Definition of Transcriptional Pause Elements in Fission Yeast

Abstract

Downstream elements (DSEs) with transcriptional pausing activity play an important role in transcription termination of RNA polymerase II. We have defined two such DSEs in Schizosaccharomyces pombe, one for the ura4 gene and a new one in the 3′-end region of the nmt2 gene. Although these DSEs do not have sequence homology, both are orientation specific and are composed of multiple and redundant sequence elements that work together to achieve full pausing activity. Previous studies on the nmt1 and nmt2 genes revealed that transcription extends several kilobases past the genes’ poly(A) sites. We show that the insertion of either DSE immediately downstream of the nmt1 poly(A) site induces more immediate termination. nmt2 termination efficiency can be increased by moving the DSE closer to the poly(A) site. These results suggest that DSEs may be a common feature in yeast genes.

Transcriptional termination of eukaryotic RNA polymerase II (Pol II) is a basic gene expression process, but its mechanism remains unclear. Two main sequence elements are likely to be involved in Pol II termination: an upstream polyadenylation signal and a downstream element (DSE) involved in polymerase pausing.

Yeast polyadenylation signals are often degenerate and redundant (13, 16), but they still direct a 3′-end processing event in which the primary transcript is cleaved and then a poly(A) tail is added to the 3′-end product (8). Polyadenylation signals appear to direct transcription termination in yeast. In the budding yeast Saccharomyces cerevisiae, mutations in the polyadenylation signal of the CYC1 gene cause transcriptional readthrough into autonomously replicating sequences, resulting in plasmid instability (28, 29). Deletion of the GAL10 polyadenylation signal impairs termination, as observed by using the transcription run-on (TRO) technique, and thereby abolishes the function of the downstream GAL7 promoter (12). Also, by using TRO, it has been observed that mutations in yeast cleavage factors affect termination, establishing the first biochemical link between the two processes (6). In the ura4 gene of the fission yeast Schizosaccharomyces pombe, mutations in the polyadenylation signal that abolish in vivo processing (16) also affect transcription termination (5). However, in other yeast genes the link is not so clear. In the S. pombe nmt1 and nmt2 genes, transcription extends for several kilobases beyond the poly(A) site, despite the presence of efficient polyadenylation signals (14). The reverse situation is observed in some S. cerevisiae genes: deletions that affect 3′-end formation do not appear to prevent efficient termination in vitro in GAL7 (17) or in vivo in FBP1 (2). These results suggest that elements in addition to the polyadenylation signal, specifically DSEs, are involved in termination.

In higher eukaryotes mutations in the polyadenylation signal also lead to reduced termination efficiency (10, 24, 31). Several different types of DSE have been defined in mammals. All share the ability to pause the polymerase. Some of them are structural features of the transcript (26) or the DNA (20). In other cases factors bound to the DNA are involved (3, 4, 9). Pause elements have been identified in budding yeast by in vitro transcription experiments only on the GAL7 and ADH2 genes (17). This suggests that DSEs could also be a common feature in S. cerevisiae genes. Finally, an orientation-specific DSE has been identified in the fission yeast ura4 by using TRO (5).

The signals and factors involved in RNA Pol I termination are more completely characterized (reviewed in reference 27). Pol I transcription ends 10 to 20 bp upstream of the binding site of a sequence specific protein, called Reb1p in yeast and TTF-I in mice. This protein causes pausing of the polymerase (22) over a T-rich release element (19). Paused polymerase is then acted on by a release factor, PTRF (18), resulting in dissociation of the ternary complex. Reb1p is also able to pause yeast Pol II in vitro (22), and in this situation it is dissociated by the action of the heterologous prokaryotic termination factor rho (23). Indeed, a rho-like activity which functions as a Pol II release factor and has an ATPase activity has been purified from Drosophila (32). All these results suggest that a rho-like mechanism could be acting in Pol II termination to release polymerases paused at DSEs.

Few Pol II pause elements have been characterized in any detail, especially in yeast. The aim of this work was to identify a new DSE in the S. pombe nmt2 gene and to compare it with the ura4 DSE (5). To this end a novel DSE was defined in nmt2, and although this pause site lacks sequence homology with the ura4 DSE, both function in the same locations and are orientation specific. The ura4 pause site is more compact than the nmt2 DSE, but both are composed of multiple, degenerate sequence elements that act together to induce full polymerase pausing. These results indicate that pausing elements could be a common feature in fission yeast genes.

MATERIALS AND METHODS

Strains and media.

The S. pombe strains used were Sp204 (nmt1::ura4 leu1-32 ade6-704 ura4-D18 h⁺), Sp92 (nmt2::ura4 leu1-32 ade6-704 ura4-D18 h⁺) (both gifts of K. Maundrell), and Sp36 (leu1-32 ade6-704 ura4-D18 h⁺). All S. pombe media, growth conditions, and maintenance were in accordance with standard methods (1).

Oligonucleotides.

A list of the oligonucleotides used for cloning purposes or PCR is given in Table 1. In the oligonucleotide pairs designed to be annealed, the one carrying the complementary sequence is indicated as 3′.

TABLE 1.

