ABSTRACT
Transcriptional regulation of IMD2 in yeast (Saccharomyces cerevisiae) is governed by the concentration of intracellular guanine nucleotide pools. The mechanism by which pool size is measured and transduced to the transcriptional apparatus is unknown. Here we show that DNA sequences surrounding the IMD2 initiation site constitute a repressive element (RE) involved in guanine regulation that contains a novel transcription-blocking activity. When this regulatory region is placed downstream of a heterologous promoter, short poly(A)+ transcripts are generated. The element is orientation dependent, and sequences within the normally transcribed and nontranscribed regions of the element are required for its activity. The promoter-proximal short RNAs are unstable and serve as substrates for the nuclear exosome. These findings support a model in which intergenic short transcripts emanating from upstream of the IMD2 promoter are terminated by a polyadenylation/terminator-like signal embedded within the IMD2 transcription start site.
Saccharomyces cerevisiae contains a family of four genes (IMD1 to IMD4) that encode IMP dehydrogenase. This enzyme is rate limiting for the de novo synthesis of guanine nucleotides. The IMD genes have been studied in the context of transcription elongation because mutations in the yeast elongation machinery confer growth sensitivity to inhibitors of IMP dehydrogenase, such as mycophenolate or 6-azauracil (6). IMD2 encodes a mycophenolate-resistant form of the enzyme that is transcriptionally induced in response to drugs that decrease intracellular guanine nucleotides (3, 7, 16, 19, 20). An optimally functioning elongation machinery is required for efficient induction of IMD2, although the precise mechanism is unknown (3, 5, 12, 19, 20, 22). Regulation of this gene is complex and seems to operate through an unusually far upstream TATA-like sequence called the guanine response element, which is positively acting (3, 4, 20). Mutation of this sequence results in loss of the drug inducibility of IMD2 (3, 4, 20).
In addition to IMD2 mRNA being upregulated when guanine nucleotide levels are low, IMD2 mRNA levels are repressed under guanine-replete conditions in which guanine nucleotides can be synthesized through a salvage pathway that bypasses IMP dehydrogenase (3, 7). Hence, the transcriptional output of IMD2 is exquisitely sensitive to guanine nucleotide levels. The proximal sensor of guanine nucleotides and the pathway that regulates IMD2 transcription are unknown. Typical sequence-specific DNA binding proteins that might serve as an upstream activator or repressor have not been identified for this gene.
In addition to the positively acting guanine response element, a distinct repressive element (RE) of approximately 100 bp that encompasses the transcriptional start site of IMD2 has been mapped, and deletion or mutation of this element results in derepression of the promoter in the absence of drug induction (4, 20). The RE also functions to repress the induction of the GAL1 transcript when placed downstream, but not upstream, of the GAL1 promoter (20).
It has recently been shown that small, unstable RNAs emanate from the IMD2 promoter region upstream of the previously characterized start site (2). These small RNAs become detectable when the exosome system is impaired by mutation. Here, we have employed a reporter system to show that a small regulatory region derived from the IMD2 transcription start site blocks full-length transcript production by a strong RNA polymerase II promoter and generates short poly(A)+ RNAs. When the RNA length is increased by moving the sequence downstream of the promoter, the prematurely shortened RNAs are stabilized, in contrast to the shortest versions of these RNAs, which are substrates for the nuclear exosome. A point mutation that ablates short transcript production results in dysregulation of IMD2. We suggest that the RE is a part of the mechanism controlling the extension of transcripts through the IMD2 open reading frame. This element resembles an efficient polyadenylation signal and terminator that overlap the IMD2 initiation region.
