Mutations of RNA polymerase II activate key genes of the nucleoside triphosphate biosynthetic pathways.

The yeast URA2 gene, encoding the rate-limiting enzyme of UTP biosynthesis, is transcriptionally activated by UTP shortage. In contrast to other genes of the UTP pathway, this activation is not governed by the Ppr1 activator. Moreover, it is not due to an increased recruitment of RNA polymerase II at the URA2 promoter, but to its much more effective progression beyond the URA2 mRNA start site(s). Regulatory mutants constitutively expressing URA2 resulted from cis-acting deletions upstream of the transcription initiator region, or from amino-acid replacements altering the RNA polymerase II Switch 1 loop domain, such as rpb1-L1397S. These two mutation classes allowed RNA polymerase to progress downstream of the URA2 mRNA start site(s). rpb1-L1397S had similar effects on IMD2 (IMP dehydrogenase) and URA8 (CTP synthase), and thus specifically activated the rate-limiting steps of UTP, GTP and CTP biosynthesis. These data suggest that the Switch 1 loop of RNA polymerase II, located at the downstream end of the transcription bubble, may operate as a specific sensor of the nucleoside triphosphates available for transcription.

Introduction

DNA transcription depends on the availability of pyrimidine (UTP and CTP) and purine (GTP and ATP) ribonucleoside triphosphate substrates, and their de novo synthesis is subjected to tight homoeostatic controls. In yeast, the first two steps of UTP biosynthesis are catalysed by Ura2, a bifunctional protein endowed with carbamoyl phosphate synthetase and aspartate transcarbamoylase activities (Potier ). These are the main rate-limiting steps of the pathway, and their allosteric control by UTP is well documented (Lacroute ; Serre ). Similar to other genes of the UTP biosynthetic pathway, URA2 transcription is induced in the presence of 6-azauracil, which depletes cells in UTP (Exinger and Lacroute, 1992). A previous study (Losson and Lacroute, 1981) has suggested that this does not depend on Ppr1, a transcriptional activator specific for the pyrimidine biosynthetic pathway (Loison ; Losson and Lacroute, 1981; Flynn and Reece, 1999), as ppr1-2, a mutation unable to activate URA1 or URA3, had no effect on Ura2 activity in cell-free extracts. Moreover, we here show that a ppr1Δ null allele remains fully competent for URA2 transcription.

In this study, two classes of mutations were found to constitutively activate pURA2∷HIS3 or pURA2∷LacZ reporter plasmids. The first class was due to short deletions immediately upstream of the URA2 mRNA 5′-ends, indicating that this region acts as negative regulatory element of URA2 transcription. A second class altered RPB1, which encodes the largest subunit of RNA polymerase II, and specifically modified the Switch 1 loop of the active site. One of these mutations, rpb1-L1397S, was investigated in more detail. It showed a genome-wide reduced occupancy of RNA polymerase II, consistent with its partial growth defect, but specifically activated URA2, IMD2/IMD3 and URA8 (encoding rate-limiting steps of the UTP, GTP and CTP biosynthetic pathways, respectively), thus raising the intriguing possibility that the Switch 1 loop of RNA polymerase II might act as a sensor of nucleoside triphosphate depletion.

Results

UTP depletion activates URA2 independently of the Ppr1 activator

The URA2 open reading frame is separated by 1231 nt from the stop codon of the upstream gene TRK1 (Figure 1A). A KpnI–BamHI cassette bearing this DNA was cloned in frame with the LacZ or HIS3 open reading frames, generating the pURA2∷LacZ reporter plasmids pFL80 and pFL80-H2, and the pURA2∷HIS3 reporter pFL81 (Materials and methods). There was a fivefold LacZ activation in wild-type cells grown under repressing (uracil) or derepressing (6-azauracil) conditions (Figure 1B), and a similar range of activation was observed in the steady-state level of URA2 mRNAs (Figure 1C), indicating that the pURA2∷LacZ reporter correctly reflects the transcriptional regulation of URA2.

Figure 1

Transcriptional regulation of URA2 in response to UTP shortage. (A) Schematic view of the TRK1–URA2 intergenic region. The corresponding DNA was cloned as a KpnI–BamHI cassette to generate the pFL80 and pFL80-H2 pURA2∷LacZ reporter plasmids. A box denotes the 5′-UTR intron. The broken arrow corresponds to the URA2 mRNAs 5′-ends. (B) β-galactosidase activities of BMA64-URA3 (WT) and FL55 (ura8-ts ura7Δ) grown without uracil (CSM) at 30 and 34 °C, in the presence of uracil (CSM+U), mycophenolate (CSM+MPA) and 6-azauracil (CSM+6AU). (C) Three independent cultures of BMA64-URA3 (WT) and FL53 (ppr1Δ) were exponentially grown without (−URA) or with uracil (+URA), or with 200 mg/l of 6-azauracil (+6AU). URA1 and URA2 mRNAs were quantified by RT–PCR, and RNA polymerase II occupancy was determined by ChIP assays, using anti-CTD antibodies (8WG16). A schematic map represents the URA1 gene, with the oligonucleotide pairs used as primers. Position 1 corresponds to the start codon.

Adding 6-azauracil blocks UTP synthesis and also reduces the cellular pool of GTP (Exinger and Lacroute, 1992). However, no pURA2∷LacZ activation was observed in the presence of mycophenolate, which specifically blocks IMP dehydrogenase (Shaw ), and thus only depletes GTP. Likewise, counteracting GTP inhibition by adding guanine to the CSM+6AU medium did not reduce the activating effect of 6-azauracil (data not shown). Finally, impairing the metabolic conversion of UTP to CTP in temperature-sensitive mutant (ura8-ts ura7Δ) reduced the expression of pURA2∷LacZ, which presumably reflects the accumulation of UTP under these conditions and implies that URA2 does not respond to CTP shortage. These data indicated that URA2 transcription is primarily activated by a shortage in UTP, with little or no effect of GTP or CTP.

