REVIEW ARTICLE Phosphorylation mechanism and structure of serine-arginine protein kinases Gourisankar Ghosh 1 and Joseph A. Adams 2 1 Department of Chemistry and Biochemistry, University of California, San Diego, La Jolla, CA, USA 2 Department of Pharmacology, University of California, San Diego, La Jolla, CA, USA Introduction The complexity of the human proteome is regulated through the alternative splicing of large precursor mRNAs [1]. Although this process plays a significant role in normal cellular development, changes or defects in alternative splicing have also been linked to human disease [1–3]. Splicing reactions occur at the Keywords mechanism; protein kinase; splicing; SR protein; structure Correspondence J. A. Adams, Department of Pharmacology, University of California, San Diego, La Jolla, CA 92093-0636, USA Fax: +1 858 822 3361 Tel: +1 858 822 3360 E-mail: j2adams@ucsd.edu (Received 7 July 2010, revised 9 November 2010, accepted 10 December 2010) doi:10.1111/j.1742-4658.2010.07992.x The splicing of mRNA requires a group of essential factors known as SR proteins, which participate in the maturation of the spliceosome. These proteins contain one or two RNA recognition motifs and a C-terminal domain rich in Arg-Ser repeats (RS domain). SR proteins are phosphory- lated at numerous serines in the RS domain by the SR-specific protein kinase (SRPK) family of protein kinases. RS domain phosphorylation is necessary for entry of SR proteins into the nucleus, and may also play important roles in alternative splicing, mRNA export, and other processing events. Although SR proteins are polyphosphorylated in vivo, the mecha- nism underlying this complex reaction has only been recently elucidated. Human alternative splicing factor [serine ⁄ arginine-rich splicing factor 1 (SRSF1)], a prototype for the SR protein family, is regiospecifically phos- phorylated by SRPK1, a post-translational modification that controls cyto- plasmic–nuclear localization. SRPK1 binds SRSF1 with unusually high affinity, and rapidly modifies about 10–12 serines in the N-terminal region of the RS domain (RS1), using a mechanism that incorporates sequential, C-terminal to N-terminal phosphorylation and several processive steps. SRPK1 employs a highly dynamic feeding mechanism for RS domain phosphorylation in which the N-terminal portion of RS1 is initially bound to a docking groove in the large lobe of the kinase domain. Upon subse- quent rounds of phosphorylation, this N-terminal segment translocates into the active site, and a b-strand in RNA recognition motif 2 unfolds and occupies the docking groove. These studies indicate that efficient regiospec- ific phosphorylation of SRSF1 is the result of a contoured binding cavity in SRPK1, a lengthy Arg-Ser repetitive segment in the RS domain, and a highly directional processing mechanism. Abbreviations CLK, cdc2-like kinase; kdSRPK1, kinase-dead form of human alternative splicing factor; LysC, lysyl endoproteinase; MAPK, mitogen-activated protein kinase; RRM, RNA recognition motif; RS domain, domain rich in Arg-Ser repeats; RS1, N-terminal region of human alternative splicing factor domain rich in Arg-Ser repeats; RS2, C-terminal region of human alternative splicing factor domain rich in Arg-Ser repeats; SRPK, SR-specific protein kinase; SRSF, serine ⁄ arginine-rich splicing factor; SRSF1, human alternative splicing factor (serine ⁄ arginine-rich splicing factor 1). FEBS Journal 278 (2011) 587–597 Journal compilation ª 2011 FEBS. No claim to original US government works 587 spliceosome, a macromolecular complex composed of five small nuclear ribonucleoproteins (U1, U2, U4, U5, and U6) and over 100 auxiliary proteins [4]. Among these many proteins, one family of splicing factors, known as SR proteins, is essential for controlling numerous aspects of mRNA splicing as well as other RNA processing events. SR proteins interact with splicing components (U1-70K and U2AF 35 ) early during spliceosome development, and help to establish the 5¢ and 3¢ splice sites [5,6]. Later, they recruit the U4 ⁄ U6ÆU5 tri-small nuclear ribonucleoprotein [7] and also enhance the second catalytic step in splicing [8]. Splicing is tightly coupled to transcription, and SR proteins have been shown to play a role by binding the C-terminal domain of RNA polymerase II and reg- ulating CDK9 [9]. SR proteins serve roles in many postsplicing events, including mRNA export [10,11], translation regulation [12,13], and genomic stability [14,15]. More than a decade ago, it was discovered that two protein kinase families [SR-specific protein kinases (SRPKs) and cdc2-like kinases (CLKs)] phosphorylate SR proteins, altering their cellular distribution and activities [16–18]. In the last several years, great strides have been made in understanding how the SRPKs rec- ognize and phosphorylate the SR proteins. In this review, we will highlight how a highly dynamic and distinct interplay between kinase and substrate is nec- essary for modification of the SR protein human alter- native splicing factor [serine ⁄ arginine-rich splicing factor 1 (SRSF1)], a prototypical SR protein involved both in constitutive and in alternative splicing [19]. These studies have shown that SRPK1 has an interest- ing feeding mechanism, whereby multiple contacts in the SR protein are utilized to catalyze a lengthy polyphosphorylation reaction. Structural features of SR proteins SR proteins derive their name from a lengthy (50–300- residue) C-terminal tail rich in Arg-Ser repeats known as the RS domain (Fig. 1). SRSF1 (also known as ASF ⁄ SF2) represents a typical arrangement for an RS domain, where Arg-Ser repeats are bracketed by smal- ler repeats and some isolated Arg-Ser pairs. In addi- tion to RS domains, SR proteins also contain one or two N-terminal RNA recognition motifs (RRMs) that modulate SR protein interactions in the spliceosome by binding short mRNA sequences (splicing enhancers) [20,21]. Numerous screening procedures have