|Home | About | Journals | Submit | Contact Us | Français|
Many steps in nuclear RNA processing, surveillance and degradation require TRAMP, a complex containing the poly(A) polymerase Trf4p, the Zn-knuckle protein Air2p, and the RNA helicase Mtr4p. TRAMP polyadenylates RNAs designated for decay or trimming by the nuclear exosome. It has been unclear how polyadenylation by TRAMP differs from polyadenylation by conventional poly(A) polymerase, which produces poly(A) tails that stabilize RNAs. Using reconstituted S.cerevisiae TRAMP, we show that TRAMP inherently suppresses poly(A) addition after only 3–4 adenosines. This poly(A) tail length restriction is controlled by Mtr4p. The helicase detects the number of 3'-terminal adenosines and, over several adenylation steps, elicits precisely tuned adjustments of ATP affinities and rate constants for adenylation and TRAMP dissociation. Our data establish Mtr4p as critical regulator of polyadenylation by TRAMP and reveal that an RNA helicase can control the activity of another enzyme in a highly complex fashion and in response to features in RNA.
The Trf4/Air2/Mtr4 polyadenylation complex (TRAMP) is critical for many RNA processing events in the cell nucleus (Anderson and Wang, 2009; Houseley and Tollervey, 2009). In addition, TRAMP is essential for the nuclear turnover of incorrectly processed RNAs (Houseley and Tollervey, 2008; Kadaba et al., 2004). TRAMP consists of three subunits that are highly conserved in eukaryotes; a non-canonical poly(A) polymerase (Trf4p or Trf5p in S. cerevisiae), a Zn-knuckle protein (Air2p or Air1p), and a RNA helicase (Mtr4p/Dob1p) (Jensen and Moore, 2005; LaCava et al., 2005; Vanácová et al., 2005; Wyers et al., 2005). TRAMP polyadenylates RNAs designated for further processing or complete 3’ to 5’ degradation by the nuclear exosome (Bonneau et al., 2009; Liu et al., 2006; Lykke-Andersen et al., 2009). TRAMP and the nuclear exosome have also been linked to chromatin maintenance, remodeling, transcriptional regulation by non-coding RNAs, and to DNA repair (Egecioglu et al., 2006; Houseley and Tollervey, 2008; San Paolo et al., 2009). In these processes, TRAMP and the nuclear exosome are thought to function analogously to their roles in RNA processing and degradation; the nuclear exosome degrades presumably nascent RNAs that are polyadenylated by TRAMP (Houseley and Tollervey, 2008).
Polyadenylation by TRAMP is distinct from mRNA polyadenylation by canonical poly(A) polymerases (PAP)(Anderson and Wang, 2009). Polyadenylation by PAP stabilizes mRNAs and requires specific sequence signals in the RNAs (Keller, 1995; Scorilas, 2002; Wilusz and Spector, 2010). In contrast, polyadenylation by TRAMP designates RNAs for degradation or processing, and TRAMP polyadenylates a large cross-section of diverse RNAs without shared sequence or apparent secondary structure and without common associated proteins (Anderson and Wang, 2009; Houseley and Tollervey, 2008).
Given the opposite goals of the two polyadenylation processes, RNAs polyadenylated by PAP must be distinct from RNAs polyadenylated by TRAMP. Indeed, recent data show a marked difference in the length of the respective poly(A) tails in vivo. While PAP typically appends several dozen to hundreds of adenosines (Keller, 1995), RNAs adenylated by TRAMP contain significantly shorter poly(A) tails (Grzechnik and Kufel, 2008; Keller, 1995; Lebreton et al., 2008). Recent data indicate that the distribution of RNAs polyadenylated by TRAMP shows a clear peak at 4 to 5 added nucleotides (Wlotzka et al., 2011).
How polyadenylation by TRAMP is controlled is unknown and not apparent from the current model of TRAMP function. In this model, Trf4p, the principal poly(A) polymerase, is assisted by Air2p, which is required for Trf4p activity in vitro (Anderson and Wang, 2009; Houseley and Tollervey, 2008). The Ski2-like RNA helicase Mtr4p is thought to unwind RNA duplexes with 3’ to 5’ polarity (Wang et al., 2008). Mtr4p (Dob1p) was the first factor implicated in aiding RNA degradation by the exosome (de la Cruz et al., 1998). The unwinding activity of Mtr4p, shown for the recombinant protein, is thought to stimulate RNA degradation (Bernstein et al., 2008; Wang et al., 2008; Anderson and Wang, 2009; Houseley and Tollervey, 2008). This basic model of TRAMP function has been instructive in assigning enzymatic roles to the components. However, the model cannot explain more intricate TRAMP functions, such as control of poly(A) tail lengths. Most likely, the yet unexplored interplay between the TRAMP components gives rise to these functions.
