Dimers of Azurin as model systems for electron transfer
Jongh, Thyra Estrid de
Citation
Jongh, T. E. de. (2006, September 12). Dimers of Azurin as model systems for electron
transfer. Retrieved from https://hdl.handle.net/1887/4554
Version:
Corrected Publisher’s Version
License:
Licence agreement concerning inclusion of doctoral thesis in the
Institutional Repository of the University of Leiden
Downloaded from:
https://hdl.handle.net/1887/4554
6
Crystal structure of a non-covalent
dim er of H 117G azurin
Thyra E. de Jongh, Anne-M arie M . Van
Roon, M iguel Prudêncio, M arcellus U bbink
and G erard W . Canters
112 C h a p te r 6
Abstract
113 C h a p te r 6
Introduction
The engineering of metalloproteins by means of site-directed mutagenesis presents interesting new opportunities for the study of structure/function relationships in proteins. The understanding of these relationships is of great importance in the development of new nanobiotechnological devices and much information on redox proteins has emerged from the study of a class of proteins known as cupredoxins or Type-1 blue copper proteins. This class is comprised of small one-electron carrier copper-containing proteins, characterized by an intense blue colour. Blue copper proteins have been isolated from a variety of plant and bacterial sources and have been the focus of extensive spectroscopic as well as structural investigations.[1]
Further information has come from site-directed mutagenesis of residues believed to have a crucial function in the protein.
It has been shown that substitution of metal coordinating residues in Type-1 copper proteins can result in the formation of novel proteins with new spectroscopic and redox properties.[2] In the case of the blue copper protein azurin from Pseudom onas
aeruginosa, mutation of the copper-coordinating residue H117 by a smaller and non-coordinating glycine residue leaves the protein with a solvent exposed aperture in the coordination sphere of the metal. In the absence of exogenous ligands this gap is occupied by solvent molecules resulting in the formation of a Type-2 like copper site. Addition of ligands such as imidazole, pyridine or chloride restores the spectroscopic properties of the native Type-1 copper site.[3-5] The H117G mutant is
of particular interest because it enables direct ‘hotwiring’ from the medium to the metal-binding site of the protein using suitable molecular ‘wires’. The feasibility of this approach was validated by the non-covalent dimerization of H117G azurin by bifunctional wires comprised of an extended alkane chain of variable length with a copper coordinating imidazole moiety on each end.[6;7]
It is of interest to explore the possible extension of this principle to the implementation of conductive molecular wires that may be applied in bioelectronic devices. To this end a proper description of the structure of H117G azurin and its ligand coordinated forms is desirable. Early attempts at crystallization of H117G azurin produced protein crystals that had suffered oxidation of the Cys112 SJ
ligand, rendering the protein incapable of metal coordination.[8] It is believed that
114 C h a p te r 6
More recently a high resolution structure of copper-containing H117G azurin was solved (S. Alagaratnam, manuscript in preparation), for which crystals were grown of the oxidized form of the protein. However, over time a gradual bleaching of these crystals, attributed to irreversible autoreduction, was observed. Autoreduction presents a serious obstacle in the successful crystallization of ligand-coordinated H117G azurin as it is known that the Cu(I) form of H117G azurin has very low affinity for exogenous ligands.[9] To circumvent these problems Cu(II)
was substituted for redox-inactive Zn(II) in the metal-binding site. This strategy is warranted by comparison of the crystal structures of Zn(II) and Cu(II) WT azurin which are superimposable with a backbone root mean square (RMS) deviation of 0.21 Å (PDB entry codes 4AZU and 1E67), showing that substitution by zinc has negligible effects on the overall protein fold.[10;11] There do exist small differences
between the immediate coordination spheres of the Cu(II) and Zn(II) ions though. The Met121 SG to Zn distance is somewhat increased and is too long for formation
of a covalent bond, whereas Gly45 O is moved slightly towards the zinc ion. It is therefore important to ensure that ligand binding can also take place with H117G azurin. We currently report the structure of a non-covalent dimer of Zn-H117G azurin formed by coordination of 1,6-di(imidazol-1-yl)hexane, as solved by X-ray diffraction to a resolution of 2.85 Å. It shows several intriguing features with regards to protein flexibility and the structure of the complex compared to WT azurin and several dimers of azurin.
