Initiation requires a multisubunit polymerase complex, termed the holoenzyme
Initiation in bacteria first requires binding of a specific initiation subunit, σ, to the core polymerase complex. The resulting complex, capable of recognizing promoter sequences and initiating transcription, is named the holoenzyme. The overall schematic of how initiation occurs is presented in Figure 1, and the conserved nucleotide sequences that form the promoter are shown in Figure 2. Analysis of numerous individual bacterial and viral, or bacteriophage, promoters recognized by σ70, the most abundant σ factor, has led to the derivation of a consensus sequence. The consensus defines the base that occurs with the highest frequency at each position with respect to the TSS. Bacterial promoters turned out to be bipartite: they contain two boxes of conserved sequences separated from each other by 16–18 nucleotides. The TSS itself is usually a purine. The consensus sequence of the box closest to the TSS is TATAAAT, and this is referred to as the –10 box. The consensus of the other box, referred to as the –35 region, is TTGACA. The relative strength of a promoter depends primarily on its affinity for the appropriate σ factor. Strong promoters, which ensure a high frequency of initiation, have sequences close to the consensus. The sequences
Promoter recognition resides entirely in the σ subunit, of which there are several distinct representatives that recognize slightly different promoters in specific gene categories (Table 1). The structure of σ70 has been resolved (Figure 3). The protein has three well-conserved domains, all of which make extensive contacts with the core enzyme. The regions in domains σ2 and σ4 that recognize the DNA promoter sequences at –10 and –35 are solvent-exposed; thus, when the σ subunit interacts with the promoter, it is sandwiched between the polymerase and the DNA. Comparisons of binding parameters and half-lives of the core enzyme and the holoenzyme with promoter and random DNA sequences reveal that the presence of the σ subunit increases the affinity of polymerase for promoters and stabilizes the DNA enzyme complex (Table 2).
Fig1. Initiation of transcription in E. coli. The transition from initiation to elongation is accompanied by conformational changes in the enzyme.
Fig2. Transcription initiation at promoters in E. coli. Examples of individual promoters from eight phage and bacterial genes recognized by σ70 are shown, aligned to highlight common sequences. N stands for any nucleotide residue; the subscript denotes the number of intervening nucleotides between the two boxes in the promoter and between the –10 region and the transcription start site. The consensus sequence was derived from a much larger database of more than 300 well-characterized promoters; it reflects the highest frequency of occurrence of a base at a specific position. The frequency of transcription initiation for any given gene depends on the needs of the cell. The more closely the promoter resembles the consensus sequence, the stronger the promoter: that is, the more frequently transcription initiates from that promoter.
Fig3. Structure of RNAP holoenzyme from T. aquaticus. (A) Conserved regions and functional assignment based on σ70: arrows point to the regions of σ that contact conserved sequences in the promoter region. (B) Top, cartoon representation of σ domains, shown in yellow, bound to core RNAP. Only one α subunit, shown in gray, is visible in this view. The same color scheme is used in the high-resolution structure on the bottom, in which the pink sphere is the active site. The pincers of the crab-claw structure formed by the β and β subunits create a 27-Å-wide channel containing the active site. During elongation, downstream DNA reaches the active site via this channel. Each σ domain makes extensive contacts with the RNAP core; all promoter-recognition determinants in σ are solvent-exposed until DNA is bound, with a spacing that is consistent with the predicted separation of their target promoter elements. Thus, the σ domains bridge the promoter DNA and the polymerase. Conformational changes in both partners are integral to initiation. [A, adapted from Young BA, Gruber TM & Gross CA (2002) Cell 109: 417–420. B, cartoon, adapted from Young BA, Gruber TM & Gross CA (2002) Cell 109: 417–420. With permission from Elsevier. B, structure, courtesy of Seth A. Darst, The Rockefeller University, New York, New York, USA.]
Table1. σ factors in E. coli. There are six factors in the σ70 family. σ70 and σS are closely related and recognize nearly identical core promoter sequences. Selectivity for specific genes may depend on protein factor binding. σ54/N constitutes a separate family on its own. N denotes any nucleotide. Housekeeping genes are defined as those that are constitutively expressed; they code for functions absolutely necessary for the life of the cell.