To examine the interplay between the TRAMP components and to illuminate the molecular basis of the poly(A) lengths control, we quantitatively analyzed individual adenylation steps by TRAMP (Trf4p, Air2p, Mtr4p) from S. cerevisiae. We found that TRAMP inherently limits the poly(A) tail length through modulation of individual adenylation rate constants and ATP affinities. This modulation depends on the number of 3' terminal adenosines, a finding that corresponds to and explains the restriction of poly(A) tails on TRAMP targets in vivo (Wlotzka et al., 2011). Most intriguingly, the RNA helicase Mtr4p elicits the modulation of the polyadenylation activity of TRAMP, by detecting the number of 3' terminal adenosines in the RNA substrate. Mtr4p impacts RNA binding, ATP affinity, and rate constants for both adenylation and TRAMP dissociation, all as a function of the number of 3’ adenosines. The modulation by Mtr4p does not depend on duplex unwinding, but on binding of the helicase to the 3' end of the RNA. Our findings establish Mtr4p as critical regulator of polyadenylation, an unprecedented function for an RNA helicase. Mtr4p controls the activity of the poly(A) polymerase Trf4p in an equally unprecedented fashion, through a series of energetically small, but highly coordinated effects on multiple reaction parameters.
To quantitatively characterize polyadenylation by TRAMP from S. cerevisiae, we reconstituted the complex from recombinant components (Fig.S1A,B). Polyadenylation activity of the reconstituted TRAMP was first measured using tRNAiMet (Fig.1A). This RNA resembles one of the physiological targets of TRAMP, and had been previously used to detect polyadenylation activity of TRAMP obtained from yeast (Kadaba et al., 2004; Wang et al., 2008). Reactions were performed under pre-steady state conditions (i.e., enzyme excess over the substrate), because the kinetic description of this reaction regime contains fewer parameters than steady-state regimes and thus provides the most accurate quantitative data. Polyadenylation timecourses were analyzed by denaturing PAGE, to resolve polyadenylated species at single nucleotide resolution (Fig.1A).
Plots of the fractions of polyadenylated species vs. reaction time revealed accumulation of species with 3 to 5 adenosines in a time window from approximately 1 to 3 minutes (Fig.1B). Over longer reaction times, the poly(A) tail grew to 15 nt and longer. The temporary accumulation of species with 3 – 5 adenosines suggested a modulation of the polyadenylation activity in response to the number of added nucleotides. To test this assertion, we determined rate constants for individual adenylation steps. A simple kinetic scheme consisting of a series of irreversible, pseudo-first order reactions faithfully described the experimental data (Fig.1C). The observed rate constants for individual adenylation steps (kobs) represent multiple physical processes, including adenylation and dissociation of TRAMP from the RNA.
Plots of rate constants vs. the corresponding number of added adenosines revealed a clear modulation of the polyadenylation activity. Rate constants increased for the first three steps, and then decreased to a fairly constant level (Fig.1D). The resulting peak in rate constants explains the temporary accumulation of RNA species with 3 – 5 adenosines in a straightforward manner: TRAMP forms these species relatively fast, but extends them only slowly.
To gain further insight into the molecular basis of the modulated polyadenylation activity, we examined the dependence of individual rate constants on the ATP concentration. We determined the functional affinity for ATP (K1/2,ATP) for each adenylation step and each adenylation rate constant at ATP and TRAMP saturation (kmax, Fig.1E, Fig.S1F–L). The observed peak in polyadenylation rate constants at A3 broadened slightly at ATP saturation (Fig.1E). Most notable was a pronounced peak of low ATP affinity at A5/A6 (Fig.1F). This peak indicates an approximately 20-fold decrease in ATP affinity for the polyadenylation reaction at A5/A6, compared to earlier and later adenylation steps. This drop in ATP affinity occurs immediately after the peak for the highest adenylation rate constant at A3, revealing that the decrease in adenylation rate constants is accompanied by a marked reduction in ATP affinity. Thus, modulation of ATP affinity and adenylation rate constants synergistically favor the temporal accumulation of species with 3 – 5 adenosines.