Materials & Methods
Protein expression and isolation
Apo-H117G azurin was produced and purified as previously described.[5] The
concentration of apo-protein was determined from the observed absorbance at 280 nm (H280 = 9.1 ± 0.1 mM-1cm-1) recorded at room temperature (RT) on a
Perkin-Elmer lambda 18 spectrophotometer.
L inkers
The bifunctional ligands 1,5-di(imidazol-1-yl)pentane (1,5-dip) and 1,6-di(imidazol-1-yl)hexane (1,6-dih) were synthesized as previously reported.[7] Purity and integrity
of the samples after prolonged storage was confirmed by 1H NMR and working
115 C h a p te r 6 NMR spectroscopy
All1H NMR spectra were recorded on a Bruker Avance DMX 600 MHz spectrometer
at 304 K with a spectral width of 12.98 ppm in 4K memory using a Watergate-filtered pulse sequence. Free induction decays were Fourier transformed using a squared sine window function. All samples were prepared in 20 mM potassium phosphate (KPi), pH 6.5 buffer supplemented with 6% D2O for locking of the signal. The binding of 1,6-dih to Zn-H117G azurin was assessed by addition of 0.7 molar equivalents of the linker to a solution of 0.5 mM apo-H117G azurin. After recording of the 1H NMR spectrum, 2 molar equivalents of Zn(SO
4)2 were added
and the sample was incubated at RT for ca. 10 minutes allowing coordination of the linker to the reconstituted protein. The different effects of coordination of 1,6-dih or 1,5-dip on the 1H NMR spectra of Zn-H117G azurin were compared by titration
of 1 mM protein in 20 mM KPi, pH 6.5 with either of the linkers.
C rystallization and X -ray data collection
A solution of 10 mg/ml apo-H117G azurin in 10 mM Tris-HCl pH 7.5 was incubated with 1.3 molar equivalents of ZnCl2 for 2 hours at RT. Excess ZnCl2 was removed by ultracentrifugation (Centricon, MWCO 10,000). The sample was then incubated with 0.6 molar equivalents of 1,5-dip or 1,6-dih. Prior to crystallization the samples were filtered through a low protein binding 0.22 Pm filter (Millipore). Crystals of (Zn-H117G)2-1,6-dih were obtained by sitting drop vapour diffusion at 295 K using equal volumes of protein and reservoir solution. Crystals suitable for X-ray crystallography appeared within 3 days, from 1 µl of protein solution (10 mg/ml in 10 mM Tris-HCl pH 7.5) mixed with 1 µl of reservoir solution containing 100 mM Tris-HCl pH 8.56 and 20% (w/v) polyethylene glycol (PEG) 8000. No crystals were obtained for (Zn-H117G)2-1,5-dip. X-ray diffraction data of (Zn-H117G)2-1,6-dih were collected on an in-house beam using a MAR345 Image Plate detector. A crystal was mounted in a cryo-loop (Hampton Research) and passed quickly through a cryo-protectant solution containing 20% glycerol, followed by flash-freezing in a nitrogen gas stream at 100K. A dataset was collected to 2.85 Å resolution. All collected data were indexed, integrated and scaled with HKL2000.[12]
116 C h a p te r 6
Structure determination and refinement
The structure of (Zn-H117G)2-1,6-dih was solved by molecular replacement using the program Molrep[13] from the CCP4 program suite[14] using the structure of
Zn-WT azurin (PDB entry 1E67)[10] as a search model. A solution was obtained with an
R-factor of 39.7 % and a correlation coefficient of 58.8. After several rounds of rigid body and restrained refinement with Refmac5[15], the mutated amino acids were built
in manually using Xtalview[16], followed by automatic solvent building using ARP/
wARP.[17] Upon inspection of a F
o-Fc difference density map, a double conformation
of the main chain running from residue G116 to M121 of molecule A could be modelled and refined, leading to an improvement of Rfree. Extra density was also observed for this flexible loop in the three other molecules present in the asymmetric unit but was considered too weak to build any amino acids. Final refinement using tight NCS restraints between residues 1 to 114 of each protein chain and refinement of translation, liberation and screw (TLS) parameters[18] resulted in a model having
an R-factor of 19.2 % (Rfree 23.3 %). The quality of the model was checked by PROCHECK[19] and WHATIF[20] [Table 4.2]. The coordinates and structure factors
have been deposited in the Protein Data Bank under accession code 2IWE. Figures 6.3 - 6.6 were made using PyMOL 0.98.[21] Figure 6.5 was generated with the help
of the color_b.py script for colouring according to B-factor (http://adelie.biochem. queensu.ca/~rlc/).