Other genes of the pyrimidine pathway, such as URA1 and URA3, are also derepressed under UTP depletion, and their activation is lost in ppr1-2, a mutation that alters the Ppr1 activator but has no effect on the Ura2 activity in cell-free extracts (Losson and Lacroute, 1981). Consistent with this observation, we found here that the ppr1Δ null allele strongly reduced the activation of URA1 and URA3 in the presence of 6-azauracil, as measured by RT–PCR quantification of their mRNAs and by their RNA polymerase II occupancy in chromatin-immunoprecipitation (ChIP) assays (shown in Figure 1C for the URA1 gene). In contrast, ppr1Δ actually increased the level of URA2 mRNA produced in the presence of 6-azauracil, which may reflect a particularly effective depletion in UTP, due to its inability to activate URA1 and URA3.

Organisation of the URA2 promoter region

It was previously thought that URA2 transcription starts some 70 nt ahead of the ATG start codon (Potier ). This is clearly inconsistent with the recent report of an untranslated intron between positions −385 and −66 relatively to the ATG (Juneau ). As shown in Figure 2, we confirmed the existence of this intron, although with different 3′-ends corresponding to positions −63, −54 or, in one case, +12. The corresponding DNA has a 5′-end consensus (−384GUAUGU−379), a canonical branch point box (−92UACUAAC−84), and a −68UAG−65 at the 3′-splice site. An intron deletion did not detectably affect the expression of pURA2∷LacZ (bottom line of Figure 2A).

Figure 2

Organisation of the URA2 promoter region. (A) Deletion mutagenesis of promoter region. LacZ expression in BMA64-URA3 (WT) transformed with pFL80 (pURA2∷LacZ), pFL80-KBΔ (KpnI–BsrGI deletion), pFL80-BSΔ (BsrGI–SalI deletion) and pFL80-INTΔ (no intron). Strains were exponentially grown without (−URA) or with uracil (+URA), or with 200 mg/l of 6-azauracil (+6AU). β-galactosidase was expressed in arbitrary units (Miller, 1972), where 1.0 is the level measured in wild-type cells grown in CSM. Broken arrows correspond to the transcriptional initiator regions, shown in black for full-length URA2 mRNA and in grey for short upstream untranslated RNAs (Thiebaut ). A canonical TATAAA box and a cluster of TCTT (UCUU) motifs (stars) are also indicated (see Supplementary data). (B) Identification of the URA2 5′-end region and of an URA2 intron. Thirteen cDNAs primed from an oligonucleotide corresponding to positions 167/188 of the URA2 open reading frame were amplified by the RNA Ligase Mediated-RACE technique (Materials and methods). The six URA2 start sites (underlined) fall between positions −589 and −563, containing the unique SalI site (gtcgac). Nine cDNAs were extracted and amplified from wild-type cells (BMA64-URA3) grown without uracil (−URA), or with 200 mg/l of 6-azauracil (+6AU), and four others were prepared from FL-M9 (rpb1-L1397S) grown without uracil (−URA). This also confirmed the existence of an intron between positions −385 and −66 (Juneau ), but with 3′-borders corresponding to positions −63, −54 or, in one case, +12.

Using RNA ligase mediated-RACE assays, we obtained and sequenced 13 cDNAs that, by construction, extended to position +188 of the URA2 open reading frame and thus corresponded to the 5′-end domains of full-length mRNAs (Figure 2B). These 5′-ends defined six positions (−588A, −579T, −576A, −571A, −565A and −563A) clustered within an A-rich domain, matching to the initiator consensus (Arich)5N(C/T)A(A/T)NN(Arich)6 deduced from a recent survey of yeast mRNAs (Zhang and Dietrich, 2005). A parallel study (Thiebaut ) has identified a second initiator region (positions −686/−657, denoted by a grey arrow in Figure 2A) upstream of the one shown here to be used for full-length URA2 mRNAs. This upstream initiator produces short unstable non-coding RNAs with the same transcriptional orientation as URA2. It is evidently not used to produce full-length URA2 transcripts, as the predicted 5′-ends were not found in the corresponding cDNAs (Figure 2B). Five TCTT (UCUU) boxes, present immediately upstream of the URA2 initiator, form a cluster of binding sites recognised by Nab3, which belongs to the transcriptional terminator system operating on non-coding intergenic RNAs (Arigo ; Thiebaut ).

The DNA upstream of the URA2 initiator prevents the progression of RNA polymerase II

Partial deletions of the TRK1-URA2 intergenic region were introduced in pURA2∷LacZ reporters (pFL80 and pFL80-H2) and tested for LacZ activation under repressing (uracil) and derepressing (6-azauracil) conditions. Deleting the first 297 nt had no effect on pURA2∷LacZ, but a BsrGI/SalI deletion lacking positions −934/−580 almost entirely blocked pURA2∷LacZ expression (Figure 2A). The corresponding DNA is therefore critical for URA2 transcription. It contains a TATAAA box (−776/−771) shown by Thiebaut to be needed for the synthesis of the URA2 mRNA and of its short-lived upstream transcripts.