We next examined whether the physiological TRAMP target, hypomethylated tRNAiMet precursor (pre-tRNAiMet) (Kadaba et al., 2004; Kadaba et al., 2006), accumulated similarly short poly(A) tails in vivo. To measure the poly(A) tail lengths of cellular pre-tRNAiMet with single base resolution, we adopted a 3’ RACE strategy (Fig.2A). We isolated total RNA from the yeast trm6-504 strain, where non-functional tRNA m1A methylase Trm6p leads to accumulation of hypomethylated pre-tRNAiMet, which is targeted by TRAMP (Kadaba et al., 2004). Following the extension of the RNA 3' ends with guanosine-inosine tails, polyadenylated pre-tRNAiMet were specifically amplified by RT-PCR. We accounted for the heterogeneity in the 3' ends of pre-tRNAiMet (Kadaba et al., 2004) by processing of the PCR products with the restriction enzyme MseI, which was possible because all precursors end with at least two 3' Us (Fig.2A). The isolated pre-tRNAiMet poly(A) tails were ligated to a piece of synthetic DNA, amplified and subjected to Sanger sequencing to delineate the number of added adenosines (Fig.2A).
The method was calibrated with a tRNAiMet processed in vitro (gel panel in Fig.2B). The corresponding sequencing chromatogram shows excellent agreement between input and final sequencing result (Fig.2B). The robustness of the method was further tested with longer, in vitro generated poly(A) tail lengths, and similar agreements were seen (H.J. et al., unpublished results). We then measured the lengths of the poly(A) tails of pre-tRNAiMet appended in vivo (Fig.2C). The corresponding Sanger chromatogram indicates accumulation of RNA species with roughly 4 adenosines (Fig.2C), in excellent agreement with our polyadenylation measurements in vitro (Fig.1). Accumulation of similarly short poly(A) tails on other TRAMP targets in vivo had been observed by others (Grzechnik and Kufel, 2008; Lebreton et al., 2008; Wlotzka et al., 2011). The striking correlation between the poly(A) tail lengths distribution of TRAMP targets in vivo and the temporary accumulation of short poly(A) tails in vitro is consistent with the notion that TRAMP displays modulated polyadenylation activity in the cell as well.
To investigate how polyadenylation activity by TRAMP was modulated, we next examined in vitro whether the modulation was specific for physiological TRAMP targets, or if TRAMP also polyadenylated simple model RNAs in a similar fashion. First, we tested a substrate consisting of a 16 bp duplex with a single nucleotide overhang at the 3’ end (Fig.3A). The protruding nucleotide was necessary to obtain appreciable levels of polyadenylation. On a 16 bp blunt end duplex, TRAMP displayed exceedingly low, unquantifiable activity (H.J. et al., unpublished results).
Adenylation rate constants (kmax) for the RNA duplex with the single nucleotide overhang displayed a clear peak, and ATP affinities for individual adenylation steps (K1/2,ATP) showed a pronounced peak of low ATP affinity (Fig.3A), as seen for the tRNAiMet substrate (Fig.1). While both peaks were slightly shifted, compared to the tRNAiMet substrate, the sharp decrease in ATP affinity coincided again with the decrease in adenylation rate constants (Fig.3A). Extending the duplex to 23 bp had little effects on overall adenylation rate constants, presence of the characteristic peak in adenylation rate constants, and the corresponding decrease in ATP affinity, although the peaks were slightly shifted, compared to the 16 bp duplex (Fig.3B). The data obtained with these simplified model substrates clearly indicated that modulated polyadenylation activity is not restricted to physiological targets, but an inherent feature of TRAMP.
The slight shifts of the peaks for polyadenylation rate constants and ATP affinities for the different substrates suggested potential effects of RNA structure on the modulation. It had been shown that RNA structure affects Mtr4p binding (Weir et al., 2010), but it was also possible that the modulation with the tested substrates was caused by Mtr4p-mediated duplex unwinding. To test whether the modulation of polyadenylation depended on duplex unwinding, we measured polyadenylation of a single-stranded RNA. If modulation required unwinding, then the absence of duplexes would eliminate or drastically change the modulation. While TRAMP displayed only weak, unquantifiable activity on a 17 nt ssRNA (Fig.S3D,E), a 24 nt ssRNA was robustly adenylated (Fig.3C). Both adenylation rate constants (kmax) and ATP affinities for the individual adenylation steps (K1/2,ATP) displayed the peaks indicating modulated polyadenylation (Fig.3C). Similar modulation was seen for other ssRNAs longer than 17 nt (H.J. et al., unpublished results). The modulated polyadenylation activity on single stranded RNA indicates that the modulation is not based on duplex unwinding by TRAMP. Notwithstanding, unwinding could still contribute to a small extent to the observed slight influence of RNA secondary structure on TRAMP activity.