revealed that the observed determinants are somewhat nonspe- cific, raising the possibility that members of the SR protein family serve redundant functions in mRNA splicing [22]. Although, in support of this idea, all SR proteins can complement splicing-deficient S100 cyto- plasmic extracts of HeLa cells [20], there are other studies showing that certain SR proteins play tissue- specific roles at various developmental stages [23,24], arguing for a specialized role for some SR proteins. Although there is no X-ray structure for an SR protein in either a phosphorylated or unphosphorylated state, a recent NMR structure shows that one of the RRMs of SRSF1 (RRM2) adopts a typical RNA-binding fold [25]. Sequence analyses suggest that SR proteins may have properties consistent with intrinsically disordered proteins, owing to an RS domain that is expected to be largely unstructured [26]. On the other hand, all RS domain RRM domain PRSPSYGRSRSRSRSRSRSRSRSNSRSRSYSPRRSRGSPRYSPRHSRSRSRT-C′ SRSF4 316 SRSF6 160 SRSF7 118 SRSF5 88 SRSF2 105 SRSF3 79 SRSF1 50 RS1 RS2 SRPK phosphorylation nuclear entry/speckle formation CLK phosphorylation nuclear dispersion Fig. 1. SR protein domain structure. All traditional SR proteins have one or two N-terminal RRMs and one C-terminal RS domain. The amino acid sequence for the RS domain of the prototype SR protein SRSF1 is displayed. Peptide mapping and cellular analyses indicate that the phosphor- ylation of two segments in this RS domain (RS1 and RS2) by SRPKs and CLKs control the subcellular distribution. SRPK structure and mechanism G. Ghosh and J. A. Adams 588 FEBS Journal 278 (2011) 587–597 Journal compilation ª 2011 FEBS. No claim to original US government works atom calculations of an eight dipeptide repeat [(Arg- Ser) 8 ] suggest that the unphosphorylated sequence adopts a helical form, with the arginines pointing out into solution for charge minimization, and a compact, ‘claw-like’ structure upon phosphorylation [27]. Appre- ciable helical content has not been detected in CD experiments for SRSF1 or its RS domain in either the phosphorylated or unphosphorylated forms [28], sug- gesting that if the Arg-Ser repeats possess helical struc- ture, it may not be highly stable in solution. Recent studies have shown that the phosphorylated RS domain is protected from dephosphorylation by the neighboring RRMs, suggesting that the RS domain may not be disordered and could pack onto other domains in the SR protein [29]. Nonetheless, although the RRMs adopt a classic RNA-binding fold, it is still not fully clear how the RS domain folds by itself or in the context of the SR protein, or how phosphorylation modifies the SR protein conformation. SR proteins are phosphorylated by two protein kinase families Early studies showed that SR proteins undergo multi- ple rounds of phosphorylation and dephosphorylation en route to spliceosome assembly [30–32]. Phosphoryla- tion was shown to occur in the RS domain and alter how the SR protein functions in the spliceosome. For example, the SR proteins SRSF1 and SRSF2 (also known as SC35) interact with the 70-kDa subunit of U1 (U1-70K) and the 35-kDa subunit of U2AF (U2AF 35 ) in a phosphorylation-dependent manner [5,6], establishing the appropriate splice sites. More than a decade ago, it was discovered that the SRPK and CLK families of protein kinases can polyphospho- rylate RS domains and alter SR protein cellular distri- bution and splicing function [16,17,33]. However, the role of RS domain phosphorylation in alternative splic- ing is not well understood. Although some studies have suggested that the RRMs are the principal driving ele- ments for alternative splicing of some precursor mRNAs [34], other studies have shown that the phos- phoryl content of the RS domain is important. For example, phosphorylation of SRSF1 controls the alter- native splicing of the caspase-9 and Bcl-x genes and induction of a proapoptotic phenotype [35]. Although further investigations are needed to provide a more forceful link between RS domain modification and splicing, it has become abundantly clear that phosphor- ylation is directly linked to the nuclear entry of SR pro- teins. It has been shown that phosphorylation of the RS domain leads to enhanced interactions with the nuclear import receptor, transportin SR, and entry of the SR proteins into the nucleus, where they largely reside in speckles [36–38]. Whereas SRPKs play a direct role in nuclear import, the CLK family controls the nuclear distribution of SRSF1 and other SR proteins. Thus, through interactions with two families of protein kinases, the cellular location and, presumably, splicing function of SR proteins can be precisely controlled. Kinetic studies on SRPK1 and SRSF1, together with crystal structures of SRPK1 bound to peptide and pro- tein substrates and the recent structures of CLK1 and CLK3, suggest an elegant mechanism of recognition and phosphorylation by these two kinases, which regu- late the biological function of SR proteins in the cell. SRPK1 structure Most of our knowledge regarding SRPKs comes from studies on SRPK1 and the yeast analog, Sky1p. SRPKs contain a well-conserved kinase domain that is bifurcated by a large, nonconserved insert domain (approximately 250 amino acids). The insert domain in SRPKs regulates subcellular localization, as its deletion changes the distribution pattern of the kinase from nuclear–cytoplasmic to exclusively nuclear [39]. In addition to this important regulatory domain, SRPKs contain N-terminal and ⁄ or C-terminal extensions, which are not conserved. Deletion of the insert and N- terminal extension does not inactivate the catalytic activity of SRPK1, suggesting that these elements play auxiliary roles [40]. The X-ray structure for SRPK1 lacking its N-terminus and most of the insert domain reveals the signature bilobal fold found in all eukary- otic protein kinases (Fig. 2A). The small lobe is com- posed mostly of b-strands, and binds the