Step-by-Step Regulation of Productive and Abortive Transcription Initiation by Pyrophosphorolysis

Abstract

An understanding of the kinetics and mechanism of bacterial transcription initiation is needed to understand regulation of gene expression and advance fields from antibiotic discovery to promoter design. The step-by-step forward kinetics and mechanism of initiation and RNA-DNA hybrid growth, made irreversible by omitting pyrophosphate (PPi) byproduct, were determined recently for E. coli RNA polymerase (RNAP)-λP_R promoter complexes. Strong position-dependences of overall rate constants (k_cat/K_m analogs) for each nucleotide-addition step were observed because of coupling of hybrid growth to disruption of promoter contacts, bubble closing, and RNAP escape. Here we investigate reversal of these steps (pyrophosphorolysis) at PPi concentrations ([PPi]) found in exponentially-growing cells. We quantify [PPi] effects on the amount and rate of synthesis of long (>10-mer, post-escape) and short (stalled, abortive) RNA to determine how PPi regulates initiation. Physiological [PPi] makes uridine incorporation and some other initiation steps significantly reversible. Physiological [PPi] reduces the fraction of RNAP-promoter complexes that productively initiate and the rate of RNA synthesis per productive complex, while increasing the fraction of complexes that abortively initiate, affecting abortive rates, and shifting the abortive-product distribution to shorter RNAs. Pyrophosphorolysis rates for some initiation complexes are orders of magnitude larger than for removal of the same nucleotide from elongation complexes because of the strong bias toward the pre-translocated state in initiation, and exhibit even stronger dependences on nucleotide identity (pyrimidine ≫ purine). Because cytoplasmic [PPi] is much higher in exponential-phase than stationary-phase cells, these [PPi] effects on initiation rates and amounts of RNA synthesis must be physiologically-relevant.

Introduction

Transcription is a fundamental life process and nuanced transcriptional regulation is consequently vital. Because this regulation largely occurs in initiation, it is very important to understand the mechanism of initiation and how the kinetics and equilibria of individual steps are determined by the sequences of the promoter and initial transcribed region (ITR) and their interaction with RNA polymerase (RNAP), as well as by physiological solution variables including temperature and concentrations of reactants ([NTP]) and byproduct (pyrophosphate, PPi). Of these variables, effects of [PPi] on initiation have been least explored. This information is needed because [PPi] in the E. coli cytoplasm increases greatly from stationary phase (low μM) to exponential growth (~1 mM), and also varies with growth conditions [1–3].

Bacterial transcription initiation begins with recognition of each unique duplex DNA promoter sequence by RNAP [4–7]. Initial binding of E. coli RNAP to the λP_R promoter directs remodeling of the duplex to open 13 base pairs (bp) and form a relatively-unstable open-promoter intermediate (I₂) [7–9]. Interactions of the RNAP jaw and other mobile elements with the downstream duplex, directed by the discriminator region of the promoter, convert I₂ at some promoters to a more stable OC (I₃, RP_O) [10, 11]. The partially-stabilized open intermediate I₃ is the λP_R initiation complex (IC) that binds NTPs and begins RNA synthesis [12, 13]. Structures of two λP_R open complexes (OC) proposed to be I₃ and the 37 °C stable OC (RPo) were recently reported [13].

Kinetic-mechanistic studies (both ensemble [12, 14–16], and single molecule [17–19]) together with structural studies [13, 20–23], have clarified key aspects of the E.coli initiation mechanism in the presence of NTPs. Analysis of the kinetic data yields overall [15, 16, 24] and step-by-step [12, 15] kinetics of irreversible hybrid growth and escape of RNAP from the promoter. After synthesis of the initiating dinucleotide, the cycle of steps to add a nucleotide to the 3’ end of the RNA in initiation is well-described by the minimal mechanism of Fig. 1 [12, 15], which is the same as in elongation [25, 26]. After the previous catalytic step, reversible translocation of the RNA-DNA hybrid establishes an equilibrium distribution of pre- and post-translocated states. In the post-translocated state the active site is accessible to the incoming NTP as a result of unfolding of the trigger helix [26, 27]. Reversible NTP binding stabilizes the post-translocated state. In this step, refolding of the trigger loop closes the active site for catalysis and aids in properly positioning the incoming NTP [27]. In the catalytic step of Fig. 1 the triphosphate of the NTP is split to add the nucleotide monophosphate (NMP) to the hybrid and form the next pre-translocated state with PPi as byproduct. Without added PPi, the catalytic step in initiation is effectively irreversible. Addition of PPi increases the rate of the reverse catalytic step (Fig. 1; pyrophosphorolysis), regenerating the NTP from PPi and the 3’-NMP [26, 28, 29].

Figure 1: Mechanism of nucleotide addition and its reversal (pyrophosphorolysis) in initiation.

In this schematic, DNA bases are represented by blue circles, RNA bases as orange triangles, PPi as a red rectangle, the incoming nucleotide triphosphate (NTP) by an orange triangle, and the active site Mg²⁺ as a green circle. The four substeps are: 1) Reversible translocation of the RNA-DNA hybrid relative to the active site. The pre-translocated state of the initiation complex ($I{C}_i^{pre-tr}$) in initiation step i is converted to the post translocated state ($I{C}_i^{\mathit{\text{post}}-tr}$) with equilibrium constant ${\text{K}}_{\text{i}}^{\text{tr}}$, forming the binding site for the i-th NTP. 2) Reversible binding of the i-th NTP to the active site (grey) with equilibrium constant ${\text{K}}_{\text{i}}^{\text{NTP}}$. 3) Catalysis of phosphodiester bond formation between the i-th NTP and the 3’ end of the RNA with rate constant ${\text{K}}_{\text{cat},\text{i}}^{+}$, with PPi (red) as byproduct still bound in the RNAP active site. 4) Release of PPi. Addition of PPi makes the PPi binding step (equilibrium constant ${\text{K}}_{\text{i}}^{\text{PPi}}$) significant and makes the catalyzed reaction reversible (pyrophosphorolysis), re-forming the i-th NTP with rate constant ${\text{k}}_{\text{cat},\text{i}}^{-}$.

In elongation by bacterial RNAP, the observation of similar kinetics of translocation and of PPi release indicates that PPi release occurs shortly before or concurrently with translocation [30]. Pre- and post-translocated complexes are in rapid equilibrium on the timescale of NTP binding. In elongation, a bias for the post-translocated state is observed, indicating that the translocation equilibrium constant of an elongation step K_tr,i > 1 [26, 30]. In initiation the opposite situation holds. Recent results demonstrate that most translocation steps in initiation are highly unfavorable, with translocation equilibrium constants K_{tr, i} ≪ 1 [12, 15]. These small K_tr,i result from translocation stresses unique to initiation, including opening of one downstream DNA bp (when not offset by upstream bp closing), steric stress [20, 21, 31], scrunching of the bubble strands [32–34], and disruption of RNAP-promoter contacts [12, 15].

Small K_tr,i of initiation steps reduce the overall 2^nd order forward rate constants k_i (the analog of k_cat/K_m for this situation) for incorporation of the next nucleotide and hybrid growth, relative to elongation. Values of k_i for most steps of nucleotide addition in initiation that begin with translocation are in the range 0.01 – 0.2 μM⁻¹ s⁻¹ at 19 and 25 °C [12], much smaller than those for initial dinucleotide synthesis (without translocation; ~ 0.6 μM⁻¹ s⁻¹ at 19 and 25 °C [12]) and for the NTP binding and catalytic steps of elongation (≥ 0.6 μM⁻¹ s⁻¹ at 24 °C [35, 36]) Conversely, translocation stress in initiation is predicted to greatly favor the reverse process of pyrophosphorolysis relative to elongation.

