Transcription, Initiation of

Transcription

1 Craig T. Martin and Olga S. Mironova, University of Massachusetts, Amherst, MA doi: / wecb606 Initiation of transcription is a complex process, requiring melting of the DNA duplex at a sequence-defined location, de novo synthesis of an initial dinucleotide, polymerization/growth from that dinucleotide to a larger oligonucleotide in an extended RNA-DNA hybrid, displacement and threading of the 5 end of the RNA into an exit channel, and release of the initial promoter contacts. These events are necessarily driven energetically by growth of the RNA-DNA hybrid. It has long been expected and is now becoming structurally clear that stress inherent in this transformation leads to instability during this phase and is a source of the abortive release of short RNA products that is characteristic of RNA polymerases. That apparently unrelated single and multisubunit RNA polymerases share extensive mechanistic (but not structural) homology, likely reflects the demands of the process. Advanced Article Article Contents Biological Background Challenges Inherent to Initiation Comparisons Yield Insights Chemical Tools and Techniques Promoter Binding Promoter Melting Template Positioning and De Novo Synthesis Abortive Cycling and Promoter Escape Conclusions While many enzymes bind substrate, catalyze a transformation in the substrate, and then release the resulting product, transcription is a process of repeated catalysis, marked at the beginning and end by very specific noncovalent transformations of both the DNA on which it acts and of the protein itself. Indeed, it is the processes at the beginning and end that define RNA polymerase as a unique molecular nanomachine. This review focuses on the mechanistic transformations occurring during initiation of transcription. In comparing this process in two apparently unrelated families of RNA polymerase, common features arise prominently. These commonalities almost certainly reflect the underlying requirements of the process itself. Biological Background Transcription is the critical first step in gene expression (1) and is carried out by DNA-dependent RNA polymerases, which synthesize RNA using DNA as a template. The transcription process consists of multiple steps, including recruitment of the enzyme to a unique site in the DNA (the promoter), de novo initiation of transcription at a specific site within the promoter, elongation of the RNA chain, and termination. In eukaryotes, access to individual promoters can be restricted by DNA superstructure, while in phages, access can be restricted by an ordered entry of the DNA into the cell. In either case, the polymerase must additionally recognize a specific site within the exposed DNA to initiate transcription. This stands in contrast to replication, where the DNA polymerase merely elongates a pre-formed primer, with no sequence specificity. Replication is also distributive that is, one DNA polymerase can extend a short stretch of DNA, dissociate, and then a different DNA polymerase can continue the synthesis. In transcription, the entire transcript is synthesized by one RNA polymerase, requiring that elongation complexes be exceptionally stable. After synthesis of the initial dinucleotide, however, RNA polymerase ternary complexes are inherently unstable and must synthesize a minimal length of RNA before the complex attains the stability characteristic of an elongation complex, a trait necessary for the processive synthesis of thousands of nucleotides of RNA. This review will consider initiation to begin with RNA polymerase binding to exposed promoter DNA and to be complete after the complex attains elongation complex stability and has lost its sequence specificity. Some nomenclature is important for the discussion of transcription complexes. As illustrated in Fig. 1, upstream refers to DNA toward the untranscribed (or recently transcribed) side of the complex, while downstream refers to DNA that is yet to be transcribed. The position of the initial encoding base is referred to as +1, while the base immediately upstream of this start site is 1. RNA polymerases use information from only one strand of the DNA duplex. This strand, to which incoming nucleoside triphosphates pair, is referred to as the template strand, while the strand that reads the same as the transcript, but serves no direct function, is referred to as the nontemplate strand. Template and nontemplate strands are also referred to as antisense and sense, respectively, while the terms coding and noncoding were used inconsistently in the early literature. Challenges Inherent to Initiation The first challenge for an RNA polymerase is to bind to the unique start site (the promoter) within long stretches of DNA. WILEY ENCYCLOPEDIA OF CHEMICAL BIOLOGY 2008, John Wiley & Sons, Inc. 1

2 nontemplate stand upstream downstream template strand 5 RNA active site Figure 1 Nomenclature in transcription. Illustration of an elongation complex transcribing to the right. An incoming nucleoside triphosphate is shown positioned at the active site. The DNA is melted (forming a bubble ) and the freed template strand is instead hybridized to the nascent RNA. Nucleotides are added to the 3 end of the RNA, leading to translocation to the right, as shown. In higher-level systems, initial control of transcription is exerted by sequestration of most of the DNA into large nucleosomal superstructures, inaccessible to RNA polymerase. Chromatin remodelling proteins then selectively and in a controlled fashion expose promoter-containing regions of DNA for the initiation of transcription (2). Once the promoter is available, the RNA polymerase must position itself more precisely at the promoter. There are countless examples of proteins that bind site-specifically to DNA and RNA polymerases appear to use similar mechanisms for sequence read-out. Additionally, RNA polymerase must bind (as a monomer) to a non-palindromic site and in order to initiate de novo synthesis, RNA polymerase must unwind (melt into) the DNA at the start site, directing the (single) template strand into the active site. Many years ago, it was noted that if RNA polymerase retains strong promoter contacts while synthesizing initial, short RNA products, then there would be a competition between foreword progression along the DNA and a return to the initial promoter bound state (3). Thus it was proposed that stress would accumulate in the complex, leading to growing instability during initial synthesis. Additionally, as noted above, early in the synthesis of RNA, the RNA primer forms only a very short RNA-DNA hybrid, and so is expected to be relatively weakly bound to the complex. Thus, one might expect the initiation phase of transcription to be inefficient, with the premature (abortive) release of short products and, indeed, this is observed to be the case. This process, known as abortive cycling, occurs in all RNA polymerases. Unlike DNA-dependent DNA polymerases, an RNA polymerase must initiate