nucleotide (ADP in the SRPK1 structure). The larger lobe is com- posed mostly of a-helices, and provides residues important for substrate binding. A short segment of the insert domain connecting the two major kinase lobes is present, and adopts short helical conforma- tions. Like other members of the CMGC group of protein kinases, SRPK1 contains a small insert within the kinase domain known as the mitogen-activated protein kinase (MAPK) insert, which connects heli- ces aG and aH (Fig. 2A). Although the X-ray struc- ture of SRPK1 was solved with a short substrate peptide, this peptide binds unexpectedly outside the active site in a groove generated by the MAPK insert and a loop connecting helices aF and aG (Fig. 2A). Later, we will discuss how this docking groove binds SRSF1 and feeds the RS domain into the active site for sequential phosphorylation. X-ray structures of the kinase domains of CLK1 and CLK3 have been reported recently [41], and are G. Ghosh and J. A. Adams SRPK structure and mechanism FEBS Journal 278 (2011) 587–597 Journal compilation ª 2011 FEBS. No claim to original US government works 589 worth noting here, given their overlapping substrate specificities with the SRPKs. Although the CLK family of protein kinases is capable of widespread RS domain phosphorylation, their structures are distinct from those of the SRPKs in several ways. Most significantly, the CLK enzymes lack a large insert domain dividing the kinase core and, unlike the SRPKs, are auto- phosphorylated on both serine and tyrosine [42]. The CLK kinases have large N-termini, as do the SRPKs, but, unlike in the SRPKs, these extensions are rich in isolated Arg-Ser dipeptides. Although the CLK family also belongs to the CMGC group of kinases, changes in the sequence of the MAPK insert and positions of helices aG and aH result in the loss of the deep sub- strate docking groove observed in SRPK1. In addition to the MAPK insert, CLK1 and CLK3 contain another small insert between stands b6 and b9 in the kinase core that interacts with a hydrophobic pocket near the hinge region connecting the kinase lobes. Maintenance of the constitutively active conformation Whereas many protein kinases are highly regulated through diverse mechanisms, SRPK family members are constitutively active and require no post-transla- tional modifications or additional protein subunits for optimal activity. Several key structural elements are essential in the maintenance of this highly active form of SRPK1. In some protein kinases, the activation loop plays a regulatory role by controlling access to the active site, and only adopts an open configuration upon phosphorylation by other protein kinases [43,44]. The activation loop of SRPK1 is comparatively short and, lacking a reversible phosphorylation site, adopts a stable conformation that allows ready access of sub- strates to the active site (Fig. 2B). Extensive biochemi- cal analyses have shown that the activation loop in SRPK1 is highly malleable [45]. Molecular dynamics simulations have shown that alternative residues can mediate contacts that are lost upon mutation of some residues in the activation segment and maintain the structural integrity of the activation segment. Thus, SRPK1 is resilient to inactivation, and exhibits robust phosphorylation activity. The extensive phosphoryla- tion that SRPK1 must execute for each SR protein is a likely explanation for the evolution of such robust activity. In addition to activation loops, all protein kinases possess a catalytic loop with a conserved aspartic acid that forms a hydrogen bond with the hydroxyl serine⁄ tyrosine of the substrate. In the case of SRPK1, the catalytic loop aspartate is ideally poised to abstract the hydroxyl hydrogen from the substrate serine, a necessary step for protein phosphorylation [46]. Several short-range interactions within the small lobe and between the large and small lobes around the active site participate in maintaining the catalytically active conformation. Two conserved interactions in all active protein kinases are also present in SRPK1: an ion pair between an invariant glutamic acid in helix aC and an invariant lysine in strand b3 in the small lobe, and a hydrogen bond between the activation loop and helix aC (Fig. 2B). Regiospecific phosphorylation of the RS domain of SRSF1 Although both the SRPK and CLK families catalyze multisite phosphorylation of SR proteins, the struc- tural data suggest that differences in critical regions Fig. 2. Structural features of SRPK1. (A) Ribbon diagram of SRPK1 in complex with ADP and a short peptide substrate (RRRERSPTR). The peptide binds near the MAPK insert. SRPK1 lacks most of its N-ter- minus (1–41) and insert domain (256–473). (B) Several conserved structural elements and contacts in SRPK1. SRPK structure and mechanism G. Ghosh and J. A. Adams 590 FEBS Journal 278 (2011) 587–597 Journal compilation ª 2011 FEBS. No claim to original US government works such as the MAPK insert and N-terminus may impart distinct regiospecificities. To address this issue, the mechanism of phosphorylation of SRSF1 by both enzymes was investigated with MS methods. As shown in Fig. 1, the RS domain of SRSF1 contains many serines throughout, and it is not clear whether these two kinases show preferences for specific residues. The mapping of phosphorylation sites in the RS domain is a vexing problem, owing to the redundancy of the Arg-Ser repeats and difficulties in separating ⁄ identify- ing the closely related polybasic fragments in tradi- tional mapping studies. This problem has been circumvented by using a modified form of SRSF1 that contains four Arg fi Lys substitutions in the RS domain. Upon phosphorylation and cleavage with lysyl endoproteinase (LysC), five fragments encompassing the complete RS domain of SRSF1 could be identified by MALDI-TOF MS [47]. These studies detected about eight phosphoserines in the N-terminal portion of the RS domain. To further define the phosphoryla- tion