Molecular and CellularBiochemistry 104: 163-168, 1991. © 1991KluwerAcademicPublishers. Printedin the Netherlands.

Termination of transcription of ribosomal RNA in Saccharomyces cerevisiae Stewart P. Johnson 1 and Jonathan R. Warner Department of Cell Biology, Albert Einstein College of Medicine, Bronx, NY10461, USA; i Present address: Dept. of Pathology, Duke University Medical Center, Durham, NC27710, USA

Key words: ribosomal R N A transcription, D N A binding protein sites Abstract

We have attempted to determine the site of termination of transcription of ribosomal R N A in the yeast, Saccharomyces cerevisiae. While a quantitative description of the termination sites of R N A polymerase I is not possible using presently available methods, we conclude that transcription of most molecules continues through a large portion of the adjacent enhancer region. There are two potential termination sites within the enhancer, one of which is near the binding site of the DNA binding protein REBI. In addition there is an apparently fail-safe termination site approximately 950 nucleotides beyond the 3' end of 35S ribosomal precursor RNA. Processing at the end of 35S RNA influences the choice of downstream termination site. Conversely downstream sequences also influence the site of termination.

Introduction

The transcription of ribosomal R N A is of particular interest both because it makes up a major portion of the transcription of the cell and because the ribosomal R N A genes are present in the genome as a tandem array. It has become clear that a processing event is responsible for the 3' end of the rRNA precursor molecules found in the nucleolus in yeast [1] as well as in higher organisms [2, 3]. Therefore, an important problem has been the identification of the actual site of the termination of transcription, in particular, because the tandem nature of the genes suggests the possibility that the termination of transcription of one gene may be coupled to the initiation of transcription at the next one. Experimentally the question is a difficult one, both for POL 1 transcripts and for POL II transcripts (reviewed in 4), because of the very rapid degradation of the 3' products of such a processing reaction. Transcription of ribosomal R N A genes by R N A polymerase I, including the question of termina-

tion, has been reviewed recently [5]. Briefly, in mouse, termination appears to occur 565 nucleotides downstream of the end of the 45S rRNA precursor in front of a group of repeated sequences termed 'Sal boxes' [6]. An additional termination site is found not far upstream of the initiation site. Whether its function is as a 'fail-safe' terminator to protect the initiation complex from errant polymerases, or whether it is related to the presence of cryptic promoters in the non-transcribed spacer (NTS) is still unclear. In Xenopus laevis, there has been some evidence that transcription reads through most of the NTS, terminating just upstream of the promoter [2], but in other Xenopus species, transcription terminates about 235 nucleotides from the end of the precursor RNA [5]. In the yeast Saccharomyces cerevisiae there is also some evidence that transcription continues beyond the end of the 35S precursor [1]. This is particularly interesting because the major enhancer of rRNA transcription lies just there [7, 8] (see Fig. 1). In this paper, we have used three independent methods to assess the site(s) of termination, and find that there

164

35 S ----

//

35 S

I

p

e

y

I'

•

I

//

E

TO TI

T2 210

T3A 690

T3B 950

5S

Tp

Fig. 1. Schematic representation of a portion of the r D N A repeat. Sequences are numbered from the 3' end of the 25S RNA. The 5S transcript and the beginning and the end of the 35S transcript are shown. The enhancer element is indicated, as are the potential termination sites, T2, T3A, T3B, and Tp, described by Van der Sande, et al. [9]. TO and T1 represent processing sites responsible for generating the 3' end of 35S precursor RNA. E and H indicate EcoRI and HindlII restriction sites. The inverted triangles represent the REB1 binding sites at the enhancer(e) and promoter(p).

are likely to be several between the end of the 35S RNA gene and the end of the 5S RNA gene. After these experiments had been completed, Van der Sande et al. reported results in general agreement with ours [9].

triphosphates added to such cells are incorporated into RNA for a brief period. The radiolabeled RNA can be hybridized to complementary sequences; the amount of radioactivity in doublestranded form is then a rough measure of the transcriptional activity of that sequence. We have utilized this protocol to measure transcription across

Materials and methods

Saccharomyces cerevisiae, strains W303 and J400, was cultured, and its RNA was prepared as described previously [7, 8]. Run-on transcription was carried out as described [8]. Construction and analysis of the products of transcription of the artificial rDNA gene have been described in detail [7].

Table I. Cells of Saccharomyces cerevisiae, strain W303a, were permeabilized and labeled with a3zP UTP as described previously [8]. R N A was prepared and hybridized to slotblot filters, where each slot contained a different species of complementary R N A , transcribed from a p G E M plasmid carrying the indicated sequences of the rDNA repeat. After hybridization, the filter was washed, treated with pancreatic RNAse to remove singlestranded tails, and subjected to autoradiography. Then, each slot was excised and its radioactivity determined in a scintillation counter.

