Module 4.5.2.2

Promoter Escape by Escherichia coli RNA Polymerase

LILIAN M. HSU

[SECTION EDITOR: I. ARTSIMOVITCH]

Posted February 12, 2008

Program in Biochemistry, Mount Holyoke College, South Hadley, MA 01075

Mailing address: Program in Biochemistry, Mount Holyoke College, South Hadley, MA 01075. Phone: (413) 538-2609, Fax: (413) 538-2327, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it .

OVERVIEW

Promoter escape refers to the early steps of transcription during which the RNA polymerase molecule—bound securely as a catalytically active open complex (59, 69, 70) and having performed the de novo synthesis of the first phosphodiester bond—must relinquish promoter contacts and undergo a concerted sequence of conformational transitions to emerge as an elongation complex. At the structural level, this conformational transition involves: (i) formation of an 8-bp RNA-DNA hybrid at the active site of an elongation complex (28, 46, 72, 99, 106), (ii) breaking of bonds between the RNA polymerase holoenzyme (especially the sigma subunit) with various upstream promoter elements (12, 47, 94), (iii) displacement or release of the sigma factor from the core enzyme (65), and (iv) translocation of the initial open complex bubble to the elongation bubble.

The above macroscopic transition encompasses a multistep dynamic process that requires the concomitant synthesis of the nascent RNA. The escape process, depending on the promoter involved, can require up to 15 steps of initial transcription (39). These steps are fraught with difficulty (and instability) and result in the release of short nascent transcripts, called abortive RNAs, by the initial transcribing complexes (ITCs) that are deemed to be stressed intermediates (94). The stress in the ITC arises as the enzyme translocates after each step of nucleotide incorporation by scrunching a base pair of unwound downstream DNA into the enzyme interior (15, 43, 83) without a concomitant rewinding of a base pair at the upstream boundary of the transcription bubble (104). Thus, with each step of initial transcription, the transcription bubble expands to harbor incremental stress. After 10 to 15 nucleotides (nt), the cumulative stress becomes sufficient to force upstream bubble rewinding, freeing the promoter DNA from its enzyme hold and allowing RNA polymerase to move into the elongation region (37, 83). Concomitant with the escape transition, the activity of RNA polymerase abruptly converts from one undergoing reiterative abortive transcription to one performing processive elongation.

The above description sets forth, in large strokes, what must happen during promoter escape. The details of how promoter escape is achieved, however, differ from promoter to promoter, as reflected by the sequence-specific pattern of abortive RNAs released from each promoter during bulk transcription. The qualitative and quantitative analysis of the abortive RNA patterns has provided a great deal of information for solving the step-by-step mechanism of promoter escape. In this chapter, I will review the salient biochemical features of promoter escape (last reviewed in reference 37) and integrate them into a kinetic, thermodynamic, and structural framework of transcription initiation.

BIOCHEMISTRY OF ABORTIVE INITIATION AND PROMOTER ESCAPE

Discovery of Abortive Initiation and Its Linkage to Promoter Escape

Promoter escape, as a process, refers to the first 10 to 15 steps of transcription that are necessary to bring about the initiation to elongation transition. That this transition, on most promoters, is not accomplished in a straightforward, single-step manner was revealed by the repetitive formation of abortive transcripts (16, 30, 35, 38, 40, 48, 52, 67). Although abortive initiation was first observed in transcription reactions containing limited nucleotide substrates (39), it was soon shown to occur even in the presence of all four nucleoside triphosphates (NTPs) (10). The observation that the abortive initiation process is not much diminished by the presence of (low levels of) heparin—which sequesters free or released RNA polymerase to prevent enzyme recycling—led to the conclusion that, during abortive initiation, RNA polymerase does not dissociate from the DNA; rather, on release of the short transcripts, the polymerase-promoter complex resumes the open complex structure to perform de novo initiation again. This picture of an RNA polymerase "idling" over the initial template region was confirmed by footprinting analyses in vitro (12, 107) and in vivo (22).

The early kinetic investigation on lacP and related mutant promoters led Gralla and coworkers to reach two important conclusions. One, the initiation of lac UV5 transcription is rate limited at the promoter escape step (11), and two, in terms of sequence agreement to consensus (33), the stronger lac promoters were poorer at promoter escape than the weaker promoters. The latter was attributed to the opposing roles required of the sigma-subunit of RNA polymerase holoenzyme during initiation and elongation. During initiation, RNA polymerase is recruited to the promoter region through sigma binding to the −35 and −10 elements; here, the closer to consensus the promoter sequence, the stronger the binding (58). However, at the other end of the initiation phase, in order to transit into the elongation phase, sigma-factor has to relinquish its hold on the promoter DNA to be released and recycled (12, 32, 100). Clearly, the more tightly bound the sigma factor to DNA, the more difficult is the release. This discrepancy between in vitro binding strength (to form initiation-competent open complexes) and in vivo promoter strength (in producing full-length RNA) was well documented by Bujard and coworkers (20). Following up on their observations, Ellinger et al. (22) found that when three artificial Eσ⁷⁰ promoters were footprinted in vivo, the consensus version caused RNA polymerase to stall excessively at the +6 to +12 region, diminishing the synthesis of full-length RNA.

