Characterization of a transitionally occupied state and thermal unfolding of domain 1.1 of σA factor of RNA polymerase from Bacillus subtilis

σ factors are essential parts of bacterial RNA polymerase (RNAP) as they allow to recognize promotor sequences and initiate transcription. Domain 1.1 of vegetative σ factors occupies the primary channel of RNAP and also prevents binding of the σ factor to promoter DNA alone. Here, we show that domain 1.1 of Bacillus subtilis σA exists in more structurally distinct variants in dynamic equilibrium. The major conformation at room temperature is represented by a previously reported well‐folded structure solved by nuclear magnetic resonance (NMR), but 4% of the protein molecules are present in a less thermodynamically favorable state. We show that this population increases with temperature and we predict its significant elevation at higher but still biologically relevant temperatures. We characterized the minor state of the domain 1.1 using specialized methods of NMR. We found that, in contrast to the major state, the detected minor state is partially unfolded. Its propensity to form secondary structure elements is especially decreased for the first and third α helices, while the second α helix and β strand close to the C‐terminus are more stable. We also analyzed thermal unfolding of the domain 1.1 and performed functional experiments with full length σA and its shortened version lacking domain 1.1 ( σA_Δ1.1 ). The results revealed that while full length σA increases transcription activity of RNAP with increasing temperature, transcription with σA_Δ1.1 remains constant. In summary, this study reveals conformational dynamics of domain 1.1 and provides a basis for studies of its interaction with RNAP and effects on transcription regulation.

The RNAP core alone is able to elongate the transcription, but it is not capable of its initiation without a σ factor. The σ factors are essential for recognition of the promoter sequence, subsequent binding of RNAP to a promoter DNA, and beginning of the transcription process. 2 The recognized crucial role of σ subunits in the transcription process was used to develop new antibacterial drugs. 3 The numbers of different σ factors are different in various species. There are species with only a single σ factor, but also with more than 100 different σ factors. 4 The σ factors are divided according to their structure into groups σ 70 and σ 54 . There are no sequential similarities between these two groups and there is also another significant difference between these two families. The factors from the σ 54 family require binding ATP activators 5  bound to free RNAP occupies the DNA binding channel. 6,7 The structure of domain 1.1 from B. subtilis consists of three α helices forming a hydrophobic core and of two short β strands arranged in a parallel β sheet. 8 The secondary structure composition is similar to previously studied domain 1.1 from Thermotoga maritima. 9 However, the structures of these two domains differ despite the sequence similarities. The first helix in the sequences has a significantly different orientation in these two structures. Surprisingly, the arrangement of the helices forming the hydrophobic core of the domain 1.1 from B. subtilis is similar to domain 1.1 from Escherichia coli 6 solved by X-ray crystallography in complex with RNAP and to the structured domain of RNAP δ subunit from B. subtilis. 10 The structure of domain 1.1 from B. subtilis was shown to be affected by dynamics at the μs-ms timescale, typical for larger structure rearrangement. 8 It was hypothesized 8 that the determined structure of the domain 1.1 from B. subtilis is in an exchange with a structure similar to domain 1.1 from T. maritima. Therefore, we decided to obtain detailed information about the low populated state of the B. subtilis with atomistic resolution. The results presented here then reveal details of the dynamic equilibrium between the two states, its dependence on temperature, and biological implications.

conformational exchange
Our first goal was to determine the quantitative parameters of the previously reported exchange in the backbone of the B. subtilis σ 1:1 domain. 8 We analyzed data provided by NMR experiments based on the Carr, Purcell, Meiboom and Gill pulse sequence (CPMG experiments). 11,12 Using the CPMG approach, we measured how exchange between conformational states contributes to the relaxation of the signal corresponding to the magnetization of 15 N in the protein backbone. The experiments were performed at five temperatures ranging from 10 C to 30 C. The exchange contribution to the relaxation rate at higher temperatures resulted in a significant attenuation of the NMR signal, preventing a detailed analysis of the CPMG data. At 25 C, we detected exchange increasing the relaxation rates by at least 2:5 s À1 for 47 out of 71 analyzed amide 15 N signals. Results are summarized in Table S1. The simplest two state model of the exchange reproduced the data well. The results of the analysis of the CPMG data of individual residues show similar values of kinetic and thermodynamic parameters suggesting that they report the same exchange event. In order to test this hypothesis, we tried to fit the available data of residues exhibiting the significant exchange together to obtain a single value of the exchange contribution k ex and of the population of the minor state p B for all residues at each temperature ( Table 1). The population of the minor state ranged from approximately 8% at 30 C to less than 1.0% at the lowest temperature. We should note that an increased χ 2 parameter was also observed at 30 C and in addition, we detected two residues (A35 and F54) which cannot be included in the global fit at 30 C, in contrast to the lower temperatures. It indicates that the dynamics is becoming more complicated and the application of two-state model may not be applicable at higher temperatures. Such a trend is expected because higher temperatures usually enhance population of additional states which can be safely neglected at lower temperatures.
However, the significant drop of quality of NMR spectra at higher temperatures did not allow us to study the dynamics beyond the twostate model.
Despite the mentioned limitations, the determined populations follow the Boltzmann's law at the temperatures 10-25 C. We T A B L E 1 Comparison of the fitted global exchange parameters at different temperatures to the dispersion profiles, the error represents the 99% confidence level estimated from Monte-Carlo simulations.