RNA synthesis in the absence of a primer (de novo), using mononucleoside triphosphates both as primer and as substrate, as shown in Fig. 2. Following the first covalent catalysis to form a dinucleotide, this small product is progressively lengthened. In subsequent reactions, the primer is then a dinucleotide, a trinucleotide, etc., such that the active site must accommodate a variety of substrate combinations at each step in early synthesis. Moreover, the enzyme must stabilize these primers, which inherently are only weakly bound to the template DNA. After this primer reaches a length of 8 10 nucleotides, the 5 end of the RNA must be dissociated from the template, to complete the transformation to an elongation complex. In the elongation phase of the overall transcription ppp de novo initiation elongation p p pp pp p pp elongating (+2) nucleotide priming nucleotide elongating (i+1) nucleotide priming (i ) nucleotide (+1) Figure 2 Initiation vs elongation. The initiating and elongating complexes are fundamentally different in the structure of the nucleic acids. Specifically, in the elongation complex, the RNA-DNA hybrid duplex provides stability against reannealing (collapse) of the (DNA-DNA) melted bubble. This stabilization is lacking or reduced in initially transcribing complexes. process, the RNA-DNA hybrid is then fixed at a certain length and subsequent synthesis is more monotonic. As noted above, the primer in polymerization is very different in the initial de novo synthesis of a dinucleotide (where the primer is a single nucleoside triphosphate) and in stable elongation (where the primer is a 8 10 nucleotide RNA in a relatively stable hybrid duplex). Consequently, one might reasonably expect that the complex must be substantially reconfigured during the initiation phase. During de novo synthesis, stability of the complex must derive primarily from the promoter DNA-protein contacts. However, in the elongation complex, these contacts are released and stability is instead achieved by formation of a topologically entwined RNA-DNA hybrid that is at least somewhat stable against collapse of the two DNA strands. Release of initial promoter contacts allows elongation to proceed in a relatively sequence-independent fashion. In summary, the requirements of the basic process require a polymerase that 1) is sequence-specific at initiation, but sequence-independent during elongation, 2) uses a nucleoside triphosphate as the priming base at initiation, but accommodates a 8 10 base pair RNA-DNA hybrid during elongation, and 3) stabilizes an initially open bubble against collapse onto weakly bound mononucleotides or short hybrids during initiation, with much less need for this during elongation, where a similar length hybrid can compete effectively against bubble collapse. Not surprisingly, large changes in the complex occur between these two states. Comparisons Yield Insights Of the well-studied DNA-dependent RNA polymerases, two distinct classes are clear. From a wealth of structural studies, it is 2 WILEY ENCYCLOPEDIA OF CHEMICAL BIOLOGY 2008, John Wiley & Sons, Inc.

3 35 σ 4 Bacterial 10ex σ 3 10 σ 2 T7 9 specificity loop Template strand connectivity RNA RNA Figure 3 Comparison of initiation complex structures for multi- and single-subunit RNA polymerases. The model for the 6 subunit (α 2 ββ ωσ) bacterial complex, colored by subunit (σ is shown in gold), contains protein and upstream DNA from structure 1L9Z and downstream DNA and RNA from the elongation complex 2O5I. The model for the single subunit phage RNA polymerase (the N-terminal domain is shown in gold) contains protein, upstream DNA, and RNA from the initiation complex 1QLN and downstream DNA from the elongation complex structure 1MSW. The latter model has been confirmed by distance measurements. Upstream promoter recognition elements are indicated. The larger initial bubble in the multi-subunit system may arise from the necessarily longer template strand connectivity. now apparent that the multi-subunit family of archeal, bacterial, eukaryotic, and plastid encoded chloroplast RNA polymerases are structurally very similar and have arisen from a common ancestor (4 6). A second class of RNA polymerases is represented best by T7 RNA polymerase, but also includes the nuclear encoded mitochondrial and chloroplast RNA polymerases, as well as the N4 virion RNA polymerase (7 10). We will refer to these as single subunit RNA polymerases (though this is not strictly true in all cases). The multi-subunit enzymes consist of a multi-subunit core with dissociable specificity factors, while the single subunit enzymes are based on a common DNA polymerase I catalytic motif, preceded by different N-terminal domains that use various modules for promoter specificity. Figure 3 illustrates the large difference in size between the multi-subunit and phage RNA polymerases. As will be described below, the striking result achieved by comparison of these two distinct classes of RNA polymerase is that despite being structurally dissimilar, they are very similar in the process by which they transition from an unstable initial complex to a stably elongating final complex. The two classes of RNA polymerase have arisen via convergent, rather than divergent, evolution, driven by the requirements of the process (of the DNA and product RNA) that have dictated very similar enzymatic mechanisms. The structure of the catalytic domain of the single subunit bacteriophage T7 RNA polymerase is that of the DNA polymerase I family, with the shape of a right hand, containing fingers, palm and thumb subdomains (11, 12). These subdomains form a nucleic acid binding cleft to accommodate the template strand. Highly conserved residues located on its walls and bottom form the catalytic active site, which cannot accommodate more than a 3 base pair RNA-DNA hybrid (13). There are neither channels for nucleic acid product nor a clear pore for substrates. To extend the RNA-DNA hybrid beyond 3 nucleotides, T7 RNA polymerase must alter its structure (12, 14, 15). The final organisation of the elongation complex resembles the overall organization of the multisubunit RNA polymerases (15). Approaching elongation, the enlarged active site cavity is adapted to an 8 base pair RNA-DNA hybrid, and final changes in structure result in formation of RNA exit and substrate entry pores. Although eukaryotic enzymes consist of more subunits than does the bacterial RNA polymerase, the core of the multi-subunit eukaryotic (and archeal) RNA polymerases shares substantial structural homology with the bacterial enzyme and comprises 5 main subunits (5, 16 18). The two largest RNA polymerase subunits form a positively charged cleft, which space nucleic acids and where the catalytic active site with metal ions is located. The remaining three core subunits play essential roles in the assembly of the complex. The wall or flap closes the cleft from the side of the exiting DNA. The mobile element, called clamp, has been suggested to control access of the template DNA to the active site and locks it in the cleft. The active site is accessible not only through the cleft but also from below, via a pore or secondary channel. This is an entrance for the incoming nucleotides and an exit for the released pyrophosphate. The exit channel for the nascent RNA chain is located between the wall and the clamp. The differences in surface residues suggest additional regulatory interactions in the eukaryotic systems not presented in bacteria. WILEY ENCYCLOPEDIA OF CHEMICAL BIOLOGY 2008, John Wiley & Sons, Inc. 3