Table2. DNA-binding properties of core RNA polymerase and holoenzyme containing σ70. The role of σ is (i) to increase the affinity for promoter sequences while decreasing the affinity for nonpromoters and (ii) to stabilize the complex on promoter sequences while destabilizing it on random sequences. Promoter binding is 100 times faster than the maximum theoretical value for a diffusion-limited second-order reaction; this can be rationalized by a one-dimensional diffusion mechanism (scanning rate ~2000 bp in ~3 s).
As Table 1 indicates, bacterial cells make use of slightly different σ factors to transcribe different groups of genes. Different bacterial species have different numbers of σ factors. The number seems to correlate with the complexity of the environment in which a given species lives: organisms with more varied lifestyles contain more σ subunits. Because all σ subunits employ the same core polymerase, there must be mechanisms that ensure the association of the appropriate factor under a given condition. Under certain circumstances, there is an orderly succession of alternative σ subunit usage. A common emergency encountered by bacteria is heat shock, a sudden exposure to an increase in temperature. A defensive response requires changes in protein, and hence RNA, synthesis. Figure 4 illustrates how bacteria respond to heat shock through the consecutive use of two factors, σ24/E and σ32/H. σ24/E is needed for transcription of the σ32/H gene; the σ32/H subunit then turns on an entire set of genes that participate in the heat-shock response. In general, it must be noted that transcription is a very tightly regulated process: numerous regulatory mechanisms are at play, in addition to the mechanisms involving σ factors.
Fig4. Alternative σ factors mediate E. coli’s global response to heat shock. The response of heat-shocked cells involves the successive use of two alternative σ factors: σ24/E is required for expression of the gene that encodes the second alternative factor σ32/H, which in turn regulates the entire σ32/H regulon, a set of bacterial genes that share common regulatory elements for transcription. The σ32/H regulon is induced by excessive unfolding of proteins in the cell, caused by heat shock. The gene encoding chaperone GroEL, a protein involved in refolding misfolded or unfolded proteins, is a member of this regulon.
In some cases, the α subunits play a role alongside σ in initiation. Each α subunit carries an extended C-terminal domain or CTD that is capable of interacting with sites upstream of the σ-sites described above. This provides another level of promoter dis crimination in the initiation of bacterial transcription.
The initiation phase of bacterial transcription is frequently aborted
The initial steps of elongation are slow and the short RNA transcripts are frequently released from the ternary complex in what is known as abortive transcription. The reaction has been studied mainly in vitro, and only very recently in vivo, by using promoter sequences that produce unusually high yields of aborted transcripts and have very slow rates of promoter escape. The use of one such promoter to visualize abortive transcripts on electrophoretic gels is illustrated in Figure 5. Note the extremely high frequency of abortion with this specific promoter: it takes the polymerase 300 trials before it can switch to stable elongation. With more conventional promoters, there are around 30–40 abortion events for each full-length transcript synthesized.
Fig5. E. coli RNAP and detection of abortive transcription in vitro and in vivo. Transcription reactions were performed on DNA templates carrying the N25anti promoter, a 100-bp transcription unit, and the tR2 terminator. This entire construct is designated N25anti-100-tR2. (A) Phage T5 N25anti promoter is a classic model system for the study of abortive initiation and promoter escape, introduced in Michael Chamberlin’s laboratory. The abortive/production ratio for this promoter is very high, ~300, as opposed to 40 for the closely related promoter N25. (B) SDS-polyacrylamide gel electrophoresis of purified commercially available preparations of E. coli RNA polymerase. The core enzyme contains α, β, and β′ subunits; the latter two co-migrate under these conditions. The holoenzyme also contains the σ subunit. This commercial RNAP preparation was used for the transcription experiments illustrated in part C. (C) Transcripts were analyzed following ethanol precipitation, electrophoresis on polyacrylamide gels containing urea, transfer to membranes, and hybridization with a 32P-labeled probe complementary to the beginning of the transcribed region. The in vivo transcription reaction was performed on a plasmid carrying the template, which was introduced into live cells. Further experiments demonstrated that the in vivo products exhibited the hallmarks of abortive transcripts as defined in vitro: altering the strength of interactions between RNAP and promoter and between RNAP and σ factor alters the yields of 11–15 nucleotide transcripts. These interactions must not be too strong to allow ending of abortive initiation and enable promoter escape. [A, C, adapted from Goldman SR, Ebright RH & Nickels BE (2009) Science 324: 927–928. With permission from American Association for the Advancement of Science. B, from Kansara SG, Sukhodolets MV (2010) PloS ONE 6: e18990. https://doi .org /10 .1371 /journal .pone .0018990.]