MATERIALS AND METHODS
Plasmid construction.The plasmids pIMD2-316, pIMD2-RE/316, pGAL-Luc, pIMD2-PGL4, and pIMD2-PGL5 have been described previously (7, 20). A PCR product containing the RE was generated using 5′-CGGGATCCTTCCGTATTCTATTCTATTCCTTGC-3′ and 5′-CGCCTGCAGAACAAAATGCGTTTATGACAGTT-3′ and blunted with T4 DNA polymerase. This was inserted in the forward orientations into the XbaI, BsiWI, or BstEII sites of luciferase in pGAL-Luc (also blunted) to generate the plasmids pREFXba-Luc, pREFBsi-Luc, and pREFBstEII-Luc and in the reverse orientation to yield pRERXba-Luc, pRERBsi-Luc, and pRERBstEII. psREF1Bsi-Luc was made by cutting a PCR product amplified using the oligonucleotides 5′-CGGGATCCTTCCGTATTCTATTCTATTCCTTGC-3′ and 5′-CGCCTGCAGAACAAAATGCGTTTATGACAGTT-3′ with Tsp590I. The upstream Tsp590I fragment was blunted and inserted into pGAL-Luc cut with BsiWI and blunted. pimd2-1 was made by introducing the RE mutation into IMD2 via site-directed mutagenesis of the plasmid pIMD2-S288C (20) with the mutagenic primers 5′-CTTATTATTTTCCATATAACCAATTTC-3′ and 5′-GAAATTGGTTATATGGAAAATAATAAG-3′. pREFimd2-1 was made by site-directed mutagenesis of pREFBsi-Luc by use of the same primers. When we sequenced these plasmids to confirm the change, we noted an A-to-G substitution 28 bp downstream from the intended change that resides in both the wild-type and mutant plasmids. Presumably, this came from the amplification of the original IMD2 sequence, but it was inconsequential in our assays. pGAL-REdown-COR was made by inserting the complementary oligonucleotides 5′-CTCAATAATACTTTTTAACTGTCATAAACGCATTTTGTT-3′ and 5′-GTACAACAAAATGCGTTTATGACAGTTAAAAAGTATTATTGA-3′, representing the downstream region of the RE, into the BsiWI site of pGal-Luc. pT3REM1 was made by cutting pGAL-Luc with BamHI and HindIII and religating the plasmid after filling in the ends.
Yeast strains.The yeast strains used in these studies are listed in Table 1. DY691, DY692, DY693, and DY1420 were made by adding the plasmids pIMD2-GL4, pIMD2-GL5, pREFXba-Luc, and pREFBst-Luc to BY4741-551682 (Δrrp6; Open Biosystems). DY896 and DY897 resulted from the addition of pIMD2-316 and pIMD2-RE/316 to DY891 (16). The strains DY1035, DY1036, DY1037, DY1038, DY1039, DY1040, DY1041, DY1215, DY1216, and DY1517 were made by transforming BY4741 with the plasmids pGAL-Luc, pIMD2-PGL4, pIMD2-PGL5, pREFXba-Luc, pRERXba-Luc, pREFBsi-Luc, pRERBsi-Luc, pREFBst-Luc, pRERBst-Luc, and pREFimd2-1. DY1409 and DY1410 were made by transforming ACY1280 (Anita Corbett, Emory University) with pREFBsi-Luc and pRERBsi-Luc. DY1415 was made by transforming BY4741 with psREF1Bsi-Luc. DY2605 was made by transforming BY4741 with pGAL-REdown-COR. DY1518 was made by integrating into DY891 the PCR product generated from pimd2-1 by use of oligonucleotides containing IMD2 sequence flanked by chromosome XII sequences 5′-GTTAAATGCTACGACTCGGCATATACTGTGCTCGTTTTTAGTGTCTGTTCTCTACAGTGCT-3′ and 5′-TTCGTTAAACGTCTGTAGAGGTGCCGAACAATTTTTGTCTGTAAACATAACACCCCATCAA-3′ and selecting for growth on medium lacking guanine. DY1519 was created similarly with a PCR product amplified from the plasmid pIMD2-S288C (20).
