Proc. Nati. Acad. Sci. USA Vol. 78, No. 5, pp. 2913-2917, May 1981 Biochemistry

Tandem termination sites in the tryptophan operon of (DNA sequence//p factor/multiple terminators) ANNA M. Wu, GAIL E. CHRISTIE, AND TERRY PLATT Department of Molecular Biophysics and Biochemistry, Yale University, 333 Cedar Street, New Haven, Connecticut 06510 Communicated by Edward A. Adelberg, February 17, 1981

ABSTRACT In vivo, transcription of tryptophan (trp) operon stop site, with proximal endpoints located 12-73 bases past trp mRNA appears to terminate at a site (trp t) 36 nucleotides after t (8). Analysis of mRNA from these strains showed that the ob- the last structural gene, and efficient function at this site requires served distal expression is due to readthrough transcription and the protein factor p. However, distal nucleotide sequences also is not, for example, the result ofreinitiation at a new promoter. seem to play a role in modulating termination at trp t. We report These data suggested that distal sequences were required for here our in vitro studies of DNA fragments carrying portions of correct termination at trp t, although the mechanisms respon- the trp termination region. Transcription ofthese DNA fragments sible for this were unclear. We report here our analysis of the in a purified system demonstrates that RNA polymerase actually trp termination region in a well-defined system in vitro, which recognizes two different termination sites. Termination at the pre- reveals a second , trp t' (located about 250 bases past viously characterized site, trp t, is only 25% efficient, and it is existence unaffected by the presence ofp factor in vitro. However, addition trp t). The ofthis site in the region removed by dele- ofp to the transcription reaction mixture reveals that termination tions that affect termination partially explains the requirement also occurs within a region that we have designated trp t', located for sequences located beyond the point corresponding to the about 250 bases past trp t. These two sites behave independently normal 3' end of the trp operon transcript (8). in vitro, whether in the tandem configuration or cloned separately, and their structural features and functional characteristics are METHODS quite different. This contrasts with the observation that termi- nation oftranscription at the end ofthe trp operon in vivo appears Preparation of DNA. Growth of cells and phage and prep- to require a p-mediated interaction between trp t and trp t'. The aration ofDNA were as described (8). Restriction endonuclease possible involvement of other factors and the significance of mul- cleavage products were separated by electrophoresis on 7% tiple termination sites is discussed. polyacrylamide gels in 50 mM Tris-borate (pH 8.3)/1 mM EDTA, eluted electrophoretically in 40 mM Tris-acetate (pH Termination oftranscription is a complex event, the essence of 8.5), and extracted with phenol before use. The 3' inset ends which is the recognition by RNA polymerase ofsignals encoded of restriction fragments were filled in by incubation of 1 pmol in the DNA template. The response to a particular signal may of fragment in 20 A.l containing 10 mM Tris HCl (pH 7.6), 10 be modulated by termination factors such as p (1), or antiter- mM MgCl2, 200 ,AM each dATP, dTTP, dGTP, and dCTP, 10 mination factors such as A N protein (reviewed in refs. 2 and mM 2-mercaptoethanol, and 5 units ofDNA polymerase I (Kle- 3). In some instances transcription termination appears to be now fragment; Boehringer Mannheim) for 15 min at 37°C. Liga- regulated in a more intricate fashion, as in transcriptional po- tion reaction mixtures contained about 0.8 pmol of vector and larity (2) and attenuation (4). At regulatory sites such as phage 1.0 pmol ofinsert in 10 ,ul containing 60 mM Tris HCl (pH 7.6), or bacterial attenuators such complex interactions are not un- 20 mM MgCl2, 10 mM dithiothreitol, 400 ,M ATP, and 1 expected, but termination at the end of a gene or gene cluster (Weiss) unit of DNA ligase (P-L Biochemicals) and were incu- has been thought to be a relatively simple event. Analysis ofthe bated 18 hr at 4°C; the mixture was used to transform E. coli region at the end ofthe trp operon in Escherichia coli supported strain W3110trpAAEJ. Ampicillin-resistant cells were screened this idea, because termination of transcription in vivo appears by colony hybridization, using an RNA probe made by tran- to occur with high efficiency at a site called trp t located only scription ofthe desired fragment. Orientation ofthe insert was 36 nucleotides beyond the trpA coding sequence (5). A G+C- determined by restriction endonuclease digestion of purified rich region of dyad symmetry in the DNA precedes this stop plasmid DNA. site, and the transcript ends with