The BsrGI/SalI DNA was submitted to error-prone amplification and recombined into the pURA2∷HIS3 reporter (pFL83), producing plasmids that were selected for HIS3 overexpression in the presence of uracil. No single-base mutation was isolated by this approach, but we obtained three very similar deletions up2Δ (−647/−577), up6Δ (−634/−579) or up10Δ (−650/−579) upstream of the SalI site (Figure 3A). These deletions overexpressed LacZ when subcloned into pURA2∷LacZ reporters (pFL80-H2 and pFL80), and a chromosomal up2Δ mutant (FL51) constitutively expressed the URA2 mRNA under-repressing conditions (Figure 3B). Moreover, up2Δ only acted in cis, as the promoter reporter (pURA2∷LacZ) was not activated in the up2Δ host (data not shown).

Figure 3

The DNA upstream of the transcription start impairs URA2 expression. (A) Schematic organisation of the upΔ mutations. Thick black line corresponds to the deleted DNA up2Δ, up6Δ and up10Δ. Stars indicate TCTT (UCUU) boxes. Broken arrows correspond to the transcriptional initiator regions as shown in Figure 2A. (B) Effect of up2Δ on pURA2∷LacZ expression and URA2 mRNA steady-state level. BMA64-URA3 (WT) was transformed with pFL80 or pFL80-up2Δ. β-galactosidase was tested as shown in Figure 2A. URA2 mRNA is expressed in arbitrary units (Miller, 1972), where 1.0 corresponds to wild-type cells grown in the absence of uracil (−URA). RT–PCR assays are based on three independent cultures of BMA64-URA3 (WT) and FL51 (up2Δ), grown with (+URA) or without uracil (−URA), or containing 200 mg/l of 6-azauracil (+6AU).

The ChIP assays of Figure 4A show that the region upstream of the URA2 start site is occupied by TFIIB and TFIIH, two components of the RNA polymerase II pre-initiation complex. The corresponding RNA polymerase II signal was recognised by anti-CTD antibodies, and also by an anti-Rpb3∷HA tag (see Figure 7D below), but not by antibodies raised against Ser2-phosphorylated CTD (Figure 4B). The CTD-Ser2P signal was only detected downstream of the URA2 initiator region. As shown in Figure 4C, this also coincided with a drop in RNA polymerase II occupancy, observed in wild-type or ppr1Δ cells grown under repressing conditions.

Figure 4

Distribution of TFIIB, TFIIH and RNA polymerase II at the URA2 locus. (A) TFIIH (Rad3-TAP) and TFIIB (Sua7-TAP) were detected by ChIP assays as described in Materials and methods. Strains D712-10C (WT RAD3∷TAP) and D714-5D (WT SUA7∷TAP) were grown on SD+aa. The schematic organisation of the corresponding DNA is presented below. Broken arrows represent the URA2 mRNA start sites and the upstream initiator, as in Figure 2A. (B) RNA polymerase II was immunoprecipitated with Dynabeads anti-mouse (Dynal Biotech), using anti-CTD (8WG16), anti-Ser2P and anti-Ser5P antibodies (Covance). Strain GR44-11C (WT) was grown on SD+aa. A schematic map indicates the oligonucleotides used as primers. (C) RNA polymerase II occupancy in wild-type, ppr1Δ and up2Δ. Strains GR44-11C (WT), FL53 (ppr1Δ) and FL51 (up2Δ) were grown at 30 °C on SD+aa with 2 g/l of uracil (+URA) or exposed to 200 mg/l of 6-azauracil (+6AU). ChIP signals were detected with anti-CTD antibodies (8WG16). A schematic map represents the URA2 gene, with the oligonucleotides used as primers.

URA2 activation in wild-type or ppr1Δ cells exposed to 6-azauracil did not increase the RNA polymerase II signal detected upstream of the initiator region, but correlated with a full occupancy of the URA2 open reading frame by RNA polymerase II (Figure 4C). Likewise, the up2Δ mutation produced a strong RNA polymerase II signal downstream of the initiator, even when grown under repressive conditions (Figure 4C). Thus, URA2 activation is not due to an increased recruitment of RNA polymerase II (as in the case of Ppr1-dependent genes such as URA1; see Figure 1C) but results from an extended RNA polymerase II occupancy downstream of the URA2 initiator region. Moreover, a relatively short DNA region, lost in up2Δ and located immediately upstream of the URA2 initiator, impairs URA2 transcription by preventing RNA polymerase II from progressing towards the URA2 open reading frame.

Mutants of the RNA polymerase II Switch 1 loop activate URA2 expression

To search for trans-acting regulator(s) of URA2, mutations constitutively expressing the chromosomal pURA2∷HIS3 reporter of strain FL52 were selected after UV mutagenesis. We obtained four mutants (FL-M9, M10, M13 and M23) that were resistant to 3-aminotriazol on uracil-supplemented medium and also overexpressed the pURA2∷LacZ reporter. They grew slowly at 30 °C, failed to grow at 16 °C and 37 °C and were somewhat sensitive to 6-azauracil (shown in Figure 5A for the FL-M9 strain). Meiotic tetrad analysis showed that these phenotypes co-segregated in a monogenic and recessive way, and complementation tests established that the corresponding mutations belonged to one and the same gene.