Parallel with the above investigations, Bujard and coworkers showed that the phage T5 early gene promoters have evolved quite different optimizations for their gene expression program (44). Of these, the N25 promoter appears to have maximized the binding (i.e., K_B) and isomerization (i.e., k₂) steps of open complex formation such that productive transcription becomes rate-limited at the next step, i.e., promoter escape (20). On this promoter, escape is greatly affected by the initial transcribed sequence (ITS) encompassing the first 20 template positions (41). Changing the native N25 ITS sequence to one called antiDSR (involving A ↔ C and G ↔ T substitutions and hereafter abbreviated as anti) decreased productive synthesis by ~10-fold. How an ITS region can exert such a large effect on gene expression was revealed subsequently by quantitative assessment of the initiation pattern—abortive and productive–at these promoters (38).

Quantitative Assessment of Abortive Initiation and Promoter Escape

The phosphorimaging technology coupled with high-resolution gel separation of short RNAs has made possible large-scale qualitative and quantitative examination of the abortive initiation-promoter escape activities at promoters (36). Applying this method to the analysis of three phage promoters, T5 N25, N25_anti, and T7 A1, each produced a set of abortive RNAs whose composition and abundance are specified by the promoter sequence (including the ITS region), yielding a distinct abortive RNA (ladder) pattern much like a bar code or fingerprint (Fig. 1). From this bar-code pattern, several parameters can be derived to quantitatively describe the abortive initiation-promoter escape process at a promoter (38, 39). The useful parameters include: the productive yield, the abortive/productive ratio, the abortive probability profile, and the maximum size of abortive transcripts (MSAT). Productive yield is the true indicator of the strength of a promoter in synthesizing full-length RNA. The abortive/productive ratio reflects the ease or difficulty at escape. The abortive probability profile describes the abortive tendency at each position and over all of the positions during initial transcription (Fig. 2). That each promoter shows a unique abortive probability profile suggests that the escape process is directed sequence specifically by the initial transcribed region. Later in this chapter, I discuss the abortive probability profile and equate it with the thermodynamic energy diagram for promoter escape at a given promoter. The MSAT indicates the length of the abortive process, at which point RNA polymerase would be poised to undergo the promoter escape transition. After the transition, RNA polymerase will have lost its reiterative transcription property because the sigma-factor has been either released from the core enzyme or displaced from its original open complex contacts (8, 57). Comparing T5 N25 and N25_anti by the above criteria, it becomes clear that the anti ITS has created a more formidable barrier for escape, such that abortive initiation now continues to the 15th position (compared with the 11th position for N25); and even then, only a tenth of the polymerase manages to escape abortive cycling and perform elongation synthesis to the full length (38).

Fig. 1A typical gel fractionation profile of abortive and productive RNAs. Shown are the products of transcription from six T5 N25-related promoter templates that differ only in their initial transcribed sequences (ITS). Compared with the native N25 promoter, N25/A1, N25_anti, and DG133 differ in sequences from +3 to +20, and DG203 and LX11 differ only from +3 to +10. Each ITS directs the synthesis of a distinct collection of abortive RNAs whose sizes (in nucleotides) are indicated by numbers on the border. A set of five reactions (lanes a to e) was performed with each promoter to show that the abortive pattern is greatly altered by the presence of GreB (lanes c and e) and unaffected by NusA (lane d) compared with the normal pattern of abortive/productive synthesis (lanes a and b). FL designates the full-length productive RNA.

Fig. 2Abortive probability profile of three promoters. The promoters analyzed are N25, N25_anti, and DG203. Abortive probability at each initial position is derived from the quantitative measurement of abortive and productive RNA levels from a slice of gel shown below each panel; the abortive yield of an RNA is further corrected by the fraction of RNA polymerase reaching that position (36). High abortive probability reflects a highly unstable ITC that tends to release its RNA and begin initiation again. Plotting the abortive probability associated with each initial position produces the abortive probability profile that conveys the ease or difficulty of traversing the initial transcribed sequence region by RNA polymerase. The journey is different on each promoter.

Factors That Influence Abortive Initiation and Promoter Escape

Factors that influence abortive initiation and promoter escape can be classified into intrinsic and extrinsic elements. The intrinsic elements are cis signals that include the promoter recognition region (PRR; sequences from −60 to −1), initial transcribed sequence (ITS; sequences from +1 to +20), and the conformational state of the template DNA. The extrinsic elements include trans factors such as the presence of activator or repressor proteins, transcript cleavage stimulatory factors, RNA polymerase mutations, and reaction conditions such as KCl and NTP concentration. As we will see, each of these elements affects abortive initiation and promoter escape by influencing one or more equilibrium/rate constants associated with the kinetic diagram of transcription initiation, which involves branched pathways due to the existence of productive and unproductive open complexes (Fig. 3). The evidence for a branched pathway is discussed in a later section in this chapter.

