector, 2003).
According to this model, one may view nuclear speckles as a snapshot of highly dynamic nuclear domains enriched with mRNA splicing factors with certain, but not all, genes exhibiting a spatial proximity to them. Because transcription is tightly coupled with downstream RNA processing steps (Maniatis and Reed, 2002; Moore and Proudfoot, 2009; Pandit et al., 2008), inhibition of transcription may halt a chain of gene expression events and arrest complexes at various RNA metabolism stages, which may be further attracted to the existing nuclear speckles to form aggregates (O'Keefe et al., 1994). This model explains why nuclear speckles, including those large aggregates induced by inhibition of transcription, harbor not only splicing factors, but also factors involved in transcription such as phosphorylated Pol II, RNA export factors, and even ploy(A)-containing RNA (Saitoh et al., 2004). The model also provides a plausible account for the complex cellular distribution of phosphorylated Pol II, which seems to depend on the general transcriptional activity in the cell: The active Pol II is distributed in a few hundred transcription foci in cells that incorporate a high level of Br-uridine into nascent RNA, but is more restricted to 20 to 50 nuclear speckles in poorly transcribing cells (Zeng et al., 1997).
If the initial formation of nuclear speckles reflects some clustered gene expression activities, for instance, at the beginning of the G1 phase, what is the underlying mechanism for active genes to become unevenly distributed in the three dimensional space of the nucleus? In other words, do some genes become grouped in a random fashion or with some specificity? It is well known that decondensed chromosomes occupy specific nuclear territories in interphase nuclei, and in response to an activation signal, induced gene loci tend to relocate to the periphery of its nuclear territory (Brown et al., 2006; Chambeyron and Bickmore, 2004). The latter observation also begs the question as to whether an induced gene randomly searches for a better environment for expression or is attracted to some specific locations for synergistic activation. Recent studies indicate that many genes are specifically engaged in long distance intra- and inter-chromosomal interactions (Kumaran et al., 2008; Sexton et al., 2007). In particular, a pair of estrogen-regulated genes becomes co-localized with nuclear speckles, and the association is tightly linked to maximal induction of gene expression (Hu et al., 2008). A similar observation was also made on several co-regulated globin genes (Brown et al., 2008). These findings suggest that specific gene networks may be organized around nuclear speckles to facilitate promoter-enhancer interactions both within the same chromosomes and between specific chromosomes for coordinated regulation of transcription and RNA processing in the nucleus.
While the specificity in establishing gene networks is unclear at this point, the question is what role SR proteins might play in organizing such hubs for gene networking in the nucleus. Earlier work has established that the nuclear distribution of SR proteins is regulated by phosphorylation (Gui et al., 1994), which appears to be the major mechanism to regulate the cellular distribution of other splicing factors in the nucleus (Misteli and Spector, 1997). In particular, SC35 seems to be the only SR protein among those tested thus far that does not shuttle out of the nucleus (Sapra et al., 2009) and it is highly resistant to detergent extraction, suggesting that SC35 may serve as a seed to nucleate the formation of nuclear speckles. Given the most recent observation that SR proteins may associate active genes through their direct interaction with the H3 tail (Loomis et al., 2009), we may now consider a tantalizing possibility that SR proteins may play a role in organizing transcription hubs to coordinate RNA transcription, splicing, and export in complex eukaryotic cells.
If SR proteins play a key part in organizing the genome in interphase cells, reorganization of gene networks may be critical for cell cycle progression. Consistent with this possibility, a recent study demonstrated that the interaction of SR proteins with chromatin is diminished in mitotic cells in which the H3 tail is modified at the Ser10 position by the activated Aurora kinase B (Loomis et al., 2009). Moreover, the SR protein SF2/ASF was found to be required for dismantling the heterochromatin mark HP1 from chromatin. These observations imply that the regulated interaction of SR proteins with chromatin may help facilitate chromosome condensation during cell cycle progression through M phase.
Interestingly, hyperphosphorylation of SR proteins by SR protein specific kinases (SRPKs) was also found to diminish the interaction of SR proteins with the H3 tail (Loomis et al., 2009) In fact, it had been observed earlier that SR proteins become hyperphosphorylated when cells enter the M phase (Gui et al., 1994; Roth et al., 1990). Consistent with SR protein hyperphosphorylation in the M phase, SRPKs are found to be largely retained in the cytoplasm of interphase cells through their interactions with molecular chaperones (Zhong et al., 2009) and 14-3-3 proteins (Jang et al., 2009), but relocate to the nucleus at onset of the M phase before the breakdown of the nuclear envelope (Ding et al., 2006). Considered together, these findings suggest that the detachment of SR proteins from chromatin may be dually regulated by Aurora kinase B and SRPKs during mitosis, both of which may contribute to cell cycle progression through the M phase.
Based on recent advances, it has become quite probable that RNA-binding SR proteins are critical to multiple steps in gene expression, from transcriptional elongation to mRNA splicing to RNA export to translation. The integration of these activities by single SR proteins may constitute the requirement of SR proteins for cell viability and proliferation. Recent findings also suggest some unexpected roles of SR proteins in organizing gene networks in the nucleus, maintaining genome stability, and facilitating cell cycle progression. Although it remains to be determined whether these functions reflect a degree of independent activities of SR proteins or are all intertwined with the function of SR proteins in pre-mRNA splicing, these recent discoveries emphasize key roles that SR proteins are likely to play in coupling various transcriptional and post-transcriptional gene expression events. Defects in these processes may directly contribute to developmental defects and onset of cancer.
The authors thank Drs. John Lis, Juan Valcarcel, Robert Singer, Bruce Hamilton and Kristi Fox-Walsh for their insightful comments on the manuscript. Work in authors’ laboratory was supported by NIH grants (GM49369 and GM52872) to X-D.F.
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.