Key words

1 RNA Polymerase I Has Only One Essential Genomic Target

Nuclear yeast RNAPs are protein complexes consisting of 12 (RNAP II ), 14 (RNAP I ) and 17 (RNAP III ) subunits. They all share a conserved architecture of the RNAP core, which catalyzes the highly accurate polymerization of RNA from single NTP molecules [1,2,3] (see review by Pilsl and Engel in this issue). Based on structural analyses and on functional in vitro assays, on the one hand the molecular mechanisms driving the RNA polymerization appear to be similar in all RNAPs. On the other hand, all three enzymes have special features, supporting their specific in vivo tasks [1, 2]. Thus, RNAP II has to transcribe thousands of different genes and produces transcripts from a few hundred to several thousand of bases in length. To account for dynamic changes in cellular gene expression RNAP II has (a) to recognize many promoters, (b) to access differently modified chromatin templates, which probably requires to adjust its elongation and termination properties. To fulfill these multiple tasks, RNAP II interacts with many different factors at each step of the transcription process [4, 5]. In contrast, RNAP III recognizes only three distinct classes of promoters and has probably a distinct transcription termination mechanism [6]. RNAP III synthesizes mainly short noncoding RNAs (typically <200 bp) with high efficiency in rapidly growing cells. Accordingly, the RNAP III transcription machinery is rather well defined [7,8,9,10]. Finally, the yeast RNAP I transcription machinery has only one known genomic target, the multicopy 9.1 kb 35S rRNA genes [11]. The rRNA genes are transcribed at very high rates accounting for up to 60% of RNA synthesis upon cellular growth [12]. Accordingly, electron micrographs of chromatin spreads show an extremely high density of RNAP I molecules at rRNA genes, whereas most RNAP II-dependent genes are only sparsely covered with polymerases [13,14,15]. Furthermore, RNAP I and RNAP III-dependent genes are constitutively transcribed in actively dividing cells and—as opposed to the majority of RNAP II transcribed genes—apparently devoid of nucleosomes (see as review [16,17,18] and Schächner et al., within this issue).

2 RNA Polymerase I Contains Additional Subunits Resembling Transcription Factors of RNAP II

RNAPs I, II, and III contain ten conserved subunits which form the catalytic core . In addition, all three enzymes contain a 2-subunit stalk structure, which is distantly related and consists of subunits A14/A43, Rpb4/Rpb7 and C17/C25 in RNAPs I, II and III, respectively. In RNAP III an additional heterotrimer C82/C34/C3 connects the stalk with the RNAP III clamp , which probably helps to open the DNA duplex [19]. The overall architecture of the three RNAPs differs mainly in vicinity of the lobe structure, which is formed by the second largest subunits Rpa135 , Rpb2 and Rpc128 , respectively. Only one subunit—Rpb9 —is bound to the RNAP II lobe, whereas the heterodimer A34.5/49 and subunit A12.2 bind to the lobe of RNAP I , and the homologous C17/C25 and C11 subunits to the lobe of RNAP III . RNAP I subunits A34.5 and A49 consist of three subdomains: a dimerization module formed by A34.5 and the N-terminal part of A49 (full length A34.5 and aa 1–110 of A49); the A49 linker (aa 105–187 of A49); and the C-terminal part of A49 (aa 187–415). The dimerization module binds to the “lobe” and “external” domains of the second largest Pol I subunit A135 on the core module side [20,21,22]. In contrast, the C-terminal part of A49 which contains a tandem winged helix can be detected at the upstream face of the clamp core in several states of transcriptional active RNAP I molecules and seems to be flexible attached [20, 21, 23,24,25,26,27,28,29,30,31]. Biochemical and cell biological experiments showed that A34.5/A49 support transcription initiation , enhance RNAP I elongation and stimulate the intrinsic RNAP I RNA cleavage activity [26, 28, 32,33,34,35,36]. RNA cleavage activity depends on the dimerization module which is located in close proximity to A12.2 whose C-terminal part is also important for efficient RNA cleavage [26]. Deletion of A12.2 results in growth inhibition at elevated temperature, sensitivity to nucleotide-reducing drugs, and inefficient transcription termination ; hampers the assembly of the RNAP I enzyme; and leads to incorporation of wrong NTPs [37,38,39,40]. The lack of A12.2 may also lead to the loss of A34.5/A49, and might influence the intrinsic stability of elongation and termination complexes [40, 41].

Based on amino acid sequence similarities, position on the enzyme and function, the heterodimer formed by A34.5 and the N-terminus of A49 was suggested to be homologous to the RNAP II transcription factor TFIIF [26, 33]. TFIIF is predominantly found at promoter-proximal regions suggesting a crucial role in transcription initiation [42, 43]. On the other hand, TFIIF was suggested to leave the promoter—at least transiently—in complex with RNAP II , likely supporting early elongation [42,43,44,45]. A role of TFIIF in RNAP II elongation is further corroborated by in vitro studies, where it increases transcription rates by suppressing RNAP II pausing [46,47,48]. Several studies propose that TFIIF promotes transcription elongation in concert with the RNA cleavage supporting factor TFIIS , which structurally resembles the C-terminus of the RNAP I subunit A12.2 (reviewed in [3, 48,49,50,51]. In the RNAP II system, TFIIF and TFIIS are independently capable to release arrested RNAP II to resume productive elongation . However, these factors may synergistically enhance resumption of RNAP II transcription especially in conditions when the paused enzyme has additionally backtracked on the template [49]. Backtracked RNAP I requires the C-terminal, RNA cleavage activating part of A12.2 to resume elongation [52]. It is, however, unknown if the heterodimer A34.5/A49 participates in this process.

