The RNA Tether from the Poly(A) Signal to the Polymerase MediatesCoupling of Transcription to Cleavage and Polyadenylation

Frank Rigo, Amir Kazerouninia, Anita Nag and Harold G. MartinsonDepartment of Chemistry and Biochemistry and the Molecular Biology InstituteUniversity of California at Los AngelesLos Angeles, California 90095-1569

SummaryWe have investigated the mechanism by which transcription accelerates cleavage andpolyadenylation in vitro. Using a coupled transcription-processing system we show that rapidand efficient 3′-end processing occurs in the absence of crowding agents like polyvinyl alcohol.The continuity of the RNA from the poly(A) signal down to the polymerase is critical to thisprocessing. If this tether is cut during transcription using DNA oligonucleotides and RNase H,the efficiency of processing is drastically reduced. The polymerase is known to be an integralpart of the cleavage and polyadenylation apparatus. RNA polymerase II pull-down andimmobilized template experiments suggest that the role of the tether is to hold the poly(A) signalclose to the polymerase during the early stages of processing complex assembly until thecomplex is sufficiently mature to remain stably associated with the polymerase on its own.

Running Title: A tether couples 3′-end processing to transcription

IntroductionThe production of mRNA in the nuclei of eukaryotes is a complex multistep process that beginswith the initiation of transcription and culminates in the export of the mature message. All ofthese steps, including all stages of transcription, processing and export, are functionallyinterconnected through a web of mutually synergistic interactions (Maniatis and Reed, 2002).Here we focus on the role of transcription in facilitating the final step of processing, cleavageand polyadenylation.

Although the ability of the poly(A) signal to modulate transcription (by causing termination)has been known now for almost two decades (Whitelaw and Proudfoot, 1986), the idea thattranscription, in turn, can affect 3′-end processing is more recent (Dantonel et al., 1997;McCracken et al., 1997). A widely accepted proposal is that RNA polymerase II, through thecarboxyl-terminal repeat domain of its large subunit (CTD), gathers processing factors and

Molecular Cell, 2005, in press

2

delivers them to the emerging transcript during transcription (Proudfoot, 2004; Bentley, 2005).However, this postulated recruitment function is difficult to distinguish experimentally from theknown function of the CTD as a required participant in the poly(A) site cleavage reaction(Hirose and Manley, 1998; Ryan et al., 2002). This CTD requirement for cleavage is manifestedin the absence of transcription and is thus distinct from any transcription-related recruitmentfunction. Interestingly, there is another connection between transcription and processing that hasreceived almost no experimental attention. This is the nascent RNA that links the processingapparatus to the polymerase. This tether is known to be functionally important in prokaryotes(Nodwell and Greenblatt, 1991) but its role in eukaryotes has not been examined.

A typical core poly(A) signal in mammals consists of two recognition elements (anAAUAAA hexamer upstream and a U or GU-rich element downstream) flanking the poly(A)cleavage site (Zhao et al., 1999). Although the only chemistry required for cleavage at thepoly(A) site is hydrolysis of a single phosphodiester bond in the RNA, the apparatus that must beassembled to do this is enormously complex (Calvo and Manley, 2003; Proudfoot, 2004).Presumably this reflects regulatory functions consistent with its connection to such far-flungactivities as transcription, capping, splicing and transport (Flaherty et al., 1997; Hammell et al.,2002; Proudfoot et al., 2002; Calvo and Manley, 2003). The ultimate consequence is anapparatus so large that if bound to the CTD it would dwarf the polymerase.

It is not known how the cleavage apparatus is assembled on the poly(A) signal, but variousdata suggest that it is a stepwise process (Chao et al., 1999; Takagaki and Manley, 2000). This isconsistent with its complexity and with the lag that reportedly precedes cleavage in vitro(Ruegsegger et al., 1998). It is also unclear what special contribution transcription might maketo assembly. For example, chromatin immunoprecipitation data from yeast suggest that the CTDmay play only a limited role in factor recruitment prior to the appearance of the poly(A) signal(Kim et al., 2004). To better understand the role of transcription in facilitating 3′-end processingwe have initiated experiments to investigate this problem in vitro. We first sought conditions inwhich transcription would yield RNA that is rapidly and efficiently processed under conditionsof ongoing transcription, and then we asked questions about the mechanism. Interestingly, ourresults highlight, not the role of the CTD, but of the RNA itself in the assembly of the cleavagecomplex. We favor a model in which the polymerase on its own cannot bind the assemblingcleavage complex with sufficient stability to support maturation, but relies on the RNA to retainthe nascent apparatus in close proximity until assembly on the polymerase is complete.

3

Results3′-end processing coupled to transcription. Fig. 1A shows 3′-end processing that is fast andefficient when coupled to transcription in vitro in a nuclear extract. Following pre-initiationcomplex formation, transcription was initiated with a pulse of [α-32P] CTP, and then chased forvarious times with a high concentration of unlabeled CTP (Fig. 1A). The gel in Fig. 1A showstranscript length steadily increasing so that by 4 min the majority of polymerases have passed thepoly(A) site. Then, after a short lag, cleaved and polyadenylated transcripts appear and rapidlyaccumulate. The % polyadenylation is given below the gel and is plotted as a function of time inFig. 1E (closed squares). Note that the burst of polyadenylation is rapidly finished. Althoughthe intensity of the poly(A) RNA band increases significantly between 20 and 60 min (Fig. 1A,lanes 10 and 11), most of this increase probably reflects the overall boost to intensity in the upperpart of lane 11 that can be attributed to laggard polymerases crossing the poly(A) site after 20min. This is consistent with the corresponding decrease in intensity below the position of thepoly(A) site in lane 11 relative to lane 10. These results suggest that most polymerases remaintranscriptionally engaged for the duration of the experiment, or at least until processing occurs.Note that little or no cleaved but non-polyadenylated RNA appears on the gel, consistent with theknown tight coupling of cleavage and polyadenylation (Manley et al., 1982; Moore and Sharp,1985). The polyadenylated RNA band is broad because of heterogeneity in poly(A) tail length(Wahle, 1995).

To confirm that the RNA was polyadenylated and accurately cleaved we oligo(dT) selectedthe RNA and characterized it by RNase protection (Fig. 1B). The oligo(dT) selected RNA (lane4) gave a protected fragment, identical to that from RNA transcribed in vivo in transfected cells(lane 1), that was not there when the poly(A) signal was inactivated by mutation (lanes 3 and 5).

Fig. 1A used the early promoter and late poly(A) signal of SV40. The cytomegalovirus(CMV) promoter and bovine growth hormone (BGH) poly(A) signal gave similar results (Fig.1C and Fig. 1E, open squares). We confirmed that processing in Fig. 1C was correct by blockingpolyadenylation and measuring the size of the cleaved but non-polyadenylated RNA produced,an approach more convenient than RNase protection. To block polyadenylation we added 3′-dATP to the coupled reaction before processing began but after a large fraction of thepolymerases had crossed the poly(A) site (4 min). Fig. 1D shows that, indeed, the broad band ofcorrectly processed RNA in lane 1 gave way, after 3′-dATP treatment, to a sharper band of

4

cleaved but non-polyadenylated RNA running faster in the gel (lane 2). This band was authenticpoly(A) site cleaved RNA—a mutant BGH poly(A) signal yielded no such band (lane 3),whereas mutant transcripts cut at the poly(A) site using RNase H did (lane 4).

Fig. 1E compares transcription-polyadenylation as a function of time for three differentnuclear extract preparations and two different promoters and poly(A) signals. All display similar3′-end processing kinetics and efficiencies.