Solvent accessibility and interface areas
The solvent-accessible surface area (ASA) was calculated using the program NACCESS 2.1.1 with a probe radius of 1.4 Å. The interface area was defined as the sum of the ASAs of the individual proteins minus the ASA of the complex.
Results & D iscussion
Non-covalent dimerization of Z n-H 117G azurin
As Zn-H117G azurin is colourless, coordination of the imidazole-based linkers can not be followed by the green-to-blue optical transition that is associated with the coordination of imidazole to Cu(II)-H117G azurin. Instead, ligand-binding was monitored by 1H NMR spectroscopy since coordination is expected to affect
residues in the immediate surrounding of the metal binding site. Addition of 0.7 molar equivalents of 1,6-dih to apo-H117G azurin had no discernible effect on the
117 C h a p te r 6
excess of Zn(SO4)2 to the sample, affected several resonances in the high-field region of the spectrum [Figure 6.1C,D]. Although none of these have been assigned for Zn-H117G azurin, in the ligand-coordinated state the resonances around 12 ppm [marked with * and *’ in Figure 6.1C,D] presumably originate from the NH protons
of histidine residues of which H35 and H46 are the most likely candidates as they are closest to the metal site. The unidentified resonance around 9.8 ppm [indicated by
Ì
, Figure 6.1] similarly shows a strong sensitivity for ligand coordination which suggests that it corresponds to a residue in the immediate vicinity of the ligand-binding site.Figure 6.1 clearly demonstrates that the presence of zinc is required for coordination of 1,6-dih whilst the formation of a higher molecular weight species is confirmed by the line broadening that occurs upon dimerization ('w = 3 - 4 Hz, where w is defined as the full width at half height). These results prove that, analogous to Cu(II)-H117G azurin, Zn-Cu(II)-H117G azurin is capable of coordinating bis-imidazolealkanes and of forming non-covalent dimers. To compare the effects on the structure of the construct with variation of the linker length, samples of Zn-H117G azurin were titrated with 1,5-dip or 1,6-dih respectively [Figure 6.2]. At a ratio of linker to protein of 1:2 complete dimerization was reached, showing that both linkers bind to the protein with high affinity and induce dimer formation [Figure 6.2C].
Figure 6.1: High field region of the 1H
118 C h a p te r 6
The recorded NMR spectra displayed somewhat different behaviour depending on the length of the linker [Figure 6.2A,B]. Although the same resonances are affected upon titration with 1,5-dip and 1,6-dih, each linker gives rise to a different new pattern of resonances. Whilst the resonances at 10.19 and 10.03 ppm disappear upon addition of either linker, titration with 1,6-dih gives rise to a single new resonance at 10.11 ppm. Titration with 1,5-dip, on the other hand, results in the appearance of several new resonances between 10.00 and 10.30 ppm. These findings indicate that addition of 1,6-dih leads to the formation of a dimeric species in which the connected proteins have adopted a very similar geometry.