If duplex unwinding was not causing the modulation of polyadenylation activity, did the helicase Mtr4p affect the modulation at all? To answer this question, we measured polyadenylation by a TRAMP complex without Mtr4p (Trf4p/Air2p). With the 16 bp duplex substrate described above, Trf4p/Air2p showed only low, unquantifiable levels of polyadenylation activity (Fig.S4A,B). The 24 nt ssRNA and the 23 bp substrates were robustly polyadenylated (Fig.4A,B). With both substrates, Trf4p/Air2p produced longer poly(A) tails than TRAMP over comparable timeframes (Fig.4A,B). Adenylation rate constants increased for the first two steps, but did not produce the characteristic peak seen with complete TRAMP (Fig.3). Similarly absent was the peak for ATP affinities (Fig.4A,B). These observations demonstrate that Trf4p/Air2p does not display the modulated polyadenylation activity seen with complete TRAMP, thus indicating a critical role of Mtr4p in the modulation. In striking correlation with our observations, Mtr4p depletion in vivo causes hyperadenylation of TRAMP targets (Houseley and Tollervey, 2006).
To illuminate how Mtr4p contributed to the modulation, we examined a TRAMP complex with a mutated Mtr4p (TRAMPMtr4-20p). The Mtr4-20p mutation, located in the helicase motif VI, strongly decreases unwinding and RNA-stimulated ATPase activities of Mtr4p, but TRAMPMtr4-20p retains polyadenylation activity (Wang et al., 2008). With all substrates tested, TRAMPMtr4-20p generated longer poly(A) tails than wtTRAMP at comparable reaction times (Fig.4C–E). With the 24 nt ssRNA and the 23 bp substrates, polyadenylation rate constants and ATP affinities showed only very broad peaks, and there was only little, if any, coordination between changes in rate constants and ATP affinity that was seen with wtTRAMP (Fig.4C,D). With the 16 bp duplex substrate, no peaks in adenylation rate constants or ATP affinities were seen (Fig.4E). These results indicate that the Mtr4-20p mutation causes a precipitous loss in the capacity of TRAMP to modulate polyadenylation activity. This observation provides further evidence that Mtr4p plays a critical role in modulating TRAMP polyadenylation activity. Moreover, the data reveal that the presence of Mtr4p in TRAMP alone is not sufficient to modulate polyadenylation. The modulation apparently requires Mtr4p with intact coordination between RNA and ATP binding sites, which is impaired in the Mtr4-20p mutant (Jackson et al., 2010; Wang et al., 2008; Weir et al., 2010).
Having implicated Mtr4p in the modulation of polyadenylation by TRAMP, we next asked how TRAMP determined at which steps to decrease adenylation rate constants and ATP affinities. A central point in this regard was whether TRAMP adjusted its activity only for adenosines that it appended or also for adenosines already present in the RNA. To distinguish between these possibilities, we measured rate constants and ATP affinities for individual adenylation steps with a 24 nt ssRNA substrate containing four 3' terminal adenosine residues (Fig.5A). Rate constants did not display the characteristic peak at A4, but were remarkably similar to those measured for steps > 4 for the substrate without 3’-terminal adenosines (Fig.5A). ATP affinities showed a peak shifted by 4 positions, compared to the substrate without the 3’ terminal adenosines (Fig.5A, lower panel). This characteristic shift was not seen with a 24 nt ssRNA substrate containing four consecutive adenosines within the sequence (Fig.5B). The data demonstrate that TRAMP adjusts its activity based on the presence of a critical number of 3' terminal adenosines, but irrespective of whether or not they are appended by TRAMP.
