|Home | About | Journals | Submit | Contact Us | Français|
We have mapped transcription termination sites for RNA polymerase I in the yeast Saccharomyces cerevisiae. S1 nuclease mapping shows that the primary terminator is the Reb1p terminator located at +93 downstream of the 3′ end of 25S rRNA. Reverse transcription coupled with quantitative PCR shows that approximately 90% of all transcripts terminate at this site. Transcripts which read through the +93 site quantitatively terminate at a fail-safe terminator located further downstream at +250. Inactivation of Rnt1p (an RNase III involved in processing the 3′ end of 25S rRNA) greatly stabilizes transcripts extending to both sites and increases readthrough at the +93 site. In vivo assay of mutants of the Reb1p terminator shows that this site operates in vivo by the same mechanism as has previously been delineated through in vitro studies.
Understanding the mechanism of transcription termination requires reproduction of the termination process in vitro where appropriately detailed experiments are feasible. For study of RNA polymerase I (PolI) termination in the yeast Saccharomyces cerevisiae, we initially used a whole-cell extract to catalogue sites of RNA 3′ end formation downstream of the 3′ end of mature 25S rRNA (15). Beginning at the 3′ end of 25S rRNA (position −1), we mapped additional termini to +93, to about +250, and to a further site in the vicinity of +350. Eventually we settled on the +93 site for further study since it possessed several attributes expected of the primary terminator for PolI transcription. For example, the +93 site behaved kinetically as a terminator rather than as a processing site. Addition of actinomycin D to a whole-cell transcription reaction abruptly stopped both transcription and accumulation of termini at +93, as expected of a true terminator. In contrast, accumulation of the 3′ end of 25S rRNA continued after transcription stopped, as expected of a processing site (15). In addition, later work using beaded templates demonstrated that transcripts are released at +93 rather than just being paused on the template (14, 16).
We have also found that both the structure and the mechanism of action of the +93 terminator are similar to those of PolI terminators that have been studied in vertebrates, indicating conservation during evolution. Similarities include the fact that in both yeast and vertebrates, termination occurs just upstream of a protein binding site (Reb1p in yeast and TTFI in mice or humans). These terminator proteins function only when bound in one orientation, and they all utilize paired Myb homology domains as DNA binding elements. In addition to the protein binding site, it is now recognized that PolI terminators contain a second DNA element which codes for the last 10 to 12 nucleotides (nt) of the terminated transcript and which is often T rich. Studies on mechanism, in both yeasts and vertebrates, show that termination at these sites occurs in two steps, pause and release; the protein binding site acting primarily as the pause element, while the upstream region controlling release. These findings have recently been reviewed (21).
Conservation of structure and function from yeasts to humans suggests that the +93 terminator performs a function important for cell survival. However, the lability of transcripts in this region and their consequent low abundance have made it difficult to study PolI termination in the living cell. The inability of some workers to detect termination at +93 has drawn the existence of a terminator at this site into question (4, 9, 23).
Due to the lack of consensus in the literature, we felt it worthwhile to reinvestigate PolI termination in vivo in S. cerevisiae, using more sensitive and quantitative technology. In this report we show, by a combination of S1 nuclease protection and reverse transcriptase-coupled PCR (RT-PCR), that termini corresponding to the +93 Reb1p terminator are readily detected in exponentially growing wild-type yeast cells. Furthermore, quantitative transcript measurement indicates that at least 90% of all PolI transcripts stop at this point. We also show by in vivo mutagenesis that the +93 terminator functions by the same rules in vivo as we have previously observed in vitro. We have identified two circumstances under which significant readthrough occurs at the +93 terminator. Transcripts reading through +93 are completely stopped by a fail-safe terminator located at +250.
The S. cerevisiae strains used were W3031a (MATa ade2-1 trp1-1 can1-100 leu2-3,112 his3-11,15 ura3-1) and HI227 (MATa leu2-3,112 ura3-52 trp1 his3Δ lys2Δ pep4-3 prb1 prc1 rnt1::URA3:rnt1-47; obtained from M. Ares).