Importantly, Figure 5 demonstrates that the same aborted transcripts produced during the in vitro reaction are also observed in bacterial cells that contain introduced plasmids carrying the same template construct.
Understanding the mechanism of abortive transcription has been challenging because it was necessary to reconcile two seemingly opposing events. On one hand, the polymerase has to move along the DNA template to be able to synthesize a short RNA chain, a process that requires sequential reading of the sequence of bases in the template. On the other hand, the polymerase has to either preserve or quickly reform contacts with the promoter to be able to re-initiate transcription once a short transcript is aborted. How can this be achieved?
Three models have been proposed to explain how abortive transcription may occur: transient excursions, inchworming, and scrunching (Figure 6). Single-molecule experiments were essential in deciding which of these models is correct. Single pair fluorescence resonance energy transfer (FRET) experiments are presented in Figure7. Additional experiments using magnetic tweezers led to the same conclusion: abortive transcription occurs through a scrunching mechanism in which a stressed intermediate, with approximately one turn of additional DNA unwinding, is formed. Stress in this intermediate provides the driving force to break the existing interactions between RNAP and promoter and between RNAP and σ initiation factor, allowing promoter escape.
Fig6. Competing models for abortive initiation. The models focus on the relative motions of RNAP and the DNA template, as indicated by arrows. The ability to synthesize numerous short transcripts that are released from the initially unstable transcribing complex, coupled with fast re-initiation of new transcripts, implies that the active site of the polymerase is moving along the DNA template in a forward direction but is still maintaining contacts with the promoter in order to allow re-initiation. In vitro experiments over the years have resulted in three models of how this might occur. (Left) In the transient excursion model, RNAP breaks its contacts with the promoter region while transcribing the initial portion of the RNA chain but quickly returns to the promoter once the short nascent transcript is aborted. (Right) In the inch worming model, the enzyme undergoes conformational changes that enlarge the footprint of the polymerase on the DNA, increasing the region over which the two molecules interact; note the slightly enlarged RNAP symbol. Thus the enzyme can continue transcribing while still maintaining its grip on the promoter. Release of the aborted transcript relaxes the RNAP to its normal dimensions. (Middle) In the most recent scrunching model, the RNAP does not change shape; its effective footprint is increased by pulling in a portion of the down stream template, resulting in a stressed DNA conformation. When the short transcript is aborted, the scrunched DNA is released, and the enzyme is ready for initiation of a new chain. The single-molecule experiments that led to the scrunching model are described in Figure 9.14. [Adapted from Herbert KM, Greenleaf WJ & Block SM (2008) Annu Rev Biochem 77: 149–176.]
Fig7. Single-pair FRET experiments demonstrating the DNA scrunching mechanism of abortive transcription. (A) Experimental approach: fluorescence resonance energy transfer or FRET between donor and acceptor dyes, attached at different positions within an open complex, was followed by confocal microscopy. Each molecule traverses the femtoliter-scale observation volume in about 1 ms; the zigzag line signifies the molecule diffusing in the observation volume. (B) Structural model of the open complex RPo showing positions of the donor and acceptor dyes, shown as green and red starbursts, respectively, used in this specific experiment. Numbers on the DNA refer to the relative position of the given base with respect to the TSS, which is position +1. Both dyes are on the DNA. The RNAP core is shown in gray, and σ70 is in yellow. Other labeling sites were used in experiments addressing the validity of the transient excursion and inchworming models, neither of which was supported by the data. (C) Expected outcome of the experiment if the process of abortive transcription occurred through scrunching of the DNA. (D) FRET data for open com plex and for initiation complex, which allows the formation of transcripts of up to 7 nucleotides; the maximal length of the aborted transcripts is determined by the presence of only two out of the four ribonucleoside triphosphates (NTPs). The data clearly indicate a decrease in the distance between fluorophores in the initiation complex: compare the positions of the two red peaks in the two complexes. Note that the scrunching model was also verified through the use of another single-molecule technique, magnetic tweezers.. [Adapted from Kapanidis AN, Margeat E, Ho SO et al. (2006) Science 314: 1144–1147. With permission from American Association for the Advancement of Science.]
