Proc. Natl. Acad. Sci. USA Vol. 90, pp. 6606-6610, July 1993 Biochemistry Transcriptional arrest of yeast RNA polymerase II by rho protein in vitro (transcription termination/RNA 3' ends/RNA processing) SHWU-YUAN WU AND TERRY PLATT* Department of Biochemistry, University of Rochester Medical Center, 601 Elmwood Avenue, Rochester, NY 14642 Communicated by Michael J. Chamberlin, April 20, 1993 (receivedfor review May lS, 1992)

ABSTRACT A promoter-independent assay utilizing initiation (3). Premature termination (attenuation) in the poly(dC)-tailed DNA templates has revealed that Saccharomy- 5'-proximal regions of several genes in mammalian systems ces cerevisiae whole-cell extracts can be proficient for tran- has been examined by either promoter-directed transcription scription by the endogenous yeast RNA polymerase H as well in cell extract systems (4-10) or promoterless transcription as for correct 3'-end RNA processing. Our attempts to examine with purified pol II (11-14). These sites may be intrinsic the fate of polymerase II itself were inconclusive, because only terminators for pol II and contain distinct signals such as trace btanscription products corresponded to the expected size RNA stem-loop structures (5, 7, 14), DNA bending in T-rich of terminated RNA species. Transcription in our processing- tracts (13), or other, structureless sequences (6). proficient extract was thus insufficient to cause termination. To While these examples illustrate an important role for test our system with a known, albeit heterologous, signal, we transcription attenuation in the regulation of eukaryotic gene examined a dC-tailed template carrying the E. col rho- expression, the fate of pol II in 3' noncoding regions after it dependent termination signal £p t' in the yeast extract. Tran- passes the poly(A) site(s) and terminates at downstream scripts from this template were not susceptible to processing, sequences is obscure, both temporally and mechanistically but addition of rho protein resulted in two distinct truncated (15). In higher , mutation studies and nuclear transcripts that could not be chased by excess unlabeled run-on experiments have suggested that the RNA processing nucleotides. These RNA species thus represented stably paused events associated with mature 3'-end formation may them- or terminated polymerase II products, and their absence when selves be involved in regulating the efficiency oftranscription a mutated unresponsive tip t' template was used affirmed that termination, despite the progress of pol II for long distances they were due to the effects of rho. E. col RNA polymerase past the poly(A) site (16-18). was also In Saccharomyces cerevisiae, RNA processing events also added to a yeast extract pretreated with a-amanitin generate the mature 3' ends of mRNA transcripts (19, 20). halted by rho at these same two sites. A mutated rho protein, Unlike the situation in higher eukaryotes, considerable cir- while only partly defective with E. coli polymerase, failed to cumstantial evidence indicates that yeast pol II does not provoke arrest when transcription was carried out by RNA proceed far beyond the polyadenylylation site before it polymerase H. Thus, functional rho and its cognate site, tip t', terminates transcription (21-23). To identify the sites and the appear necessary and sufficient to elicit the production of signals specifying transcription termination for pol II, we truncated transcripts by RNA polymerase II in a yeast whole- have developed a coupled system containing the 3' RNA cell extract. The ability of rho to halt the eukaryotic enzyme processing and pol II activities. In such a system, utilizing strengthens the likelihood that a rho-like helicase may be dC-tailed templates to achieve promoter-independent initia- involved in RNA polymerase H transcription termination. tion by pol II, no substantial level oftranscription termination is detectable under conditions either allowing or blocking the Transcription termination is required to complete RNA syn- processing events. However, we show that on a template thesis and prevent transcriptional interference with down- carrying the bacterial rho-dependent termination site trp t', stream genes. It