Effects of PPi on the amounts and rates of productive and abortive RNA synthesis and on the reversible kinetics of the individual steps of transcription initiation and escape of RNAP from the promoter have not previously been systemically investigated. Scaffolded promoter constructs with artificial mismatches that mimic the initiation complex and artificially paused complexes have been employed to investigate pyrophosphorolysis in initiation at or near the escape point [28]. Pyrophosphorolysis in elongation has been characterized in both scaffolded [30, 37, 38] and continuously-elongating complexes [39, 40]. The sensitivity of the elongation complex to [PPi] is highly sequence-specific for several bacterial RNAP, including E. coli [30, 38]. Elongating transcripts with a 3’ U are the most sensitive to pyrophosphorolysis, with the rank order U ⋙ C > A > G. These preferences are proposed to arise from differences in the strength of the favorable interaction of the 3’ base with the folded trigger helix [30, 38, 41]. These interactions stabilize the pre-translocated state, increasing the rate of pyrophosphorolysis at that position. The penultimate 3’ nucleotide has a smaller but significant effect, in the opposite order (G > A > C ≫ U). Overall 2^nd order rate constants estimated from published results [38] for pyrophosphorolysis in elongation are very small (< 10⁻⁵ μM⁻¹ s⁻¹; Table S2) for both E. coli RNAP and T.th. RNAP. For continuously-elongating complexes, addition of 1 mM PPi reduces the average elongation rate to approximately half the baseline (1 μM PPi) rate [39].

Previously we investigated initiation of a single round of long RNA synthesis at two model promoters (λP_R, T7A1 [14]). We found there are two phases of initiation and two approximately equal subpopulations of initiating complexes, designated productive and nonproductive. These findings are consistent with earlier studies of initiation time courses where long RNA synthesis was single-round [42]. Fixed-time experiments in which synthesis of long RNA is multi-round are incapable of distinguishing these subpopulations. In the faster phase of initiation (< 10 – 20 s for the NTP concentrations investigated at 37 °C), productive complexes initiate and processively extend the RNA-DNA hybrid without stalling or aborting. RNAP escapes from the promoter after synthesis of a 10-mer (λP_R) or 7-mer (T7A1) by these productive initiation complexes. In this faster phase of initiation, nonproductive initiation complexes stall after synthesizing a short RNA (10-mer or less for λP_R, 7-mer or less for T7A1). At longer times, in a second phase of initiation, some of these stalled nonproductive complexes release their RNA and reinitiate (abortive synthesis) at a rate which decreases in general with increasing RNA length and with decreasing temperature [12, 14, 15, 42].

We subsequently determined the step-by-step kinetics and mechanism of initial transcription at the model λP_R promoter [12, 15]. We found that in initiation at this promoter, overall 2^nd order rate constants (k_cat/K_m analogs) for steps of nucleotide addition involving translocation exhibit a repeating pattern of large and small forward rate constants which we interpreted in terms of the stepwise disruption of RNAP-promoter contacts and collapse of the upstream initiation bubble. We deduced that in-cleft contacts are broken first, followed by disruption of −10 contacts and then upstream (−35) contacts, allowing promoter escape. These previous studies were done in the absence of PPi, where all catalytic steps are irreversible.

Here, from initiation kinetic measurements at 37 °C and 19 °C, we determine the large effects of [PPi] in the low-mM range on the fraction of RNAP- λP_R complexes that are productive and synthesize a long (post-escape) RNA, and on the rate of long RNA synthesis by these productive complexes. From rapid-mixing experiments at 19 °C we determine the large effects of [PPi] on each step of productive initiation. Analysis of these 19 °C kinetics in the context of our previous results for irreversible initiation without added PPi reveals that [PPi] effects result from pyrophosphorolysis and yield composite 2^nd order rate constants for each pyrophosphorolysis (back reaction) step from 16-mer to 3-mer. We compare the magnitudes and nucleotide (ultimate, penultimate) dependences of pyrophosphorolysis rate constants with previous observations of PPi effects on elongation steps [38]. From manual-mixing experiments at 37 °C we determine the large effects of [PPi] on the fraction of the OC population that is nonproductive, on the length-distribution of the short RNAs synthesized by nonproductive complexes (NPC) before stalling, and on the abortive synthesis rates of individual short RNAs.

We conclude that effects of [PPi] on initiation result from pyrophosphorolysis and that PPi effects on steps of initiation parallel those observed in elongation, but are even more profound. Based on our results and analysis, we propose that physiological [PPi] affects amounts and rates of productive (and also abortive) initiation in a promoter-specific manner. Because cytoplasmic [PPi] varies with growth conditions and attains low mM concentrations in exponential growth, we deduce that [PPi] is a global regulator of transcription initiation and gene expression.

Results

Time Courses of Transcription Initiation by Productive and Nonproductive Complexes at 37 °C and Physiological Levels of Pyrophosphate (PPi)

The kinetics of transcription initiation by E. coli RNAP at the λP_R promoter were investigated over a wide range of [PPi] (0.1 mM – 3 mM) that bracket that observed in exponentially-growing E. coli [1–3]. OC were formed between RNAP and a 122 bp linear λP_R promoter (−80 to +42) with a modified ITR (Fig. 2A) to halt transcription at 16-mer when CTP is withheld. All experiments include the polyanionic competitor heparin to ensure that long RNA synthesis is single-round. For Fig. 2 experiments, these pre-formed OC were manually mixed with NTPs at 37 °C to initiate transcription. To obtain efficient initiation, concentrations of the two initiating NTP (ATP, UTP) were 200 μM, while that of the labelling NTP (GTP) was 10 μM unlabelled, 17.5 nM α-³²P-GTP. Use of a low GTP concentration ensures efficient incorporation of the radiolabel and improves detection of transient intermediates in 16-mer RNA synthesis [12, 15]. RNAP escapes from its contacts with the λP_R promoter DNA upon synthesis of 11-mer RNA [14, 43]. Long (post-escape) RNA is defined as the sum of all RNA species ≥ 11-mer (referred to as 11+ RNA) observed on gels like Fig. 2B.

Figure 2: Effects of PPi on Synthesis of Long RNA at 37 °C.

Panel A) The nontemplate-strand sequence of the core region of the modified λP_R promoter (in a 122 bp (−80 to −42) promoter fragment) used in this and previous studies of step-by-step initiation kinetics [12]. Panel B) Single-nucleotide resolution PAGE separations of ³²P-labelled RNA products at different times in initiation (10–480 s after addition of 200 μM ATP, 200 μM UTP, 10 μM GTP, 17.5 nM α-³²P-GTP to preformed OC) in manual-mixing experiments at 0 mM and 2mM PPi. Omission of CTP causes a stop at 16-mer, with subsequent readthrough to a second stop at 31-mer. Different run-times were used for the two gels. The partial lane at left is from an unrelated experiment. Panel C) Time courses of synthesis of long RNA (11-mer or longer, designated as 11+) in single-round experiments at ≤ 3 mM PPi at 37 °C. Amounts of 11+ RNA are normalized by the total amount of OC. Single exponential fits are shown as solid curves. Error bars show one standard deviation from the average of normalized data (see Methods) from 3–4 experiments. Panel D) Half-times t_1/2 (t_1/2 = 0.69/k₁₁₊) for synthesis of 11+ RNA (black) and amounts of 11+ RNA synthesized (red) are plotted for each PPi concentration at 37 °C. Experimental uncertainties are approximately ±20% for all points. Without added PPi, t_1/2 is too small to determine by manual mixing. Fast-mixing experiments at 37 °C gave t_1/2 ≈ 1.8 s (k₁₁₊ 0.4 s⁻¹) [12]. Dashed lines are included as a visual guide.

Figure 2B shows representative gel separations of quenched samples taken at the indicated times in initiation without added PPi and with 2 mM added PPi. The gel at left, from an experiment without added PPi, is similar to those obtained for a different set of NTP concentrations (200 μM ATP,GTP; 10 μM UTP, 17.5 nM α-³²P-UTP) and analyzed previously [14]. These experiments demonstrate two kinetic phases of initiation and two subpopulations of complexes (productive, nonproductive). In the faster phase, complete in the first 10–20 s without added PPi for these conditions, RNAP in productive complexes escapes from the promoter to synthesize an 11+ RNA, while nonproductive complexes stall after synthesis of a shorter RNA. At 37 °C the kinetics of this first phase are too fast to determine by manual mixing, but were previously determined in fast-mixing (rapid quench flow, RQF) experiments with a <10 ms deadtime [12]. Manual mixing experiments like those in the left panel of Fig 2B without added PPi provide information about the endpoint amounts of 11+ and shorter RNAs synthesized in this faster phase. At longer times release of some short RNAs from nonproductive complexes (e.g. 7-mer in Fig. 2B left panel) results in re-initiation and abortive RNA synthesis which continues for the duration of this 480 s experiment. Heparin does not inhibit re-initiation and abortive synthesis by nonproductive complexes because RNAP remains bound to promoter DNA during this process [14]. In this slower kinetic phase, significant readthrough of the stop at +16 is observed, as in previous studies [14], and is accounted for in calculations of amounts of 11+ RNA.

