SUPPLEMENTARY INFORMATION

The crystal structure of DNA mismatch repair protein MutS binding to a G:T mismatch

Meindert H. Lamers*†, Anastassis Perrakis‡†, Jacqueline H. Enzlin*§, Herrie H. K. Winterwerp*, Niels de Wind*||, Titia K. Sixma*

Figure A Surface charge representation48 of MutS dimer. Electrostatic charges (red for negative, blue for positive) shown between –10 and +10 kT. The figure illustrates the overall negative charge of MutS and the existence of channels through the protein. The positive interface of the clamp domains with DNA is just visible, as well as an additional positive groove at the top of the clamp domains.

Figure B Sequence alignment of E. coli MutS to human MSH2, MSH3 and MSH6. The alignment used is the multi sequence alignment of ref 50. Secondary structure elements (:-helix, : -strand, : 310-helix) are displayed and coloured according to domain structure (see Fig. 2). Conservation displayed by grey boxes for similarity, grey background for identity. Disordered loop indicated by blue dashed line under the sequence. All DNA contacts (including van der Waals interactions) from the specific mismatch-binding domain shown in blue background with white lettering. Non-specific DNA binding residues denoted by light pink filled circles for mismatch binding monomer, dark pink open circles for second monomer. All dimer contacts from mismatch binding monomer in light green background, from second monomer in dark green boxes. Sequence variants found in HNPCC patients (see Supplementary information Fig. D) are indicated in red boxes in MSH2 and MSH6 sequences respectively.

Figure C Domain movements between the two monomers of MutS, after superposition of the core domains. A linear interpolation between the coordinates of the two superimposed monomers was done in Cartesian coordinate space to create all the (fictional) intermediate stages for the animation. The animation is intended only to visualise differences in orientation of domains and does not claim to represent real-time movements. DNA and protein are represented as cartoon, coloured as in figure 2; Phe36 and ADP as ball-and-stick models. Regions that are disordered in the non-mismatch binding monomer become transparent as the animation goes 'towards' that monomer. The same is done for the ball-and-stick model of the ADP molecule which is placed in a random place in the non-mismatch binding monomer to facilitate the animation. The figure was drawn with BOBSCRIPT/Raster3D and animated with the ImageMagick tools.

Figure D Missense mutations found in HNPCC families localised to MutS structure. MSH6 mutations (green spheres) on mismatch-binding monomer (right), MSH2 mutations (blue spheres) on non mismatch binding monomer (left). Orientation as in Fig. 1, colouring of backbone from grey to red with increasing conservation50. Missense variants were assembled (June 2000) from the database of the international collaborative group on HNPCC and the human gene mutation database MSH2 mutations: N127T, E198G, C199R, A305T, G322D, S323C, C333R L440P, D506Y, R524P, E562V, P622L, A636P, H639R, E647K, G674D, G692R, C697F; A834T. MSH6 mutations: S144I, S285I, Y556F, G566R, Q698E, D803G, Y850C, P1087T/R, T1225M. See. Supplementary information Fig.B for mapping to MutS. This mapping provides a basis for interpretation of additional functional studies that will be required for definitive classification as polymorphism or pathogenic mutation.