urg1: A Uracil-Regulatable Promoter System for Fission Yeast with Short Induction and Repression Times

Background The fission yeast Schizosaccharomyces pombe is a popular genetic model organism with powerful experimental tools. The thiamine-regulatable nmt1 promoter and derivatives, which take >15 hours for full induction, are most commonly used for controlled expression of ectopic genes. Given the short cell cycle of fission yeast, however, a promoter system that can be rapidly regulated, similar to the GAL system for budding yeast, would provide a key advantage for many experiments. Methodology/Principal Findings We used S. pombe microarrays to identify three neighbouring genes (urg1, urg2, and urg3) whose transcript levels rapidly and strongly increased in response to uracil, a condition which otherwise had little effect on global gene expression. We cloned the promoter of urg1 (uracil-regulatable gene) to create several PCR-based gene targeting modules for replacing native promoters with the urg1 promoter (Purg1) in the normal chromosomal locations of genes of interest. The kanMX6 and natMX6 markers allow selection under urg1 induced and repressed conditions, respectively. Some modules also allow N-terminal tagging of gene products placed under urg1 control. Using pom1 as a proof-of-principle, we observed a maximal increase of Purg1-pom1 transcripts after uracil addition within less than 30 minutes, and a similarly rapid decrease after uracil removal. The induced and repressed transcriptional states remained stable over 24-hour periods. RT-PCR comparisons showed that both induced and repressed Purg1-pom1 transcript levels were lower than corresponding P3nmt1-pom1 levels (wild-type nmt1 promoter) but higher than P81nmt1-pom1 levels (weak nmt1 derivative). Conclusions/Significance We exploited the urg1 promoter system to rapidly induce pom1 expression at defined cell-cycle stages, showing that ectopic pom1 expression leads to cell branching in G2-phase but much less so in G1-phase. The high temporal resolution provided by the urg1 promoter should facilitate experimental design and improve the genetic toolbox for the fission yeast community.


INTRODUCTION
The experimental manipulation of expression levels from specific genes is a key genetic approach to elucidate gene function in model organisms. A range of regulatable promoter systems have been described for use in the fission yeast, Schizosaccharomyces pombe [1,2]. The most widely applied system is based on the thiamine-repressible nmt1 promoter and two weakened derivatives [3][4][5]. While these promoters offer a wide choice of transcription levels, they take ,15-21 hours to reach maximum induction once thiamine is removed from the medium, and ,2-4 hours for full repression after thiamine addition. A truncated version of the nmt1 promoter shows altered characteristics [6]: it reaches maximum expression within 3 hours but requires a temperature shift for induction, which is expected to trigger a cellular stress response [7].
Other promoter systems have been described for fission yeast. One system is based on ctr4 [8], which is strongly induced within 3 hours in the absence of copper. The addition of a copper chelator as an inducing agent, however, leads to a large transcriptional response [9]. The inv1 promoter fully induces transcription within one hour in the absence of glucose in sucrosebased medium, but induction is only transient as sucrose is hydrolyzed to glucose, leading to inv1 repression after ,2 hours [10]. The fbp1 promoter is also glucose-repressible [11]. Changes in carbon sources, however, lead to substantial transcriptional and metabolic responses ( [12]; LLM and JB, unpublished observation). The hsp16 promoter is activated within a few hours by heat shock or other stresses [13], conditions that will also trigger substantial stress responses [7]. The ectopic CaMV promoter is induced by tetracycline [14] or by anhydrotetracycline that is a superior inducing agent [15]. This promoter is regulatable in both minimal and rich media and is fully induced within 12 and 9 hours in the two media types, respectively [15]. The CaMV35S promoter shows low basal expression levels under repressed conditions [15], which should make it useful to study essential proteins.
Since the fission yeast cell cycle is completed within 2-3 hours, the lack of a promoter system that can be rapidly regulated, similar to the GAL system for budding yeast [16], is a serious drawback for many experiments. We used genome-wide expression data to identify conditions that lead to a strong and rapid regulation of few specific genes. This approach culminated in the development of the uracil-regulatable urg1 promoter system, which allows tight expression control of ectopic genes with otherwise minimal side effects on genome-wide gene expression.

