Yeast Pol II start-site selection: the long and the short of it

The formation of a pre-initiation complex (PIC) is the first step that takes place during transcription by RNA polymerase II (Pol II). The TATA-binding protein binds to the TATA box, thereby positioning the polymerase in the PIC in such a way that it can unwind downstream DNA and lock the template strand in the active site (Bushnell et al, 2004). In metazoans, initiation takes place 25–30 base pairs (bp) downstream of the TATA box (Smale & Kadonaga, 2003), whereas in Saccharomyces cerevisiae, start sites are spread across 40–120 bp downstream of the TATA box (Hampsey, 1998). The rules that control selection of the Pol II start point are not well understood. In metazoans, start sites often occur at a conserved initiator sequence (Smale & Kadonaga, 2003), whereas yeast initiators are more divergent (Zhang & Dietrich, 2005) and the polymerase has been proposed to ‘scan' the downstream sequences for optimal start sites (Giardina & Lis, 1993; Kuehner & Brow, 2008). Several articles recently published by the Brow, Lacroute, Libri and Reines laboratories show that start-site selection has an important role in regulating the expression of genes of the S. cerevisiae nucleotide-biosynthetic pathways (Jenks et al, 2008; Kuehner & Brow, 2008; Kwapisz et al, 2008; Thiebaut et al, 2008).

IMD2 and URA2 encode the initial enzymes in the pathways that lead to the synthesis of GTP and UTP, respectively. The transcription of these genes is repressed by the end product NTP (Escobar-Henriques & Daignan-Fornier, 2001; Losson & Lacroute, 1981); however, despite years of searching, no repressors or activators have been identified. Studies on IMD2 identified a guanine-responsive TATA box and a repressive element spanning the messenger RNA (mRNA) start site, yet how these cis-elements regulate IMD2 expression was unclear (Escobar-Henriques et al, 2003; Shaw et al, 2001). Transcription profiling of strains disabled for nuclear exosome function revealed the existence of transcripts that originate upstream of the IMD2 and URA2 mRNA start point (Davis & Ares, 2006; Thiebaut et al, 2008), and additional work has shown that these transcripts terminate before reaching the coding region (Fig 1; Jenks et al, 2008; Kuehner & Brow, 2008). Termination is directed by components of the non-poly(A) termination pathway, and include the RNA-binding proteins Nrd1 and Nab3, the helicase Sen1 and components of the nuclear exosome (Conrad et al, 2000; Steinmetz et al, 2001; Vasiljeva & Buratowski, 2006). This termination complex recognizes specific sequences in the nascent transcript, and directs its termination and degradation. Deletions or point mutations in the IMD2 or URA2 terminators—located upstream of the coding region—result in gene expression even under repressing conditions such as high NTP concentration, showing that the upstream non-coding transcripts are essential for complete repression (Kuehner & Brow, 2008; Kwapisz et al, 2008; Thiebaut et al, 2008).

Figure 1.

Map of the upstream and downstream transcripts that regulate the expression of IMD2 and URA2. mRNA, messenger RNA; ORF, open reading frame.

What factors determine whether Pol II initiates at the upstream or downstream sites? Presumably, as Pol II scans the sequences downstream of the TATA box, potential start sites pass through the active site (Fig 2). Under GTP-replete or UTP-replete conditions, upstream start sites predominate, whereas depletion of the specific NTP leads to initiation at the downstream site. The upstream IMD2 transcripts begin with the dinucleotide sequence GG, which led Steinmetz and colleagues to propose that the concentration of initiating nucleotide is an important determinant of start-site selection (Steinmetz et al, 2006). Recent results from the Brow laboratory show that mutations from G to A in position +1 result in the production of upstream transcripts in low-guanine conditions (Kuehner & Brow, 2008), which supports this hypothesis. When there is insufficient GTP to form the first phosphodiester bond, as would be necessary in the case of IMD2, the polymerase continues to scan downstream, eventually finding the mRNA start site. In the case of URA2, the situation is more complex: the upstream transcripts start with an A, which would argue against its regulation by the initiating NTP (Kwapisz et al, 2008; Thiebaut et al, 2008). However, the first few nucleotides of these transcripts are U-rich, which might trigger pausing and allow time for short abortive transcripts to be released from the PIC during scanning.

Figure 2.

The pre-initiation complex scans the downstream promoter DNA looking for an initiation site (indicated by a question mark). bp, base pairs; mRNA, messenger RNA; PIC, pre-initiation complex; Ter, non-poly(A) terminator.

