Ional Effects on the Circular Dichroismpositions from the crystal structure with the rest of the system being deleted due to computational demands. The missing hydrogen atoms were added using GausView5 [25]. Continuum solvent model with a dielectric constant of 4 was used to approximately represent the protein environment. The B3LYP functional, with three basis sets (6-31G(d), 6-31G(d,p) and 631++G(d,p)), was used as it was previously demonstrated that this model provides reasonable results for tryptophan zipper proteins [26]. A comparison of the different basis sets is provided in Figure S1 in Supporting Information S1 and here we will focus only on B3LYP/6-31G(d) results of the wild-type and all seven tryotophan mutants. The MD simulations were carried out with the GROMACS code (version 4.3.1) [27] and Gromos43b1 force field for 20 nanoseconds (ns) for the wild-type enzyme and all seven tryptophan mutants. The system was prepared from the crystal structure of HCAII using Gromacs utilities 
for system preparation. The hydrogen atoms were added and geometry was energy minimized according to the protonation states of the ionogenic groups. Consequently the entire system was minimized using the steepest descent algorithm. The system was solvated using ?rectangular SPC water box placed 10 A from the edges of the protein, neutralized and the solvent was equilibrated for 50 picoseconds (ps). Production MD was run for 20 ns in NPT ensemble at 310K applying Berendsen thermostat [28]. The electrostatic interactions were treated by Particle Mesh Ewald method [29]. The quality of the simulations was monitored by RMSDs (Figure S2 in Supporting Information S1). The structures of the seven mutants were prepared from the crystal structure of the wild- type enzyme using the What if web interface (http://swift.cmbi.ru.nl/servers/html/index.html) [30]. Consistent with the JW-74 cost experimental CD studies of HCAII tryptophan mutants [8], the following structures: W5F, W16F, W97C, W123C, W192F, W209F and W245C were prepared. The received structures were additionally energy minimized to avoid local stretching interactions. All structures for MD simulations were prepared as in the case for the wild-type enzyme. The protein structure of HCAII was visualized using VMD [31]. The experimental CD spectra were taken from [8].Results and Discussion CD Spectrum of the Wild-type HCAII Based on the Crystal StructureThe near-UV CD spectrum of the wild-type enzyme Thiazole Orange calculated with the matrix method using the crystal structure in comparison to the experimental spectrum is presented in Figure 2A (the computed spectrum is shown in blue and the experimental spectrum is shown in black). The calculated spectrum is characterized by a spectral minimum (at 263 nm) and represents the correct spectral sign and overall shape, however, the magnitude at the spectral minimum is deeper than the experimental one (by 94 deg.cm2.dmol21). The position of the minimum of the calculated spectrum is blue-shifted by 7 nm in respect to the experimental spectrum as in the calculations done by Hirst et al. performed with the same model and parameters [9]. The achieved level of agreement is reasonable for the semiempirical matrix method we apply, however applying potentially more accurate methods such as TDDFT on the system is not feasible at present. In the experimental spectrum, there are features above 280 nm due to the fine vibration structure, not reproduced in the calculated spectrum. At present, howeve.Ional Effects on the Circular Dichroismpositions from the crystal structure with the rest of the system being deleted due to computational demands. The missing hydrogen atoms were added using GausView5 [25]. Continuum solvent model with a dielectric constant of 4 was used to approximately represent the protein environment. The B3LYP functional, with three basis sets (6-31G(d), 6-31G(d,p) and 631++G(d,p)), was used as it was previously demonstrated that this model provides reasonable results for tryptophan zipper proteins [26]. A comparison of the different basis sets is provided in Figure S1 in Supporting Information S1 and here we will focus only 
on B3LYP/6-31G(d) results of the wild-type and all seven tryotophan mutants. The MD simulations were carried out with the GROMACS code (version 4.3.1) [27] and Gromos43b1 force field for 20 nanoseconds (ns) for the wild-type enzyme and all seven tryptophan mutants. The system was prepared from the crystal structure of HCAII using Gromacs utilities for system preparation. The hydrogen atoms were added and geometry was energy minimized according to the protonation states of the ionogenic groups. Consequently the entire system was minimized using the steepest descent algorithm. The system was solvated using ?rectangular SPC water box placed 10 A from the edges of the protein, neutralized and the solvent was equilibrated for 50 picoseconds (ps). Production MD was run for 20 ns in NPT ensemble at 310K applying Berendsen thermostat [28]. The electrostatic interactions were treated by Particle Mesh Ewald method [29]. The quality of the simulations was monitored by RMSDs (Figure S2 in Supporting Information S1). The structures of the seven mutants were prepared from the crystal structure of the wild- type enzyme using the What if web interface (http://swift.cmbi.ru.nl/servers/html/index.html) [30]. Consistent with the experimental CD studies of HCAII tryptophan mutants [8], the following structures: W5F, W16F, W97C, W123C, W192F, W209F and W245C were prepared. The received structures were additionally energy minimized to avoid local stretching interactions. All structures for MD simulations were prepared as in the case for the wild-type enzyme. The protein structure of HCAII was visualized using VMD [31]. The experimental CD spectra were taken from [8].Results and Discussion CD Spectrum of the Wild-type HCAII Based on the Crystal StructureThe near-UV CD spectrum of the wild-type enzyme calculated with the matrix method using the crystal structure in comparison to the experimental spectrum is presented in Figure 2A (the computed spectrum is shown in blue and the experimental spectrum is shown in black). The calculated spectrum is characterized by a spectral minimum (at 263 nm) and represents the correct spectral sign and overall shape, however, the magnitude at the spectral minimum is deeper than the experimental one (by 94 deg.cm2.dmol21). The position of the minimum of the calculated spectrum is blue-shifted by 7 nm in respect to the experimental spectrum as in the calculations done by Hirst et al. performed with the same model and parameters [9]. The achieved level of agreement is reasonable for the semiempirical matrix method we apply, however applying potentially more accurate methods such as TDDFT on the system is not feasible at present. In the experimental spectrum, there are features above 280 nm due to the fine vibration structure, not reproduced in the calculated spectrum. At present, howeve.