List of oligonucleotides

Name	Location (gene and nucleotide positiona)	Sequence
3229	ura4, 1762–1732	GCTTGTGATATTGACGAAACTTTTTGACATC
6331	ura4, 1622–1607	ATGCTAGCATCATTACAAGTCTAA
6934	ura4, 1351–1368	TTTGGTTGGTTATTGAAA
DSE-C	ura4, 1661–1678	ATGTAAAATACCATGTAG
3′DSE-C	ura4, 1678–1661	CTACATGGTATTTTACAT
mut-A	ura4, 1661–1678	ATCATAAATACCATGTAG
3′mut-A	ura4, 1678–1661	CTACATGGTATTTATGAT
mut-B	ura4, 1661–1678	ATGTAATTAACCATGTAG
3′mut-B	ura4, 1678–1661	CTACATGGTTAATTACAT
mut-C	ura4, 1661–1678	ATGTAAAATAGGTTGTAG
3′mut-C	ura4, 1678–1661	CTACAACCTATTTTACAT
mut-D	ura4, 1661–1678	ATGTAAAATACCATCATG
3′mut-D	ura4, 1678–1661	CATGATGGTATTTTACAT
CATGTAG		AGTACTCATGTAGAATTC
3′CATGTAG		GAATTCTACATGAGTACT
REB1		AGGTAAGGGTAATGCAC
3′REB1		GTGCATTACCCTTACCT
3a	nmt2, 1653–1686	GTCATATTGATTGTATTAATTGCATTTAAACCGG
3′3a	nmt2, 1686–1653	CCGGTTTAAATGCAATTAATACAATCAATATGAC
3b	nmt2, 1687–1720	TTAAAAAAACTATTGATAGTATAATCGTAAGGAC
3′3b	nmt2, 1720–1687	GTCCTTACGATTATACTATCAATAGTTTTTTTAA
3c	nmt2, 1721–1754	ACATGCTTTAAATTGGATGCATCATATACATCTA
3′3c	nmt2, 1754–1721	TAGATGTATATGATGCATCCAATTTAAAGCATGT
3d	nmt2, 1679–1703	TAAACCGGTTAAAAAAACTATTGAT
3′3d	nmt2, 1703–1679	ATCAATAGTTTTTTTAACCGGTTTA
3e	nmt2, 1704–1727	AGTATAATCGTAAGGACACATGCT
3′3e	nmt2, 1727–1704	AGCATGTGTCCTTACGATTATACT
8759	nmt2, 1653–1673	AGCCTAGGTCATATTGATTGTATTAATTGC
8760	nmt2, 1219–1194	AGCCTAGGTCTTTACAAGTACTTTTCATTAACAG
9074	nmt2, 2141–2161	AGCCTAGGTTCAATGGATATGGCGAAGTCG
9094	nmt2, 2134–2109	TAAACCACCAATCGTCTGTATTAGC
9314	nmt2, 1217–1238	AGAAAATAGTGTACTAGCAACC
A34	nmt2, 1754–1735	TAGATGTATATGATGCATCC
A35	nmt2, 1755–1774	CTCTGATATTGCTGCGTATA
A36	nmt2, 1855–1836	GATGGCAGTTTCTTTATTTC
A37	nmt2, 1856–1875	GATTGACAAATCATTAATGC
A39	nmt2, 1954–1934	TTTGTGCTGAGTACGAAATG
A40	nmt2, 1955–1976	AAAATAGAATAAAAACTAACAG
AVN2-1	nmt2, 2240–2218	TGATTCACAAAACGCTCGTCTTG
AVN2-2	nmt2, 2241–2262	ATTTCTTGTCCTGTTGCATAAG
AVN2-3	nmt2, 2340–2320	AGGATGATGCTATGGAGAAGG
DELTA4	nmt2, 1468–1443	GATCTCGCTTACAGTTTACTCTAAAC
9921		GTAGGTGCTATTTTAGGCCTCGAGTATTTTACTAACTTCTTTTAG
9922		CTAAAAGAAGTTAGTAAAATACTCGAGGCCTAAAATAGCACCTAC

^aPosition +1 corresponds to the start of the HindIII site upstream of the ura4 and nmt2 ORFs. 

Plasmids and M13s.

All the cloning processes used the standard techniques described by Sambrook et al. (30). The M13 derivatives used as probes in the run-on experiments were previously described by Hansen et al. (14), except the probe called 5′ that contains a 116-bp AccI-XhoI fragment from the 5′ region of nmt1.

The PAC assay was described by Birse et al. (5). The PAC vector was cut with NheI and blunt ended, and all the various oligonucleotides, restriction fragments, and PCR products were inserted into it. Their orientation was confirmed by DNA sequencing. Fragment C of ura4 DSE and all its mutated versions, the nmt2 DSE fragments 3a through 3e, and the Reb1p binding site were obtained by annealing kinased oligonucleotides (sequences are given in Table 1). nmt2 fragments 1 and 2 were obtained by amplifying a bigger fragment by PCR by using oligonucleotides 9314 and 3′3a and then digesting with BglII and EcoRV. The blunt-ended BglII-EcoRV fragment is fragment 1. Fragment 2 extends from EcoRV to the end of the PCR fragment defined by oligonucleotide 3′3a. The next nmt2 fragments were obtained by PCR with the indicated pairs of oligonucleotides: fragment 3, 8759 and A34; fragment 4, A35 and A36; fragment 5, A37 and A39; fragment 6, A40 and 9094; fragment 7, 9074 and AVN2-1; and fragment 8, AVN2-2 and AVN2-3. Fragment 9 is a HindIII-XbaI blunt-ended fragment obtained from RF M13 DNA containing probe 8 (14). Fragment 10 was obtained in the same way from M13 nmt2 probe 9.