Figure 5

Properties of Switch 1 loop mutants. (A) GR44-11C (WT) and D711-13B (rpb1-L1397S) cells spotted on SD+aa or SD+aa supplemented with mycophenolate (50 mg/l) or 6-azauracil (20 and 200 mg/l). FL53 (ppr1Δ) and FL51 (up2Δ) were used as controls. Cells were microphotographed in exponential cultures grown in SD+aa at 30 °C. (B) Expression of pURA2∷LacZ. β-galactosidase was assayed as shown in Figure 2A, in strains BMA64-URA3 (WT) and FL-M9-URA3 (rpb1-L1397S) transformed with the pFL80 or pFL80-up2Δ pURA2∷LacZ reporter plasmids. (C) GR44-11C (WT) or YGH2 (WT), D711-13B (rpb1-L1397S), FL51 (up2Δ) and D876-10D (rpb1-L1397S up2Δ) were exponentially grown in CSM. Total RNA was extracted as described in Materials and methods. Steady-state levels of URA1, URA2 and URA3 mRNAs were quantified by RT–PCR and expressed in arbitrary units, where 1.0 corresponds to the wild-type level of URA1 mRNA in CSM medium for URA1 and URA3, and to the wild-type level of URA2 mRNA in CSM.

We then transformed FL-M9 with a yeast genomic library and obtained one plasmid restoring growth at 37 °C. The corresponding insert harboured RPB1, which encodes the largest subunit of RNA polymerase II, and three surrounding genes. Further subcloning showed that growth at 37 °C correlated with an intact RPB1 gene, and complementation tests with the temperature-sensitive allele rpb1-1 (Scafe ) firmly established that all four mutants were due to rpb1 mutations. In vivo gap repair (Rothstein, 1991) indicated that these mutations belonged to the SwaI–PshAI segment of RPB1. Finally, DNA sequencing revealed single amino-acid replacements corresponding to rpb1-L1397S (M9), rpb1-S1401P (M10) and rpb1-F1402L (M13 and M23), which were also present in the chromosomal DNA of the original mutant strains.

The above-mentioned data led to the rather unexpected conclusion that URA2 activation occurs by trans-acting mutations altering RNA polymerase II itself, at the level of its Switch 1 loop fold. RT–PCR and LacZ assays confirmed that rpb1-L1397S highly expressed URA2 (Figure 5B and C). Moreover, an rpb1-L1397S up2Δ double-mutant had the same constitutive expression as up2Δ alone, and a pURA2∷LacZ reporter bearing the up2Δ allele (plasmid pFL80-up2Δ) was expressed at the same level when tested in wild-type or in an rpb1-L1397S host strain (Figure 5B and C). In other words, rpb1-L1397S and up2Δ have epistatic effects on URA2 expression, thus strongly suggesting that they are defective in the same mechanism downregulating URA2 in response to uracil.

The six amino-acid segment occupied by L1397, S1401 and F1402 corresponds to Rpb1-α47b, one of the two α helices forming the Switch 1 loop domain of the RNA polymerase II active site (Gnatt ). Moreover, the rpb1-G1388V allele, altering the Rpb1-α47a helix (Berroteran ), also resulted in a high constitutive expression of URA2 (data not shown). This high clustering was not anticipated in a UV mutagenesis, as the latter evidently affects the whole yeast genome. Our genetic screen was presumably not saturating, and we cannot exclude that URA2 activation may, for some reason, be a general property of partly defective RNA polymerase II mutants. We thus tested the pURA2∷LacZ reporter (pFL80) in 10 other slow-growing mutations partly impairing Rpb1 (rpb1-G1437D, rpb1-E1351K and rpb1-H1367D) or Rpb2 (rpb2-R857K, rpb2-E836A, rpb2-D978A, rpb2-P1018S and rpb2-G1142D) or lacking the non-essential subunits Rpb4 (rpb4Δ) or Rpb9 (rpb9Δ) (Woychik and Young, 1989; Scafe ; Woychik ). Except for a modest derepressing effect of rpb2-E836A and rpb2-R857K, their β-galactosidase activity was equal to or lower than the wild-type control (data not shown), thus strongly suggesting that URA2 activation, at least to a large extent, is primarily due to changes in the Switch 1 loop.

The Switch1 loop is highly conserved in all eukaryotic and archaeal RNA polymerases (Figure 6A). Together with the α25 Bridge helix, α36 Trigger helix and the α46/47 Loop of Rpb1, it forms an identical fold in the RNA polymerase of Sulfolobus solfataricus (Hirata ) and in yeast RNA polymerase II (Gnatt ). This fold wraps the downstream end of the transcription bubble and holds the DNA template strand by the invariant R1386-E1403 Switch 1 dipole (Figure 6B and C). As discussed elsewhere (Zaros ), this fold is stabilised by the C-end of the Rpb5 subunit, itself strongly conserved from archaea to eukaryotes. Bacterial RNA polymerases, on the other hand, have no Rpb5 and their Switch 1 loop is considerably extended but nevertheless adopts a very similar spatial orientation (Vassylyev ).

Figure 6

Organisation of the Switch 1 loop in RNA polymerase II. (A) Sequence alignments with yeast, archaeal and bacterial subunits. The Switch 1 loop domains (Gnatt ) of Rpb1, Rpa190 and Rpc160 (S. cerevisiae) were aligned with the corresponding region of M. jannaschii (subunit A) and of the β′ subunits of E. coli, Thermus aquaticus and Synechocystis 6803. Amino-acid conservation is shown in red (all species), gold (Eukaryotes and Archaea) and blue (Prokaryotes only). Circles denote the mutated amino acids of rpb1-G1388V, rpb1-L1397S, rpb1-S1401P and rpb1-F1402L. (B) Spatial organisation of the Bridge, Trigger and Switch 1 fold in S. cerevisiae, S. solfataricus and T. thermophilus on the basis of PDB files 2NVZ, 2PMZ and 205I, respectively. These domains are shown on the Rpb1 sequence, with the same colour code and with numbers indicating the corresponding amino-acid positions. (C) Organisation of the Switch 1 loop in RNA polymerase II, using a different orientation, and showing the whole 10-subunit structure of RNA polymerase II (without Rpb4 and Rpb7). The template and non-template DNA strands and the nascent RNA are in dark green, light green and orange, respectively. A box shows the details of Switch 1 loop, underscoring the DNA-binding positions R1386 and E1403 by space-filling.