several uridine residues- In Vitro Transcription. Wild-type RNA polymerase and the both these features are commonly found in prokaryotic termi- rpo2O3 polymerase were purified according to Burgess and Jen- nators (2, 6). Because introduction of the rho201 mutation re- drisak (9). [a-32P]GTP (10-30 Ci/mmol; 1 Ci = 3.7 X 10'° sults in readthrough transcription (termination is reduced by becquerels) was from New England Nuclear. p protein was pu- 60%), we inferred that correct termination at trp t is p-depen- rified as described below. The standard 10-,ul transcription re- dent in vivo (5). action mixture contained 20 mM Tris-acetate (pH 7.9), 0.1 mM Mutations in this region were obtained by using a selection EDTA, 0.1 mM dithiothreitol, 4 mM magnesium acetate, 150 for readthrough into distally placed lactose operon genes; these mM KCl, 200 ,AM each ATP, CTP, and UTP, 5-10 ,Ci of [a- were all deletions, ranging in size between 185 and 900 base 32P]GTP, 0.1-0.2 pmol of DNA template, 0.02-0.1 ,ug of RNA pairs (bp) (7). Surprisingly, sequence analysis ofeight deletions polymerase (0.04-0.2 pmol), and 0. 1-0.5 ,ug of p. After incu- revealed that five were located distal to the normal transcription bation for 20 min at 37°C, reaction mixtures were diluted in 100 ,ul of0.3 M sodium acetate, 1 mM EDTA, and carrier tRNA at The publication costs ofthis article were defrayed in part by page charge 0.5 mg/ml, extracted with phenol, and precipitated with payment. This article must therefore be hereby marked "advertise- ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact. Abbreviations: bp, base pair(s); nusA, nusA protein. 2913 2914 Biochemistry: Wu et al. Proc. Natl. Acad. Sci. USA 78 (1981) ethanol, and products were separated on 5% acrylamide/7 M of [a-32P]GTP was added as tracer. Transcription in the pres- urea gels. RNA was eluted either electrophoretically or by in- ence ofpwas carried out at 37°C for 20 min, the reaction mixture cubation of gel slices at 370C in 0.33 M KCI/10 mM Tris HCl was extracted with phenol twice, and the product was separated (pH 7.6)/1 mM EDTA. T1 ribonuclease digestion and two-di- from precursors by chromatography on Sephadex G-50. The mensional analysis of the fragments were performed as de- 5-100 pmol of RNA from such a transcription were incubated scribed (5). in 30 ,ul of 50 mM Hepes (pH 8.3)/10 mM MgCl2/3.3 mM di- Purification of p. p was purified by a modification (unpub- thiothreitol/10% (vol/vol) dimethyl sulfoxide/15% (vol/vol) lished) of the method of Calva and Burgess (10). Nucleic acid glycerol/5 ,tM ATP containing 30 pmol of[32P]pCp (>8000 Ci/ and associated proteins from a crude extract from E. coli W3110 mmol) and 10 units of RNA ligase (P-L Biochemicals) for 18 hr cells were precipitated with Polymin P. Ammonium sulfate was at 4°C. The samples were extracted with phenol, extracted with added to the supernatant to 50% saturation, then the resulting ether, and chromatographed on Sephadex G-50. After ethanol precipitate was collected and washed twice with buffer con- precipitation, the p-terminated transcripts were separated by taining ammonium sulfate at 35% saturation. The pellet was electrophoresis on a preparative 5% acrylamide/7 M urea gel. redissolved, applied to a Bio-Rex-70 column, and eluted with For complete alkaline hydrolysis, RNA was incubated in 0.2 a 0.1-0.5 M NaCl gradient. Further purification on poly(U)- M NaOH for 18 hr at 37°C. Partial alkaline hydrolysis was per- Sepharose and agarose 5'-(p-aminophenylphosphoryl)uridine formed by incubating RNA in 50 mM NaHCO3, pH 9.0/1 mM 2'(3') phosphate (Miles) removed additional contaminants, in- EDTA for 5-15 min at 950C. cluding ribonuclease (see ref. 10). p-containing fractions were detected by the ability to cause termination at trp t' in vitro. RESULTS The final protein product yielded a single band on a sodium dodecyl sulfate/polyacrylamide gel and was free ofribonuclease Construction of a Plasmid Carrying the trp Promoter and activity. More than 500 Ag ofp could be obtained from 25 g of Termination Region. Plasmids containing the trp termination cells. pwas active at trp t' at salt concentrations between 25 and region downstream from the trp promoter were constructed in 200 mM KC1, although total synthesis by RNA polymerase was two steps. First, the trp promoter was isolated as a 380-bp depressed above 150 mM KCl. Hinfl-Hae III restriction fragment from pLD102 (11). EcoRI- Labeling and Analysis of RNA 3' Termini. For end labeling cleaved pBR322 and purified promoter fragment were incu- the standard transcription reaction mixture was scaled up 10- bated with DNA polymerase and dNTPs to fill in the 3' ends fold, and all four triphosphates were present at 200 ,uM; 1-5 ,uCi generated by Hinfl and EcoRI, and the mixture was ligated. The