Some of the fragments already described were introduced into the unique StuI site of plasmid p41 (14), and their orientation was confirmed by DNA sequencing. p41+nmt2-SB was described by Hansen et al. (14).

p43B and p43+ura4 DSEs were constructed by Hansen et al. (14). p43+nmt2 DSE was obtained cloning nmt2 fragment 3 inside the PmlI site of p43B.

p2V was described by Hansen et al. (14). p2VΔ4 was obtained by deleting most of the area covered by probe 4 by using PCR. Thus, p2V was amplified with oligonucleotides 8760 and Delta4 and the product was self ligated. This introduces a unique AvrII site (contained in oligonucleotide 8760) that was used to introduce fragment C from ura4 DSE, creating p2VΔ4+C. The nmt2 polyadenylation signal was deleted by PCR with oligonucleotides 8759 and 8760, cutting the PCR product with AvrII (sites for this restriction enzyme are contained in both oligonucleotides), and religation. In the resulting plasmid, called p2VΔpA nmt2, the polyadenylation sequence was replaced by an AvrII restriction site. The ura4 polyadenylation signal was amplified with oligonucleotides 6934 and 6331 and cloned into the AvrII site of p2VΔpA, producing p2Vura4-pA. A longer portion of the ura4 3′ end region containing the whole polyadenylation signal and DSE was obtained by PCR with primers 6934 and 3299. This fragment was cloned into the AvrII site of p2VΔpA to create p2Vura4-pA+DSE.

The synthetic intron was obtained by annealing oligonucleotides 9921 and 9922. pUI was then created by inserting this intron into the StuI site of pURA4 (16). A minimal ura4 polyadenylation signal containing site-determining element 2 (SDE2) and an efficiency element (EE) was obtained by amplifying ura4 DNA with oligonucleotides 6934 and 3′ DSE-C. The resulting PCR fragment was digested with the restriction enzymes StyI and RsaI, and the 134-bp blunt-ended StyI-RsaI fragment was cloned into the StuI site of pUI, creating pUIpA. This plasmid was digested with XhoI and blunt-ended, and annealed oligonucleotides containing the ura4 DSE-C fragment, the nmt2 DSE fragment 3b, and the Reb1p binding site were introduced. All constructs were confirmed by DNA sequencing.

Transformation and RNA analysis.

S. pombe transformation was performed by the dimethyl sulfoxide-enhanced method (15), but with the addition of 20 μg of boiled salmon sperm carrier DNA together with the transforming plasmid. Total RNA was prepared by using a protocol for the isolation of RNA from ascospores (21). The method of Northern blot analysis was previously described (16). After hybridization, filters were washed twice in 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)–0.1% sodium dodecyl sulfate at room temperature for 20 min and then visualized and quantified with a Molecular Dynamics PhosphorImager.

TRO analysis was carried out as described previously (5, 14).

RESULTS

A small mutation abolishes the ura4 DSE pause activity.

We have employed the poly(A) competition (PAC) assay developed by Birse et al. (5) to identify and characterize transcriptional pause elements in S. pombe. In this assay (Fig. 1A), signals derived from the 3′ end of the ura4 gene are introduced into the nmt1 open reading frame. The ura4 polyadenylation signal contains two SDEs (SDE1 and -2) and an EE (16). A region immediately 3′ to the EE, the DSE, contains a strong pause site which is necessary to achieve complete transcription termination (5). All three polyadenylation elements were mutated in the PAC vector, and the 5′-end portion of the DSE was replaced by a NheI restriction site. Since these modifications weaken the ura4 3′-end formation signal, the wild-type ura4 poly(A) site is used in only 20% of the transcripts. Most transcripts use the downstream nmt1 poly(A) site, although some are processed by cryptic signals present in ura4 (PAC) (Fig. 1B). If a sequence that is cloned into the NheI site between the two polyadenylation signals works as a transcriptional pause element, it will increase the time that the weak upstream signal is available for processing. This will enhance the levels of 3′-end formation at the upstream polyadenylation site. The percent 3′-end formation was calculated as the amount of the smaller ura4 3′-end band relative to the total amount of RNA specific for the probe.

FIG. 1.

A small mutation abolishes the ura4 DSE pause activity. (A) Diagram of the PAC assay (5). The upper part shows the arrangement of polyadenylation signals and pause elements in the ura4 3′-end region. In PAC all three polyadenylation elements are mutated (grey boxes) and the most active part of the pause element (5′ DSE) is removed and replaced by a NheI site, where the fragment of interest (X) is cloned. The nmt1 disruption strain Sp204 is transformed with the resulting plasmid. The transcripts processed by ura4 signals are labelled as ura4 and the ones by the downstream nmt1 polyadenylation signal are referred to as nmt1. (B) Northern blot analysis. The sequence of the C fragment of the ura4 DSE and the sequence changed in every trinucleotide mutation (mut-A to mut-D) are shown in bold. Below them the sequence used to replace each trinucleotide is shown. The Reb1p binding site was obtained by annealing the oligonucleotides pAGGTAAGGGTAATGCACA and pGTGCATTACCCTTACCT. Ten micrograms of total RNA extracted from given constructs was separated on a 1.5% denaturing formaldehyde gel, transferred to a nylon membrane, and hybridized with an nmt1-specific 0.8-kb HindIII-StuI random primer-labelled probe. The percent 3′-end formation was calculated as the amount of the smaller ura4 3′-end band relative to the total amount of RNA specific for the probe. The variation observed in levels of total RNA between the different lanes reflects different levels of induction of the nmt1 promoter during cell growth. This does not affect the relative efficiency of ura4 poly(A) site usage.