4 Bacteria and archea have only one RNA polymerase. In eukaryotes there are three polymerases, each transcribing different class of genes. All three classes of nuclear RNA polymerases utilize transcription factors to recognize DNA regions and to specifically initiate transcription. Bacterial RNA polymerase, in contrast, employs only one dissociable specificity factor, sigma (σ). Chemical Tools and Techniques A wide variety of tools are used to study transcription initiation. Gel electrophoretic analysis of the RNA products provides key insight, particularly with respect to assaying initiation and to understanding abortive product patterns. The time course of reactions is often followed by manual mixing, with quenching by addition of urea or acid (19), but can also include quench-flow kinetic analysis at shorter times (20). Reactions can either be run with all four nucleoside triphosphates on DNA templates that end a short distance from the start site (runoff) or can be run with a limited set of nucleoside triphosphates to limit synthesis to a position just preceding the first missing nucleotide (see below). Crosslinking experiments provide information on proteinnucleic acid interactions (21, 22), while various footprinting approaches have been used to characterize movement of the polymerase and dynamics of the bubble. Exonuclease and hydroxyl radical footprinting provide estimates of the protein-dna contacts, while footprinting with potassium permanganate can delineate regions with exposed single strands, as in a transcription bubble (23 26). Footprinting with the chemically tethered reagent FeBABE allows identification of regions of the DNA or RNA in proximity to the label placed site-specifically within the protein (27 29). Fluorescent base analogs, as exemplified by 2 aminopurine, can also provide information on the size, positioning, and dynamics of the melted bubble (30, 31). In combination with the above approaches, walking experiments are crucial, to allow halting of a complex at a specific position along the path to promoter clearance. In these experiments, the enzyme is provided only a subset of the four nucleoside triphosphates, with the polymerase halting when it encounters the first position requiring one of the missing substrates. Although complexes halted in the initially transcribed region are necessarily unstable, if dissociation is the slowest step, the bulk of the population at steady state is the desired complex. With enzyme (or DNA) bound to solid supports, one can also do walk-chase experiments, in which the enzyme is halted at a specific position, all substrate is washed away, and then a different subset of nucleotides is added to walk to a new translocational position. This approach is more useful in the elongation phase, where complex lifetimes are sufficiently long. More recently, x-ray crystallography has provided an absolute wealth of information on the various RNA polymerases, and at various stages of the transcription process (6, 32, 33). These structures now allow more direct probing of mechanism using the tools above, but also using structural tools such as fluorescence resonance energy transfer (FRET) to test specific models of structural dynamics (34, 35). Finally, a variety of single molecule approaches are beginning to provide exciting information not available in ensemble measurements (36, 37). Promoter Binding Successful transcription initiation by RNA polymerase depends on sequence specificity and binding affinity. Sequence selectivity in binding commonly arises from hydrogen bonding interactions between amino acid residues and the major groove of DNA (direct readout) or from sequence-dependent features of intrinsic DNA structure (indirect readout), such as its ability to deform upon interaction with proteins (38). Since the polymerase must ultimately transcribe away from the promoter, too strong a binding, however, might be counterproductive for efficient overall transcription. T7 RNA polymerase shows some non-specific binding to DNA, but binds tightly to its promoter through specific protein- DNA interactions. The very small 21 base promoter element (39) can be subdivided into three parts: an upstream duplex recognition element from position 17 to 5, which provides most of the binding specificity (40 42); an AT-rich melting region, which extends from position 4 to 1 and provides a low energetic barrier to melting (43); and the initiation region of the promoter, from position +1 to about +4, which encodes the initial transcript. Most, if not all, of the binding energetics arises from interactions with upstream recognition element (43). In particular, the specificity loop of the fingers subdomain interacts with the bases between 11 and 8 through the major groove and defines the primary sequence specificity of the promoter (40 42). The N-terminal domain of T7 RNA polymerase also recognizes the DNA sequence in the minor groove between bases 17 to 13. ThecoreofE. coli RNA polymerase (subunits β, β,and two α subunits) is not able to bind promoter specifically, but does show nonspecific DNA binding activity. Binding of the σ subunit to the core enzyme, forming the holoenzyme, induces conformational changes both in the core RNA polymerase and in σ itself (32, 44, 45). These changes allow loading of the linker between σ regions 3 and 4 (referred to as σ 3 and σ 4 ) into the RNA-exit channel towards the active site and exposure of σ s DNA binding determinants (46). Promoters may contain consensus 10, extended 10 (from 17 to 13) and 35 upstream recognition elements. Each of the above promoter elements is recognized by one of the four independently folded domains of σ and are recognized differently by different σ factors. The highly conserved regions of σ 2 to σ 3 are involved in direct interactions with the major groove of the promoter DNA (47 49), with σ 2 interacting in a base-specific manner with the promoter 10 (duplex) element (50). Region σ 3,which is critical for recognition of the 10 extended region, faces the major groove of the promoter element. Direct binding of σ 4 to the major groove of the 35 promoter element induces DNA bending (36 around the recognition helix) and thus orients UP elements (upstream of the 35 element) towards the α subunit of RNA polymerase (51). The α subunit binds to the AT-rich DNA minor groove of the UP elements in a sequence-specific 4 WILEY ENCYCLOPEDIA OF CHEMICAL BIOLOGY 2008, John Wiley & Sons, Inc.