A. pWU5 HincII EcoRI HinclI Sal I EcoRI trp p trpC trpA trp t trp t' + + I !U , 0U I vART38 Readthrough RNA -4 545 trp t' terminated 470-520 trp t terminated F 4 245

HincII B. pWU11 EcoRI Sal I EcoRI trp p, trpC I trpA ,trp t -U_. RT38 Readthrough RNA I- 465 trp t terminated I -4 245

C. pWU15 EcoRI EcoRI p trp t' * trp trpC .

Readthrough RNA F- 1 355 trp t' terminated - 240-290 FIG 1. Plasmids and templates generated for use in transcription studies. Solid lines represent E. coli trp DNA; broken lines are pBR322 DNA. The trp promoter (trpp), terminators (trp t and trp t'), and parts of trp structural genes (trpA and trpC) are indicated. Boxes indicate the EcoRI sites, which result from insertion of the terminator fragments into the EcoRI site of pTP119. (A) pWU5 contains a 500-bp insert spanning the trp ter- mination region and including the sequences removed by the RT38 deletion, as shown. Lines below the map indicate transcripts produced from an 800-bp Hae III-Sal I template when terminated at trp t (245 nucleotides) or trp t' (470) or continued to the end ofthe template (545). (B) The pWU11 plasmid is identical to pWU5 except for the RT38 deletion. The triangle indicates the location of the deletion. Thus, pWU11 carries trp t but not trp t'. The transcripts produced from a 720-bp Hae III template are indicated. (C) pWU15 contains a 200-bp Hincd fragment spanning trp t' (see A) inserted into the EcoRI site of pTP119. The template is a 610-bp Hae III fragment, and transcription yields the indicated RNA products. Biochemistry: Wu et al. Proc. Natl. Acad. Sci. USA 78 (1981) 2915 resulting plasmid, pTPil9, carries the trp promoter followed by a single new EcoRI site (generated by fusion of the Hae III end from the promoter fragment with a filled EcoRI end from pBR322). Because normal termination at the end ofthe trp op- 9nti* eron requires distal.sequences (7, 8), we chose to clone a 500- -.1 . bp Hae III fragment from this region that extends 387 nucleo- - tides beyond trpA and encompasses the DNA removed by sev- A* W.. eral ofthe smaller deletions affecting termination (5, 8). EcoRI 0.-..... linkers were ligated to this terminator fragment, which was then inserted into the EcoRI site ofpTPll9 to yield pWU5 (Fig. 1A). Transcription ofpWU5 Reveals a Secondp-Dependent Ter- mination Site. An 800-bp Hae III-Sal I fragment from pWU5 FIG. 3. Two-dimensional analysis of T1 ribonuclease fragments carrying the trp promoter and terminator region (Fig. LA) was from RNA corresponding to the termination region of the trp operon. Analysis ofRNA transcripts synthesized in vitro with [a-32P]GTP was used as the template in transcription reactions. trp-promoted described previously (pH 3.5 electrophoresis from left to right, homo- transcription that terminates at trp t should yield a 245-base chromatography from bottom to top; see refs. 5 and 8). (Left) Analysis transcript; the readthrough (runoff) transcript should be 545 of trp t terminated RNA, band A of Fig. 2. (Right) Analysis of p-ter- bases long. Fig. 2, lane 1, shows that in the absence ofp, 25% minated transcripts extending to trp t', band C of Fig. 2. The oligo- of the trp-promoted transcription halts at trp t (band A), and nucleotides present in band A are a subset ofthose present in band C, the remaining 75% reads through to the end of the template and correspond to sequences from the trp leader-trpC regions (11) and the 3' end ofthe operon (5, 8). Further analysis (data not shown) dem- (band B). When p is included in the reaction mixture, no effect onstrated that the additional oligonucleotides in C are derived from is seen at trp t, where termination is still.25% efficient. How- sequences between trp t and trp t', establishing that A and C are the ever, the remaining 75% of the transcripts are terminated at a transcripts indicated in Fig. 1A. A similar analysis ofband B (Fig. 2) new site, which we have designated trp t' (Fig. 2, lane 2, band confirmed its identity as the readthrough transcript (data not shown).