Birse et al. (5) demonstrated that a small (18-bp) sequence within the 5′ DSE region called fragment C has full pausing activity, when cloned in PAC (DSE-C) (Fig. 1B). The sequence of fragment C contains two copies of the pentanucleotide ATGTA. To identify if this or other sequences have an important role in the activity of fragment C, mutated versions with three changed nucleotides were tested in the PAC assay (Fig. 1B). As indicated, mutations in the 5′ end and central portions of fragment C (called mutations A, B, and C) have quite strong effects on the levels of usage of the upstream ura4 poly(A) site (reducing them from 95% to an average of 40%). Strikingly, mutation of the last GTA trinucleotide (mutation D) abolishes fragment C function. However, the 3′-end region does not work by itself, as an oligonucleotide containing the last 7 bp of C (CATGTAG) surrounded by random sequence has very little pausing activity (28%) (Fig. 1B). We conclude that the whole fragment C is required to achieve a full pausing activity and that the last GTA trinucleotide is absolutely required but is not sufficient. The dramatic effect observed when just three nucleotides were mutated suggests that this trinucleotide could be part of the binding site of an unknown factor involved in the pausing activity of fragment C.

To validate the PAC assay using a well-characterized heterologous pause element, a 17-bp fragment containing the binding site of S. pombe Reb1 protein (25, 33) was cloned in both orientations. The homologous S. cerevisiae Reb1p is an essential component of the RNA Pol I termination mechanism (for a review, see reference 27). It blocks Pol I in vitro in an orientation-sensitive way and it is also able to block Pol II in vitro (22). The S. pombe Reb1p binding site has a partial activity in the PAC assay in the forward orientation (REB1-F) (43%) but has no activity in the reverse orientation (REB1-R) (11%), suggesting that Reb1p also blocks S. pombe Pol II in vivo and that this interaction is orientation dependent.

Identification of a novel DSE in nmt2.

The PAC assay was used to identify new downstream elements in a different S. pombe gene, nmt2. Transcription termination in this gene was previously analyzed by the TRO technique (14). In this procedure, permeabilized yeast cells are briefly incubated with radioactively labelled [α-³²P]UTP, which is incorporated into the nascent transcript. Total RNA is isolated, partially hydrolyzed, and hybridized to immobilized single-stranded M13 probes covering the region of interest. The signal for each probe is proportional to the average polymerase density across the DNA fragment. The profile of polymerases across the 3′-end region of chromosomal nmt2 and the convergently transcribed avn2 is shown in Fig. 2A (14). It is noticeable that there is a drop in polymerase density over the probe containing the poly(A) sites (probe 3) and over the next one (probe 4), but there is then an increase of polymerase density over probe 5, suggesting the presence of a pause element in that area.

FIG. 2.

Identification of a novel DSE in nmt2. (A) TRO profile obtained from the chromosomal copy of nmt2 (14). The nmt2 ORF is depicted in black and the avn2 ORF is in white. The nmt2 poly(A) [p(A)] sites are indicated by an arrow. The region covered by each TRO probe is indicated by vertical bars. Signals are corrected relative to the first probe and the length of each probe is reflected in the diagram. The fragments from the nmt2 3′ end tested in the PAC assay are indicated by bars and numbered from 1 to 10. The location of fragment 3 is indicated, with +1 designated as the HindIII site upstream of the nmt2 ORF, and the sequences of its central part (3b) and its subfragments (3d and 3e) are indicated. (B) Northern blot analysis, depicted as described for Fig. 1B.

Ten fragments extending from the 3′ flanking region of nmt2 immediately downstream of the polyadenylation signal up to the middle of the convergently transcribed gene avn2 were tested in the PAC vector (Fig. 2). Fragments 1 and 2 cover the area defined by probe 4. Fragment 1 does not stimulate the usage of the upstream ura4 poly(A) site and the upper band probably reflects a cryptic polyadenylation activity of the fragment introduced. It has been previously shown that probe 4 has some low-level polyadenylation activity (14). Fragment 2 covers the second half of probe 4 and has a good pausing activity (79%). Fragment 3 overlaps with fragment 2, covering the end of probe 4 and the beginning of probe 5. It has a very high activity (95%). The rest of probe 5 (fragment 4) and fragments 5 to 10 have little or no activity in this assay, despite the high polymerase density measured on some of them. These high TRO signals could reflect increased speed of the polymerase rather than polymerase accumulation. In previous experiments the region covered by probes 7 and 8 gave strong TRO signals when the pause region delimited by probes 5 and 6 was deleted (14). The results obtained here with PAC indicate that this region which also coincides with the 3′-end portion of avn2 does not have a very strong intrinsic pausing activity. The high polymerase density detected by TRO could reflect a stronger hybridization signal due to the higher CG content of coding sequences or increased speed of the polymerases.

We chose to analyze in greater detail fragment 3 since it has the major pausing activity in this intergenic region according to the PAC assay and coincides with a buildup in the density of polymerases based on the TRO analysis (14). It should be noted that fragments 2 and 3 overlap to some extent, although the common sequence (3a) has no activity alone (see below). The 102-bp fragment 3 was divided into three 34-bp subfragments (3a to 3c), which were then tested in the PAC assay. Only the central subfragment, 3b, has partial pausing activity (55%). Two new fragments containing half of the 3b fragment and some adjacent nucleotides (3d and 3e) were also tested to determine if the low activity of 3b (compared to that of the whole fragment 3) was due to the loss of critical sequence from the contiguous fragments 3a and 3c. Only 3e has some activity (34%), even less than 3b. In summary, these results indicate that the nmt2 DSE is less compact than that of ura4, requiring more elements for full activity, even though these elements cannot work alone. There is no clear sequence homology between fragment C of ura4 DSE and fragment 3b of nmt2 DSE, although the latter contains two copies of the trinucleotide GTA. Both fragments 3 and 3b of the nmt2 pause site were tested in the reverse orientation (3R and 3bR) to see if nmt2 DSE is orientation sensitive like ura4 (5). As indicated by the negative result (Fig. 2B), this pause element is substantially orientation specific.