Yeast strains used in these studies
Northern analysis.Cells were grown in liquid media, collected at the logarithmic growth phase, washed once with water, and frozen. Total RNA was isolated from thawed cell pellets by hot phenol extraction and quantitated by measuring absorbance at 260 nm. Total RNA (30 μg) was resolved on a 1% (wt/vol) agarose gel and blotted onto Zeta-probe GT nylon (Bio-Rad) or Hybond XL (Amersham). Filters were baked at 85°C for 2 h and then prehybridized for a minimum of 3 h at 42°C in 5× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 5× Denhardt's solution, 50% (vol/vol) formamide, 1% (wt/vol) sodium dodecyl sulfate, and 100 μg/ml salmon sperm DNA. Filters were hybridized under the same conditions with 60 to 70 μCi of 32P-labeled DNA probe for 4 h at 30°C. Filters were washed at least once at 22°C in 0.1× SSC-0.1% sodium dodecyl sulfate for 15 min each. Washed filters were exposed to Kodak X-Omat or Pierce CL-Xposure film. Probes were labeled with Klenow DNA polymerase (Promega, Madison, WI), random hexamer primers (Invitrogen), and [α32-P]dATP (Amersham Biosciences). The “promoter-proximal” probe (Fig. 1A) was a PCR product made using pGalLuc (21) as the template and the primers 5′-ATACTTTAACGTCAAGGAGAAAAAACC-3′ and 5′-TGTTCACCTCGATATGGCATCTGTAA-3′. The “downstream” probe shown in Fig. 1A was a PCR product from the 3′ end of the luciferase gene generated using the oligonucleotides 5′-TCGCGGTTGTTACTTGCATG-3′ and 5′-TTCCATCTTCCAGGGATACG-3′ (20). The IMD2 downstream probe (see Fig. 7) was amplified from pIMD2-S288C (20) by use of 5′-GTGGTATGTTGGCCGGTACTACCG-3′ and 5′-TCAGTTATGTAAACGCTTTTCGTA-3′, which amplify a 401-bp fragment from the 3′ end of the IMD2 open reading frame. The SED1 probe was amplified from yeast genomic DNA by use of 5′-CCGAATTCCACTGATTGCTCCACGTCAT-3′ and 5′-CCGGATCCTTACACGCAACGCGTAAGAA-3′. A 220-nucleotide (nt) marker RNA for Northern analysis was generated with phage T3 RNA polymerase from the plasmid pT3REM1 cut with BsiWI.
The RE induces short transcripts when placed at sites downstream of the GAL1 promoter. (A) Schematic of the plasmids transformed into yeast (rows 1 to 9) or used to generate a marker RNA (row M) synthesized in vitro. The restriction sites in the luciferase sequence into which the RE was inserted are indicated at the top. The orientation of the RE is shown with white arrows. The promoter-proximal and downstream probes are shown as thick black underlines. ORF, open reading frame. (B) Northern blots using the downstream luciferase probe and strains with plasmid constructs 1 to 7 (top panel) or the promoter-proximal luciferase probe and strains with plasmid constructs 1 to 9 (lower panel). RNA was isolated from strains DY1035 (lanes 1), DY1036 (lanes 2), DY1037 (lanes 3), DY1038 (lanes 4), DY1039 (lanes 5), DY1040 (lanes 6), DY1041 (lanes 7), DY1215 (lane 8), and DY1216 (lane 9) at 0, 0.5, or 2 h following galactose induction and subjected to Northern blotting. Orientation of the RE is indicated by black arrows above the lanes. A 220-nt RNA prepared by in vitro transcription of BsiWI-cut pT3REM1 (panel A) was run in lane M. Short RNAs are indicated with white underlines.
RNase H cleavage.Sixty micrograms of total cellular RNA was mixed with 20 μmol oligonucleotide 1 (5′-GCCTTATGCAGTTGCTCTCC-3′) or oligonucleotide 2 (5′-CGGGATCCATATAGAAAATAATAAGAAAAGTAAG-3′) or 30 μmol oligo(dT100) in (50 mM Tris, pH 7.5, 100 mM NaCl) and incubated at 65°C for 5 min. Samples were diluted threefold with 20 mM Tris, pH 7.5, 100 mM KCl, 10 mM MgCl2, 0.1 mM EDTA, 0.1 mM dithiothreitol, 50 μg/ml bovine serum albumin and digested with 6 units of RNase H (Invitrogen) for 30 min at 23°C before RNA was ethanol precipitated and prepared for Northern analysis.
