cryo-EM structures of the E. coli replicative DNA polymerase reveal its dynamic interactions with the DNA sliding clamp, exonuclease and τ.

Materials All chemicals and oligonucleotides were purchased from Sigma-Aldrich (Gillingham, United Kingdom) and chromatography columns from GE healthcare (Little Chalfont, United Kingdom).

Protein expression and purification

To increase binding to the clamp, amino acid residues 920–924 of E. coli PolIIIα were changed by site directed mutagenesis from QADMF to QLDLF, while in the exonuclease residues 182–187 were changed from QTSMAF to QLSLPL, based on sequences described in ( Wijffels et al., 2004 ). All proteins were expressed in E. coli (DE3) BL21. PolIIIα, clamp and exonuclease were expressed and purified as described before ( Toste Rêgo et al., 2013 ). τ 500 was purified by Histrap HP column, Resource S column, and a Superdex 75 gel filtration column. His-tags were removed by proteolytic cleavage with human rhinovirus 3C protease. Proteins were flash frozen in liquid nitrogen and stored at -80°.

Gel filtration analysis

Proteins were analyzed by gel filtration using a 2.4 mL Superdex 200 Increase column (GE healthcare) in 25 mM Hepes pH 7.5, 150 mM NaCl, and 2 mM DTT. PolIIIα-clamp-exonuclease complex was assembled at 10, 1, and 0.1 μM and 50 μL injected onto the column. Electro mobility shift assay A DNA substrate identical to the substrate used for the cryo-EM samples was used, with the exception of a 6-carboxyfluorescein (6-FAM) at the 5' end of the primer strand and a phosphorothioate link at the 3' terminal bond to prevent exonuclease digestion. 5 nM DNA was incubated with 2.5 μM polymerase ( E. coli Pol I (Klenow fragment), Pol II, Pol IIIα, or Pol IV) for 10 min at room temperature. Reaction mixtures contained 20 mM Tris pH 7.5, 4% glycerol, 5 mM DTT, 40 μg/ml BSA, and 40 mM Potassium Glutamate. Half of the sample was separated on a native 6% acrylamide gel and imaged on a Typhoon laser scanner (GE Healthcare). The remaining half of the sample was analyzed on a denaturing 4–12% SDS acrylamide gel and stained with Coomassie blue. Sample preparation for cryo-EM The PolIIIα-clamp-exonuclease-τ 500 protein complex was assembled from the individual components to a final concentration of 15 μM and purified on a 2.4 mL Superdex 200 Increase gel filtration column (GE Healthcare) in 25 mM Hepes pH 7.5, 50 mM Potassium Glutamate, 3 mM Magnesium Acetate, and 2 mM DTT. The peak fraction (∼4 μM) was retrieved and incubated for 5 min with 20 μM of a 25 bp DNA substrate with a 4 nt overhang (template: 5′-TCAGGAGTCCTTCGTCCTAGTACTACTCC-3′, primer: 5′-GGAGTAGTACTAGGACGAAGGACTC-3′) for 5 min at room temperature. Subsequently, 0.1 volume of 0.05% (V/V) Tween 20 was added and incubated for another 5 min before the samples were pipetted onto glow-discharged holey carbon cryo-EM grids (Quantifoil Cu R1.2/1.3), and frozen in liquid ethane using a Vitrobot (FEI, Hillsboro, OR).