| Structural analysis of the minor state
The NMR structure determination of proteins is most typically based on inter-atomic proton-proton distances estimated from measured nuclear Overhauser effect (NOE). 13 As supplementary structural information, chemical shifts of backbone nuclei and occasionally scalar couplings or residual dipolar couplings (RDCs) 14 The dependence of equilibrium constant K a (A), forward k AB (B), and backward k BA (C) rate constants on temperature.
these residues are in a much different local environment in the minor vs. major state.
Positions of the methyl groups are also depicted in Figure 2 and color-coded according to jΔωj. In proteins, methyl groups are sensitive indicators of structure and dynamics 18 and they often report on events within the hydrophobic core. The CPMG experiments measured with σ 1:1 samples including stereospecifically labeled methyl groups provided us a complete set of 13 C methyl chemical shifts of the minor state. The most significant changes were identified for 13 C γ -proS of V15, 13 C δ -proS of L19, both methyls in I34, and 13 C δ -proR of L55. All these methyls are located in a proximity of aromatic rings in the major state ( 13 C γ -proS of V15 is close to F41, other mentioned methyls are in the proximity of Y51). It can be expected that the significant disturbance of their chemical shifts is induced by a change of the distance and orientation to the aromatic rings, known for a strong effect on the chemical shifts. 19 Generally, the results indicate a larger structural rearrangement affecting the hydrophobic core of the σ 1:1 structure.

| Exchange rates of amide protons
In addition to the characterization of the minor state we performed an experiment to monitor a proton-deuterium exchange of backbone amides of the major state at 1 C after a quick replacement of the protonated buffer for its deuterated equivalent. Decays of amide signal intensities in time were converted to protection factors (available in Figure S9). The exchange rate of residues outside the secondary structure elements were too fast to be detected by our approach. The protection factors of residues within the secondary structure elements were in the order of 10 2 -10 3 while the protection factors of structured proteins are expected to be in the orders 10 6 -10 9 . 25  2.5 | Thermal unfolding of the domain 1.1 As the NMR data revealed a presence of less ordered/disordered state(s), we complemented our study by differential scanning calorimetry (DSC) and circular dichroism spectroscopy (CD) measured at multiple temperatures which allowed us to study thermal unfolding of σ 1:1 . The DSC data were interpreted with a simple two-state model and a more sophisticated sequential Zimm-Bragg model. 26 The agreement of the calorimetry data with the two state model was poor unless the van't Hoff enthalphy is fitted separately from the overall heat (data not shown). The Zimm-Bragg model provided a better, but not perfect agreement with the DSC data (results shown in Figure S10). The optimized Zimm-Bragg model was further crossvalidated using CD data 27  where a disturbance of the hydrophobic core of the protein resulted in significant phenotypic changes compared to the wild-type.
Although the mutated residue I48 is weakly conserved in primary σ factors, it is often replaced with leucine or valine with similar biophysical properties. In B. subtilis this position corresponds to L55, which is located in the C-terminal α helix and its sidechain was shown in our study to be highly affected by the conformational exchange ( Figure 2 and Table S3). A parallel can be found also with the ω subunit of RNAP, where its flexibility is essential for its function, and even silent mutations (mutations that do not change amino acids but codons; subsequently, due to differential availability of aminoacylated tRNAs, the protein is folded differently) that reduce this flexibility compromise its interplay with RNAP and its biological function. 33 The higher tendency of the minor state to form α helix 2 and the β-sheet suggests that these secondary structure elements represent a core of the σ 1:1 structure. Interestingly, the less ordered helices 1 and 3 interact in E. coli RNAP holoenzyme with an α helix in a linker between domains 1.1. and 1.2, and with the Gp2 inhibitor produced by the bacteriophage T7. 6 Helices 1 and 3 are oriented towards to the β 0 clamp whose motion was proposed to eject σ 1:1 from the RNAP cleft. 28

| Conclusions
We characterized a previously detected low populated state of the T. maritima appears to be incorrect. Instead, the studied minor state was identified to be more flexible than the major state and it has a lower propensity to form a structured conformation, especially for helices 1 and 3. We have shown that the disorder induced by the elevated temperature increases the transcriptional output. We hypothesize that the conformational plasticity of σ 1:1 plays a role in binding and ejection of the domain 1.1 from the binding channel of RNAP.