Results

Complementary rDNA sequences a

Relative transcription b

Measurement of the rate of transcription of a particular piece of DNA can be difficult if the lifetime of the product is very short. As a way around that problem, run-on transcription assays have been developed for mammalian cells [10, 11], based on the observation that, in isolated nuclei, RNA polymerase molecules will continue a transcription in which they are involved for a few hundred nucleotides, but will not initiate new chains. Furthermore, processing and degradation of this RNA is minimal. We have developed a run-on transcription protocol for Saccharomyces, in which the membrane is permeabilized with detergent, leaving the cell wall intact [8]. Radiolabeled nucleoside

- 496 + 96 + 288 + 401 + 774 + 992 + 1484 + 1797

12.3 c 52.2 1.9 3.1 2.2 1.9 0.95 0

to to to to to to to to

+ + + + + + + +

101 289 400 671 969 1179 1644 1910

Position is stated relative to the 3' end of 25S rRNA. 5S R N A is transcribed from + 1255 to + 1137. b Calculated as cpm relative to the U content of the hybridized RNA. cThis value is artifically low probably reflecting the saturation of the filter with mature r R N A molecules. In other experiments, the relative transcription of this region was 2 to 3 fold greater than + 96 to + 289.

165

the r D N A spacer region using as hybridization probes anti-sense RNAs synthesized in vitro from p G E M plasmids that contained small fragments of the spacer DNA. As shown in Table 1, the region of the spacer from 96 to 289 bp downstream of the 3' end of 25S r R N A is as transcriptionally active as sequences coding for 25S rRNA. Beyond this region transcription drops off markedly, although a low level of transcription is still measurable 1 kb beyond the end of 25S RNA. These results suggest that 8090% of the R N A polymerase I molecules are terminating within or, perhaps slightly beyond, the EcoRI-HindIII enhancer element, and the remainder are terminating much further downstream. This is in relative agreement with the results of similar experiments carried out by Van tier Sande etal. [9]. These results suggest that termination actually occurs within or shortly downstream of the r R N A enhancer.

01 oo N 4-

~!" o ~!4-

tn r,,, (D 4-

t'~ r,,. O1 4"

l

n

i

i

0

4"

O1

4"

0

÷

I~

4-

1315

705 475

i

Transcripts from the NTS1 region of spacer rDNA As another approach, we have searched for transcripts of the spacer region, using anti-sense RNAs radiolabeled in vitro. R N A was isolated from an exponentially growing culture of cells, separated by denaturing gel electrophoresis, and transferred to a nylon filter. The R N A was cross-linked to the filter by U V irradiation and then probed sequentially with complementary R N A s synthesized from p G E M plasmids containing discrete segments of the r D N A spacer. As shown in Fig. 2, a probe from + 96 to + 289 hybridizes to four RNAs. These RNAs are approximately 240, 320, 470, and 850 bases in length. When the filter was stripped and reprobed with R N A complementary to + 292 to + 404, the smallest of these was no longer observed. Similarly, a probe of + 405 to + 675 hybridizes to the two largest RNAs, and a probe of + 778 to + 972 hybridizes to only the largest of the RNAs. An antisense probe that spans + 915 to + 1060 hybridizes to none of these RNAs (not shown). The simplest explanation for these results is that these R N A s possess a common 5' end, perhaps that generated by the processing event that generates

Fig. 2. Transcriptsof the NTS sequences. Total RNA prepared from log-phasecellswas subjectedto Northern analysis, and the resulting filter was hybridizedsequentially to a series of radioactive anti-sense RNA probes from the NTS, as indicated. The relative intensity between the lanes is arbitrary, depending on the specificactivityof the probe and the length of exposure of the autoradiograph. The relative concentration of the different RNA species is apparent from comparing the intensity of the bands in the first lane. the end of 25S rRNA. If that were the case, the 3' end of the smallest of the RNAs would be near the T2 site [9] and that of the largest R N A near the T3B site [9]. The two other RNAs do not correlate to any of the 3' end generating sites previously described and might represent intermediates in the degradation of the longest transcript.