Since Mtr4p modulated polyadenylation, we next probed whether and how the 3’-terminal nucleotides were detected by Mtr4p during polyadenylation. A recent crystal structure of Mtr4p indicated a potential base recognition site, outside the helicase core (Molecule B in Weir et al., 2010, Fig.6A,B). This structure suggested that E947, which is highly conserved in Mtr4p orthologs, contacts adenosine-specific groups on the fourth base from the 5’ end of the RNA bound in the structure (Fig.6C). Reasoning that E947 might be involved in the identification of the critical number of 3’ terminal adenosines, we replaced E947 with an alanine. TRAMP with Mtr4p(E947A) (TRAMPMtr4p(E947A)) produced longer poly(A) tails than wtTRAMP over identical reactions times (Fig.6D). The peak in adenylation rate constants seen with TRAMPMtr4p(E947A) was significantly broader than with wtTRAMP. No clear peak at all was seen for ATP affinities, which also were much lower for later steps (7–10) than for wtTRAMP (Fig.6E). Thus, TRAMPMtr4p(E947A) markedly diminished the modulation of polyadenylation, similar to the Mtr4-20p mutation in TRAMPMtr4-20p (Fig.S6B). We conclude that E947 is important for modulating polyadenylation. This notion is consistent with a scenario where Mtr4p directly binds the 3’ terminal nucleotides, and upon detection of 3' terminal adenosines, alters the polyadenylation activity of Trf4p.
To further understand how the polyadenylation activity was modulated, it was important to examine whether multiple binding and dissociation events were required until 4 adenosines were added. To answer this question, we determined the processivity of TRAMP for individual adenylation steps (Fig.S7). The processivity is the probability of TRAMP adding the next adenosine vs. dissociating from the substrate (Fig.7A). This probability is directly related to the average number of steps per binding event (Ali and Lohman, 1997, Fig.7A).
Plots of processivity vs. number of added adenosines revealed a steady increase in processivity until P = 0.64 ± 0.10 at A4, followed by a slight decrease to P = 0.40 ± 0.03 at A10 (Fig.7B). These data indicate that TRAMP dissociates roughly four times faster than it adds the adenosine for the first step, thus using about 5 binding events to add the first adenosine. For subsequent steps, dissociation and adenylation are roughly equally fast, i.e., TRAMP adds roughly 2 nucleotides per binding event (Fig.7B). The data show that TRAMP undergoes multiple binding and dissociation cycles to append 4 to 5 adenosines.
We next examined the effects of Mtr4p on TRAMP processivity. TRAMP without Mtr4p (Tr4p/Air2p) showed lower processivity than wtTRAMP for the first four steps. Subsequent steps displayed a slightly higher processivity than wtTRAMP (Fig.7C). To understand the influence of Mtr4p on a more quantitative level, we calculated forward and dissociation rate constants for each adenylation step for TRAMP with and without Mtr4p (Fig.7D,E). Mtr4p enhances polyadenylation rate constants for steps 1 – 3, and then slows these rate constants for subsequent steps (Fig.7D). In addition, Mtr4p enhances TRAMP dissociation for the first step. For subsequent steps, Mtr4p decreases dissociation rate constants, thus prolonging the time TRAMP remains bound to the RNA (Fig.7E).
To visualize the multifaceted, coordinated effects of Mtr4p on the Trf4p activity, we calculated energetic contributions of Mtr4p to adenosine addition, TRAMP dissociation and ATP affinity for individual adenylation steps (Fig.7F,G). Compared to the reaction without Mtr4p, the helicase enhances adenylation rate constants and promotes tighter ATP binding for the first two steps (Fig.7H). For steps 4 and higher Mtr4p slows adenylation rate constants and weakens ATP binding (Fig.7H). Mtr4p slows TRAMP dissociation from the RNA, except for the first step (Fig.7H). In energetic terms, the impact of Mtr4p is greatest on ATP affinities and adenylation rate constants. In general, however, Mtr4p imparts rather small changes on the individual rate constants. Yet, numerous small effects multiply over many steps and thus significantly alter the polyadenylation pattern, compared to the reaction without or with impaired Mtr4p. The coordination between changes in rate constants and changes in ATP affinity provides additional synergy to favor a temporal accumulation of short poly(A) tails (Fig.7H).