occurs when the RNA polymerase dissoci- E. coli termination factor rho can halt RNA elongation by the ates from the DNA template and the nascent RNA is re- yeast transcription complex, suggesting the possible exis- leased. Termination mechanisms are classified into two tence of rho-like termination factors in yeast and other major categories (1). occurs as RNA eukaryotic organisms. polymerase stops in direct response to signals encoded in the RNA transcript and/or the DNA template. Alternative mech- anisms require the action of some additional trans-acting MATERIALS AND METHODS factor(s) to terminate RNA elogation by the polymerase DNA Templates with Deoxycytidylylated 3' Ends. Fifty mi- molecules. Among identified bacterial termination factors, crograms of CsCl-purified DNA was linearized, ethanol- the involvement of Escherichia coli rho protein in transcrip- precipitated, and dissolved in 50 MI of 0.2 M potassium tion termination is best understood (2). cacodylate, pH 7.2/1 mM CoCl2/2 mM 2-mercaptoethanol/ In eukaryotes, the study of RNA polymerase II (pol II) 0.2 mM dCTP with 60 units ofterminal deoxynucleotidyltrans- termination is complicated by 3'-end RNA processing, which ferase (Ratliff Biochemicals, Los Alamos, NM). The mixture produces mature mRNA 3' ends via specific endonucleolytic was then incubated at 37°C for 40 min, followed by ethanol cleavage and polyadenylylation. To distinguish between ter- precipitation and wash. After digestion by the second restric- mination and RNA processing events, nuclear run-on studies tion enzyme, the DNA templates specifically poly(dC)-tailed in vivo or transcription experiments in vitro are required. at one 3' end were purified from agarose gels and dissolved in However, the former are tedious and time-consuming, and 50 ,ul of doubly distilled water. The lengths of the poly(dC)- the latter are difficult due to the complexities of pol II tails averaged 100 nt, estimated by comparisons with a DNA size marker of 123 nt (Bethesda Research Laboratories). The publication costs ofthis article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" Abbreviation: pol II, RNA polymerase II. in accordance with 18 U.S.C. §1734 solely to indicate this fact. *To whom reprint requests should be addressed. 6606 Downloaded by guest on September 27, 2021 Biochemistry: Wu and Platt Proc. Natl. Acad. Sci. USA 90 (1993) 6607 Transcription in Vitro. Yeast whole-cell extracts were RESULTS prepared as described (20), except that we used only 180 mg Transcription in Vitro by Endogenous pol II in a Processing- ofammonium sulfate per ml in the fractionation step, instead Proficient Extract. To avoid the complications of promoter of220 mg. Ten microliters oftranscription mixture contained initiation for yeast pol II transcription in vitro, we took 50-100 ng of poly(dC)-tailed DNA template, 0.5-2 ,ul of advantage of promoter-independent transcription protocols extract (20 mg ofprotein per ml), 8 mM Hepes/KOH (pH 7.0), utilizing linear DNA templates extended at one 3' end by a 50 mM potassium acetate, 5 mM magnesium acetate, 1 mM poly(dC) tail (25, 26). We prepared 3' dC-extended CYCI dithiothreitol, three nonradioactive NTPs at 500 ,uM, and S (Fig. 1A) to test transcription in yeast whole-cell ,uCi of [a-32P]CTP or -UTP (800 Ci/mmol, NEN; 1 Ci = 37 extracts, which are proficient in 3' processing of many yeast GBq) brought to the desired concentration with unlabeled transcripts, including ones corresponding to the CYCI gene nucleotide. Incubation was at 30°C for 20 min unless other- (19, 20). There was sufficient endogenous RNA polymerase wise specified. In the chase experiments using yeast CYCI as activity to support transcription (Fig. 1B, lane 1), and its templates, the reaction was carried out with 20 ,uM inhibition by a-amanitin (10 pg/ml) (lane 2) indicates that it [a-32P]CTP for 10 min and then with 500 ,uM unlabeled CTP was due to pol II. for another 10 min. For rho-dependent termination, a 5-min Since plasmid stability assays have suggested that a tran- preincubation of DNA template and RNA polymerase was scriptional block occurs within 100 nt beyond the CYCI followed by the addition ofwild-type or mutant rho (replaced poly(A) site (23), we