DNA segment capture by Smc5/6 holocomplexes

Abstract

Three distinct structural maintenance of chromosomes (SMC) complexes facilitate chromosome folding and segregation in eukaryotes, presumably by DNA loop extrusion. How SMCs interact with DNA to extrude loops is not well understood. Among the SMC complexes, Smc5/6 has dedicated roles in DNA repair and preventing a buildup of aberrant DNA junctions. In the present study, we describe the reconstitution of ATP-dependent DNA loading by yeast Smc5/6 rings. Loading strictly requires the Nse5/6 subcomplex which opens the kleisin neck gate. We show that plasmid molecules are topologically entrapped in the kleisin and two SMC subcompartments, but not in the full SMC compartment. This is explained by the SMC compartment holding a looped DNA segment and by kleisin locking it in place when passing between the two flanks of the loop for neck-gate closure. Related segment capture events may provide the power stroke in subsequent DNA extrusion steps, possibly also in other SMC complexes, thus providing a unifying principle for DNA loading and extrusion.

Fig. 1. DNA entrapment in Smc5/6.

a, Schematic view of a supercoiled circular DNA substrate (left) bound to SK rings (with topological DNA entrapment, top; nontopological DNA entrapment, middle; no DNA entrapment, bottom) and of the Smc5/6 ring subunits (middle). The positions of the cysteines that were engineered for crosslinking of the hinge, Smc6–Nse4 and Smc5–Nse4 interfaces are indicated in green, purple and orange, respectively. Close-up views of the positions of cysteines are shown at the Smc6–Nse4 (top right) and Smc5–Nse4 (bottom right) interfaces. b, Two schematic representations of the SK compartment depending on the positioning of the Nse4 kleisin subunit. Three cysteine pairs—necessary to covalently close it—are indicated with colored handlebars. A cartoon showing the crosslinked complex with the colored compartment after denaturation is shown in the dashed box. Positions of hinge as well as N- and C-terminal ATPase domains are still shown as half-ovals and circles, respectively. c, Coisolation of crosslinked Smc5/6 proteins with plasmid DNA by agarose gel electrophoresis. The results were obtained with an octameric Smc5/6 holocomplex harboring cysteine pairs for crosslinking of the SK ring, comparing linear and circular DNA substrates. Traces of contaminating linear DNA from E. coli are marked with an asterisk. BsaI, BsaI restriction endonuclease. d, As in c but with titration of the protein complex and comparison of entrapment efficiencies between the wild-type (WT) and ATP-hydrolysis-deficient (EQ/EQ) complexes. Controls without either ATP or the Nse5/6 loader are also included. The plasmid substrate is largely supercoiled (sc), but a relaxed (Rel) form is also visible. Source data

Fig. 2. DNA entrapment in Smc5/6 subcompartments.

a, Schematic representation of subcompartments formed during ATP-dependent head engagement. The scheme on the left shows the complete Smc5/6 hexamer with the SK compartment highlighted as in Fig. 1. Combinations of cysteine pairs (colored handlebars as indicated) lead to covalent closure of K and S compartments, with the latter being split into upper (S^up) and lower (S^lo) compartments by Nse3 crosslinking as indicated. Note that the schemes on the right show only the crosslinked subunits that remain attached after denaturation. Schemes in dashed boxes indicate compartments after denaturation as in Fig. 1b. XL, Crosslink; E, Engaged. b, Coisolation of crosslinked Smc5/6 proteins with plasmid DNA by agarose gel electrophoresis. Results obtained with protein preparations harboring cysteine pairs for crosslinking of the K (top), S (second top), S^up (third top) or S^lo (bottom) compartments, as in Fig. 1d. Hex, Hexamer. c, Schematic representation of a DNA segment-capture state explaining the findings in b. Source data

Fig. 3. Opening of the neck gate by Nse5/6.

Crosslinking of purified Smc5/6 hexamers with cysteines at the Smc6–Nse4 interface is shown in the presence and absence of ligands. Detection of crosslinked species was by SDS–PAGE and Coomassie staining. Loss of crosslinking suggests that the gate was open. XL, Crosslink. Equivalent experiments for this interface containing ATPase head mutations as well as for other cysteine pairs are shown in Extended Data Fig. 6. Source data

Fig. 4. DNA clamping is essential for DNA entrapment.