The ribosomal minigene construct 105LT contains a 2,721-bp fragment of ribosomal DNA (rDNA), including an entire intergenic spacer and PolI promoter (see Fig. Fig.5A).5A). The spacer-promoter fragment is bounded on its upstream end by an NheI restriction cut (175 bp upstream of the 3′ end of the 25S region) to which has been fused an 8-bp linker containing a unique NotI restriction site (destroying the NheI site). On the downstream end, a TaqI site at +25 was inactivated by insertion of a 10-bp linker to create a unique XhoI site at position +36 relative to transcription initiation (insertion of this XhoI linker has been described elsewhere ). A 53-bp piece of rDNA containing the Reb1p terminator (23 bp upstream of the termination site and 30 bp downstream) was amplified by PCR using primers which placed an XhoI site upstream and a novel KpnI site downstream. This XhoI-KpnI terminator fragment was then joined to the +36 XhoI site of the spacer-promoter fragment. Finally, the +36 XhoI site was opened and a 222-bp SalI fragment from the ADE8 gene was inserted to move the Reb1p terminator a convenient distance from transcription initiation. The orientation of the SalI stuffer was chosen to recreate a unique XhoI site at +36. The final NotI-KpnI construct was then inserted between NotI and KpnI in the polylinker of pRS426 (2μm, URA3) (22).
DNA sequence surrounding the 3′ end of the rDNA coding region was amplified by PCR using an upstream primer designed to fuse three C residues to a run of four G’s about 70 bp upstream of the 3′ end of 25S rRNA and thus create a novel Pspa1 site. On the downstream end, the amplified fragment extends to a natural HincII site in the rDNA at position +415. This PspaI-HincII fragment was cloned into the polylinker of pBluescript. To make end-labeled probe, the plasmid was digested with Pspa1 and then filled in with the Klenow fragment of DNA polymerase plus [32P]dCTP and unlabeled dGTP. This yielded a 3′-labeled DNA strand complementary to the sequence shown in Fig. Fig.11 in which two C’s at positions −70 and −71 are radioactive, with a final nonradioactive G at −72. After end labeling, the probe was digested with BssHII (cutting in the polylinker beyond the HincII site), and the fragment was denatured and strand separated by gel electrophoresis. Probe 1 contains rDNA originally cloned as pBD4 (2). This sequence matches the rDNAs of strains W3031a and HI227 up to position +179, at which point there is a 24-bp deletion in pBD4.
Probe 2 was made in the same way as probe 1 except with rDNA from strain W3031a. Thus, probe 2 matches the rDNAs of both W3031a and HI227 from −72 to the HincII site at position +415.
The 284-bp XhoI-KpnI fragment from 105LT was subcloned into the polylinker of pBluescript. To make probe, the plasmid was opened with XhoI, 3′ end labeled by fill-in with Klenow DNA polymerase and radioactive deoxynucleoside triphosphate (dNTPs), and digested again with BssHII, and the labeled fragment was denatured and strand separated by gel electrophoresis.
In a typical experiment, 20 μg of total yeast RNA (DNase I treated) was mixed with 5 to 10 ng of single-stranded, 3′-end-labeled DNA probe, ethanol precipitated, and dissolved in 50 μl of 3 M sodium trichloroacetate–5 mM sodium EDTA–50 mM piperazine-N,N′-bis(2-ethanesulfonic acid) (pH 7). The mixture was denatured for 5 min at 65°C and then hybridized at 37°C overnight. For S1 nuclease digestion, hybrids were ethanol precipitated and dissolved in 200 μl of 1.5 mM ZnSO4–750 mM NaCl–30 mM sodium acetate (pH 5)–20 μg of Escherichia coli tRNA per ml–1,000 U of S1 nuclease per ml. Hybrids were digested for 30 min at 30°C, and digestion was stopped by addition of 20 μl of 200 mM sodium EDTA–2% sodium dodecyl sulfate. Hybrids were phenol extracted, ethanol precipitated, dissolved in 10 μl of deionized formamide containing marker dyes, and electrophoresed on denaturing gels containing 6.4% acrylamide and 8 M urea. Gels were dried onto a paper support and autoradiographed on X-ray film, and selected bands were quantitated with a PhosphorImager.