A possible artifact of the PAC assay is that sequences defined as pause elements may actually act as polyadenylation signals that increase the usage of the upstream poly(A) sites. To control this possibility, nmt2 fragment 3, ura4 fragment C, and the Reb1p binding site were cloned in the forward orientation inside the StuI restriction site placed in the middle of the nmt1 open reading frame (ORF) located in plasmid p41 (Fig. 3A). The resulting constructs were used to transform the nmt1-defective strain Sp204. If the test fragment contains a polyadenylation signal, a shorter, truncated messenger is expected. This is the case when a 260-bp ScaI-BglII fragment containing the whole nmt2 polyadenylation signal (14) is cloned in p41 (Fig. 3B, lane p41+nmt2−SB). Only the truncated band is produced. If the inserted plasmid has no polyadenylation activity, transcription reaches the nmt1 polyadenylation signal, producing a readthrough transcript with the length of the nmt1 mRNA (Fig. 3B, lane p41) plus the length of the sequence introduced. This is the case in the construct which includes nmt2 fragment 3 (Fig. 3B, lane p41+nmt2-3), as well as the ura4 C element (lane p41+ura4-C) and the Reb1p binding site (lane p41+Reb1). The greater length of the 102-bp nmt2 fragment results in an RT product with slightly lower mobility. We conclude that these pause elements do not have any detectable polyadenylation activity.

FIG. 3.

nmt2 and ura4 DSEs have no intrinsic polyadenylation activity. (A) Diagram of the nmt1 gene in plasmid p41. Sequences of interest (X) are cloned in the unique StuI restriction site, and the nmt1 disruption strain Sp204 is transformed with the resulting constructs. Truncated transcripts (T) are due to 3′-end formation directed by the test fragment. Readthrough transcripts (RT) are due to 3′-end formation directed by the wild-type nmt1 poly(A) signals located downstream. (B) Northern blot analysis. The conditions and probe were as for Fig. 1B.

Study of the nmt2 and ura4 DSEs by TRO analysis.

To further characterize the nmt2 and ura4 DSEs, we studied their activities at the nascent level (by TRO analysis). First we obtained the polymerase density profile across some selected PAC constructs. One of the probes used (called 5′) is located at the 5′ region of nmt1, before the ura4 sequences present in the PAC vector (Fig. 4A). The rest of the probes (from 1 to 6) are located downstream of the ura4 sequences. The ratio between the upstream probe and the downstream ones indicates the ability of the sequence tested in the PAC assay to induce termination. In the PAC vector there is a high polymerase density in the downstream probes with a buildup of signal over the 3′ end of the nmt1 gene (probe 2) as well as extended transcription beyond the poly(A) signal. This indicates that the modified ura4 sequences do not induce efficient termination. In the construct containing the ura4 fragment C (DSE-C), the density of polymerases over the downstream probes is drastically reduced, indicating that this short element alone works together with the poly(A) signal to induce transcription termination. The mutation in fragment C called D (mut-D), which abolishes the function of this fragment at the steady-state level (Fig. 1B), gives a profile similar to that obtained with the PAC vector alone (Fig. 4A), indicating that its pausing activity is also impaired. Finally, nmt2 fragment 3 has a behavior at the nascent transcription level similar to that of ura4 fragment C. Thus, low polymerase density is observed over the downstream probes, which correlates well with the almost exclusive usage of the upstream poly(A) site in the steady-state studies (Fig. 2B). The fact that both ura4 and nmt2 DSEs activate the upstream poly(A) signal in the PAC assay as well as altering the polymerase elongation pattern in the TRO assay suggests that these elements act as true polymerase pause elements. This idea is reinforced by the fact that specific mutations in ura4 fragment C affect both processes.

FIG. 4.

Study of the nmt2 and ura4 DSEs by TRO analysis. (A) TRO analysis of some PAC-derived constructs (PAC, DSE-C, mut-D, and nmt2-3) transforming the nmt1-defective strain Sp204. The signals were corrected for the M13 background (M13mp18 without insert) and for the U content of the nascent RNA hybridizing to each probe and expressed relative to the signal for probe 5′. The black box indicates nmt1 coding sequence and the white box shows the ura4 sequences introduced in the PAC vector (Fig. 1A). p(A), poly(A). Probe 5′ location is indicated. The rest of the probes are depicted under their location across the gene. Their widths indicate the probe size. (B) TRO analysis of the plasmids p43B, p43+ura4 DSE (14), and p43+nmt2 DSE transforming the nmt1-defective strain Sp204. The signals were corrected for the M13 background (M13mp18 without insert) and for the U content of the nascent RNA hybridizing to each probe and expressed relative to the signal for probe 1. ura4 5′-DSE (p43+ura4 DSE) and nmt2 DSE fragment 3 (p43+nmt2 DSE) were cloned in the PmlI site 112 bp downstream of the nmt1 poly(A) site.