5 manner. Interaction with the α subunit moderately distorts the DNA structure (52). The α subunit also interacts nonspecifically with promoters lacking UP elements (53). In contrast to the direct targeting of bacterial and viral RNA polymerase promoters, eukaryotic DNA packaged into nucleosomes is generally inaccessible to the basal transcription machinery (2). RNA polymerase transcription initiates only on naked, or nucleosome-depleted, DNA. Recruitment of the general transcription factors (GTFs) to the promoter is highly dependent on gene-specific regulatory factors near the promoter (54). Most eukaryotic core promoters represent a set of several sequence elements such as the TATA box, BRE (TFIIB-recognition element), Inr (Initiator element) and DPE (downstream promoter element) (55). These serve as binding sites for the basal transcription machinery, promoting the assembly of a functional preinitiation complex, analogous to the bacterial closed complex. In eukaryotes, the TATA box is located 25 to 30 bp upstream of the transcription start site ( bp in yeast) and is the recognition element for TBP (TATA box Binding Protein). TBP binds the minor groove of an 8 base TATA element, inducing kinks in the DNA (56, 57). Such unwinding does not nucleate the initial bubble, but rather initiates association of the enzyme and the basal factors to the promoter DNA between positions 43 and +24 in the following order: TFIID/TFIIA, TFIIB, TFIIF, TFIIE and TFIIH (58). The exact recognition mechanisms of other promoter elements (BRE, Inr and DPE) are less clear, though their recognition partners are identified (55). The small single-subunit RNA polymerases appear optimized for simple promoter binding, with only minor tuning by external factors. Bacterial RNA polymerases use different σ factors to recognize different promoter DNA sequences. Bacterial promoters are longer and contain additional regulatory elements. Like the bacterial system, the core eukaryotic RNA polymerase does not show promoter specificity and depends on GTFs for binding. Thus eukaryotic GTFs carry out the functions of the corresponding regions of the bacterial σ factor. A unique feature of eukaryotic DNA is its chromatin organization, representing another primary regulatory level. Thus, the tuning of gene expression in higher organisms is achieved by a sophisticated multi-level promoter recognition mechanism. Promoter Melting Melting of promoter DNA requires an energetically unfavorable unstacking of DNA bases. For most promoters, ATP hydrolysis is not required and so melting must ultimately arise from, or at least be assisted by, the favorable energy of promoter binding. In principle, RNA polymerase binding to its promoter can be coupled to DNA bending (43, 59 62) and/or twisting (63), which can then nucleate melting of the DNA duplex, as both of these processes result in partial unstacking of the bases. Thermal fluctuations (assisted by negative supercoiling) might also lead to transient melting, forming an open complex that can then be trapped by subsequent synthesis. In either case, the stability of the resulting initial transcription bubble can then derive both from the protein s stabilization of the unstacked duplex bases at the ends of the bubble (the protein can serve as a surrogate for stacking by providing a hydrophobic residue) and from interaction with the template and/or nontemplate single strands (44, 64, 65). One might expect the RNA polymerase to actively guide the template strand into the active site, but in at least the T7 system, promoter binding and melting serves merely to tether the now single stranded template DNA near the catalytic site (66, 67). Although promoter binding provides energy to help drive melting, it is not always sufficient to drive complete melting. This is best characterized in the E. coli system, in which σ 70 dependent promoters typically form stable open complexes, while polymerase binding to σ 54 dependent promoters leaves the promoter in a primarily closed state. Binding of the σ 54 RNA polymerase holoenzyme to its promoter results in a stable closed complex, incapable of promoter melting. ATP hydrolysis provided by an activator belonging to AAA+ protein family is required to convert it to an open complex (68, 69). At other promoters, the initially unstable closed complex of E. coli isomerizes into the open complex in a reaction that includes at least two kinetically significant intermediates and involves conformational changes in both the promoter DNA and the RNA polymerase (23, 70). A model based on the E. coli holoenzyme structure (49) has suggested that the downstream double-stranded DNA may be pushed into its active site and is kinked or bent near the 12 to 11 promoter region. This kink or bend is necessary to bring the 10 promoter element and σ 2 into contact. The β-subunit loop seems not to allow the double stranded DNA through, thus stabilizing a melting-competent DNA orientation. Another model suggests that DNA starts opening only after