rpb1-L1397S enhances RNA polymerase II occupancy downstream of the URA2 initiator

Consistent with the partial growth defects of rpb1-L1397S, a twofold reduction in RNA polymerase II occupancy was seen for rpb1-L1397S in single-gene ChIP assays at ADH1 or PYK1 (Figure 7A). This extended to the whole genome, with an average occupancy level of 62% (R2=0.80) relatively to wild type (Figure 7B). As expected, RNA polymerase II occupancy was increased on the URA2 open reading frame. Moreover, this increased occupancy was only seen downstream of the URA2 initiator region (Figure 7C). Similar results were obtained in ChIP assays done in RPB3∷3HA rpb1-L1397S cells, showing that this effect is independent of the precipitating antigen (Figure 7D). In contrast, two RNA polymerase II alleles (rpb2-P1018S and rpb9Δ) with partial growth defects similar to those of rpb1-L1397S significantly reduced RNA polymerase II signal, upstream and downstream of the initiator region. Taken together, these data clearly show that rpb1-L1397S activates URA2 by extending RNA polymerase II occupancy downstream of the initiator region, thus recapitulating the effect seen above for up2Δ or for wild-type and ppr1Δ cells exposed to 6-azauracil.

Figure 7

RNA polymerase II occupancy in rpb1-L1397S. (A) RNA polymerase II occupancy at ADH1 and PYK1 in D535-4D (RPB3∷HA) and D535-9D (rpb1-L1397S RPB3∷HA). Cells were grown in SD+aa at 30 °C. RNA polymerase II was immunoprecipitated with anti-HA antibodies. A schematic map represents the oligonucleotide pairs used as primers. Position +1 corresponds to the initiator ATG. (B) Genome-wide ChIP assays. Three independent cultures of YGH2 (wild-type) and D711-13B (rpb1-L1397S) were grown in YPD at 30 °C. Chromatin was extracted and analysed as described previously (Ghavi-Helm ). (C) Distribution of RNA polymerase II at the URA2 locus. Data were extracted from genome-wide analysis shown in Figure 7B above. The wild-type and rpb1-L1397S profiles are shown in grey and black, respectively. (D) Effect of rpb1-L1397S, rpb2-P1018S and rpb9Δ on RNA polymerase II occupancy at URA2. Experimental conditions were as shown in Figure 7A. The strains used were D535-4D (RPB3∷HA), D535-9D (rpb1-L1397S RPB3∷HA), YMW304-8C (rpb2-P1018S RPB3∷HA) and YWM305-2D (rpb9Δ RPB3∷HA).

rpb1-L1397S extends RNA polymerase II downstream of IMD2, IMD3 and URA8

We have seen above that rpb1-L1397S generally reduced RNA polymerase II occupancy, except for URA2. This was also consistent with genome-wide transcriptome assays, showing a moderate but significant overexpression of URA2 in rpb1-L1397S, compared with an isogenic wild-type control (Figure 8A). Furthermore, these transcriptome and genome-wide RNA polymerase II occupancy assays identified two other genes, IMD2 (IMP dehydrogenase) and URA8 (CTP synthase), which were both overexpressed and over-occupied by RNA polymerase II in rpb1-L1397S. A few additional genes were also overexpressed in the rpb1-L1397S transcriptome, but with no higher RNA polymerase II occupancy of their open reading frames (data not shown). Conversely, rpb1-L1397S enhanced RNA polymerase II at ADE12 (adenylosucinate synthase), but had no detectable effect on its mRNA level.

Figure 8

Effect of rpb1-1397S on IMD2, IMD3, ADE12, URA7 and URA8. (A) Enrichment of RNA polymerase II transcripts in D535-9D (rpb1-L1397S) compared with an isogenic wild-type obtained by complementation with pFL36-RPB1. The histogram represents median percentile ranks of Cy3/Cy5 fluorescence ratios. Data are based on three independent cultures of each strain. The URA2, URA8, ADE12 and IMD2-4 transcripts are individualised by arrows. (B) RT–PCR of individual mRNAs. GR44-11C (WT), D711-13B (rpb1-L1397S) and FL51 (up2Δ) were exponentially grown in SD+aa. Total RNA was extracted as described in Materials and methods. Individual mRNAs were quantified by RT–PCR and expressed in arbitrary units, where 1.0 corresponds to the wild-type level. (C) RNA polymerase II occupancy at IMD2 and IMD3. Data were extracted from the genome-wide analysis shown above (Figure 7B) except for IMD2 in wild-type cells grown in CSM+MPA, taken from Ghavi-Helm . (D) RNA polymerase II occupancy at URA7, URA8 and ADE12. Data were extracted from the genome-wide analysis shown above (Figure 7B).

Saccharomyces cerevisiae has two CTP synthases (URA7 and URA8) and three active IMP dehydrogenase genes (IMD2, IMD3 and IMD4), with IMD1 as a pseudo-gene inactivated by a frame-shift mutation in most laboratory strains (Ozier-Kalogeropoulos ; Nadkarni ; Hyle ). URA7 and URA8 are sufficiently different to be distinguished by individual RT–PCR assays, which clearly showed that rpb1-L1397S activated only URA8 (Figure 8B). RT–PCR assays also confirmed the overexpression of the IMP dehydrogenase mRNA in rpb1-L1397S but did not clearly distinguish between IMD2, IMD3 and IMD4, due to their very similar nucleotide sequences. However, our genome-wide ChIP assays readily discriminated between these three genes. This revealed that rpb1-L1397S strongly activated IMD2, with some effect on IMD3 (Figure 8C) and no effect at all on IMD4 (not shown). We considered the possibility that the activation of CTP synthase and IMP dehydrogenase genes may be an indirect effect of URA2 overexpression, reflecting an increased cellular pool of UTP. However, the activation of URA2 in up2Δ was not accompanied by a parallel increase in URA8 or IMP dehydrogenase mRNA, thus strongly arguing for a direct activating effect of rpb1-L1397S (Figure 8B).