RESULTS AND DISCUSSION
Microarrays are ideal to screen for genes that are distinctly regulated under selected conditions that otherwise have little effect on global gene expression. In a study to determine how commonly used media supplements affected global transcriptional patterns in fission yeast, we identified the three 'uracil-regulatable genes' urg1 (SPAC1002. 19), urg2 (SPAC1002.17c), and urg3 (SPAC1002.18) whose transcript levels were highly increased when the pyrimidine base uracil was present in the medium, a condition that affected the expression of only ,0.5% of all genes ( Figure 1A). This effect was not transient: the increased transcript levels were maintained after 24 hours in the continued presence of uracil (data not shown). Notably, the three urg genes were clustered together on chromosome I ( Figure 1B). urg1 encodes a protein of the GTP cyclohydrolase II family of enzymes involved in riboflavin biosynthesis [17]. urg2 encodes a protein similar to the budding yeast Fur1p uracil phosphoribosyltransferase of the pyrimidine salvage pathway [18], while urg3 encodes a protein of unknown function with a DUF1688 domain.
To analyse the regulation of the urg genes in more detail, we determined their expression profiles at different times after addition and removal of uracil ( Figure 2). All three genes were rapidly induced in uracil-containing medium, showing highly increased transcript levels after 5 minutes of uracil addition and peaking in transcript levels within 30 minutes. Similarly, transcript levels of all three genes rapidly dropped after transfer to medium without uracil. These results encouraged us to develop a new promoter system for the rapid and specific regulation of ectopic genes. We focussed on urg1 as it is the most strongly regulated gene in response to uracil ( Figure 1A; Figure 2). The difference in relative regulation between urg1, urg2, and urg3 seems to mainly reflect differences in basal expression levels: while all three genes show similar absolute expression signals in rich medium (containing uracil), the expression signals in minimal medium are ,3-and 4-fold higher for urg2 and urg3, respectively, compared to urg1 based on Affymetrix chip data [19]. Cells deleted for urg1 were viable and showed wild-type growth rates, both in the presence and absence of uracil (data not shown).
We analyzed available microarray data for urg1 expression patterns under different conditions. In vegetative cells growing in the absence of uracil, urg1 shows close to background signals on microarrays and is among the 10% most lowly expressed genes [19], and it is marginally periodically expressed during the cell cycle [20]. The urg1 transcripts were induced ,10-to 20-fold in response to cadmium and t-butylhydroperoxide but not in response to heat shock, sorbitol, MMS, H 2 O 2 , or menadione [7,21]. The urg1 transcripts were also induced ,10-to 20-fold in late meiosis and, most strongly, during nitrogen starvation, where transcript levels increased .100-fold [22][23][24], which is similar to the response in uracil described above. It is possible that urg1 is involved in recycling uracil as an alternative nitrogen source. Interestingly, urg1 was even more highly expressed in an ura4 deletion background than in a wild-type background in the presence of uracil, and conversely, ura4 was more highly expressed presence (Y-axis) or absence (X-axis) of uracil, whose transcriptomes were competitively hybridized on the same microarray. The urg1, urg2, and urg3 genes are most strongly induced in response to uracil, a condition that otherwise triggers only minor gene expression changes. The grey dots reflect transcripts that were flagged 'absent' during initial data processing [37]. (B) Genomic arrangement of urg1, urg2, and urg3 genes on chromosome I. Arrowheads indicate transcriptional direction. The SPAC1002.21 open reading frame between urg1 and urg3 may not be a real gene: it is annotated as 'dubious' in S. pombe GeneDB and does not seem to be expressed in any conditions based on microarray data (unpublished observations). doi:10.1371/journal.pone.0001428.g001 in an urg1 deletion background (unpublished microarray data). Thiamine, which represses the nmt1 promoter, has no influence on urg1 expression levels ( [25]; and unpublished data).