In contrast, the 1H NMR spectrum of the (Zn-H117G)
2-1,5-dip complex suggests
that the sample is a mixture of two or more species with slightly different conformations near the metal binding site. The shorter length of the 1,5-dip linker may be insufficient to accommodate each of the imidazole moieties in exactly the same way. This observation is consistent with results obtained by EPR spectroscopy previously reported by van Pouderoyen et al. which showed that dimerization with 1,6-dih gave rise to a single pure Type-1 EPR spectrum, whereas the spectrum Figure 6.2: High field region of the 1H NMR spectra of 1 mM Zn-H117G azurin
119 C h a p te r 6
recorded for dimers dimerized with 1,5-dip appeared to be a superposition of two or more slightly different Type-1 like spectra.[7]
Crystal structure of (Zn-H117G)2-1,6-dih azurin
The crystal structure of (Zn-H117G)2-1,6-dih azurin was solved to 2.85 Å resolution [Figure 6.3, Table 6.1]. The asymmetric unit contained a total of four protein molecules arranged as a pair of nearly identical dimers. Each of the molecules within the dimer is oriented longitudinally along the axis formed by the 1,6-dih linker with the metal centres positioned at an intramolecular Zn-to-Zn distance of 16.1 Å. The four molecules within the asymmetric unit are superimposable with an averaged RMSD of 0.4 Å for the CD trace of the entire protein backbone, further lowered to 0.2 Å when the loop region between residues 116 and 120 is excluded, indicating a well conserved core structure. The overall protein fold of each of the separate proteins is in good agreement with that of WT azurin. The crystal structure of (Zn-H117G)2-1,6-dih, however, shows several interesting features.
120 C h a p te r 6
Table 6.1: Data collection, refinement and model statistics Data collection
Space group P1
Unit cell parameters a x b x c (Å) 42.66 x 49.72 x 66.07 Resolution (Å)1) 25-2.85 (2.9-2.85)
Measured reflections 28364
Unique reflections 11979
Completeness (%) 88.2
Rmerge2) (%) 8.7 (34.1)
Average I/sigma (I) 9.36 (1.83)
5HÀQHPHQWVWDWLVWLFV
R-factor (%) 19.2 (32.5)
Free R-factor (%) 23.3 (44.7) Average total B-value protein (Å2) 30.5
Model statistics
Number of TLS groups 4
Number of monomers in the asymmetric unit 4 Number of protein residues 512 Number of solvent molecules 3
Number of Zn ions 4
Number of ligands 2
Ramachandran plot (%)
Most favoured region 89.5 Additionally favoured region 10.2 Generally favoured region 0.2
Disallowed region 0.1
R.m.s. deviation from ideality
Bond lengths (Å) 0.013
Bond angles (º) 1.070
1)V alues of reflections recorded in the highest resolution shell are shown in parentheses, 2)R
121 C h a p te r 6
Ligand loop dynamics
Although the overall protein structure of each of the subunits is very similar to WT azurin, the protein surface area close to the metal-binding site is severely affected by the removal of the coordinating histidine residue. The loop on which residues 116 to 119 are located has been displaced from the protein interior, increasing its solvent accessibility [Table 6.2]. This loop has adopted slightly different conformations in each of the four subunits found in the asymmetric unit, all of which are characterized by relatively high B-factors indicative of increased flexibility [Figure 6.4, Figure 6.5]. This high degree of flexibility is understandable given that the loop is comprised of a series of small, flexible amino acids (Gly-Gly-Ser-Ala) and is consistent with the finding by Jeuken et al. that for Cu(I)-H117G azurin the residues 116-120 display increased backbone dynamics in solution compared to WT azurin.[22] Furthermore,
a recently elucidated structure of Cu-H117G azurin shows a similar high degree of flexibility. The displacement of the loop region from the protein interior is in part inherent to the H117G mutation as it was also observed for free Cu-H117G azurin, but becomes more pronounced in the ligand-coordinated state.