The right panel of Fig. 2B shows the qualitatively different behavior of both short and long RNA synthesis in the presence of 2 mM PPi. This gel reveals slow kinetics of synthesis of 11+ RNA by productive complexes at 2 mM PPi, occurring over 480 s and accurately determined in these manual mixing experiments. In addition, the fraction of the OC population that synthesizes an 11+ RNA is visibly reduced at 2 mM PPi as compared to 0 mM PPi, while the fraction of OC synthesizing only short RNA increases at 2 mM PPi.

Much more extensive readthrough of the stop at 16-mer is observed in these 2 mM PPi experiments than at 0 mM added PPi, resulting in synthesis of 31-mer at the next stop point where CTP is required [15]. No significant readthrough of the 31-mer stop (and no run-off 42-mer RNA) is observed. Readthrough bands are included in the calculations of amounts of 11+ RNA for Fig. 2. Other than this, readthrough does not affect the interpretation of these 37 °C experiments.

Effects of PPi on the Time Required for Long (11+) RNA Synthesis and on the Fraction of Productive OC at 37 °C

Results of the experiments in Fig. 2B and other 37 °C experiments at [PPi] from 0 – 3 mM are analyzed in Fig. 2C, which plots the amount of 11+ RNA synthesized (normalized to total pre-formed OC) vs. time at these different [PPi]. Fig. 2C is truncated at 240 s for clarity; Fig. S1 extends these plots to include the 480 s time points. With increasing [PPi], the fraction of OC that synthesize 11+ RNA decreases and the time required for single-round synthesis of 11+ RNA increases greatly. Fig. 2D quantifies these trends.

Fig. 2D indicates that without added [PPi], for the λP_R promoter construct and conditions investigated, about half (40–50%) of OCs are productive, capable of synthesizing 11+ RNA upon NTP addition in the time interval examined (480 s). A similar result was obtained previously at a different NTP condition [14]. As [PPi] is increased to 1 – 2 mM, Fig. 2D shows that the fraction of OCs that behave as productive is reduced to about 20% of the OC population, while at 3 mM PPi no OC behave as productive and 11+ RNA synthesis is eliminated.

Previous fast-mixing experiments performed in the absence of PPi for the conditions investigated here found that the approach of the amount of 11+ RNA to its plateau value exhibits first order kinetics with a t_1/2 of ~1.8 s (rate constant k₁₁₊ ≈ 0.4 s⁻¹) after an initial short lag (~0.4 s) [12]. Fig. 2D shows that addition of PPi drastically slows the synthesis of 11+ RNA, allowing t_1/2 values to be determined by manual mixing. At 1 mM PPi, t_1/2 of 11+ RNA synthesis is ~25 s, a ~14-fold increase in t_1/2 as compared to 0 mM added PPi. At 2 mM PPi, t_1/2 is ~70 s, almost a 40-fold increase as compared to 0 mM added PPi. For comparison, a ~2-fold effect of 1 mM PPi was observed on the rate of in vitro elongation relative to 1 μM PPi [39].

Fast-mixing initiation kinetic experiments, performed at 19 °C with a rapid-quench-flow (RQF) mixer and described in the following sections, are needed to determine step-by-step kinetics and mechanism of long RNA synthesis as a function of [PPi] and test the hypothesis that the large effects of PPi are the result of pyrophosphorolysis. These step-by-step kinetics are more accurately determined at 19 °C than at higher temperatures because of the larger number and greater amplitudes of short RNA transients and because rates of abortive synthesis by nonproductive complexes are reduced at this lower temperature [12, 15]. On the other hand, 37 °C is a better choice of temperature to characterize PPi effects on initiation by nonproductive complexes because abortive rates are larger, transients in long RNA synthesis are reduced, and the rate of long RNA synthesis is larger [12] so one obtains a better separation of contributions to initiation from nonproductive vs productive complexes.

Fast Kinetic Analyses: Calculation of Rate Constants for Step-Wise Conversion of 11-mer to 16-mer Without Added PPi at 19 °C

Forward rate constants (2^nd order; k_cat/K_m analogs) for each nucleotide incorporation step of irreversible initiation by productive OC up to and including 11-mer synthesis were previously determined for the λP_R promoter without added PPi at 19 °C and higher temperature [12, 15]. In our previous research, all RNA lengths beyond the escape point were defined as full-length and their amounts as a function of time were summed to obtain the kinetics of synthesis of full-length (11+) RNA. An extension of that mechanism and step-by-step kinetic analysis to the induced pause after 16-mer synthesis in the absence of added PPi is now needed to accompany the quantitative analysis of stepwise pyrophosphorolysis rates. This extension of the minimal kinetic mechanism is given in Fig. S2; all forward steps are irreversible when no PPi is added. The left panels of SI Fig. S3 show the irreversible kinetic data for 12-mer to 16-mer at 19 °C and 0 mM added PPi. Amounts of 12-mer and 13-mer RNA increase to maxima at around 10 s and then decrease, 14-mer and 15-mer RNA increase to maxima at 15–20 s before decreasing, and 16-mer RNA increases to a stable plateau at times > 50 s because of the slow rate of readthrough (quantified below) for these 19 °C experiments without added PPi. Analysis of these results yields composite 2^nd order rate constants ${\text{K}}_{\text{i}}^{+}$ for each post-escape step of nucleotide incorporation up to 16-mer (Table S1). Values of ${\text{K}}_{\text{i}}^{+}$ are plotted in Fig. 3A, and the good fits to the time courses at the two NTP conditions investigated in that research [15] using these rate constants are shown in Fig. 3B and C. Labelling of these longer RNAs at multiple positions makes it possible to quantify their small mole amounts and determine rate constants accurately.

Figure 3: Composite 2^nd-Order Rate Constants and Fits to Kinetic Data for Post-Escape Steps of Nucleotide Incorporation at the λP_R Promoter Without Added PPi at 19 °C.

Panel A plots 2^nd-order rate constants ${\text{K}}_{\text{i}}^{+}$ (k_cat/K_m analogs) for steps of nucleotide incorporation after promoter escape (11-mer synthesis) up to the pause at 16-mer induced by withholding CTP. These ${\text{K}}_{\text{i}}^{+}$ values are reported in SI Table S1 together with previously-determined ${\text{K}}_{\text{i}}^{+}$ for steps up to the escape point [12, 15]. Panels B and C compare fast-mixing data (averages of 3 experiments) with fits (solid curves) of kinetic data at two sets of NTP concentrations, obtained using these composite rate constants. Amounts of RNA lengths shorter than 11-mer are consolidated for clarity.