Specific initiation by PolI on the minigene 105LT was detected by using primer 1 (5′-CTCGAGGGTCTTGACGAAC-3′), a 19-nt oligonucleotide which specifically anneals to transcripts of the XhoI linker region and yields an extension product of 41 nt on correctly initiated RNA. In a typical assay, 5 to 20 μg of total yeast RNA was precipitated together with about 1.5 × 105 counts of each desired primer, dissolved in 7 μl of 90 mM Tris-HCl (pH 7.5)–12 mM KCl, denatured at 65° for 5 min, annealed at 42°C for 30 min, and placed on ice; 7 μl of a mixture containing 90 mM Tris-HCl (pH 7.5), 12 mM KCl, 7 mM dithiothreitol, 14 mM MgCl2, 180 mM dNTPs, 36 μg of actinomycin D per μl, and 100 U of Moloney murine leukemia virus reverse transcriptase was added, primers were extended for 45 min at 42°C, 5 μl of 30 mM sodium EDTA–600 mM sodium acetate (pH 5.2)–10 μg of RNase A per ml was added, and the mixture was incubated at 42°C for 10 min. Reactions were stopped by addition of 5 μl of 0.2% sodium dodecyl sulfate–600 mM sodium acetate–20 μg of proteinase K per ml and incubation at 65°C for 15 min. Nucleic acids were ethanol precipitated, dissolved in 10 μl of deionized formamide with marker dyes, and electrophoresed on 8% acrylamide–8 M urea denaturing gels.
Fifteen-microgram aliquots of total yeast RNA (DNase treated) were annealed separately with 100 pmol of primer P1 (5′-AACAAAGGCTTAATCTCAGC-3′), P2 (5′-AGAGACTTGTTCAGTCTACT-3′), P3 (5′-TGGTACACTCTTACACACTA-3′), or P4 (5′-AAAGCCCTTCTCTTTCAACC-3′), reverse transcribed into cDNA as described above, phenol extracted, and dissolved in 100 μl of water. For PCR amplification, 1 to 10 μl of each cDNA was mixed with 100 pmol of the same primer used for cDNA synthesis plus 100 pmol of 32P-labeled primer P5 (5′-CTAGCAACGGTGCACTTGG-3′), in a total volume of 100 μl containing 5 U of cloned Pfu polymerase (Stratagene), 20 nmol of each dNTP, 20 mM Tris-HCl (pH 8.8), 10 mM KCl, 10 mM (NH4)2S)4, 2 mM MgSO4, 0.1% Triton X-100, and 0.1 mg of bovine serum albumin per ml. Samples were layered with paraffin oil and subjected to amplification cycles consisting of 94°C for 1 min, 55°C for 1 min, and 72°C for 2 min. Aliquots of 10 μl were removed every five cycles for analysis by gel electrophoresis. For quantitative-competitive PCR, multiple reactions containing equal amounts of cDNA and various amounts of an internal standard DNA template (illustrated in Fig. Fig.3A)3A) were set up. Amplification was allowed to proceed until the earliest point at which products could be detected, and samples were then gel electrophoresed. Gels were dried, and band intensities were quantitated on a PhosphorImager.
For a typical RNA preparation, yeast cells were grown in 25 ml to an optical density of 0.5 to 1.0. NaN3 was added to 100 mM, and pelleted cells were frozen in liquid nitrogen for storage. For RNA isolation, a pellet of cells from a 25-ml culture was suspended in 0.5 ml of 10 mM Tris-HCl (pH 7.5)–10 mM EDTA–0.5% sodium dodecyl sulfate to which was added 0.5 ml of glass beads and 0.5 ml of water saturated phenol. The mixture was heated to 65°C, vortexed for 2 min and then incubated a further 60 min at 65°C. Samples were chilled on ice and centrifuged, and the aqueous layer was reextracted with 0.5 ml of phenol plus 0.5 ml of CHCl3. After ethanol precipitation, RNA was digested with DNase I (1 μg/ml) for 30 min at 37°C, phenol extracted, precipitated, and dissolved in water.
A common laboratory strain of yeast (W3031a) was grown to mid-log phase in rich medium, and total RNA was extracted. Aliquots of RNA were then hybridized with a single-stranded, end-labeled DNA probe which has a labeled 3′ terminus at −72 relative to the 3′ end of 25S rRNA and extends to +179, where its sequence diverges from the sequence of the rDNA in strain W3031a (probe 1; termini illustrated in Fig. Fig.1).1). The hybrids were digested with S1 nuclease, and protected probe fragments were displayed on a denaturing urea-acrylamide gel as shown in Fig. Fig.2A,2A, lane 1.