122 C h a p te r 6
It is interesting to consider what may cause the different loop conformations found in WT and H117G azurin. In both Zn- and Cu-WT azurin the loop residues G116, H117 and S118 form several hydrogen bonds to nearby located water molecules but only H117 is directly connected to the protein framework by hydrogen bonds between its backbone N and O atoms and residues 114, 120 and 121. Since the HÆG substitution does not affect the composition of the protein backbone, these hydrogen bonds could have been formed also in the structure of H117G azurin. Nonetheless, it is apparent from both the Cu-H117G and (Zn-H117G)2-1,6-dih azurin structures that the conservation of the main chain itself is insufficient for maintaining the WT loop conformation. It has been postulated that the imidazole side chain of His117 exhibits some S-electron overlap with Phe114 that helps to stabilize the WT conformation.[23] Additional loop stabilization could be achieved
by coordination of the NG of the imidazole side chain to the metal centre, although
the available crystal structures of apo-azurin (PDB entry 1E67) which show a fully buried loop despite the absence of a metal, indicate that this interaction is not required.[24]Finally, in the WT conformation the region between residues
123 C h a p te r 6
displaced loop conformations are all situated in the disallowed region of the plot. These conformations are therefore not accessible to the WT protein.
Interface packing and solvation
Another interesting aspect of the (Zn-H117G)2-1,6-dih azurin structure concerns the relative orientation of the connected moieties in the complex. Thus far, most structures of P. aeruginosa azurin have shown a highly conserved interaction surface involving packing of the proteins through an area of the protein commonly referred to as the hydrophobic patch. In these structures the proteins are generally packed together in a fashion in which the D-helices on the two subunits are situated on opposite sides of the complex with respect to its longitudinal axis. It has been shown that as long as sufficient conformational freedom is provided, also covalent dimers of azurin have a tendency to adopt this orientation.[2] In contrast, the structure of
the non-covalent dimer of (Zn-H117G)2-1,6-dih azurin shows both D-helices on the same side [Figure 6.6].
124 C h a p te r 6
Similar, although not quite identical, orientations have been observed before for the C112D and F114A mutants of azurin (PDB access codes 1AG0 and 1AZN respectively).[23;25] Both these mutations have direct bearing on the metal-binding
site of the protein, either through coordination to the metal (C112) or more indirectly through van der Waals contact with H117 (F114). In all structures the residues involved in the interaction surface are largely conserved indicating that hydrophobicity accounts for association of the proteins whereas the specificity of the interaction is more closely regulated by the interplay of forces such as van der Waals interactions or intermolecular hydrogen bonds. Mutations in the vicinity of the dimer interface may easily shift the balance between these to favour alternative conformations. In the dimer interface of (Zn-H117G)2-1,6-dih azurin a single intermolecular hydrogen bond is formed between Asn42-O and Ser118-OJ and
no additional electrostatic interactions are present, suggesting that the orientation of the dimer interface is not significantly stabilized by any other forces than the hydrophobic shielding itself.
Although the area involved in the interaction surface of the dimer is largely retained, the solvent accessibility of the dimer interface is somewhat greater than in structures of azurin that display the classical hydrophobic interaction packing [Table 6.2].[2;26]
This is related both to the previously mentioned exposure of the loop region to the solvent and to the way in which the monomers are packed together.
Table 6.2: Solvent accessibility and interface area in complexes of azurin Accessible Surface Area
(Å2) per monomer†
Interface Area (Å2)
Free Complex
wild type (1E5Y) 6579 6066 1026 N42C-BMME (1JVL) 6547 6050 995 N42C-disulfide bridged (1JVO) 6524 6033* 983*
(Zn-H117G)2-1,6-dih 6687 6205 964
† T he term ‘free’ is used to indicate that the coordinate file used contained only a single
monomer and additional monomers have been removed from the asymmetric unit. To calculate the ASA for each of the monomers in a ‘complex’, the coordinate files have been edited to contain only those molecules that share the contact surface investigated.