Fig. 3Kinetic scheme of transcription initiation. At escape rate-limited promoters, transcription initiation follows a branched pathway due to the formation of two types of open complex. In this scheme, R represents RNA polymerase; P, promoter DNA; RP_c, the closed complex; RP_o, the productive open complex that can escape rounds of abortive cycling to synthesize full-length RNA; RP_o', the unproductive open complex that abortively initiates without undergoing escape; and TEC, ternary elongation complex. The numbers 2, 3, 4, 5, and 6 represent initial transcribing complexes (ITCs) with a nascent RNA of the same length. K_B is the equilibrium binding constant for closed complex formation; k₂ and k₋₂ are forward and reverse rate constants for open complex formation and collapse; k_E is a composite rate constant of promoter escape that measures the rate of productive RNA synthesis from an open complex. An arrow underneath each ITC is of different length to represent the different levels of abortive RNA released.

Intrinsic elements.

• PRR (Promoter Recognition Region). Among the cis factors, the PRR is the master regulator for abortive initiation and promoter escape; it governs open complex stability and, in turn, sets the rate-limiting step for transcription initiation at a given promoter (11, 19, 21, 22, 44, 103). A promoter that has maximized its K_B and k₂ steps (i.e., a consensus promoter) forms highly stable open complexes that can initiate at high frequency but escape (from the sigma-contacts) with difficulty. It is by default rate limited at the promoter escape step, with the rate limitation manifested in elevated levels of abortive transcripts (21, 49, 103). By contrast, rrnB P1 forms an open complex of extremely short half-life (due to a large k₋₂); it is not rate-limited at promoter escape and produces few abortive RNAs (3, 29). The abortive behavior of most Eσ⁷⁰ promoters has not been as thoroughly analyzed and linked to their K_B, k₂, and k₋₂ constants. At this time it is not known whether each promoter, with its distinct K_B × k₂ value, may occupy a point on the continuum between the consensus promoter and rrnB P1 (44, 58).

• ITS (Initial Transcribed Sequence). For a promoter that is rate limited at the escape step, the escape process is further modulated by the ITS sequence. For T5 N25 promoter, randomized ITS sequences gave rise to 25-fold differences in productive synthesis (39). An inverse correlation between the productive yield and the initiation frequency among the 40-odd N25-random ITS promoter variants suggests that a failure at escape relegates the polymerase to repetitive abortive initiation. Each ITS variant further shows a distinct abortive probability profile, suggesting that the ITS sequence directly governs the promoter escape process. As discussed in a later section in this chapter, kinetic investigation reveals a role of the ITS in changing k_E, the composite rate constant of escape, and the extent of partitioning of RNA polymerase into the productive and unproductive complexes (17, 88, 102).

• Template supercoiling. Although most quantitative analyses of abortive initiation and promoter escape have employed short DNA fragments containing a single promoter, promoters embedded in a supercoiled plasmid also undergo repetitive abortive initiation to reach the promoter escape transition (16). Lim et al. (53) investigated the effect of varying superhelicity (from a supercoiling density of 0 to −0.093, with −0.063 being the physiological density) on abortive initiation and promoter escape and found that increasing negative superhelical density led to a consistent decrease in abortive RNA synthesis from galP1 and galP2 promoters. In the case of galP1, the decrease in abortive synthesis directly correlates with the increase in productive synthesis at all superhelical densities, suggesting that supercoiling has a role in facilitating promoter escape. For galP2, the productive synthesis showed a dependence on optimal superhelical density. At higher than physiological superhelicity, both abortive and productive synthesis decreased, indicating that the promoter had become inactivated. Thus, supercoiling can affect abortive-productive transcription; the exact outcome, however, will vary from promoter to promoter. This latter conclusion parallels our understanding of the effect of supercoiling on Escherichia coli global gene expression (79). Mechanistically, a study employing single-molecule nanomanipulation showed that supercoiling, whether positive or negative and depending on the promoter involved, affects the open complex parameters (rate of formation and lifetime) mechanically (through torque) in the melted DNA (82). Since promoter escape involves translocation of the transcription bubble—first by unwinding the downstream bubble, followed by rewinding the upstream bubble—it is expected that the rate of movement of the bubble would be greatly affected by the degree of supercoiling in the template DNA.

Extrinsic elements.

Many extrinsic elements are now known to influence abortive initiation and promoter escape. The examples cited below illustrate the principle that they exert their effect by changing one or another rate/equilibrium constant in the transcription initiation rate diagram (Fig. 3). LacI normally represses gene expression from the lac promoter. However, LacI binding at the primary operator overlaps the initial transcribed sequence region, allowing RNA polymerase to synthesize abortive RNAs but without escaping into the elongation region (51); thus, LacI binding inhibits k_E. Bacillus subtilus phage φ29 p4 protein normally functions as an activator by binding to the upstream region of a promoter. However, when bound upstream of the A2c promoter, p4 represses productive RNA synthesis by elevating abortive synthesis (63). The effect of p4 can be rationalized as having raised K_B (through upstream binding) sufficiently to create a new rate-limiting step at k_E. GreB protein has been shown to facilitate promoter escape in vitro and in vivo (35); it exerts its effect by enhancing the cleavage rescue of backtracked transcripts (39). RNA polymerase mutations that alter the abortive initiation-promoter escape efficiencies all map to the σ-subunit. Notably, mutations in σ-region 2.2 or σ3.2 linker decrease the stability of the open complexes and, thereby, reduce the tendency of abortive initiation (13, 34, 86, 103).