segment in SRSF1, a wide series of truncation derivatives were made, and their phosphoryl contents were assessed by MS [48]. These studies showed defini- tively that SRPK1 is a regiospecific protein kinase, preferring to phosphorylate up to 12 serines in the N-terminal region of the RS domain of SRSF1 (RS1) (Fig. 1). Single turnover kinetic studies have shown that RS1 is phosphorylated very efficiently within 1–2 min. In comparison, CLK1 does not show this regiospecificity, and instead can phosphorylate all 20 serines in the RS domain of SRSF1 [47]. Furthermore, CLK1 appears to be able to completely phosphorylate the RS domain of SRSF1 even if RS1 is prephosph- orylated by SRPK1. This sequential phosphorylation of RS1 (by SRPK1) and the C-terminal region of the RS domain of SRSF1 (RS2) (by CLK1) segments is biologically relevant, as it has been demonstrated that SRSF1 lacking RS2 translocates to the nucleus but is neither additionally phosphorylated nor dispersed in the nucleus by CLK1 [40]. These studies provide a model in which SRPK1 phosphorylates RS1, leading to translocation of SRSF1 from the cytoplasm to nuclear speckles, whereas CLK1 phosphorylates RS2, leading to broad nuclear dispersion of the SR protein. Mechanism of RS domain phosphorylation Although it is not uncommon for protein kinases to exhibit somewhat relaxed substrate specificities and phosphorylate more than one site in their protein target [49–52], SRPK1 possesses the distinct ability to effi- ciently insert numerous phosphates in close proximity in RS domains. In general, protein kinases recognize local charges flanking the site of phosphorylation [53]. Random library searches have shown that SRPK pre- fers to phosphorylate serine, but not threonine, that is next to arginines [54]. These studies were performed with a biased peptide library (arginine fixed in the P-3 position), that contained a single serine for modifi- cation. In contrast, SRPKs phosphorylate many con- secutive serines in a richly electropositive substrate. To accomplish this task, SRPK would need to maneuver deftly through a substrate whose charge is dramatically changing after each round of phosphorylation. The question that this raises is whether these splicing kinases must re-engage the RS domain after each phosphoryla- tion reaction through a sequence of dissociation–associ- ation steps (distributive phosphorylation), or whether the RS domain can stay attached during subsequent rounds of phosphorylation, simply translating through the active site (processive phosphorylation) (Fig. 3A). There are examples of protein kinases that catalyze multisite phosphorylation using either mechanism and sometimes a combination of both. For example, the nonreceptor protein tyrosine kinase Src phosphorylates up to 15 tyrosines in the protein Cas by a processive mechanism [49,50]. In contrast, the dual specific protein kinase MEK activates MAPK through a two-site phos- phorylation mechanism that is fully distributive [52,55]. Finally, the yeast cyclin–CDK complex from budding yeast (Pho80–Pho85) appears to phosphorylate five serines in the transcription factor Pho4, by a semipro- cessive mechanism [56]. The question of how a splicing kinase modifies an SR protein was originally addressed for SRPK1 and its substrate SRSF1, with a start-trap protocol [57]. In this experiment, a peptide inhibitor or a kinase-dead form of SRPK1 (kdSRPK1) is added at the start of the reac- tion to a preformed enzyme–substrate complex in single turnover experiments (i.e. [SRPK1] > [SRSF1]). If the enzyme phosphorylates the RS domain in a distributive manner, then free enzyme and phospho-intermediates of the SR protein will be generated during the reaction that can be trapped by the inhibitor or kdSRPK1 and lead to reaction inhibition [47,48,57,58]. However, if the mechanism is processive, then no free enzyme or phospho-intermediates will be released, and the peptide inhibitor or kdSRPK1 will not be able to stop the reac- tion. For SRSF1, it was found that SRPK1 phosphory- lates, on average, five to eight of the 12 available serines in RS1, using a processive reaction before the enzyme dissociates and continues in a distributive man- ner. These findings suggest that SRPK1 may use a dual-track mechanism, incorporating both processive and distributive phosphorylation steps (Fig. 3A). Such G. Ghosh and J. A. Adams SRPK structure and mechanism FEBS Journal 278 (2011) 587–597 Journal compilation ª 2011 FEBS. No claim to original US government works 591 a process is expected to require a stable enzyme–sub- strate complex. Indeed, competition and single turnover analyses indicate that the SRPK1–SRSF1 complex dis- plays unusually high affinity, with a K d between 50 and 100 nm [48,57]. It is likely that this initial high affinity is diminished during subsequent phosphorylation steps, driving a shift from processive to distributive phos- phorylation. Accordingly, it has been shown that SRPK1 inefficiently pulls down phosphorylated SRSF1, whereas the unphosphorylated SR protein is robustly pulled down [29,59]. Although this mechanism has been established with the use of SRSF1 that has a rather short RS domain, it remains to be seen whether processivity is a general feature of SRPKs and other SR proteins with much larger RS domains. It is inter- esting to note that expanding the number of Arg-Ser repeats in SRSF1 leads to enhanced processivity, sug- gesting that other SR proteins could also be phosphor- ylated by this mechanism [60]. Directional phosphorylation of RS domains Although SRPK1 can processively phosphorylate sev- eral serines in SRSF1, it is not clear how this enzyme attaches phosphates in close succession to a highly charged substrate. DNA polymerase, a classic proces- sive enzyme, adds nucleotide triphosphates in a rigid 5¢fi3¢ direction, and initiates strictly at a DNA pri- mer [61]. To investigate whether SRPK1 is likewise directional, an engineered protease footprinting