In this study, we have shown that the RNA helicase Mtr4p regulates polyadenylation in the TRAMP complex. By modulating individual adenylation steps, Mtr4p facilitates a temporary accumulation of RNAs with short poly(A) tails of only 3 to 4 adenosines in vitro. Strikingly, TRAMP targets in vivo accumulate similarly short poly(A) tails. We demonstrate that hypomethylated pre-tRNAiMet, a prototypical TRAMP target, accumulates poly(A) tails with roughly 4 nucleotides (Fig.2). Previous reports showed similarly short poly(A) tails on other TRAMP targets (Grzechnik and Kufel, 2008; Lebreton et al., 2008), and a recent analysis of a wide range of TRAMP targets found a distribution of poly(A) tails with a pronounced peak at 4 to 5 nucleotides (Wlotzka et al., 2011). The remarkable agreement between these observations made in vivo and our in vitro data suggests a physiological role of Mtr4p in the control of the lengths of poly(A) tails produced by TRAMP. This notion further concurs with data showing hyperadenylation upon Mtr4p depletion in vivo (Houseley and Tollervey, 2006). Removal of Mtr4p in vitro abolishes modulation and eliminates poly(A) tail length restriction (Fig.4). The striking correlation between several lines of experiments in vivo and in vitro is consistent with a pivotal, physiological role of Mtr4p in the regulation of polyadenylation by TRAMP. This role is in addition to the previously shown function of Mtr4p as exosome co-factor (Houseley and Tollervey, 2008; LaCava et al., 2005). Defects in the function as exosome co-factor might also contribute to changes in poly(A) lengths of TRAMP targets seen in vivo with functionally impaired Mtr4p.
The multiple functional roles of Mtr4p mark this RNA helicase as central player in the control of critical steps of the TRAMP-exosome machinery. Regulation of Mtr4p could simultaneously affect multiple steps in TRAMP/exosome-mediated RNA decay and processing. The Mtr4p-mediated restriction of poly(A) tail lengths might also prevent RNAs processed by TRAMP from binding to poly(A) binding protein (Pab1p), which requires at least 12 A for stable binding in yeast (Sachs et al., 1987). In addition, the short poly(A) tails appended by TRAMP may serve as a specific signal on the RNA for subsequent processing by the exosome (Bernstein et al., 2010).
Mtr4p modulates the polyadenylation activity of Trf4p even on simple model substrates (Figs.1,,33,,4).4). These results indicate that the modulation by Mtr4p is an inherent feature of TRAMP, and not conferred by certain substrates. The regulation of Trf4p by Mtr4p represents a new type of function for an RNA helicase; the modulation of another enzyme in response to features in the RNA, here the presence or emergence of a certain number of 3’ adenosines. The modulation seen with ssRNA shows that this function of Mtr4p does not depend on duplex unwinding (Fig.3). Modulation of Trf4p by Mtr4p therefore differs from unwinding-based effects of RNA helicases on viral RNA polymerases and on RNA degradation by bacterial and mitochondrial degradosomes (Borowski et al., 2010; Carpousis et al., 2009; Piccininni et al., 2002).
Notwithstanding, RNA structure slightly influences the degree by which Mtr4p modulates Trf4p (Fig.3). This effect is probably mainly caused by the impact of RNA structure on TRAMP binding, as reflected by differences in TRAMP affinities for structured vs. unstructured RNAs (Figs.S1,S3–5). Effects of RNA structure on TRAMP binding are consistent with the impact of secondary structure on RNA binding by Mtr4p, as seen previously by others (Weir et al., 2010). Nevertheless, duplex unwinding might still contribute slightly to the observed modulation of polyadenylation by Mtr4p.
Although the modulation of Trf4p by Mtr4p does not depend on duplex unwinding, helicase activity is most likely important to stimulate exosome function on structured substrates (LaCava et al., 2005; Wang et al., 2008). RNA degradation by the exosome is thought to require ssRNA much longer than the short poly(A) tails found on TRAMP targets (Bonneau et al., 2009), and Mtr4p-catalyzed duplex unwinding may be necessary to generate sufficiently long stretches of ssRNA (Anderson and Wang, 2009).
The ability of Mtr4p to modulate Trf4p activity through a series of energetically rather minor adjustments (Fig.7H) is a notable and new paradigm for a regulatory function by an RNA helicase. It is interesting to speculate about the benefits of this mode of regulation, compared to a total shutoff of polyadenylation after a set number of adenylation steps. Numerous small adjustments might provide TRAMP with the flexibility to accommodate diverse substrates (Wlotzka et al., 2011), and to coordinate polyadenylation with subsequent processing steps (Callahan and Butler, 2010).