To test the ability of nmt2 DSE to promote termination in a different sequence environment and to compare both DSEs downstream from a heterologous polyadenylation signal, we used the plasmid p43B (14). p43B contains the whole nmt1 gene and sequences downstream (Fig. 4B). As for nmt2, transcription from nmt1 extends far beyond the poly(A) site, as shown by the TRO profile of plasmid p43B (14) (Fig. 4B). When the 134-bp ura4 5′ DSE was cloned in the PmlI restriction site downstream from the nmt1 poly(A) site (p43+ura4 DSE) (Fig. 4B), run-on signals dropped to low levels 1.23 kb beyond the poly(A) site (14). We therefore cloned the 102-bp nmt2 fragment 3 in the PmlI restriction site of the plasmid p43B (p43+nmt2 DSE), and the TRO profile obtained is shown in Fig. 4B. As indicated, the nmt2 pause site gave a result very similar to that obtained with the ura4 DSE. In all cases the nmt1 polyadenylation signal was fully active, producing a unique mRNA species (data not shown). These data demonstrate that both nmt2 and ura4 DSEs are defined sequence elements that can induce complete transcription termination when placed downstream of a heterologous polyadenylation signal in a similar way. This result indicates that they are functionally equivalent elements.

The nmt2 gene has a strong polyadenylation signal, as determined in a PAC assay using the p41 plasmid (Fig. 3) (14). We demonstrated in these studies that nmt2 also contains a strong DSE analogous to that of ura4. So why is the presence of both elements not enough to achieve complete transcription termination close to the polyadenylation site as observed for ura4 (5)? To address this question we attempted to modify the TRO profile of the nmt2 gene by making a variety of changes to the elements present in its 3′-end region. The plasmid p2V (14) containing the whole nmt2 gene and its derivatives were used to transform the nmt2-deficient strain Sp92, and TRO profiles were obtained (Fig. 5). In p2V (Fig. 5A) the usual TRO profile of lower density over probes 3 and 4 and polymerase accumulation over probes 5 and 6 was observed. Note that the density of polymerases over the avn2 ORF is lower in the plasmid than in the chromosomal copy (compare Fig. 2A with Fig. 5A), although there is still a high polymerase density across the whole intergenic region. In ura4 the DSE is immediately adjacent to the polyadenylation signal (5), while the nmt2 DSE (defined by fragment 3) is around 200 bp away. To test if distance could be the cause of the lack of interaction between these two elements, a 184-bp deletion covering most of probe 4 was made by PCR. The TRO profile of the resulting plasmid, p2VΔ4 (Fig. 5B), showed a low signal on probe 3 (which contains the polyadenylation site) and higher signals over the downstream probes as in plasmid p2V. However, the polymerase density on the three probes after the deletion (5 to 7) was lower than that of the wild-type gene. This result indicates that nmt2 termination efficiency can be increased by moving the DSE closer to the poly(A) site. To determine if the ura4 DSE could induce a more efficient termination in combination with the nmt2 polyadenylation signal, the ura4 DSE fragment C was cloned just downstream of the polyadenylation signal in the previously constructed p2VΔ4, producing p2VΔ4+C (Fig. 5C). This combination gave a profile similar to that of p2VΔ4, though with lower densities over probes 6 to 8. We conclude that the ura4 DSE does not induce the same degree of termination as in its normal ura4 gene environment. To test if the nmt2 polyadenylation signal is less effective than that of ura4 for termination purposes, the region covered by probes 3 and 4 [including all the nmt2 poly(A) signals] was deleted and replaced by ura4 poly(A) signals (p2Vura4pA) (Fig. 5D). The introduced ura4 polyadenylation signal is therefore adjacent to the nmt2 DSE, in an arrangement similar to that of the nmt2 polyadenylation signal and DSE in the plasmid p2VΔ4. The TRO profile observed is similar to that of p2VΔ4, indicating that the ura4 polyadenylation signal does not stimulate termination in combination with the nmt2 DSE any better than the nmt2 polyadenylation signal in the same environment. Finally the whole ura4 3′-end region containing the polyadenylation signal and the whole DSE was used to replace the nmt2 polyadenylation signal (p2Vura4pA+DSE) (Fig. 5E). In the wild-type ura4 gene and in a construct in which transcription starts from the nmt1 promoter, termination occurs within the fragment used in this last construct (5). Even so, in this nmt2 gene environment there is still a relatively high TRO signal on the downstream probes, giving a pattern similar to the one obtained with plasmid p2Vura4pA. This result indicates that there are additional features of the nmt2 gene environment that preclude efficient termination. In all cases the nmt2 and ura4 polyadenylation signals worked efficiently, producing a unique mRNA (data not shown). In summary, these experiments suggest that the distance between the polyadenylation signal and the DSE is important to improve Pol II transcription termination in nmt2. However there are unknown factors independent of the nature of DSE or polyadenylation signal that influence termination efficiency.

FIG. 5.

Manipulation of the nmt2 polymerase profile. Schematic diagrams of each of the constructs based on plasmid p2V used to transform the nmt2 disruption strain Sp92. The nmt2 ORF is shown in black and the convergently transcribed avn2 ORF is shown in white. p(A), poly(A). The grey box shows the location of nmt2 DSE fragment 3. In panel B the deleted nmt2 sequence is indicated by dotted lines. In panels D and E the inserted ura4 poly(A) signal is indicated by an open box and the ura4 DSE is shown by a black box. Underneath each diagram the TRO hybridization signals are shown approximately aligned with each probe location. Probes are the same as the ones showed in Fig. 2A. Only for the wild-type plasmid p2V (A) are the thirteen probes shown; for the rest of the constructs (B through E) just the first eight probes are shown. LEU2 indicates transcription levels of the LEU2 marker from the plasmid, and M13 indicates the background hybridization from a single-stranded phage without an insert. For each construct, transcription levels of probes 2 to 8 are corrected for the M13 background signals and the U content and plotted with respect to probe 2.