tech- nique was employed [58]. In these experiments, a lysine is placed in the center of RS1 of SRSF1, and several additional Lys fi Arg mutations in RRM2 are then inserted. When the resulting substrate is cleaved with LysC, two fragments easily identified on a gel can be obtained that correspond to the N-terminal and C-ter- minal halves of RS1. This method permits a fast and quantitative method for sorting phosphates placed on either the N-terminal or C-terminal end of RS1. By monitoring of the phosphorylation reaction in single turnover mode and conversion of the substrate into N-terminal and C-terminal fragments with LysC at various reaction stages, it can be shown that SRPK1 phosphorylates RS1 in a C-terminal to N-terminal direction (Fig. 3B). Furthermore, by alteration of the position of the cleavage site in the RS domain, the ini- tiation region at the C-terminal end of RS1 (initiation box) can also be identified. Interestingly, although SRPK1 prefers to start phosphorylation in the initia- tion box (Ser221–Ser225), mutations in this region do not halt catalysis, indicating that the enzyme possesses the flexibility to move to other sites [58]. This adapt- ability is likely to be an important feature of SRPK1 function, as the RS domains in other SR proteins are larger and more diverse (Fig. 1). In addition to rapid RS1 phosphorylation, SRPK1 is capable of phosphor- ylating about three serines in RS2, although about 100-fold more slowly than the serines in RS1 [60]. This overwhelming specificity for RS1 over RS2 is a result SRPK1 SRSF1 Processive phosphorylation Distributive phosphorylation Directional phosphorylation A B Fig. 3. Mechanism of SRSF1 phosphoryla- tion by SRPK1. (A) Dual-track mechanism. Start-trap analyses indicate that SRPK1 can phosphorylate up to eight serines in RS1, using a processive mechanism in which the kinase stays attached to the substrate after each round of phosphorylation. The remain- ing serines in RS1 are modified in a distribu- tive manner, in which the kinase and substrate dissociate after each phosphoryla- tion event. (B) Directional phosphorylation. Mapping studies show that SRPK1 is a directional kinase that initially binds to an ini- tiation box (Ser221–Ser225) in the center of the RS domain, and then moves in an N-ter- minal direction to maximally phosphorylate RS1. The bold and light arrows indicate that processivity is progressively diminished as SRPK1 translates from the C-terminus to the N-terminus and dissociation becomes favored over forward catalysis. SRPK structure and mechanism G. Ghosh and J. A. Adams 592 FEBS Journal 278 (2011) 587–597 Journal compilation ª 2011 FEBS. No claim to original US government works of SRPK1’s preference for long Arg-Ser repeats, as adding such repeats greatly increases phosphorylation rates in RS2. These findings suggest that SRPK1 scans RS domains in a search for long Arg-Ser stretches, and is clearly capable of docking at additional sites on the basis of local sequence factors. Overall, SRPK1 moves in a well-defined C-terminal to N-terminal direction along the RS domain of SRSF1, and possibly could use a similar mechanism for other SR proteins, although it may be capable of recognizing different and, possibly, multiple initiation boxes. Docking interactions guide multisite phosphorylation Studies on the SRPK1-dependent phosphorylation of SRSF1 have uncovered a remarkable catalytic mecha- nism, displaying very unusual features. How SRPK1 achieves multisite and directional phosphorylation at the molecular level has recently been revealed through the X-ray structures of SRPK1 bound to either a short peptide substrate (Fig. 2A) or the core region of SRSF1 (RRM2-RS1) (Fig. 4A). These two structures show that SRPK1 possesses a docking region in the large lobe that can accept a portion of the RS domain. This acidic docking groove in the kinase accommo- dates basic peptides about six to seven residues in length. Mutation of several acidic residues within the docking groove (e.g. Asp564, Glu571, and Asp548) eliminates processive phosphorylation and strong directional preferences within the RS domain [48]. The peptide-bound form of SRPK1 allowed identification of a small segment preceding the RS domain of SRSF1 [(RVKVDGPR(191–198)] as the cognate substrate site that specifically interacts with the docking groove. Mutations of two basic residues in this segment (R191A and K193A) altered the catalytic mechanism, suggesting the importance of this region in SR protein A B Fig. 4. Model describing how the RS domain of SRSF1 is threaded into the active site of SRPK1. (A) X-ray structure of the SRPK1–SRSF1 complex. SRSF1 retained the central RRM2 and RS1 segments and lacked RRM1 and RS2. Only a portion of RS1 is well defined in the complex (N¢-RS1, residues 204–210), and resides in the elec- tronegative docking groove. The dotted cir- cles present the possible path of the segment of SRSF1 disordered in the crystal from N¢-RS1 to RRM2 and the active site. The surface rendition of CLK1 is shown in the right panel. The dotted circles represent a possible path of the p-RS1 peptide sub- strate on the kinase. (B) Feeding mecha- nism. The N-terminal portion of RS1 (N¢-RS1) initially binds in the docking groove, and the C-terminal portion (initiation box) occupies the active site. Representative Arg-Ser pairs in both segments are repre- sented as green hexagons. The dotted line represents intervening Arg-Ser-rich regions in RS1. In the presence of ATP, RS1 is phosphorylated in a C-terminal to N-terminal direction until residues 191–198 (b4of RRM2) occupy the docking groove. Electro- positive side chains from the P+2 pocket stabilize the phosphates on RS1. G. Ghosh and J. A. Adams SRPK structure and mechanism FEBS Journal 278 (2011) 587–597 Journal compilation ª 2011 FEBS. No claim to original US government works 593 phosphorylation [40]. However, a subsequent structure that cocrystallized with a truncated form of SRSF1 (RRM2-RS1) revealed