Values of rate constants ${\text{k}}_{\text{12}}^{+}$ and ${\text{k}}_{\text{13}}^{+}$ for the first two post-escape steps (synthesis of 12-mer, 13-mer) are progressively larger than ${\text{k}}_{\text{11}}^{+}$ for the 11-mer escape step, but ${\text{k}}_{\text{14}}^{+}$ smaller than ${\text{k}}_{\text{13}}^{+}$ (Fig 3A; Table S1). The progression to larger ${\text{K}}_{\text{i}}^{+}$ values resumes for 15- and 16-mer rate constants (${\text{k}}_{\text{15}}^{+}$ and ${\text{k}}_{\text{16}}^{+}$) Table S1). While these post-escape rate constants (ranging from 0.08 μM⁻¹ s⁻¹ to 0.2 μM⁻¹ s⁻¹) are at the high end of the range of pre-escape rate constants (Table S1), all are less than previously reported rate constants for initial dinucleotide synthesis (without translocation [12]; ~ 0.6 μM⁻¹ s⁻¹ at 19 and 25 °C) and for the combined NTP binding and catalytic steps of elongation (≥0.6 μM⁻¹ s⁻¹ at 24 °C [35, 36])

Fast Kinetic Analyses: Profound Effects of PPi on the Step-by-Step Kinetics of Initiation by Productive λP_R OC at 19 °C

Effects of PPi on the step-by-step kinetics of transcription initiation by productive complexes were investigated by fast-mixing (RQF) experiments at 19 °C. Fig. 4A shows representative PAGE gels comparing time courses of initiation of RNA synthesis at 0 mM and 2 mM added PPi, and a representative gel of an RQF experiment at 1 mM PPi is shown in SI Fig. S4. Phosphorimager analysis of these gels provides quantitative single-nucleotide resolution of ³²P-labelled RNA products as a function of time after NTP addition to preformed λP_R open complexes (which, at 19 °C are predominantly I₃, the initiation complex) [12]. Long RNA synthesis in these experiments is single-round because heparin is added with the NTPs to bind any RNAP that dissociates from the DNA.

Figure 4: Effects of [PPi] on the Step-by-step Kinetics of Initiation at 19 °C.

Panel A) Representative gels comparing single-nucleotide resolution PAGE separations of ³²P-labelled RNA products from RQF initiation experiments at the λP_R promoter at 2 mM PPi, 200 μM ATP, UTP, 10 μM GTP with results at 0 mM added PPi and the same NTP concentrations [12]. Synthesis of 16-mer RNA in these experiments is single-round. Panel B) Time courses of amounts of 5-mer (Top), 9-mer (Middle), and 12-mer (Bottom) RNA at 0 mM (Black), 1 mM (Green), and 2 mM PPi (Red). Points shown are averages of 3 experiments. Amounts are normalized to the total amount of 11+ RNA synthesis in each single-round experiment before averaging. Different vertical scales are used for each RNA length. Dashed lines are included as a visual guide.

Visual comparison of the 0 mM and 2 mM PPi gels in Fig. 4A reveals that the kinetics of long RNA synthesis by productive complexes are almost an order of magnitude slower at 2 mM added PPi than at 0 mM added PPi. Transients in short RNA intermediates on the pathway to long RNA synthesis by productive complexes are shifted to longer times and broadened at 2 mM added PPi. Additionally, the amount of short abortive products is much greater at 2 mM added PPi than at 0 mM added PPi. Quantitative comparisons are made in subsequent figures. More readthrough of the 16-mer stop, resulting in 31-mer synthesis, is observed at 2 mM PPi than without added PPi (Fig 4A; see also Fig 2B). A first order readthrough step with rate constant k_rt is included in the overall initiation mechanism in Fig S2, and values of k_rt are obtained at each [PPi].

Fig. 4B quantifies the kinetic behavior of two pre-escape RNAs (5-mer, 9-mer) and a post-escape RNA (12-mer) as a function of [PPi] (0, 1, 2 mM) at 19 °C. Total amounts of these RNA lengths, including initiation from both productive and nonproductive complexes, are plotted as a function of time. Results for 5-mer and 9-mer at 0 mM added PPi were analyzed previously [12, 15]. Results for 12-mer at 0 mM added PPi are from experiments analyzed in Fig. 3. Significant transients are observed at these three lengths, in part because the next base added is G and the low concentration of GTP in the reaction mixture reduces the rate of extension to 6-mer, 10-mer, and 13-mer respectively. With increasing [PPi], rates of formation and of subsequent decay of all three of these intermediate RNA species decrease because each step of RNA synthesis becomes more reversible. This is expected from the observed behavior of long RNA with increasing [PPi] in both the 37 °C experiments in Fig. 2 and in these 19 °C experiments in Fig. 4A.

In general, after the increase and decrease in amount of a short RNA (≤ 10-mer for λP_R) in the transient phase, a non-zero plateau or slow increase is observed, where the plateau indicates the amount of RNA synthesized in the first round by NPC that stall at that length, and the slow increase is from re-initiation after slow release of this first short RNA by NPC (abortive initiation) [14, 15]. Without added PPi, Fig. 4B shows that at 0 mM added PPi there is little synthesis of 5-mer and 9-mer RNA by NPC, either initially or abortively, so the plots return to zero after the transient. This is because λP_R NPC at this this low-GTP, 19 °C condition primarily synthesize 3-mer, with smaller amounts of 4-, 7-, 8- and 10-mer, and abortive rates are only significant for 3-mer and 4-mer (Figs. S5–S6; see below). After the transient phase the amount of 12-mer (a post-escape RNA) decays to zero as expected.

As [PPi] is increased to 1mM and 2 mM. Fig. 4B shows that post-transient amounts of 5-mer, 9-mer, and 12-mer all increase greatly. Figs. S5–6 show similarly large increases for 3-mer and 4-mer, but not for most other RNA lengths. Abortive rates remain small for 5-mer and 9-mer at 1 mM and 2 mM PPi. For 12-mer, which is post-escape and therefore synthesized only by productive complexes, the return of the transient peak to zero at 0 mM added PPi indicates that synthesis of 13-mer is irreversible for this condition. Development of a significant 12-mer plateau at 1 mM and 2 mM PPi demonstrates unequivocally that 13-mer synthesis is reversible at these PPi concentrations, and that the bottleneck at 16-mer causes long-time stable plateaus (on the time scale of these experiments) in the amounts of preceding long RNA intermediates (11-, 13-, 14-, and 15-mer in Fig. S5–6).

Fig. S7 plots time courses of long (11+) RNA synthesis at 19 °C from the above RQF measurements at 0 mM, 1 mM and 2 mM PPi. The fraction of OC that are productive and synthesize an 11+ RNA is reduced by these PPi concentrations and the half time of 11+ RNA synthesis is increased, as observed at 37 °C (Fig. 2 C, D). The effect of PPi on the half time of 11+ RNA synthesis is not as large at 19 C as at 37 C. At 19 °C t_1/2 increases from approximately 7 s at 0 mM PPi to 20 s at 1 mM PPi and 34 s at 2 mM PPi, while the fraction of OC that are productive is halved at these mM PPi concentrations (Fig. S7).

Rate Constants of Pyrophosphorolysis Steps of Initiation and Escape Span a 1000-fold Range

Second order rate constants ${\text{K}}_{\text{i}}^{-}$ for the overall process of PPi uptake and pyrophosphorolysis are obtained from the analysis of the kinetic data (including that of Fig. 4A) at 1 and 2 mM PPi according to the mechanism of Fig S2, which includes the first order kinetics of readthrough of the 16-mer stop to form 31-mer with rate constant k_rt. These reverse rate constants (${\text{K}}_{\text{i}}^{-}$) are plotted together with forward rate constants (${\text{K}}_{\text{i}}^{+}$) on a linear scale in Fig. 5A, and ${\text{K}}_{\text{i}}^{-}$ values are plotted on a logarithmic scale in Fig 5B to illustrate the wide range (>10³) in values of ${\text{K}}_{\text{i}}^{-}$ for different steps of initiation. All rate constants (${\text{K}}_{\text{i}}^{+}$, ${\text{K}}_{\text{i}}^{-}$, k_rt) are given in Table S1. Observed time courses of all RNA intermediates in initiation by productive complexes (and of the readthrough 31-mer) are compared with predictions from the fitted rate constants at each [PPi] (0, 1 and 2 mM) in Fig. 5C.

Figure 5: Step-by-step Variations in 2^nd Order Pyrophosphorolysis Rate Constants in Initiation at 19 °C; Comparison with 2^nd Order Nucleotide Incorporation Rate Constants.

Panel A: Black data points and curve are composite 2^nd order rate constants ${\text{K}}_{\text{i}}^{+}$ for each forward step of nucleotide incorporation that begins with translocation. Red data points and curve are composite 2^nd order pyrophosphorolysis rate constants ${\text{K}}_{\text{i}}^{-}$ for each reverse step (See Figs. 1 and S2 for mechanism; rate constants ${\text{K}}_{\text{i}}^{+}$ are from Fig 3A and Table S1). Panel B: Log scale plot of pyrophosphorolysis rate constants ${\text{K}}_{\text{i}}^{-}$ from Panel A, separated by shading into steps with large, moderate, and small pyrophosphorolysis rate constants. Panel C: Comparison of RQF data (averages of 3 experiments) with fits (solid curves) for all RNA intermediates (3–15mer) in synthesis of 16-mer RNA (black) and subsequent slow readthrough (rt). Fits are shown for 0 mM PPi, 1 mM PPi, and 2 mM PPi.