How can we physically imagine Mtr4p to exert its modulating effects in the context of TRAMP? Guided by a recent crystal structure of Mtr4p, which suggests contacts between protein and nucleobases via residues located outside the helicase core (Weir et al., 2010), we show that at least one of these residues, E947, is important for the modulation. Mutation of E947 markedly diminishes modulation by Mtr4p (Fig.6). This convergence of structural and biochemical data suggests that Mtr4p directly reads out the sequence at the 3' terminus in the context of TRAMP, most likely by binding the 3' terminus in a conformation similar or identical to that seen in the crystal structure (Weir et al., 2010). This notion is supported by the diminished modulation seen with the Mtr4-20p mutation (Fig.4C–E). This mutation impairs the coupling between ATP and RNA binding, and thus interferes with ATP-dependent contacts to the RNA backbone that are established by residues in the helicase core (Weir et al., 2010).
Although Mtr4p starts to modulate Trf4p as soon as a single 3’ terminal adenosine is present, the effects increase until four to five 3’ terminal adenosines are detected (Fig.7H). This observation suggests that Mtr4p binding to four to five adenosines is needed for the restriction of the poly(A) tail length. Binding of Mtr4p to 4 to 5 adenosines is consistent with the crystal structure (Weir et al., 2010). In addition, the binding site of isolated Mtr4p was recently shown to encompass approximately 5 nucleotides (Bernstein et al., 2010). Mtr4p was also shown to bind RNAs with adenosines tighter than other sequences (Bernstein et al., 2010; Bernstein et al., 2008). In TRAMP, this increased affinity might contribute to the Mtr4p-induced slight slowing of TRAMP dissociation from terminal adenosines (Fig.7E,H).
Multiple lines of structural and biochemical evidence thus coalesce around a model where Mtr4p, in the context of TRAMP, binds to the 3' end of a substrate RNA, and thereby interrogates the sequence for 3’ terminal adenosines. Probing of the nucleobases involves the helical bundle domain, wherein E947 is located. If fewer than three to four 3' terminal adenosines are detected, Mtr4p modestly stimulates Trf4p activity (Fig.7H). Upon detection of roughly four 3’terminal adenosines Mtr4p restricts further adenylation by TRAMP. Addition of four adenosines by TRAMP involves multiple binding and dissociation events (Fig.7B), which provide repeated opportunities for Mtr4p to interrogate the 3’ terminal bases and to adjust the Trf4p activity. Which molecular events take place during this adjustment will be a central question in the further investigation of TRAMP function.
Recombinant TRAMP complex was generated by combining lysate of E.coli BL21(DE3) co-expressing Trf4p/Air2p (pETDuet-His6-AIR2-TRF4-FLAG) and lysate of E.coli BL21(DE3) pLysS expressing Mtr4p (pET15b-His6-MTR4). Clarified lysates from the two strains were combined in volumes containing a roughly five-fold molar excess of Mtr4p over Trf4p/Air2p, followed by purification with cobalt-sepharose (TALON® metal affinity resin, Clontech). Samples were further purified with FLAG M2 affinity gel (Sigma) and eluted with FLAG peptide (Sigma). The FLAG peptide was removed from the TRAMP preparation using NAP-25 columns (GE Healthcare). Equimolar ratio of all components in the final TRAMP preparation was verified by size exclusion chromatography and sucrose density gradient centrifugation. Integrity of the components was determined by SDS-PAGE (Fig.S1A,B). TRAMP concentrations were measured by Coomassie staining using BSA as standard. Aliquots were frozen in liquid nitrogen and stored at −80°C. TRAMPMtr4-20p, TRAMPE947A and Trf4p/Air2p were purified and stored using identical procedures. For construction of expression plasmids see supplementary materials.
S. cerevisiae tRNAiMet was transcribed in vitro using T7 RNA polymerase (Senger et al., 1992) and purified with denaturing PAGE. All other RNAs were synthetic oligonucleotides purchased from Dharmacon. For sequences see Supplementary Materials. All RNAs were 5'-radiolabeled with P32 using T4 polynucleotide kinase (PNK, NEB) and purified with denaturing PAGE. Duplex substrates were generated and purified as described (Yang and Jankowsky, 2005).