that the N-terminal part of the RS domain rather than residues 191–198 was bound to the docking groove (Fig. 4A). This was surprising, as this RS segment [N¢-RS1; SYGRSRSRSR(201–210)], binds to a pocket far from the active site (Fig. 4A), but eventually undergoes phosphorylation, as deter- mined by mapping studies [58]. These two kinase struc- tures appeared to offer differing perspectives on which regions outside the RRMs bind in the docking groove. In the RRM2-RS1-bound structure, the docking groove binds an N-terminal segment of RS1 (resi- dues 201–210), whereas in the peptide-bound structure, the docking groove binds sequences that are more N-terminal from N¢-RS1 (residues 191–198). As prior mapping studies showed that SRPK1 moves along the RS domain in a C-terminal to N-ter- minal direction (Fig. 3B), it is possible that the struc- ture of the SRPK1–SRSF1 complex changes as a function of phosphorylation, and that the two X-ray structures present two distinct states along the catalytic pathway. This model was tested with mutant forms of SRPK1 and SRSF1 that differentially cross-link as a function of ATP. A cysteine placed in the docking groove of SRPK1 (K604C) cross-links with a cysteine substituted in the segment preceding the RS domain (K193C) only in the presence of ATP. In comparison, a second mutant form of SRSF1 in which a cysteine is inserted in N¢-RS1 (R204C) cross-links with the dock- ing groove cysteine in the absence of ATP. When con- sidered in light of the directional phosphorylation mechanism, these structural observations can be used to propose a model for substrate phosphorylation in which the Arg-Ser repeat motif constitutes a mobile docking element, where the part of RS1 that is to be phosphorylated (N¢-RS1; residues 204–210) first serves as a docking sequence placing a C-terminal serine from the initiation box at the active site (Fig. 4B). As each serine undergoes phosphorylation, the docking motif moves by two residue increments towards the N-termi- nus. Each Arg-Ser tract from the docking groove is sequentially displaced by an N-terminal tract with the originally identified docking motif in the docking groove at the end of the reaction. In essence, the entire RS1 motif is fed through the active site of the kinase until the furthest N-terminal docking motif (resi- dues 191–196) ‘hits’ the kinase docking groove. Inter- estingly, residues 191–196 lie in b-strand 4 of RRM2, so that it must unfold in order to occupy the docking groove, a result supported by CD and mutagenesis experiments [28,62]. Although the C-terminal residues of RS1 are poorly defined in the structure, a single phosphoserine resulting form a small impurity in the cocrystallized nucleotide analog (AMPPNP) was found in the basic P+2 pocket of the kinase (Fig. 4B). Muta- tions in this pocket (R515A, R518A, and R561A) reduce the rate of phosphate incorporation into the N- Sky1p SRPK1 CLK1 Fig. 5. Surface electrostatic properties of SRPKs and CLKs. Ribbon (top) and electro- static surface presentations (bottom) for Sky1p, SRPK1 and CLK1 are displayed. All three molecules were crystallized as trun- cated proteins. The nonconserved N-termi- nal and spacer domains were deleted in Sky1p and SRPK1. The N-terminal RS domain was deleted in CLK1. SRPK structure and mechanism G. Ghosh and J. A. Adams 594 FEBS Journal 278 (2011) 587–597 Journal compilation ª 2011 FEBS. No claim to original US government works terminal portion of RS1 [48], suggesting that the P+2 pocket stabilizes the growing phosphorylated RS domain. Although structural studies on SRPK1 are the most advanced at this time, it is likely that other SR-direc- ted protein kinases will use aspects of the above ‘feed- ing’ mechanism. For example, the yeast SRPK, Sky1p, contains a similar charged docking groove to that of SRPK1, which plays a role in the recognition of its cognate substrate Npl3 (Fig. 5). Although Npl3 lacks a classic RS domain, it has a single RS dipeptide at the very C-terminus of its RGG (Arg-Gly-Gly-rich) domain. In vitro studies on Sky1p and Npl3 have shown that the RGG domain contains multiple dock- ing motifs, at least one of which is essential for the interaction of Npl3 with Sky1p [63]. Although Sky1p modifies a rather distinct substrate as compared with SRSF1, it appears that the mobile docking element may be a conserved feature in SR and SR-like pro- teins and their kinases. In comparison with SRPK1, the X-ray structures of the CLKs revealed no deep groove that would fit a peptide with geometric com- plementarity (Fig. 4A). Moreover, the corresponding segment that would constitute the SRPK1 docking groove is shallow and dispersed, with both acidic and basic charge patches (Fig. 5). This is comparable to the highly acidic nature of the SRPK1 docking groove. This charge distribution suggests that the hypo- phosphorylated RS domain with alternate positive and negative charges could interact with CLK with high efficiency as compared with the unphosphory- lated RS domain. That is, the product of SRPK1 phosphorylation might be the substrate of CLK. We showed that CLK1 will readily phosphorylate approx- imately seven serines in RS2 in SRSF1 when it is prephosphorylated in RS1 by SRPK1 [47]. In compar- ison, SRPK1 can phosphorylate about three serines in RS2 but very inefficiently [60]. The differences between SRPK1 and CLK1 are likely to be rooted in differences in docking elements and charge dispersal (Fig. 5). Whereas SRPK1 catalyzes a very strict, directional mechanism, owing to its electronegative docking groove, CLK1 lacking such a groove ran- domly phosphorylates the RS domain of SRSF1 [29]. Our understanding of how CLKs modify RS domains will be greatly advanced with the generation of a CLK:RS domain structure and further investigations into its substrate specificity. Conclusion Recent structural and mechanistic studies on the splic- ing kinase SRPK1 have uncovered a novel phosphory- lation mechanism, in which a