A reporter gene designed to identify mutations in 3′-end formation.

We finally used these well-characterized transcription pause signals to develop a reporter gene suitable to identify factors involved in the processes of transcriptional pausing or termination. The first step was to choose an intronless gene whose inactivation has a strong phenotypic effect. The ura4 gene itself is a good candidate. By using a unique blunt-ended StuI restriction site in the coding sequence of the ura4 gene in pURA4 (16), a synthetic intron that produces a functional mRNA following splicing was inserted. The synthetic intron was designed by combining the 5′-end region of a previously described synthetic intron (11) and the 3′ end of the long intron present in bsw1 (7) separated by StuI and XhoI restriction sites (Fig. 6A). This intron was then inserted into the StuI site of pURA4 to create pUI. This plasmid and its derivatives were introduced into S. pombe Sp36, which carries a deletion of the chromosomal ura4 gene. The ura4 mRNA produced by pUI has the same mobility as pURA4, as determined by Northern analysis (Fig. 6B), although at this level of resolution we cannot distinguish between spliced and unspliced transcripts. A minimal ura4 polyadenylation signal containing SDE2 and the EE (16) was cloned in the StuI site inside the intron to create pUIpA. If the polyadenylation signal inside the intron is recognized by the processing machinery, then a truncated transcript should be produced (Fig. 6A). If this is not the case then transcription would extend to the downstream ura4 poly(A) signal, producing an unspliced mRNA that may be processed to the wild-type ura4 mRNA. Only the latter transcript is able to give a functional ura4 product, making the ura4 mutant strain able to grow in media lacking uracil. In pUIpA all three bands are produced (Fig. 6B). The inserted ura4 polyadenylation signal had little activity in this intronic environment and most of the transcripts used the downstream polyadenylation signal. Although splicing is slightly inhibited, as indicated by the presence of the unspliced band, most transcripts are correctly spliced and the transformed strain can grow without uracil. To determine if the DSEs from ura4 and nmt2 increase the usage of this minimal polyadenylation signal, both were cloned in the XhoI site of pUIpA in either orientation. Fragment C from ura4 was cloned in forward orientation to create plasmid pUIpA+C−F. The presence of the pause element greatly increases the usage of the upstream polyadenylation signal, producing only truncated mRNAs (Fig. 6B). As expected growth in the absence of uracil is dramatically impaired. Only very small colonies appear after long incubation. The same element in reverse orientation (pUIpA+C−R) (Fig. 6B) does not enhance the usage of the minimal poly(A) signal to the same extent, resulting in the presence of longer transcripts that are mainly unspliced. Small amounts of spliced mRNAs can be seen at long exposures which allow the cells to grow without uracil. The 34-bp fragment 3b from nmt2 also had a full pausing activity in this new location in the forward orientation, producing only truncated transcript and impaired growth without uracil (pUIpA+3b−F) (Fig. 6B). It also showed orientation-specific activity like the ura4 DSE. Although the mRNA produced by the plasmids transformed with the plasmid pUIpA+3b−R showed an abundant transcript of apparently the same size as that of the wild-type mRNA, the presence of this sequence results in aberrant splicing, since very little growth is observed in media without uracil (Fig. 6B). Finally, the Reb1p binding site has a similar orientation-specific pattern. pUIpA+Reb−F contains this sequence in the forward orientation and results in an increase in usage of the minimal polyadenylation signal but is somewhat less effective than the Pol II pause elements (Fig. 6B). pUIpA+Reb−R contains the Reb1p binding sequence in reverse orientation and shows a much weaker effect. There is growth without uracil in both cases.

FIG. 6.

Reporter gene designed to identify mutations in 3′-end formation. (A) Diagram of the intron-based reporter gene. The sequence of the artificial intron cloned into the StuI site inside the ura4 ORF is shown. The splicing consensus sequences are shown in bold type. The ura4 ORF is represented by white boxes, and the intron is shown by the line that joins both boxes. Inside the StuI site a polyadenylation signal containing the SDE2 and EE polyadenylation sequences from ura4 was introduced (grey box). The arrowhead marks the polyadenylation (pA) site present in SDE2. The DSE sequences to be tested are depicted as a striped box inserted into the XhoI site. Those constructs were used to transform ura4 deletion strain Sp36. 3′-end processing inside the intron produces a truncated (T) transcript, while the nonprocessed transcript uses the downstream wild-type ura4 polyadenylation signal, producing an unspliced mRNA (US) that may be processed to produce the mature wild-type (WT) ura4 mRNA. This is the only mRNA able to produce the active ura4 product and to enable the cells to growth in a minimal medium without uracil (URA+). (B) Northern blot analysis. Total RNA from the depicted constructs was extracted, separated in a 1.5% formaldehyde gel, transferred to nylon, and hybridized with a ura4-specific 1.2-kb HincII-EcoRV random primer-labelled probe. The ability (+) or inability (−) to grow at a normal rate in a medium without uracil is indicated for each transformed strain.