Besides proteins, solution factors also affect the relative abortive/productive ratio in in vitro transcription studies. The most basic factor is salt concentration; for each escape-rate-limited promoter, an optimal KCl concentration exists in favor of escape (36). The rather high concentration of KCl, ranging from 150 to 250 mM, serves to select stable open complexes that, on subsequent initiation and RNA chain growth, can undergo transcription bubble rewinding to enter the elongation phase. Another important solution factor is the concentration of the NTP substrates. High [NTP], especially the initiating nucleotide, not only stabilizes the open complex lifetime leading to more frequent initiation (38, 82), but also favors the partitioning equilibrium toward the formation of productive open complexes (102).

In recent years, the role of the stringent response alarmone, ppGpp, has been increasingly elucidated. ppGpp binds near the RNA polymerase active site, requiring DksA stabilization through the secondary channel, to affect transcription initiation (2, 78). The binding of ppGpp or DksA alone was sufficient to decrease the open complex lifetime several fold on lacCONS or rrnB P1 promoter (76, 82). Together, however, ppGpp and DksA synergistically amplified the inhibitory effect on rrnB P1 transcription but positively stimulated the amino acid biosynthetic operon promoters, e.g., P_argI (76, 77). For P_argI, ppGpp/DksA raised the isomerization rate constant very significantly without changing the rate of escape or the abortive/productive ratio (77). The ability of ppGpp/DksA to alter the abortive/productive synthesis, however, was demonstrated on a T7 A1 promoter where it decreased abortive synthesis 15-fold but productive synthesis only ~2-fold (78). Thus, ppGpp/DksA appears to modulate promoter activity at different steps on different promoters. Many of the extrinsic factors that affect abortive initiation and promoter escape act through the secondary channel of RNA polymerase; these include NTP (5), Gre factors (23, 50, 73, 91), and DksA and ppGpp (76, 78).

Mechanism of Abortive RNA Formation

Structural impediment.

At least three factors govern the synthesis of abortive RNAs. The foremost arises from the structure of RNA polymerase holoenzyme where the σ-region 3.2 peptide linking the surface-exposed σ₂σ₃ and σ₄ domains dips toward the enzyme active site before looping around to traverse the RNA exit channel (68, 101). In this arrangement, the σ3.2 linker physically obstructs the entrance of the exit tunnel and is predicted to impede the entry of a transcript that has reached 5 to 6 nt in length, resulting in their abortive release. Several experiments have been done to confirm the obstructive role of sigma in causing abortive initiation. Murakami et al. (68) deleted the entire C-terminal portion of σ⁷⁰, including the σ3.2 linker (σ⁷⁰ residues 1–503), and found that the ΔC-holoenzyme can still initiate transcription on a galP1 extended −10 promoter (which normally lacks a requirement for −35/ σ4 binding for promoter function) but now produces mostly full-length RNA. Kulbachinskiy and Mustaev (49) created a more targeted deletion of σ3.2 linker (missing amino acids 513-519 of the loop region) and found that the σ-subunit is no longer within the cross-linking range (≤18 Å) of an initiating ATP analogue bound at the i site (49, 89). Like the wild-type polymerase, the Δ(513–519)σ⁷⁰ holoenzyme still transcribes a T7 A1cons promoter to yield an abortive ladder of 16 nt. However, it now exhibits an altered abortive transcription pattern similar to that seen with RNA polymerase containing σ3.2 linker point mutation S504F or P506L (34). These mutations presumably rearrange the trajectory of the σ3.2 linker connection and result in a less stable open complex (86). Together, the above data suggest the loop of σ⁷⁰ 3.2 linker poses a physical barrier to the advancing nascent transcript and is partially responsible for the abortive release of 6 to 8 nt RNA (103). Blockage to initial transcript progression also exists in T7 RNAP (15, 106) and yeast RNA Pol II (9).

RNA polymerase backtracking.

During transcription, RNA polymerase occasionally, but in a sequence-specific manner, undergoes backtracking and becomes paused or arrested on the template DNA (1, 45, 55, 56). On the backtracked complex, the 3'-OH end of the nascent RNA is extruded into the secondary channel and cannot be further elongated (45). To reactivate the transcription complex, RNA polymerase utilizes an intrinsic hydrolytic activity that cleaves a backtracked transcript at the active site, generating a new 3'-OH that can be reextended (74, 96). The transcript cleavage activity of RNA polymerase performs the proofreading function during transcription (108) and is greatly stimulated by Gre factors—GreA or GreB– bound in the secondary channel to enhance the catalytic ability of RNA polymerase (6, 7, 24, 26, 50, 73, 91).

