Rules of engagement: co-transcriptional recruitment of pre-mRNA processing factors David L Bentley The universal pre-mRNA processing events of 50 end capping, splicing, and 30 end formation by cleavage/polyadenylation occur co-transcriptionally. As a result, the substrate for mRNA processing factors is a nascent RNA chain that is being extruded from the RNA polymerase II exit channel at 10���30 bases per second. How do processing factors find their substrate RNAs and complete most mRNA maturation before transcription is finished? Recent studies suggest that this task is facilitated by a combination of protein���RNA and protein��� protein interactions within a ���mRNA factory��� that comprises the elongating RNA polymerase and associated processing factors. This ���factory��� undergoes dynamic changes in composition as it traverses a gene and provides the setting for regulatory interactions that couple processing to transcriptional elongation and termination. Addresses Department of Biochemistry and Molecular Genetics, University of Colorado School of Medicine, UCHSC at Fitzsimons, Mail Stop 8101, PO Box 6511, Aurora Colorado 80045, USA Corresponding author: Bentley, David L (David.Bentley@UCHSC.edu) Current Opinion in Cell Biology 2005, 17:251���256 This review comes from a themed issue on Nucleus and gene expression Edited by Christine Guthrie and Joan Steitz Available online 13th April 2005 0955-0674/$ ��� see front matter # 2005 Elsevier Ltd. All rights reserved. DOI 10.1016/j.ceb.2005.04.006 Introduction: how does coupling to transcription enhance pre-mRNA processing? How mRNA processing is facilitated by coupling to RNA polymerase II (pol II) transcription is a fascinating pro- blem that is the subject of several excellent reviews [1���5]. Transcription-coupled processing differs from uncoupled processing in that the substrate RNA is a growing and progressively folding structure rather than a static full- length pre-mRNA. The importance of coupling is sug- gested by the fact that processing of full-length synthetic pre-mRNAs in injected oocytes is less efficient than co- transcriptional processing in vivo [6]. In vivo, introns can be removed and the poly(A) site cleaved by the time polymerase has transcribed only 1 kb beyond the proces- sing sites, probably within 30s [7]. By contrast, in vitro processing uncoupled from transcription usually takes 20 min. Optimal processing is achieved by coupling with transcription by RNA pol II and not other poly- merases because pol II is uniquely equipped with an unusual domain on its large subunit, called the C-terminal domain (CTD), that provides a landing pad for mRNA processing factors. Coupling of pol II transcription with processing can influence processing reactions in at least three ways. First, localization: in its simplest form, coupling positions mRNA processing factors at the elongation complex, raising their local concentration in the vicinity of the nascent transcript. Second, kinetic coupling: the rate of transcript elongation can have profound effects on RNA folding and the assembly of RNA���protein complexes and has been shown to affect the choice between alternative processing sites [8,9]. Third, allostery: contacts between mRNA processing factors and the pol II elongation com- plex can allosterically activate or inhibit mRNA proces- sing factors [10]. Here I will review recent progress in our understanding of how the factors that carry out mRNA capping, splicing and 30 end formation engage the pol II elongation complex. The pol II C-terminal domain: a recruitment platform The CTD is an essential domain in the large subunit of pol II, but is absent from the related subunits of RNA polymerases I and III. This domain comprises tandem heptads whose consensus sequence, Y1S2P3T4S5P6S7, is identical across animals, plants and some protozoa. The in vivo functional unit of the CTD appears to be a pair of tandem heptads [11]. A recent proteomic analysis iden- tified over 100 yeast proteins that bind to the phos- phorylated CTD [12 ]. The CTD is more than a passive landing pad, however. Among its numerous roles, the CTD can allosterically regulate capping enzymes and regulate transcriptional elongation and termination [13]. CTD deletion prevents efficient co-transcriptional cap- ping, splicing and 30 end formation in metazoans [14,15]. Although it is essential for co-transcriptional processing, the CTD is dispensable for processing uncoupled from transcription in injected Xenopus oocytes [6], suggesting that processing at the site of transcription differs from post-transcriptional processing elsewhere in the nucleus. Important clues to how the CTD works has come from in vitro systems, in which it can stimulate processing reac- tions, in some cases even in the absence of ongoing transcription [16,17 ]. www.sciencedirect.com Current Opinion in Cell Biology 2005, 17:251���256