To develop a new regulatable promoter system, we cloned urg1 promoter fragments of different sizes (232, 675, and 924 bp upstream of urg1 start codon) into the pFA6a-kanMX6-P3nmt1 module [26], replacing the nmt1 with urg1 promoter fragments. We then applied PCR-based gene targeting with these three new cassettes to put the pom1 gene [27] under the control of the urg1 promoter fragments. The 232 bp fragment showed constitutively active transcription, whereas the 675 and 924 bp fragments both led to similarly regulated transcription in response to uracil (data not shown). These data suggest that the first 232 bp upstream of the ATG start codon are sufficient for active transcription whereas the sequences between 232 and 675 bp are required to downregulate transcription in the absence of uracil.
Based on these data, we cloned the 675 bp fragment containing the functional urg1 promoter (Purg1) into several PCR-based targeting vectors for straightforward integration of the promoter upstream of selected genes in their normal chromosomal locations ( Figure 3). The available modules contain the kanMX6 or natMX6 dominant markers, allowing selection for cells resistant to the antibiotics G418 or nourseothricin (NAT), respectively [26,28]. NAT allows easier selection on minimal medium (without uracil), which can be advantageous in situations where constructs with the active urg1 promoter lead to sick or dead cells. If required, the products of genes placed under Purg1 control can also be N-terminally tagged with 3HA, GST, or GFP (S65T) [26]. All the modules shown in Figure 3 can be amplified using the same forward primer, but they require different reverse primers (Table 1). Genomic integration of Purg1 ensures more controlled and homogeneous expression levels compared to analyses involving multi-copy plasmids, which show great variations in copy number. The 675-bp Purg1 fragment was used in all experiments below.
We tested the urg1 promoter system by placing pom1 under the control of Purg1 using the pFA6a-kanMX6-Purg1 cassette ( Figure 3; Table 2, strain JB381). The expression profile of pom1 driven by the urg1 promoter (Purg1-pom1) closely reflected the profiles of the urg genes, showing similar timing of induction and repression upon addition and removal of uracil, respectively ( Figure 2). Maximal Purg1-pom1 induction was reached within 10 minutes, which was ,2.2-fold higher than pom1 expression levels driven from its own promoter (based on microarray data). Rapid induction and repression time of Purg1-pom1, similar to urg1 under its own promoter, are also evident from PCR-based assays reported before ( Figure S9 in [19]). The ,10-fold induction of pom1, however, was lower than for the urg genes themselves. Several factors could contribute to this difference in relative regulation. Some of this difference is due to higher basal pom1 expression: qRT-PCR data showed that Purg1-pom1 is ,3.8-fold higher expressed than urg1 in the absence of uracil ( Figure 4A). Moreover, changes in half-live of transcripts can affect relative transcript changes; the 39-untranslated region (UTR) of urg1 contains an AU-rich element (ARE) consensus sequence [29], consistent with posttranscriptional control contributing to strong changes in mRNA levels [30,31]. It is possible that inserting the 39-UTR of urg1 behind genes already under the control of Purg1 would support lower basal transcript levels and larger relative transcript changes after uracil addition. Alternatively or in addition, the genomic context could influence the low expression of urg1 genes under repressed conditions. Consistent with this possibility, the expression levels of urg1 and urg2 increase in several silencing mutants [32], suggesting that this genomic region is relatively silent.
To compare the quantitative regulation by Purg1 with two widely used nmt1-based promoters (the strong P3nmt1 and weak P81nmt1; [3,5]), we performed qRT-PCR analysis of pom1 under control of these three ectopic promoter systems (Table 2, strains JB381, JB151, and JB178). Figure 4A compares the regulation of pom1 under control of either Purg1 or P3nmt1 with the regulation of urg1 and nmt1 under control of their native promoters. Both urg1 and nmt1 are more tightly regulated (higher induced and lower repressed transcript levels) than Purg1-pom1 and P3nmt1-pom1. The nmt1 gene under activating conditions is among the most highly expressed genes in the S. pombe genome [19,33], and expression levels driven by Purg1 are arguably closer to physiological levels for most genes. Note that the relative mRNA levels for ectopic genes put under control of regulatable promoters will be strongly affected by features such as chromatin context and mRNA stability, and different genes may show different regulation. Figure 4B compares the regulation of pom1 under control of  The forward primer is identical for all modules described here and is the same as for previously described modules containing nmt1-derived promoters [26]. The genespecific portion of the primer is typically chosen to correspond to sequences 100-200 bp upstream of the start codon. A web-based tool for automated primer design is available for these primers [35].