We assume in these fits that the enzymatic Michaelis constant for PPi (K_{m, PPi}) is larger than 2 mM (i.e. a small PPi binding constant 1/K_{m, PPi} < 500 M⁻¹) so that a single composite rate constant ${\text{K}}_{\text{i}}^{-}$ (the analog of k_cat/K_m) is sufficient to describe the kinetics up to 2 mM PPi. This assumption is justified by the expectation that PPi binding should be weak and by the quality of the fits obtained. In these fits, newly-determined forward rate constants for post-escape steps and previously-determined forward rate constants for pre-escape [12] steps are fixed at their values without added PPi in order to test whether pyrophosphorolysis by itself, without other PPi effects, can explain all the observed changes in the kinetics of short and long RNA synthesis by productive complexes at 1 and 2 mM PPi.

Fig. 5B shows that pyrophosphorolysis ${\text{K}}_{\text{i}}^{-}$ values span a very wide range (over 3 orders of magnitude). Steps 4, 7, and 15 have relatively large ${\text{K}}_{\text{i}}^{-}$ values (0.01 – 0.1 μM⁻¹s⁻¹). Steps 6, 9, 10, 12, 13, and 16 are in the intermediate range of ${\text{K}}_{\text{i}}^{-}$ values (0.0005 – 0.01 μM⁻¹s⁻¹). Steps 3, 5, 8, 11, and 14 have very small ${\text{K}}_{\text{i}}^{-}$ values (<0.0005 μM⁻¹s⁻¹) (Fig. 5B). Values of ${\text{K}}_{\text{i}}^{-}$ for this last group of steps are of similar magnitude to pyrophosphorolysis rate constants in elongation by E. coli RNAP, which are on the order of 10⁻⁵ μM⁻¹ s⁻¹ or less (Table S1) as calculated from published results [38]. Values of ${\text{K}}_{\text{i}}^{-}$ for nine of the fourteen initiation steps quantified are much larger than for elongation. As discussed below, ${\text{K}}_{\text{i}}^{-}$ values of these initiation and post-escape steps correlate well with the identity of the 3’ RNA base, as previously reported for elongation steps [36].

Effects of PPi on Synthesis of Short RNAs by Nonproductive Complexes

For the reasons stated above, we chose 37 °C manual-mixing experiments [14] like those of Fig. 2 to investigate effects of [PPi] on synthesis of short RNA by nonproductive complexes (NPC). Results in Fig. 6 reveals that PPi at physiological concentrations has very large effects on the fraction of complexes that are nonproductive, on the initial amount of short RNA of each length synthesized by NPC and on the rate of abortive synthesis of these RNA lengths. NPC initiate a first round of RNA synthesis but are incapable of escaping the promoter to make a 11+ RNA and stall after synthesis of a short, pre-escape length (i.e. 3-mer to 10-mer) RNA. The time required for the initial round of short RNA synthesis by NPC at 37 °C is not accurately known, but is probably similar to that for long (11+) RNA synthesis by productive complexes (t_1/2 ≈ 1.8 s without added PPi; Fig. 2D). Stalled NPC slowly release their RNA and reinitiate (abortive initiation). At 37 °C the nonproductive fraction of the OC population that stalls at each RNA length is quantified by extrapolating the linear abortive-synthesis kinetic data to time t = 0 [14]. The bar graph of Fig. 6A quantifies the effect of PPi on the fraction of λP_R initiating complexes detected (combined productive and nonproductive complexes) in our assays that are nonproductive at 37 °C for the NTP concentrations investigated here and compares with results obtained at 19 °C from experiments like those in Fig 4.

Figure 6: Effects of [PPi] on the Fraction of Nonproductive Complexes (NPC) and on the Initial RNA Length Distribution and Subsequent Abortive Rates of these NPC.

Panel A) Fraction of detected initiating complexes that are nonproductive (stalling after synthesizing a short RNA (<11-mer) at 37 °C and 19 °C. Panel B) The fraction of detected NPC that stall after synthesis of each RNA length up to the point of escape in the initial round of short RNA synthesis at 37 °C and 0 mM added PPi (Yellow), 1 mM PPi (Blue), 2 mM PPi (Green) and 3 mM PPi (Red). Panel C) Abortive synthesis rates, expressed as RNA synthesized per detected NPC per second, for each short RNA length as a function of [PPi]. Color scheme as in Panel B.

Figure 6B shows the changes in the length distribution of the initial short RNAs synthesized by NPC before stalling as [PPi] is increased from 0 to 3 mM. (Values are listed in Table S2). In the absence of PPi, this length distribution is broad, with a peak at 4- and 5-mer but with significant populations of all RNA lengths up to 10-mer. Increasing [PPi] increases the population fraction of NPC that initially synthesize a short RNA (3-, 4-, and 5-mer) before stalling, and greatly reduces the fraction that make a longer RNA (6-mer to 10-mer). Populations of 8-mer and 10-mer disappear at 2 mM PPi and no RNA longer than 5-mer is synthesized by NPC at 3 mM PPi (Fig. 6B). These trends in the length distribution of short RNAs at 37 °C can be explained qualitatively by reference to the 19 °C rate constants in Fig. 5 and Table S1. At 1 mM and 2 mM PPi, for the NTP concentrations investigated here, step 4, 6, and 7 are predicted to be thermodynamically reversible and steps 9 and 10 are predicted to be somewhat reversible. At 3 mM PPi, steps 4, 6, 7 and 10 are all predicted to be reversible, and steps 5, 9, and 11 are predicted to be somewhat reversible. With so many reversible steps beyond 5-mer, it is not surprising that most NPC synthesize only 3-mer, 4-mer or 5-mer RNA at 3 mM PPi (Fig. 6B, C).

Fig. 6C summarizes abortive initiation rates at 37 °C, determined from manual-mixing experiments extending to 480 s as a function of [PPi] up to 3 mM. These rates are for one abortive cycle, including release of the initial short RNA, regeneration of the original OC without dissociation of RNAP from the promoter, and re-synthesis of another short RNA. (Values are listed in Table S3.) Abortive rates at 19 °C from rapid (RQF) mixing experiments extending to 150 s are smaller and less well determined, but trends with [PPi] appear similar to those at 37 °C in Fig 6 C. Amounts of RNA of a given length observed as a function of time are normalized by the total amount of NPC making RNA of any short length (3-mer to 10-mer), and expressed as RNA per total NPC per second.

In the absence of added PPi, each cycle of abortive initiation is a much slower process than productive initiation, which occurs on a time scale of a few seconds for the conditions investigated (Fig. 2D; [12]). Averaged for all short RNAs detected (3-mer to 10-mer) at the low U condition investigated previously, the time required for abortive synthesis by a nonproductive complex at 37 C is ~80 s [14]. The results in Fig. 6C for the high U condition investigated here are very similar (~80 s).

Fig. 6C shows that, without added PPi, the abortive synthesis rate is largest for 7-mer RNA. The rank order of abortive synthesis rates is 7-mer > 4-mer > 5-mer> 6-mer ≈ 3-mer > 9-mer ≈ 8-mer > 10-mer. This rank order, obtained for a “high UTP” condition (200 μM ATP, UTP; 10 μM GTP), differs from the order obtained previously at a different NTP condition (“low UTP”: 200 μM ATP, GTP; 10 μM UTP): 3-mer > 7-mer > 6-mer ≈ 4-mer > 5-mer > 8-mer > 10-mer > 9-mer), and shows that multiple factors (including but not limited to NTP concentration and RNA length) determine the rank order of abortive rates. Addition of 1 mM PPi has modest effects on rates of abortive synthesis of most RNAs (Fig. 6C), except for 7-mer (large reduction) and 3-mer (moderate increase). The average time required for a round of abortive synthesis is ~130 s as compared to ~80 s for 0 mM added PPi.