prepared three related poly(dC)-tailed by bovine serum albumin in reaction mixtures lacking rho) CYCI templates. Each ofthese carried 470 nt of3' noncoding and nucleotide mix (10 ,uM [a-32P]UTP/2.5 mM ATP/500 ,uM sequence, with end extensions from either the coding region GTP/500 ,uM CTP). After the reactions were incubated at or vector sequences (Fig. 1A). Templates 1 and 2 were 30°C for 3 min, a chase was carried out with 500 ,uM identical in their 5' regions, but the former had an additional nonradioactive UTP for 10 min. For pretreatment with 130 nt of vector sequence at the 3' terminus. Template 2 and a-amanitin, 2 p.l ofextract and 1 ,ul ofa-amanitin solution (100 3 contained the same 3' regions, but the latter had an extra 90 5 nt (of the coding sequence) in its 5' portion. When these pg/ml) were incubated on ice for min and then used for templates were incubated in extracts under transcription rho-dependent termination as described above. Wild-type E. conditions (see Materials and Methods), we detected runoff coli was purified from cells carrying the overpro- and processed transcripts in each case with the sizes pre- ducing vector p39ASE (24). Reactions were stopped and the dicted by processing at the major in vivo polyadenylylation mixtures were phenol/chloroform-extracted before ethanol site (28) (Fig. 1B, lanes 3-6). The consistency in the lengths precipitation (20). The samples were then digested by DNase of these processed RNAs indicates that transcription prefer- I, ethanol-precipitated, washed, and finally dissolved in 10 ,ul entially started at the tailedjunctions, as previously reported of 45% formamide/125 mM EDTA/0.025% xylene cyanol/ (26). To test that the processing activity was as expected (19, 0.0251% bromophenol blue) for electrophoresis in 7 M 20), we added a pre-RNA (made by SP6 RNA polymerase) urea/5% polyacrylamide (19:1 acrylamide/N,N-methylene- corresponding to the full-length transcript of poly(dC)-tailed bisacrylamide weight ratio) gels in 0.1 M Tris base/0.1 M template 2 and found that the products generated during boric acid/2 mM EDTA. ongoing transcription (lane 4) corresponded to those obtained

B 1 2 3 4 5 6

runoff (1) _. -runoff (3) I_O runoff (2)* 3' (1)"' 3' (2)- - 3'+poly(A) (3) A 247 330 Poly(A) site -5' (3) 1 poly(A) (1,2) :' .,,W.. 823 2500nt jATG TAAInI1IZ Z 2; 505 -1-1--/e -I. Kpnl EooRV Sail EcoRI 5' (1,.2) --g

Template Runoff 5' 3' v t 4. 1 d(C)n- - -- 610 160 45ont 10 130 2 d(C)n-- v No 480 160 320 nt Mik. 10 ..:20-ft 3 d(C)n l -No 570 250 320 nt 'W. FIG. 1. pol II-specific transcription and RNA 3'-end processing in vitro. (A) Map of dC-extended CYCI templates and the predicted sizes of the runoff as well as the processed transcripts. Open box indicates the CYCI coding region. Solid and dashed lines represent the CYCI noncoding region and pBluescript SK(+) vector sequence, respectively. Numbers above the map indicate the positions of the CYCl poly(A) and restriction enzyme sites; numbers below the dashed lines are the sizes of the vector sequences. (B) Transcription reactions in the yeast extract. (Left) Typical reactions using 2 01 of extract and 100 ng of dC-tailed template 2 in the absence (lane 1) or presence (lane 2) of a-amanitin (10 ,ug/ml). (Right) dC-tailed CYCI template 1 (lane 3), template 2 (lane 4), or template 3 (lanes 5 and 6). The limiting nucleotide (100 ,uM) was CTP for lanes 1-5 and GTP for lane 6: the former gave the lowest background, with the drawback that tRNA species (intense bands near the bottom of lanes 3-5) were also labeled by the CCA-end exchange activity in the extract (27). This, however, also served as an internal control for RNA recovery, and lane 6 shows that there are no other products obscured by the strong tRNA signal. Some minor RNA bands represented nucleotide-specific transcriptional pausing (compare lanes 5 and 6); these could be chased by the addition of high concentrations of the appropriate nonradioactive nucleotides. The 5' and 3' processing and the polyadenylylation [poly(A)] of RNA cleavage products are indicated. The uncleaved runoff transcripts are also marked, and the respective dC-tailed CYC1 template is given in parentheses. Downloaded by guest on September 27, 