We wondered whether the rapidity of 3′-end processing seen in Fig. 1 requires on-goingtranscription. To answer this we took advantage of the lag before processing begins (see Fig.1A, lane 3). We pulsed, chased for 3.5 min and then, before processing began, added α-amanitinto stop transcription. We also added 3′-dATP, as above, to block poly(A) tail growth andhighlight the poly(A) site cleavage event per se (Niwa et al., 1990; Cooke et al., 1999). Fig. 2A,lane 1 confirms that shortly after α-amanitin and 3′-dATP were added the majority of transcriptsremained uncleaved. However, efficient poly(A) site cleavage rapidly ensued (Fig. 2A, lanes 2-4) and displayed similar kinetics to a parallel reaction in which both transcription andpolyadenylation were allowed to proceed (lanes 5 and 6). The data are plotted as squares andtriangles in Fig. 2B. The line in Fig. 2B, however, is a direct reproduction of the dashed line inFig. 1E—a previous experiment using the same extract under continuous transcriptionconditions. It can be seen that the rate of 3′-end processing is similar both in the absence andpresence of on-going transcription. Therefore, once the polymerase has crossed the poly(A)signal ongoing transcription is no longer required for rapid and efficient 3′-end processing.Moreover, poly(A) tail growth does not contribute to the speed or efficiency of poly(A) sitecleavage in our coupled system.

Is there any role for transcription in this system, beyond merely producing the RNA that is tobe processed? For example, perhaps processing in these extracts is efficient simply because theextracts are unusually effective at processing per se, or because they produce RNA having somespecial property that facilitates processing. To address this we made 32P labeled RNA undercoupling conditions, used gel extraction to purify RNA of sufficient length to contain thepoly(A) site, and then added this back to a coupled reaction in which transcription had beeninitiated with a cold pulse (Fig. 2C). We also added 3′-dATP to facilitate detection of anycleaved RNA (i.e. so a sharp band low in the gel would be produced rather than a broadpolyadenylated band overlapping the unprocessed precursor). The results do not show anypoly(A) site cleavage of the pre-made RNA over at least 30 min (Fig. 2C). In contrast, when the

5

larger amount of RNA synthesized in situ in an identical parallel sample was labeled, it could beseen to undergo fast and efficient cleavage at the poly(A) site (Fig. 2D). Thus coupled, but notuncoupled, processing under these conditions is fast and efficient.

To show that the RNA in Fig. 2C had not been damaged by the purification procedure, wesubjected some of the same RNA to standard uncoupled processing conditions in the presence ofpolyvinyl alcohol (PVA). Fig. 2E shows that this gel-purified RNA is fully capable of poly(A)site cleavage when conditions that drive uncoupled processing are used. Although we haveincluded PVA in some of our coupled processing reactions (e.g. Fig. 1), we now realize that,while it does improve efficiencies, it is not a required ingredient for coupling. Thus, coupledprocessing proceeds quickly in the absence of PVA (Fig. 2D) whereas uncoupled processingrequires PVA (compare Figs. 2C and 2E). Taken together these results show that some propertyof the ternary transcription complex itself or of the associated RNA (as opposed to on-goingtranscription) allows rapid processing to occur even in the absence of crowding agents such asPVA.The RNA tether from the poly(A) signal to the polymerase mediates coupling. To evaluatethe role of the transcription complex in coupling we decided to focus on its defining feature—thenascent RNA. Specifically, we severed the RNA tether between the poly(A) signal and thepolymerase to see if this would impair coupling. To sever the tether we added to thetranscription mixture short DNA oligonucleotides complementary to sequences downstream ofthe poly(A) signal (Fig. 3A). Hybrid formation by these oligos with their RNA targets then ledto cutting by the RNase H endogenous to the extract.

Fig. 3B shows the outline and the results of such an experiment. Five different oligos wereused. Three targeted cutting to positions 77, 158 and 397 nt downstream of the SV40 latepoly(A) site (see maps beside the gels in Fig. 3B). The other two oligos (one for each panel ofFig. 3B) were controls, not complementary to any part of the RNA. All oligos were added withthe chase (30 s into the reaction) before any polymerases had reached the poly(A) signal (see Fig.1A, lane 1). Thus RNase H cutting began as soon as the oligo target was extruded from thepolymerase. After 3.5 min of chase, during the lag before cleavage begins (Fig. 1A, lane 3), α-amanitin and 3′-dATP were added to facilitate a quantitative evaluation of the results (bypreventing new transcripts from entering the processing pool during the time course, and byblocking polyadenylation so that cleaved RNA appears as a distinct band not overlapping theRNase H-cut RNA higher in the gel).

6

Lanes 4-6 and 13-15 in Fig. 3B show that poly(A) site cleavage remained fast and efficient inthe presence of the control oligos. In contrast the 77 oligo almost completely eliminated poly(A)site cleavage at the 10 min time point (lane 2), and even after an hour (lane 3) only very littlepoly(A) site cleavage was observed. Targeting the oligos farther and farther downstreamallowed a progressive rescue of poly(A) site cleavage (lanes 7-12).

We wanted to verify that the RNase H-cut RNA was not intrinsically incapable of beingcleaved at the poly(A) site (unlikely because RNase H cutting occurred in sequences from thecloning vector). Therefore, we gel extracted the RNase H-cut RNA from bands like those inlanes 1 and 10 of Fig. 3B and subjected it to standard uncoupled processing in the presence ofPVA (Fig. 3B had no PVA). Fig. 3C shows that both 77-cut and 397-cut RNA were processedefficiently in PVA. Moreover, the efficiency of this uncoupled processing was comparable forboth RNAs, showing that cutting farther downstream rescues coupled processing (Fig. 3B) forsome reason other than the length increase per se of the cut RNA.

The SV40 late poly(A) signal is strong (Carswell and Alwine, 1989), having enhancerelements both upstream and downstream of the hexamer and G/U-rich core elements (Lutz andAlwine, 1994; Bagga et al., 1995). To assess the tether requirement for a weak poly(A) signal,composed of core elements only, we carried out the experiment of Fig. 3D. Lanes 1 and 2 showthat the coupled in vitro system can support processing of this weak poly(A) signal. Lane 5shows that when the tether was cut 129 nt downstream of the poly(A) site, little poly(A) sitecleaved RNA was produced.

The results of Fig. 3 suggest a model in which a tether is required to hold the poly(A) signalclose to the CTD during the early stages of cleavage complex assembly until this complex issufficiently mature to remain stably associated with the CTD on its own (Fig. 3A). We can alsoenvision a “structural” model that invokes a 3′-end processing complex that is large whencoupled to transcription, and that includes hundreds of nucleotides of downstream RNA.According to this model, the complex needs to be large to support functions related to coupling,and cutting interferes with its assembly. Coupled processing in the structure model would berescued upon cutting farther downstream (Fig. 3B) because assembling this large structure (onthe polymerase, in the simplest version of the model) would get easier as the 3′ extension on theRNA gets longer (not true, recall, for uncoupled processing of these RNAs, Fig. 3C). Thiscontrasts with the tether model for which any RNase H-cut RNA, long or short, gets lost and cannever be processed efficiently if it is cut before becoming stably associated with the polymerase.

7

For the tether model, targeting the RNase H farther downstream rescues processing because alarger number of transcription complexes are able to assemble a mature cleavage complex beforethe RNase H target is extruded.

We can distinguish between these two models by determining how efficiently the RNase H-cut RNA gets processed. In Fig. 3B the amount of RNase H-cut RNA decreases slowly withtime once cutting is complete (e.g. lanes 2-3, 8-9 and 11-12). This decrease could reflect non-specific degradation, uncoupled processing, and/or coupled processing. Non-specificdegradation can be accounted for by reference to RNase H-cut RNA having a mutant poly(A)signal. The structure and the tether models can then be distinguished by comparison of RNAscut by RNase H at the 397 and the 77 oligo positions. Both of these RNAs are similarlyaccessible to uncoupled processing (Fig. 3C), but the structure model predicts that the longer397-cut RNA will more easily assemble the large apparatus required for coupled processing andwill, therefore, get processed and decrease in amount more rapidly than the 77-cut RNA. Ananalysis of the data in Fig. 3B (after normalizing to parallel data for mutant RNAs) indicates,however, that this is not the case, disfavoring the structure model.