The observation that abortive initiation can be diminished or overcome in the presence of Gre factors led to the realization that, before their release from the enzyme, the abortive products were transiently associated with backtracked initial complexes (25, 35). The presence of GreA stimulates the cleavage of 2 to 3 nt from the 3' end of the backtracked RNA—whereas GreB stimulates the cleavage of up to 10 nt—and allows reextension of the 5' piece of RNA (7, 73). For highly abortive promoters, GreB was found to be more effective at cleavage-rescue; once cleaved, the 5'-RNA can be stably extended to the full length (39). The analysis of GreB activity on these highly abortive promoters suggests that cleavage-rescue is only successful on initial backtracked complexes that have retained an RNA-DNA hybrid region of ≥5 bp. The backtracked complexes held by a hybrid ≤5 nt apparently readily release the short RNAs such that their levels are unaffected by the presence of GreB (39).

The preponderance of backtracking during initiation can be attributed to the strain generated in the initial transcribing complexes by the scrunching mechanism of translocation (15, 43, 83), leading to a distorted (and strained) transcription bubble; thus, the initial transcribing complexes are stressed intermediates (94). Before reaching the escape transition, however, the accumulating stress can be relieved by one of two means of stabilizing the transcription bubble: either by backtracking or by propagating the strain away from the active center of the polymerase so that the enzyme can bring in the next nucleotide for incorporation (37). In this regard, the degree of backtracking and abortive release (measured by the abortive probability) is a direct reflection of the amount of stress associated with an initial transcribing complex formed at a given position.

Moribund complexes.

A third source of abortive RNA comes from the moribund complexes that, by an inverse pulse-chase assay, were shown to be trapped in abortive transcription continuously and unable to undergo escape (48). Their existence in vitro has been demonstrated for a broad list of promoters suggesting that RNA polymerase partitions at the binary complex stage to give rise to (at least) two types of open complexes—one more adept at promoter escape to give rise to productive RNA, and the other less adept at promoter escape and trapped predominantly at abortive cycling (93, 97, 102). The assessment of these complexes differs slightly depending on the analysis and the promoter involved. In general, the productive open complexes are more capable at full-length RNA synthesis, but only after going through prior rounds of abortive initiation (93, 102). The less-productive (or unproductive) open complexes perform mostly abortive synthesis and, on some promoters, are moribund, eventually forming dead-end complexes (87).

The occurrence of RNA polymerase partitioning at the open complex stage suggests a branched pathway for transcription initiation (Fig. 3). In vitro, the extent of branching varies according to promoter sequences and reaction conditions such as temperature, salt, and NTP concentrations—with low salt and low initiating nucleotide concentration favoring the formation of moribund complexes (93, 97, 102). The degree of branching adopted at a promoter also depends on the reversibility of the various binary complexes formed at that promoter. High reversibility directly accounts for the low fraction of moribund complexes found on T7 A1 (97). For λ P_R AL promoter (AL indicating the initial transcribed sequence is an A-less cassette), the reversibility is activated by the presence of high concentrations of GreA and the initiating nucleotide (88). Based on the in vitro role of GreA/B on abortive initiation, Susa et al. (98) identified and characterized a group of cellular E. coli promoters whose in vitro and in vivo expression is regulated by GreA/B; these results led to the suggestion that the branched pathway of transcription initiation also operates in vivo.

The existence of moribund complexes can account for the abundance of abortive RNAs synthesized. The formation of moribund complexes, however, is not well understood. What little we know is that their formation appears to be promoter sequence related and structurally based. For lac UV5 promoter, the unproductive complex is reluctant to give up its protein contacts at the −24 region; the relinquishment of this interaction is necessary for promoter escape (95). For λ P_RAL and T7 A1, the RNA polymerase active site appears to be slightly backtracked or forwardtracked, respectively, in the moribund fraction (97).

Promoter Escape Is Coincident with Promoter Release, Not Necessarily Sigma Release

Biochemically, the completion of promoter escape is signaled by the cessation of abortive synthesis seen in a gel assay (36, 38). Conformationally, this event is documented as changes in DNase I and Exo III protection where the footprint of an open complex (from −50 to +20) shrinks to that of an elongation complex (from −5 to +25) (12, 94) through several intermediate stages of protection (47, 61, 62). Although the above studies do not directly monitor the disposition of the sigma subunit on the enzyme, the changes observed were deemed to coincide with the release of the sigma subunit. Recent analyses, however, have demonstrated the association of sigma factor with many elongation complexes (4, 8, 64, 66, 84) and point to the likelihood that the above changes signal promoter release, i.e., the polymerase relinquishing its promoter contacts, and additionally, that sigma release during a transcription cycle occurs in two phases (65, 90). The first phase involves a triggering event very early in initiation that leads to the release of 30 to 40% of sigma (from the T7 A1 promoter), followed by a second phase of stochastic release at a rate dependent on the (greatly reduced) affinity of the remaining sigma-core enzyme complexes (27, 90). The trigger phase may be linked to the synthesis of a critical length of nascent transcript, such as the 8 to 9 nt RNA on a poly-d(AT) template, which brings about the cessation of abortive synthesis (32, 90). The activity and conformation changes accompanying promoter release, therefore, represent changes associated with the trigger phase.