RESULTS
We previously showed that a DNA sequence called the RE and encompassing IMD2's initiation site prevents productive transcription from the GAL1 promoter in yeast when it was placed downstream of the promoter (20). The RE did not function, however, when it was located upstream of GAL1, the usual position of a recognition site for a sequence-specific DNA binding repressor. This prompted us to examine the nature of this repression and to ask how far downstream from a promoter the RE could function to block transcription. We generated a family of transcription units on plasmids containing the GAL1 promoter, each with the RE placed progressively further downstream from the promoter in either the natural or inverted orientation (Fig. 1A). Strains were induced with galactose for various times and analyzed by Northern blotting using a downstream probe in the luciferase-encoding segment (Fig. 1A). The RE prevented accumulation of the expected transcripts when it was placed ≈30 (the BamHI site), 80 (the XbaI site), or 190 (the BsiWI site) bp downstream of the GAL1 initiation site, and it did so in an orientation-dependent manner (Fig. 1B, top panel).
Since transcription of IMD2 is dependent upon efficient elongation by RNA polymerase II (3, 5, 10-12, 19, 20, 22), we tested whether the RE presented a block to transcription by probing for prematurely shortened transcripts by use of a promoter-proximal region of DNA (Fig. 1A). For three distal insertions (XbaI, BsiWI, and BstEII [645 bp downstream]), short RNAs were detectable, but only when the RE was in the correct orientation (Fig. 1B, bottom panel). Based upon the sizes of these RNAs relative to a 220-nt reference transcript generated in vitro by use of T3 phage RNA polymerase (construct M in Fig. 1A), as well as the resulting pattern of a nested set of short RNAs as the RE was moved downstream, we conclude that the 3′ ends of the short RNAs fall within the RE. When the RE was in the reverse orientation, transcription from the GAL1 promoter was unperturbed. The most promoter-proximal insertion (BamHI) also generated short transcripts, but they were more difficult to detect under these conditions (Fig. 1B, bottom panel, construct 2; also see below). Hence, the RE did not prevent initiation from the GAL1 promoter but resulted in the generation of short RNAs at the expense of full-length RNAs, and the further downstream from the transcription start site that the RE was placed, the longer and more abundant the shortened RNAs became.
The RE encompasses sequences both up- and downstream of the IMD2 mRNA's transcription start site. An upstream piece of the RE (−53 to +4) was inserted downstream from the GAL1 promoter and assayed for blocking activity (Fig. 2, lanes 4 to 6). Compared to the full-length RE (Fig. 2, lanes 1 to 3), this 5′ portion of the element did not generate short transcripts or prevent full-length RNAs from being synthesized. Similarly, a downstream piece of the RE (+6 to +53) was completely inactive at generating short transcripts. Hence, DNA sequences just upstream and just downstream of the natural IMD2 initiation site are required for the blocking function of the RE. DNA at the 5′ end of the RE contains three tandem TATTC repeats (Fig. 3A) which attracted our attention. We tested if this sequence was important for the RE's function by splitting it from the rest of the RE. Neither the repeats nor the remainder of the element was active in blocking full-length RNA production, indicating that the repeats were required for RE function (data not shown). Thus, the sequence requirements for RE function are complex, involving DNA both up- and downstream of the IMD2 initiation site.
Transcribed and nontranscribed parts of the RE are required for its function. Strains containing plasmids with the GAL1 promoter driving transcription into the RE inserted into the BsiWI site of the tester plasmid (DY1040; lanes 1 to 3), the upstream portion of the RE (DY1415; lanes 4 to 6), or the downstream portion of the RE (DY2605; lanes 7 to 9) were grown in raffinose and challenged with galactose for the indicated lengths of time before RNA was prepared for Northern analysis with the promoter-proximal probe or the SED1 probe. FL indicates the position of the major full-length RNAs, and the arrow shows the position of shortened RNAs.