The CTD is phosphorylated on Ser5 residues of the heptads by the TFIIH-associated kinase when transcrip- tion initiates, and later the Ser 2 residues are phosphory- lated by the kinases CTK1 (yeast) and PTEFb (positive transcription elongation factor b or CDK9) [18]. Decora- tion of the CTD with phosphates is also fashioned by several phosphatases which differ in their preferences for Ser2 or Ser5 phosphate: Fcp1 [19], Ssu72 [20 ] and SCPs [21] The ratio of Ser5:Ser2 phosphorylation is high at the 50 end and lower at the 30 end [22]. A technical limitation is that analysis of Ser2 phosphorylation has relied on one monoclonal antibody, H5, which has highest affinity for heptads phosphorylated on both Ser2 and Ser5 [23]. CTD conformation is modified by the peptidyl-prolyl isomerase Pin1 (Ess1 in yeast). How this enzyme affects recruitment of mRNA processing factors is unknown however, it does affect CTD kinase and phosphatase activity by remodelling the substrate [24]. The CTD is also O-glycosylated in a manner that is mutually exclusive with serine phosphorylation, suggesting that there could be cross-talk between these two modifications [25]. CTD peptides do not assume fixed conformations regard- less of phosphorylation state [26 ]. Rather, the CTD is quite malleable in the grip of its binding partners. There are three identified protein domains that recognize CTD heptads: the CTD interacting domain [27], the WW domain [28] and the FF domain [29], but other proteins that lack these domains can also bind directly to the CTD. There are three available structures of phosphorylated CTD peptides bound to different partners: Pin1, which binds via a WW domain the guanylyltransferase Cgt1 and the CTD interacting domain of Pcf11, a cleavage/poly- adenylation factor [28,30 ,31]. These structures show quite different CTD conformations, although the prolyl bonds are exclusively trans isomers [32]. The Pcf11 CTD- interacting domain, which comprises eight a-helices in a superhelical arrangement, binds Ser 2-phosphorylated heptads by an induced-fit mechanism [26 ,30 ]. The extreme flexibility of the CTD helps explains how it can bind so many partners. The changing pattern of covalent and non-covalent modifications between the 50 and 30 ends of the gene may constitute a CTD code [33] that is translated into a sequence of processing factor binding and release reactions as transcription proceeds. The code seems to be quite a loose one, however, as Ser 2 phosphorylation is not essential for viability in yeast [34]. Capping enzymes: processing factors with transcriptional functions The first mRNA processing factors to be recruited to the CTD during the transcription cycle are the capping enzymes RNA triphosphatase, guanylyltransferase and 7-methyltransferase. Yeast guanylyltransferase and 7-methyltransferase bind directly and independently to the phosphorylated CTD and phosphorylation of Ser 5 at the promoter by Kin28, a subunit of TFIIH, is necessary for their recruitment [22,35]. Hence, recognition of a Ser- 5-phosphate ���code word��� permits recruitment of capping enzymes at the right time to perform co- transcriptional capping. CTD binding not only localizes the mammalian guanylyltransferase but also allosterically regulates it, reducing the Km for GTP [10]. Removal of Ser5 phosphate by phosphatases early in elongation is coupled to release of capping enzymes. Guanylyltransfer- ase is released within the first 500 bases of the gene, whereas 7-methyltransferase remains bound throughout the length of the gene [22,35] (Figure 1 and 2). Capping enzymes participate in a network of interactions that regulate early steps in transcription [36]. The yeast 7-methyltransferase (Abd1) stabilizes pol II on some promoters and both 7-methyltransferase and guanylyl- transferase (Ceg1) stimulate early elongation [37,38], whereas the RNA triphosphatase (Cet1) inhibits re-initia- tion [39]. Stimulation of transcription by Abd1 occurs in a methylation-defective mutant and is therefore indepen- dent of capping itself [38]. Human capping enzyme (CE) stimulates promoter escape (Figure 1) by countering the negative elongation factor NELF, and CE recruitment is 252 Nucleus and gene expression Figure 1 MT MT MT GT CTD High Ser5 PO4:Ser2 PO4 MeGpppN P5 P5 P5 P2 P2 P5 C/P Rat1 C/P Rat1 CBC Pol II Current Opinion in Cell Biology P5 P5 GT P5 GT P5 Co-transcriptional recruitment of pre-mRNA processing factors at the 50 end of a gene. Capping enzymes (guanylyltransferase [GT] and 7-methyltransferase [MT]), cap binding complex (CBC), cleavage/ polyadenylation factors (C/P), and the RNA 50���30 exonuclease, Rat1, are indicated. Processing factors interact with pol II elongation complex via the CTD (green line) and probably also via the nascent RNA (red line). Phosphorylation of Ser2 and Ser5 residues in the CTD heptad repeats are marked P2 and P5. Capping enzymes stimulate early steps in pol II transcription including promoter clearance denoted by the thick arrow [17 ,38]. Following addition of the cap, MeGppp, GT is released from the elongation complex. Current Opinion in Cell Biology 2005, 17:251���256 www.sciencedirect.com