Similar protocols

Protocol publication

[…] For crystallization, 50 μM MutSΔC800/MutLLN40 complex was incubated with 25 μM DNA containing a G:T mismatch (27-bp: TGCCAGGCACCAGTGTCAGCGTCCTAT annealed with ATAGGACGCTGACACTGGTGCTTGGCA or 100-bp, same sequence as above) for 25 min on ice. AMP-PNP was subsequently added to a concentration of 1 mM and the protein was crystallized at 4°C using vapor diffusion in 9–12% PEG-8000, 100 mM Tris pH 7.0, 200 mM MgCl2, and 80–450 mM sodium malonate. Microseeding was used to increase crystal nucleation. Crystals were cryoprotected in mother liquor supplemented with 25% ethylene glycol and 100 mM KCl before flash-cooling in liquid nitrogen. Diffraction data were collected at 100 K at beamline ID-29 at the ESRF or beamline PX-III at the SLS.Crystallographic data were processed with XDS () or iMOSFLM () and scaled using Aimless from the CCP4 suite (). Crystal structures were solved in consecutive steps of finding domains using Phaser (). Several search models were used, but best results were obtained with domains from chain A of PDB entry 1W7A as search models for MutSΔC800 and chain A from PDB entry 1BKN for MutLLN40, while clear density for residues 150–164 of MutLLN40 allowed building as in PDB entry 1NHJ. The search process was improved by going back and forth between the different datasets to find missing domains. Initial structure solution was performed starting from crystal form 1 as follows: first, a search model consisting of residues 267–800 of chain A of PDB entry 1W7A (MutS) was searched twice using Phaser, which resulted in a solution with these chains forming a tilted MutS dimer. Next, this solution was used together with a search model consisting of chain A of 1BKN (MutLLN40), which placed this protein against the ATPase domain of one MutS subunit. Then, the second MutLLN40 was found with Phaser using a brute rotation search of 15° around the angle that would orient this MutLLN40 on the other side of the MutS dimer in a similar manner as the first, and automated translation, packing and refinement steps by Phaser indeed placed the MutLLN40 in the symmetrically equivalent position. One connector domain (residues 128–266 of chain A of 1W7A) was then found with Phaser, and the second connector domain was placed using similar steps as for the second MutLLN40 search. Thus the search identified the equivalent dimeric counterpart three times for separate parts of the complex (the main MutS chain, MutLLN40 and connector domains). The resulting MutS-MutLLN40 complex structure could then be used as a search model in all crystal forms and easily identified equivalent complexes in each of those (present three times in the asymmetric units in the 6.6 Å and 7.6 Å datasets). Finally, for crystal form 1, an additional ‘half complex’ was found with Phaser using one MutS chain and one MutLLN40 chain of the existing complex structure. This second complex forms a symmetry-generated dimer over a twofold axis, with similar MutS-MutLLN40 interfaces, but the MutS clamp domains in this crystallographic dimer could not be modeled. This second conformer forms a more compact MutS dimer, probably due to crystal packing, but since it has identical interfaces with MutLLN40 we focussed on the main conformation throughout this paper. Excellent quality of the structure solutions after molecular replacement with the complete but unrefined models is evident from the Phaser statistics: TFZ = 9.0/LLG = 996 for 4.7 Å; TFZ = 14.2/LLG = 899 for 6.6 Å; and TFZ = 13.0/LLG = 795 for the 7.6 Å dataset.Refinement was first performed using rigid body refinement in REFMAC5 (, ), for which the following domains of MutS were defined: residues 128–266, 267–765, 766–800; and for MutL: residues 20–204, 205–331. Next, limited restrained refinements were performed, first using ProSMART-generated external restraints () to the PDB_REDO-optimized () entries of chain A of 1W7A and chain A of 1BKN in order to ensure consistency with prior observations, followed by TLS and jelly-body refinement in latter stages. PDB_REDO-optimized homologues were used for external restraint generation in order to maximize reliability of the prior structural information. All refinements were performed using REFMAC5 (, ). During refinement, clear density became visible for missing residues 150–164 of the MutLLN40 subunits, which followed the conformation of PDB entry 1NHJ. Interestingly, this conformation was different from that in the MutL search model state, indicating this to be real signal, and not due to bias from the search model. Also, AMP-PNP could be placed in density in the nucleotide binding sites of MutS. During intermediate stages, PDB_REDO and MolProbity () were used to correct geometry and perform side-chain flips. After refinement, all structures were in the 97th–100th Clashscore and 98th–100th MolProbity score percentiles. Refinement and data collection statistics can be found in . Figures and videos were generated with MacPyMOL (, interpolations between conformations were created with LSQMAN () and electrostatic surface with CCP4mg (). Protein interface areas were calculated using PISA () for which the missing loop of residues 126–131 of MutLLN40 in interface 2 was modeled as in PDB entry 1NHJ. […]

Pipeline specifications

Software tools XDS, iMosflm, CCP4, REFMAC5, MolProbity, PyMOL, CCP4mg
Databases PDB_REDO
Applications Small-angle scattering, Protein structure analysis
Chemicals Adenosine Triphosphate