Structural biology: Pivotal findings for a transcription machine.

RNA polymerases are intricate molecular machines that transcribe DNA into RNA, combining RNA synthesis with the precise movement of a DNA template across their active site. Eukaryotic cells (those of animals, plants and fungi) have several RNA polymerases, each dedicated to the production of specific RNAs. RNA polymerase I (Pol I) synthesizes the ribosomal RNA component of the cell's protein-producing factories and so is crucial for cell survival, growth and proliferation; malfunction of Pol I can cause cell death or support the unrestrained proliferation characteristic of cancer cells1. In two groundbreaking papers in this issue, Fernández-Tornero et al.2 (page 644) and Engel et al.3 (page 650) present the first crystal structures of the complete 14-subunit yeast Pol I at 3.0 and 2.8 ångströms resolution, respectively. The structures provide unprecedented insight into Pol I-specific features, potential mechanisms in transcription, and evolutionary conservation of the structures and functions of Pol enzymes*. Structural analyses of bacterial Pol, eukaryotic Pol II and archaeal Pol have detailed the architecture of these enzymes, the interactions between their subunits and their inner workings during transcription4. Pol I and Pol III share overall architecture with Pol II5. However, each Pol contains specific subcomplexes and features that influence its ability to transcribe a particular subset of genes: whereas Pol I produces rRNAs, Pol II generates messenger RNAs and Pol III synthesizes small non-coding RNAs, including transfer RNAs and 5S rRNA6. Pol I is the most productive of the eukaryotic polymerases. To achieve high-throughput transcription, multiple Pol I complexes transcribe the ribosomal DNA, and these are densely packed along each template and are highly processive (likely to traverse the entire template). The Pol I crystal structures reveal various distinguishing features that have the potential to influence the enzyme's output, partly by facilitating its productive association with the DNA template. The structures confirm that the zinc-ribbon domain at the carboxy terminus of Pol I subunit A12.2 inserts into, and forms an integral part of, the enzyme's active-site region7 (Fig. 1). (By contrast, TFIIS, the functional counterpart of A12.2 in Pol II, only transiently associates with the active site of paused Pol II.) Within the active site, this zinc ribbon can stimulate the removal of faulty and redundant RNA sequences to prevent Pol I arrest and consequent 'traffic jams' along the template, thus increasing transcription efficiency. The stability of A12.2 in Pol I is influenced by its interaction with the (TFIIF-like) dimerization domain of the Pol I-specific subcomplex A49–A34.5 (ref. 8). The structural data now rationalize this, revealing the contact points between A12.2 and the A49–A34.5 amino-terminal dimerization domain, as well as extensive interactions of the A34.5 C terminus as it wraps around the outer face of the A135 subunit, which help to anchor the subcomplex. Procession of RNA polymerases along a DNA template is facilitated by a closed-clamp component. The structures of Pol I reveal that the A43–A14 subcomplex, which comprises a fixed stalk (Fig. 1), contributes to a permanently closed state of the clamp and, therefore, to the high processivity of Pol I. By contrast, the clamps of other RNA polymerases are mobile elements. In Pol II, for example, attachment of the Rpb4–Rpb7 stalk locks the clamp in a closed state over the complex of RNA and DNA template during transcription, but this stalk is detachable9, 10. Intriguingly, both teams' crystals are dimers of Pol I, in which the stalk of each Pol I inserts into the DNA-binding cleft of the other Pol I, through the A43 C-terminal 'connector' domain, thus making extensive contacts with the cleft and the coiled-coil motif of the clamp. The dimers have an unusually wide cleft (Fig. 1), perhaps partly because of this A43-connector insertion. The cleft is too wide to anchor the RNA–DNA-template complex, particularly near the active site. This widening contributes to further rearrangements near the active site. (For example, crucial 'aspartate-loop' interactions are configured differently from those in Pol II; the 'bridge' helix contributing to DNA movement through the active site is unfolded in the middle and kinked; and there is partial blockage of the gate to the exit channel for newly synthesized RNA.) Furthermore, the wide cleft is occupied by a Pol I-specific extended loop of A190, which the authors refer to as the expander3 or DNA-mimicking loop2. Because of its location, this loop would interfere with DNA loading at the active site. In one of the three Pol I structures presented by Fernández-Tornero et al., no loop is detectable at the active site, hinting that it is unlikely to be essential for stabilization of the expanded cleft, although not excluding a role in its establishment. Fernández-Tornero and colleagues' crystals display varying degrees of cleft widening. Comparative structural modelling of RNA polymerases suggests that the Pol I cleft widens as a result of relative pivoting of 'core' and 'shelf' modules (which are formed mainly by the largest subunits, A135 and A190) at the base of the cleft, near the active site2, 3 (Fig. 1). Engel et al. draw parallels to similar domain pivoting in inhibitor-bound or paused bacterial Pol, in which a pivoted or ratcheted state is associated with cleft opening and coupled rearrangements of domains near the active centre, inactivating the polymerase11, 12. Because the new structures imply that the DNA template must be loaded into Pol I that has a closed clamp, perhaps the open or shut status of the cleft contributes to DNA-loading efficiency. It is possible that binding of the DNA template in the open cleft of a Pol I monomer triggers cleft closure, potentially coupled with relocation of the expander loop, rendering the enzyme active. Cleft closure by pivoting of the core and shelf modules presumably occurs concomitantly with refolding of the bridge helix, opening of the RNA-exit gate and the approach of A135 to anchor the DNA template in the active site. An understanding of the exact rearrangements will hinge on structural analysis of Pol I engaged in transcript elongation and, therefore, in complex with DNA and RNA. Engel et al. propose that regulatory factors binding at the core–shelf interface might facilitate cleft closure. They speculate that Rrn3 (a factor that tethers Pol I to proteins bound specifically to promoter DNA sequences) triggers cleft closure by binding Pol I near the RNA-exit channel3, 13. This attractive possibility awaits confirmation, perhaps through analysis of a Pol I–Rrn3 co-crystal. Conversely, factors that terminate transcription by Pol I might induce cleft opening. In all probability, the regulation of transcription by modulation of the core–shelf interface, which is seen in bacterial Pol, is also a feature of eukaryotic RNA polymerases3. Solving the crystal structure of the complete Pol I complex is a triumph, providing a wealth of information with which to build a picture of the specific mechanisms and control of rRNA-gene transcription in eukaryotes and also to explore the general mechanisms of transcription by all RNA polymerases. Another tour de force will be necessary to solve the structure of Pol I in transcription-elongation mode and, further, that of the complete Pol I pre-initiation complex, incorporating Rrn3, the core promoter-binding factors (Rrn6, Rrn7 and Rrn11 with TBP) and the rDNA promoter sequences. Such structures, together with those presented by Fernández-Tornero et al. and Engel et al., will yield information that is vital for establishing when and where crucial protein and DNA contacts are made, disrupted and rearranged as Pol I steps through the transcription cycle. Download references