Experimental Procedures

Protein Preparation
Clones of psNUC, stNUC, and saNUC (Uniprot accession numbers: A5WF35, A7J283, and A3EZR7, respectively) were synthesized by GeneART and subcloned into pGEX 6P-1. Human, mouse, and Pseudomonas aeruginosa FAN1 clones in pGEX 6P-1 were prepared by the DNA-cloning service at the University of Dundee. Proteins were expressed as glutathione S-transferase fusions in E. coli and purified by means of glutathione-Sepharose affinity chromatography. Proteins were cleaved off the affinity column with 3C protease and applied to Superdex 75 or 200 gel filtration in 300 mM NaCl, 20 mM Tris (pH 8), and 0.5 mM tris(2-carboxyethyl)phosphine (TCEP), depending on the target protein size. FAN1 protein purification required an additional heparin step to remove DNA contamination prior to gel filtration.

Crystallography
Crystals were obtained by the sitting drop vapor diffusion method using the Morpheus crystallization screen (Molecular Dimensions; Supplemental Information). Diffraction data were collected at 100 K on beamline IO4 at the Diamond Light Source and processed with Denzo/Scalepack (HKL Research). The structures of selenomethionine containing derivatives of stNUC and psNUC were determined by automated SAD/model building using PHENIX (Adams et al., 2010). saNUC was solved by molecular replacement with PHENIX using stNUC as a search model. Models were improved and refined using Coot, Refmac, and PHENIX.

Activity Assays
DNA substrates were prepared as in MacKay et al. (2010) with the exception of Jbm5, prepared as in Kvaratskhelia et al. (2001). Purified recombinant proteins were preincubated with the radiolabelled substrate in binding buffer (Supplemental Information) at the reaction temperature for 10 min to allow binding to occur. The reaction was started by the addition of divalent metal ions and stopped by mixing with a 2-fold excess of EDTA. After addition of 66% formamide, the samples were heat denatured and analyzed by denaturing PAGE (12% polyacrylamide and 8 M urea in Tris-borate-EDTA [TBE] buffer). Gels were dried, exposed to storage phosphor screens, quantified by a Typhoon FLA 9500 (GE Healthcare) phosphorimager, and analyzed with the ImageQuant software.

Gel Mobility Shift Assays
Varying amounts of purified recombinant protein were incubated with 1 nM radiolabelled substrate at 20°C for 30 min in binding buffer (Supplemental Information). Following addition of 2.5% Ficoll-400, samples were run on 8% native polyacrylamide gels in TBE buffer. Gels were dried, exposed to storage phosphor screens, quantified by a Typhoon FLA 9500 (GE Healthcare) phosphorimager, and analyzed with the ImageQuant software.

Sedimentation-Velocity AUC
Sedimentation-velocity experiments were performed in a Beckman Optima Xli analytical ultracentrifuge, using aluminum double-sector sapphire cells in an An-50 Ti rotor. The rotor speed was 50,000 rpm, and the temperature was maintained at 277 K. Prior to centrifugation, protein samples were dialyzed exhaustively against the buffer blank (50 mM Tris/HCl [pH 8.0], 300 mM NaCl, and 0.50 mM TCEP). Interference images were collected every 180 s during the sedimentation run. The data recorded from moving boundaries were analyzed by the program SEDFIT in terms of both discrete species and continuous distribution function of sedimentation coefficient (c(s)) and molar mass (c(M)), allowing determinations of sedimentation coefficient and shape-independent molecular weight (Brown and Schuck, 2006).

Model Building
The FAN1 VRR-Nuc domains (human, mouse, and Pseudomonas) have the same predicted secondary structure using PSIPRED (McGuffin et al., 2000) and therefore likely the same topology as standalone domains except for an insertion consisting of two confidently predicted helices. Twenty models of hFAN1 and pFAN1 were initially built based on the combined crystal structures of the archaeal HJR, HJC (PDB code 1GEF; Nishino et al., 2001), and the SdaI restriction endonuclease (PDB code 2IXS; Tamulaitiene et al., 2006) using the “modeler” option (Sali and Blundell, 1993) in QUANTA (Accelrys). The different models were calculated by varying the initial model and optimizing the objective function using conjugate gradients and molecular dynamics with simulated annealing using the CHARMm force field. The models with the lowest objective function were chosen for further refinement. This consisted of introducing the two helices insertion using the manual modeling options in QUANTA, taking care to ensure proper packing of hydrophobic residues.