A mutation in the RE inactivates its blocking effect. (A) Map of the RE. The IMD2 transcription start site (+1) is indicated with a bent arrow above the sequence. The T-to-C mutation is indicated, and the region deleted in the “Δ” construct (see Fig. 7A) is delimited by a bracket. The TATTC triplet is underlined. (B) RNA was prepared from DY1035 (no RE), DY1040 (+ RE), DY1041 (+ revRE), and DY1517 (+ mtRE) after the indicated times in the presence of galactose. Lanes 1 to 3 each received 7.5 μg RNA, and lanes 4 to 12 each received 15 μg RNA. Northern blots were probed with the promoter-proximal probe (Fig. 1A). The open arrowhead shows the position of the major full-length (FL) RNA, and the arrow shows the position of shortened RNAs.
Using an in vivo screen in which the RE located upstream of the green fluorescent protein (GFP) reading frame blocked GFP production, we identified a single base mutation in the RE that resulted in the restoration of GFP expression (data not shown). The change was a single T-to-C transition at −7 relative to the natural IMD2 transcription start site (Fig. 3A). To test if this mutation restored GFP expression because it inactivated the RE's blocking activity, we engineered this mutation into the RE in a GAL-RE-luciferase reporter plasmid. Compared to the wild-type RE (Fig. 3B, lanes 4 to 6), the mutated RE was completely incapable of generating short transcripts in yeast (Fig. 3B, lanes 10 to 12), and full-length RNA was generated with an efficiency comparable to that seen with the no-RE (lanes 1 to 3) or inverted-RE (lanes 7 to 9) test plasmid.
To determine if the short transcripts were polyadenylated, we mixed an ≈290-nt RNA generated by the RE (Fig. 1, construct 6) with oligonucleotides complementary to sequences in the body of the transcript (oligonucleotide 1 or oligonucleotide 2 [Fig. 4]) or oligo(dT), and the hybrids were digested with RNase H. Cleavage of the RNA:DNA hybrids was detected by Northern blotting (Fig. 4). Hybridization to internal oligonucleotides 1 and 2 resulted in the expected shortening of the RNA commensurate with the location of the hybridization sequences (lanes 4 and 5). Hybridization to oligo(dT) resulted in a slight but reproducible shortening of the transcript in an RNase H- and oligo(dT)-dependent manner (lane 1 versus lanes 2 and 3; also data not shown). We conclude that the RE-dependent RNA bears a short poly(A) tail.
RE-generated short RNA contains a poly(A) tail. RNA was isolated from strain DY1040 (plasmid construct 6 in Fig. 1A) after 90 min of exposure to galactose. Aliquots were hybridized to oligonucleotide (oligo) 1, oligonucleotide 2, poly(dT), or no oligonucleotide as indicated, treated with RNase H or buffer, separated by electrophoresis, blotted, and hybridized to the promoter-proximal probe. A schematic of the RNA and the expected products of RNase H digestion in the presence of the indicated oligonucleotides is shown to the right.
Processing of mRNAs involves cleavage and polyadenylation of the precursor transcript in a process that is coupled to transcription elongation and termination (reference 1; reviewed in reference 17). We examined the effect on short transcript synthesis of mutations in RNA15, which encodes an essential part of the polyadenylation machinery and is necessary for proper termination (17). Mutations in this gene have been shown to reduce the abundance of certain mRNAs, since the aberrant RNAs resulting from a failure to terminate are degraded by the nuclear exosome (18, 24).
The GAL-RE reporter plasmids (Fig. 1A, constructs 6 and 7) were moved into a strain with the temperature-sensitive rna15-2 mutant (18). Cells from wild-type and mutant strains were induced with galactose, and RNA was analyzed by Northern blotting with the promoter-proximal probe (Fig. 5). At the permissive temperature, the rna15-2 strain showed a reduced amount of RE-dependent short transcript relative to wild-type cells (lane 2 versus lane 6). At the restrictive temperature, no short transcripts were detectable in the mutant strain (lane 8). (We confirmed this when twice the usual amount of RNA from the mutant strain was analyzed [Fig. 5, lanes 5′ to 8′].) This was specific for the short transcript, since the levels of the full-length RNA resulting from placing the RE in the inverse orientation downstream from the GAL1 promoter were similar between mutant and wild-type cells (lane 12 versus lane 16).