Judging from the widely variable abortive RNA pattern on different promoters, the trigger length of nascent RNA that leads to promoter release may correspond to the size of the longest abortive transcript (38, 39). This size varies according to the promoter and the initial transcribed sequences (see Fig. 1). For rrnB P1, promoter release is accomplished by 3 to 4 nt in vitro with stochastic release of the sigma factor occurring over the next 50 nt in vivo (29, 80). On T7 A1, promoter release occurs after the 9th nt, followed by stochastic release (or sigma retention) (4, 62, 90). On lac UV5 and λ P_R', the longest abortive RNA is 11 and 12 nt, respectively; after releasing the promoter contacts, the σ⁷⁰ subunit on both complexes binds immediately to an adjacent −10-like sequence to induce promoter-proximal pausing (8, 42, 66, 71, 84). This "sigma-hopping" phenomenon may be the result of the −10-like DNA being scrunched and brought to the σ₂ domain on the enzyme (85). On N25 and its various random-ITS derivatives, the initial transcribed sequence determines where promoter release occurs. The default trigger length is 15 nt, but N25 has optimized an ITS sequence for early promoter release at 12 nt (39). Interestingly, a subset of the random-ITS variant promoters produces very long abortive transcripts (VLATs) of 16 to 19 nt in length. Their lengths suggest that they are produced during the promoter escape transition; their abortive release, however, results not from polymerase backtracking, but most likely from hyper-forward translocation (14).

In recent years, several groups have investigated the in vivo relevance of the promoter escape rate limitation and the two-step mechanism of sigma release during cellular gene expression. Using ChIP analysis applied to a select number of E. coli promoters, Wade and Struhl (105) showed that σ⁷⁰ is largely associated with the promoter region during logarithmic growth, suggesting that sigma is released quickly after the initiation-elongation transition. At the highly transcribed rrn promoters, which command ~70% of all cellular transcription activity during exponential phase, σ⁷⁰ was found to undergo stochastic release and sigma switching in vivo (80). Sigma switching verifies that release and exchange of sigma factors occur in vivo. On a genome-wide basis, Reppas et al. (81) probed the distribution of σ⁷⁰ and transcription fragments by using ChIP-chip analysis on the same high-density microarray. Their results confirm the preferential association of σ⁷⁰ with promoter regions, suggesting that once RNA polymerase has gone through the initiation-elongation transition, sigma is rapidly released. However, a low level of σ⁷⁰ was found to be associated with elongating RNA polymerase under specific growth conditions. Most surprisingly, Reppas et al. (81) found ~23% of RNA polymerase bound at the promoter region were not associated with a transcript. These "poised" RNA polymerase molecules may be rate limited at promoter escape in vivo. Thus, in E. coli, the initiation to elongation transition is highly variable and may be rate limiting.

KINETICS AND THERMODYNAMICS OF PROMOTER ESCAPE

Kinetics Analysis

Promoter escape can be the rate-limiting step to productive transcription at many promoters. Conventionally, to measure the rate of promoter escape, a transcription reaction is performed under single-cycle conditions by preforming the open complexes; upon the addition of nucleotide substrates, time-point aliquots are withdrawn to monitor the level of productive RNA synthesis. The results fit to the equation A = A_o (1 − e^−kt), where A is the amount of productive RNA synthesized at time t, A_o is the plateau level of synthesis, and k is the composite rate of promoter escape that encompasses the (fast) rates of the abortive initiation steps prior to escape and elongation steps after escape. This equation describes a single-order event suggesting that promoter escape involves a unimolecular transition. For many promoters analyzed this way (e.g., lac UV5, malT, argI, livJ, and N25_anti), the half-life of escape ranged from 1 to ~10 min (3, 60, 92).

The recent demonstration that RNA polymerase partitions into the productive versus unproductive open complexes at many escape-limited promoters indicates that transcription initiation follows a branched pathway, as depicted in Fig. 3 (48, 93, 102), with the productive complexes giving rise to full-length RNA and the unproductive complexes performing mostly abortive synthesis. The escape rate measurement is based on the level of productive complexes formed at a promoter and should be independent of whether the enzyme follows a branched or sequential (i.e., only the productive branch) pathway of transcription. However, the existence of the branched pathway brings in a level of complexity that the unproductive complexes can convert to the productive fraction through k₋₂ and k₂ steps. Thus, if exchange occurs during the reaction time frame, the k_E measurement may now include the k₋₂ contribution and the interpretation of the kinetic results is more complicated. Furthermore, the productive fraction formed at a promoter varies with reaction conditions. For example, T7 A1 follows a branched pathway in low salt, but predominantly the sequential pathway in high salt (97). On λ P_R, transcript cleavage factors GreA and GreB facilitate productive transcription by converting the unproductive complexes to the productive conformation (88). Thus, in examining the rate of promoter escape, one has to pay attention to the reaction conditions.