To test the structure model explicitly, we carried out additional experiments (Fig. 4). In Fig.3B the oligos had been added early, before the transcribing polymerases reached the poly(A) site,for maximum effect. However, for Fig. 4, to restrict our attention to structural issues, the oligoswere not added until transcription was stopped with α-amanitin, just before the start ofprocessing. We then allowed 5 min for the RNase H to cut, and finally took time points to askwhether 397-cut RNA is processed more rapidly than 77-cut RNA as required by the structuremodel.

Fig. 4A shows the results from such an experiment. Lanes 1-3, 7-9 and 13-15 are exactly thesame as Fig. 3B except, importantly, as indicated by the time line. Fig. 4A confirms both theinhibition of processing by RNase H cutting (lanes 1-3 and 7-9 vs. 13-15) and rescue from thisinhibition by cutting farther downstream (lanes 1-3 vs. 7-9). However, these effects are mutedrelative to Fig. 3B because the oligos were added later, the overall time window was only half asbig, and the first time point was taken at a later time, after a significant proportion of whateverprocessing would occur had already taken place. The effects of oligo cutting on processing aresummarized in the upper panel of Fig. 4B where we plot the amount of poly(A) cleaved RNAproduced after completion of RNase H cutting (i.e. after 5 min of oligo) over both short and longtime intervals. The data show that, like coupled processing itself, the effects of cutting the tether

8

are not dependent upon on-going transcription at the time of cutting.Recall, the purpose of the experiments in Fig. 4 was to test the prediction of the structure

model that the 397-cut RNA will get processed more efficiently than the 77-cut RNA. Thedifferences in RNase H-cut band intensities in Fig. 4A are not sufficient to see by eye, but can berevealed by quantitation. The heavy bars in Fig. 4B (lower panel) represent the range of valuesobtained for the decrease in these RNase H-cut RNAs over the two different time intervals (afternormalizing to the mutant RNAs in lanes 4-6 and 10-12). It is clear from the overlapping datasets that there is no significant difference in the rates at which these RNAs disappear. Inparticular, there is no support for the requirement in the structure model that the 397-cut RNAshould disappear faster than the 77-cut RNA.

Since the “rescue” that occurs upon moving the cutting downstream to the 397 position (Fig.4B, upper panel) cannot be accounted for by superior processing of the 397-cut RNA, thisprocessing must arise from the ternary complexes stalled by α-amanitin at positions precedingthe 397 location on the template. Indeed, material can be seen to be disappearing from thisregion of the gel in lanes 10-12 of Fig. 3B (these differences are difficult to see in Fig. 4A for thereasons noted above). Thus, rescue presumably arises from ternary complexes in which the 397target has not yet been exposed and in which the poly(A) signal remains tethered to thepolymerase.RNA binds more tightly to the ternary complex after assembling a cleavage complex. Thetether model is based on the idea that the poly(A) signal does not become firmly associated withthe polymerase until late in the processing complex assembly pathway. To evaluate this idea weexamined the relative tendencies of the various RNA species in Fig. 3B to remain associatedwith the ternary complex in pull-down experiments. The bands in Fig. 3B are of twotypes—those arising from cleavage at the poly(A) site, and those resulting from cutting byRNase H. The poly(A) site cleaved RNA, of course, is representative of RNA that wassuccessful in assembling a cleavage apparatus on its poly(A) signal. In contrast, the longerRNase H-cut RNAs, which got cut before cleavage could occur at the poly(A) site, arerepresentative of RNAs that were not successful in assembling a functional cleavage apparatus.We asked to what extent these two classes of RNA are pulled down with the ternary complex.

We began by using immobilized templates (Fig. 5B). To facilitate quantitation we includedan internal standard in our reactions in the form of an oligonucleotide targeted to a region 5′ ofthe poly(A) signal (see Fig. 5A). The concentration of this oligo was chosen so that only a small

9

proportion of the transcripts would be cut by RNase H at this up-stream location. Theexperiments then consisted of determining how the two classes of transcript described above [i.e.cleaved at the poly(A) site or cut within the tether by RNase H] partition between pellet andsupernatant relative to partitioning of the 5′ fragment. Transcription on the immobilizedtemplates was initiated with a pulse of [α-32P] CTP in the usual way (Fig. 5B) except that thepulse was lengthened to give more counts in the transcripts and the oligos were added during thepulse. Then α-amanitin and 3′-dATP were added during the processing phase of the reaction, asbefore, to facilitate quantitation. Finally, magnetic selection was used to separate the templatesand their associated ternary complexes from any RNA that was released.

An inconvenience of these experiments is the presence in HeLa nuclear extracts of TTF2, anATP-dependent, poly(A)-independent transcript release factor (Jiang et al., 2004). We haveconfirmed, using activation by ATP but not by AMPPNP (Xie and Price, 1997) that our extractscontain such an activity (data not shown). The properties of this protein have been likened tothose of E. Coli Mfd (Hara et al., 1999; Jiang et al., 2004) which preferentially attacks stalledpolymerases (Park et al., 2002). Unfortunately, our need to block poly(A) tail growth [so as toresolve poly(A) cleaved transcripts from those cut by RNase H] requires the use of 3′-dATP—which stalls transcription (regardless of whether α-amanitin is also present). Althoughwe have shown that this stalling does not interfere with coupling (Figs. 2A and B), many of thestalled transcription complexes get released from the immobilized templates before they can beisolated for analysis. Moreover, the problem is exacerbated by the fact that magnetic beadsreduce processing efficiency (Yonaha and Proudfoot, 2000) which necessitates longer incubationtimes. Consequently in Fig. 5B most transcripts are actually released (see lanes 1 and 3 of thegel in Fig. 5B) so that the supernatant reflects primarily the overall composition of the samplerather than the composition of a preferentially released fraction. Fortunately, however, thesignificant observations in this experiment come from the pellets, which still contain sufficientmaterial to quantitate.

The gel in Fig. 5B shows that the pellet of a transcription-processing reaction is substantiallyenriched for poly(A) site cleaved RNA (compare lane 1 with lane 2a). The pellet has twice asmuch poly(A) site cleaved as 5′-cut RNA (line graph 2) whereas the supernatant has only onethird as much (line graph 1), for an enrichment of 6 fold (1.91/0.315) in this experiment (4 foldon average). Note that these transcripts are cleaved but, because of the 3′-dATP, they are notpolyadenylated. Therefore, their preferential association with the immobilized templates is not

10

the result of poly(A) tail binding proteins.If the preferential association of poly(A) site cleaved RNA with ternary complexes is

mediated by the cleavage apparatus, then RNAs that are unable to assemble this apparatus shouldnot preferentially associate with the ternary complex, even if they contain a poly(A) signal. Totest this we ran a parallel sample to the above in which we replaced the non-complementaryoligo with the 77 oligo. Importantly, Fig. 5B, lanes 1 and 3 show that cutting the tether using the77 oligo uncouples transcription and processing for immobilized templates just as for circularplasmid DNA (compare with Fig. 3B, lanes 1-6). As predicted by the tether model, the 77-cutRNA, unlike the poly(A) site cleaved RNA, is distributed similarly between supernatant (0.93relative to 5′, line graph 3) and pellet (0.81 relative to 5′, line graph 4), and is therefore notpreferentially retained on the ternary complex. Moreover, these same results show that whetherRNA contains a poly(A) signal or not, if it is not processed it behaves pretty much the same,since the ratio of RNA cut at the 77 oligo position to that cut at the 5′ position is not muchdifferent between supernatant and pellet (0.93 and 0.81). On average, RNA cleaved byprocessing is enriched 5 fold (4.4/0.86) in the pellet compared to the similar, poly(A) signal-containing RNA cut 77 nt downstream. Thus, it is apparent that cutting the tether with RNase Hinterferes with a process that causes the poly(A) signal to become associated with thepolymerase.