Short transcript accumulation is dependent upon RNA15. Strains DY1040 and DY1041 (RNA15; lanes 1 to 4 and 9 to 12, respectively) and DY1409 and DY1410 (rna15-2; lanes 5 to 8 and 13 to 16, respectively) were grown in the presence of galactose for 0 or 90 min at the permissive (23°C) and restrictive (37°C) temperatures. RNA was prepared for Northern blotting, and filters were hybridized with the promoter-proximal probe (Fig. 1). As a loading control, the filters were stripped and reprobed with the SED1 probe. Thirty micrograms of RNA was loaded into lanes 1 to 12. Sixty micrograms was loaded into lanes 5′ to 8′ to emphasize the low levels of short transcripts seen for this strain.
We next considered the possibility that the short RNAs generated from the GAL-RE reporter series were substrates for degradation by the nuclear exosome. Recent work indicates that short intergenic transcripts around the genome, including one upstream of the IMD2 promoter region, are substrates of a nuclear exosome nuclease (2). We introduced GAL-RE-luciferase reporter plasmids into a strain of yeast deleted for RRP6. Loss of RRP6 led to a dramatic increase in the abundance of the shortest transcripts derived from RE insertions (Fig. 6A, lanes 1 to 3 versus lanes 4 to 6, and lanes 7 to 9 versus lanes 10 to 12). A longer RNA that resulted from the RE's placement further downstream (645 bp [Fig. 1A, construct 8]) was more abundant in wild-type cells than were the shorter RNAs (Fig. 6B, lanes 2 and 3, versus Fig. 6A, lanes 1 to 3 and 7 to 9). The longer RNAs were not additionally stabilized by RRP6 deletion (Fig. 6B, lanes 2 and 3 versus lanes 5 and 6). We conclude that promoter-proximal RE-dependent short transcripts are degraded by the nuclear exosome; however, once they reach a threshold size (between 180 and 290 nt), they no longer serve as efficient substrates for this nuclease.
Short RE-generated transcripts are stabilized in RRP6 deletants. (A) Plasmids containing the RE in the correct orientation in two promoter-proximal regions (constructs 2 and 4, lanes 1 to 6 and 7 to 12, respectively) or in the opposite orientation (construct 3, lanes 13 to 18) were introduced into RRP6 (lanes 1 to 3, 7 to 9, and 13 to 16) or Δrrp6 (lanes 4 to 6, 10 to 12, and 16 to 18) strains. Strains DY1036 (lanes 1 to 3), DY691 (lanes 4 to 6), DY1038 (lanes 7 to 9), DY693 (lanes 10 to 12), DY1037 (lanes 13 to 15), and DY692 (lanes 16 to 18) were grown in raffinose, and galactose was added for 0, 30, or 90 min before RNA was prepared for Northern blotting with the promoter-proximal probe (upper panel), autoradiographed, stripped, and reprobed with the SED1 probe (lower panel). (B) A plasmid containing the RE inserted 645 bp downstream from the GAL1 promoter (construct 8 in Fig. 1A) was introduced into yeast strains containing wild-type RRP6 (DY1215; lanes 1 to 3) or Δrrp6 (DY1420; lanes 4 to 6). RNA for Northern analysis was harvested from cells at the indicated times after exposure to galactose and probed with the promoter-proximal probe (Fig. 1A).
We tested the role of the RE in guanine-mediated repression of IMD2. First, we exploited strains that lacked endogenous IMDs but which contained a plasmid-borne copy of either IMD2 or IMD2 lacking its RE (Fig. 3A and 7A). Northern blots were probed with an IMD2 fragment derived from the 3′ end of IMD2. As shown previously (20), deletion of the RE derepressed full-length IMD2 mRNA levels in the absence of guanine (Fig. 7A, lane 1 versus lane 2). (Multiple bands are presumably due to the heterogeneous polyadenylation of IMD2 mRNA.) Following the addition of guanine to the growth medium, transcript levels from intact IMD2 were repressed as expected (Fig. 7A, lane 3 versus lane 4). Guanine did not, however, repress the expression of IMD2 when its RE was deleted, rendering it effectively immune to repression by this metabolite (Fig. 7A, lane 5 versus lane 6).