2021 6608 Biochemistry: Wu and Platt Proc. Natl. Acad. Sci. USA 90 (1993) by cleavage and polyadenylylation of the equivalent pre- nt downstream of the poly(A) site, we must conclude that RNA (data not shown). under our conditions, good transcription in an extract profi- Will these extracts also support transcription termination, cient for processing is not sufficient to elicit transcription and if so, how would a terminated transcript be identified? If termination. processing is tightly coupled to termination, a unique short 3' Yeast pol II Transcription Is Susceptible to Arrest by E. coli fragment of 100-150 nt should remain after cleavage. Exam- rho Factor. Since the helicase mechanism of the bacterial ination of the RNA products in the transcription reaction transcription termination factor rho (29) provides a paradigm failed to reveal any short species in this range (Fig. 1B, lanes for termination in eukaryotic cells, we attempted an exper- 5 and 6) that might be derived from a transcript that had been iment using heterologous components. In the simplest ex- processed and terminated not far beyond the poly(A) site. periment, we added E. coli rho factor to a yeast extract This failure is uninformative, since distal cleavage products transcription reaction with a poly(dC)-tailed CYCI template could have considerable 3' heterogeneity and would be and failed to see any effect of the bacterial protein (data not susceptible to degradation by 5' -- 3' exonuclease digestion shown). We then tested a poly(dC)-tailed template containing due to the lack of a protective 5' cap structure. The only 104 nt of the 5' portion of trp t' bacterial termination region identifiable 3'-cleavage RNA products corresponded to those with a 14-nt leader sequence, followed by 140 nt ofadditional predicted from processing of full-length runoff RNAs. sequence derived from the pGEM-3Z vector (Fig. 3 Upper). In a time-course experiment, we observed little if any lag As a control, we used a nearly identical template, pRMlla, between transcription and RNA processing: the latter was carrying a mutated trp t' region defective in its ability to cause detectable even within the first minute, when many RNA termination (30). When rho factor was included in pol II polymerase molecules had not reached the end of the tem- transcription reactions with each ofthese two poly(dC)-tailed plate (data not shown). This experiment also revealed some templates (Fig. 3 Lower), two major transcripts (marked by minor transcripts whose sizes made them possible candidates arrowheads) of 120 and 130 nt were produced from the for terminated or paused transcripts according to in vivo data. wild-type template (lane 2) but not from the mutated template To examine them further, we performed transcriptions with (lane 4). These two rho-mediated transcripts correspond low concentrations of the labeled nucleotide (20 ,uM precisely in size to the doublet termination transcripts pre- [a-32P]CTP) for 10 min (Fig. 2) in the presence of either UTP viously seen on the RMt' template (30); congruence with (lane 3) or 5-bromo-UTP (lane 4). Incorporation of the latter rho-dependent products using E. coli polymerase will be blocks RNA processing of the substrate (unpublished data) shown below. To further confirm their identity, we annealed and was followed in this experiment by a chase with unla- a trp t'-specific deoxyoligonucleotide to the transcription beled CTP at 500 AM for an additional 10 min (lane 5). Two products and then treated with RNase H. A single digestion faint transcripts of about 220 nt still remained after the chase product of 65 nt was obtained, in good yield, consistent with (marked by arrows), as if they were unable to be further a 5' terminus at the dC-template junction (data not shown). extended by yeast pol II. Although these species are consis- RNase A treatments (10 pg/ml, 10 min at 30°C) eliminated tent with a very low-level arrest of pol II on CYCI about 60 these products, indicating that they were not in the form of RNADNA hybrids, and the absence of transcription with M 1 2 3 4 5 a-amanitin (10 p,g/ml) confirmed that the transcription was pol II-specific (data not shown). Therefore, the signals de- tected in the extract transcription were trp t' transcripts synthesized by yeast pol II molecules that initiated specifi- cally at the dC-templatejunction and were halted in response