To explore this further using a different method we carried out a polymerase pull-downexperiment. Transcription was carried out in the presence of the 397 oligo as shown in the timeline of Fig. 5C (essentially as for lane 11 of Fig. 3B). Then RNA polymerase II was pulled downusing an antibody to the N-terminus of its large subunit. Recall that the 397 oligo allows somerescued processing to occur but that the 397-cut RNA itself is not efficiently processed (Fig. 4).Lanes 1 and 2 of Fig. 5C show that most of the 397-cut RNA was left in the supernatant butabout half of the poly(A) cleaved RNA appeared in the pellet. Almost no RNA appeared in thepellet if an irrelevant antibody or naked beads were used (Fig. 5C, lanes 4 and 6). In each of fiveindependent repeats of this experiment the poly(A) site cleaved RNA was enriched in the pelletafter pulling down the polymerase—on average 2.6 fold.

It is interesting that processed transcripts remain associated with the polymerase. This couldreflect the persistence of the entire processing apparatus on the polymerase even after cleavage,or it could reflect the action of a sub-set of factors that bind the cleaved 5′ fragment. To evaluatethese alternatives we focused on CstF which binds the GU-rich element of the poly(A) signal

11

downstream of the cleavage site. Since CstF would not be expected to remain associated withthe processed transcript unless much of the apparatus remains intact we pulled down its RNA-binding subunit (CstF 64) to see if the cleaved transcripts come along. A 5′ oligo was alsoincluded for normalization as in Fig. 5B. Fig. 5D, lanes 2 and 4 show that, indeed, a substantialamount of poly(A) site cleaved transcript, polyadenylated or not, was pulled down by CstF. Incontrast, transcripts cut by RNase H at the 5′ oligo position were not significantly pulled down(lanes 2 and 4), and an irrelevant antibody pulled down nothing at all (lane 6). Apparently, thecleavage apparatus remains at least partially intact even after half of the RNA sequenceresponsible for its initial assembly has been removed.

DiscussionWe have described an in vitro system in which fast, efficient and accurate cleavage andpolyadenylation is coupled to transcription. In addition we have shown, for two differentpoly(A) signals (Figs. 3B and 3D), that an intact tether of nascent RNA from the poly(A) signalto the polymerase (Fig. 3A) is required for this coupled processing. We also observed this tetherrequirement for 3′-end processing when we used a transcription unit that exhibits active splicing(data not shown). We suggest that the immature cleavage apparatus is unable to cling securely tothe polymerase on its own and requires a tether to hold the poly(A) signal close to thepolymerase until a mature, and stable, processing complex has formed (Figs. 4 and 5). A similar

idea has been suggested on the basis of experiments performed in vivo in which a ribozyme

rather than RNase H was used to cut the tether (David Bentley, personal communication). Thesimplest model for cleavage apparatus assembly thus appears to be that the poly(A) signal isextruded from the polymerase and then collaborates with the CTD to recruit factors (Kim et al.,2004) and to assemble a complex that does not bind strongly to either the RNA or the CTDalone. Of course, some factors may be recruited to the polymerase in advance of the appearanceof the poly(A) signal (Calvo and Manley, 2003).

But why have the decisive stages of assembly been designed to be so fragile? An attractivepossibility is that this is a manifestation of the previously proposed check-point activity of thepoly(A) signal (Orozco et al., 2002). Perhaps a tenuous assembly scheme allows the nascentprocessing apparatus to sample multiple inputs before committing to a final course of action.

Though the assembly process is initially tentative, once mature, the association of theapparatus with the polymerase apparently survives even cleavage at the poly(A) site (Fig. 5B and

12

C). Adamson et al. (2005) have made a similar observation. Both Adamson et al. and we lookedat cleaved transcripts for which polyadenylation had been blocked. However, oncepolyadenylation occurs, the transcripts appear to be released (Yonaha and Proudfoot, 2000),consistent with earlier results on uncoupled processing (Zarkower and Wickens, 1987).Interestingly, these presumptively released polyadenylated transcripts appear to retain theirassociation with CstF (Fig. 5D, lane 4) consistent with a function for CstF beyond the cleavagereaction itself (Moreira et al., 1998).

We must emphasize that the tether is only part of the coupling story. There are severalmethods for preparing nuclear extract and a variety of conditions that can be used fortranscription. But although the tether exists in all of these, many of the conditions fail to giverapid and efficient 3′-end processing, concurrent with transcription. Thus, the tether is requiredfor coupling, but it is not sufficient.

We began this study by optimizing for rate and efficiency of processing concurrent withtranscription. Mindful that under the right conditions processing can be fast and efficient in vitroeven when not coupled (Zarkower and Wickens, 1987) we sought functional connections beyonda mere precursor-product relationship between the processing and the transcription (functionalcoupling). Moreover, we wanted a criterion that points uniquely to the coupled state. Forexample, both the cap and the CTD are required even for efficient uncoupled processing(Flaherty et al., 1997; Hirose and Manley, 1998; Ryan et al., 2002) so a requirement for thesecannot be used as an indicator of coupling in vitro. The tether, however, is a unique signature ofthe coupled state, so we directed our initial attention to the tether, and cut it using RNase H to seeif this would disrupt coupling.

Disrupting a functional connection to investigate functional coupling requires some cautionso that the only difference is the disrupted functional connection itself. Thus, the failure of RNAto be processed efficiently after removal from the ternary complex by RNase H (Figs. 3B and 4)provides strong evidence of coupling because it is not likely that the state of the RNA has beenaltered beyond its removal from the ternary complex. Yonaha and Proudfoot (2000) have madea comparable observation using immobilized templates. In contrast, reduced efficiency ofprocessing after removal of the RNA from the ternary complex by phenolextraction—occasionally applied as a criterion for functional coupling (Adamson et al.,2005)—is not a sufficient criterion because free RNA is unlikely to fold and associate withproteins in the same way as RNA extruded from a polymerase. Very likely, newly extruded

13

RNA is packaged in a way that makes it a preferred substrate for the next step of mRNAproduction, but this is really coupling in the broader precursor-product sense. For this reasonstudies directed at functional coupling often compare the processing of RNA produced by RNApolymerase II with the processing of RNA produced by T7 RNA polymerase under the sameconditions (Mifflin and Kellems, 1991; Ahuja et al., 2001). However, even here caution isrequired because, as discussed earlier, mere involvement of the polymerase as a participant in theprocessing reaction is not synonymous with functionally coupling processing to transcription.

We have been careful to show that processing proceeds concurrently with transcription in oursystem (Fig. 1A). Although for technical clarity we sometimes stopped transcription prior toprocessing (e.g. Fig. 3B), the overall conditions were not otherwise changed, and care was takento demonstrate that this did not have a significant effect on processing (Fig. 2A and B). In onerecent study on functional coupling, transcription and processing were carried out undermarkedly different conditions (Adamson et al., 2005). Apparently it was necessary to stoptranscription with EDTA and then add PVA in order to obtain robust poly(A) site cleavage.However, at least for the SV40 late poly(A) site, we have found that PVA promotes rapid andefficient processing in reactions that have been uncoupled by RNase H cutting (data not shown).This is expected given the known ability of PVA to drive uncoupled poly(A) site cleavage(Zarkower and Wickens, 1987; McLauchlan et al., 1988). Indeed, a hallmark of the coupledprocessing we report here is that its rate and efficiency are not dramatically different in thepresence or absence of PVA (e.g. compare Fig. 3B, lanes 14-15 with Fig. 1E, diamonds). Insummary, after conditions have been changed, caution must be exercised in concluding that whathappens next is functionally coupled to what happened before.

Although the present study has a number of features in common with previous studies on thecoupling of 3′-end processing to transcription, it is difficult to make direct comparisons. Forexample, like us, Mifflin and Kellems (1991) observed processing that was fast and efficientwhen coupled to transcription, and Yonaha and Proudfoot (1999; 2000) and Ahuja et al. (2001)observed functional interactions between processing and transcription. Nevertheless, it is notpossible to say if a tether requirement would have been evident in those studies because they allemployed crowding agents like PVA to enhance the processing. In some cases (but not always,e.g. Fig. 3D), crowding agents can mask the effects of uncoupling by accelerating the uncoupledreaction to levels that can even exceed those of coupled processing. Therefore, the rate andefficiency of processing per se are not necessarily reliable indicators of coupling, at least when

14

PVA is involved. Instead, evidence of functional interactions must be sought. We think it likelythat PVA mimics an activity in nuclear extract that accelerates processing—but that this activityonly acts on processing that is coupled to transcription (i.e. with the tether still intact). Incontrast, PVA appears to promote rapid processing indiscriminately.