Previous studies have shown that IMD2 is controlled by an initiation switch between short transcripts, synthesised from an upstream initiator, and full-length mRNAs starting from a downstream initiator (Escobar-Henriques ; Steinmetz ; Kopcewicz ; Jenks ). RNA polymerase II is confined to the upstream initiator when wild-type cells are grown under repressing conditions (Steinmetz ), but extended to the entire IMD2 gene in rpb1-L1397S or in derepressed wild-type cells that have been exposed to mycophenolate (Figure 8C). In URA8 and perhaps in ADE12, Thiebaut have provided evidence for a transcriptional switch reminiscent of the IMD2 case. Again, RNA polymerase II was mostly restricted to the upstream part of the gene in wild-type cells grown under repressing conditions, but extended to the entire URA8 open reading frame in rpb1-L1397S (Figure 8C). A similar effect may apply to ADE12, encoding the first step of the ATP biosynthetic pathway, where rpb1-L1397S moderately increased RNA polymerase II occupancy downstream of the promoter region (Figure 8D).

Discussion

Previous studies have shown that URA1 or URA3 are specifically activated by Ppr1, a Zn activator of the Gal4-type which responds to increased concentration of the orotate and/or dihydro-orotate precursors of UTP (Loison ; Losson and Lacroute, 1981; Flynn and Reece, 1999). This leads to an increased RNA polymerase II occupancy at the URA1 and URA3 promoters, lost in the ppr1Δ-null allele, indicating a ‘classical' mode of gene-specific activation on the basis of a more effective recruitment of RNA polymerase II by its pre-initiation complex. URA2, which encodes the main rate-limiting enzyme of UTP biosynthesis, is activated when UTP is depleted by 6-azauracil addition. However, the ppr1Δ-null allele is fully competent for URA2 transcription. Moreover, and in contrast to URA1 or URA3, URA2 activation does not change the amount of RNA polymerase II residing at or upstream of the URA2 initiator region, but operates instead by extending RNA polymerase II occupancy to the URA2 open reading frame.

Three cis-acting mutations constitutively expressing URA2 were selected in this study. They corresponded to very similar deletions (up2Δ, up6Δ and up10Δ) removing a 55–70 nt DNA region upstream of the URA2 initiator. The up2Δ mutation was studied in more detail. It extended RNA polymerase II to the entire URA2 open reading frame, with no change in the RNA polymerase II signal upstream of the URA2 initiator, thus recapitulating the pattern seen in UTP-depleted wild-type or ppr1Δ cells. Further studies (Thiebaut ) have shown that the corresponding DNA is transcribed from an initiator region located some 100 nt upstream of the main URA2 mRNA 5′-ends identified here, producing unstable RNAs with the same transcriptional orientation as URA2, which were no longer detected in up2Δ. Taken together, these data suggest that URA2 may be regulated by an attenuation mechanism related to the one recently described for IMD2 (Escobar-Henriques ; Davis and Ares, 2006; Steinmetz ; Kopcewicz ; Jenks ).

A genome-wide selection for trans-acting mutations constitutively expressing URA2 yielded three tightly clustered rpb1 mutations (rpb1-L1397S, rpb1-S1401P and rpb1-F1402L), corresponding to the Rpb1α-47b helix of the Switch 1 loop. rpb1-G1388V, which was initially selected for its ability to alter transcription initiation in favour of downstream sites (Berroteran ) and belongs to the Rpb1-α47a helix, also activates URA2 (see also Thiebaut ). In contrast, other partly defective RNA polymerase II mutations failed to derepress URA2, indicating that URA2 activation is by no means a general property of RNA polymerase II mutants. Similar to up2Δ, rpb1-L1397S activates URA2 by extending RNA polymerase II occupancy to the entire open reading frame, with no change in the signal upstream of the URA2 initiator, thereby reproducing the effect seen in up2Δ and in UTP-depleted wild-type cells. Moreover, up2Δ and rpb1-L1397S had epistatic effects on URA2 expression, indicating that rpb1-L1397S is defective in the attenuation mechanism lost in up2Δ itself. Finally, rpb1-L1397S and wild-type URA2 mRNAs had similar 5′-ends, as determined by 5′-RACE assays, and therefore use the same transcription initiator.

The above-mentioned data indicate that changes in the Switch 1 loop specifically activate URA2, despite their adverse effects on growth and their reduced genome-wide RNA polymerase II occupancy. Moreover, we were unable to identify mutants encoding a specific regulator of URA2, although it remained possible that such mutants are lethal because they perturb some essential cellular process. The Switch 1 loop forms an invariant R1386(α47a)-E1403(α47b) dipole, holding the DNA template strand at positions +2/+3 downstream of the catalytic Mg2+ (Gnatt ). This domain is highly conserved among eukaryotic and archaeal RNA polymerases (Cramer ; Hirata ) and a related structure exists in the bacterial enzyme, where it also binds DNA at positions +2/+3 downstream of the catalytic Mg2+ (Vassylyev ; Zaros ).

