Computational protocol: Sequential conformational transitions and α-helical supercoiling regulate a sensor histidine kinase

Similar protocols

Protocol publication

[…] The time-resolved scattering data of recorded at BioCARS and at cSAXS were merged to allow analysis of the entire time range, from ca. 100 ns to ca. 5 s. Since the X-rays at BioCARS are polychromatic, whilst the X-rays at cSAXS are monochromatic the cSAXS curves were convoluted with the BioCARS undulator spectrum according to established methods, , –. As an illustration the same operation was performed on the steady state difference scattering for the isolated YF1-LOV domain (Supplementary Fig. ), showing that this procedure reproduces the polychromatic difference curve with high accuracy. The time dependent scattering data is considered to be a linear combination of time independent basis spectra according to Eq. ().1qΔI(q,t)=BS(q)⋅C(t)Where ΔI(q,t) is the experimentally recorded data, BS(q) is the time independent basis spectra and C(t) is the concentration/contribution of the respective basis spectrum in the experimental data. To remove high frequency noise prior to spectral decomposition the experimental data was filtered by approximating it by a 40 degree polynomial, closer to the actual information content in the data. The concentration of the species were assumed to follow sequential mono exponential kinetics. The data could be well represented by two components, one with a rise time τ 1 ≈ 2 μs and one with the rise time τ 2 ≈ 250 ms. The time dependent concentration of the two species is defined by Eqs. (a) and (b).2aC1=τ2τ1-τ2e-t∕τ1-e-t∕τ2 2bC2=1+τ2e-t∕τ2-τ1e-t∕τ1τ1-τ2 To perform structural analysis of the difference scattering data a range of trial structures were generated using MD simulations (see subsequent section). For each of these structures the theoretical X-ray scattering was calculated using CRYSOL . The scattering was calculated in 101 points, in the range 0 Å−1 < q < 1 Å−1, explicitly accounting for the hydrogen atoms present in the structure. The theoretical scattering was convoluted with the undulator spectrum, to allow comparison between experimental and theoretic scattering. The structures were considered to belong to a pool of dark or light candidate structures, depending on the state of FMN and cysteine 62 and the conformation of the LOV domain in the simulation. Every possible light-dark combination was calculated and scored against the experimental data using a previously used R factor, defined by Eq. ().3R=∑qΔIexperiment-cΔItheory2∑qΔIexperiment2where the scaling parameter c was allowed to vary. To avoid over fitting by allowing c to scale completely free the experimental scattering was first approximately scaled to the theoretical scattering by comparing absolute (rather than difference) scattering curves. Based on the laser energy titration performed it was also known that the scattering curve represented ca. 70–75% turnover (Supplementary Fig. ). The scaling parameter c was therefore allowed to vary so that 0.35 ≤ c ≤ 1.6. To avoid a disproportional influence on the fit by flexible regions, the histidine residues in the His-tag and residues 1–10 were removed after the simulation, but before calculating the scattering. After the difference scattering of all possible trial pairs were compared to the experimental data and the 100 pairs with lowest R pairs were accepted as good fits. [...] To generate physically reasonable trial structures molecular dynamics simulations were utilized. The simulations were performed using GROMACS 5.0.4 and the Charmm27 force field. Force field parameters for FMN and the flavin-cysteinyl adduct were manually adapted from to the GROMACS format. As a starting structure the YF1 crystal structure (pdb ID: 4GCZ) was used. Missing residues were added using MODELLER. The ADP found in the crystal structure was removed. There were no Adenosine nucleotides present during the simulations. The protein was placed in a cubic box, 1 nm larger than the protein in all directions and solvated with TIP3P water. The system was neutralized by adding 38 sodium ions. After adding ions the system underwent initial energy minimization until all forces were below 1000 kJ mol−1 nm−1. Subsequently the system was minimized for 100 ps in the NVT and NPT ensembles. During equilibration all non hydrogen atoms were position restrained with force constants of 1000 kJ mol−1 nm−2.The final production runs included restraining the alpha carbons of the sensory domain to the positions where they were found according to ref. . This was achieved by aligning alpha carbons of the best dark and light LOV domain structures from to the LOV domain of YF1. The position of the alpha carbons was fixated using force constants of 10,000 kJ mol−1 nm−2. This generates two simulation trajectories, one where the LOV domain is locked in the dark state, and one where the LOV domain is locked in the light state. The simulations were run for ca. 450 ns. The atomic coordinates were saved every 10 ps to generate structures for the structural fitting routine. All the simulations were run with periodic boundary conditions in all directions, all bonds were constrained using the LINCS algorithm and a time step of 2 fs was used. Particle Mesh Ewald (PME) electrostatics with fourth order interpolation and with a grid spacing of 0.16 were used. The cutoff scheme was Verlet with a 1.0 nm cutoff, cutoffs for short-range electrostatic and van der Waals interactions were 1.0 nm as well. During the production run pressure control was achieved using the Parrinello-Rahman barostat (τ P = 2 ps, P = 1 bar) and temperature control was achieved via the modified Berendsen (velocity-rescale) thermostat (τ T = 0.1 ps, T = 300 K). […]

Pipeline specifications

Software tools CRYSOL, GROMACS, MODELLER
Applications Small-angle scattering, Protein structure analysis
Organisms Escherichia coli, Dipturus trachyderma