Experimental ProceduresPlasmids. In pSV40E/P the transcription unit and several kb of surrounding sequence areidentical to pAP〈117cat〉 (Tran et al., 2001). In pSV40E/L a SmaI-BamHI fragment containing L(Tran et al., 2001) replaces the HpaI-BamHI fragment in pSV40E/P that contains P. In themutated P and L poly(A) signals, the AATAAA hexamers have been changed to AgTAct andAAgtAc respectively.

For pCMV/BGH a 1778 bp PCR fragment was made from pcDNA3 (Invitrogen) using oligos#1 and #2 as primers. Two cloning steps were then carried out, essentially to replace theEcoO109I-SalI segment of pAP〈117cat〉 containing the SV40 early promoter with the 5′ portionof this fragment (up to the XhoI site) containing the CMV promoter, and to replace the SalI-BamHI segment of pAP〈117cat〉 containing the P poly(A) signal with the remainder of the PCRfragment containing the BGH poly(A) signal. For the mutant BGH poly(A) signal the AATAAAhexamer was changed to AgTAct.DNA oligomers (5′→3′).1) CGGATCGGGAGATCTCCCGATCCCCTATGG2) GGACTTTCCACACCCTAACTGACACACATTCC3) CTCAGACAATGCGATG4) 77 oligo: GTAGGGAGTATTGGG5) 158 oligo: TGGGAGTGGAATGAG6) 397 oligo: CGGAATTCCGGATGAGCATTCATCAGGCGGGC7) CTCATTCCACTCCCACCCGGGCAAGCTTTTCAGGAGCTAAGG8) CAACTAGAATGCAGTG9) -147 oligo: CGAGGTCGACAGTGGTACTCGTGGGCCAGC10) -181 oligo: CCATCTTCTGCCAGG11) AAACAAATAGGGGTTCCGCGCACATTTCCC12) GGTATCGATAAGCTGATCTCATGCACCATTCGCoupled processing assay. HeLa nuclear extract was prepared as described (Tran et al., 2001).

15

This is a very crude extract. For the success of our previously reported elongation assay("signaling", Tran et al., 2001) we found it essential to conduct the final centrifugation at a muchlower speed than called for in the earlier protocol that we followed (Flaherty et al., 1997). Wecontinued that practice for these studies although we have not determined whether it is asimportant for coupled processing as it is for signaling. We did find, however, that for efficientcoupled processing it was particularly important to achieve complete cell lysis and to remove allmaterial above the nuclear pellet after centrifugation.

A typical pulse-chase assay began with 3 µl of nuclear extract that was mixed with anti-RNase (Ambion), DTT, MgCl2, sodium citrate, DNA and water up to 5.9 µl. Amounts ofmagnesium, citrate and nuclear extract were individually optimized for each extract preparation.The mixture was pre-incubated at 30°C for 30 min and then pulsed with 3 µl containing 20 µCiof [α-32P] CTP, nucleotide triphosphates and creatine phosphate. Then 3.6 µl of chase mix wasadded containing a high concentration of non-radiolabeled CTP. Final concentrations in astandard pulse-chase assay (unless otherwise noted) were as follows: 4.8% glycerol, 4.8 mMHEPES (pH 7.9), 24 mM KCl, 48 µM EDTA, 2.1 mM DTT, 24 µM PMSF, 10 U anti-RNase, 4mM MgCl2, 3 mM sodium citrate (pH 6.7), 0.3 µg DNA, 200 µM each of ATP, UTP and GTP,20 mM creatine phosphate and 2 mM CTP. PVA or DNA oligonucleotides, when used, wereusually added with the chase. When α-amanitin and 3′-dATP were used, they were added in a 1µl volume. The final concentrations of these additions, if used, were: 2.1% PVA, 8 ng/µl DNAoligo, 37 ng/µl α-amanitin and 400 µM 3′-dATP.

In vitro transcription was terminated by the addition of a “stop solution”: 65 µl of 10 mMTrisHCl, 10 mM EDTA, 0.5% SDS, and 100 µg proteinase K (Ambion). After 30 min at 30°Cthe RNA was extracted with 350 µl TRIzol (Invitrogen), 70 µl chloroform, then precipitated with4 µl of 5 mg/ml glycogen (Ambion) and 350 µl isopropanol (30 min, room temperature), andfinally run on a 5% polyacrylamide gel. Following electrophoresis, results were recorded andanalyzed using a PhosphorImager with ImageQuant software (Molecular Dynamics). NoPhotoshop was used.Immobilized template experiments. PCR was carried out using oligos #11 and #12 as primerson pSV40E/L DNA following the manufacturer’s protocol for ThermalACE (Invitrogen) exceptthat 2 U each of Taq and Pfu polymerase were used. Primer #11 was biotinylated. The PCRproducts were purified using agarose gels and a Qiagen Gel Extraction kit. The eluted DNA wasthen bound to Dynabeads M-280 Streptavidin (Dynal) using the manufacturer’s protocol. The

16

beads were then washed 2 times with 400 µl of Dynal Washing Buffer, followed by washingonce, rotating 10 h and washing again 4 times with 400 µl of 10 mM Tris, 1 mM EDTA, pH 8.Attachment was confirmed by digesting the bead-bound DNA with restriction enzymes.Approximately 0.5 pmol of DNA was attached per mg of bead. Magnetic selection, followingtranscription, was allowed to proceed for 2 min. The bead fraction (pellet) was then washedtwice with vortexing in 50 µl of buffer D (nuclear extract dialysis buffer), and these washes werecombined with the original supernatant to which 65 µl of the stop solution was added. The beadswere also incubated in stop solution at 30ºC for 30 min to liberate the template-associated RNA,and then the beads were removed prior to isolation of the RNA.Antibody pull-down experiments. For the polymerase pull-down, 20 µl of protein A or Gmagnetic beads (Dynal) were incubated overnight at 4°C with 20 µl of anti-RNAP II (N-20,Santa Cruz Biotechnology), anti-E1B hybridoma supernatant (2A6, Dass et al., 2001), or bufferalone. The beads were washed twice with 100 µl PBS (phosphate-buffered saline), mixed with13.5 µl of transcription mixture and then incubated for 20 min at room temperature. Then thebead-bound fraction (pellet) was magnetically selected, washed gently with 100 µl of PBS, andthe RNA in the pellet and supernatant isolated as above. The CstF pull-down was the sameexcept that 40 µl of anti-CstF 64 (a mixture of 3A7 and 6A9, Wallace et al., 1999) or anti-E1Bhybridoma supernatant were used in a 12 min pull-down, and the wash was combined with thesupernatant for analysis.

AcknowledgmentsWe thank Clint MacDonald for the 3A7, 6A9 and 2A6 antibodies, David Tsao for plasmids, andDavid Bentley for communicating unpublished results. This work was supported by NIH grantGM50863.