* T he ASA of the N42C-disulfide bridged azurin complex and its interface area correspond
125 C h a p te r 6
The positions of the protein ligands are fairly conserved with respect to Zn-WT azurin with average RMS deviations of 0.19 (Gly45), 0.19 (His46) and 0.24 Å (Cys112) [Table 6.3]. The position of Met121 is more substantially affected by the mutation with an RMSD of 0.79 Å. Met121 is, however, not considered a ligand in the structure of Zn-WT azurin, due to an increase in the SG121 - Zn distance
compared to Cu(II)-WT azurin.[10]
Table 6.3: Bond lengths and angles between ligands and the metal centre (M) Cu WT azurin (1E5Y) Zn WT azurin (1E67) Zn-H117G-1,6-dih azurin Bond lengths (Å) Gly45-M 3.02 (0.06) 2.32 (0.07) 1.99 (0.01) His46-M 2.14 (0.07) 2.07 (0.04) 2.06 (0.02) Cys112-M 2.29 (0.01) 2.30 (0.02) 2.34 (0.01) Met121-M 3.25 (0.06) 3.38 (0.07) 3.51 (0.15) His117/Linker-M 2.10 (0.07) 2.01 (0.03) 2.07 (0.01) Bond angles (o) SJ112 - M - NG46 133 (2) 128 (3) 136 (1) SJ112 - M - NG117/Linker 122 (1) 121 (1) 98 (13) SJ112 - M - SG121 113 (1) 103 (1) 99 (2) NG46 - M - SG121 72 (1) 70 (2) 67 (2) NG46 - M - NG117/Linker 105 (2) 110 (3) 118 (10) SG121 - M - NG117/Linker 91 (4) 83 (1) 81 (2) O45 - M - NG46 75 (1) 84 (1) 89 (4) O45 - M - SG121 145 (1) 152 (2) 152 (3) O45 - M - NG117/Linker 87 (2) 98 (2) 102 (4) O45 - M - SJ112 99 (1) 101 (1) 107 (1)
The bond lengths and angles were obtained by averaging over all the molecules within the asymmetric unit.
Considerations for electron transfer in complexes of H117G azurin
For WT azurin much importance has been attributed to the formation of a hydrogen bond network between H117 NHand water molecules in the protein interface that
has been implicated in mediating fast electron self-exchange between the interacting proteins.[2;11;27;28] In complexes of H117G azurin that involve coordination of linkers
126 C h a p te r 6
Conclusions
Site-directed mutagenesis of the copper ligand H117 of the blue copper protein azurin (P. aeruginosa) creates a solvent exposed aperture in the protein which can accommodate a variety of ligands, thereby permitting ‘hotwiring’ of the protein.[4;5;7] Hotwiring of redox proteins is a promising strategy in the development
of efficient bionanotechnological devices which requires a thorough understanding of the structure/function relationships of protein-wire constructs. We report the crystallization of a ligand-coordinated form of H117G azurin which has formed a non-covalent dimer. Crystallization was aided by substitution of Cu(II) for Zn(II) in the metal-binding site of the protein. NMR spectroscopy confirmed coordination of the bifunctional 1,5-dip and 1,6-dih linkers to Zn-H117G azurin. Coordination of 1,6-dih led to the formation of dimers in which the connected proteins have adopted nearly identical conformations, whereas coordination of 1,5-dip produced dimers with slight conformational differences between the subunits. The (Zn-H117G)2 -1,6-dih azurin dimer was crystallized and a structure was solved to 2.85 Å resolution. It shows increased flexibility in the loop region between residues 116-120, surrounding the site of mutation, compared to WT azurin. The displacement of this loop from the protein interior may facilitate the coordination of larger exogenous ligands. The structure represents a novel type of dimer in which the proteins have adopted a relative orientation distinct from those previously reported for other dimers of azurin.
References
[1.] A. Messerschmidt, R. Huber, K. Wieghardt, T. Poulos, Handbook of metalloproteins, 2001.
[2.] I. M. C. van Amsterdam, M. Ubbink, O. Einsle, A. Messerschmidt, A. Merli, D. Cavazzini, G. L. Rossi, G. W. Canters, Nature Structural Biology 2002,9 48-52.