Elongation in bacteria must overcome topological problems
In bacteria, the process of elongation is simpler than in eukaryotes, principally because it does not, for example, have to deal with nucleosomes. The nucleoid structure in bacteria appears to provide little resistance to the polymerase; there remains, however, the complication of superhelical stress.
It has been known for years that when an enzyme, such as RNA polymerase, tracks the DNA helix, superhelical stress is created in the DNA if it is topologically con strained. The conventional view of transcription elongation assumed that the polymerase spiraled around the DNA as it tracked the double helix. Such a movement would, of course, entangle the transcript around the DNA. We now realize that, in the cell, the polymerase is probably stationary and exerts both a linear pulling force and a rotary force on the DNA to thread it through its active center. The two kinds of motions—lateral translocation and rotation—on the topologically constrained DNA create positive superhelical torsion ahead of the advancing polymerase and negative supercoiling in the wake of the enzyme (Figure 8).
Fig8. Topological consequences of transcription elongation in a topologically constrained DNA template. The DNA is represented as having fixed ends to satisfy the topological requirements for superhelicity. The transcription bubble does not rotate in the immobile RNA polymerase catalytic center, creating domains of positive and negative supercoiling, represented as DNA with different twist. Superhelical stress accumulates quickly because transcription of one helical turn, around 10 bp, leads to the creation of one super helical turn. DNA structure cannot tolerate too much twist, so the stress can be relieved by the creation of writhe, which is coiling of the DNA axis.
As elongation proceeds along the topologically constrained loop domains in the bacterial nucleoid or in the eukaryotic nucleus, the buildup of positive supercoiling would be expected to stop transcription eventually, as the great energy cost of unwinding the overwound DNA would prohibit further elongation. Although it is true that topoisomerases will relieve the stress, a topoisomerase molecule must find the stressed region and then attach before relaxation can occur. Thus it is likely that, given the high rate of transcription, considerable stress will accumulate. The idea that high levels of transcriptional activity necessarily lead to high levels of superhelical stress is, however, a misconception. High levels of transcription usually reflect high frequency of initiation, which creates an array of polymerases tracking each other along the gene. Such an array will partially neutralize the positive and negative stresses that are produced by each polymerase: that is, the negative stress in the wake of one polymerase will be partially neutralized by the positive stress created in front of the following polymerase, and so on. Nevertheless, transcription in a constrained system will always encounter at least transient effects of induced changes in supercoiling.
The topological consequences of transcription have been often overlooked, as most in vitro transcription systems use either short linear DNA templates, in which the torsion dissipates from the free DNA ends, or circular plasmids, where the positive and negative supercoiling propagating from opposite directions around the circle eventually cancel each other out.
There are several mechanisms for transcription termination in bacteria
Bacteria use several mechanisms to terminate transcription. The “classical” two mechanisms are the intrinsic (or sequence-dependent) and the Rho-dependent. In Escherichia coli, the number of genes using one mode or the other is about equally divided. However, in general, there is a wide range of relative usage of these two modes among bacterial species. Here, we will describe termination in E. coli, where the process is best understood:
1. Intrinsic or sequence-dependent termination: Intrinsic termination is entirely determined by the nucleotide sequence at the 3′-end of the gene and does not require any other factors. The sequence of the template near the termination site should give rise to an RNA transcript capable of forming a hairpin. This will be assured if the DNA contains an inverted repeat (Figure 9). An additional sequence feature is essential for intrinsic termination: the inverted repeat should be followed by a string of six or seven As. The formation of a hairpin in the RNA transcript causes the polymerase to pause and strips the RNA from the complex before further transcription can occur. The weak base pairing between the A string in the template and the complementary U string in the transcript contributes to dissociation of the complex: the pausing caused by the hairpin gives the string of A–U pairs many chances to dissociate. Proteins such as NusA stabilize the RNA hairpin, thereby lengthening the pause; this helps the dissociation process.