Much larger effects on rates of appearance and amounts of short RNAs per NPC are observed at higher [PPi]. Fig. 6C shows that abortive rates of longer RNAs (6-mer to 10-mer) are greatly reduced at 2 mM PPi and become too small to measure at 3 mM PPi. Concurrently, abortive rates of 3-mer and 4-mer increase greatly. At least in part because of this shift to shorter RNAs, which are more easily released and where the conformational changes to re-form the initiation complex are presumably smaller, the time required for a round of abortive synthesis, averaged for all RNA lengths, is greatly reduced from 130 s at 1 mM PPi to ~30 s at 2 mM PPi and ~20 s at 3 mM PPi. An alternative normalization of abortive rates for each RNA length (Table S4), using the initial amount of the NPC of that length instead of using the total amount of NPC as in Fig. 6C, shows similar trends for each RNA length with increasing [PPi].

A minority of OC (variable between experiments but typically 20–30%) are not detected as either productive or nonproductive in both low-UTP [14] and low-GTP initiation assays at all PPi concentrations investigated. These OC may fail to initiate at all or may stall during or after synthesis of the initial pppApU dinucleotide. Any pppApU synthesized in the low GTP assays reported here is unlabeled, and even when labeled in experiments at low UTP is difficult to detect [12, 15]. The fraction of undetected complexes does not exhibit a significant trend with PPi concentration.

Discussion

[PPi] Regulates Many Aspects of Initiation by Productive and Nonproductive Complexes

With increasing [PPi], all steps of initiation become more reversible, reducing the net forward rate of RNA synthesis per RNAP-promoter complex engaged in productive or nonproductive (abortive) initiation. From the rate constants reported here, the effects of any choice of PPi and NTP concentrations (below their K_m) on the rate of productive initiation at the λP_R promoter and ITR investigated here are readily calculated. Additionally, PPi facilitates stalling of RNAP prior to promoter escape, converting what were productive complexes without added PPi to nonproductive complexes that are incapable of promoter escape and the transition to elongation. Hence an increase in [PPi] in the physiological range reduces the total amount as well as the rate of long RNA synthesis, while increasing amounts and rates of short (abortive) RNA synthesis. An increase in PPi also shifts the length distribution of the short RNA made initially and in following abortive cycles by nonproductive complexes, reducing the fraction of nonproductive complexes that make 6-mer or longer RNA and increasing the fraction that make 5-mer or shorter RNAs.

In addition to PPi concentration, which is highly variable in vivo (as discussed below), factors investigated in this research and found to be important for regulation of initiation by PPi include RNA sequence, because pyrimidine steps are reversed at faster rates than purine steps, and translocation stress, which determines the bias in initiation for the pre-translocated state as a result of translocation stress including the coupling of translocation to disruption of interactions between RNAP and the upstream regions of promoter DNA. The bias toward the post-translocated state in elongation reduces the pyrophosphorolysis rate because only pre-translocated complexes can undergo pyrophosphorolysis. Unlike NTP binding, which occurs only to the post-translocated state, PPi appears to bind with similar affinity to pre- and post-translocated states so the extent of translocation is not a function of PPi concentration [30, 38]. Hence PPi binding does not preferentially stabilize ${\text{IC}}_{\text{i}+1}^{\text{pre-tr }}$ relative to ${\text{IC}}_{\text{i}+1}^{\text{post-tr }}$.

Both productive and abortive initiation from other promoters should be affected by [PPi] in similar ways to those described here for the λP_R promoter. These [PPi] effects will certainly depend on the ITR sequence and probably also on the promoter sequence that determines the stability of the open complex. Translocation stresses in initiation are presumably larger for stable OC like λP_R than for unstable OC like those formed at ribosomal promoters [14], and this difference is predicted to affect both the escape point [14] and the bias in each step for the pretranslocated state [12, 15]. Consideration of these factors leads to the prediction that initiation from promoters with unstable OC should be less sensitive to PPi than initiation from promoters with stable OC like λP_R.

Pyrophosphorolysis Rates in Initiation Vary Greatly with RNA 3’-end Sequence

Second-order pyrophosphorolysis rate constants ${\text{K}}_{\text{i}}^{-}$ of initiation and post-escape steps of long RNA synthesis span a wide range from < 10² M⁻¹s⁻¹ to 6 × 10⁴ M⁻¹ s⁻¹ (Fig. 5; Table S1). These rate constants correlate well with the identity of the 3’ RNA base. All three large and one moderate pyrophosphorolysis rate constants are for steps where the RNA initially ends in 3’U and UTP is the product. The U-excision steps with large pyrophosphorolysis rate constants are distributed between early initiation (step 4), late initiation (step 7) and post-escape (step 15). All five steps with negligibly-small pyrophosphorolysis rate constants and five of the steps with moderate pyrophosphorolysis rate constants are steps where the 3’ end base is a purine and ATP or GTP is the product. These steps also are distributed throughout the ITR: steps 6, 9, 10, 12, 13, and 16 have moderate ${\text{K}}_{\text{i}}^{-}$ values and steps 3, 5, 8, 11, and 14 have relatively small ${\text{K}}_{\text{i}}^{-}$ values. Aside from the strong correlation between rate constant and the 3’ end sequence of the RNA, we have found no other correlation of rate constant with position in the ITR. We deduce that the lack of a detectable dependence of ${\text{K}}_{\text{i}}^{-}$ values for the same 3’ RNA dinucleotide sequence on ITR position results from the strong pre-translocation bias present in all steps of initiation, as discussed quantitatively below.

In addition to the large differences in pyrophosphorolysis rate constants for pyrimidine vs. purine bases, second nucleotide effects are also present. The smallest rate constant for a 3’-U pyrophosphorolysis step and two of the smallest rate constants for a 3’-A pyrophosphorolysis step occur where the next base in the pyrophosphorolysis direction is U. All the larger rate constants for purine pyrophosphorolysis steps occur where the next base in the pyrophosphorolysis direction is another purine, regardless of the position within in ITR.

The dependence of pyrophosphorolysis rate constants on the identity of the 3’ RNA nucleotides matches what has been reported for elongation. In elongation, transcripts ending in U are the most PPi sensitive, followed by C, A, and G [30, 38]. Furthermore, the penultimate 3’ nucleotide has also been shown to influence PPi sensitivity, with roughly the opposite effect as the final 3’ nucleotide (e.g. a penultimate U making the RNA strand overall less PPi sensitive). These penultimate-nucleotide effects are also present in pyrophosphorolysis rate constants of initiation steps. The steps that are predicted to undergo slow but significant pyrophosphorolysis, excluding step 16, all have the 3’ sequence AG, AA, or GA. All steps predicted to undergo negligible pyrophosphorolysis have the 3’ sequence UG, UA, or GG. Step 16, the only step with a 3’ U that is not predicted to undergo fast pyrophosphorolysis, is the only step with the sequence UU. These sequence effects closely correspond to those reported for elongation.

This correlation of pyrophosphorolysis rate constants with the identity of the 3’ RNA base contrasts with the situation for RNA synthesis (forward) rate constants (Fig. 5A and Table S1), which as reported previously [12, 15] show no significant correlation with the identity of the nucleotide incorporated. Forward rate constants for pre-escape steps instead exhibit a pattern with three cycles of large and small values (3-mer to 5-mer, 6-mer to 9-mer, 10-mer to 11-mer). We previously interpreted these small forward rate constants in terms of difficulties in translocation resulting from coupling of translocation to sequential disruption of RNAP-promoter contacts with the discriminator strands, with the −10 region and with the −35 region [12, 15]. We also found that the corresponding step-by-step forward rate constants for initiation by T7 RNAP [44] exhibit a clear pattern with ITR position and proposed a structural interpretation in terms of disruption of RNAP-promoter contacts [15].