RE mutation and guanine repression of IMD2. (A) Strains DY896 and DY897, which bear plasmids with IMD2 containing (+) and lacking (Δ) the RE, respectively, were grown in the absence (−) or presence (+) of 0.5 mM guanine. Total RNA was isolated after 30 min and subjected to Northern blotting with the downstream probe from the 3′ end of the IMD2 open reading frame (see Materials and Methods) or the SED1 probe. (B) Similarly, strains DY1523 and DY1522, which contain plasmid-borne IMD2 and IMD2 with the T→C substitution at −7, respectively, were grown in the absence or presence of 0.5 mM guanine as indicated. RNA was isolated and analyzed by Northern blotting as in panel A. (C) Strains DY1525 and DY1524, which contain wild-type IMD2 and IMD2 with the T→C substitution at −7 integrated into chromosome XII, respectively, were grown in the presence or absence of 0.5 mM guanine. RNA was isolated and subjected to Northern blotting as in panel A. We presume that the heterogeneous distribution of these bands is due to the lack of the natural polyadenylation element (thus far unmapped) in this construct, which ends 117 bp downstream from the IMD2 stop codon.
To examine this at higher resolution, we exploited the mutation described above, which prevents the RE from generating short transcripts in the GAL1-driven reporter. The T→C change was built into a plasmid-borne copy of IMD2 and introduced into yeast lacking all IMD genes. The point mutation resulted in an increase in full-length IMD2 mRNA in the absence of guanine (Fig. 7B, lane 1 versus lane 2 and lane 3 versus lane 5) indicating that this position is important for much of the full RE's activity. Guanine repression was readily detected for intact IMD2 (lane 3 versus lane 4), but full repression was impaired in the plasmid with IMD2 containing the T→C change in the RE (Fig. 7B, lane 5 versus lane 6), at least in terms of absolute mRNA levels present after guanine challenge (compare lanes 4 and 6). We take this as additional in vivo evidence that the RE region is important for the regulated expression of IMD2.
Finally, we integrated wild-type IMD2 and IMD2 with the T→C change into a relatively gene-poor region of chromosome XII to study guanine repression in a true chromatin background. Again the point mutation resulted in derepression under basal conditions (Fig. 7C, lane 1 versus lane 2). It also failed to show strong guanine-mediated repression (lane 5 versus lane 6) compared to what was seen for the wild type (lane 3 versus lane 4). These results indicate that the mutation associated with short transcript formation in the GAL reporter plasmid is important for basal and repressed activity of the promoter.
DISCUSSION
The regulation of IMD2 is unusual and complex. While a number of DNA sequences have been identified as important, no traditional activator or repressor proteins have been identified, nor has the pathway of guanine regulation been elucidated. Here, we expand upon earlier findings that the region around the transcription initiation site is a sequence that represses full-length IMD2 mRNA synthesis. The transcription of IMD2 can be considered privileged in that it is highly active when guanine nucleotide levels and total poly(A)+ mRNA synthesis are strongly reduced by mycophenolate treatment (19). This strategy makes biological sense, since synthesis of IMP dehydrogenase is required to overcome the effect of the drug, and transcriptional upregulation of the enzyme needs to occur even during nucleotide depletion. The need for transcriptional induction under nucleotide-limiting conditions could help explain why this gene's transcription is dependent upon an optimally functioning transcription elongation machinery.
The known dependence of IMD2 transcription upon efficient elongation led to the hypothesis that the RE modulates elongation. Our use of the strong inducible GAL system to study the RE allowed us to detect foreshortened RNAs and study the sequence and genetic determinants of RE function. Recently, undetectable short transcripts initiating upstream of the IMD2 initiation site were revealed when the nuclear exosome was impaired (reference 2 and K. A. Kopcewicz and D. Reines, data not shown). We propose that the RE element is involved in the synthesis of such short RNAs.