.- 3' L tip t' pGEM3Z - poly(A) 14 104 140

5' M 1 2 3 4

.4

FIG. 2. Transcriptional chasing. Reactions using template 2 (Fig. 1) were performed as described in Materials and Methods with 0.5 FIG. 3. pol II transcription arrest by bacterial rho protein on a pl of extract added. A total transcription reaction with normal wild-type but not a mutated trp t' template. (Upper) The trp t' nucleotide substrates (lane 3) was then passed over oligo(dT)- construct used in the reactions. A 14-nt leader sequence (L) precedes cellulose to obtain bound (lane 1) and unbound (lane 2) fraction. the 104-nt trp t' region, which is followed by 140 nt derived from Substitution of 5-bromo-UTP for UTP was also tested, without (lane pGEM-3Z (Promega). (Lower) Effect of rho on pol II transcription. 4) or with (lane 5) a chase with excess unlabeled CTP. The unbound Reactions were performed with 2 j. of yeast extract under chase fraction (lane 2) reveals multiple bands in addition to the runoff and conditions as described in Materials and Methods. rho (37 nM) was processed RNAs. Similar bands in the 5-bromo-UTP reaction (lane included in lanes 2 and 4. dC-pRMt' in lanes 1 and 2 was the wild-type 4) were transiently paused transcripts as judged by their disappear- template and dC-pRMlla in lanes 3 and 4 contained unresponsive trp ance during the chase (lane 5), and arrows indicate the possible t' region (30). Arrowheads indicate the transcripts produced by rho terminated or stably paused transcripts. Lane M, size markers activity. Lane M, size markers [32P-labeled Hpa II fragments of [32P-labeled Msp I fragments ofpBR322 DNA: 622, 527,404, 309, 242 pBluescript II KS(-) (Stratagene): 242, 190, 147 (doublet), 118, and (doublet), 217, 201, 190, 180, 160, 147, 123, 110, 90, 76, 67, and 34 nt]. 110 nt]. Downloaded by guest on September 27, 2021 Biochemistry: Wu and Platt Proc. Natl. Acad. Sci. USA 90 (1993) 6609 to E. coli rho factor at sites that are used by the bacterial M 1 2 3 4 5 67M enzyme. Response to a Defective rho Factor Is More Severe with pol U than withE. coliRNA Polymerase. To provide one more test of whether the observed responses were genuine effects of _ _~~~~~~~~~~~~~~~~~~~~~~~~P rho, rather than some peculiar artifact, we compared tran- scription by both enzymes using either wild-type rho or a VW... mutant rho protein (S4) constructed by site-directed muta- genesis. The S4 protein has residues 77 and 78 both altered from aspartic acid to alanine and requires a different purifi- cation (P. Spear and T.P., unpublished work); the active sample is nevertheless free of RNase contamination, and retains ATPase activity and partial termination function as assayed under promoter-initiated transcription conditions rho

(30). The effects of wild-type and S4 rho proteins were Pretreated _ + compared in reactions with E. coli RNA polymerase (Fig. 4, extract lanes 1-3) and pol II (lanes 4-6). While E. coli RNA poly- E. coli RNAP + merase was halted by the mutant rho factor S4 at the two Yeast pol 11 4 + distal sites and with slightly decreased efficiency (lane 3), the FIG. 5. Responses of both polymerases to rho factor are nearly addition ofS4 to the pol II transcription reaction produced no identical. The components of each reaction are indicated below the such transcripts (lane 6). This experiment further confirms autoradiograph. E. coli RNA polymerase (RNAP) behavior was that the observed arrest ofpol II transcription was due to an modified in the yeast extract pretreated with a-amanitin, compared with its rho response in buffer alone (lane 5 vs. lane 2). Lanes 1 and event mediated by wild-type rho factor. 