Figure LegendsFigure 1. Transcription-coupled 3′-end processing is fast, efficient, accurate and reproducible.(A) Circular plasmid DNA was transcribed for increasing lengths of time and the RNA wasdisplayed on a gel. The % poly(A) refers to the ratio of polyadenylated RNA to all RNAextending past the poly(A) site. This assay contained PVA and the [MgCl2] was 5 mM.(B) Wt and mt refer to a poly(A) signal with intact or mutated hexamer respectively. For lanes2-7 RNA was isolated from a 15 min, 5 fold transcription-processing reaction as for (A) that had2 µM of cold CTP in place of [α-32P] CTP. One third of the RNA was set aside as the Input and

17

the remainder was oligo(dT) selected using a Poly(A) Purist MAG kit (Ambion). The FreeRNA was taken from a first round of selection and the Bound RNA was from the second. TheInput, Free, and Bound fractions were then digested with DNase I (Roche), hybridized at 65°C,and subjected to RNase protection using RNase T1 (Chao et al., 1999). The mole % ofprocessing in lane 2 is 30% of the total. The DNA used here, pSV40E/L′, had the same promoterand poly(A) signal as in (A) but set in a different plasmid background that provided an intron toallow for expression in vivo. Lane 1 shows a control using cytoplasmic RNA, isolated fromtransfected cells as previously described (Park et al., 2004). The probe used was a run-offtranscript from a derivative of pSV40E/L′ containing an inserted T7 RNA polymerase promoter.Since only RNase T1 (specific for G residues) was used in the RNase protection a single probesufficed for both wt and mt RNAs. The intensities of lanes 6 and 7 were reduced in ImageQuantfor purposes of comparison.(C) This assay was like that in (A) but using DNA with a different promoter and poly(A) signal,and at [citrate] and [MgCl2] of 5 and 6 mM respectively.(D) Cleavage at the BGH poly(A) signal is accurate. Reactions were performed as in (C) withpCMV/BGH having either a wildtype or mutant poly(A) signal. For samples receiving 3′-dATPto block polyadenylation, α-amanitin was also added, simply because this had become part of astandard procedure (see Fig. 2). Lane 4 was as for lane 3 except that oligo #3 was added withthe chase to direct RNase H cutting to the BGH poly(A) site.(E) Transcription-polyadenylation as a function of time for different extracts, promoters andpoly(A) signals. The % polyadenylation is quantitated as for Fig. 1A. The data for extract 1 arefrom an experiment that differed from the standard assay in having a 15 min pre-incubation in avolume of 6.9 µl, a chase of 2.6 µl containing PVA, and a final [ATP] of 500 µM. PVA was alsoused in the experiment for extract 2. The data for extract 3 are from the gels in Figs. 1A and 1Cand include some time points not shown in Fig. 1A.Figure 2. Coupling requires a ternary complex but not ongoing transcription.(A) Ongoing transcription and poly(A) tail growth are not required for coupling. The assay wasdone using pSV40E/L, extract 1 and PVA.(B) Quantitative comparison of processing efficiency with and without transcription. The datapoints are from Fig. 2A and the line is from Fig. 1E.(C) Exogenous RNA added to a coupled reaction does not get processed. Pre-made 32P-labeledRNA was isolated from a 10 fold coupled processing reaction of pSV40E/L in extract 3. RNA

18

running slower than poly(A) cleaved RNA was purified from a 5% polyacrylamide gel and15,000 CPM of the RNA was added to a standard (i.e. containing no PVA) coupled processingreaction in extract 3 (along with α-amanitin to 34 ng/µl and 3′-dATP to 372 µM) that was pulsedwith cold rather than hot CTP. The resulting gel was exposed 3 days to the phosphor screen.(D) Exogenous RNA is not inhibitory. A reaction was carried out in parallel with the abovewhich differed only in that the coupled reaction to which the gel-purified RNA was added waspulsed with 32P as usual. This gel was exposed only for 8 h to the phosphor screen and the newlymade RNA accounts for over 97% of the signal.(E) To demonstrate that the gel-purified RNA is capable of under-going processing under specialconditions it was incubated at 37°C with PVA for 2 h under standard uncoupled processingconditions (e.g. Wahle and Keller, 1994). Final amounts or concentrations in 30 µl were 2.5 µgtRNA, 0.67 mM 3′-dATP, 17 mM creatine phosphate, 1.9 mM DTT, 10 U anti-RNase (Ambion),42 µM PMSF, 2.1% PVA, 8.3% glycerol, 8.3 mM Hepes (pH 7.9), 42 mM KCl, 83 µM EDTA,12.5 µl extract 3 and 6.25 µl PBS.Figure 3. Severing the RNA tether from the poly(A) signal to the polymerase disrupts coupling.(A) Cartoon of a ternary elongation complex in the process of assembling a cleavage andpolyadenylation apparatus. A DNA oligonucleotide is shown hybridized to a target in the RNAthereby directing RNase H to cut the tether.(B) Severing the tether prevents coupled 3′-end processing. The oligonucleotide names refer tothe distance from the principal poly(A) cleavage site to the predominant RNase H cutting site(Wu et al., 1999). The control oligos for lanes 4-6 and 13-15 were the 77 oligo-complement andoligo #7 respectively.(C) RNase H-cut RNA is not intrinsically resistant to processing. RNase H-cut RNA wasgenerated as in (B) using a 10 fold coupled processing reaction. The bands were gel purified andincubated under uncoupled processing conditions for 2 h with PVA. The % processing given ismole % of the total.(D) Tether requirement for a weak poly(A) signal. This assay used PVA, with [citrate] and[MgCl2] of 4 and 5 mM respectively. Lane 4 was as for lane 3 except that oligo #8 was addedwith the chase to direct RNase H cutting to the poly(A) site. The 129 oligo here is the same asthe 158 oligo in Fig. 3B, but cloning has placed the identical cut site closer to the poly(A) site inthis construct.Figure 4. The RNA tether mediates coupling.

19

Wt and mt refer to a poly(A) signal with intact or mutated hexamer respectively. Threeindependent experiments like that shown in part A were carried out (except that lanes 13-15 werelacking in one). The increases in poly(A) cleaved RNA [as a % of all RNA extending past thepoly(A) site] between 5 min and either 10 or 30 min in the presence of the 77 and 397 oligos andthe control oligo (77-oligo complement) are plotted as the standard deviations in the upper panelof part B. The decreases in RNase H-cut RNA are plotted similarly in the lower panel. In eachexperiment the % of decrease in the RNase H-cut RNA was adjusted by subtracting the amountof decrease observed for the corresponding poly(A) signal mutant RNA.Figure 5. Processed RNA remains associated with the polymerase and the processing apparatus.(A) Cartoon of an elongation complex. The template for all parts of this figure was pSV40E/L.(B) Processed RNA is preferentially retained with the template. Transcription was initiated inextract 2. At 30 s 0.005 µg of the -181 (5′) oligo and 0.05 µg of either the control oligo (oligo#7, lanes 1 and 2) or the 77 oligo (lanes 3 and 4) were added. PVA causes all cut RNA, shortand long, to remain template-associated and was therefore not used. Bound RNA (pellet, lanes 2and 4) was separated from released RNA (supernatant, lanes 1 and 3) by magnetic selection. Theaverages given are the mean ± the difference from the mean for two independent experimentsusing two different 5′ oligos (-181 or -147).(C) Processed RNA is preferentially retained by the polymerase (extract 2). The error shown isthe standard deviation.(D) Processed RNA remains preferentially associated with CstF (extract 4). This experimentwas carried out using 5 mM MgCl2, 4 mM citrate and 0.013 µg of the -181 (5′) oligo.

ReferencesAdamson, T. E., Shutt, D. C., and Price, D. H. (2005). Functional coupling of cleavage and

polyadenylation with transcription of mRNA. J Biol Chem 280, 32262-32271.

Ahuja, D., Karow, D. S., Kilpatrick, J. E., and Imperiale, M. J. (2001). RNA Polymerase II-

dependent Positional Effects on mRNA 3' End Processing in the Adenovirus Major Late

Transcription Unit. J Biol Chem 276, 41825-41831.

Bagga, P. S., Ford, L. P., Chen, F., and Wilusz, J. (1995). The G-rich auxiliary downstream

element has distinct sequence and position requirements and mediates efficient 3' end pre-

mRNA processing through a trans-acting factor. Nucleic Acids Res 23, 1625-1631.

Bentley, D. L. (2005). Rules of engagement: co-transcriptional recruitment of pre-mRNA

processing factors. Curr Opin Cell Biol 17, 251-256.

