DNA Entry into and Exit out of the Cohesin Ring by an Interlocking Gate Mechanism

Abstract

Structural maintenance of chromosome (SMC) complexes are proteinaceous rings that embrace DNA to enable vital chromosomal functions. The ring is formed by two SMC subunits, closed at a pair of ATPase heads, whose interaction is reinforced by a kleisin subunit. Using biochemical analysis of fission-yeast cohesin, we find that a similar series of events facilitates both topological entrapment and release of DNA. DNA-sensing lysines trigger ATP hydrolysis to open the SMC head interface, whereas the Wapl subunit disengages kleisin, but only after ATP rebinds. This suggests an interlocking gate mechanism for DNA transport both into and out of the cohesin ring. The entry direction is facilitated by a cohesin loader that appears to fold cohesin to expose the DNA sensor. Our results provide a model for dynamic DNA binding by all members of the SMC family and explain how lysine acetylation of cohesin establishes enduring sister chromatid cohesion.

Graphical abstract

Figure 1

Fission Yeast Pds5 and Wapl and Their Effect on Cohesin Loading (A) Purification schemes for recombinant Pds5 and Wapl and analysis of the purified proteins by SDS-PAGE and Coomassie blue staining. (B) Pds5 forms a complex with Wapl. Purified Pds5 and Wapl were mixed and analyzed by gel filtration. Fractions were analyzed by SDS-PAGE, and proteins visualized by silver staining. The elution positions of two size markers (aldolase and conalbumin) are indicated. (C) Pds5 and Wapl interact with the cohesin complex. Recombinant cohesin tetramer (Psm1, Psm3, Rad21, and Psc3, “4”) or trimer (Psm1, Psm3, and Rad21, “3”) complexes were immunoprecipitated in the presence or absence of Pds5 and/or Wapl. Bound proteins were analyzed by western blotting. 1/20 of the input and Rad21-HA immunoprecipitates are shown. (D) Wapl stimulates loading of the cohesin holocomplex by Mis4-Ssl3. Schematic of the in vitro cohesin-loading assay and a gel image of the recovered DNA. All reactions contained cohesin, Mis4-Ssl3, plus the indicated proteins and were incubated for 15 min. The graph shows means and standard deviations from at least three independent experiments. See also Figure S1 for an analysis of how Pds5 counteracts loading.

Figure 2

Pds5-Wapl Promotes Cohesin Unloading from DNA (A) Schematic of the in vitro cohesin-unloading assay. (B) The gel and graph show DNA recovered before (0 min) and after (60 min) the second unloading incubation, in which the indicated proteins were included. See also Figures S2A–S2C for indication that DNA and protein remained intact during the incubation. (C) Time-course analysis of the unloading reaction, including the indicated protein concentrations. (D) Pds5 counteracts spontaneous cohesin unloading. In vitro unloading reactions were carried out in the presence of the indicated concentrations of Pds5. All graphs show means and standard deviations from at least three independent experiments. See also Figure S2D, which shows that Mis4-Ssl3 has no detectable role during unloading.

Figure 3

ATP Hydrolysis Dependence of Cohesin Unloading (A) Cohesin-unloading reactions were performed in the presence of the indicated compounds: “E,” EDTA; “A,” ATP; “γS,” ATP-γ-S. The graph shows means and standard deviations from three independent experiments. (B) Effect of the Walker B mutations on the cohesin ATPase. The indicated cohesin tetramer complexes were incubated with or without Mis4-Ssl3 and RC-DNA. ATP hydrolysis was assayed and quantified by thin-layer chromatography. “1B” and “3B” denote Psm1(E1161Q) and Psm3(E1128Q) mutant cohesin, respectively. (C) Walker B mutant cohesin complexes are DNA-loading proficient. Loading assays were performed and quantified with the indicated cohesin complexes and cofactors. Means and standard deviations from three independent experiments are shown. See also Figures S3A–S3C for the various purified cohesin complexes, a time course of loading, and confirmation of the topological nature of DNA binding by Walker B mutant cohesin. (D) Walker B mutant cohesin loading remains ATP hydrolysis dependent. The loading assays were carried out with or without ATP in the presence of additional ATP (“A”) or ATPγS (“γS”). The quantification denotes mean and errors derived from two experiments. Addition of ATPγS on top of ATP was used to confirm that ATPγS does in fact bind to and inhibit the SMC ATPase. (E) Walker B mutant cohesin is unloading defective. The indicated cohesin complexes were loaded onto DNA (time “0”), and comparable amounts used as substrate for a 30 min unloading incubation with or without Pds5-Wapl. Shown are means and standard deviations from three independent experiments. See also Figures S3D–S3F for the differential effect of divalent cations on cohesin’s ATPase, loading and unloading.