The pUIpA vector results indicate that ura4 and nmt2 DSEs work in the same orientation-specific way when placed in quite different sequence environments. The ability of the nmt2 and ura4 pause sites to confer full activity on the 3′-end formation signals when placed in the intronic location (pUIpA+C−F and pUIpA+3b−F) provides a clear phenotype (no growth without uracil), especially when the reporter gene is integrated in a single copy in the genome (data not shown). This phenotype may make these vectors a powerful tool to identify mutations in the signals or elements involved in 3′-end formation. Revertants able to grow in a medium without uracil should have a defect in the process of recognizing the minimal ura4 polyadenylation signal. Although mutations in the polyadenylation signal and factors could produce increased readthrough, it is also likely that factors acting at the DSE level could be identified as well. Such mutations may provide new information about the elusive mechanism of RNA Pol II pausing and/or termination. Mutants with mutations in factors involved in 3′-end formation isolated by using this vector are currently being characterized.

DISCUSSION

We have identified a novel DSE in the 3′-end region of the nmt2 gene using a PAC assay. It is a defined element that works in a variety of sequence environments. It improves transcription termination of nmt1 when placed downstream of its poly(A) site, like the ura4 DSE (14). The nmt2 DSE causes an accumulation of polymerases before the actual site of termination, confirming its role as a transcriptional pause element. As a final confirmation, it also works in a different PAC assay designed to identify 3′-end formation mutants. The nmt2 DSE, like that of ura4 (5), is orientation specific. The fact that both DSEs work in several different locations in the same manner argues that they may be different versions of the same element. However, there is no clear sequence homology between them. This is not surprising considering the degenerate and elusive nature of yeast polyadenylation signals (13). The mutational study of fragment C from the ura4 DSE shows that its 3′-end region is absolutely required, although it does not work alone. To achieve full activity, cooperation of the rest of the fragment is required. The nmt2 DSE is a less compact element than that of ura4 but shows similar behavior: there are sequences that have no activity by themselves but that are required for full pausing activity. Therefore, like polyadenylation signals, multiple and redundant sequence elements act together to achieve full activity. The core of one of these elements could be the GTA trinucleotide identified by mutagenesis in fragment C. The presence of a similar DSE in nmt2 indicates that these signals could be a general feature in S. pombe genes.

The fact that transcription termination is produced far away from the nmt2 poly(A) site even though this gene has a strong polyadenylation signal and DSE is intriguing. Multiple factors, some described in this work and some still unknown, appear to be involved. The distance between the two elements is important. We have demonstrated that termination can be improved by placing the DSE closer to the polyadenylation signal. However, the nature of the polyadenylation signal and DSE is not critical, since those elements are interchangeable between genes. Thus, combinations of nmt2 polyadenylation signal plus the ura4 DSE (p2VΔ4+C) and of the ura4 polyadenylation signal plus the nmt2 DSE (p2Vura4pA) do not significantly improve transcription termination compared to the wild-type situation. Finally, when the nmt2 polyadenylation signal is replaced by an ura4 fragment that contains all the polyadenylation signal and DSE required for full transcription termination in its wild-type location (5), there is still no complete termination. This argues that there are additional factors in the nmt2 environment that preclude efficient termination. Overall chromatin structure or different subunit composition of the polymerase holoenzyme initiating on the nmt2 promoter may account for these differences.

Pol II transcriptional termination is a poorly understood process. There is a clear link between it and the 3′-end formation process, since polyadenylation signals are required for termination (31) and their strength correlates with termination efficiency (10). Recently it has been demonstrated that cleavage factors are required for termination (6). The best understood termination mechanism of a eukaryotic polymerase is that for Pol I (for a review, see reference 27). Pol I termination signals consist of a binding site for a DNA binding protein and an upstream release element. The DNA binding protein in S. cerevisiae is called Reb1p, and its function is to pause the polymerase by blocking its advance. This activity is orientation sensitive, so that the binding site does not work when reversed. It is also able to stop yeast Pol II and III in vitro (22), suggesting that a similar pausing mechanism could be operating in the genes transcribed by Pol II. Some cases of DNA binding proteins bound to DSEs that block mammalian Pol II in vivo have been described (3, 4, 9). The fact that the binding sequence of the S. pombe Reb1p increases the usage of the upstream weak polyadenylation signal in both PAC assays and is orientation specific suggests that a mechanism of this type could also be operating for S. pombe Pol II in vivo. Unfortunately no DNA binding proteins that bind to the ura4 DSE (fragment C) or the nmt2 DSE (fragment 3b) have been detected. We therefore cannot rule out the possibility that pausing activity works at the RNA level. This might also explain the orientation specificity of the DSEs and their degenerate nature.

The second step of Pol I termination involves a release activity that dissociates the paused ternary transcription complex. This activity has been identified in mouse and it has been demonstrated that this factor, called PTRF, interacts not only with Pol I and the Reb1p homologue TTF-I but also with the region in the 3′ end of the transcript that corresponds to the release element (18). It has also been shown by in vitro experiments that yeast Pol II paused by Reb1p can be released by the prokaryotic terminator factor rho (23). This result suggests that a similar mechanism could be used by eukaryotic Pol II. Indeed, a factor that releases RNA Pol II transcripts in an ATP-dependent manner has been purified from Drosophila (32). It is tempting to speculate that the release machinery in Pol II may be related to the 3′-end formation event and thus, the polyadenylation signal or the cleavage event may be a starting point for a hypothetical Pol II release factor.

The results presented in this paper describe the development of a reporter gene designed to search for mutations that affect 3′-end formation dependent on DSEs. Hopefully we will be able to employ this system to identify factors that act at the level of transcriptional pausing. Together with polyadenylation factors these pausing activities mediate transcriptional termination of RNA Pol II.