Data collection and image processing

All data was collected using a Titan Krios electron microscope (FEI) operated at 300 kV equipped with a K2 summit direct electron detector (Gatan, Pleasaston, CA). Although this detector was mounted after a Gatan Imaging Filter (GIF), the filter was not used to remove any inelastic scattering. Images were collected in single-electron counting mode at a calibrated magnification of 28.571x (1.76 Å/pixel), using a flux of 2 e/Å 2 /sec and a total dose of 40 e/Å 2 over a total of 20 frames. Frames were aligned and averaged using whole-image movement correction using MOTIONCORR ( Li et al., 2013 ). Contrast transfer function parameters were calculated using CTFFIND3 ( Mindell and Grigorieff, 2003 ). All subsequent particle picking and data processing was performed using Relion-1.3 ( Scheres, 2012 ), with the exception of the generation of the initial model, which was done using Eman2 ( Tang et al., 2007 ). A total of 1350 micrographs were recorded from which >550,000 particles were picked automatically in Relion. After 2D classification, a large number of spurious particles as well as particles that show free polymerase or free clamp were removed, yielding a dataset of ∼90,000 particles. After 3D classification a another ∼27,000 were removed to yield a final dataset of 63,215 particles. From these, six 3D classes were calculated that were subsequently merged into the final three 3D classes of 'DNA-free' (16,970 particles), 'DNA-bound' (5663 particles) and 'DNA-bound, no tail' (40,582 particles). Particle-based movement correction and per-frame B-factor weighting to account for radiation damage and unresolved particle movement was performed in the later stages of refinement using the particle polishing option in Relion ( Scheres, 2014 ). Reported resolutions are based on the gold-standard FSC-0.143 criterion ( Scheres and Chen, 2012 ) and FSC-curves were corrected for the convolution effects of a soft mask using high-resolution noise-substitution ( Chen et al., 2013 ). All density maps were sharpened by applying a negative B-factor that was estimated using automated procedures ( Rosenthal and Henderson, 2003 ). We believe that the resolution of these reconstructions is limited by both the relatively small size of the complex (250 kDa), which hampers accurate alignment and classification, and the inherent flexibility of this four-protein and DNA complex. Still, the maps are of excellent quality, with individual helices, β-sheets, and loops clearly visible in the map ( Figure 1C ). Fitting of the crystal structures into the cryo-EM map Individual crystal or NMR structures were manually placed into the cryo-EM map in PyMOL ( Schrödinger, LLC 2010 ) and subsequently rigid-body fitted into the density using Coot ( Emsley et al., 2010 ). PDB codes of the fitted structures are: PolIIIα: 2HNH ( Lamers et al., 2006 ), clamp: 2POL ( Kong et al., 1992 ), exonuclease: 1J54 ( Hamdan, et al., 2002 ), τ 500 : 2AYA ( Su et al., 2007 ). The C-terminal tail of Eco PolIIIα that is lacking in the crystal structure (2HNH) was modeled as described in ( Toste Rêgo et al., 2013 ). The PolIIIα structure was divided into five domains that were further fitted independently into density as rigid bodies (see Figure 1—figure supplement 3B,C ). These domains were: PHP (residues 1–280), palm-fingers (residues 281–432 + 510–810), thumb (residues 433–509), tip-of-fingers (residues 811–928) and C-terminal tail (residues 929–1160). Clamp binding motifs of PolIIIα and exonuclease were manually built into the clamp in Coot guided by the crystal structures of clamp-bound peptides from Pol II and Pol IV, ( Bunting et al., 2003 ; Georgescu et al., 2008b ; Jeruzalmi et al., 2001 ). The DNA substrate was generated with Coot, and the last four base pairs of the DNA were adjusted guided by the DNA from Taq Pol IIIα ( Wing et al., 2008 ). Comparison of DNA polymerase structures The following crystal structures of C family DNA polymerases were used to compare DNA binding and τ binding.

DNA bound Taq PolIIIα

(PDB code: 3E0D [ Wing et al., 2008 ]), τ bound Taq PolIIIα (PDB code: 4IQJ [ Liu et al., 2013 ]), DNA bound G. kaustophilus PolC (PDB code: 3F2B [ Evans et al., 2008 ]). Crystal structures of bacterial DNA polymerases in complex with DNA were used to compare the distance between the polymerase active site and the opening to the clamp. The following structures were used: T. aquaticus DNA Pol I (PDB code: 1QTM [ Li et al., 1999 ]), E. coli Pol II (PDB code: 3K57 [ Wang and Yang, 2009 ]), E. coli PolIIIα (this work), and E. coli Pol IV (PDB code: 4IRD [ Sharma et al., 2013 ]). For the structures of Pol I, Pol II, and Pol IV, the sliding clamp (PDB code: 2POL [ Kong et al., 1992 ]) was manually placed close to the clamp binding sequences in the different polymerases, taking care not to cause any clashes with other parts of the polymerase.

Materials All chemicals and oligonucleotides were purchased from Sigma-Aldrich (Gillingham, United Kingdom) and chromatography columns from GE healthcare (Little Chalfont, United Kingdom).