Figure 4

Wapl Opens the Rad21N-Psm3 Ring Interface (A) Pds5 and Wapl do not affect the cohesin ATPase. ATP hydrolysis was monitored in the presence of the indicated proteins and RC-DNA. Pds5 and Wapl by themselves did not show detectable ATP hydrolysis (data not shown). Means and standard deviations from three independent experiments are shown. (B) Schematic and results of the “Rad21N dissociation” experiment. Cohesin carrying TEV-cleavable Rad21, or wild-type cohesin as control, was incubated with Pds5-Wapl, in presence or absence of ATP (T), ADP (D), or ATP-γ-S (γS), then Rad21 was cut by TEV protease. The resultant cohesin was recovered by immunoprecipitation, and the Rad21 subunit and its two cleavage products were monitored by western blotting. See also Figure S4 for the full image of the western blot, an experiment examining the activity of Mis4-Ssl3 in this assay, and demonstration of DNA binding by Pds5, Wapl, and Psc3. (C) Schematic and results of the “DNA release” experiment in which cohesin-DNA complexes are first isolated on magnetic beads. The unloading reaction was then initiated in the presence of the indicated components. The gel image shows the supernatant “S” and bead “B” fractions following the unloading incubation. The percentage of released DNA was quantified.

Figure 5

Pds5-Wapl Is both Loader and Unloader (A) Cohesin-loading reactions (15 min) were carried out without the Mis4-Ssl3 cohesin loader, but including Pds5-Wapl. Reactions omitting cohesin or ATP are shown as controls. See also Figure S5A for confirmation of topological cohesin loading by Pds5-Wapl. (B) Comparison of cohesin loading over time in the presence of Mis4-Ssl3, Pds5-Wapl, or both. See also Figures S5B–S5D for titration of Mis4-Ssl3 and Pds5-Wapl and illustration that Pds5-Wapl-dependent reactions require Psc3. (C) Following loading by Pds5 and Wapl for 30 min, cohesin unloading was initiated by increasing the salt concentration to 150 mM NaCl in the presence of the indicated compounds: “E,” EDTA; “A,” ATP; “γS,” ATPγS. (D) Cohesin-loading reactions were performed for 15 min with indicated loaders and adenosine derivatives. (E) Schematic and quantification of DNA released from cohesin rings following TEV-protease cleavage of either Psm3 (3 TEV) or Rad21 (21 TEV) after cohesin had been loaded by either Mis4-Ssl3 or Pds5-Wapl. See also Figures S5E–S5G for controls for loading and cleavage of the two TEV-cleavable cohesin complexes.

Figure 6

Psm3 Acetylation Acceptor Lysines are DNA Sensors during Loading and Unloading (A) Cohesin-loading reactions were performed using wild-type (WT), Psm3K106Q (KQ), and Psm3K105Q,K106Q (KKQQ) cohesin tetramer complexes in the presence of the indicated loading factors. See also Figure S3A for the purification of the mutant cohesin complexes. (B) ATPase activity of the indicated cohesin tetramers with or without added Mis4-Ssl3 and/or DNA. (C) Wild-type and KQ cohesin was loaded onto DNA, and equivalent concentrations of cohesin-DNA complexes were used as substrate in unloading reactions initiated by Pds5-Wapl (PW). Retained DNA was quantified over time; the means and standard deviations from three independent experiments are shown.