20

Calvo, O., and Manley, J. L. (2003). Strange bedfellows: polyadenylation factors at the

promoter. Genes & Dev 17, 1321-1327.

Carswell, S., and Alwine, J. C. (1989). Efficiency of utilization of the simian virus 40 late

polyadenylation site: effects of upstream sequences. Mol Cell Biol 9, 4248-4258.

Chao, L. C., Jamil, A., Kim, S. J., Huang, L., and Martinson, H. G. (1999). Assembly of the

cleavage and polyadenylation apparatus requires about 10 seconds in vivo and is faster for

strong than for weak poly(A) sites. Mol Cell Biol 19, 5588-5600.

Cooke, C., Hans, H., and Alwine, J. C. (1999). Utilization of splicing elements and

polyadenylation signal elements in the coupling of polyadenylation and last-intron removal.

Mol Cell Biol 19, 4971-4979.

Dantonel, J. C., Murthy, K. G., Manley, J. L., and Tora, L. (1997). Transcription factor TFIID

recruits factor CPSF for formation of 3' end of mRNA. Nature 389, 399-402.

Dass, B., Attaya, E. N., Michelle Wallace, A., and MacDonald, C. C. (2001). Overexpression of

the cstf-64 and cpsf-160 polyadenylation protein messenger rnas in mouse male germ cells.

Biol Reprod 64, 1722-1729.

Flaherty, S. M., Fortes, P., Izaurralde, E., Mattaj, I. W., and Gilmartin, G. M. (1997).

Participation of the nuclear cap binding complex in pre-mRNA 3' processing. Proc Natl Acad

Sci USA 94, 11893-11898.

Hammell, C. M., Gross, S., Zenklusen, D., Heath, C. V., Stutz, F., Moore, C., and Cole, C. N.

(2002). Coupling of termination, 3' processing, and mRNA export. Mol Cell Biol 22, 6441-

6457.

Hara, R., Selby, C. P., Liu, M., Price, D. H., and Sancar, A. (1999). Human transcription release

factor 2 dissociates RNA polymerases I and II stalled at a cyclobutane thymine dimer. J Biol

Chem 274, 24779-24786.

Hirose, Y., and Manley, J. L. (1998). RNA polymerase II is an essential mRNA polyadenylation

factor. Nature 395, 93-96.

Jiang, Y., Liu, M., Spencer, C. A., and Price, D. H. (2004). Involvement of transcription

termination factor 2 in mitotic repression of transcription elongation. Mol Cell 14, 375-385.

Kim, M., Ahn, S. H., Krogan, N. J., Greenblatt, J. F., and Buratowski, S. (2004). Transitions in

RNA polymerase II elongation complexes at the 3' ends of genes. EMBO J 23, 354-364.

Lutz, C. S., and Alwine, J. C. (1994). Direct interaction of the U1 snRNP-A protein with the

upstream efficiency element of the SV40 late polyadenylation signal. Genes & Dev 8, 576-586.

21

Maniatis, T., and Reed, R. (2002). An extensive network of coupling among gene expression

machines. Nature 416, 499-506.

Manley, J. L., Sharp, P. A., and Gefter, M. L. (1982). RNA synthesis in isolated nuclei:

Processing of adenovirus serotype 2 late messenger rna precursors. J Mol Biol 159, 581-599.

McCracken, S., Fong, N., Yankulov, K., Ballantyne, S., Pan, G., Greenblatt, J., Patterson, S. D.,

Wickens, M., and Bentley, D. L. (1997). The C-terminal domain of RNA polymerase II

couples mRNA processing to transcription. Nature 385, 357-361.

McLauchlan, J., Moore, C. L., Simpson, S., and Clements, J. B. (1988). Components required for

in vitro cleavage and polyadenylation of eukaryotic mRNA. Nucleic Acids Res 16, 5323-5344.

Mifflin, R. C., and Kellems, R. E. (1991). Coupled transcription-polyadenylation in a cell-free

system. J Biol Chem 266, 19593-19598.

Moore, C. L., and Sharp, P. A. (1985). Accurate cleavage and polyadenylation of exogenous

RNA substrate. Cell 41, 845-855.

Moreira, A., Takagaki, Y., Brackenridge, S., Wollerton, M., Manley, J. L., and Proudfoot, N. J.

(1998). The upstream sequence element of the C2 complement poly(A) signal activates mRNA

3' end formation by two distinct mechanisms. Genes & Dev 12, 2522-2534.

Niwa, M., Rose, S. D., and Berget, S. M. (1990). In vitro polyadenylation is stimulated by the

presence of an upstream intron. Genes & Dev 4, 1552-1559.

Nodwell, J. R., and Greenblatt, J. (1991). The nut site of bacteriophage lambda is made of RNA

and is bound by transcription antitermination factors on the surface of RNA polymerase. Genes

& Dev 5, 2141-2151.

Orozco, I. J., Kim, S. J., and Martinson, H. G. (2002). The poly(A) signal, without the assistance

of any downstream element, directs RNA polymerase II to pause in vivo and then to release

stochastically from the template. J Biol Chem 277, 42899-42911.

Park, J. S., Marr, M. T., and Roberts, J. W. (2002). E. coli Transcription repair coupling factor

(Mfd protein) rescues arrested complexes by promoting forward translocation. Cell 109, 757-

767.

Park, N. J., Tsao, D. C., and Martinson, H. G. (2004). The Two Steps of Poly(A)-Dependent

Termination, Pausing and Release, Can Be Uncoupled by Truncation of the RNA Polymerase

II CTD. Mol Cell Biol 24, 4092-4103.

Proudfoot, N. (2004). New perspectives on connecting messenger RNA 3' end formation to

transcription. Curr Opin Cell Biol 16, 272-278.

22

Proudfoot, N. J., Furger, A., and Dye, M. J. (2002). Integrating mRNA processing with

transcription. Cell 108, 501-512.

Ruegsegger, U., Blank, D., and Keller, W. (1998). Human pre-mRNA cleavage factor Im is

related to spliceosomal SR proteins and can be reconstituted in vitro from recombinant

subunits. Mol Cell 1, 243-253.

Ryan, K., Murthy, K. G., Kaneko, S., and Manley, J. L. (2002). Requirements of the RNA

polymerase II C-terminal domain for reconstituting pre-mRNA 3' cleavage. Mol Cell Biol 22,

1684-1692.

Takagaki, Y., and Manley, J. L. (2000). Complex protein interactions within the human

polyadenylation machinery identify a novel component. Mol Cell Biol 20, 1515-1525.

Tran, D. P., Kim, S. J., Park, N. J., Jew, T. M., and Martinson, H. G. (2001). Mechanism of

poly(A) signal transduction to RNA polymerase II in vitro. Mol Cell Biol 21, 7495-7508.

Wahle, E. (1995). Poly(A) tail length control is caused by termination of processive synthesis. J

Biol Chem 270, 2800-2808.

Wahle, E., and Keller, W. (1994). 3' end-processing of mRNA. In RNA processing: A practical

approach, S. J. Higgins, and B. D. Hames, eds. (New York, Oxford University Press), pp. 1-34.

Wallace, A. M., Dass, B., Ravnik, S. E., Tonk, V., Jenkins, N. A., Gilbert, D. J., Copeland, N.

G., and MacDonald, C. C. (1999). Two distinct forms of the 64,000 Mr protein of the cleavage

stimulation factor are expressed in mouse male germ cells. Proc Natl Acad Sci USA 96, 6763-

6768.

Whitelaw, E., and Proudfoot, N. (1986). α-thalassaemia caused by a poly(A) site mutation

reveals that transcriptional termination is linked to 3' end processing in the human α2 globin

gene. EMBO J 5, 2915-2922.

Wu, H., Lima, W. F., and Crooke, S. T. (1999). Properties of cloned and expressed human

RNase H1. J Biol Chem 274, 28270-28278.

Xie, Z., and Price, D. (1997). Drosophila factor 2, an RNA polymerase II transcript release

factor, has DNA-dependent ATPase activity. J Biol Chem 272, 31902-31907.

