A Structure-Based Mechanism for DNA Entry into the Cohesin Ring

Abstract

Despite key roles in sister chromatid cohesion and chromosome organization, the mechanism by which cohesin rings are loaded onto DNA is still unknown. Here we combine biochemical approaches and cryoelectron microscopy (cryo-EM) to visualize a cohesin loading intermediate in which DNA is locked between two gates that lead into the cohesin ring. Building on this structural framework, we design experiments to establish the order of events during cohesin loading. In an initial step, DNA traverses an N-terminal kleisin gate that is first opened upon ATP binding and then closed as the cohesin loader locks the DNA against the ATPase gate. ATP hydrolysis will lead to ATPase gate opening to complete DNA entry. Whether DNA loading is successful or results in loop extrusion might be dictated by a conserved kleisin N-terminal tail that guides the DNA through the kleisin gate. Our results establish the molecular basis for cohesin loading onto DNA.

Keywords: ABC-ATPase; DNA loop extrusion; DNA-protein crosslink mass spectrometry; Mis4/Scc2/NIPBL; S. pombe; SMC complexes; chromosome segregation; cohesin; cryo-electron microscopy; sister chromatid cohesion.

Graphical abstract

Figure 1

Cohesin ATPase Head Engagement Leads to a DNA Gripping State (A) Schematic of purification and labeling of wild-type (WT) and Walker B mutant (EQ) cohesin to measure FRET between the Psm1 and Psm3 ATPase heads. Purified complexes were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) followed by Coomassie blue (CBB) staining or in-gel fluorescence detection. (B) Head FRET efficiencies of EQ-cohesin with the indicated additions were calculated by dividing the Alexa 647 intensity at its emission peak by the sum of Alexa 647 and Dy547 intensities. Results from three independent repeats of the experiment and their means and standard deviations are shown. (C) Head FRET efficiencies of WT cohesin in the presence of the Mis4-Ssl3 loader, a 3-kb circular plasmid DNA and the indicated nucleotides and phosphate analogs. Results from four independent repeats of the experiment and their means and standard deviations are shown. (D) Schematic of the DNA gripping experiment. Following incubation and washes, bound protein was analyzed by SDS-PAGE and immunoblotting, and the DNA was visualized by agarose gel electrophoresis. (E) Salt sensitivity of cohesin-DNA complexes following assembly with hydrolyzable or non-hydrolyzable ATP on linear DNA and DNA loops. Following incubation and washes, the products were analyzed as in (D). See Figure S1 for further characterization of cohesin’s DNA gripping state.

Figure 2

Overview Structure of Cohesin during Its Loading onto DNA (A) Schematic of EM sample preparation. The DNA gripping was separated by sucrose gradient centrifugation. The protein and DNA composition of each fraction were analyzed by SDS-PAGE followed by silver staining and agarose gel electrophoresis. Fractions 7 and 8 were used for EM analysis. (B) Superposed image of the negative staining 3D reconstruction and cryo-EM map of the cohesin core complex. (C) Two views of the 3.9-Å resolution cryo-EM map of the core complex with a transparent surface containing the atomic model (center) and a solid surface rendering (right). Three examples of secondary structure elements with resolved amino acidic side chains are shown. See also Figure S2, which documents the negative staining and cryo-EM data collection and image processing.

Figure 3

Molecular Mechanism of the Cohesin Loader (A) Psm3 and Mis4 topologically embrace DNA in the gripping state. Shown are an atomic model of Psm3, Mis4, and DNA built into the cryo-EM map (left) as well as Coulombic surface coloring for the protein component. Blue represents positively and red negatively charged amino acids. (B) Positively charged residues on the Mis4 surface, colored black, line the DNA path. The inset displays the cryo-EM map and atomic model to illustrate contacts made by R487 and R874 with the DNA. K873 and K877 are highlighted in gold. (C) Atomic model and cryo-EM map surrounding Psm3 K105 and K106 acetyl acceptor lysines, their orientation with respect to DNA, and a Mis4 acidic patch. (D) DNA gripping experiment comparing salt resistance of WT and acetyl acceptor lysine mutant (K105Q/K106Q) cohesin in the presence of ADP⋅BeF₃⁻. (E) Comparison of CLMS contacts between initial binding and the DNA gripping state. Crosslinks between Mis4 and Psm3 (golden lines) and between Mis4 and Rad21 (red lines) were mapped onto an expanded atomic model of the DNA gripping state. Insets show crosslinks between Rad21 and the Psm3 neck (blue lines). (F) Hypothetical sequence of ATP hydrolysis-controlled Mis4 conformational changes before and after gripping state formation. The behavior of the Rad21 N terminus is explored in Figure 7. See Figure S3 for additional analyses of the DNA gripping state.