Regulatory non-coding RNAs have been described previously in yeast. For example, the non-coding SRG1 transcript negatively regulates expression from the downstream SER3 gene by destabilizing the assembly of a PIC on the downstream SER3 promoter (Martens et al, 2004). This is not the case for URA2, however, as expression levels of both upstream and downstream transcripts are reduced by the mutation of an upstream TATA box (Kwapisz et al, 2008). In addition, Pol II is present at the URA2 promoter even under repressing conditions (Kwapisz et al, 2008). A single TATA box has been identified for IMD2; however, mutations of this element reveal the presence of downstream cryptic TATA-like sequences (Kuehner & Brow, 2008). The increase in downstream start sites seen when the regulatory NTP is depleted does not seem to come at the expense of upstream start sites, indicating that either Pol II is always scanning downstream and initiates there only when NTP levels drop or that there is increased recruitment to the promoter. In the first case, an unknown signal might stabilize the PIC at the downstream start site. Evidence for the second possibility was provided by experiments that showed a fivefold increase in IMD2 mRNA even when the upstream start sites and terminator were deleted. Therefore, additional regulatory steps that involve the recruitment and/or stabilization of the PIC might be required for the regulation of IMD2 and URA2.

The way in which Pol II scans for an initiation site is unknown. One model proposes that the transcription factor IIH (TFIIH) helicase activity pumps DNA downstream of the TATA box into the active site (Hampsey, 2006; Miller & Hahn, 2006; Roberts, 2006). The central cleft of the polymerase cannot accommodate a transcription bubble of more than 20 bases, and therefore it is thought that the single-stranded DNA loops out of the central cleft at the upstream end of the transcription bubble (Fig 2). This model is consistent with the idea that Pol II maintains contact with the general transcription factors (GTFs) during the scanning process, and with permanganate-footprinting results that show the melting of promoter DNA extending from 20 bp downstream of the TATA box to more than 100 bases downstream (Giardina & Lis, 1993). Although considerable energy is needed to unwind such an extended transcription bubble and to scan for distal start sites, some could be stored in the stressed DNA loops that extend from the polymerase active-site cleft.

In addition to the potential initiator sequences, both Pol II and the GTFs TFIIF and TFIIB are important for start-site selection (Hampsey, 2006). Several rpb1 mutations near the active site reduce upstream start sites and increase downstream start sites, resulting in constitutive expression of URA2 and IMD2 (Kuehner & Brow, 2008; Kwapisz et al, 2008; Thiebaut et al, 2008). Similar results were obtained with mutations in the TFIIB B finger (Kuehner & Brow, 2008; Thiebaut et al, 2008). By contrast, deletion of the non-essential RPB9 subunit gene results in a shift to upstream start sites, thereby preventing the induction of IMD2 (Jenks et al, 2008). Taken together, these results argue that the PIC acts as a sensor that monitors NTP concentration and uses this information during scanning to find the appropriate start site. Kwapisz and colleagues propose that the Switch 1 loop—located at the downstream end of the transcription bubble—acts as the sensor (Kwapisz et al, 2008). This loop is positioned in the DNA-binding channel rather than in the funnel through which NTPs are thought to enter the active site (Cramer et al, 2001), although NTPs have also been proposed to enter the active site through the DNA channel (Gong et al, 2005), in which case the Switch 1 loop would be well situated to act as an NTP sensor.

The regulation of nucleotide synthesis by start-site selection has previously been observed in bacteria (Sorensen & Neuhard, 1991; Wilson et al, 1992). Transcription from the PyrC promoter initiates at a C residue when pyrimidine levels are high; however, these transcripts are not translated owing to a hairpin that forms at the 5′ end and blocks translation. When pyrimidine levels are low, transcription initiates at a site several nucleotides downstream, and the shorter mRNA, which is unable to form a stable hairpin, is translated. In principle, this is the same as the mechanism used by IMD2 and URA2; however, in yeast the alternative start sites are much further apart. In addition, the alternative RNA sequences included in the upstream transcript have different downstream targets: the ribosome in the bacterial case and the non-poly(A)-termination machinery for the yeast genes.

Given the diversity in yeast start sites, it is not surprising that several of the four articles highlighted here offer preliminary evidence indicating that other yeast genes are regulated by alternative start sites. URA8 and ADE12 are crucial enzymes in the synthesis of CTP and ATP, respectively, and both genes have upstream non-coding transcripts (Thiebaut et al, 2008). This indicates that the synthesis of all four NTPs is regulated by a similar feedback mechanism. From the genomic analysis of known GTP starts, Kuehner and Brow propose that genes in the yeast ribosome-biosynthetic pathway are also regulated by the availability of GTP. Although there is no evidence of short non-coding RNAs to support this model, the authors note that the regulation by availability of the initiating NTP is similar to the regulation of ribosomal RNA synthesis in Escherichia coli (Gaal et al, 1997). Further analysis of non-coding RNAs in yeast and other organisms is bound to lead to more examples of regulation by start-site selection.