The relation of this domain to the entry route of NTPs in RNA polymerase II is unclear (Landick, 2005). One possibility is that NTPs reach the catalytic site of RNA polymerase II by its funnel-shaped pore (Cramer ; Kettenberger ; Westover ; Landick, 2005), but it has also been suggested that NTPs enter through the large DNA channel, transiently binding the DNA template at nucleotides +2, +3 and possibly +4 (Gong ), which implies that NTPs pass by the Switch 1 loop before reaching the catalytic Mg2+. Hence, the Switch 1 loop may act as an NTP-sensing module, enabling RNA polymerase II to adopt a processive mode of elongation when enough NTP is available. This would optimise the efficiency of transcription, which may account for the cold- and heat-sensitive defects of rpb1-L1397S, rpb1-S1401P and rpb1-F1402L and their increased sensitivity to 6-azauracil and mycophenolate.

Strikingly, rpb1-L1397S also extended RNA polymerase II occupancy downstream of the IMD2/IMD3 (IMP dehydrogenase) and URA8 (CTP synthase) promoters, with some evidence for a similar effect on ADE12 (adenylosuccinate synthase). This makes biological sense, as the corresponding genes encode rate-limiting enzymes in the de novo synthesis of GTP, CTP and ATP. This suggests the rather non-canonical view that nucleoside triphosphate shortage or RNA polymerase II Switch 1 loop mutations (assumed here to mimic that shortage) specifically activate the expression of genes that are themselves critical for the de novo synthesis of NTPs. In other words, the transcriptional response to NTP shortage would not rely on dedicated activators or repressors (which have been vainly searched for in the case of URA2 and IMD2) but would be mediated by an NTP-sensing mechanism built in the structure of RNA polymerase II itself.

In IMD2, there is good evidence that RNA polymerase II molecules are recruited on a common pre-initiation complex, followed by a start site switch leading to the alternative production of upstream transcripts ended by Nab3/Nrd1-dependent termination or full-length mRNAs transcribed from a downstream initiator region. Start site selection is presumably dictated by the nucleotide composition of the surrounding DNA, with upstream transcription starting at Gs, whereas IMD2 transcription starts in a G-poor DNA region and is therefore favoured by low GTP pools (Escobar-Henriques ; Davis and Ares, 2006; Steinmetz ; Kopcewicz ; Jenks ). The URA2 start choice, however, depends on a different mechanism, as upstream initiation does not occur preferentially at T's and is not sensitive to the abundance of uracil (Thiebaut ). Conversely, we found that 3 out of 10 URA2 cDNAs produced under activating conditions started with a T, which argues against the simple idea of a selection driven by the nature of the starting nucleotide. Nevertheless, the upstream initiator is followed by an ∼60 nt T-rich domain, continued by 40 nt that are conspicuously poor in T's and harbour the URA2 mRNA 5′-ends. This pattern is conserved among all Saccharomyces species sequenced to data (see Supplementary data), and we would be surprised if it did not provide some form of UTP-sensing. Further studies are clearly needed to better understand how RNA polymerase II responds to NTP shortage, and how this response is ultimately converted into a specific control of the IMD2, URA2 and URA8 genes.

Materials and methods

Plasmids

Newly constructed plasmids are listed in Supplementary data. Plasmids pFL35-II, pFL36-CII (CEN6 LEU2) and pFL39-CII (CEN6 TRP1), corresponding to pFL35, pFL36 and pFL39 (Bonneaud ), with an extended polylinker of 90 nucleotides (Supplementary data). pFL80 and pFL82 are pURA2∷LacZ pURA2 reporters derived from pFL39-CII and pFL36-CII, respectively. They harbour a KpnI–BamHI–NotI pURA2∷LacZ cassette of 4392 nt. The KpnI–BamHI segment (1254 nt) is formed by the TRK1–URA2 intergenic region, preceded by GGTACCAGCACAACGCTCTAA (the KpnI site is in italics) and followed by ATGGATCC (BamHI). It is cloned in-frame to the initiator ATG of a BamHI–NotI LacZ cassette. pFL83 is pFL80 where the LacZ open reading frame has been replaced by a BamHI–NotI HIS3 (S. cerevisiae) cassette. As pFL80 poorly complemented trp1 mutant strains, which overexpressed the pURA2 promoter (due to some transcriptional interference between LacZ and TRP1), we also used pFL80-H2, a pURA2∷LacZ reporter with a lower LacZ expression. This plasmid was selected after random insertion of genomic HindIII fragments in the unique HindIII site downstream of LacZ (Supplementary data). The HindIII fragment of pFL80-H2 corresponded to positions 7679–8251 of the TEL1 gene and was inserted in a transcriptional orientation opposite to the one of TRP1.

Plasmids pFL80-KB, pFL80-BS and pFL80-CS correspond to KpnI/BsrGI, BsrGI/SalI and ClaI/SalI deletions of pFL80, respectively. pFL80-INTΔ was precisely deleted of the DNA comprised between nucleotides −384 to −63, where 1 corresponds to the URA2 start codon. The pFL83-upΔ2, -up6Δ and -up10Δ plasmids were obtained by error-prone amplification of the KpnI–BamHI cassette of pFL83, which was recombined into the original vector by co-transformation with linearised pFL83 DNA deleted of its BsrGI–SalI fragment. Mutant plasmids were selected on 3-aminotriazol in the presence of uracil, amplified in Escherichia coli, retransformed and verified for their overexpression. The corresponding upΔ alleles were transferred to pFL80 by KpnI–BamHI subcloning.