2
A 25-mer universal sequence is used to anneal to Purg1 due to the AT-rich nature of this sequence; the gene-specific portion is therefore reduced to 75 bp for 100-mer primers, which does not seem to affect targeting efficiency. The complement start codon is indicated in italic. For regulated expression of full length proteins, the genespecific portion of the primer corresponds to the complement of the N-terminal codons of the target gene (without start codon). 3 The reading frames of the tag sequences are indicated. These primers are the same as for the corresponding modules containing nmt1-derived promoters [26]. For Nterminal tagging of full-length proteins, the gene-specific portion of the primer corresponds to the complement of the N-terminal codons of the target gene (including start codon). Note that the 39 portions of these primers are specific to the tags and correspond to the complement of the C-terminal tag codons (without stop codon). A web-based tool for automated primer design is available for these primers [35].  Purg1, P3nmt1, or P81nmt1. Under induced conditions, Purg1-pom1 was ,3.5-fold more expressed than P81nmt-pom1 but ,3.5-fold lower expressed than P3nmt1-pom1. Under repressed conditions, Purg1-pom1 was ,1.7-fold more expressed than P81nmt1-pom1 but ,2.7-fold lower expressed than P3nmt1-pom1. These data are consistent with gene expression ratios observed after competitively hybridizing samples to DNA microarrays (induced conditions, Purg1-pom1/P3nmt1-pom1: 2.9, Purg1-pom1/P81nmt1-pom1: 0.3; repressed conditions: Purg1-pom1/P3nmt1-pom1: 1.6, Purg1-pom1/P81nmt1-pom1: 0.6). Our nmt1 data are also similar to data from a previous comparative analysis [1]. We conclude that for both induced and repressed conditions, the gene expression levels attained from Purg1 are between those attained from P81nmt1 and P3nmt1. These findings are corroborated by phenotype data. Deletion of pom1 leads to aberrantly positioned polarized growth and cytokinesis [27], while overexpression of pom1 under the control of P3nmt1 leads to branched cells after 19 hours and to depolarized, round cells within 25 hours [34]. In the absence of uracil, the Purg1-pom1 cells looked like wild-type cells ( Figure 5A), suggesting that basal transcription from the urg1 promoter provides sufficient pom1 expression to prevent the defects associated with pom1 mutants. The pom1 expression levels are ,3.8-fold below average transcript levels [19], and repressed Purg1-pom1 levels were ,90% of the native pom1 levels (based on microarray data). We conclude that the basal expression level from Purg1 can be sufficient to fully support gene function, at least for relatively lowly expressed genes such as pom1.
Induction of Purg1-pom1 expression by uracil addition led to increasing numbers of bent and branched cells already after two hours ( Figure 5A), although these cells did not become round even after 25 hours of Purg1-pom1 expression, consistent with weaker pom1 expression than in P3nmt1-pom1 cells. In contrast, P81nmt1-pom1 cells did not show any cell branching even 25 hours after uracil addition, consistent with weaker pom1 expression than in Purg1-pom1 cells. These findings are consistent with expression levels driven by Purg1 being between those of P81nmt1 and P3nmt1, and they illustrate the dramatic decrease in timing of transcriptional induction when using Purg1.
As a further proof-of-principle, we used the Purg1-pom1 cells to perform a cell-cycle experiment that would be difficult with the nmt1 promoter. Available data suggest that Pom1p can activate growth during the G2-phase of the cell cycle: 1) cells deleted for pom1 cannot activate a second growth site at the new end and thus fail to initiate bipolar growth during G2-phase [27]; 2) Pom1p kinase activity is cell-cycle regulated and is higher in G2-phase than in G1-phase [34]; and 3) overexpression of pom1 leads to branched cells, indicating mislocalized growth sites (see above).