binding of double stranded DNA to the active site channel (23). A minimal E. coli holoenzyme fragment (the N-terminal 314 amino acids of β, combined with amino acids of σ) has been shown to be sufficient for promoter DNA strand separation (44). Since this fragment is too small to actively bend or torque the DNA, these data suggest that DNA bending is the result, rather than the cause, of melting (once a melted bubble has formed, there is no longer a driving force to retain colinearity of the flanking duplexes). The sequence-specific interaction between upstream DNA and enzyme may assist the isomerization to an open complex (71). The σ 1 domain serves as a molecular placeholder near the active center cleft, mimicking double stranded DNA (72). Wrapping of upstream DNA may facilitate the ejection of σ 1 from the active site (73) and this is thought to allow DNA to descend further into the channel (74). The nucleation of promoter melting is initiated by the highly conserved A-T base pair at position 11 (75), flipping out of the double helix with the help of σ 2 (76). Once separated, the two DNA strands take different paths, with the template strand approaching the active site of the RNA polymerase and the non-template strand being held by conserved aromatic residues in σ 2 that had previously been implicated in DNA melting. Strand separation proceeds up to position +3. Binding of TBP to TATA element, base pairs upstream of the start site, as shown in Fig. 4, triggers assembly WILEY ENCYCLOPEDIA OF CHEMICAL BIOLOGY 2008, John Wiley & Sons, Inc. 5

6 T7 RNA polymerase Spec Loop N-Term Subdomain 4 bases +1 p pp p pp Bacterial RNA polymerases σ 4.2 σ bases p p pp pp Eukaryotic RNA polymerases TATA Box TBP bases bases ( in yeast) p pp p pp Figure 4 The initial bubble. The size of the bubble and its distance from upstream promoter recognition elements scaleswiththesizeofthecomplex. DNA upstream of the +1 start site is melted to span from the promoter recognition determinants to the catalytic site. DNA downstream is melted only minimally. of the eukaryotic preinitiation complex. One of the models describes the complex as being in dynamic equilibrium between open and closed conformations and ATP hydrolysis by TFIIH is required to sustain promoter opening (77). Another model suggests that the ATPase/helicase subunit of TFIIH introduces negative superhelical tension in the DNA (58). Both TFIIF and TFIIE participate in promoter melting. Because of the similarity of TFIIF and TFIIE positions to the location of σ near the active center cleft (78), both factors may take part in the initiation process to promote or stabilize DNA strands melting. The single stranded DNA template can now bend and is positioned within pol II active center, where it interacts with B-finger of TFIIB. It has also been proposed that TFIIF helps direct DNA into the active site of pol II. Like σ 3, TFIIE or TFIIF may trap the single-stranded bubble and promote the insertion of the single-stranded region into the active site of the enzyme. Then promoter DNA is melted between nucleotides 9 and+2. The single subunit T7 RNA polymerase carries out all of the steps of promoter DNA melting without the assistance of activator proteins or external sources of energy. Binding of T7 RNA polymerase to its promoter leads to a bending of the DNA. This bending is centered at the start site and is coupled to DNA melting (59, 60, 79). The intercalating β-hairpin loop of the N-terminal domain serves as a wedge between the DNA strands (stabilizing the unstacked base pair face) and may nucleate DNA melting and/or flipping of the 4 base (12). The nontemplate strand departs the active site and may bind to the top of the fingers subdomain (11). In the absence of NTPs, the initially melted bubble consists of 7 8 bases, sufficient for initiation of RNA synthesis (79). The overall bend angle in the open complex and the in ternary complex is about the same (80), similar to that seen in the E. coli initiation complex (61). Bending of DNA upstream, in regions that do not melt open, can serve to allow a larger protein-dna interaction interface. Thus binding of TBP, for example, bends upstream duplex DNA, allowing for a larger length of interaction with the polymerase complex. Bending of DNA at the locus of melting can, in principle, facilitate melting, as both bending and melting (and untwisting) of the DNA require unstacking of the DNA bases. Alternatively, bending can be the result of melting. Fully duplex DNA is maintained linear by the requirement of base stacking, such that once DNA is locally melted there is no reason to expect that the DNA stems on either side should remain colinear (unbent). Negative supercoiling of DNA can facilitate such bends, either directly or by facilitating the untwisting which is necessary for melting. Finally, while the stability of the melted bubble in the initiation complex must derive from interactions with protein, in 6 WILEY ENCYCLOPEDIA OF CHEMICAL BIOLOGY 2008, John Wiley & Sons, Inc.

