Structural origins of Escherichia coli RNA polymerase open promoter complex stability

Abstract

The first step in gene expression in all organisms requires opening the DNA duplex to expose one strand for templated RNA synthesis. In Escherichia coli, promoter DNA sequence fundamentally determines how fast the RNA polymerase (RNAP) forms "open" complexes (RPo), whether RPo persists for seconds or hours, and how quickly RNAP transitions from initiation to elongation. These rates control promoter strength in vivo, but their structural origins remain largely unknown. Here, we use cryoelectron microscopy to determine the structures of RPo formed de novo at three promoters with widely differing lifetimes at 37 °C: λP_R (t_1/2 ∼10 h), T7A1 (t_1/2 ∼4 min), and a point mutant in λP_R (λP_R-5C) (t_1/2 ∼2 h). Two distinct RPo conformers are populated at λP_R, likely representing productive and unproductive forms of RPo observed in solution studies. We find that changes in the sequence and length of DNA in the transcription bubble just upstream of the start site (+1) globally alter the network of DNA-RNAP interactions, base stacking, and strand order in the single-stranded DNA of the transcription bubble; these differences propagate beyond the bubble to upstream and downstream DNA. After expanding the transcription bubble by one base (T7A1), the nontemplate strand "scrunches" inside the active site cleft; the template strand bulges outside the cleft at the upstream edge of the bubble. The structures illustrate how limited sequence changes trigger global alterations in the transcription bubble that modulate the RPo lifetime and affect the subsequent steps of the transcription cycle.

Keywords: RNA polymerase; cryo-EM; open complex; promoter DNA; transcription.

Fig. 1.

Promoter DNA constructs used for cryo-EM studies and overall cryo-EM structures of RPos. (A) Promoter sequences studied by cryo-EM (nt-strand DNA [Top strand], light gray and t-strand [Bottom strand], dark gray). Numbers above the DNA sequences denote positions with respect to the transcription start site (+1, black arrow). Shaded colors highlight key promoter regions: −35 element (yellow), −10 element (magenta), and the discriminator (pale green; bp, base pair). (Top) λP_R (−60 to +30: −60 to +20 are native λP_R sequences; sequences downstream of +20 originate from the plasmid construct used in extensive kinetic and DNA footprinting studies of λP_R (11)). (Middle) λP_R-5C (−60 to +30) is a single base pair inversion of λP_R G_-5 to C. (Bottom) T7A1 (−66 to +20). While T7A1 and λP_R share the same −35 element sequence, key differences exist in both the −10 element and the sequence and length (seven versus six nucleotides, respectively) of the discriminator. At the upstream end of the constructs used here, the T7A1 promoter has both proximal and distal UP elements (tight-binding αCTD binding sites), and λP_R (and λP_R-5C) has a distal UP (3). (B) Eco RNAP subunits are shown as transparent surfaces [αI-NTD, αII-NTD, and ω: light gray (NTD, N-terminal domain); αCTD: pale green; β: pale cyan; β’: light pink; and σ⁷⁰: light orange], with the active site Mg²⁺ shown as a sphere (pale yellow). Promoter DNA is shown as the cryo-EM difference density (−35 element, −10 element, and discriminator colored as in A). Particle classification revealed two distinct RNAP-DNA complexes populated at λP_R and a single class at T7A1 and at λP_R-5C. The number of particles in each class and nominal resolution are shown. In λP_R class I, good map density allowed all bases in the nt- and t-strands to be modeled. For comparison, disordered bases in the other RPo are shown as spheres positioned approximately at the corresponding position of the phosphate backbone in λP_R class I. Outside of the transcription bubble, duplex DNA was modeled from −45 to +23 in λP_R class I, −37 to +13 in λP_R class II, −37 to +15 in λP_R-5C, and −47 to +15 in T7A1.

Fig. 2.

Comparison of differences in base stacking in the transcription bubble. The schematic illustrates the following: 1) the position of bases in each open complex, with missing bases shown as dashes and 2) base stacking pairs indicated by a star symbol. DNA backbone and bases colored as in Fig. 1. (A) λP_R class I. (B) λP_R class II. (C) λP_R-5C. (D) T7A1.