Probe protection due to the mature 3′ end of 25S rRNA is the predominant signal in lane 1, and we designate the leading edge of this cluster of bands as position −1 in the numbering scheme shown in Fig. Fig.1.1. The next most prominent bands are in a cluster located from about +72 to +93 beyond the 3′ end of 25S. A higher-resolution gel of the +72 to +93 cluster (Fig. (Fig.2B)2B) shows that the longest band corresponds to position +93. This is the same location as identified previously as the site of PolI termination at the Reb1p terminator by in vitro studies (6, 14). The shorter bands in the +72 to +93 cluster probably represent 3′ processing products of primary terminated RNA. 3′ processing of terminated transcripts has been observed in mammals (10) as well as in amphibians (12), and we have observed 3′ processing of terminated and released transcripts in yeast whole-cell extracts (16a). As shown below (Fig. (Fig.5),5), in vivo mutagenesis of the Reb1p terminator is consistent with the +93 band representing paused transcripts whereas the shorter transcripts may be released and 3′ processed transcripts. When quantitating termination at the +93 site, we routinely sum all of the bands in the +72 to +93 cluster and compare their intensities to the intensity of the cluster at −1.
In Fig. Fig.2A,2A, lane 1, the only significant band beyond the +93 cluster is a faint band located at +179. This is the point where the sequence of probe 1 diverges from the rDNA sequence of strain W3031a. The practical result of this divergence is that all transcripts which read through the Reb1p terminator are truncated at this point and collected into a single band. In addition, if any rDNA contamination survived the RNA purification, probe protection due to this contaminant could contribute to the +179 band. PhosphorImager quantitation of Fig. Fig.2A2A and similar experiments shows that the intensity of the 3′ 25S signal is 200- to 300-fold greater than the intensity of the +93 cluster of bands. The intensity of the +179 readthrough band ranges from 1 to 10% of the +93 signal in different experiments. We conclude from this experiment that RNA termini corresponding to the Reb1p terminator are readily detectable in wild-type yeast cells. Furthermore, this analysis detects 10% or less readthrough past the +93 site, consistent with it being the primary terminator of PolI transcription.
S1 nuclease analysis is useful for precisely locating RNA termini. However, accurate quantitation of PolI termination efficiency by this method is difficult due to the low amount of RNA extending to +93 in the presence of a much larger amount of 25S rRNA. We initially thought to circumvent the problem of excess 25S rRNA by use of a shorter DNA probe whose labeled 3′ end would be located beyond the 25S sequence. After several failures with such a probe, we realized that a large stable hairpin is present in the RNA in this region (this is the hairpin recognized by RNase III during the processing that forms mature 25S rRNA ). The labeled terminus of any practical short probe must compete for hybridization against the formation of a more stable RNA-RNA hybrid.
Ultimately we decided to use RT-PCR to measure PolI termination efficiency. As illustrated in Fig. Fig.3A,3A, we first synthesized four short primers, P1 to P4. An excess of each primer was hybridized to a separate, uniform aliquot of total yeast RNA and was then extended with reverse transcriptase to make cDNA. The relative abundance of cDNA extending from each primer position was then measured by PCR amplification using a radioactive common 5′ primer (P5) paired with the same 3′ primer originally used to make each cDNA. Parallel amplification reactions were run with each primer pair, and aliquots were removed every five cycles to measure the progress of amplification. As shown in Fig. Fig.3B,3B, the product of the P1-P5 primer pair, corresponding to the 3′ end of 25S rRNA, amplifies rapidly and uses up the radioactive P5 primer by 15 cycles. Amplification by the P2-P5 primer pair, corresponding to cDNA extending to +93, proceeds more slowly, consistent with +93 transcripts being several orders of magnitude less abundant than 25S rRNA. The P3-P5 primer pair (detecting cDNA from transcripts extending beyond +93) amplifies even more slowly, consistent with most transcripts stopping at +93. Finally, primer pair P4-P5 barely makes any product, even at 30 cycles. The small amount of product amplified by P4-P5 is about the same as is amplified from a mock cDNA reaction in which either the reverse transcriptase or the P4 primer was omitted. Thus, this small amount of product is probably due to residual DNA in the RNA preparation.