7 the elongation complex, the RNA-DNA hybrid is of sufficient length that, at least thermodynamically, bubble collapse now must compete with dissociation of the (more) stable hybrid. Thus protein-mediated stabilization of the bubble may no longer be required. Template Positioning and De Novo Synthesis De novo synthesis is a requirement unique to RNA polymerase (canonical DNA polymerases extend only pre-formed primers), and can be subdivided into distinct sub-steps: (i) positioning of the template strand (position +1) in the active site, (ii) binding of initiating and elongating nucleoside triphosphates, and (iii) formation of the first phosphodiester bond (catalysis). While promoter binding and DNA melting serve to place position +1 of the template strand near the active site, evidence suggests that the template strand is not precisely directed into the active site. In the case of T7 RNA polymerase, the bases ( 4 to 1) between the upstream promoter duplex and the template DNA can be replaced by a polyethylene glycol linkage or the backbone disrupted altogether, with only minor effects on initiation site fidelity (66, 67). In the latter case, the template strand is presumably tethered near the start site via its connection to the bound downstream duplex. In either case, transcription appears to initiate with the nearest downstream C in the template strand, encoding a +1 guanine. In the larger multisubunit polymerases, the template strand seems to serve a similar purpose, tethering rather than precisely guiding the initial templating bases into the active site (81). As shown in Fig. 4, the details, however, are different and presumably reflect the larger protein surface over which these interactions exist (see also Fig. 3). While G is dominant as the first incorporated base in T7 RNA polymerase, A is preferred in the bacterial system (18). It is tempting to speculate that the preference for a purine as the priming base arises from its potentially stronger stacking interactions with the elongating base. That stacking interactions must be important is evident from the consideration that while base pairing provides substrate specificity with respect to the template strand, it is clearly insufficient to stably position either the priming or the elongating substrate NTP. During initiation, the elongating nucleoside triphosphate should enjoy the same stabilizing interactions that it does during stable elongation: base pairing, base stacking, and interactions with the triphosphates. The initiating nucleotide is stabilized only by the first two, as the triphosphates is not required (82). Not surprisingly, the K m for the initiating (priming) nucleotide is significantly higher than for elongating nucleotides (83). Given the weak nature of the initial substrate binding and the fact that growth of the bubble and translocation are not required until after the synthesis of a trinucleotide RNA (79, 84, 85), it is also not surprising that RNA polymerases initiate very well with dinucleotides as primers, with the first elongating NTP at position +3 (86, 87). Abortive Cycling and Promoter Escape Although during elongation, transcribing RNA polymerases are very stable against premature dissociation of the product RNA, during synthesis of the first 8 10 nucleotides of RNA, complexes are not necessarily stable and release premature (abortive) RNA products with reasonable frequency. This process, known as abortive cycling, likely arises from a variety of factors. Most notably, short RNA-DNA hybrids are expected to be less stable than longer ones, such that spontaneous release of shorter RNAs is more likely. At the same time, during the synthesis of the initial, growing transcript, the complex must change from one optimized for de novo synthesis of a dinucleotide, to one with a much longer RNA-DNA hybrid and a nascent RNA chain exiting the complex. Thus it was proposed that during the transition, a stress might build in the system, leading to the observed instability (3). This might predict maximal abortive release of RNA products corresponding to the point of maximal stress. Neither trend is universally observed. Indeed, the pattern of abortive products is highly sequence dependent, in ways that are not necessarily obvious (88 90). To add to the complexity of the system, at any stage in transcription, one would expect a kinetic competition between (minimally) release of the transcript RNA and incorporation of the next nucleotide. Thus any sequence dependence must include at least these factors. With the revelation of structures for initiating and elongating forms of T7 RNA polymerase, we now know that there is indeed a large structural change in the protein complex accompanying the transition from initiation to elongation. In the elongation complex, the RNA-DNA hybrid is about 8 bases in length and the transition occurs at about this point in the initial synthesis. Returning to the fundamental demands of the system, the transition should also include release of upstream promoter contacts (promoter clearance) and this is observed. Two models have been proposed for the structural changes occurring in an initially transcribing RNA polymerase (11, 14). Based on the structure of T7 RNA polymerase complexed with a three base nascent transcript, Cheetham and Steitz (14) proposed a scrunching model in which as synthesis proceeds down the template, the catalytic and promoter recognition sites maintain the same relative orientation by accumulating or scrunching the transcribed template in a pocket on the enzyme. In contrast, in the inchworming model, it was proposed (12) that the duplex recognition portion of the enzyme and the polymerase portion of the enzyme reside on different domains and can move independently of each other as RNA synthesis proceeds. It is now understood that in T7 RNA polymerase, the structure requires that growth of the hybrid pushes on an N-terminal domain within the protein, which rotates as a rigid body relative to the catalytic C-terminal domain, to make room for the growing hybrid (12, 15, 34, 91 93). This domain is a part of the promoter binding surface, such that the bound promoter initially rotates with this domain relative to the C-terminal catalytic domain, consistent with the inchworming model. During this movement, the upstream edge of the initial bubble remains WILEY ENCYCLOPEDIA OF CHEMICAL BIOLOGY 2008, John Wiley & Sons, Inc. 7