166

T7 rRNA

RR 10 ~ / [ Enh

I

I

RR 60

I

I

I

I

~

I

I

_A_

I

I

T7

seq

Eco E

RR 61

RR 63 ~ / /

RR 62 ~

RR 58 ~ - ~ / /

//

- -

//

//

A RR 117 ! ~ - ~ / /

I

I

-"

II

Fig. 3. Artificial rRNA transcription unit containing bacteriophage T7 sequences. The original construct, RR10, has been described in detail [8]. Briefly, the sequences shown are carried on a centromere plasmid, YCp50, that is maintained at about one copy per cell. Enh is the 190bp enhancer sequence; T7 is the 600bp fragment of bacteriophage T7 that is used to identify the transcript of this gene, T7 rRNA. EcoE is the 591 bp EcoRi 'E' fragment that contains the T0 and T1 processing sites. The TO and perhaps the T1 sites were removed by deleting the sequence from - 149 to + 37 (large triangle), to generate plasmid RR60, in which there are no NTS sequences beyond + 100. The other constructs are as follows: RR61: Includes the EcoRI-HindIII enhancer fragment, i.e. sequences from + 100 to + 290. RR63: Includes the entire NTS. RR62: Includes the entire NTS except the enhancer fragment. RR58: Includes the entire NTS except nucleotides + 108 to + 112 (arrowhead), that comprise the REB1 binding site. RR117: Includes the entire NTS except nucleotides + 218 to + 246 (arrowhead), that include the major run of T residues within the enhancer.

Expression of an r D N A mini-gene that lacks a 3' processing site W e have previously described the use of an r D N A mini-gene to study the p r o m o t e r and e n h a n c e r functions of the spacer r D N A [7, 8; see Fig. 3]. T h e transcript can be identified because it carries sequences of b a c t e r i o p h a g e T7. Its 3' end is generated by the same processing event that p r o d u c e s the 35S precursor R N A . In o r d e r to ask h o w far transcription continues, we have deleted the sequence f r o m - 1 4 9 to + 37, r e m o v i n g the T0 site and p e r h a p s the T1 sites (Fig. 1). This was then substituted for the f r a g m e n t present in R R 1 0 to generate R R 6 0 (Fig. 3). As shown in Fig. 4, transcription

f r o m this minigene terminates at multiple sites within the plasmid sequences. W h e n the minigene sequence was extended by adding the E c o R I - H i n d I I I e n h a n c e r f r a g m e n t (RR61), most of the transcripts still read t h r o u g h into the plasmid sequences, but a novel R N A species, R2, is p r o d u c e d whose 3' end is near the H i n d l I I site. W h e n the complete spacer r D N A is present d o w n s t r e a m of the mini-gene (RR63), there are three p r o m i n e n t R N A species, as well as a n u m b e r of m i n o r ones. In addition to an R N A that comigrates with R2, there is R1, whose 3' end is near the E c o R I site, and R3, whose 3' end is approximately 650 bases downstream of R2 (Fig. 1). W h e n the e n h a n c e r f r a g m e n t was deleted f r o m this construct ( R R 6 2 ) , neither R1

167 nor R2 were present. Almost all the transcript was in the form of R3, which is shorter by 190 nucleotides in this case, since it lacks sequences from the enhancer. These results show that specific sequences are necessary for termination to occur. In particular the presence of sequences determining termination at the end of R3 appear to be 100% effective. When they are present, no longer transcripts are observed. The 3' end of R3 appears to coincide with termination site T3B described by Van der Sande et al. [91 (Fig. 1). Precise identification of the 3' ends of R1 and R2 is more problematical. It appears that R2 terminates at the site T2 described by Van der Sande et al. [9], but it may be beyond rather than before the stretch of T residues at + 217 to + 246. The 3' end of R1 is slightly downstream of the EcoRI site that limits the 5' end of the enhancer fragment. Close analysis of these results yields some curious conclusions. For instance, comparison of RNAs from YCpRR61 and YCpRR63 suggests that the production of R1 depends on sequences downstream of the HindIII site, well over 150 bases away. We have identified a protein, termed REB1, that binds both to the enhancer region and to the promoter region of the NTS [12]. When a 6 bp deletion of the REB1 binding site within the enhancer is introduced into the test plasmid (yielding RR58), R1 is no longer produced (Fig. 4). Similarly, Van der Sande et al. [9] have identified an additional termination site Tp, near the promoter ( - 3 0 0 from the initiation site.) When they disrupted the REB1 binding site (at - 2 0 0 from the initiation site) termination activity of Tp was abolished. Thus, in two instances, disruption of a REB1 binding site abolished a termination site. The T2 site has been mapped to a point immediately upstream of a stretch of T residues that is present at + 217 to + 246, within the enhancer fragment. It should be noted that nuclease S1 analyses of this site show considerable protection into this T-rich region [9]. Deletion of this run of T residues (yielding RR117) leads to a decrease in the relative amount of R2, although one now sees a somewhat larger R N A that is a light band in RR63 but is normally obscured by the presence of R2

n-

~

n-

n-

n-

n-

R3

R2

--

R1

--

Fig. 4. Products of transcription units lacking TO. RNA prepared from strain J400 carrying the plasmids described in Fig. 3 (except RR10, see ref. 8) was subjected to Northern analysis and probed with labeled RNA complementary to the T7A sequence present in the construct. Bands R1, R2, and R3 are indicated. Their sizes were estimated from known marker bands.