The translocation step is readily reversible and for yeast pol II (the only RNAP for which translocation rate constants have been determined [25]) is likely to be in rapid equilibrium on the time scale of NTP binding. Rate constants for NTP binding by this or other RNAP have not been determined, but predictions from molecular dynamics simulations [45, 46] are ≤ 1 μM⁻¹ s⁻¹, similar to the average value obtained for NTP binding to F1 ATPase [47]. For steps of initiation with 200 μM NTP, this is a pseudofirst order NTP binding rate constant of 200 s⁻¹, sufficiently less than the reverse translocation rate constant (680 s⁻¹) that it is a good approximation to treat the translocation step as equilibrated on the time scale of NTP binding. For steps with 10 μM NTP, the translocation step is fully equilibrated on the time scale of NTP binding.

Given rapid equilibrium of translocatipn steps, the contribution of translocation to each ${\text{K}}_{\text{i}}^{+}$ value is determined by the translocation equilibrium constant ${\text{K}}_{\text{i}}^{\text{tr}}$, defined as the ratio of concentrations of post- and pre-translocated states [12, 15]. Each ${\text{K}}_{\text{i}}^{+}$ value contains a factor ${\text{K}}_{\text{i}}^{\text{tr}}/\left(1+{\text{K}}_{\text{i}}^{\text{tr}}\right)$, interpreted as the fraction of complexes in the post-translocated state which binds the NTP. Likewise each ${\text{K}}_{\text{i}}^{-}$ value contains the factor $1/\left(1+{\text{K}}_{\text{i}}^{\text{tr}}\right)$, interpreted as the fraction of complexes in the pre-translocated state, to which PPi must bind before pyrophosphorolysis. We previously deduced that ${\text{K}}_{\text{i}}^{\text{tr}}\ll 1$ for all initiation steps, varying by more than 10-fold for different steps. For this situation, applicable in initiation but not in elongation, ${\text{K}}_{\text{i}}^{+}$ is proportional to ${\text{K}}_{\text{i}}^{\text{tr}}$ while ${\text{K}}_{\text{i}}^{-}$ is independent of ${\text{K}}_{\text{i}}^{\text{tr}}$, so translocation stresses in initiation affect only ${\text{K}}_{\text{i}}^{+}$, not ${\text{K}}_{\text{i}}^{-}$.

Pyrophosphorolysis Rates in Initiation Greatly Exceed those for the Same Pair of 3’ Nucleotides in Elongation, Amplifying Differences in Rate for Different Sequences

Pyrophosphorolysis rate constants (Table S1) for most if not all initiation steps greatly exceed those estimated from the published E. coli elongation results [38]. A numerical comparision can be made for GU and UG 3’ dinucleotide sequences. For excision of 3’ U when the adjacent base is G, the ratio of pyrophosphorolysis rate constants for initiation and elongation is ≥ 10⁴, while for excision of G with a neighboring U it is ≥10². For these GU vs UG 3’ dinucleotide sequences, pyrophosphorolysis ${\text{K}}_{\text{i}}^{-}$ values differ by 6-fold in elongation but by >10² fold in initiation (Table S1). Hence initiation appears to amplify the effects of the 3’ nucleotides observed in elongation. In elongation, the difference in pyrophosphorolysis rates for GU vs UG is thought to arise from differences in interactions of 3’ U and G with the folded trigger loop (TL) in the pre-translocated state, with 3’ U – TL interactions being more favorable than 3’ G – TL interactions. If and how this preference might be amplified in initiation is unknown.

[PPi] Regulates the Amount and Rate of Short (Abortive) RNA Synthesis by Nonproductive Complexes (NPC)

Evidence has been presented for two distinct sub-classes of initiation complexes: productive and nonproductive [14, 42, 48]. Productive complexes initiate without aborting, carrying out processive RNA synthesis which culminates in escape of RNAP from the promoter and conversion of the initiation complex to an elongation complex. NPC, on the other hand, stall before escape, presumably as the result of a translocation defect, slowly release the short RNA they synthesized and start over, making another short RNA in an abortive cycle. Abortive transcripts may have regulatory roles in vivo. The shortest abortive RNA have been reported to prime initiation, in some cases changing the transcription start site (TSS) [49, 50]. The molecular differences between productive and nonproductive initiation complexes remain unclear.

Addition of PPi makes complexes that were productive at 0 mM added PPi behave as nonproductive, stalling after synthesis of an initial short RNA, slowly releasing that RNA and re-forming the initiation complex to carry out another round of short RNA synthesis (abortive initiation). NPC presumably are defective in translocation [15]. [PPi] acts like an additional translocation defect by increasing the rate of the back reaction, reducing translocation especially at 3’ dinucleotide sequences like GU in the ITR investigated here, and thereby converting productive complexes to nonproductive.

Predicted Consequences of Large Changes in E. coli Cytoplasmic [PPi] with Growth Conditions on Gene Expression

The [PPi] in the E. coli cytoplasm increases greatly from stationary phase (low μM) to exponential growth (~1 mM), and also varies with growth conditions [1–3]. In stationary phase where the gene expression rate is very small, eliminating most production of PPi from nucleic acid synthesis and tRNA aminoacylation, [PPi]_stationary is in the low μM range [51], comparable to K_m of the E. coli pyrophosphatase [51]. In exponential growth, on the other hand, [PPi]_exponential is in the low mM range [2]. This thousand-fold increase in [PPi] appears to be the direct result of gene expression and protein synthesis, which are capable of making PPi at a sufficiently rapid rate to overwhelm the ability of the constitutive pyrophosphatase to degrade it, resulting in a cytoplasmic [PPi] that increases greatly in the transition from stationary phase to exponential growth (i.e. with increasing growth rate).

The predicted large effects of mM PPi concentrations in vivo on tRNA aminoacylation and therefore on protein synthesis led to the proposal that much of this anionic PPi is bound in the cytoplasm [52]. A possible analogy for this would be Mg²⁺, where the free concentration is much less than the total concentration because of strong binding of Mg²⁺ to polyanionic ribosomal RNA phosphates. No comparable polycation exists to bind anionic PPi. If PPi were bound to individual anion binding sites of proteins these sites would have to be present on a significant fraction of all proteins, because the total cytoplasmic protein concentration is only ~5 mM. No evidence exists for such widespread PPi binding sites on E. coli proteins, so it is likely that the total and free [PPi] are similar and are in the mM range in exponential growth, where PPi effects on initiation from the λP_R promoter are large. In vivo ³¹P NMR experiments may be capable of determining free and total [PPi] in the E. coli cytoplasm [53].

Additional research is also needed to determine if initiation from promoters with less stable OC than λP_R (e.g. T7A1 [14] and rrnBP1 [54] is less affected by [PPi] than λP_R. For promoters with unstable OC, steps with translocation are predicted to exhibit a smaller bias toward the pretranslocated state. This would result in smaller pyrophosphorolysis rate constants for each 3’ end RNA base sequence, more comparable to pyrophosphorolysis rate constants in elongation.

Conclusions

In this research we quantified the very large effects of changes in [PPi] in the physiological range (up to low mM) on productive and abortive transcription initiation by E. coli RNAP at the λP_R promoter. This information was not previously available for any RNAP or promoter DNA. We found that, as in elongation, uridine incorporation and some other initiation steps become significantly reversible at physiological PPi and NTP concentrations. We also found that both the fraction of RNAP-promoter complexes that productively initiate and the rate of RNA synthesis per productive complex decrease with increasing [PPi]. Concomitantly the fraction of complexes that abortively initiate increases and the abortive product distribution shifts shorter RNAs. Pyrophosphorolysis rates for some initiation complexes are orders of magnitude larger than for removal of the same nucleotide from elongation complexes because of the strong bias toward the pre-translocated state in initiation, and exhibit even stronger dependences on nucleotide identity (pyrimidine ≫ purine). Initiation from promoters with unstable open complexes, which are predicted to exhibit less of a bias for the pre-translocated state and which may transition from initiation to elongation at a shorter RNA-DNA hybrid length [14], should be less affected by [PPi] and should behave more like elongation complexes in this regard. Given that cytoplasmic [PPi] is much higher in exponential-phase than stationary-phase E. coli and varies with growth conditions, these large [PPi] effects must be physiologically-relevant.