1 2 3 4 C). Although the trp t' terminated products are heterogeneous in size, it is clear that this region contains a strong site for p- dependent transcription termination. RNAs eluted from bands A, B, and C were digested with Ti ribonuclease and the frag- ments were analyzed two-dimensionally (see Fig. 3), which ver- ified that they correspond to the transcripts shown in Fig. 1A. To investigate these sites from another viewpoint, we carried out transcriptions using the mutant rpo2O3 polymerase. This mutation arose as a reversion restoring termination at the end ofthe trp operon in a strain carrying the p mutation rho201 (12). We therefore might expect to see a difference in the response of the rpo2O3 enzyme at the p-dependent site trp t' compared to the wild-type enzyme. Fig. 2 (band A, lanes 1-4) illustrates that the major difference actually occurs at trp t, where the efficiency of termination is greater with rpo2O3 polymerase (45%) than with wild type (25%). The behavior at trp t' is qual- itatively the same for both enzymes (band C, lanes 2 and 4). A Separation ofthe Tandem Terminators. We subcloned each ofthe two termination sites separately next to the trp promoter in order to characterize them independently. To subclone trp t we utilized the RT38 deletion (7), which removes sequences required for trp t' function. The 315-bp Hae III fragment from ART38 (8) corresponding to the wild-type 500-bp fragment was inserted into the EcoRI site ofpTPll9 by using EcoRI linkers as described above. The template from this plasmid (pWU1l, FIG. 2. Transcription from a template carrying the wild-type trp Fig. 1B) contains the trp t fragment in the same orientation as termination region. Purified Hae HI-Sal I fragment from pWU5 was the promoter. trp t' was cloned by inserting a 200-bp HincIH used in standard transcription reactions. The products oftranscription fragment from pWU5 (see Fig. 1A) into pTPll9 plasmid by using wild-type RNA polymerase are shown- without (lane 1) or with using EcoRI linkers, yielding pWU15 (Fig. LC). (lane 2) the addition of p. Band A represents material terminated at trp t~,band B is readthrough to the end of the template fragment. Band Fig. 4 shows products from transcription ofHae III templates C consists of transcripts terminated at trp t' in the presence of p. Also from pWUll (trp t) and pWU15 (trp t') in the presence and shown are the products of transcription using the. mutant rpo2O3 poly- absence of p. Lanes 1 and 2 show that trp t still functions as a merase, either alone (lane 3) or with p (lane 4). Bands were cut out, and weak p-independent terminator in vitro. As before, trp t' is their radioactivities were measured and the counts were normalized efficient and completely p-dependent (lanes 3 and 4). Thus, in to reflect the guanosine content of each transcript before the percent- the behavior of each terminator on its own is the same age of termination or readthrough was determined. Two-dimensional vitro, analysis of oligonucleotides (Fig. 3) confirms that bands A, B, and C as it is when carried in the wild-type tandem configuration. correspond to the 245-, 545-, and 470-base transcripts diagrammed in Location of the p-Dependent Termination Site trp t'. p-de- Fig. LA. The high molecular weight products are presumably due to pendent termination at trp t' is heterogeneous. Preparative end on transcription of the template. The disappearance of the high transcription reactions generally yielded the bands numbered molecular weight transcripts in the presence of p was variable, (com- 1-5 in Fig. 2. Each band in turn was a cluster containing several pared Fig. 2 with Fig.- 4) and we did not pursue this problem. In ad- 3' termini. After end labeling using and RNA ligase, dition, the rpo2O3 mutation causes polymerase to terminate more read- [32P]pCp ily (i2, i3), and we believe that this explains the reduction of high RNA from each band was eluted and subjected to complete molecular weight products in lanes 3 and 4. digestion with ribonuclease T1 to yield only the 3'-labeled ter- 2916 Bio6hemistry: Wu et al. Proc. Natl. Acad. Sci. USA 78 (1981)