The RT-PCR data in Fig. Fig.3B3B are qualitatively in agreement with the S1 nuclease analysis shown in Fig. Fig.2A.2A. They confirm that transcripts extending to +93 are less abundant than 25S rRNA and that most transcripts do not extend beyond the +93 site. However, the more sensitive RT-PCR method verifies that there is detectable readthrough past the +93 site, and furthermore, this readthrough does not extend beyond +250. Beyond +250, the PCR method detects only a trace of DNA contamination that escaped DNase treatment during the initial RNA preparation.
To determine absolute transcript abundances, we used the same cDNAs and primer pairs to perform competitive PCR. As shown in Fig. Fig.3A,3A, we constructed a DNA internal standard which has a 75-bp insert of plasmid DNA situated so that it will be amplified by all primer pairs. Known amounts of this internal standard DNA were then mixed with each of the cDNAs, and both cDNA and internal standard were amplified with the appropriate primer pair in the same reaction. When the amplified signal obtained from both the cDNA and the internal standard were the same, we assumed that the amount of internal standard added matched the amount of cDNA (Fig. (Fig.3C).3C). In cases where the internal standard did not exactly match the cDNA, we interpolated between the two nearest internal standard amounts to estimate cDNA concentration. If we assumed that most of the stable RNA in a yeast cell is 17S and 25S rRNA and that reverse transcriptase synthesized a single cDNA molecule for each molecule of 25S rRNA, then we added 0.075 pmol of 25S cDNA to the P1-P5 amplication. As shown in Fig. Fig.3C,3C, we estimated by competitive PCR that about 0.15 pmol of 25S rRNA was in the reaction, a difference of only twofold from the first estimate.
The results of competitive PCR quantitation are shown in Table Table1.1. We estimated on the basis of S1 nuclease protection (Fig. (Fig.2A)2A) that 25S rRNA is about 300-fold more abundant than transcripts extending to +93, while the estimate by PCR is 3,000-fold difference. It is likely that the lower estimate by S1 analysis is due to the fact that the experiment was not done under conditions of probe excess and the PCR estimate is the correct estimate. Quantitative PCR agrees with S1 analysis that about 90% of transcripts stop at +93 with about 10% readthrough beyond this point. No transcripts extend beyond +250.
The low abundance of transcripts extending beyond the 3′ end of 25S rRNA makes it difficult to study their termini. Recently it has been shown that the stability of these transcripts can be greatly increased by inactivation of the RNT1 gene product, an RNase III which is involved in the processing steps that form the 3′ end of mature 25S rRNA (4). Figure Figure4A4A shows S1 nuclease protection analysis of RNA from strain HI227 (rnt1-47), a strain that carries a temperature-sensitive mutation in RNT1 (a generous gift from M. Ares). RNA was isolated from HI227 cells grown at the permissive temperature (24°C) and at intervals after shifted to the nonpermissive temperature (37°C). S1 protection was done with the same probe (probe 1) as used previously for Fig. Fig.2A.2A.
S1 analysis of RNA from HI227 grown at the permissive temperature (Fig. (Fig.4A,4A, lane 1) looks similar to that seen with RNA from wild-type W3031a analyzed at a similar probe-to-RNA ratio (Fig. (Fig.2A,2A, lane 1). In RNA from HI227, a cluster of bands is seen in the +72 to +93 region and the apparent readthrough signal at +179 is about 6% of the +72 to +93 cluster intensity. However, shifting HI227 to 37°C has a dramatic effect on 3′ end formation (Fig. (Fig.4A,4A, lanes 2 to 4) which is graphically summarized in Fig. Fig.4B.4B. The intensity of the 3′ 25S signal decreases, the intensity of the +72 to +93 cluster increases nearly to that of the 3′ 25S signal, and readthrough to the +179 divergence point also increases dramatically. Increased stability of the longer transcripts now makes it easy to detect events at the Reb1p terminator and beyond.