Yeast strains

Yeast mutants are listed as Supplementary data. Plasmid shuffle assays were done in the presence of 5-fluoro-orotic acid to counter-select URA3 plasmids (Boeke ). Except when stated otherwise, they were constructed in BMA64-1A and BMA64-URA3, a URA3+ revertant of BMA64-1A (Baudin-Baillieu ). Yeasts were grown on SD+aa, corresponding to the standard synthetic dextrose medium supplemented with histidine, tryptophan and adenine sulphate (20 mg/l), leucine and lysine (30 mg/l). In β-galactosidase assays, SD+aa was replaced by the complete synthetic medium CSM (manufactured by BIO101) supplemented with a complete set of amino acids but lacking uracil. CSM supplemented with 2 g/l of uracil (CSM+U) fully represses URA2, whereas adding 6-azauracil at 20 mg/l (CSM+ 6AU) fully activates URA2.

FL50 was obtained by inserting the integrative plasmid pFL81 (TRP1 pURA2∷HIS3) at the EcoRV site of the trp1-1 allele in W303-1B. When grown in the presence of uracil, this strain has an increased sensitivity to 3-aminotriazol, due to the repression of its pURA2∷HIS3 reporter. FL52 is FL50 transformed with pFL82 (LEU2 pURA2∷LacZ). FL-M9, M10, M13 and M23 are FL52 mutants selected from UV-irradiated cells (about 1% survival) plated for 3 days at about 4 × 106 cells on leucine omission plates with uracil (2 g/l) and 50 or 100 mM of 3-aminotriazol. β-galactosidase was assayed in a loop of cells put on Whatman paper soaked in 2 ml of Z buffer (Miller, 1972) containing X-Gal (1 g/l) and a drop of zymolase. Dark blue clones were tested again, using freshly grown cells suspended in 0.6 ml of Z buffer and vortexed for 1 min with an equal volume of glass beads. Cells were refrigerated in an ice bath, treated with a second round of glass bead extraction, pelleted by centrifugation at 14 000 r.p.m. and tested for β-galactosidase after 10 min at 37 °C. Mutant strains with an at least fivefold increase in β-galactosidase were retained for further analysis. A fifth mutant was a fur1 allele partly defective for UMP pyrophosphorylase, impairing the metabolic conversion of uracil into UMP.

FL51, with the up2Δ allele integrated on the chromosome by homologous recombination, was constructed by subcloning the KpnI–BamHI fragment of pFL81-up2Δ into pFL35-II. The resulting plasmid was linearised at the BsmI site of the pURA2 promoter region, to direct its integration by homologous recombination with the pURA2 region of BMA64-URA3. The corresponding transformants harboured the TRP1 cassette flanked by wild-type and up2Δ tandem copies of pURA2. Tryptophan auxotrophic clones spontaneously occurred by recombination between these pURA2 copies and were selected by nystatin enrichment (Snow, 1966), yielding single copies of pURA2. Among them, up2Δ mutants were identified by their weak resistance to 5-fluorouracil (10−5 M), and were further checked by PCR amplification.

ChIP and genome-wide ChIP-chip assays

Chromatin immunoprecipitation assays were as described elsewhere (Ghavi-Helm ). All experiments were performed on three independent cultures of 100 ml, harvested at an OD600 of 0.3–0.5. Cells were grown in SD+aa with or without uracil, or UTP-depleted by adding 200 mg/l of 6-azauracil to log-phase cells grown in SD+aa and grown for three additional doubling times. RNA polymerase II was immunoprecipitated with Dynabeads anti-mouse IgG (Dynal Biotech), using anti-CTD antibodies (8WG16), or anti-hemagglutinin A antibodies (12CA5) in the case of Rpb3-HA tagged strains. Sua7-TAP and Rad3-TAP proteins were directly immunoprecipitated on Dynabeads anti-mouse IgG. Phosphorylated variants of the RNA polymerase CTD domain were immunoprecipitated with Dynabeads anti-mouse IgM, using anti-Ser2P (H5, Covance) and anti-Ser5P (H14, Covance) antibodies. ChIP signals were calculated by the immunoprecipitation/input signal. The value 1.0 was arbitrarily given to the reference signal provided by amplifying the GAL1 gene. Genome-wide ChIP assays, on the basis of three independent cultures in YPD or CSM medium for mycophenolate experiments (10 mg/l) medium, were described elsewhere (Harismendy ) and were based on DNA arrays with over 40 000 oligonucleotide probes covering 12 Mb of the yeast genome (Ghavi-Helm ).

RNA and transcriptome assays

Total RNA was extracted from three independent, exponential cultures with hot phenol and reverse-transcribed using 1 μg of total RNA, Super-Script II reverse transcriptase (Invitrogen) and random hexamers as primers. Controls without reverse transcriptase showed negligible levels of DNA contamination. DNA was quantified by real-time PCR amplification (Applied Biosytems, System SDS Software) using primers listed in Supplementary data. mRNA levels are calculated as a ratio of measured mRNA and ACT1 mRNA. The 5′-ends of URA2 mRNAs were mapped by sequencing cDNAs obtained by the RNA ligase mediated-RACE technique (using an RLM-RACE kit from Ambion), according to the protocol provided by the manufacturer. Yeast micro-arrays were probed against total RNA extracted from three independent cultures of strains BMA64-URA3 (wild type) and FL-M9 (rpb1-L1397S) grown on SD+aa at 30 °C, to an OD600 of 0.55. cDNA synthesis, purification, indirect labelling, microarray hybridisation, scanning and analysis were done according to standard protocols (Soutourina ), with two independent hybridisations for each RNA batch.

Supplementary Information 1

Supplementary Information 2

Supplementary Information 3