1 2 3 4 nucleotide (labeled by transfer of32P from pCp). Partial alkaline hydrolysis and analysis of the resulting "wandering spot" pat- terns allowed alignment with the previously determined DNA sequence. The major sites oftermination at trp t' are shown in h Fig. 5.

DISCUSSION RT _ Termination oftranscription at the end ofthe E. coli tryptophan operon is not a simple event. Although initial analysis revealed RT a classic terminator structure and a homogeneous 3' end for the transcript (5), the location of deletions affecting termination .t' distal to the point corresponding to the 3' end ofthe transcript t raised unexpected questions about the mechanism of termi- nation (8). We have now discovered a second termination site (trp t') located about 250 nucleotides beyond the first (trp t), in a region affected by all of the distal deletions. This explains the previously puzzling failure to find terminator mutations in trp t alone (7)-only by eliminating the function at trp t' could readthrough transcription into distal lac genes occur. Our char- acterization in vitro of these tandem termination sites reveals that the structural features and functional characteristics of the two sites differ considerably. The sites on a DNA template at which termination can occur FIG. 4. Transcription response at the subcloned termination sites. are of two major types-those that require p for termination in This gel shows the transcription products from pWU11 (lanes 1 and vitro and those that do not. The p-independent terminators 2) and pWU15 (lanes 3 and 4) templates using wild-type RNA poly- share some striking features: a G+C-rich region of dyad sym- merase. RT, readthrough transcript. The subcloned trp t site is a weak metry is followed by a series ofthymidines corresponding to the terminator in the absence (lane 1) or presence (lane 2) ofp in the tran- uridines at the 3' terminus scription reaction mixture. With the isolated trp t' region, transcrip- ofthe transcript (2, 6). Transcription tion without p (lane 3) continues to the end ofthe template fragment, studies using a variety of templates, nucleotide analogs, and a whereas with p (lane 4) termination at trp t' occurs. The degree ofhet- mutant polymerase have helped delineate the roles ofthese two erogeneity observed in p-dependent termination varied in our reac- features (4, 13, 14). In several systems (13,15,16), formation of tions regardless ofwhether pWU5 or pWU15 templates were used, and a strong RNA hairpin has been shown to induce a pause in the it appeared to be highly sensitive to the relative amounts ofDNA tem- continuation oftranscription. Weak base-pairing ofthe uridine- plate, RNA polymerase, and p. rich terminal region of the transcript with the DNA template enhances release (6, 13, 17), thus resulting in termination of minal oligonucleotides. The terminal oligonucleotides were synthesis. separated either on 20%o acrylamide/7 M urea gels or by a stan- Because the DNA immediately following the trpA gene con- dard two-dimensional electrophoresis/homochromatography. tains a region of strong dyad symmetry followed by four thy- Complete alkaline hydrolysis identified the terminal 3'-OH midines, this site should function as a p-independent terminator