4 had no added rho, lane 3 had no added E. coli RNA polymerase. Sites of pol II Arrest Correspond to Sites of rho-Dependent The a-amanitin-pretreated extract (see Materials and Methods) was Termination with E. coli RNA Polymerase. The buffers and used in lanes 3-5. Transcription reactions performed with untreated salts in the yeast whole-cell extract differed considerably yeast extracts with (lane 6) or without (lane 7) rho proteins had no from our standard conditions for assaying rho activity. As added bacterial enzyme. Transcripts produced by E. coli RNA shown above (Fig. 4, lane 2), under extract conditions, E. coli polymerase in the presence of rho are bracketed. Arrowheads RNA polymerase was terminated by rho at about 75, 90, and represent the rho-dependent termination transcripts common to all 110 nt into the trp t' region, as well as at two distal positions rho-containing reaction mixtures. Lane M, size markers (as in Fig. about 120 and 130 nt from the 5' end; the overall efficiency 4). was also not very good. To try to resolve the differential responses ofthe E. coli and yeast polymerases (lane 2 vs. lane (lane 6) was very similar to that of the bacterial enzyme. The 5), we asked whether the extract itself(above and beyond pol longest rho-induced product seems to correspond to a long II activity) might be influencing the response; for this bac- but inefficient pause site for both polymerases (faintly evi- terial transcription reactions were carried out in yeast ex- dent in lanes 1, 4, and 7); this is not evident for the shorter tracts pretreated with a-amanitin. Template incubated in product. pretreated extract alone elicited no transcriptional activity These experiments support our hypothesis that the rho- (Fig. 5, lane 3), demonstrating the absence of all yeast RNA mediated arrest ofpol II is functionally similar to the bacterial polymerase activities. When E. coli polymerase was added, mechanism of rho-dependent termination. It also appears good transcription was observed (lane 4), and the further that some activities (other than pol II) in the yeast extract inclusion of rho shifted termination to the two distal loca- result in better rho termination and usage of the two distal tions, with a substantial improvement in overall termination sites in both systems. We surmise that some factors in the efficiency (lane 5 vs. lane 2). The pattern displayed in lane 5 yeast extract can mediate the coordination of rho and RNA is consistent with that observed in the promoter-specific polymerases and may compensate in particular for any ad- bacterial termination system (30). The pol II response to rho verse effects of the promoterless initiation on dC-extended templates. The simplest explanation for the fact that both pol 1 2 3 M 4 5 6 II and bacterial enzyme are halted at similar sites on the trp t' template is that rho factor activity is functionally inter- changeable in the two systems. Whether the response with either polymerase depends on a specific interaction between rho and the transcribing elongation complex has yet to be determined.

DISCUSSION Whole-cell extracts from S. cerevisiae possess endonucleo- lytic cleavage and polyadenylylation activities that produce the mature 3' ends ofyeast mRNA transcripts (19, 20). While FIG. 4. Effect of a mutant rho factor on the bactenal ana yeast this final step in mRNA maturation is qualitatively similar to RNA polymerases. Parallel transcription reactions were carried out that seen in higher eukaryotic organisms, it differs in two with dC-tailed trp t' template in extract buffer (for theE. coli enzyme, notable respects: the highly conserved AAUAAA metazoan lanes 1-3) and in extracts (for pol II, lanes 4-6). The samples in lanes signal is functionally absent in yeast (and no single signal has 1 and 4 had no added rho, wild-type rho was present for lanes 2 and been clearly shown to serve in its place), and yeast mRNA 5, and rho mutant S4 was present for lanes 3 and 6 (rho concentra- tions, 37 nM). TheE. coi RNA polymerase transcripts caused by rho transcripts have never been seen to extend far beyond the action are bracketed. Arrowheads indicate rho-generated transcripts polyadenylylation site. Circumstantial evidence supports the common to both systems. Lane M, size markers (32P-labeled Msp I postulate that yeast RNA polymerase II most likely termi- fragments of pBR322: 242, 217, 201, 190, 180, 160, 147, 123, 110, 90, nates transcription <200 nt beyond the poly(A) addition site 76, and 67 nt). (21-23). Downloaded by guest on September 27, 2021 6610 Biochemistry: Wu and Platt Proc. Natl. Acad Sci. USA 90 (1993) We have established