Methods

Reagents, Buffers, and Gels

Reagents for buffers and stock solutions were the highest available grade and were used as received. All solutions were prepared using 18 MΩ deionized water from a Barnstead EPure system. NTPs and dNTPs (Boston Bioproducts, Thermo Fisher, New England Biolabs) used in transcription assays and PCR reactions were 99% pure and used as received. Enzymes for PCR reactions were purchased from NEB and used according to the manufacturer’s protocols.

Storage buffer (SB) for core RNAP, σ⁷⁰ and RNAP holoenzyme is 50% v/v glycerol, 0.01 M Tris (pH 8.0), 0.1 M NaCl, 0.1 mM EDTA, and 0.1 mM DTT, 0.05 mg/ml BSA. Transcription buffer (TB) is 40 mM Tris, 5 mM MgCl₂, 60 mM KCl, 1 mM DTT, and 0.05 mg/mL BSA, adjusted to pH 8.0 at the experimental temperature (19 °C, 37 °C).

5X initiation solution (IS) for manually-mixed transcription assays at 37 °C assayed with α-³²P-GTP is 1 mM ATP, 1 mM UTP, 50 μM unlabeled GTP, 87.5 nM α-³²P-GTP, 0.25 mg/ml heparin in TB, and from 0 to 15 mM PPi (five times the final [PPi] concentration). 2X initiation solution (IS) for rapid quench flow (RQF) transcription assays at 19 °C assayed with α-³²P-GTP is 400 μM ATP, 400 μM UTP, 20 μM unlabeled GTP, 35 nM α-³²P-GTP, 0.1 mg/ml heparin in TB and from 0 to 4 mM PPi (two times the final [PPi]).

Quench Solution (QS) for manual mixing and RQF transcription assays is 8 M urea and 15 mM EDTA in TB. QS with added dyes (QSD) for polyacrylamide gel electrophoresis (PAGE) has 0.05% xylene cyanol and 0.05 bromphenol blue in QS. TBE buffer for PAGE is 90 mM Tris-Borate (pH 8.3) and 2 mM Na₂EDTA. All transcription gels are 20% acrylamide-(bis)acrylamide (19:1), and were made using the UreaGel system (National Diagnostics).

RNA Polymerase and λP_R Promoter DNA

RNAP core enzyme (α₂ββ’ω) is overexpressed from pVS10 plasmid and purified via Ni affinity chromatography [15]. The σ⁷⁰ subunit is overexpressed from pIA586 and purified using Ni affinity chromatography [15]. Holoenzyme is reconstituted using a 1:2 ratio of core to σ⁷⁰. Filter binding activity assays [14] performed on preparations of RNAP holoenzyme used here show that 50% ± 10% of RNAP molecules form a stable open complex with the λP_R promoter at 37 °C and 0 mM added PPi. All RNAP concentrations reported here refer to this active fraction. Open complexes of RNAP and λP_R promoter DNA are formed as previously described [14, 15] by incubating a 2:1 mole ratio of active RNAP and promoter DNA for 1 hour at the temperature of the experiment. For fast-mixing experiments the initial OC concentration was 100 nM; for manual mixing experiments the initial OC concentration was 12.5 nM. Sequences of PCR primers and λP_R promoter DNA fragments used to prepare the 122 bp promoter DNA fragment used in all kinetic studies reported here and previously are given in Table S5.

Initiation Kinetics Assays

Manually-mixed initiation kinetics assays are performed at 37 °C and analyzed as described previously [15]. Assays are designed to obtain single-round RNA synthesis by productive complexes. Briefly, 5X initiation solution (IS) is manually mixed 1:4 with a solution of pre-formed OC at time zero to obtain a final OC concentration of 10 nM. Reactions are quenched with QS and RNA is separated via PAGE.

Rapidly quenched initiation kinetics assays are performed at 19 °C and analyzed as described previously [14]. Assays are designed to obtain single-round RNA synthesis by productive complexes. Briefly, 2X initiation solution (IS) is mixed 1:1 with a solution of pre-formed OC at time zero using a KinTek Corp Rapid Quench Flow (RQF) apparatus to obtain a final OC concentration of 50 nM. Reactions are quenched with QS and RNA is separated via PAGE.

Transcription gels are transferred to a phosphorimaging cassette and analyzed using a Typhoon 9000 phosphorimager (18 h exposure). Peak area is converted to moles of observed product as described previously [12]. To obtain good labelling efficiency without the use of high concentrations of labelled NTP that cause background problems and limit the ability to detect small RNAs on the gels, we use a low concentration of α-³²P-GTP (17.5 nM) and a reduced concentration (10 μM) of the corresponding unlabeled NTP. The probability of incorporation of the radiolabeled NTP at each position is determined by the ITR sequence, and was accounted for using incorporation probabilities [12]. Quantitative results reported here are from the average of three RQF or manually mixed experiments at each set of NTP concentrations investigated, and the uncertainties reported are one standard deviation from the mean.

Quantifying Synthesis of Short RNAs by Nonproductive Complexes in 37 °C Manual Mixing Experiments

Nonproductive complexes (NPC) are defined as those OC that stall after synthesizing a short (pre-escape length) RNA in the initial phase of the kinetics [12, 14, 15]. At 37 °C many of the NPC release their RNA and reinitiate to begin a cycle of abortive synthesis. As previously described [14], amounts of NPC that stall at each RNA length after synthesis of an initial short RNA are determined in manual-mixing experiments from the intercepts of plots of the linear increase in the amount of that short RNA vs time from abortive synthesis in the time range after long RNA synthesis by productive complexes is complete.

Quantifying Synthesis of Short RNAs by Nonproductive Complexes in 19 °C Fast Mixing Experiments

Without added PPi, at 19 °C, the kinetics of synthesis of the first short RNA of each length by NPC are well-described as a first order (single-exponential) approach to a plateau value [12, 15]. These fitted amounts were subtracted from the total amount of RNA of a given length present at each time to determine transient amounts of that RNA length in initiation by productive complexes [15]. Addition of PPi broadens transient RNA peaks from initiation by productive complexes and reduces RNA synthesis rates, introducing ambiguity in the above approach. As an alternative, the full time course of initial and subsequent rounds of RNA synthesis by NPC at 19 °C was modeled in Kintek Explorer [55] as reiterative synthesis of each RNA length by a subpopulation of NPC, according to the simple mechanism NPC_i → NPC_iRNA_i → NPC_i + RNA_i where the first step is [PPi]-dependent RNA synthesis and the second, rate-determining step is RNA release and restoration of the initiating NPC, assumed to be [PPi]-independent. RNA synthesis and release rate constants were adjusted manually to best replicate the observed NPC behavior. Results of this analysis are shown as red curves in SI Figs. S5–S6. The breakover point between the initial exponential phase and the subsequent linear abortive phase was used to determine the initial amount of RNA synthesized by NPC and hence the amount of NPC synthesizing that RNA length for use in Fig. 6A.

Determination of Rate Constants for Forward (post-escape) Steps of RNA Synthesis

To determine forward rate constants ${\text{K}}_{\text{i}}^{+}$ for synthesis of 12-mer to 16-mer for Fig. 3, previously published 0 mM PPi RQF experiments [15] at final NTP concentrations of 200 μM ATP and GTP, 10 μM UTP, and 17.5 nM α-³²P-UTP were analyzed together with the data presented here to extend the transient analysis to 16-mer, instead of 11-mer. In conjunction with the 0 mM PPi experiments shown here, the forward rate constants for steps 11–16 were determined by fitting the transient RNA produced in the two NTP concentrations studied to the mechanism in Fig. S2, an extension of the mechanism reported reported previously [12, 15]. Because these are post-escape species, no correction for synthesis by NPC is required for this analysis.

Determination of Rate Constants for Individual Steps of Pyrophosphorolysis (for 3-mer to 16-mer)

Second order pyrophosphorolysis rate constants at each [PPi] were determined at 19 °C using Kintek Explorer [55] to fit all lengths of RNA transients in 16-mer RNA synthesis to the mechanism in SI Fig. S2. These transients (shown in SI Figs. S3 and S5–S6) were determined by subtracting the predicted amounts of RNA synthesized by NPC from the total amount of RNA at each time. Forward rate constants, determined at 0 mM PPi as previously described [15] were held constant in these fits.