In contrast to the situation in a wild-type cell, inactivation of Rnt1p results in considerable readthrough at the Reb1p terminator. We used probe 2 (Fig. (Fig.1)1) to determine where these readthrough transcripts terminate. Probe 2 was derived by PCR from the genomic rDNA of strain W3031a and matches the rDNA sequences of both W3031a and HI227 to a HincII site at position +415 beyond the 3′ end of 25S rRNA. Probe 2 has additional vector sequence beyond the divergence point to allow full-length transcripts to be distinguished from undigested probe. Figure Figure4C4C shows an S1 protection assay in which probe 2 was hybridized with RNA from HI227 cells grown at 37° for 4.5 h. The S1 protection pattern shows bands corresponding to 3′ 25S and the +72 to +93 cluster similar to bands seen with probe 1 in the same RNA sample (Fig. (Fig.4A,4A, lane 3). In Fig. Fig.4C,4C, however, readthrough past the Reb1p terminator appears as a cluster of bands of which the longest maps to a position about +250 bp beyond the 3′ end of 25S. Quantitation of Fig. Fig.4C4C shows that no transcripts are detectable beyond the +250 site to the point where probe 2 diverges from rDNA sequence at +415. Therefore, +250 appears act as a completely efficient fail-safe terminator for transcripts that escape the Reb1p terminator.
In their original analysis of RNA synthesized after heat shock of a rnt1 temperature-sensitive strain, Elela et al. observed a disappearance of signal from 3′ 25S rRNA and saw no evidence of termini at +93 (4). Instead, they saw only the appearance of a cluster of bands mapping approximately to +250. We have been able to duplicate these results by using a small amount of very hot probe and doing the S1 nuclease analysis under conditions of large RNA excess (data not shown). These are conditions predicted by Lopata et al. (17) to favor formation of triple and higher-order hybrids in which the longest transcripts in the mixture suppress detection of shorter transcripts. When the amount of probe is increased the amount of RNA is decreased, termini corresponding to 3′ 25S and +93 are readily detected in RNA from rnt1 cells (Fig. (Fig.44).
We have previously assayed a series of mutants of the Reb1p terminator for their effects on PolI termination in vitro (15, 16). To test these same mutations in the living cell, we constructed a ribosomal minigene, 105LT (Fig. (Fig.5A).5A). 105LT has an entire intergenic spacer region, including a PolI promoter and enhancer, upstream of a 314-bp minigene. The minigene contains a unique XhoI site at +36 to allow use of primers to specifically monitor transcription initiation from the minigene promoter. The XhoI site also allows radioactive end labeling of a single-stranded DNA probe (probe 3) to measure RNA 3′ end formation on 105LT by S1 nuclease protection. The Reb1p termination site is 285 bp downstream of initiation, and the terminator consists of a 53-bp fragment of rDNA that was previously found to direct wild-type levels of termination in vitro (16). The 53-bp terminator fragment contains 23 bp of T-rich sequence upstream of termination plus 30 bp downstream (including the Reb1p binding site). Probe 3 contains 38 nt of additional length beyond the point where it diverges from the sequence of 105LT to allow readthrough transcripts to be distinguished from undigested probe.
105LT was transformed into W3031a on a 2μm yeast vector. Figure Figure5B,5B, lanes 2 and 3, shows the patterns of S1 protected bands obtained with RNA from two independent transformants. In each lane, there is a band at +247 that we interpret as the primary termination product. Extending down from +247 is a cluster of bands which presumably represent 3′ processing products of the primary terminated transcript. As an internal size standard, we have cut the plasmid used to make S1 probes with BfrI and transcribed the truncated DNA with T7 RNA polymerase. This is expected to yield an RNA that is 3 nt longer than the primary termination product that we observe in purified in vitro termination systems. The band, labeled +247, migrates detectably faster (lanes 2, 3, 5, and 6) than the band protected by the BfrI transcript (lane 8) and thus, within the resolution of this gel, maps to the location previously mapped in vitro as the primary site of Reb1p- directed termination.
Termination on 105LT is leaky. PhosphorImager analysis, summarized at the bottom of Fig. Fig.5B,5B, shows that about 33% of total transcripts read past the Reb1p terminator and appear as a band of 284 nt (lanes 2 and 3).