Taking these data together, we would predict that Pom1p can promote cell branching when overexpressed in G2-phase but not when overexpressed in G1-phase. To test this hypothesis, we combined Purg1-pom1 with the temperature-sensitive cdc10 and cdc25 mutants, which arrest in G1-and G2-phases, respectively, at the restrictive temperature of 36uC (Table 2, strains JB509 and JB511). We incubated the cdc10 Purg1-pom1 and cdc25 Purg1-pom1 strains at 36uC to enrich for cells in G1-and G2-phases, respectively. After two hours, we added uracil to the medium to induce pom1 expression and incubated the cells for another two hours at 36uC. As predicted, the cdc25 Purg1-pom1 strain showed an about 7-fold higher proportion of branched cells than the cdc10 Purg1-pom1 strain ( Figure 5B-D). It is possible that the few branched cells in the cdc10 background reflect that two hours at 36uC was not sufficient to completely arrest all cells in G1-phase. These data support the notion that Pom1p can activate growth during the G2but not during the G1-phase of the cell cycle. Note that this type of experiment would be very complicated or impossible with the nmt1 promoter system due to the long induction times.

Conclusion
We believe that the urg1 system will prove to be a popular and valuable addition to the genetic toolbox available to fission yeast researchers. Besides regulation of the three urg genes, clustered together on chromosome I, the addition of uracil has only minimal effects on global gene expression and should affect cellular physiology less than the changes in carbon sources required for the budding yeast GAL promoter system. This specific effect on gene expression will also make it easier to interpret regulatory effects of genes under urg1 promoter control in genome-wide studies. The urg1 promoter system could also be used to control gene expression in specialized situations, such as to induce ectopic genes in nitrogen-starved cells, a condition that leads to urg1 induction in the absence of uracil. The pom1 gene under Purg1 control is fully induced and repressed within ,10 minutes of uracil addition and removal, respectively. Both induced and repressed Purg1-pom1 transcript levels are intermediate between those from the weakest and the strongest nmt1 promoter driving pom1.
Probably, Purg1 will be most useful to rapidly induce transcription of selected genes, e.g. to provide a pulse of expression during a defined cell-cycle stage. As most promoters, Purg1 supports substantial basal expression levels even when 'switched off'. As for the nmt1 promoter, regulation of urg1 transcripts themselves is tighter and stronger compared to the regulation of ectopic transcripts by Purg1. This difference could reflect local chromatin environment and/or additional posttranscriptional control. In any case, the half-lives of different transcripts will affect changes in transcript levels, and addition of the 39-UTR of urg1 might promote a tighter regulation of ectopic transcripts. Future refinements of the urg promoter system, including manipulations of the promoter sequence and analysis of uracil concentration effects, may further increase its usefulness.

Strains and yeast experiments
Strains used in this study are listed in Table 2. Strain JB381 was constructed using the new pFA6a-kanMX6-Purg1 module ( Figure 3) and transformed as described [26]. 100-mer primers were designed using PPPP [35] such that 160 bp of the native pom1 promoter were replaced with Purg1. Transformed cells were checked for correct integration by colony PCR using a forward primer in Purg1 (59-ATAAATAAGGGAGGAAATCCATACG-39), whose 59-end is located 203 bp upstream of the ATG start codon of urg1, and a reverse primer complementary to pom1. Strain JB383 was created by PCR-based gene deletion [26]. Strains JB509 and JB511 were created by crossing JB381 with JB506 and JB508. Cells were grown at 32uC in Edinburgh Minimal Medium (EMM) [36], adding either uracil at 0.25 mg/ml to induce Purg1, or 15 mM thiamine to repress the nmt1 promoter. For the experiment in Figure 2, JB381 cells were grown to ,5610 6 cells/ ml before uracil addition; after four hours, cells were filtered, washed once in 32uC EMM without uracil, and incubated in 32uC EMM without uracil for another 30 minutes. For Figure 5A, JB381 cells were grown in EMM without uracil, before adding uracil and growing for two hours. For the experiment in Figure 5B-D, JB506, JB508, JB509 and JB511 cells were grown at 25uC to ,5610 6 cells/ml, shifted to 36uC and grown for two hours, at which time uracil was added, and grown for another two hours at 36uC.