8 Initially transcribing Stably elongating Upstream binding Upstream binding Upstream binding Abortive RNA Figure 5 Abortive cycling. The initially transcribing complex is relatively unstable and abortive products are released prior to attainment of a stable elongation complex. Initial growth of the hybrid pushes on a region of the protein involved in upstream DNA binding, contributing to instability in the complex. fixed (the wedge stabilizing the duplex face is a part of the N-terminal translocating domain), while the downstream edge grows to allow incorporation of subsequent nucleotides (94), as shown in Fig. 5. Thus the initial bubble does indeed grow in size during this phase (consistent also with the scrunching model). After synthesis beyond an 8 nucleotide transcript, upstream promoter contacts are released and the upstream edge of the bubble collapses (94, 95). Release of promoter contacts necessarily removes the intercalating hairpin (residues ) from the upstream edge of the bubble, allowing spontaneous collapse of that region of the bubble. Collapse of the upstream edge of the bubble, in turn, is shown to drive displacement of the 5 end of the RNA. At the same time, rotation of the N-terminal domain (no longer constrained by bound promoter) allows refolding of a coupled region of the protein that now becomes a part of an RNA exit channel. Thus, the 5 end of the RNA is resolved from the hybrid just as an RNA exit channel forms to accommodate it. In bacterial RNA polymerase, the RNA exit channel is already present in the open complex, although it is occupied by the σ 3 -σ 4 linker. The growing RNA:DNA hybrid first clashes with this region, as above, and causes displacement of linker from the gateway of the RNA exit channel. The dissociation of σ 3 from the β-flap of RNA polymerase destabilizes σ 4 interactions with the β-flap and the 35 promoter element, allowing the enzyme to release the promoter and displace σ 4 from the β-flap (96). Functionally, this is very similar to the hybrid-driven translocation of the N-terminal promoter binding domain of T7 RNA polymerase, which serves both to weaken promoter contacts and ultimately, to allow formation of an RNA exit channel. Thus σ in the bacterial enzyme serves a function similar to that of the N-terminal domain of T7 RNA polymerase. In the bacterial enzyme, the transition into the elongation stage does not require the complete release of σ, since the binding of σ 2 and σ 3 to the RNA polymerase does not interfere with the paths of the nucleic acids in the elongation complex (32, 45, 97). Recent studies have measured the movement of various regions of E. coli RNA polymerase relative to upstream and downstream DNA, during the transition from initiation to elongation (98). With initial RNA synthesis, downstream DNA is pulled in towards the protein, as expected in any model. At the upstream edge, promoter DNA remains fixed relative to the protein promoter contacts are maintained. Topological analysis of DNA circles has also revealed expansion of the transcription bubble at its leading edge in the abortive complex (62). Thus, as in the T7 system, the initial bubble grows with initial transcription and then contracts upon release of promoter contacts. Most interestingly, the distances from position 20 in the middle of the promoter DNA to labels in σ 4 upstream and to labels in σ 2 downstream also do not change (see Fig. 3). Thus it seems that protein structural changes in this system may be limited to hybrid-driven rearrangements near the σ 3 -σ 4 linker, and that the single stranded (bubble) DNA between the upstream and downstream duplexes must adopt new single stranded conformations to accommodate expansion of the bubble, as predicted by the scrunching model. In summary, in both the T7 and bacterial systems, there is evidence both that the single stranded DNA within the bubble must move to accommodate growth of the bubble and that the protein must undergo some structural changes to accommodate growth of the hybrid. Whether either of these changes contributes energetic stress to the system, leading to abortive cycling, remains to be determined. In eukaryotic systems, the bubble grows in size during initial transcription, as in the bacterial and single subunit enzymes (99, 100). The N-terminal domain of TFIIB may play a role similar to that of the σ 3 -σ 4 linker, as it extends deep into the enzyme and similarly presents a barrier to growth of the RNA DNA hybrid beginning at about 5 bases. It has been proposed that the energy for disruption of initiation-specific interactions is partially provided by TFIIH. This energy is also used to maintain the transcription bubble before promoter release and/or to disassemble the initiation complex (101). Thus competition may lead either to ejection of TFIIB, with potential release of promoter contacts and exposure of the RNA exit channel, or to abortive RNA release and recycling of initiation (102). The former would mark the completion of the rearrangement of the ternary complex (81, 102) and is essential for the initial collapse of the upstream region of the transcription bubble (100). Another model suggests that ejection of TFIIB coincides with growth of an 8 nt RNA, and that the polymerase itself senses the size of the transcript (103). In any case, promoter escape occurs after growth of the bubble to about bases, with at least a 7 8 base transcript (100, 103). Other factors that can govern promoter escape and abortive initiation are transcriptional activators and repressors, cofactors and eukaryotic RNA polymerase general transcription factors (101, 104). Bacterial Gre factors (GreA and GreB) can facilitate promoter escape. In the elongation stage both factors stimulate intrinsic RNA cleavage activity of RNA polymerase, releasing RNA from the stalled complex. However, their mechanism of 8 WILEY ENCYCLOPEDIA OF CHEMICAL BIOLOGY 2008, John Wiley & Sons, Inc.