an in vitro yeast pol II transcription result is that our extracts lack some additional factor(s) system utilizing linear DNA templates 3'-extended with required for efficient pol II termination. An alternative is the poly(dC). It was known that with partially purified mamma- less attractive possibility that the yeast polymerase does not lian pol II, efficient and selective initiation occurs on 3'-dC- in fact terminate efficiently in the expected regions of our extended templates, forming a very stable elongation com- templates, even in vivo. plex whose subsequent behavior can be readily studied, In conclusion, we have described an in vitro system derived particularly the response at "intrinsic" termination sites in from yeast whole-cell extracts that is proficient in both tran- several genes (11-14). We demonstrated that a partially scription and in 3' RNA-processing activities. Under our fractionated whole-cell extract previously shown to contain conditions, the 3' endonucleolytic cleavage is temporally RNA 3'-end-processing activities also contained pol II tran- inseparable from (ifnot coupled to) transcription. Although we scription activity, as measured by a-amanitin-sensitive syn- hoped that we might detect a definitive signal representing thesis from dC-extended DNA templates. Transcription ini- transcripts terminated by pol II, the only candidates are quite tiation from the tailedjunction gave rise to RNA species from faint and remain merely suggestive. Transcription in our the CYCI templates corresponding to runoff transcription processing-proficient extract is thus not sufficient to produce and to products resulting from cleavage and polyadenylyla- termination. Nevertheless, E. coli rho factor is capable of tion at the cognate poly(A) site. We observed no shorter arresting the progress of yeast pol II, dependent on the 3'-cleavage RNA products that might have been generated by presence of a cognate E. coli recognition signal and on normal pol II termination in conjunction with processing. RNA cleavage and polyadenylylation occurred rapidly in our assay rho function. This strengthens the hypothesis that a rho-like conditions, with no noticeable time lag after transcription helicase factor may well be involved in specifying the fate of (data not shown). We cannot tell whether processing remains eukaryotic pol II molecules after they have finished transcrip- independent from transcription, rather than being facilitated tion past the site(s) of polyadenylylation. by it or coupled to it. However, we can conclude from these We are grateful to Caroline Kane for advice on poly(dC)-tailing and observations that on its own, transcription in a processing- transcription and Fred Zalatan for providing pRMt' and pRMlla proficient extract is not sufficient to elicit termination to a DNAs. Special appreciation is given to Keith W. Nehrke and Peggy significant extent. We surmise that termination requires some Spear for preparing wild-type and S4 rho proteins, respectively. We additional factor(s) in addition to the core RNA polymerase also thank Scott Butler, Parag Sadhale, and the current members of our laboratory for helpful criticism and discussion. Support for this and RNA-processing activities. work was provided in part by Public Health Service Grant GM22830 By contrast, the heterologous system with E. coli rho to T.P. and an Elon Huntington Hooker graduate fellowship from the factor yielded clear results, suggesting that pol II can suc- University of Rochester to S.-Y.W. cumb to the action of an RNA-dependent ATP-driven heli- 1. Platt, T. (1986) Annu. Rev. Biochem. 55, 339-372. case activity (30) in a specific manner. The diminished 2. Platt, T. & Richardson, J. P. (1992) in Transcription Regulation, eds. termination with either the mutant trp t' template or an McKnight, S. L. & Yamamoto, K. R. (Cold Spring Harbor Lab. Press, altered rho protein demonstrate that the effect is due to the Plainview, NY), pp. 365-388. normal properties of rho and not an artifact of the heterolo- 3. Young, R. A. (1991) Annu. Rev. Biochem. 60, 689-715. 