4 IAAI CCC.CCCG C C G C C AGITTCCGICTGGCAGTGCIATTTTAACTTTCTTTAATG

'4~~~~~~~~~~~~~~~~~l -RT38 t t,p L: t trpA X trp t ,l trp t' 0 50 , ' 250 315

5 4 3 2 C C I C A A A A T A T A T T T T C C C T C I A T C G G C G CT T A A TTTG A A CT A G C A C I TIC TIT A C I ITT C A G FIG. 5. Location ofthe termination sites at the end ofthe trp operon. The central diagram shows the location oftrp tand trp t'. The top line shows the DNA sequence of trp t (5); the area of dyad symmetry is boxed and the precise point of termination is indicated by the arrow. On the bottom line is the DNA sequence across trp t' (8). The numbers 1-5 correspond to the numbered bands in Fig. 2. Longer arrows indicate the prominent 3' termini; shorter arrows show lesser points oftermination. The distal deletions (such as RT38), which result in readthrough transcription at the end of the trp operon (7, 8), thus appear to remove sequences required for termination at trp t'. Biochemistry: Wu et al. Proc. Natl. Acad. Sci. USA 78 (1981) 2917

in vitro. The G +C-rich RNA hairpin causes a pronounced pause functions as a termination site in vitro; thus, only if the pro- in the progression ofRNA synthesis as polymerase traverses the cessing and termination sites are coincident is this a plausible trp t region in vitro (16). However, as can be seen in Figs. 2 explanation. and 4, only 25% of the trp-promoted transcripts terminate at Several questions about the function and arrangement ofsites trp t. The inefficiency ofthis site is probably due to the presence in the trp termination region remain unanswered at present. of only four uridines at the end of the transcript; most strong The two sites we have identified are clearly different in their p-independent terminators have six to eight terminal uridines structural features (both primary and secondary), as well as in (6). This conclusion is supported by studies using the rpo203 their function in vitro. Whether the discrepancy between their polymerase, which has been shown to terminate more effi- behavior in vitro and in vivo is due to position effects (i.e., re- ciently than wild-type polymerase at a mutant attenuator site moval or addition of nearby critical sequences during cloning) encoding only four consecutive uridines (13). Fig. 4 shows that or to the absence of other factors is not known. Interestingly, the rpo2O3 enzyme also terminates better at trp t (45% instead a tandem arrangement of termination sites apparently occurs of 25%). Further characterization of the properties of trp t in at the end of the trp operon in Salmonella typhimurium. As in vitro and a comparison with several other terminator sites are E. coli, a region ofstrong dyad symmetry follows the Salmonella reported elsewhere (14). trpA gene and distal sequence has been shown to be required Far less is known concerning the relationship of structure to for termination oftranscription in vivo (22). Transcription stud- function at sites at which polymerase is completely dependent ies indicate that there are several p-dependent termination sites on p for termination in vitro. Three p-dependent termination within this distal region (B. Nichols and C. Yanofsky, personal sites have been previously characterized: AtR1 (15), the tRNATYr communication). Multiple terminators also occur at the end of terminator (18), and a site within the A cro gene called AtRo (10). the tRNATYr gene in E. coli (18). In this case, a 178-nucleotide These sites share the following features: (i) dyad symmetry pre- sequence is repeated three times with several single base ceding the termination point (albeit weaker than that usually changes from repeat to repeat. Polymerase and p are extremely found at p-independent terminators), (ii) and A+U-rich se- sensitive to the specific sequence changes in the repeats, re- quence at the terminus of the RNA, and (iii) the common se- sulting in quite different responses at the three sites (18). It has quence C-A-A-T-C-A-A.in AtRi and the tRNATYr terminator and been suggested that clustered p-dependent termination sites the subset A-T-C-A-A in AtRo just preceding the stop site. occur at AtR2 (3) and AtRL (23). Further research should show trp t', like these other -dependent termination sites, occurs whether multiple terminination sites are a common feature of in an A+T-rich region of DNA; A and U are the preferred ter- prokaryotic gene