Construction of Purg1 modules
To construct the modules of Figure 3, the urg1 promoter (Purg1) was amplified from S. pombe genomic DNA by PCR using the following primers: urg1F675 (59-AAAAGATCTCGAT-TAGCGTGACACGGATT-39) and urg1R (59-AAATTAAT-TAACCTTTGTTCAGTGGCAAGCAT-39) containing BglII and PacI sites (underlined) for cloning into the corresponding sites of the pFA6a-MX6 vectors [26,28]. For the smaller and larger urg1 promoter fragments tested, we used the following two forward primers instead: urg1F232 (59-AAAAGATCTGCGCTTTCATT-GATAGTATCTG-39), urg1F924 (59-AAAAGATCTTGCACT-CAGCGTAAAGTCAAG-39). All PCR amplifications were carried out with HiFi-Platinum Taq (Invitrogen), and ligations were carried out using T4 DNA ligase (Roche) and transformed into DH5a competent cells (Invitrogen). Plasmid DNA was extracted using the Illustra GFX Micro Plasmid kit (GE Healthcare). uracil was then added to the same culture and cells were grown for another two hours (right). Cells show no morphological aberrations under repressed conditions but form misplaced growth sites after activation of Purg1. (B) cdc10 and cdc10 Purg1-pom1 cells grown for two hours at restrictive temperature without uracil and two hours at restrictive temperature with uracil. Activation of Purg1 has little effect on cell morphology. (C) cdc25 and cdc25 Purg1-pom1 cells grown for two hours at restrictive temperature without uracil and two hours at restrictive temperature with uracil. Activation of Purg1 leads to misplaced growth sites. (D) Histogram comparing percentage of branched cells when Purg1-pom1 is activated in either cdc10 or cdc25 backgrounds. doi:10.1371/journal.pone.0001428.g005

Microscopy
Light microscopy was carried out using a Carl Zeiss Axiostar equipped with a Canon Powershot A95 camera. Images were captured with the Canon ZoomBrowser EX software. For Figure 5, at least 200 cells were counted.

Microarray analysis
RNA was extracted and processed for microarray hybridization as described before ( [37]; www.sanger.ac.uk/PostGenomics/ S_pombe). Cy3 and Cy5 (GE Healthcare) incorporation was carried out using the Invitrogen Superscript direct cDNA labelling system according to manufacturer's instructions. For the timecourse experiment in Figure 2, all timepoints were pooled and used as a common reference for each timepoint. Microarrays were scanned using an Axon GenePix 4000B scanner and analyzed with GenePix 6.0 software. Quality control and data normalization was carried out as described [37]. Results were visualized with GeneSpring GX 7.3 (Agilent). The processed microarray data are available from our website: www.sanger.ac.uk/PostGenomics/ S_pombe.

Quantitative RT-PCR
For the qRT-PCR experiment in Figure 4, cells were grown for two hours either in the presence or absence of uracil (strain JB381), or for 21 hours in the presence or absence of thiamine (strains JB151 and JB178). RNA was isolated and purified as described [37] and treated with Turbo DNA-free (Ambion). Reverse transcription reactions were performed using Superscript III (Invitrogen). qRT-PCR reactions were carried out using Taqman specific probes (Sigma) and Platinum qPCR mix with ROX (Invitrogen) on an Applied Biosystem 7900HT system according to manufacturer's instructions. All primers and probes were designed using Primer3 software ( [38]; http://primer3. sourceforge.net). Expression levels for urg1, nmt1, and pom1 were determined from two repeats against a standard curve. Arbitrary expression units were calculated using a standard curve for each probe and primer set from serial dilutions of S. pombe genomic DNA. The following fluorescent probes with 6-FAM as 59-end reporter and TAMRA as 39-end quencher were used: P-C1223.02: 59-TTATTC-CAAGCGTTTGGGCATCATC-39; P-C2F7.03c: 59-CCTTTAC-CGAATTTGCCAATGGAAT-39; P-C1002. 19