Characterizing highly dynamic conformational states: The transcription bubble in RNAP-promoter open complex as an example

Abstract

Bio-macromolecules carry out complicated functions through structural changes. To understand their mechanism of action, the structure of each step has to be characterized. While classical structural biology techniques allow the characterization of a few "structural snapshots" along the enzymatic cycle (usually of stable conformations), they do not cover all (and often fast interconverting) structures in the ensemble, where each may play an important functional role. Recently, several groups have demonstrated that structures of different conformations in solution could be solved by measuring multiple distances between different pairs of residues using single-molecule Förster resonance energy transfer (smFRET) and using them as constrains for hybrid/integrative structural modeling. However, this approach is limited in cases where the conformational dynamics is faster than the technique's temporal resolution. In this study, we combine existing tools that elucidate sub-millisecond conformational dynamics together with hybrid/integrative structural modeling to study the conformational states of the transcription bubble in the bacterial RNA polymerase-promoter open complex (RPo). We measured microsecond alternating laser excitation-smFRET of differently labeled lacCONS promoter dsDNA constructs. We used a combination of burst variance analysis, photon-by-photon hidden Markov modeling, and the FRET-restrained positioning and screening approach to identify two conformational states for RPo. The experimentally derived distances of one conformational state match the known crystal structure of bacterial RPo. The experimentally derived distances of the other conformational state have characteristics of a scrunched RPo. These findings support the hypothesis that sub-millisecond dynamics in the transcription bubble are responsible for transcription start site selection.

FIG. 1.

D-A labeled promoter constructs representing two reaction coordinates. Examples of dyes’ attachment points in RPo for two reaction coordinates (pdb code: 4XLN). (a) Scrunching coordinates: dyes attached to bases upstream and downstream relative to the transcription bubble; (b) bubble coordinates: dyes attached to bases within the transcription bubble. The arrows show the general directions in which distance changes are expected in the two reaction coordinates. Dye AVs are also shown as green (D) and red (A) partial spheres.

FIG. 2.

Qualitative assessment of smFRET dynamics using BVA. Proximity ratio (PR) histograms and BVA plots of σ(PR) versus PR show that (a) the free promoter is either characterized by a single PR population with no smFRET dynamics or with no PR population at all, due to quenching of FRET, and (b) RNAP bound to promoter DNA induces a unique PR subpopulation that exhibits smFRET dynamics. Depiction of differently labeled D-A constructs in free promoter (a) and RPo (b) states was generated by calculating D and A AVs (green and red surfaces, respectively) on top of DNA labeling positions in the context of RPo crystal structure (pdb code: 4XLN).

FIG. 3.

Comparison of measured apparent D-A distances, ${\overline{r}}_{D-A,E}$, with expected D-A distances, ${\overline{r}}_{D-A}$. The experimentally derived ${\overline{r}}_{D-A,E}$ values of different conformational states for each D-A labeled lacCONS constructs are compared with the expected D-A distances from RPo crystal structure (pdb code: 4XLN) in free DNA (black dot) and in RPo (red dot). Two conformational states were found (with sub-millisecond underlying interconversion dynamics) in the RNAP-bound fraction. Their derived distance values are shown (magenta open squares) for the subset that defines a conformation closest to the known RPo structure. Distances for the other conformation are also shown (blue open diamonds). The values of the free promoter fraction are also presented (grey triangles; for the D-A labeled promoter constructs that do not exhibit quenched FRET). Quenched FRET constructs had an expected D-A distance below 0.5R₀ (shown as shaded area). The error bars in the measured distances represent experimental standard error (Eq. S21 of the supplementary material). The error bars in the expected distances represent the D-A standard deviation, as calculated from all possible donor and acceptor positions in space (dye AVs).

FIG. 4.

FPS analysis of transcription bubble conformational states. FPS analysis was performed for each conformational state against the known RPo structure (pdb code: 4XLN). (a) One set of apparent mean D-A distances, ${\overline{r}}_{D-A,E}$ (RPo state), permits reconstruction of the template (T; red) and nontemplate (NT; green) strands in a 3D organization that matches the 3D organization of the T (yellow) and NT (cyan) strands in the crystal structure. (b) The second ${\overline{r}}_{D-A,E}$ set (scrunched RPo state) yielded reconstruction of the template (T; red) and nontemplate (NT; green) strands in a 3D organization that does not match the 3D organization of the T (yellow) and NT (cyan) strands in the crystal structure. Solid spheres represent the mean positions of the D (green) and A (red) dyes. The set of mean D-A distances, ${\overline{r}}_{D-A}$, probed is shown by gray dotted lines connecting D and A mean positions. The mean position is always within the dyes’ AVs (3 examples of dye AVs are shown).