2.4 Methods and Materials
49
2.4.6 Structure Determination and Refinement
Data collection, integration, scaling, and empirical absorption correction were carried
out in the Rigaku CrystalClear-2.054 program package. The structure was solved
in 0.82Å resolution by direct method using the software of SIR201155 and well
refined by Full-Matrix-Least-Squares against F2 by SHELXTL97 [55]. The nonhydrogen atoms were anisotropically refined and hydrogen atoms were placed at
idealized positions and refined using the riding model. The absolute configuration
was determined using the method of Flack [56]. The statistics of data collection and
final refinement were shown in Table 2.4.
2.4.7 Computational Studies
2.4.7.1 Methods for Computational Studies
Simulation settings: All simulations were carried out using the Gromacs 4.5.4 [57].
Our recently developed residue-specific force field [58, 59] RSFF2 [60] was used to
treat each peptide, except for the tether, which was described using the generalized
Amber force field (GAFF) [60, 61] with the restrained electrostatic potential (RESP)
[62] charges. Each starting structure was initially constructed using the HyperChem
software, and solvated in a truncated octahedron box (30 Å in length) with 647709 TIP3P water molecules (depending on the peptides). Energy minimization was
carried out using steepest descent method. Then, the initial periodic box volume was
equilibrated using a 3 ns MD simulation in an NPT ensemble near 300 K and 1 atm.
Subsequently, the initial structures for REMD were obtained at regular intervals
from a 30 ns NVT MD trajectory at 600 K. For each REMD simulation, 24 replicas
were used with temperature range from 300 K to 600 K. By frequently exchanging
the replicas of different temperatures, REMD can speed up the barrier crossing and
achieve higher efficiency in conformational sampling [63].
The electrostatics were treated using the particle-mesh Ewald (PME) method [64]
with a real-space cutoff of 0.9 nm and van der Waals interaction cutoff at 0.9 nm
with the long-range dispersion correction for energy and pressure in all simulations.
A velocity rescaling thermostat [65] with τ T = 0.2 ps and a Berendsen barostat [66]
with τ P = 0.5 ps were used to maintain constant temperature and constant pressure
(for NPT simulations), respectively. All bonds involving hydrogen were constrained
using LINCS [67], and a time step of 2 fs was used. At the same time, the mass of water
oxygen atom was reduced from 16 to 2 amu to increase the sampling efficiency [68]
without altering the thermodynamics equilibrium properties. 24 replicas of the system
were simulated simultaneously at temperatures from 300 to 600 K. The intermediate
temperatures were chosen following a recent study [69] to obtain uniform exchange
rate, and exchanges were attempted between neighboring replicas every 1.0 ps. The
49
2.4.6 Structure Determination and Refinement
Data collection, integration, scaling, and empirical absorption correction were carried
out in the Rigaku CrystalClear-2.054 program package. The structure was solved
in 0.82Å resolution by direct method using the software of SIR201155 and well
refined by Full-Matrix-Least-Squares against F2 by SHELXTL97 [55]. The nonhydrogen atoms were anisotropically refined and hydrogen atoms were placed at
idealized positions and refined using the riding model. The absolute configuration
was determined using the method of Flack [56]. The statistics of data collection and
final refinement were shown in Table 2.4.
2.4.7 Computational Studies
2.4.7.1 Methods for Computational Studies
Simulation settings: All simulations were carried out using the Gromacs 4.5.4 [57].
Our recently developed residue-specific force field [58, 59] RSFF2 [60] was used to
treat each peptide, except for the tether, which was described using the generalized
Amber force field (GAFF) [60, 61] with the restrained electrostatic potential (RESP)
[62] charges. Each starting structure was initially constructed using the HyperChem
software, and solvated in a truncated octahedron box (30 Å in length) with 647709 TIP3P water molecules (depending on the peptides). Energy minimization was
carried out using steepest descent method. Then, the initial periodic box volume was
equilibrated using a 3 ns MD simulation in an NPT ensemble near 300 K and 1 atm.
Subsequently, the initial structures for REMD were obtained at regular intervals
from a 30 ns NVT MD trajectory at 600 K. For each REMD simulation, 24 replicas
were used with temperature range from 300 K to 600 K. By frequently exchanging
the replicas of different temperatures, REMD can speed up the barrier crossing and
achieve higher efficiency in conformational sampling [63].
The electrostatics were treated using the particle-mesh Ewald (PME) method [64]
with a real-space cutoff of 0.9 nm and van der Waals interaction cutoff at 0.9 nm
with the long-range dispersion correction for energy and pressure in all simulations.
A velocity rescaling thermostat [65] with τ T = 0.2 ps and a Berendsen barostat [66]
with τ P = 0.5 ps were used to maintain constant temperature and constant pressure
(for NPT simulations), respectively. All bonds involving hydrogen were constrained
using LINCS [67], and a time step of 2 fs was used. At the same time, the mass of water
oxygen atom was reduced from 16 to 2 amu to increase the sampling efficiency [68]
without altering the thermodynamics equilibrium properties. 24 replicas of the system
were simulated simultaneously at temperatures from 300 to 600 K. The intermediate
temperatures were chosen following a recent study [69] to obtain uniform exchange
rate, and exchanges were attempted between neighboring replicas every 1.0 ps. The