5.2.2.1 Phasing of medium-resolution Thermus-MnSOD structure
When the resolution limits were extended to 2.4 Å, the number of unique reflections observed increased to 24,547, a more than six-fold increase for both the native and platinum derivatives. There was again no mention of a low-resolution cut-off. Also lacking are statistics on data-set completion, redundancy and average signal strength. Some X-ray data statistics from Thermus-MnSOD at 2.4 Å and data from other structural projects are given in a table for X-ray data collected at the University of California, San Diego Multi- wire Detector Facility (Xuong et al., 1985). The data given in this table are generally in agreement with those presented in the Thermus-MnSOD structure. However, the presentation of the data is incomplete and ambiguous. Lower resolution limits are not given, no indication of data completeness is given, high-resolution shells are not treated separately, and data are given for entire multi-crystal projects (including metal derivatives) rather than separate data for individual crystals.
The method used to derive structure factors from measured intensities is not expressly given. Most likely, structure factors were estimated using Gaussian distribution approaches (Wilson, 1949) or Bayesian distributions (French & Wilson, 1978). The strongest 12,789 reflections, about half the total number of reflections, were used to refine the positions of the Pt binding using low-resolution results as a starting point. It is unclear whether the low resolution data set used was derived from crystals grown at pH 5.7, pH 7.0 or averaged information from both pH ranges. This allowed for the refinement of FH (structure factors of metal component) for the 2.4 Å platinum structure factors. There is no mention as to whether FP (structure factors of the protein component) were calculated de
novo, or whether the FP were derived from low resolution SIR and MIR experiments as a starting point. The phases were then refined and extended to 22,922 reflections, although the approach used for extension and refinement of phases is not given or whether the initial structure solution contributed phase information. Weighted Fourier maps were generated then averaged using the NCS operators derived from the low-resolution data.
This may have been a suboptimal approach to use, as it entails transferring information about phases from the low-resolution data set derived from a poorly diffracting crystal measured on older equipment to a dataset with inherently better characteristics collected on more modern equipment. It may have been more appropriate to repeat the SIR and MIR experiments to 2.4 Å. In the low-resolution study the use of a dimercurial soak allowed the generation of phase information by MIR and these maps were of better quality than the SIR maps.
5.2.2.2 Refinement of medium resolution Thermus-MnSOD structure
The starting point for structural refinement was the backbone of Ec-FeSOD which had been overlaid onto the low-resolution density. Building was done using primary amino- acid sequences of MnSODs from Thermus aquaticus and Bacillus stearothermophilus. There are point differences between these sequences and that for Thermus-MnSOD. These differences are likely to introduce bias into the structure. In particular side chains may have been fitted into ambiguous electron density incorrectly.
The protein structure was built into the NCS-averaged map density, but unaveraged map density was also available for interpretation. The use of NCS-averaged maps for building may introduce bias (Fabiola et al., 2006). Where the differences in atomic positions between the NCS partners are negligible, this bias in the NCS maps will also be negligible, as the “true” density will be averaged with other “true” density. When there are real differences in atomic positions between NCS partners within a crystal the use of NCS maps can introduce errors as “true” density is averaged with “noise”, regions of low density and “wrong” density such as solvent. The regions of a protein that are the most convergent in NCS maps are buried residues, and the most divergent will be residues that have multiple rotamers and are exposed to the solvent. This may have been compounded by inaccurate primary amino-acid sequence. Modern practice is in early cycles of building to use NCS-averaged maps to build the buried core of the protein and then, after the structure has improved, to do the final stages of building in un-averaged maps.
The building of this structure does not appear to have been an iterative process between real-space refinement and refinement of phases, which may mitigate the effect of using incorrect amino-acid sequences. As more of a structure is built this should lead to improved phasing and hence better maps. There is no refinement of the structure against the X-ray data or against any geometric parameter, even though at the time there were computer programs available at the time. There is also no validation of the structure against the X-ray data (such as an R-factor) or against geometric parameters (such as a Ramachandran plot). There is only mention of three non-protein atoms: the active-site manganese, its bound water-derived molecule and a water peak associated with one histidine. At 2.4 Å resolution, it is not unreasonable to expect to see well ordered waters such as those associated with the dimer-dimer interface and the solvent-access funnel. At this resolution the metal bound at the penta-coordinate active site was designated to be Mn3+ with four ligands derived from the protein and the fifth ligand being a solvent. No evidence of a sixth ligand was detected. Some of the rebuilt regions are described in the body text, but there is no indication as to how many of the residues have definite
assignment and how many are modelled as alanine or glycine. There is also no indication as to how well the NCS-related atoms agree.
The authors state that the coordinates would be submitted to the PDB, but it appears that this structure was superseded by the 1.8 Å structure.
5.2.3 The 1.8 Å resolution structure of Thermus-MnSOD