2. Rho-dependent termination: Factor-dependent termination makes use of Rho factor, a hexameric ATP-dependent DNA–RNA helicase. The steps in this termination process are schematically depicted in Figure 10. The process begins when Rho binds to a Rho utilization site or rut site on the RNA transcript. rut sites have highly skewed nucleotide content, being very rich in C and poor in G; otherwise, there is no recognizable sequence similarity among rut sequences in different transcripts. Once Rho binds to the rut sequence, it undergoes a conformational transition from an open ring to a closed ring that embraces the RNA (Figure 11). In an ATP-dependent process, Rho is propelled along the RNA transcript in a 5′ → 3′ direction; its DNA–RNA helicase activity pulls the RNA from the DNA in the transcription bubble, thus terminating transcription.
Fig9. Intrinsic termination in E. coli. Transcription terminates when an inverted repeat in the RNA transcript forms a hairpin that is abutted by a string of uracils; the low stability of the RNA–DNA heteroduplex, due to A-U base pairing, contributes to the destabilization of the RNA–DNA hybrid in the transcription bubble. The RNA hairpin interacts directly with the RNA polymerase, destabilizing the triple DNA RNA-polymerase complex. Additionally, the elongation protein factor NusA interacts with both the polymerase and the RNA hairpin structure to aid in dis sociation of the complex.
Fig10. Rho-dependent termination of transcription in E. coli. Rho factor is a hexameric ATP-dependent DNA–RNA helicase that acts on DNA–RNA hybrids exclusively. Transcription termination occurs in four steps. (Step 1) The N-terminal domain of Rho, shown in blue, binds the Rho utilization or rut sequence on the transcript. (Step 2) The C-terminal domain of Rho, shown in yellow, binds the mRNA downstream of rut, and the ring closes. (Step 3) The C-terminus cyclically hydrolyzes ATP to propel itself along the mRNA in a 5′ to 3′ direction. (Step 4) Helicase action leads to disassembly of the transcription complex. (Inset) Sequence of Rho utilization or rut site of λ tR1 terminator. Note the highly skewed nucleotide content, with 42% of the residues being C and only 4% being G. This is a typical characteristic of the rut sequences: the efficiency of usage of these sites for termination increases with the length of the C-rich/G-poor region. rut sites are found at varying distances from the actual termination sites. [Adapted from Kaplan DL & O’Donnell M (2003) Curr Biol 13: R714 R716. With permission from Elsevier.]
Fig11. Crystal structure of Rho termination factor. Top-down view shows protomers labeled A–F. In the front view, the six subunits or protomers, represented in different colors, pack in an open hexameric ring. This is the conformation in which Rho binds to the rut sequence on the mRNA, after which the structure closes into a ring that embraces the RNA. The transition between open and closed ring conformations is accompanied by rotational shifts of subunits with respect to their neighbors. Shown at the bottom is a drawing of the relative rise and offset of adjacent Rho subunits as they wind about the axis of the ring, shown as a vertical line. Color scheme corresponds to the top-down and front views. Gap between monomers is 12 Å and helical pitch is 45 Å. [Adapted from Skordalakes E & Berger JM (2003) Cell 114: 135–146. With permission from Elsevier.]
Recently, several other termination mechanisms have been recognized. One involves a protein, Mutation frequency decline (Mfd), which is predominantly involved in the process of transcription-coupled DNA repair. Mfd is a large protein with ATP hydrolase and helicase activities. Mfd recognizes a stalled RNA polymerase (RNAP) and removes it from DNA, at the same time recruiting the excision repair machinery. Unlike the two classical termination mechanisms described above that involve RNA, Mfd acts by binding simultaneously to DNA and RNAP; it uses the energy of ATP hydrolysis to remove the RNAP from the DNA template by translocating along the chain. New single-molecule results show that Mfd can also bind and translocate along DNA without first binding RNAP. In this mode of action, the protein “patrols” the DNA, moving more slowly on the template. If the polymerase moves at a normal speed, Mfd cannot catch up; a stalled RNAP is, however, overtaken and released.