structure and whether they simply ensure minal nucleotides (Fig. 5). Due to the composition of trp distal against readthrough transcription into adjacent genes or serve DNA, sequences similar to the (C-A)-A-T-C-A-A prototype can in a more complex regulatory capacity. be found, but there is no obvious relationship between such We thank our colleagues for expert and helpful comments and crit- sequences termination A striking and the points of (see Fig. 5). icisms throughout the course of this work and Kathy Fiasconaro for deviation of trp t' from other termination sequences, p-depen- cheerful and efficient typing of the manuscript. This research was sup- dent or independent, is the lack of any strong dyad symmetry ported by National Insitiute of Health Grant GM-22830 to T.P. across this region. In addition, the p-dependent termination observed here is quite heterogeneous, occurring across 50 nu- 1. Roberts, J. (1969) Nature (London) 224, 1168-1174. cleotides (Fig. 5). 2. Adhya, S. & Gottesman, M. (1978) Annu. Rev. Biochem. 47, Other factors probably contribute to the specificity of ter- 967-996. 3. Greenblatt, J. (1981) Cell 24, 8-9. mination at trp t'. In particular, the E. coli nusA protein (nusA) 4. Yanofsky, C. (1981) Nature (London) 289, 751-758. has been shown to mediate termination and antitermination 5. Wu, A. M. & Platt, T. (1978) Proc. Nati. Acad. Sci. USA 75, events in several other systems (3, 19, 20). Greenblatt et al. 5442-5446. (20) have proposed that nusA is in fact an integral component 6. Rosenberg, M. & Court, D. (1979) Annu. Rev. Genet. 13, of the normal elongating RNA polymerase complex in E. coli 319-353. and is involved in oftermination signals in the cell. 7. Guarente, L. P., Beckwith, J., Wu, A. M. & Platt, T. (1979) J. recognition Mol. Biol. 133, 189-197. This hypothesis is strongly supported by the effect of nusA on 8. Wu, A. M., Chapman, A. B., Platt, T., Guarente, L. P. & Beck- polymerase behavior at the trp termination region. Although with, J. (1980) Cell 19, 829-836. nusA alone has no effect, when p and nusA are included to- 9. Burgess, R. R. & Jendrisak, J. J. (1975) Biochemistry 14, gether, the heterogeneity we have reported here for termina- 4634-4638. tion at the trp t' site is almost eliminated (there is no effect at 10. Calva, E. & Burgess, R. R. (1980)J. Biol. Chem. 255, 11017-11022. the trp t site) (21). 11. Christie, G. E. & Platt, T. (1980)J. Mol. Biol. 143, 335-341. 12. Guarente, L. P. & Beckwith, J. (1978) Proc. Natl.Acad. Sci. USA The participation of additional factors may explain the puz- 75, 294-297. zling results we have observed in vivo. Readthrough at the nor- 13. Farnham, P. J. & Platt, T. (1980) Cell 20, 739-748. mally efficient trp t site can be caused by the introduction of 14. Christie, G. E., Farnham, P. J. & Platt, T. (1981) Proc. Natl. a mutation in p (5,12) or by the removal of distal sequencein Acad. Sci. USA 78, in press. the trp t' region (8). The most direct interpretation ofthese data 15. Rosenberg, M., Court, D., Shimatake, H. & Wulff, D. L. (1978) is that there is a interaction between these two Nature (London) 272, 414-423. p-dependent 16. Farnham, P. J. & Platt, T. (1981) Nucleic Acids Res. 9, 563-577. termination sites in vivo. Alternatively, these data can be ac- 17. Martin, F. & Tinoco, I. (1980) Nucleic Acids Res. 8, 2295-2300. counted for by a processing event that generates the observed 18. Kupper, H., Sekiya, T., Rosenberg, M., Egan, J. & Landy, A. 3' terminus in vivo. For example, if trp operon transcription (1978) Nature (London) 272, 423428. normally terminates at trp t' and this is followed by cleavage 19. Greenblatt, J., McLimont, M. & Hanly, S. (1981) Nature (Lon- of the message to yield the observed 3' end, deletion of trp t' don), in press. and the adjoining processing signals would lead to readthrough 20. Greenblatt, J. & Li, J. (1981) Cell 24, 421428. 21. Platt, T. (1981) Cell 24, 10-23. transcription. In order to explain the readthrough transcription 22. Nichols, B. P., Blumenberg, M. & Yanofsky, C. (1981) Nucleic observed in the rho201 strain, we would further have to pos- Acids Res. 9, 1743-1755. tulate that processing depends on pfunction. However, we have 23. Calva, E., Rosenvold, E. C., Szybalski, W. & Burgess, R. R. shown (Fig. 2) that the presumed in vivo terminator, trp t, also (1980) J. Biol. Chem. 255, 11011-11016.