Key words

1 Cryo-Electron Microscopy : The New Standard in Transcription Research

The visualization of macromolecular complexes is essential to our understanding of their function. This is especially true for eukaryotic RNA polymerases (Pol) I, II, and III. These enzymes play a pivotal role within the central dogma of molecular biology by synthesizing the 35S ribosomal RNA precursor (Pol I), messenger and many noncoding RNAs (Pol II), and tRNAs , 5S rRNA , U6 snRNA as well as other small, structured RNAs (Pol III).

Almost 20 years ago, advances in cryo-crystallography allowed solving the structure of RNA polymerase II , first in its 10-subunit form [1], later comprising all 12 subunits [2], providing insights into the function of this molecular machine at an unprecedented level of detail. This gave rise to a number of follow-up studies, resulting in structures of an actively elongating Pol II form [3, 4] and elongation [5, 6] or initiation [7, 8] factor–bound Pol II. Thereby, the molecular mechanisms of transcription , as well as regulatory and catalytic functions of transcription factors , could be deciphered.

However, X-ray diffraction analysis relies on the availability and quality of the analyzed crystals. It is therefore not surprising that a number of transcription factors could never be successfully studied in complex with their respective polymerase by crystallography. This includes the general Pol II initiation factors TFIIF and TFIIE, which are involved in initiation complex formation and promoter DNA melting [9]. A crystal structure of Pol I was solved 10 years after Pol II in an inactive conformation [10, 11], a Pol III crystal structure is still lacking to date. Crystallization depends on high amounts of purified material. Whereas conformational heterogeneity usually is problematic for crystal formation, cryo-EM allows for visualization of different functional states of proteins in vitrified ice, therefore capturing close-to-native states. Technical improvements such as the development of direct electron detectors [10], highly stable microscopes and improved processing software [11, 12] pushed previous limitations of the technique and led to the often quoted “resolution revolution” in cryo-EM [13]. Within this book, we describe protocols for a sample preparation and single-particle cryo-EM screening workflow, adapted to RNA polymerase complexes (Chapter 6 ).

Here, we aim to briefly outline the benefits and limitations of single-particle cryo-EM for the analysis of transcription complexes at the example of RNA polymerase I in the context of functional characterization reviewed by Merkl et al. in the same issue.

2 Pol I Specific Subunits Resemble Built-in Transcription Factors

Yeast Pol I has a molecular weight of 590 kDa and consists of 14 protein subunits. A core of ten subunits includes the large subunits A190 and A135, the subcomplex AC40/AC19 (shared with Pol III), the common subunits Rpb5 , Rpb6 , Rpb8 , Rpb10 , and Rpb12 (also included in Pol II and Pol III) and the subunit A12.2. Pol I also contains the specific subunit complex A14/A43 forming the “stalk” and the specific heterodimer subcomplex A49/A34.5. In Fig. 1, we present a “hybrid model” of Pol I, constructed using the software package COOT [14]. The model combines structural information obtained from cryo-EM reconstructions of elongation and initiation complexes and the dimer crystal structure (see below). The Pol I subunit complex A49/A34.5 structurally and functionally resembles a built-in version of subunits Tfg1/2 constituting the Pol II initiation factor TFIIF [15] and is involved in open complex stabilization as well as promoter escape. Nevertheless, subcomplex A49/A34.5 may also contribute to transcript cleavage activity [16] and may play a role in elongation [17], as reviewed by Merkl et al. in this book. The Pol I subunit A12.2 displays features of the Pol II subunit Rpb9 , as well as the Pol II elongation /cleavage factor TFIIS , explaining the intrinsic ability of Pol I for transcript cleavage [18] and efficient recovery from deep backtracks [19].

Fig. 1
figure 1

Ribbon model of RNA polymerase I highlighting specific subunits and –domains. Hybrid model constructed using COOT [14] for demonstration purposes. The structure of the Pol I elongation complex (PDB 5M3F) was extended by adding (a) a model of the C-terminal domain of subunit A12.2, (b) the “connector” domain of subunit A43 from the crystal structure (both from PDB 4C2M) and (c) the linker and tandem-winged helix domains [15] from an ITC reconstruction (PDB 5 W66). The “front view” looks along the incoming (“downstream”) DNA. Subunits not visible in the front view but present in the model are AC19, Rpb10 , and Rpb12 . Cyan spheres depict coordinated zinc atoms

Full size image

The constitutive association of subunits A12.2 and A49/A34.5 may be a symptom of the enzyme’s adaptation to transcribing one specific, extraordinarily long gene. A differential regulation of Pol I transcription similar to the conditional, gene-specific regulation of Pols II and III is most likely not required. Instead, a binary activity switch can be achieved by posttranslational modification of the polymerase or its initiation factors [20,21,22,23].

3 The Pol I Transcription Cycle Visualized In Vitro

Throughout transcription of a DNA template, Pols pass through three main phases, which are more or less conserved throughout various organisms. First, the Pol is recruited to the template sequence, melts the DNA duplex with or without the help of additional factors and begins the synthesis of an RNA strand (initiation ). Thereafter, the initial RNA is extended according to the DNA template strand (elongation ). Finally, Pol dissociates from its DNA template and releases its RNA product (termination ) [24].

The Pol I transcription cycle requires few basic transcription factors [25, 26]. Basal initiation in yeast can be achieved with only the factor Rrn3 and the Core Factor (CF) complex , consisting of proteins Rrn6 , Rrn7 , and Rrn11 [27, 28]. Elongation may involve regulatory factors in vivo, but can commence in vitro without them [25, 26]. Termination finally requires a Myb-domain containing protein, Nsi1 in yeast [29, 30]. Pol I can adopt a dimeric state specific to this enzyme, [31], in which both molecules are inactivated [32] and prevented from binding initiation factors Rrn3 and CF [33,34,35]. Dimerization is reversible [33] and may play a role in storage of Pol I molecules during starvation [36, 37]. Using cryo-EM in many variations, our understanding of the Pol I transcription cycle has been significantly improved in recent years. A structural description of the Pol I transcription cycle allowed a detailed structure–function analysis and the comparison to other transcription systems [38,39,40,41].