Figure 7

Cohesin Hinge Interactions and a Unified Model for Cohesin Loading and Unloading (A) Purification of a Psm1-Psm3 hinge dimer. See also Figure S6A, which shows DNA binding by the Psm1-Psm3 hinge. (B) The Psm1-Psm3 hinge interacts with Mis4-Ssl3. Co-immunoprecipitation of the cohesin loader, or Pds5 as control, with the cohesin hinge was analyzed by western blotting. (C) A supramolecular complex between the Psm1-Psm3 hinge, Psc3, and the cohesin loader. Psc3 was immunoprecipitated, and co-precipitation of the cohesin hinge and the cohesin loader were analyzed. (D) A unified model for DNA entry and exit into and out of the cohesin ring. See Discussion for details. See also Figure S6B for alternative models for DNA entry into the cohesin ring.

Figure S1

Pds5 Antagonizes Mis4-Ssl3 Interaction with Cohesin, Related to Figure 1 (A) Cohesin’s ATPase activity was monitored in the presence of Mis4-Ssl3, DNA, and the indicated proteins. Shown are the means and standard deviations from three independent experiments. (B) Co-immunoprecipitation of cohesin (30 nM) and Mis4-Ssl3 (50 nM) were analyzed in the presence of additional “+,” 50 nM and “++,” 100 nM Pds5 protein.

Figure S2

Characterization of Pds5-Wapl-Mediated Cohesin Unloading, Related to Figure 2 (A–C) The total DNA contained in the reaction was visualized by agarose-gel electrophoresis (A), and cohesin subunits were analyzed by SDS-PAGE and silver staining or western blotting, respectively, following the cohesin-unloading incubation (B and C). This confirmed that all macromolecular components remained intact. (D) Mis4-Ssl3 is dispensable for Pds5- and Wapl-dependent cohesin unloading. Cohesin-DNA complexes were recovered by anti-Pk-immunoprecipitation of the Psm3 subunit. Western blotting revealed that Mis4-Ssl3 is not part of the cohesin-DNA complex retrieved by immunoprecipitation. Then a DNA release assay was carried out in the presence of the indicated components (100 nM Psc3, with or without 400 nM of Pds5 and Wapl and 100 nM Mis4-Ssl3). DNA released into the supernatant (“S”) is shown next to the DNA that remained bound to the beads (“B”), the percentage of released DNA was quantified.

Figure S3

The Role of ATP Hydrolysis for Cohesin Loading and Unloading, Related to Figure 3 (A) Purifications of wild-type and Walker B mutant (1B, 3B, or 1B3B), as well as lysine 105 and 106 mutant (3KQ or 3KKQQ) cohesin tetramer complexes and a cohesin tetramer containing TEV protease cleavable Psm3 (3TEV). Proteins from the final gel-filtration step were analyzed by SDS-PAGE and Coomassie blue staining. (B) Time course of the cohesin-loading reaction, comparing wild-type and double Walker B mutant (1B3B) cohesin complexes. (C) Confirmation that Walker B mutant cohesin complexes topologically entrap DNA. A schematic of the reaction is shown. DNA-cohesin complexes were immobilized on magnetic beads, then bound cccDNA was linearized by restriction enzyme PstI treatment, which cuts once in pBluescript. Release of linear DNA into the supernatant (“S”) or its retention on the beads (“B”) was analyzed by gel electrophoresis. The gel image shows a typical result of the experiment. (D) Little impact of different divalent cations on cohesin loading. Cohesin-loading reactions were carried out with the indicated divalent cations (1 mM) and Mis4-Ssl3. (E) Inhibition of the cohesin ATPase by Ca²⁺ or Mn²⁺. The effect of the indicated divalent cations on cohesin’s ATPase activity, in the presence of Mis4-Ssl3 and DNA, was measured. Panels (B), (D), and (E) report means and standard deviations from three independent experiments. (F) Ca²⁺ inhibits cohesin unloading by Pds5 and Wapl. A cohesin-unloading reaction was carried out in the presence of Ca²⁺ instead of Mg²⁺.