Yonaha, M., and Proudfoot, N. J. (1999). Specific transcriptional pausing activates

polyadenylation in a coupled in vitro system. Mol Cell 3, 593-600.

Yonaha, M., and Proudfoot, N. J. (2000). Transcriptional termination and coupled

polyadenylation in vitro. EMBO J 19, 3770-3777.

Zarkower, D., and Wickens, M. (1987). Specific pre-cleavage and post-cleavage complexes

23

involved in the formation of SV40 late mRNA 3' termini in vitro. EMBO J 6, 4185-4192.

Zhao, J., Hyman, L., and Moore, C. (1999). Formation of mRNA 3' ends in eukaryotes:

mechanism, regulation, and interrelationships with other steps in mRNA synthesis. Microbiol

Mol Biol Rev 63, 405-445.

wt mtwt wt mt

In vitro

wt mt

Oligo(dT) selection

Bound FreeIn v

ivo

Input

Pro

be +

RN

ase

Pro

be a

lone

Uncleaved (300 nt)Cleaved (242 nt)

Probe (350 nt)

5 10 15 20 30 60

CMVpromoter

BGH poly(A) site

No

3'-d

AT

P

Mar

ker

wt wt mt

Time (min) 1

% Poly(A) 0 0 0 1 9 16 21 26 29 36 43

SV40 earlypromoter

SV40 latepoly(A) site

Poly(A) RNA

Poly(A) site

32Ppulse

0 30 sTime

Chase

15 min transcription (extract 3)

Oligo(dT) selection

RNase protection

32Ppulse

0 30 s

Chase

4 min

(-/+) 3'-dATP

32Ppulse

0 30 sTime

Chase

~1.1 kb

~1.2 kb

20 min

Poly(A) RNA

Poly(A) site Poly(A) RNA

Poly(A) sitecleaved

2 4 6 8 10 12 14 16 20 60

+ 3'-d

AT

P

Extract 3

Extract 3

1 2 3 4 5 6 7 8 9

1 2 3 4

Promoter Poly(A)signal

Figure 1

A B

C D

E

846 bp

pCMV/BGH

pSV40E/L

934 bp

pCMV/BGH

% Poly(A) 0 21 31 40 47 54

1 2 3 4 5 6Lane # 7 8 9 10 11

Extract 3

Time (min)

E) Figure

With α-ama, 3'-dATP

4.5 9 14 24 2414(min)

Noadditions

α-ama3'-dATP

Chase32P

pulse

Coldpulse

Pre-made 32P RNA

-4 0-2Time (min)

Chase

0.5 5 10 30

Long exposurePre-made 32P RNA

0.5 5 10 30

32Ppulse

0 30 sTotaltime

Chase

4 min

α-ama3'-dATP

Poly(A) RNA

Poly(A) sitecleaved

α-ama3'-dATP

Chase

Pre-made RNA

-4 0-2Time (min)

Chase32P

pulse

α-ama3'-dATP

Poly(A) sitecleaved

Poly(A) sitecleaved

Short exposureNascent 32P RNA

1 2 3 4 5 6

32Ppulse

Control

(min)α-ama3'-dATP

(min)

A

Figure 2

B

C D E

(+/-)

Total time

2 hr

E) Figure

0.5 10 60 60100.5

77 oligo Control oligo

Oligos for RNase H cutting

3'

5'

397 oligo

Inpu

t

Inpu

t

2 hr

2 hr

Uncoupled processing

32Ppulse

-4 -3.5Time (min)

Chase +DNA oligo

0

SV40 earlypromoter

Weakpoly(A) site

No

3'-dA

TP

wt wt mt wt3'

5'

129 oligo

+ 3'-d

AT

P

129

olig

o

32Ppulse

0 30 s

Chase

4 minM

arke

r

32Ppulse

0 s 30 s

Chase +DNA oligo

4 minGel purify theRNase H-cutRNA bands.

77 oligo RNase H-cut

397 oligo RNase H-cut

20 min

α-ama3'-dATP

SV40 earlypromoter

SV40 latepoly(A) site

Poly(A) RNA(~1 kb)

Poly(A) sitePoly(A) sitecleaved

Poly(A) site cleaved

60 60 6010 10 100.5 0.5

158 oligo 397 oligo Control oligo

158

397 Oligos for RNase H cutting

Oligos for RNase H cutting

3'

5'

Poly(A) sitePoly(A) site

cleaved

0.5

27140510% Poly(A)

% Poly(A) 281922114216102

CPSF CstF

RNase H

DNA ol igo

Poly(A) signal

Tether

1 2 3 4 5

1 2 3 4 5 6

7 8 9 10 11 12 13 14 15

A

Figure 3

B

D

C

Oligo to direct RNase H cutting77 nt past poly(A) site.

RNase H-cut RNA

77 oligo

α-ama3'-dATP (min)

α-ama3'-dATP (min)

Poly(A) sitecleaved

(-/+)α-ama

3'-dATP

Lane #

Lane #

Extract 2

pSV40E/L

756 bp

Extract 3

pSV40E/P

4446% Poly(A) cleaved

E) Figure

Figure 4

12%

10%

8%

6%

4%

2%

0D

ecre

ase

inR

Nas

e H

- cut

RN

A(w

t rel

ativ

e to

mt)

32Ppulse

-5 -4 30 min

Chase

0

α-ama3'-dATP

DNA oligo

Extract 2

7 7

cont

rol o

ligo

7 7

397

397

305 10

Time points

Between5 & 10 min

cont

rol o

ligo

Between5 & 30 min

Incr

ease

inpo

ly(A

) cle

aved

RN

A- 2%

- 4%

- 6%

- 8%

7 7

7 7

397397

Time interval

30

77 oligo 397 oligo Control oligo

77

397

Oligos for RNase H cutting

Poly(A) sitecleaved RNA

10530105301053010530105

wt mt wt mt wt

RNase H-cut RNA

α-ama3'-dATP

(min)

A B

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

E) Figure

0.315S

PSPS

77 oligoControl o l igo

5' Oligo

77 oligo

S P S P S P3'

5'

α-RNAPII No antibodyα-E1B

1 2 3 4 5 6

2 3 41

397 oligo

Magneticbead

32P pulse

0

Chase

10 min

α-ama3'-dATP

38 min

Magneticselection

32Ppulse

0 30 s

Chase +397 oligo

4 min 15 min

Pull-down S

P

Immobilized template

RNAPII pull-down

Poly(A) sitecleaved

Poly(A) sitecleaved

SV40 earlypromoter

SV40 latepoly(A) site

RNase H-cut (77 oligo)

3'

α-ama3'-dATP

5' oligo

2 3 41

Poly(A) sitecleaved

5'

S P S P S P

5 6

3'

Poly(A) RNA

32Ppulse

0 2 min

Chase +DNA oligo

5 min 15 min

Pull-downS

CstF pull-down

α-ama(+/-) 3'-dATP

P

α-E1B

+ 3'-dATP No 3'-dATP

2 min

S

PDNA oligosat 30 s

Figure 5

B

C

D

Poly(A) sitecleaved

397 oligo

Poly(A) site cleaved (control oligo)

A

CPSF CstF

397 oligo

Poly(A) signal

77 oligo5' oligo

Tether

1.91P

5'

6.1 4.4 ± 1.6

Poly(A) site

0.87 0.86 ± 0.01

1

2

5' oligo

77 oligo

Poly(A) site5' oligo

PelletSupernatant

3

4

0.93S

0.81P

2a 4a

Control oligo

77 oligo

P P

This expt. Average

77 oligo5' oligo

Poly(A) site77 oligo

5.1 ± 1.9

0.29S

0.78P

2.7 2.6 ± 0.8

Poly(A) site397 oligo

PelletSupernatant

This expt. Average1

2

846 bp 34800-299

+ 3'-dATP

α-CstF 64

Longer exposureof lanes 2 & 4

5' oligo

E) Figure