Specifically, single-particle cryo-EM analyses showed how Pol I monomers are bound by the initiation factor Rrn3 [42], which allows for recruitment of CF subunit Rrn7 [37, 43, 44] that is related to the general Pol II initiation factor TFIIB [45, 46]. The interaction with Rrn3 prevents the formation of transcriptionally inactive Pol I dimers. In baker’s yeast (Saccharomyces cerevisiae ) Pol I dimerization depends on the specialized, nonconserved “connector” domain of subunit A43, suggesting that this mode of regulation is specific to S. cerevisiae [37]. However, recent cryo-EM studies of Pol I from fission yeast (Schizosaccharomyces pombe ) demonstrated that inactive dimers can form independent of the connector domain utilizing divergent structure elements [47]. It is therefore possible that different organisms evolved different dimer interfaces, but use the same strategy of Pol I hibernation by dimerization during periods of starvation.

Monomeric Pol I bound to Rrn3 can be recruited by promoter-engaged CF . Whereas the structure of CF was determined by X-ray crystallography [48], its interaction with the Pol I–Rrn3 complex was apparently too flexible for this method to succeed. Instead, three independent studies used cryo-EM to visualize a reconstituted initially transcribing complex (ITC) that relies on an artificially stabilized, mismatched transcription bubble with a short initial product RNA [48,49,50]. Comparison of these studies shows that seemingly similar experimental cryo-EM approaches may yield different, though highly complementary results. This demonstrates the influence of sample preparation and experimental conditions on the outcome of a cryo-EM experiment and the strength of single-particle cryo-EM to elucidate dynamic processes. Specifically, some complexes exist in conformations that fail to bind Rrn3 [49], which is required essential to the initiation process and dissociates after promoter escape [42, 51, 52]. These complexes may represent later initiation intermediates. In another structure, the RNA primer is lost and the C-terminal domain of subunit A12.2 is inserted into the funnel domain of Pol I, as would be expected during RNA cleavage events [50]. A third reconstruction showed local resolution differences suggesting high flexibility in distal CF regions [48]. More recently, consolidating structures of close-to-native preinitiation complexes were reported [53, 54]. A closed and an open complex could be reconstructed from a small fraction of recorded particles but yielded important information about the roles of a flexible loop in Rrn3 in CF engagement, and the role of the linker/tWH domain of subunit A49 in template DNA melting, open bubble stabilization, and Rrn3 dissociation [53].

Following initiation , Pol I adopts an actively elongating conformation. This conformation was never successfully crystallized but was reconstructed by single-particle cryo-EM [55, 56]. It is still under debate, whether dissociation/transient association of the A49/A34.5 heterodimer and formation of a 12-subunit polymerase has a physiological function under some circumstances. Chromatin immunoprecipitation (ChIP ) data does not suggest subunit depletion along the 35S gene body in vivo [57], although a lack of density in cryo-EM reconstructions of S. cerevisiae and S. pombe Pol I ECs suggests that subdomain relocation can take place [47, 58]. Upon initiation , the central DNA-binding cleft contracts and tightly binds downstream DNA and the DNA–RNA hybrid region, thus facilitating processive elongation . Contraction upon activation is apparently common to Pol I regulation in different organisms [47] but differs from Pol II [3] and III [59]. The various Pol I EC reconstructions demonstrated the versatile nature of the specific subunits A12.2, A49, and A34.5 [47, 55, 56, 58], and the importance of a Pol I-specific arginine residue in the “bridge-helix” in stalling at DNA lesions caused by UV radiation [60]. Furthermore, the physiological relevance of a contracted Pol I state was confirmed by the 3D-reconstruction of the enzyme from ex vivo purified, actively transcribing Pol I using electron cryo-tomography [55]. Interestingly enough, the underlying preparation technique has been used to describe Pol I functionality since the 1960s [61].

Structural information on transient and highly dynamic states of the Pol I transcription cycle , such as termination , are still lacking to date.

4 Outlook

Continuing investigation of the Pol I transcription system currently aims at defining the structural basis of promoter recruitment in the context of TATA-binding protein [62] and yeast upstream activating factor (UAF) which are important to achieve full initiation levels under physiological conditions [27, 63]. The structural investigations of these complexes and their interplay with the Pol I enzyme will be the key to understand this initiation systems in comparison with Pol II [64, 65] and Pol III [66, 67]. Ultimately, such comparative structure–function analyses will enable us to understand the particular adaptation of the Pol I transcription system to pre-rRNA transcription .

Cryo-EM has now firmly established itself as the method of choice to analyze the structure of such mega-Dalton sized, dynamic, macromolecular complexes. However, the method has also proven to be prone to certain errors. Especially the lack of tools for intrinsic validation of density interpretation may be an issue for initial sequence assignment at resolutions worse than 4 Å. Continuing method development at the experimental and the computational level continues and may soon overcome issues of flexibility and density assignment. For the time being, however, the interpretation of cryo-EM reconstructions should be carefully evaluated and supported by functional data or mutational analysis. Results can also be strengthened by crystallization or small angle X-ray scattering analysis of flexible domains [68], native mass spectrometry [15], hydrogen–deuterium exchange mass-spectrometry evaluation [69], or single-molecule FRET analysis [70]. Especially distance restrains obtained from protein crosslinking coupled to mass spectrometry provides architectural information and may assist low-resolution density assignment [71, 72].