Thermodynamic Considerations

High levels of abortive initiation imply the existence of thermodynamic (energy) barriers during the promoter escape process. On average, Eσ⁷⁰ promoters abort over a 10-nt span, indicating that the initiation-elongation transition involves multiple unstable intermediate states that are described quantitatively by the abortive probability profile (Fig. 2) (38). The abortive probability reflects the likelihood that an ITC positioned at a given template position will release its RNA (36). From thermodynamic stability considerations, one can equate the abortive probability of a position with the standard free energy of formation of an ITC (ΔG^o_ITC). The reasoning is as follows.

According to Greive and von Hippel (31), the standard free energy of formation of an elongation complex is the sum of three component parts: ΔG^o_bubble, ΔG^o_hybrid, and ΔG^o_RNAP. ΔG^o_bubble is a destabilizing component requiring energy input to keep a transcription bubble open, whereas ΔG^o_hybrid, depending on the length of the RNA-DNA hybrid, offers compensating stabilizing influence. The third component, ΔG^o_RNAP, includes all of the protein-nucleic acid interactions such as polymerase binding to the RNA-DNA hybrid and the template and nontemplate strands of the bubble, the single-stranded 5'-nascent RNA with the protein exit channel, the 3'-dsDNA with the downstream binding site. The protein-nucleic acid interactions lend further stability to the elongation complexes.

The above consideration can be applied to assess the thermodynamic stability of the initiation complexes—open complexes and initial transcribing complexes— where the ΔG^o_RNAP component includes, in addition, the interaction of 5'-dsDNA with σ₃ and σ₄ domains, as well as protein-protein interactions between the sigma subunit with the core enzyme. Thus, for an open complex,

ΔG^o_{open complex} = ΔG^o_bubble + ΔG^o_hybrid + ΔG^o_RNAP

where ΔG^o_hybrid is zero, and if ΔG^o_RNAP greatly outweighs the ΔG^o_bubble (for an open complex with negligible k₋₂), a highly stable open complex results.

As transcription begins, an open complex is turned into an initial transcribing complex. The standard free energy of formation of an ITC, compared with the open complex, would sustain changes (i.e., ΔΔG^o) to all three components. For an ITC, the transcription bubble expands due to scrunching, introducing additional destabilization (i.e., ΔΔG^o_bubble = +). The RNA-DNA hybrid begins to form and, as it grows in length, will add increasing stabilization to the ITC; thus, ΔΔG^o_hybrid is a negative value. Although, before the hybrid gets to 8 bp in length, the stabilization is probably insufficient to override the other destabilizations occurring in the complex. As initial transcription proceeds, ΔG^o_RNAP undergoes substantial changes. Owing to the unfavorable interaction of protein with scrunched bubble strands of the DNA as well as the displacement of various sigma contacts, this change would be in the destabilizing direction (i.e., ΔΔG^o_RNAP = +). Overall, the ΔΔG^o_ITC is a + value, leading to the formation of a stressed intermediate. The magnitude of the +ΔΔG^o_ITC change varies with each ITC, because sequences of base pairs within the changing parts of the hybrid and transcription bubble differ in thermodynamic stability.

At any initial position, the stressed intermediate represents the posttranslocated form of an ITC that harbors freshly introduced scrunching-induced strain. Before the next nucleotide incorporation step can occur, an elongation complex must reach structural equilibrium at each template position (31, 75). We propose that a structural equilibration step occurs for each initial transcribing complex as well, before it can incorporate the next nucleotide. For an ITC, the structural equilibration is achieved by one of two means—either by propagating the scrunching-induced strain away from the catalytic center in the downstream direction, leading to backtracking, abortive release of the RNA, and resumption of the stable open complex conformation; or by propagating the scrunching-induced strain in the upstream direction to achieve a stable active site conformation that can accept the next NTP (37). The structural equilibration steps are depicted in Fig. 4A. How well the structural equilibration occurs in an ITC is reflected by the abortive probability associated with each position.

Fig. 4Thermodynamic considerations of abortive initiation and promoter escape. Thermodynamic considerations lead to two postulates. (A) Abortive potential predicts the occurrence of structural equilibration of ITCs at each initial position. Shown here are the ITC conformations associated with two successive steps of incorporation, for example, the seventh and eighth step in the N25_anti transcription, where the seventh step is one of high abortive potential and the eighth step is one of low abortive potential. At the nth initial position, at least four conformations of the ITC can be distinguished by their superscript designations: pre, the pretranslocation conformation; sc, the posttranslocation, scrunched conformation; bt, the backtracked conformation; and fi, the forward-incorporating conformation. ITC_n^pre and ITC_n^sc are related through the scrunching translocation mechanism, whereas ITC_n^sc, ITC_n^bt, and ITC_n^fi are related through the structural equilibration steps to stabilize the ITC. ITC_n^sc harbors high strain energy; it can be stabilized through one of two means: by backtracking to form ITC_n^bt, which subsequently releases its RNA to resume the stable open complex structure (RP_o), or by propagating the strain energy away from the active site, forming a stable structure (ITC_n^fi) competent at incorporating the next nucleotide and moving the transcription complex forward. The lengths of the b versus a arrows are drawn to indicate its relationship with abortive probability. (B) The free-energy diagram of promoter escape at a promoter is related to its abortive probability profile. The example shown is that of the N25_anti promoter. RP_o, open complex; EC, elongation complex. See text for discussion on the relationship of abortive probability to the free-energy minimum of an ITC.