Fig. 3.

Differences in Eσ⁷⁰ interactions with the nt-strand discriminator region between λP_R class II (A) (six-base discriminator) and T7A1 (B) (seven-base discriminator). (Left) Overall view of RPo (similar to Fig. 1B with the same color scheme and abbreviations), with RNAP subunits shown in the surface representation and DNA as atomic spheres. The boxed area is magnified in the middle. (Middle) Magnified view showing the interactions with the nt-strand from the discriminator to +1. RNAP subunits are shown in the backbone worm (β: pale cyan; β’: light pink; and σ⁷⁰: light orange); the sidechains of atoms within 4.5 Å nucleic acid atoms (λP_R [−6 to +1] or T7A1 [−7 to +1]) are shown as sticks. Atomic distances within 3.5 Å and interactions with the π electrons of the DNA bases (sulfur-π and cation-π) are shown by black dashed lines. The DNA carbon atoms are colored green. (Right) Schematic comparing the network of interactions between the nt-discriminator (and +1) and Eσ⁷⁰. Favorable interactions within 3.5 Å (hydrogen bonds and salt bridges) are shown by heavier lines (see color key) than those within 4.5 Å (polar, ionic, van der Waals, and π [sulfur, cation, and π]). The corresponding cryo-EM density is shown in SI Appendix, Fig. S10. Although the nature and extent of interactions with the nt-strand discriminator differ between λP_R and T7A1, the DNA backbone is largely solvent exposed. Few favorable interactions exist with the sugar atoms (at upstream and downstream ends of the discriminator region), and only one positively charged amino acid falls within 4.5 Å the DNA phosphate backbone of either promoter.

Fig. 4.

Scrunching in the t-strand. (A) λP_R class I RPo flips out T_-11(t) at the ds/ss junction. (B) T7A1 flips out two bases, T_-12(t) and A_-11(t), scrunching the t-strand at the upstream entrance to the active site channel. Residues within 4.5 Å nucleic acid atoms of the t-strand are shown as sticks. Eco Eσ⁷⁰ subunits shown in backbone worm (β: pale cyan; β’: light pink; and σ⁷⁰: light orange). Single-stranded bases from −11 to −9 (A; λP_R) or −12 to −9 (B; T7A1) are shown as sticks (same color coding as Fig. 3) and also transparent atomic spheres (hot pink). The same σ and β residues stabilize the A_-10(t)/T_-9(t) or T_-10(t)/G_-9(t) stacking pairs in λP_R and T7A1, respectively, noted and shown as transparent CPK atoms (σ: orange and β: cyan). The corresponding cryo-EM density is shown in SI Appendix, Fig. S12.

Fig. 5.

λP_R class I t-strand interactions and positioning in the active site. (A, Left) Overall representation of RPo (same color scheme and abbreviations) as in Fig. 3. The boxed area is magnified to the right. (B) Schematic detailing the interactions between the t-strand and residues within 4.5 Å. (C) Comparison of the t-strand in λP_R class I RPo and in RPinit. (Left) Cryo-EM density (blue mesh) defining the modeled position of single-strand bases on the t-strand (−3 to +2), the downstream ss/ds junction at +3 (shown as sticks), and the Mg²⁺ bound in the RPo active site (yellow sphere). The conserved RNAP bridge helix (β’: pink) is shown as a point of reference. Colors are the same as in Fig. 1. (Right) Result of aligning λP_R class I RPo with a high-resolution X-ray structure of RPinit [2.9-Å resolution and PDB 4Q4Z (32)]. Initiating triphosphate ribonucleotides (dark beige) bound at +1 (iNTP = ATP) and +2 (i+1NTP = CMPCPP) pair with the corresponding t-strand bases. The backbone and bases from −3 to +3 in RPinit (beige) largely superpose with the equivalent positions in λP_R, with the exception of G_-1(t) in which a change in base tilt leads to a clash (red circle) with the iNTP (ATP).