Figure S4

Pds5-Wapl Facilitates Psm3-Rad21 Interface Opening, Related to Figure 4 (A) The original western blot images from Figure 4B. Red rectangles denote the selected areas. (B) Comparison of Pds5-Wapl and Mis4-Ssl3 in the Rad21-N dissociation assay. Cohesin carrying TEV-cleavable Rad21 was incubated with Mis4-Ssl3 (“MS”) or Pds5-Wapl (“PW”), in presence or absence of DNA and/or ATP, then Rad21 was cut by TEV protease. The resultant cohesin was recovered by immunoprecipitation and the Rad21 subunit and its two cleavage products were monitored by western blotting. (C) Pds5 and Wapl are DNA-binding proteins. The indicated concentrations of Pds5 and/or Wapl were mixed with a 3.0 kb linear double-stranded DNA (lds DNA) substrate and incubated at 32°C for 10 min. Psc3 was similarly analyzed. The DNA-protein complexes were analyzed by 0.8% agarose gel electrophoresis.

Figure S5

Control Experiments for Pds5-Wapl-Dependent Cohesin Loading, Related to Figure 5 (A) Confirmation of topological cohesin loading by Pds5-Wapl. Cohesin-loading reactions were carried out with Pds5-Wapl for 15 min, then bound DNA was digested and analyzed as described in Figure S3C. (B) Stimulation of cohesin loading by Pds5-Wapl. Cohesin-loading reactions were carried out including the indicated concentrations of Mis4-Ssl3 and Pds5-Wapl. The band intensities were quantified. (C and D) Psc3 is required for both Pds5- and Wapl-dependent cohesin loading and unloading, respectively. The DNA loading and release assays were carried out using a purified cohesin trimer (Murayama and Uhlmann, 2014), devoid of the Psc3 subunit, supplemented with the indicated components. Means and standard deviations from three independent experiments are shown; the reactions with Mis4-Ssl3 were included twice. (E) TEV-protease cleavage of Psm3-TEV (3) and Rad21-TEV (21) cohesin complexes. The various cohesin complexes were incubated with TEV protease at 16°C for 1.5 hr, then analyzed by SDS-PAGE and Coomassie staining. The wild-type cohesin tetramer (W) served as a comparison. A size marker is shown and the band identities are indicated (N, M, and C indicate N-terminal, middle, and C-terminal protein fragments, respectively). (F) Confirmation of comparable loading of the cohesin complexes shown in (E) onto DNA using Mis4-Ssl3 or Pds5-Wapl. (G) Western blot analysis of Rad21 and Psm3 cleavage during the experiment shown in Figure 5E. Aliquots (0.5 μl out of 15 μl) of the immunoprecipitate were analyzed by SDS-PAGE followed by western blotting against Rad21 or against the C-terminal Pk epitope on Psm3. Fragment annotation is as in (E).

Figure S6

DNA Binding by the Psm1-Psm3 Hinge and Alternative Possibilities for DNA Entry Mediated by Mis4-Ssl3 or Pds5-Wapl, Related to Figure 7 (A) The Psm1-Psm3 hinge contains DNA-binding activity. The indicated concentrations of Psm1-Psm3 hinge were mixed with a 3.0 kb linear double-stranded DNA (lds DNA) substrate and incubated at 32°C for 10 min. The DNA-protein complexes were analyzed by 0.8% agarose-gel electrophoresis. (B) Alternative possibilities for DNA entry mediated by Mis4-Ssl3 or Pds5-Wapl. See the Discussion in the main text for details.