Previous in vitro work has identified three informative mutants of the Reb1p terminator. The sequence of these mutations is shown in Fig. Fig.5C.5C. Mutant 1 changes two G residues in the Reb1p binding site to T’s. This alteration eliminates Reb1p binding as assayed by gel shift analysis and concurrently eliminates PolI termination in vitro (15). When the same mutations are introduced into 105LT, they also abolish all termination in vivo with a corresponding increase in readthrough (Fig. (Fig.5B,5B, lane 4).
Mutant 2 changes three A residues in the Reb1p binding site to C’s. When assayed in vitro these changes increase the binding of Reb1p about 10-fold and also increase pausing and overall termination efficiency (15). Introduction of these changes into 105LT has a similar effect in vivo since it decreases readthrough from about 33% to about 13% of total transcripts (lane 5).
Mutant 3 changes three T residues to G’s within the upstream T-rich element of the Reb1p terminator. In vitro work indicates that this change inhibits transcript release while having little effect on pausing caused by Reb1p binding (16). It is a reasonable interpretation that the primary termination product at +247 represents transcripts that have paused but not released from the template. Conversely, protected bands shorter than +247 are likely the result of 3′ processing of released transcripts. At a wild-type terminator, the ratio of primary to processed products is 0.3 to 0.4 (Fig. (Fig.5B,5B, lanes 2 and 3). When the three G substitutions are introduced into the upstream T-rich element, however, this ratio changes dramatically to 1.7 (lane 6). This change in ratio is consistent with transcript release being impaired by mutant 3 in vivo, similar to its effect in vitro.
Analysis of these three mutant terminators supports the conclusion that termination directed by Reb1p proceeds in vivo by essentially the same mechanism as previously defined by test tube studies (21). Termination requires the Reb1p binding site, increasing the affinity of this site for Reb1p also increases termination efficiency, and introducing G residues into the T-rich upstream element impairs transcript release. Primer extension analysis (data not shown) reveals that none of the terminator mutants has any effect on transcription initiation by PolI.
We have shown in this report that in wild-type S. cerevisiae, most PolI transcription terminates at the Reb1p terminator 93 bp downstream of the end of mature 25S rRNA. Evidence that the +93 site is a terminator and not a processing site can be summarized as follows.
(i) In a crude whole-cell extract, where processing of the 3′ end of the 25S rRNA is observed, pulse-chase experiments show that the +93 site behaves as a terminator, not a processing site (15). In such whole-cell extracts Rnt1p must be functioning (since 25S processing is observed), but when transcription is stopped with actinomycin D, readthrough transcripts are not processed at the +93 site. This fits with the fact that there is no stem-loop structure near +93 for Rnt1p to recognize.
(ii) In a purified termination system, containing only PolI, recombinant Reb1p, and template, transcripts paused at +93 are precursors to released (terminated) transcripts. Transcripts that read through +93 are not precursors to anything (14).
(iii) As shown in this report (Fig. (Fig.5),5), the terminator at +93 follows the same rules in vivo as were previously defined in vitro. Termination is influenced both by the Reb1p binding site and by an upstream T-rich element. And increasing the affinity of the Reb1p binding site increasing termination efficiency and reduces readthrough. Thus, if the +93 site is a terminator in vitro, it is likely a terminator in vivo as well.
The +93 terminator has a structure and mechanism of action that has been conserved from yeasts to humans (21). Conservation of structure and function over such a large stretch of evolutionary time suggests that termination elements of this type play some important role in the cell’s physiology. Therefore, it is curious that S. cerevisiae maintains a termination element at +93 that appears to be less strong than it could be. Even in wild-type cells, a low but significant amount of transcription reads through the +93 site (Table (Table1)1) and the terminator is easily deranged to allow more readthrough. For example, placing the +93 terminator on an extrachromosomal plasmid increases readthrough to about one-third of all transcripts (Fig. (Fig.5C).5C). Moreover, damaging processing at the nearby 3′ 25S rRNA terminus induces readthrough even on the chromosomal rDNA (Fig. (Fig.4).4). In some situations, both in vitro (14, 15) and in vivo (Fig. (Fig.5C),5C), we have been able to eliminate nearly all of this readthrough simply by increasing the binding affinity of the Reb1p site. In fact, one copy of this higher-affinity Reb1p binding site is normally present in the chromosomal rDNA at position −215 upstream of the PolI promoter. It is a puzzle why S. cerevisiae maintains a relatively weak Reb1p binding site at the +93 terminator when a higher-affinity site is seemingly available.