Polyadenylation reactions were performed at 30°C in a temperature-controlled heating block in a buffer containing 40 mM MOPS pH 6.5, 100 mM NaCl, 0.5 mM MgCl2, 5% glycerol, 0.01% Nonidet P-40, 2 mM DTT, 0.7 U/µl Protector RNase Inhibitor (Roche). Prior to the reaction, radiolabeled RNA (0.5 nM final concentration) was incubated for 5 min with the indicated concentration of TRAMP. Reactions were started by addition of equimolar ATP and MgCl2 at the concentrations indicated. The order of addition of TRAMP, RNA and ATP-MgCl2 did not affect the observed modulation of the polyadenylation activity (Fig.S1C–E). Aliquots were removed at times indicated, and the reaction was stopped by addition of an equal volume of 80% formamide and dye markers. Samples were applied to denaturing PAGE and run to single-nucleotide resolution. Gels were dried, individual bands were visualized on a Storm PhosphorImager (GE Healthcare) and quantified using the ImageQuant software (GE Healthcare).
Individual adenylation rate constants at a given ATP concentration were determined by describing the polyadenylation reaction as a series of irreversible 1st order reactions :
(A0: RNA substrate, A1…n: species with 1…n adenosines, k1…n: pseudo-first order rate constants for individual adenylation steps). Derivation of explicit kinetic descriptions of timecourses for all individual adenylated species (described in Supplementary Material) yielded:
If all steps have identical rate constants (ki = ku, i = 1…n), Eqs. (1) adopt the form published by Lucius et al. (2003). To calculate rate constants from the measured timecourses, we implemented Eqs. (1) in a computer routine using Mathematica 6 (Wolfram Research). Calculated rate constants were verified by simulating timecourses with the determined values.
Total RNA was prepared from the trm6-504 strain, and G:I tails were appended with poly(U) polymerase (NEB). Reverse transcription primer was added and first strand synthesis was performed with MMLV-RT (NEB). Obtained cDNA were amplified with Taq polymerase (Roche) and tRNAiMet-specific primers. Specific amplification of pre-tRNAiMet was verified by Sanger sequencing (data not shown). To account for the heterogeneity of pre-tRNAiMet 3' ends, the RT-PCR products were digested with MseI (NEB), which removed the tRNA moiety from the 3' G-tailed poly(A) tails of the pre-tRNAiMet (Fig.2). A dsDNA linker was then ligated to the separated poly(A) tails, and ligated products were amplified with Herculase polymerase (Stratagene). PCR products were purified and sequenced by ACGT Inc. (Wheeling, IL).
The procedure was calibrated using in vitro transcribed tRNAiMet. Several 3' terminal U were appended to the RNA using TRAMP and UTP instead of ATP. The species containing three 3' terminal U was isolated on denaturing PAGE and then polyadenylated with TRAMP. The reaction was analyzed on denaturing PAGE to visualize the poly(A) tail distribution (Fig.2B). The purified RNA was subjected to the procedure outlined above for the total RNA from the trm6-504 strain, thus allowing a direct correlation between in vitro data and the poly(A) tail length analysis by Sanger sequencing (for more detailed protocols see Supplemental Materials).
To determine the processivity for individual adenylation steps, polyadenylation reactions were performed as described above using 0.5 nM of 24 nt single-stranded substrate and 150 nM wtTRAMP or Trf4p/Air2p. However, at defined times after the reaction start (t1), 10 µM of a scavenger RNA (73 nt RNA of unrelated sequence) was added to prevent re-binding of TRAMP to the substrate (Fig.S7). Control reactions confirmed complete prevention of TRAMP re-binding (data not shown). After scavenger addition, aliquots were removed from the reaction at defined times. Samples were applied to a 15% denaturing polyacrylamide gel and run to single-nucleotide resolution as described above. Gels were dried, individual bands were visualized on a Storm PhosphorImager (GE Healthcare) and quantified using the ImageQuant software (GE Healthcare). Distributions of all polyadenylated species at a given time (t1) before and after scavenger addition (when this distribution no longer changed, t2 ~ 10 min after scavenger addition, Fig.S7) were determined, and processivity was calculated according to:
To calculate processivities from the measured distributions, the equations were implemented into a computer routine using the Mathematica 6 platform. Derivation of Eqs. (2), more detailed descriptions of the experimental steps and representative data are described in the Supplementary Material.
We thank the members of our laboratories and Dr. Mike Harris (Case Western) for helpful discussions, and Drs. Timothy Nilsen and Pat Maroney (Case Western) for advice on the determination of the poly(A) tail length of pre-tRNAiMet in vivo. This work was supported by grants from the NIH to E.J. (GM067700) and J.T.A. (GM069949). UPG is supported by a DFG postdoctoral fellowship (GU1146/1-1).
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.