4. Bengal, E., Goldring, A. & Aloni, Y. (1989) J. Biol. Chem. 264, gous system. The initial difference between the bacterial and 18926-18932. yeast systems became reconciled by the effects of some 5. Toohey, M. G. & Jones, K. A. (1989) Genes Dev. 3, 265-282. non-pol II transcription activities existing in the yeast ex- 6. Wiest, D. K. & Hawley, D. K. (1990) Mol. Cell. Biol. 10, 5782-5795. tract. These activities cause the bacterial enzyme to use more 7. Mechti, N., Piechaczyk, M., Blanchard, J.-M., Jeanteur, P. & Lebleu, B. (1991) Mol. Cell. Biol. 11, 2832-2841. distal termination sites and terminate as efficiently as pol II 8. London, L., Keene, R. G. & Landick, R. (1991) Mol. Cell. Biol. 11, in the extract, similar to what has been observed in the 4599-4615. optimized promoter-specific E. coli transcription system 9. Innis, J. W. & Keliems, R. E. (1991) Mol. Cell. Biol. 11, 5398-5409. using similar DNA templates (30). The kinetic coupling model 10. Wiest, D. K., Wang, D. & Hawley, D. K. (1992) J. Biol. Chem. 267, 7733-7744. ofJin et al. (31), in which a faster polymerase (or slower rho) 11. Reines, D., Wells, D., Chamberlin, M. J. & Kane, C. M. (1987) J. Mol. causes reduced termination, provides a simple explanation Biol. 196, 299-312. for the shift to distal sites, which themselves would elicit 12. Kerppola, T. K. & Kane, C. M. (1988) Mol. Cell. Biol. 8, 4389-4394. strong pauses. If pol II elongation were slightly faster than 13. Kerppola, T. K. & Kane, C. M. (1990) Biochemistry 29, 269-278. 14. Bengal, E. & Aloni, Y. (1991) J. Virol. 65, 4910-4918. elongation by E. coli polymerase, the yeast enzyme could 15. Proudfoot, N. J. (1989) Trends Biochem. Sci. 14, 105-110. outstrip rho, reducing the proximal stops and halting only 16. Whitelaw, E. & Proudfoot, N. (1986) EMBO J. 5, 2915-2922. when paused most stably at the distal sites. In this light, 17. Logan, J., Falck-Pedersen, E., Darnell, J. E. & Shenk, T. (1987) Proc. slower tracking along the transcript by the S4 mutant rho Natl. Acad. Sci. USA 84, 8306-8310. 18. Connelly, S. & Manley, J. L. (1988) Genes Dev. 2, 440-452. would account for its effects, as well. 19. Butler, J. S. & Platt, T. (1988) Science 242, 1270-1274. We do not know whether rho simply acts as a juggernaut 20. Butler, J. S., Sadhale, P. P. & Platt, T. (1990) Mol. Cell. Biol. 10, catching up with the elongation complex to exert helicase- 2599-2605. mediated RNA release, independent of other effects, or 21. Snyder, M., Sapolsky, R. J. & Davis, R. W. (1988) Mol. Cell. Biol. 8, 2184-2194. whether its activity depends on additional interactions be- 22. Osborne, B. I. & Guarente, L. (1989) Proc. Natl. Acad. Sci. USA 86, tween rho and the polymerase molecule. The arrest of yeast 4097-4101. pol lI by E. coli rho factor may involve the conserved 23. Russo, P. & Sherman, F. (1989) Proc. Natl. Acad. Sci. USA 86, domains found in the (B and (3' subunits of E. coli RNA 8348-8352. 24. Nehrke, K. W., Seifried, S. E. & Platt, T. (1992) Nucleic Acids Res. 20, polymerase and the two largest subunits of eukaryotic RNA 6107. polymerase 11 (3). In this case, the inability of S4 rho to 25. Kadesch, T. R. & Chamberlin, M. J. (1982) J. Biol. Chem. 257, 5286- impede pol II transcription would suggest that the E. coli 5295. termination factor prefers the cognate RNA polymerase for 26. Dedrick, R. L. & Chamberlin, M. J. (1985) Biochemistry 24, 2245-2253. 27. Manley, J. L. (1984), in Transcription and Translation: A Practical its function. If a rho counterpart does exist in yeast, its Approach, eds. Hames, B. D. & Higgins, S. J. (IRL, Washington, DC), recognition signal must be unrelated to that in the bacterial pp. 71-88. ' system, since reactions with CYCI templates revealed no 28. Russo, P., Li, W.-Z., Hampsey, D. M., Zaret, K. S. & Sherman, F. rho-dependent effects. The positive results with rho using the (1991) EMBO J. 10, 563-571. 29. Brennan, C. A., Dombrosk, A. J. & Plait, T. (1987) Cell 48, 945-952. trp t' template suggest that any significant degree of termi- 30. Zalatan, F. & Platt, T. (1992) J. Biol. Chem. 267, 19082-19088. nation in the homologous yeast system should have been 31. Jin, D. J., Burgess, R. R., Richardson, J. P. & Gross, C. A. (1992) Proc. detectable. The most reasonable explanation for the negative Natl. Acad. Sci. USA 89, 1453-1457. Downloaded by guest on September 27, 2021