The process of making RNA obliges the DNA-dependent RNA polymerases to perform a remarkable number of separate tasks:
They scan duplex DNA to find the sites for initiating transcription.
At these sites, the promoters, they select the DNA strand that will serve as the transcription template and expose it for base-paired binding of their ribonucleoside triphosphate substrates.
They polymerize RNA chains through distinctive processes of initiation, promoter escape and elongation.
They discriminate against the incorporation of deoxyribonucleotides into RNA.
They recognize specific DNA sites for disassembling the elongating transcription complex, releasing the nascent transcript and disengaging from the DNA template.
They also do many things that are less obviously intrinsic to a nucleic acid polymerase:
They abortively and repetitively initiate transcription at the promoter, generating short ribo-oligonucleotides.
They elongate RNA chains at non-uniform rates, pausing at characteristic sites.
They detect incorrect nucleotide addition to the elongating transcript and increase the accuracy of transcription by proofreading.
They set intrinsically unequal rates of transcription at different promoters by DNA sequence-determined variations of binding affinity, rates of promoter opening and ease of escape from the promoter.
They exert force on transcription-obstructing DNA-bound proteins, clearing them out of their path as they elongate RNA chains.
They exert force on their DNA templates, with diverse consequences for the architecture of the eukaryotic nucleus and the prokaryotic nucleoid/chromoid.
They monitor their DNA templates for damage.
They recruit accessory proteins for several of these activities, particularly transcript elongation, termination and DNA repair, and they also interact with other molecular machines in order to couple transcription with subsequent RNA processing.
This is an exciting time for understanding and admiring transcription in its machine-like and mechanistic terms. Within the past approximately 10 years, the determination of the structure of the multisubunit RNA polymerases has profoundly transformed the way in which ideas about transcription mechanisms are formulated and tests of these ideas are designed. The determination of the structure of the eukaryotic RNA polymerase II from budding yeast and, most recently, of an archaeal RNA polymerase has made the common evolutionary roots of the multisubunit enzymes vividly apparent. At the same time, comparison of structures of single-subunit nucleic acid polymerases with multisubunit RNA polymerases establishes the existence of catalytic mechanisms that are common to all nucleic acid polymerizations.
During the same period, the spectacular development of methods for examining single molecules of RNA polymerase in action has opened up entirely new possibilities for probing the mechanical and motor-like aspects of RNA polymerases. The process of directly observing RNA synthesis one molecule at a time reveals insights that are difficult to retrieve from, or are entirely obscured in, observations of ensembles in bulk solution. At the same time, technical advances have significantly increased the power of longer established analytical approaches (e.g., fast reaction kinetics).
This book presents a synthesis of these streams of endeavor. Overview chapters that focus on the mechanism–structure interface and the structure– machine interface introduce the two sections of the book, while individual chapters within each section concentrate more specifically on particular processes – kinetic analysis, single-molecule spectroscopy, and termination of transcription, for example. Seen from the perspective of (nearly) 50 years ago, the detail in which every step of transcription is currently understood is remarkable, and the ways in which that detail illuminates every aspect of gene regulation is enormously satisfying. From the current perspective, the dominant sense is of unanswered questions, of experiments addressing key aspects of mechanism that remain open to competing interpretation, of further technical development that would yield insights currently just out of reach, of molecular computations that have not yet been done – i.e., of a work in progress and a field of activity urgently engaged in finding fascinating new questions to answer.
The current understanding of the mechanistic and machine-like aspects of transcription has been formed primarily through work with the bacterial RNA polymerases. While the common evolutionary and mechanistic basis of all transcription can now be appreciated at the structural level, research on the eukaryotes has been dominated by the challenge of enumerating and understanding elaborations of the core transcription machinery with extrinsic initiation, elongation and termination factors and complexes (the core transcription initiation factors of budding yeast alone comprise nearly 50 polypeptide chains with an aggregate mass of more than 2.5 × 106 MDa), the dominant role of chromatin structure and modification in regulation of transcription, direct coupling of transcription with post-transcriptional RNA processing and, recently, the role of small RNAs in these processes. Many questions about the RNA polymerase machine that are specific to eukaryotic transcription, especially transcription of nucleosomal chromatin and the important role of elongation factors, are open to lines of experimentation and analysis that are described here for the bacterial enzymes. For these lines of inquiry, the work on bacterial transcription that is presented here points to, and lights up, the path.