By the above reasoning, the abortive probability of an ITC can be equated to the free-energy minimum occupied by that ITC. The height of the free energy well reflects the relative tendency of forward versus backward structural equilibration that occurs at a given step; the higher the backward equilibration, the higher the energy bar. The abortive probability profile of a promoter can be further equated with the thermodynamic free-energy diagram of promoter escape. In Fig. 4B, we show N25_anti as an example. Currently, we don’t know how one energy bar is connected to the next. According to Fig. 4A, there are two steps involved between each pair of bars: one of nucleotide incorporation, and the other, scrunching translocation. Using direct observation of abortive initiation and promoter escape, Margeat et al. (54) found for the lacCONS promoter that abortive product synthesis (i.e., forward incorporation) and RNA polymerase active-center forward translocation during initial transcription (i.e., scrunching translocation) are fast, whereas the abortive product release and RNA polymerase active-center reverse translocation are slow. The rapid rate of abortive transcription in general suggests that the energy barrier between any two initial steps is not high. The question then is what drives the formation of the more highly stressed state of an ensuing ITC (such as between the 6 to 7 nt and the 12 to 13 nt transition in Fig. 4B).

BIOPHYSICAL STUDIES OF ABORTIVE INITIATION AND PROMOTER ESCAPE

The abortive initiation-promoter escape process is a dynamic one involving the formation of many high energy intermediates (i.e., ITCs) during the initiation-elongation transition. The instability of the ITC was predicted to arise from a mechanism of translocation during initiation involving DNA scrunching (37). With the upstream edge of the transcription bubble firmly anchored on the sigma subunit during the abortive initiation phase, DNA scrunching would create a strained transcription bubble. Successive steps of scrunching results in cumulative strain energy eventually sufficient to break loose the promoter DNA-sigma interactions and allow escape to occur.

At the time of this proposal, DNA scrunching has only been documented in a crystal structure of T7 RNA polymerase containing a 3-nt initial transcript (15). It is not known that E. coli RNA polymerase performs scrunching translocation. The probability of capturing an E. coli RNA polymerase crystal structure as an initial transcribing complex seems remote at present (18). To probe this issue, Revyakin et al. (83) used single-molecule nanomanipulation coupled to real-time videomicroscopy to monitor the occurrence of scrunching (which, by further unwinding of the transcription bubble, would bring about shortening or lengthening of positively or negatively supercoiled DNA, respectively) on T5 N25 promoter. By this method, scrunching was indeed observed at single-base-pair resolution to accompany both abortive and productive transcription. The degree of scrunching is RNA length dependent and corresponds to (n − 2) bp. Thus, scrunching is not involved in the synthesis of the 2-nt product, but thereafter is obligatory for downstream incorporation and promoter escape. N25 undergoes promoter escape after the synthesis of an 11-nt RNA. Thus, on the verge of promoter escape, the N25 ITC would have scrunched 9 bp of downstream DNA, generating a strain energy of ~18 kcal/mol that, in turn, can drive the promoter escape transition. That scrunching is the mode of translocation during the initiation stage was further confirmed by single-molecule FRET analysis of E. coli RNA polymerase movement on lacCONS promoter (43).

CONCLUDING REMARKS

Promoter escape, occurring at the initiation-elongation transition, is a key step in regulating productive gene expression. Most E. coli promoters that have been studied show some level of rate limitation at this step, as revealed by the synthesis and release of abortive RNAs. The degree of rate limitation exhibited by a promoter is directly linked to its sequence agreement to the consensus features that ensure the formation of a highly stable open complex (103). With a stable open complex comes the difficulty of escape; this point was made clear by the extensive interactions between promoter DNA and RNA polymerase protein in the structure of a bacterial RNA polymerase open complex (69). Escape requires the disruption of most, if not all, of these interactions. By commencing RNA synthesis, RNA polymerase generates two factors internally to mediate the disruption process. One is the growing nascent RNA that, upon reaching 6 to 8 nt in length, will begin to displace the sigma subunit-core enzyme interactions (49). The second factor results from the scrunching mechanism of translocation used by RNA polymerase during initial transcription to accrue strain energy in the growing transcription bubble, which eventually displaces the sigma-promoter DNA interactions (83). Here, the initial transcribed sequence region of a promoter comes into play to direct the escape process (39). The sequence composition of the ITS affects both factors above to alter the escape efficiency at a promoter, but the mechanism is not fully understood. The great progress made so far on abortive initiation-promoter escape research positions us well to look into the kinetic and thermodynamic details of this important transitional phase in transcription.