9 action in the late stages of initiation remains unclear. It has been proposed, that GreA and GreB may shift the equilibrium between productive and unproductive complexes (105), or that GreA stimulates promoter release by its transcript cleavage activity (106). However, TFIIS, the eukaryotic homologue of GreB, has been shown not to participate in promoter escape. In all three systems, the extent of abortive cycling depends on DNA sequences in the promoter recognition region as well as in the initially transcribed region. It has further been proposed that two distinct types of open promoter complex can be formed at a given promoter: one that is capable of promoter escape (and therefore, productive synthesis) and one that is not, and the equilibration of these two types of complexes varies from promoter to promoter (107, 108). Conclusions In summary, RNA polymerase is a complex molecular machine that must rearrange itself to meet the often competing demands of initiation and elongation. Recent structural and mechanistic studies paint a picture in which growth of the RNA is accompanied by an expansion of the initial bubble. Growth of the RNA is also used to disrupt at least some initial promoter contacts, facilitating promoter release. Disruption of the initial structure then allows collapse of the upstream edge of the melted bubble, which in turn, facilitates displacement of the 5 end of the nascent RNA into an RNA exit channel within the protein. Although the details differ between the single and multi-subunit enzymes, the basic mechanism is very similar, suggesting that these systems have converged to a similar mechanism, not because of the details of the protein, but because of the energetic and structural requirements of the process. References 1. Orphanides G, Reinberg D. A unified theory of gene expression. Cell 2002;108: Morse RH. Transcription factor access to promoter elements. J. Cell Biochem. 2007;102: Carpousis AJ, Gralla JD. Cycling of ribonucleic acid polymerase to produce oligonucleotides during initiation in vitro at the lac UV5 promoter. Biochemistry 1980;19: Hirata A, Klein BJ, Murakami KS. The X-ray crystal structure of RNA polymerase from Archaea. Nature 2008;451: Ebright RH. RNA polymerase: structural similarities between bacterial RNA polymerase and eukaryotic RNA polymerase II. J. Mol. Biol. 2000;304: Kornberg RD. The molecular basis of eukaryotic transcription. Proc. Natl. Acad. Sci. USA. 2007;104: Davydova EK, Santangelo TJ, Rothman-Denes LB. Bacteriophage N4 virion RNA polymerase interaction with its promoter DNA hairpin. Proc. Natl. Acad. Sci. USA 2007;104: Amiott EA, Jaehning JA. In vitro analysis of the yeast mitochondrial RNA polymerase. Methods Mol. Biol. 2007;372: Cermakian N, Ikeda TM, Miramontes P, Lang BF, Gray MW, Cedergren R. On the evolution of the single-subunit RNA polymerases. J. Mol. Evol. 1997;45: Murakami KS, Davydova EK, Rothman-Denes LB. X-ray crystal structure of the polymerase domain of the bacteriophage N4 virion RNA polymerase. Proc. Natl. Acad. Sci. USA. 2008;105: Cheetham GM, Steitz TA. Insights into transcription: structure and function of single-subunit DNA-dependent RNA polymerases. Curr. Opin. Struct. Biol. 2000;10: Cheetham GM, Jeruzalmi D, Steitz TA. Structural basis for initiation of transcription from an RNA polymerase-promoter complex. Nature 1999;399: Sousa R, Chung YJ, Rose JP, Wang BC. Crystal structure of bacteriophage T7 RNA polymerase at 3.3 Å resolution. Nature 1993;364: Cheetham GM, Steitz TA. Structure of a transcribing T7 RNA polymerase initiation complex. Science 1999;286: Tahirov TH, Temiakov D, Anikin M, Patlan V, McAllister WT, Vassylyev DG, Yokoyama S. Structure of a T7 RNA polymerase elongation complex at 2.9 A resolution. Nature 2002;420: Cramer P. Multisubunit RNA polymerases. Curr. Opin. Struct. Biol. 2002;12: Iyer LM, Koonin EV, Aravind L. Evolutionary connection between the catalytic subunits of DNA-dependent RNA polymerases and eukaryotic RNA-dependent RNA polymerases and the origin of RNA polymerases. BMC Struct. Biol. 2003;3: Young BA, Gruber TM, Gross CA. Views of transcription initiation. Cell 2002;109: McClure WR. Rate-limiting steps in RNA chain initiation. Proc. Natl. Acad. Sci. USA. 1980;77: Anand VS, Patel SS. Transient state kinetics of transcription elongation by T7 RNA polymerase. J. Biol. Chem. 2006;281: Mustaev A, Zaychikov E, Grachev M, Kozlov M, Severinov K, Epshtein V, Korzheva N, Bereshchenko O, Markovtsov V, Lukhtanov E, et al. Strategies and methods of cross-linking of RNA polymerase active center. Methods Enzymol. 2003;371: Guo Q, Nayak D, Brieba LG, Sousa R. Major conformational changes during T7RNAP transcription initiation coincide with, and are required for, promoter release. J Mol. Biol. 2005;353: Davis CA, Bingman CA, Landick R, Record MT, Jr, Saecker RM. Real-time footprinting of DNA in the first kinetically significant intermediate in open complex formation by Escherichia coli RNA polymerase. Proc. Natl. Acad. Sci. USA. 2007;104: Saecker RM, Tsodikov OV, Capp MW, Record MT Jr. Rapid quench mixing to quantify kinetics of steps in association of Escherichia coli RNA polymerase with promoter DNA. Methods Enzymol. 2003;370: Brieba LG, Sousa R. T7 promoter release mediated by DNA scrunching. Embo. J. 2001;20: Muller DK, Martin CT, Coleman JE. T7 RNA polymerase interacts with its promoter from one side of the DNA helix. Biochemistry 1989;28: Bown JA, Owens JT, Meares CF, Fujita N, Ishihama A, Busby SJ, Minchin SD. Organization of open complexes at Escherichia coli promoters. Location of promoter DNA sites close to region 2.5 of the sigma70 subunit of RNA polymerase. J. Biol. Chem. 1999;274: Miller G, Hahn S. A DNA-tethered cleavage probe reveals the path for promoter DNA in the yeast preinitiation complex. Nat. Struct. Mol. Biol. 2006;13: Mukherjee S, Brieba LG, Sousa R. Structural transitions mediating transcription initiation by T7 RNA polymerase. Cell 2002;110: WILEY ENCYCLOPEDIA OF CHEMICAL BIOLOGY 2008, John Wiley & Sons, Inc. 9