Perhaps S. cerevisiae can get by with a relatively weak terminator at +93 because it has a backup, fail-safe terminator further downstream at +250. S1 nuclease analysis of RNA from rnt1-47 cells (Fig. (Fig.4C)4C) and RT-PCR of RNA from wild-type cells (Table (Table1)1) indicates that no transcripts get past this fail-safe site. The existence of terminators at both +93 as well as +250 helps explain some of the results reported in the literature.
Results of nuclear run-on experiments from two different laboratories (7, 23) agree that PolI density is negligible downstream of the EcoRI-HindIII fragment which spans positions +98 to +285 downstream of 25S rRNA (Fig. (Fig.1).1). These results are consistent with termination at either +93 or +250. The first attempts to map PolI termination sites more precisely (9, 23) studied transcription on extrachromosomal minigenes. Judging by our results with extrachromosomal plasmids (Fig. (Fig.5C),5C), they probably had considerable readthrough of the +93 site in those experiments. For reasons not entirely clear, these workers missed the +93 site (partly because they used a probe in later work that could not detect the +93 site), but they clearly detected the +250 site. A subsequent study (7), also performed with minigenes on plasmids, detected multiple termini but clearly identified both the +93 and +250 sites. The next report was our work examining 3′ end formation in whole-cell extracts where we identified the +93 and +250 sites plus a further downstream site that is apparently not utilized in vivo (15). An additional experiment using plasmid minigenes in vivo also showed formation of a discrete RNA 3′ end that was dependent on a functional Reb1p binding site at +93 (8). Finally, the work of Elela et al. (4) showed that inactivation of Rnt1p causes the appearance of transcripts extending to about +250, although the +93 site was missed due to a technical problem with the probe.
When PolI termination is examined in different species, it is noteworthy that leakiness of the primary PolI terminator (the first terminator encountered by the polymerase) is observed in every reported case. And without exception, this leakiness of the primary terminator is rescued by the presence of one or more backup, fail-safe terminators. For example, in the frog Xenopus laevis, the primary terminator (T2) has a point mutation that renders it leaky, and transcripts are detected across most of the intergenic spacer. However, a strong terminator (T3) located just upstream of the PolI promoter apparently prevents any of this leaky transcription from damaging the succeeding promoter initiation complex (11, 13). This promoter-proximal terminator is present in all vertebrate rDNAs that have been examined (20). In a closely related species of frog, X. borealis, the T2 terminator is slightly leaky but has a backup copy located shortly downstream to prevent readthrough (12). In another vertebrate, mouse rDNA has at least eight tandem copies of the PolI terminator located downstream of the 28S rRNA region (5). Finally, another species of yeast, Schizosaccharomyces pombe, contains a Reb1 protein closely related to S. cerevisiae Reb1p (24). There are three binding sites for S. pombe Reb1p downstream of the 25S rRNA region, and S1 analysis indicates that RNA 3′ end formation (presumably termination) occurs at each site. Considerable readthrough occurs at the first site, a small amount occurs at the second site, and no readthrough is detected at the third site (18, 19). Seen against this background, the arrangement of PolI terminators in S. cerevisiae is unremarkable and fits the pattern seen in other eukaryotes.
It has been proposed that processing at the 3′ end of 25S rRNA is cotranscriptional (1). Obviously this conclusion cannot be entirely correct or we could not detect 3′ ends at +93 using a probe end labeled within the 25S rRNA sequence (Fig. (Fig.2A,2A, lane 1), nor could we detect transcripts extending to +93, using a PCR primer located within the 25S rRNA sequence (Fig. (Fig.3A).3A). It will require a different experimental design to determine if all transcripts proceed to +93 prior to being processed or if a fraction are processed cotranscriptionally.
We thank M. Ares for the gift of strain HI227 and communication of unpublished results. We also thank B. Sollner-Webb for pointing out the importance of probe to RNA ratios in nuclease protection experiments.
This work was partially supported by GM41792 awarded to R.H.R. P.G. was the recipient of a Fogarty International Scholarship.