| Sample preparation
Cloning procedure of gene encoding σ 1:1 was described elsewhere. 8 Expression and purification of samples are described in Appendix S1.
A purity and stability of the samples was verified prior to every NMR measurement. A special attention was paid to avoid a contamination of glycerol which increases viscosity of the solvent and affects the measured relaxation rates. Four samples of wild-type σ 1:1 were prepared in this study, each differing in the isotopic labeling scheme, while the buffer composition was the same, the samples contained 20 mM sodium phosphate buffer, 10 mM NaCl, 3 mM NaN 3 , and the pH was 6.6. The first sample of 1 mM concentration was uniformly labeled with 15  F I G U R E 5 Functional characterization of σ 1:1 in dependence on temperature. Multiple round transcription were performed with RNAP reconstituted with σ A or σ A_Δ1:1 . Representative primary data are shown above the graph, the bands are radiolabeled transcripts resolved on a polyacrylamide gel. The graph shows averages of three independent experiments AE SD. To facilitate comparison of the differential effects, transcription activity at 30 C for both σ variants was set as 1.

| Differential scanning calorimetry
The DSC thermograms of σ 1:1 were collected on a Microcal PEAQ-DSC Automated (Malvern) instrument. The concentration of protein was 0.75 mg/mL in a phosphate buffer identical to the NMR buffer (20 mM NaPi, 10 mM NaCl, 3 mM NaN 3 , pH 6.6). Triplicate measurement was performed. Both buffer and sample solutions were degassed before measurements. The following measurement parameters were used: the temperature scan range was 4-85 C, scan heat rate was 60 C=h, pre-scan thermostat was set to 5 min, post-scan thermostat was 0 min, feedback mode was set to high. Evaluation of the DSC data was performed using MicroCal PEAQ-DSC software (Malvern). were recorded in the range of 5-90 C in 5 C temperature steps. Each spectrum was recorded with total of three scans.

| Data analysis
Data acquired from NMR measurements were converted and processed using the software NMRPipe, 50 non-uniformly sampled data were processed using NMRPipe, version 9.9, and SMILE 2.0beta. No extrapolation was used in the processing of the non-uniformly sampled data and identical signal downscaling factor was used for independently processed spectra in relaxation or CEST series. The analysis and visualization of spectra were done in the software NMRFAM-Sparky. 51 In CPMG data sets signal intensities obtained from 2D spectra measured in relaxation series were converted into the effective relaxation rates R 2,eff using the Octave 3.8.2 program 52 employing the function leasqr from the package optim, by fitting peak intensities to a mono-exponential decay using a nonlinear least-squares approach: where ν CPMG = 1=4τ, with 2τ being the interval between consecutive refocusing pulses of the CPMG sequence that is applied during a con- Uncertainties in the exchange parameters were established by 5000 Monte Carlo simulations. The dependence of the population of the minor state p B on temperature was used to obtain the change of enthalpy ΔH and entropy ΔS upon transition to the minor state following the Boltzmann's law: where R is the molar gas constant, T is the absolute temperature, and the fraction p B = 1 À p B ð Þrepresents the equilibrium constant K a .
The change of enthalpy and entropy for a transition from the ground state (ΔH AB and ΔS AB , respectively) or from the minor state (ΔH BA and ΔS BA , respectively) to a transition saddle point was determined following the Eyring equation: where k B is the Boltzmann's constant, h is the Planck constant, and T is the thermodynamic temperature. Standard deviations of thermodynamic parameters were estimated using the smooth Bootstrap method.
In CEST data sets, the peak intensities were collected from 2D spectra in B 1 position series. Uncertainties in intensities were estimated in the same way as mentioned above. CEST profiles were generated and exchange parameters were extracted using the software ChemEx 43 (http://www.github.com/gbouvignies/chemex). Global/ per-residue fitting as well as uncertainty estimation was done as described for the CPMG data set.
The chemical shift perturbation was calculated as where n was the number of available chemical shift disturbance for backbone nuclei in each residue and the parameters α i were 0.32, 0.19, 0.12, and 1.00 for 13 CO, 13

Reconstitutions were carried out in buffer
The screens were scanned with Typhoon 5 (Cytiva). The amounts of the 145-nucleotide-long transcripts (originating from the cloned promoter) were quantified with ImageQuantTL software (Cytiva). All calculations and data fitting were done using SigmaPlot from Jandel Scientific.

CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest.

PEER REVIEW
The peer review history for this article is available at https://www. webofscience.com/api/gateway/wos/peer-review/10.1002/prot.

DATA AVAILABILITY STATEMENT
The additional data that support the findings of this study are available from the corresponding author upon reasonable request.