[3.] T. den Blaauwen, C. W. G. Hoitink, G. W. Canters, J. Han, T. M. Loehr, J. Sanders-Loehr, Biochemistry 1993,32 12455-12464.
[4.] T. den Blaauwen, G. W. Canters, Journal of the American Chemical Society 1993,115 1121-1129.
127 C h a p te r 6
[6.] G. van Pouderoyen, T. den Blaauwen, J. Reedijk, G. W. Canters, Biochemistry 1998, 37 7656.
[7.] G. van Pouderoyen, T. den Blaauwen, J. Reedijk, G. W. Canters, Biochemistry 1996, 35 13205-13211.
[8.] C. Hammann, G. van Pouderoyen, H. Nar, F. X. G. Ruth, A. Messerschmidt, R. Huber, T. den Blaauwen, G. W. Canters, Journal of Molecular Biology 1997, 266 357-366.
[9.] L. J. C. Jeuken, P. van Vliet, M. P. Verbeet, R. Camba, J. P. McEvoy, F. K. Armstrong, G. W. Canters, Journal of the American Chemical Society 2000, 122 12186-12194.
[10.] H. Nar, R. Huber, A. Messerschmidt, A. C. Filippou, M. Barth, M. Jaquinod, M. van de Kamp, G. W. Canters, E uropean Journal of Biochemistry 1992, 205 1123-1129.
[11.] H. Nar, A. Messerschmidt, R. Huber, M. van de Kamp, G. W. Canters, Journal of Molecular Biology1991, 221 765-772.
[12.] Z. Otwinowski, W. Minor, Macromolecular Crystallography, Pt A 1997, 276 307-326.
[13.] A. Vagin, A. Teplyakov, Journal of Applied Crystallography 1997, 30 1022-1025.
[14.] S. Bailey, Acta Crystallographica Section D-Biological Crystallography 1994, 50 760-763.
[15.] G. N. Murshudov, A. A. Vagin, E. J. Dodson, Acta Crystallographica Section D-Biological Crystallography1997, 53 240-255.
[16.] D. E. McRee, Journal of Structural Biology 1999, 125 156-165.
[17.] V. S. Lamzin, K. S. Wilson, Acta Crystallographica Section D-Biological Crystallography1993, 49 129-147.
[18.] M. D. Winn, M. N. Isupov, G. N. Murshudov, Acta Crystallographica Section D-Biological Crystallography2001, 57 122-133.
[19.] R. A. Laskowski, M. W. MacArthur, D. S. Moss, J. M. Thornton, Journal of Applied Crystallography1993, 26 283-291.
[20.] G. Vriend, Journal of Molecular Graphics 1990, 8 52-&.
[21.] W. L. DeLano, The PyMO L Molecular Graphics System (http://www.pymol. org), San Carlos, CA, USA 2002.
128 C h a p te r 6
W. Canters, Journal of Molecular Biology 2000, 299 737-755.
[23.] L. C. Tsai, L. Sjolin, V. Langer, T. Pascher, H. Nar, Acta Crystallographica Section D-Biological Crystallography1995, 51 168-176.
[24.] H. Nar, A. Messerschmidt, R. Huber, M. van de Kamp, G. W. Canters, Febs Letters1992, 306 119-124.
[25.] S. Faham, T. J. Mizoguchi, E. T. Adman, H. B. Gray, J. H. Richards, D. C. Rees, Journal of Biological Inorganic Chemistry 1997, 2 464-469.
[26.] P. B. Crowley, M. A. Carrondo, Proteins-Structure Function and Bioinformatics 2004, 55 603-612.
[27.] K. V. Mikkelsen, L. K. Skov, H. Nar, O. Farver, Proceedings of the National Academy of Sciences of the U nited States of America1993, 90 5443-5445. [28.] H. Nar, A. Messerschmidt, R. Huber, M. van de Kamp, G. W. Canters,
13