(Fig. 4). Thus, the T-rich region may influence the efficiency of T2.

Discussion

One almost feels impelled to invoke the Heisenberg Uncertainty Principle for the study of the termination of transcription. Each experimental approach introduces its own perturbation into the system, thus casting into doubt the conclusions. In our case, although we have used three separate approaches, each has its potential drawbacks. a) Run-on transcription is said to proceed only a few hundred nucleotides, not to intiate new transcripts at either correct or incorrect sites, and not to be subject to post-transcriptional modifications such aas rapid degradation [10, 11]. The data rigorously supporting each of

168 those claims is sparse, especially for the Saccharomyces cerevisiae system. b) Northern analysis of transcripts assumes that the products observed are the normal products of transcription, and that those containing NTS sequences are rare simply because they are degraded rapidly. An alternative possibility is that they are aberrant, rare transcripts unrepresentative of normal transcription events. c) Finally, the deletion of the 3' processing signal, as we have done in the R R plasmids shown in Fig. 3, may itself lead to aberrant termination. Indeed, in studies of POL II transcription it has been shown that deletion of the usual poly adenylation signal can alter termination at downstream sequences [4]. Nevertheless a number of conclusions can be safely drawn from the data: a) In the presence of TO and all the downstream sequences, substantial transcription continues into the enhancer, but probably less than 10% of the molecules continue beyond the HindIII site marking the end of the enhancer at + 292. b) The presence of sequences within the enhancer element lead to termination. Sequences both at the REB1 binding site and the T-stretch are seen to lead to inefficient but specific termination sites, at least in the absence of TO (Fig. 4). c) Sequences both upstream and downstream can affect termination events. Thus, in the presence of TO most transcripts terminate within the enhancer (Table 1), but in its absence, many read through the enhancer (Fig. 4). Furthermore termination leading to the transcript R1 (Fig. 4) occurs only in the presence of sequences relatively far downstream. d) Little if any transcription is seen beyond T3B. Either T3B represents a fail-safe termination site, or transcripts beyond it are degraded immediately.

sistance of Mary Studeny. This work has been supported by grants from the NIH # GM25532 and CA13330.

Acknowledgements

Address for offprints: J.R. Warner, Department of Cell Biol-

The authors acknowledge fruitful discussions with Bernice Morrow and the invaluable technical as-

References 1. Kempers-Veenstra AE, Oliemans J, Offenberg H, Dekker AF, Piper PW, Planta R J, Klootwijk J: 3'-End formation of transcripts from the yeast rRNA operon. EMBO J 5: 27032710, 1986 2. Labhart P, Reeder RH: Characterization of three sites of RNA 3' end formation in the Xenopus ribosomal gene spacer. Cell 45: 431-443, 1986 3. Bartsch I, Schoneberg C, Grummt I: Purification and characterization of TTF I, a factor that mediates termination of mouse ribosomal DNA transcription. Mol Cell Biol 8: 3891-3897, 1988 4. Proudfoot NJ: How RNA polymerase II terminates transcription in higher eukaryotes. Trends Bio Sci 14: 105-110, 1989 5. Reeder RH: Regulatory elements ofthe generic ribosomal gene. Curt Op in Cell Biol 1: 466-474, 1989 6. Grummt I, Maier U, Ohrlein A, Hassouna N, Bachellerie JP: Transcription of mouse rDNA terminates downstream of the 3' end of 28S RNA and involves the interaction of factors with repeated sequences in the 3' spacer. Cell 43: 801-810, 1985 7. Johnson SP, Warner JR: Unusual enhancer function in yeast rRNA transcription. Molec Cell Biol 9: 4986--4993, 1989 8. Elion EA, Warner JR: An RNA polymerase I enhancer in Saccharomyces cerevisiae. Molec and Cell Biol 6: 20892097, 1986 9. van der Sande CAFM, Kulkens T, Kramer AB, de Wijs IJ, van Heerikhuizen H, Klootwijk J, Planta ILl: Termination of transcription by yeast RNA polymerase I. Nucl Acids Res 17: 9127-9146, 1989 10. McKnight GS, Palmiter RD: Transcriptional regulation of the ovalbumin and conalbumin genes by steroid hormones in chick oviduct. J Biol Chem 254: 9050-9058, 1979 11. Weber J, Jelinek W, Darnell Jr. JE: The definition of a large viral transcription unit late in Ad2 infection of HeLa cells: mapping of nascent RNA molecules labeled in isolated nuclei. Cell 10: 611-616, 1977 12. Morrow BE, Johnson SP, Warner JR: Proteins that bind to the yeast rDNA enhancer. J Biol Chem 264: 9061-9064, 1989

ogy, Albert Einstein College of Medicine, Bronx, NY 10461, USA