The RE resembles a potent polyadenylation and/or termination site rather than a pause site of the type seen for Drosophila heat shock genes (13). A polyadenylation/termination function would account for our observation of a discrete polyadenylated RNA species. It would also explain (i) why the RE sequence is inactive when placed upstream of the GAL1 promoter, (ii) why the RE is so strongly orientation dependent, and (iii) why its inactivation by mutation or removal leads to the restoration of full-length RNA synthesis. This particular polyadenylation/termination site would be unusual considering its strength and its location at a transcription initiation site. One of its functions could be to protect IMD2 initiation from upstream polymerase traffic. Oncoming transcription from an upstream intergenic promoter, SRG1, has been shown to negatively regulate the downstream promoter SER3 in yeast (14, 15). Thus, when the upstream promoter is active, the downstream one is inactivated and vice versa. Whether a similar regulated occlusion event operates at IMD2 is unclear (2). Our evidence is consistent with the idea that the RE could be involved in terminating intergenic RNAs upstream of IMD2, thereby protecting the IMD2 promoter from occlusion by oncoming transcription.
It is interesting that the two insertion sites of the RE that are promoter proximal (constructs 2 and 4) give rise to labile short transcripts, while two more promoter-distal insertions (constructs 6 and 8) yield relatively stable RNAs (Fig. 1B and 6). Hence, there seems to be length dependence in the targeting of these foreshortened RNAs for exosome destruction. In this regard, it has been suggested by Steinmetz and coworkers that the termination of short promoter-proximal transcripts operates via a Nrd1-dependent system, while longer transcripts are substrates for conventional polyadenylation-coupled termination (23). The RE may be important in keeping the intergenic IMD2 promoter-associated RNAs from becoming full-length poly(A)+ transcripts shipped to the cytoplasm.
While the precise role of the RE at IMD2 is enigmatic, we can exclude a number of possible functions for it. The DNA sequence itself does not block elongation by purified RNA polymerase II in vitro as a traditional elongation factor SII-dependent arrest site does (data not shown). If the sequence binds a protein, it does not serve as a simple steric block to elongation unless it does so in a highly polar manner. The RE alone does not serve as an alternative promoter because it is insufficient in either orientation to initiate transcripts, nor does it pirate initiation from an adjacent promoter, e.g., GAL1. Additional sequences are required for regulated elongation through the RE, since mycophenolate treatment of cells bearing the GAL1-RE reporters does not show relief of blockage (data not shown). This is not surprising, since the guanine response element that lies further upstream is required for promoter strength and induction (3, 4). In sum, regulation of IMD2 appears to involve the embedding of a polyadenylation/termination signal in the heart of a promoter, where it may serve as a switch to control productive mRNA synthesis.
ADDENDUM IN PROOF
A recent paper (E. Steinmetz, C. Warren, J. Kuehner, B. Panbehi, A. Ansari, and D. Brow, Mol. Cell 24:735-746, 2006) examining the genome-wide role of the yeast Sen1 termination-related protein describes the repressive element region of IMD2 as a Sen1-dependent terminator. Steinmetz et al. demonstrate terminator function using Pol II density distribution mapping. Our independent results are consistent with their work. Taken together, our mutual findings strengthen the idea that this DNA element is involved in the switch from an upstream start site to a downstream start site during regulation of this promoter and that derepression of IMD2 by a point mutation (Fig. 7) is likely due to terminator readthrough by Pol II from the upstream start site.
ACKNOWLEDGMENTS
We thank Randy Shaw and Young Lee for plasmid and strain construction and early observations of RE function and J. Boss, C. Moran, and A. Corbett for helpful comments.
This work was supported by NIH grant GM46331.
FOOTNOTES
- Received 17 November 2006.
- Returned for modification 19 December 2006.
- Accepted 31 January 2007.
- Copyright © 2007 American Society for Microbiology