Another mechanism involves the action of RNase J1. The enzyme possesses both a 5′-3′ exonuclease and an intrinsic endonuclease activity. Depletion of the enzyme in living Bacillus subtilis cells leads to a significant increase in RNAP density along some transcription units without a concomitant increase in the level of transcription, consistent with the accumulation of unresolved stalled polymerases.
A general principle appears to apply in termination events, both in bacteria and in eukaryotes: an allosteric transition from the processive elongation complex to a dis sociation-prone pre-termination complex.
Antisense transcription in bacteria is widespread and might have numerous functions
It is a well-established fact that bacterial genomes are densely packed with protein coding genes. Recent genome-wide mapping of transcriptional start sites revealed that a significant fraction of promoters drives the transcription of RNA that do not have protein-coding capacities. The transcribed RNAs are mostly natural antisense molecules (asRNAs) which overlap other transcriptional units. A 2000 study found antisense transcription occurring at as many as ~4,000 open-reading frames in E. coli. The present estimate is that asRNAs originate from at least ~37% of all transcription start sites (TSSs). In agreement with such estimates, it was recently demonstrated that a substantial fraction of the TSS discovered by transcriptome studies in a wide array of bacterial taxa is not associated with protein-coding genes. Internal parts of coding regions are massively transcribed in both sense and antisense orientation, in what has been termed pervasive transcription.
Bacterial asRNAs show enormous differences in length, genomic location, abundance, and regulation. The main categories of bacterial asRNAs are presented in Figure 12. There are long forms spanning several genes, shorter 5′-overlapping or 3′-overlapping, internally located asRNAs.
Fig12. Overview of the main categories of asRNAs in bacteria. (A) The short asRNAs have a defined 3′ end and depending on where in the gene sequence they originate, they fall into three distinct categories (top three schematics). A special case are asRNAs to internal genes within an operon, which may cause decoupled expression of genes 5′ and 3′ of the asRNA binding site (bottom schematic). (B) Long asRNAs usually do not have a strong terminator and can thus be present as a set of molecules of different lengths.
What do we know about the possible roles of asRNAs, if any? Many asRNAs have been found to play roles in the regulation of gene expression. Some asRNAs may form duplexes with their target molecules, thus altering the secondary structure of both interacting partners. These structural changes influence the stability and half life of the molecules, often leading to rapid and complete degradation of both transcripts. However, degradation does not necessarily occur. Actually, cases have been described where the duplex formation leads to a stabilized form of mRNA.
Some asRNAs modulate protein translation, through a variety of mechanisms. The duplex formed between an asRNA and the 5′-end of its mRNA partner leads to the occlusion of the ribosome binding site and the start codon, thus preventing initiation of translation. A well-studied example of such action is presented in Figure 13. It must be noted that if the distance between the start codon and the TSS of the asRNA is somewhat large (~20 nt), the duplex formation between the two RNA partners may have the opposite effect, that is, stabilization of the mRNA.
Fig13. The Vibrio cholerae MtlS asRNA represses the synthesis of MtlA. MtlA is a mannitol transporter protein that is specifically expressed in the presence of the sugar mannitol. The TSS of the mRNA is situated only 5 nt upstream of the start codon; thus, binding of the asRNA fully covers the ribosome binding site, repressing translation.
There are also gene expression mechanisms that act through affecting transcription of the target genes. In view of all these molecular mechanisms, are there global functions of asRNAs? Excessive pervasive transcription of likely detrimental to cell viability; thus, cells have developed mechanisms to control the formation of asRNAs at the transcriptional level. These are too many and rather complex to be listed here, but the interested student is referred to a comprehensive review by Georg J & Hess WR [(2018) Microb Spectrum 6: RWR-OO29].
Understanding transcription in bacteria is useful in clinical practice
Today, the scientific community has acquired deep understanding of how transcription occurs in bacteria. This knowledge is absolutely necessary for the development of drugs to combat bacterial infections.