A Foreword is a good place to relate how and where this very large endeavor started. The activity of DNA-dependent RNA polymerase was discovered at the University of Chicago's Argonne Cancer Research Hospital in 1959 and announced in a brief note in the Journal of the American Chemical Society in August of that year by S. B. Weiss and his assistant L. Gladstone. They showed that incorporation of 32P-labeled CTP into acid-insoluble (i.e., polymeric) form in a preparation of rat liver nuclei required all four ribo NTPs, ATP, GTP, CTP and UTP. The product of their synthesis was degraded by pancreatic ribonuclease but not by deoxyribonuclease I. Moreover, degradation with alkali of the radioactive product made with CTP 32P-labeled in the alpha position distributed radioactivity to all four ribonucleoside 2′ and 3′ monophosphates. This implied the synthesis of RNA polymers of complex sequence, as opposed to the mere addition of CMP to the ends of nucleic acid chains, either adventitiously to DNA or as the matured CCA adduct of tRNAs. RNA synthesis was strongly inhibited by pyrophosphate but indifferent to orthophosphate, distinguishing the enzyme from polynucleotide phosphorylase. The role of DNA in RNA synthesis was, however, not resolved.
A year later, A. Stevens, then a postdoctoral fellow at NIH, and J. Hurwitz, A. Bresler and R. Diringer at the NYU School of Medicine separately announced the existence of a comparable activity in extracts of Escherichia coli. RNA synthesis by the abundantly active bacterial extracts was readily shown to be profoundly dependent on DNA. In contrast, Weiss could not separate the mammalian RNA polymerase activity from DNA and also turned to work with a bacterial enzyme (from Micrococcus luteus). With the bacterial preparations, J. J. Furth, Hurwitz and M. Goldman and also A. Stevens, followed by Weiss and T. Nakamoto soon showed the correspondence of the relative incorporation of (AMP+UMP) to (GMP+CMP) into synthesized RNA with the guanine- cytosine content of added DNA. Weiss and Nakamoto extended the analysis of the DNA template–RNA product relationship to the level of nearest neighbor nucleotide pairs by essentially copying an elegant analytical strategy devised for DNA polymerase (by J. Josse, A. D. Kaiser and A. Kornberg) that had been published just months before.
Using CsCl density centrifugation, B. D. Hall and S. Spiegelman had just provided the definitive proof that RNA made in phage T2-infected E. coli was specific to the infecting virus by showing that it was able to form DNA–RNA hybrid duplexes with phage DNA. Weiss, Nakamoto and I adopted this approach to show that the RNA synthesized in vitro by the bacterial RNA polymerase generated a polymeric product that was fully complementary to the eliciting double-stranded T2 phage DNA and corresponded, in that sense, to RNA made in the phage-infected cell. This (simple) experiment showed that DNA, which was still widely referred to as the “primer” of RNA synthesis, in fact was its template. The same series of experiments yielded the additional information that the newly synthesized RNA was released from its template and that transcription did not separate template DNA strands. However, both strands of T2 DNA were transcribed, yielding RNA that was self-complementary. In similar experiments with the RNA polymerase activity from E. coli and phage ϕX174 DNA, M. N. Hayashi and S. Spiegelman at the University of Illinois, as well as M. Chamberlin and P. Berg at Stanford, also found both strands of their phage DNA templates transcribed. In contrast, the RNA isolated from cells infected with diverse phages was soon found to be DNA-strand-selected, or “asymmetric,” which is consistent with the requirements of messenger RNA function in instructing protein synthesis. Was something missing from the in vitro RNA synthesis system, or was it conceivable that mRNA might have to be selected after transcription by an additional process, with unusable transcripts rapidly disposed of? The first alternative implied an ability to select specific sites on DNA for starting transcription; the alternative hypothesis clearly failed on two counts: it was too elaborate and inelegant to be plausible, and it implied that transcription yields its functional products at the cost of an energy-consuming futile cycle. Thus, the hunt for DNA strand-selective “asymmetric” transcription was on. Of course, it can now be appreciated that the dichotomy was, to some extent, false. The human genome is pervasively transcribed, with both complementary DNA strands frequently serving as transcription templates. Moreover, cellular processes for very fast disposal of nonfunctional/unusable transcripts do exist.
In any case, the question of template strand selection in transcription was soon answered by J. Marmur's group at Brandeis, Hayashi and Spiegelman, as well as G. P. Tocchini-Valentini and co-workers at Chicago and the International Laboratory of Genetics and Biophysics in Naples, with bacterial RNA polymerase preparations that yielded strand-selective transcription, implying the ability of the enzyme to select specific DNA sites for production of RNA in vitro. In hindsight, the something missing from, or inactivated in, the polymerase preparations used for the initial experiments must have been σ, the initiation-specific subunit of bacterial RNA polymerases, discovered several years later by R. R. Burgess and A. A. Travers at Harvard, and E. K. F. Bautz and J. J. Dunn at Rutgers. Within the next year, partially purified bacterial RNA polymerase had been prepared in several laboratories, including those already referred to, W. Zillig's group and others. Those first experiments also provided what we now understand to have been a straightforward and simple demonstration of the existence of genes under positive transcriptional control: “asymmetric” transcription of phage T4 DNA, and the DNA of large-tailed phages infecting Bacillus subtilis, selectively yielded transcripts that correspond to RNA produced at the outset of phage infection, the so-called early RNA. This is, in a sense, the historical baseline for the “modern synthesis” represented by this book.
E. P. Geiduschek
Division of Biological Sciences, UCSD, La Jolla