Figure 4

A Hybrid Structural Model of the Cohesin Complex in the Gripping State (A) Atomic model of the cohesin core docked into the negative-stain EM envelope. An atomic model of the hinge and coiled coil is placed into the rod-shaped extension. The overall density accommodates a large portion of the Psm3 coiled coil, whereas parts of Psm1 remained invisible (dashed lines). The structure of Psc3 derived from multibody refinement of the cryo-EM structure is shown, and its variable positions are indicated. The likely position of Ssl3 bound to the Mis4 N terminus is indicated. (B) Protein crosslinks between the atomic models in the gripping state, supporting the assignments in (A). (C) Comparison of protein crosslinks within and between the SMC coiled coils in the initial binding and gripping state. See Figure S4 for supporting analyses of the hybrid structural model.

Figure 5

The Kleisin Path in the Gripping State (A) Crosslinks of Rad21 with Mis4 and Psc3 in the gripping state mapped onto their atomic models. Rad21 amino acids 360–431 are modeled based on the crystal structure of human SA2 bound to Rad21 (PDB: 4PK7). (B) Crosslinks from (A) mapped onto the structures of gripping state complex components, suggesting a likely kleisin path (red line). (C) Schematic of the kleisin circularization experiment. A gripping reaction was performed with Rad21 carrying N- and C-terminal CLIP and SNAP tags using a DNA loop substrate on beads. The CLIP and SNAP tags were covalently crosslinked by SC-Cy5. (D) In-gel Cy5 detection of the experiment in (C). SC-Cy5 was added to the input proteins or following gripping state assembly on the DNA beads. Beads were then washed with buffer or SDS (left). After the SDS wash, DNA beads were treated with PstI restriction endonuclease (center) or TEV protease (right), and bead-bound and supernatant fractions were analyzed. See Figure S5 for supporting information regarding the kleisin path.

Figure 6

The DNA Trajectory into the Cohesin Ring (A) Schematic of the DPC-MS workflow. See the main text for details. (B) A representative mass spectrum of a peptide containing a diazirine mass tag. The diagnostic 159u ion is highlighted. (C) DNA crosslinks of a 125-bp linear DNA in the gripping state, shown on the surface of the hybrid model (light blue), compared with crosslinks observed with a 3-kb circular plasmid DNA (medium blue). Crosslinks in common are shown in dark blue. (D) DNA crosslinks in the initial DNA binding state (light blue) are compared with those in the gripping state (medium blue); those in common are shown in dark blue. Arrowheads highlight crosslinks along the SMC coiled coils and hinge. (E) A model of the DNA trajectory from initial binding toward the gripping state, based on the observed DNA contacts. (F) DNA gripping experiment using head-head crosslinked cohesin. Psm1-SNAP Psm3-CLIP cohesin was treated with SC-Cy5 to close the head gate before the gripping reaction. DNA-bound proteins were analyzed by immunoblotting and in-gel Cy5 detection. See Figure S6 for additional DPC-MS results.

Figure 7

A Kleisin N-terminal Tail Guides DNA into the Cohesin Ring (A) Schematic of the kleisin N-gate FRET construct. FRET efficiencies at the kleisin N-gate were recorded under the indicated conditions using a 3-kb plasmid DNA as a substrate. ADP⋅BeF₃⁻ was used in the gripping incubation. Results from three independent repeats and their means and standard deviations are shown. (B) Sequence alignment of the cohesin N-tail. Positions of DNA crosslinking in the initial binding and the DNA gripping state are indicated. (C) Atomic model of the Rad21 N-tail (left), showing the conserved K10 and K11 residues relative to the DNA. A magnified view around K25 is also shown (right), including the cryo-EM density. (D) Comparison of WT and N17-cohesin in a DNA gripping experiment. Following reaction with a bead-bound DNA loop substrate and washes, the bead-associated proteins and DNA were analyzed by immunoblotting and gel electrophoresis. (E) Comparison of ATP hydrolysis by WT and N17-cohesin in the presence of the loader and a 3-kb plasmid DNA. Shown are the means and standard deviations from three independent experiments. (F) Loading of WT and N17-cohesin onto a 3-kb plasmid DNA. Following the loading reaction, cohesin was immunoprecipitated and washed with buffer containing 750 mM NaCl, and recovered DNA was analyzed by agarose gel electrophoresis. Shown are the means and standard deviations from three independent experiments. (G) A model for DNA entry into the cohesin ring. The kleisin N-tail guides DNA through the kleisin N-gate before DNA reaches the ATPase heads. ATP hydrolysis and passage through the head gate completes DNA entry. See Figure S7 for further analyses of the kleisin N-tail and Video S1 for an animated model of DNA entry into the cohesin ring.