long section of the substrate’s RS domain is fed into the active site through a docking groove in the large lobe (Fig. 4). This mechanism has similarities to polymerase-type chain reactions, where the enzyme binds in a defined region and then proceeds in a directional manner. SRPK1 starts in a narrow initiation box that is defined by the length of a greater binding channel encompass- ing the docking groove and the active site, a total dis- tance that can accommodate RS1 of the SR protein SRSF1. After initiation, the driving force for the direc- tional reaction is likely to involve a combination of repulsive interactions between the phosphoserines and the electronegative channel, and attractive electrostatic interactions between the phosphoserines and an elec- tropositive P+2 pocket. Whether discrete initiation and extension reactions like those found in SRSF1 are common within the SR protein family awaits further investigations. Although SRSF1 is a prototype for the family and the first to be investigated at a refined mechanistic level, it possesses a relatively small RS domain as compared with others in the SR protein family. It will be interesting to learn how the catalytic principles uncovered for SRSF1 apply to SR proteins with considerably larger RS domains with multiple, lengthy Arg-Ser repeats. Acknowledgements This work was supported by NIH grants to J. A. Adams (GM67969) and G. Ghosh (GM084277). References 1 Tazi J, Bakkour N & Stamm S (2009) Alternative splicing and disease. Biochim Biophys Acta 1792, 14– 26. 2 Venables JP (2006) Unbalanced alternative splicing and its significance in cancer. Bioessays 28, 378–386. 3 Faustino NA & Cooper TA (2003) Pre-mRNA splicing and human disease. Genes Dev 17, 419–437. 4 Jurica MS & Moore MJ (2003) Pre-mRNA splicing: awash in a sea of proteins. Mol Cell 12, 5–14. 5 Wu JY & Maniatis T (1993) Specific interactions between proteins implicated in splice site selection and regulated alternative splicing. Cell 75, 1061–1070. 6 Kohtz JD, Jamison SF, Will CL, Zuo P, Luhrmann R, Garcia-Blanco MA & Manley JL (1994) Protein–protein interactions and 5¢ -splice-site recognition in mammalian mRNA precursors. Nature 368, 119–124. 7 Roscigno RF & Garcia-Blanco MA (1995) SR proteins escort the U4 ⁄ U6.U5 tri-snRNP to the spliceosome. RNA 1, 692–706. G. Ghosh and J. A. Adams SRPK structure and mechanism FEBS Journal 278 (2011) 587–597 Journal compilation ª 2011 FEBS. No claim to original US government works 595 8 Chew SL, Liu HX, Mayeda A & Krainer AR (1999) Evidence for the function of an exonic splicing enhancer after the first catalytic step of pre-mRNA splicing. Proc Natl Acad Sci USA 96, 10655–10660. 9 Lin S, Coutinho-Mansfield G, Wang D, Pandit S & Fu XD (2008) The splicing factor SC35 has an active role in transcriptional elongation. Nat Struct Mol Biol 15, 819–826. 10 Huang Y, Yario TA & Steitz JA (2004) A molecular link between SR protein dephosphorylation and mRNA export. Proc Natl Acad Sci USA 101, 9666– 9670. 11 Huang Y & Steitz JA (2001) Splicing factors SRp20 and 9G8 promote the nucleocytoplasmic export of mRNA. Mol Cell 7, 899–905. 12 Sanford JR, Gray NK, Beckmann K & Caceres JF (2004) A novel role for shuttling SR proteins in mRNA translation. Genes Dev 18, 755–768. 13 Sanford JR, Ellis JD, Cazalla D & Caceres JF (2005) Reversible phosphorylation differentially affects nuclear and cytoplasmic functions of splicing factor 2/alter- native splicing factor. Proc Natl Acad Sci USA 102, 15042–15047. 14 Labourier E, Rossi F, Gallouzi IE, Allemand E, Divita G & Tazi J (1998) Interaction between the N-terminal domain of human DNA topoisomerase I and the argi- nine-serine domain of its substrate determines phos- phorylation of SF2 ⁄ ASF splicing factor. Nucleic Acids Res 26, 2955–2962. 15 Xiao R, Sun Y, Ding JH, Lin S, Rose DW, Rosenfeld MG, Fu XD & Li X (2007) Splicing regulator SC35 is essential for genomic stability and cell proliferation dur- ing mammalian organogenesis. Mol Cell Biol 27, 5393– 5402. 16 Gui JF, Lane WS & Fu XD (1994) A serine kinase reg- ulates intracellular localization of splicing factors in the cell cycle. Nature 369, 678–682. 17 Colwill K, Pawson T, Andrews B, Prasad J, Manley JL, Bell JC & Duncan PI (1996) The Clk ⁄ Sty protein kinase phosphorylates SR splicing factors and regu- lates their intranuclear distribution. EMBO J 15, 265– 275. 18 Duncan PI, Howell BW, Marius RM, Drmanic S, Dou- ville EM & Bell JC (1995) Alternative splicing of STY, a nuclear dual specificity kinase. J Biol Chem 270, 21524–21531. 19 Black DL (2003) Mechanisms of alternative pre-messen- ger RNA splicing. Annu Rev Biochem 72, 291–336. 20 Caceres JF & Krainer AR (1993) Functional analysis of pre-mRNA splicing factor SF2 ⁄ ASF structural domains. EMBO J 12, 4715–4726. 21 Zuo P & Manley JL (1994) The human splicing factor ASF ⁄ SF2 can specifically recognize pre-mRNA 5¢ splice sites. Proc Natl Acad Sci USA 91, 3363–3367. 22 Long JC & Caceres JF (2009) The SR protein family of splicing factors: master regulators of gene expression. Biochem J 417, 15–27. 23 Wang HY, Xu X, Ding JH, Bermingham JR Jr & Fu XD (2001) SC35 plays a role in T cell development and alternative splicing of CD45. Mol Cell 7, 331– 342. 24 Xu X, Yang D, Ding JH, Wang W, Chu PH, Dalton ND, Wang HY, Bermingham JR Jr, Ye Z, Liu F et al. (2005) ASF ⁄ SF2-regulated CaMKIIdelta alternative splicing temporally reprograms excitation–contraction coupling in cardiac muscle. Cell 120, 59–72. 25 Tintaru AM, Hautbergue GM, Hounslow AM, Hung ML, Lian LY, Craven CJ & Wilson SA (2007) Struc- tural and functional analysis of RNA and TAP binding to SF2 ⁄ ASF. EMBO Rep 8, 756–762. 26 Haynes C & Iakoucheva LM (2006) Serine ⁄ arginine- rich splicing factors belong to a class of intrinsically disordered proteins. Nucleic Acids Res 34 , 305–312. 27 Hamelberg D, Shen T & McCammon JA (2007) A pro- posed signaling motif for nuclear import in mRNA pro- cessing via the formation of arginine claw. Proc Natl Acad Sci USA 104, 14947–14951. 