Finally, the C-terminus of A49 structurally and functionally resembles the tandem winged helix of TFIIE [33]. As TFIIE in RNAP II transcription , the C-terminal domain of A49 binds DNA and supports promoter-dependent transcription initiation in vitro [28] and RNAP I promoter recruitment in vivo [32]. In contrast to TFIIE, the A49 subunit stays associated with the enzyme after promoter clearance in vivo [32], and supports elongation of RNA from a DNA/RNA scaffold in vitro [33]. Recent studies suggest a more dynamic association of the heterodimer A34.5/A49 to the RNAP I lobe, since the heterodimer was absent from the core enzyme and A12.2 C-terminus was rearranged when RNAP I elongation was artificially blocked by addition of a nonhydrolyzable nucleotide [53].

A specific challenge for all elongating RNAPs is the transcription of chromatin templates. Since RNAP I and III transcribe nucleosome-depleted chromatin templates ( [13,14,15] see short review of Schächner et al. this issue). This indicates that nucleosomes are displaced from the chromatin template in the initial round of transcription , and it is possible that the lobe associated subunits of RNAP I and III may be involved in the process of nucleosome depletion.

3 Nuclear RNAPs Transcribe Chromatin Templates

In eukaryotic cells, nuclear DNA is assembled into repeated units called nucleosomes consisting of 146 bp of DNA wrapped around an octameric complex of histone proteins [54]. Nucleosomes generally provide a strong barrier for elongating RNAP II in vitro [46, 55, 56]. DNA attached to nucleosomes recoils on the octamer, locking the enzyme in an arrested state [57] thereby providing four major superhelical pausing sites [58]. Additional factors are required for passage of RNAP II through this barrier (see below). The mechanism how purified RNAP II complexes passes nucleosomes in vitro was thoroughly studied [59, 60]. Whether assembled nucleosomes stay associated or are evicted during RNAP II transcription depends on the formation of a small intranucleosomal DNA loop and on the transcription efficiency (rate) [61]. Accordingly, various RNAP II complexes can remodel chromatin to a different extent [59, 60].

It was suggested, that RNAP II can only pass nucleosomes if uncoiling of the DNA from the surface of the octamer is facilitated and if transcription elongation factors keep the polymerase in a transcriptionally competent state [62]. TFIIF and TFIIS may prevent the release of RNAP II at nucleosomal barriers and thereby support transcription through a nucleosome in a synergistic manner [63]. Passage of purified RNAP II through in vitro assembled nucleosomes was also supported by elongation factor Spt4/Spt5 [64] and in the presence of TFIIS together with Spt4/5 and elongation factor Elf1 [65]. Insights in the molecular mechanism how Spt4/5 together with Elf1 facilitate progression of RNAP II through a nucleosome were recently obtained using high resolution structures by cryo-EM of different stalled elongation complexes [65]. Other factors that were reported to partially disassemble nucleosomes similar to Spt4/Spt5 are the histone chaperone FACT and Paf1c [66,67,68].

Much less is known about how RNAP I and RNAP III interact with nucleosomes in vitro. However, in vivo there is ample evidence that RNAP I and RNAP III genes are largely devoid of nucleosomes (reviewed in [16,17,18], see short review of Schächner et al. in this issue). It is an open question how RNAP I and RNAP III deal with nucleosomal genes in the initial round of transcription . In contrast to RNAP II which has only subunit Rpb9 associated to the lobe structure, yeast RNAP I and RNAP III have the TFIIF- and TFIIS-homologous subunits A34.5/A49, A12.2 (RNAP I ), and C37/C53, C11 (RNAP III ) tightly associated to the lobe. Similar to TFIIF and TFIIS in RNAP II transcription , the homologous Pol I subunits A34.5/A49 and A12.2 facilitate RNAP I passage through nucleosomes [35]. Depletion of either the heterodimeric subunits A34.5/A49 or the cleavage supporting activity of A12.2 resulted in both reduced Pol I processivity and impaired passage though nucleosomes [35]. It is tempting to speculate that the homologous RNAP III subunits could play a similar role in RNAP III chromatin transcription . Furthermore, additional factors like FACT , Spt4/Spt5, or Paf1c have all been suggested to support RNAP I elongation [69,70,71,72]. Appropriate in vitro transcription system using highly purified factors and defined nucleosome templates will be the key to elucidate details of the molecular mechanisms of RNAP I and RNAP III transcription in the context of chromatin (see chapter by Merkl et al. in this issue).