28 Ngo JC, Giang K, Chakrabarti S, Ma CT, Huynh N, Hagopian JC, Dorrestein PC, Fu XD, Adams JA & Ghosh G (2008) A sliding docking interaction is essen- tial for sequential and processive phosphorylation of an SR protein by SRPK1. Mol Cell 29, 563–576. 29 Ma CT, Ghosh G, Fu XD & Adams JA (2010) Mechanism of dephosphorylation of the SR protein ASF ⁄ SF2 by protein phosphatase 1. J Mol Biol 403, 386–404. 30 Xiao SH & Manley JL (1998) Phosphorylation–dephos- phorylation differentially affects activities of splicing factor ASF ⁄ SF2. EMBO J 17, 6359–6367. 31 Mermoud JE, Cohen PT & Lamond AI (1994) Regula- tion of mammalian spliceosome assembly by a protein phosphorylation mechanism. EMBO J 13, 5679–5688. 32 Cao W, Jamison SF & Garcia-Blanco MA (1997) Both phosphorylation and dephosphorylation of ASF ⁄ SF2 are required for pre-mRNA splicing in vitro. RNA 3, 1456–1467. 33 Colwill K, Feng LL, Yeakley JM, Gish GD, Caceres JF, Pawson T & Fu XD (1996) SRPK1 and Clk ⁄ Sty protein kinases show distinct substrate specificities for serine ⁄ arginine-rich splicing factors. J Biol Chem 271, 24569–24575. 34 Zhu J & Krainer AR (2000) Pre-mRNA splicing in the absence of an SR protein RS domain. Genes Dev 14, 3166–3178. 35 Massiello A & Chalfant CE (2006) SRp30a (ASF ⁄ SF2) regulates the alternative splicing of caspase-9 pre- mRNA and is required for ceramide-responsiveness. J Lipid Res 47, 892–897. SRPK structure and mechanism G. Ghosh and J. A. Adams 596 FEBS Journal 278 (2011) 587–597 Journal compilation ª 2011 FEBS. No claim to original US government works [...]... (1997) The activating dual phosphorylation of MAPK by MEK is nonprocessive Biochemistry 36, 5929–5933 Adams JA (2001) Kinetic and catalytic mechanisms of protein kinases Chem Rev 101, 2271–2290 Wang HY, Lin W, Dyck JA, Yeakley JM, Songyang Z, Cantley LC & Fu XD (1998) SRPK2: a differentially expressed SR protein- specific kinase involved in mediating the interaction and localization of pre-mRNA splicing factors... (2005) Interplay between SRPK and Clk ⁄ Sty kinases in phosphorylation of the splicing factor ASF ⁄ SF2 is regulated by a docking motif in ASF ⁄ SF2 Mol Cell 20, 77– 89 41 Bullock AN, Das S, Debreczeni JE, Rellos P, Fedorov O, Niesen FH, Guo K, Papagrigoriou E, Amos AL, Cho S et al (2009) Kinase domain insertions define distinct roles of CLK kinases in SR protein phosphorylation Structure 17, 352–362 42... dispens- SRPK structure and mechanism 50 51 52 53 54 55 56 57 58 59 60 61 62 63 able for processive phosphorylation by Src and anchorage-independent cell growth J Biol Chem 281, 20689– 20697 Pellicena P & Miller WT (2001) Processive phosphorylation of p130Cas by Src depends on SH3–polyproline interactions J Biol Chem 276, 28190–28196 Schulze WX, Deng L & Mann M (2005) Phosphotyrosine interactome of the ErbB-receptor... Manley JL (2003) Regulation and substrate specificity of the SR protein kinase Clk ⁄ Sty Mol Cell Biol 23, 4139–4149 43 Hubbard SR (1997) Crystal structure of the activated insulin receptor tyrosine kinase in complex with peptide substrate and ATP analog EMBO J 16, 5572–5581 44 Russo AA, Jeffrey PD & Pavletich NP (1996) Structural basis of cyclin-dependent kinase activation by phosphorylation Nat Struct... studies of the dual phosphorylation of mitogen-activated protein kinase J Biol Chem 272, 19008–19016 Jeffery DA, Springer M, King DS & O’Shea EK (2001) Multi-site phosphorylation of Pho4 by the cyclin-CDK Pho80–Pho85 is semi-processive with site preference J Mol Biol 306, 997–1010 Aubol BE, Chakrabarti S, Ngo J, Shaffer J, Nolen B, Fu XD, Ghosh G & Adams JA (2003) Processive phosphorylation of alternative... multi-site phosphorylation of the splicing factor ASF ⁄ SF2 by SRPK1 J Mol Biol 376, 55–68 Koizumi J, Okamoto Y, Onogi H, Mayeda A, Krainer AR & Hagiwara M (1999) The subcellular localization of SF2 ⁄ ASF is regulated by direct interaction with SR protein kinases (SRPKs) J Biol Chem 274, 11125– 11131 Ma CT, Hagopian JC, Ghosh G, Fu XD & Adams JA (2009) Regiospecific phosphorylation control of the SR protein. .. kinetic analysis of ASF ⁄ SF2 phosphorylation by SRPK1 and Clk ⁄ Sty J Biol Chem 280, 41761–41768 48 Hagopian JC, Ma CT, Meade BR, Albuquerque CP, Ngo JC, Ghosh G, Jennings PA, Fu XD & Adams JA (2008) Adaptable molecular interactions guide phosphorylation of the SR protein ASF ⁄ SF2 by SRPK1 J Mol Biol 382, 894–909 49 Patwardhan P, Shen Y, Goldberg GS & Miller WT (2006) Individual Cas phosphorylation. .. Z, Hurwitz J & O’Donnell M (1998) Processivity of DNA polymerases: two mechanisms, one goal Structure 6, 121–125 Huynh N, Ma CT, Giang N, Hagopian J, Ngo J, Adams J & Ghosh G (2009) Allosteric interactions direct binding and phosphorylation of ASF ⁄ SF2 by SRPK1 Biochemistry 48, 11432–11440 Lukasiewicz R, Nolen B, Adams JA & Ghosh G (2007) The RGG domain of Npl3p recruits Sky1p through docking interactions...G Ghosh and J A Adams 36 Caceres JF, Misteli T, Screaton GR, Spector DL & Krainer AR (1997) Role of the modular domains of SR proteins in subnuclear localization and alternative splicing specificity J Cell Biol 138, 225–238 37 Kataoka N, Bachorik JL & Dreyfuss G (1999) Transportin-SR, a nuclear import receptor for SR proteins J Cell Biol 145, 1145–1152 38 Lai MC,... Lin RI, Huang SY, Tsai CW & Tarn WY (2000) A human importin-beta family protein, transportin-SR2, interacts with the phosphorylated RS domain of SR proteins J Biol Chem 275, 7950–7957 39 Ding JH, Zhong XY, Hagopian JC, Cruz MM, Ghosh G, Feramisco J, Adams JA & Fu XD (2006) Regulated cellular partitioning of SR protein- specific kinases in mammalian cells Mol Biol Cell 17, 876–885 40 Ngo JC, Chakrabarti . ARTICLE Phosphorylation mechanism and structure of serine-arginine protein kinases Gourisankar Ghosh 1 and Joseph A. Adams 2 1 Department of Chemistry and. pro- tein substrates and the recent structures of CLK1 and CLK3, suggest an elegant mechanism of recognition and phosphorylation by these two kinases, which