Identification of putative DNA-binding residues in Smc5 and Smc6. a, The positions of the selected residues schematically displayed for the DNA-clamping state (Extended Data Fig. 7a). b, Positively charged residues on the Smc5 head mutated to alanine (A) or glutamate (E) in isolation or in combination as indicated. The mutant alleles were tested for function by plasmid shuffling. Counterselection against a pCEN(URA3) plasmid carrying a wild-type (WT) SMC5 allele by addition of 5-FOA revealed smc5 mutants resulting in a growth defect. c, As in b for residues in Smc6. d, Salt-stable DNA binding with wild-type and mutant Smc5/6 as determined by protein immobilization using a TwinStrep tag on Smc6 (bottom) and detection of coisolated plasmid DNA (‘recovered plasmid’) by agarose gel electrophoresis (top). e, DNA entrapment in the SK ring of wild-type, DNA-binding and ATP-hydrolysis-defective Smc5/6 mutants. Standard DNA entrapment salt conditions (150 mM NaCl) were used as in Fig. 1d (top) or buffer with reduced ionic strength (150 mM KOAc) as for the salt-stable DNA-binding assay in b. Source data

Fig. 5. DNA passes through the Smc6–Nse4 gate.

a, Coisolation of crosslinked Smc5/6 proteins with plasmid DNA by agarose gel electrophoresis. Results were obtained with protein preparations containing combinations of cysteine pairs and an Smc6–Nse4 fusion protein. Schematic drawings of crosslinked and denatured protein species are shown below the gel, with the closed compartments indicated. Different observed step sizes of DNA ladders are caused by differences in the size of the crosslinked protein species. XL, Crosslink. b, Schematic representation of DNA clamping in the presence of an Smc6–Nse4 fusion protein. The length of the linker allows it to wrap around the DNA strand clamped on ATP-engaged heads. c, As in a with Smc5/6 holocomplexes at 400 nM, and with or without post-entrapment opening of the linker using the human rhinovirus 3C protease. d, Coisolation of crosslinked Smc5/6 proteins with plasmid DNA by agarose gel electrophoresis, comparing a protein preparation harboring cysteine pairs for crosslinking of the SK ring with one in which the hinge cysteine pair is replaced by a SpyTag–SpyCatcher fusion. Topological DNA entrapment in Smc5/6 is not prevented by the hinge fusion, unlike what is shown for cohesin. Source data

Fig. 6. Smc5/6 entraps DNA in vivo.

The scheme on the left shows an outline of the procedure. Results from southern blotting on the right show that crosslinkable versions (6C) of both cohesin and Smc5/6 entrap the mini-chromosome (catenated monomers (CM)), but only cohesin does so in a cohesive manner (catenated dimers (CD)). Control strains lacking one of the ring cysteines (5C) abolish entrapment in both cases. Source data

Fig. 7. Model for chromosome loading and DNA translocation by alternating between a holding and a segment-capture state.

Schematic representation of the main findings presented in the present study. The segment-capture state (middle) is an essential intermediate in both DNA loading (left) and DNA translocation (right). Note that two possibilities for this state are indicated, differing in their degree of arm opening toward the hinge.

Extended Data Fig. 1. Schematics of Smc5/6 complex architecture.

(a) Overall architecture of the hexameric Smc5/6 core complex and the Nse5/6 dimer (adapted from). For simplicity the Nse5/6 complex is shown separately with multiple arrows denoting various contact points with the core complex. (b) Simplified schematics focusing only on changes in Smc5/6 dimer architecture during the ATPase cycle and upon binding to the Nse5/6 loader and the DNA substrate.

Extended Data Fig. 2. Analysis of Smc5/6 complexes harboring multiple engineered cysteines.

(a) Crosslinking of Smc5/6 hexamers harboring multiple engineered cysteines as indicated. Identification of covalently closed ring species and intermediary crosslinking products by protein gel analysis on 2 types of gels (left: 4–12 % (w/v) gradient gel, right: 3–8 % (w/v) gradient gel) for proper separation of small and large proteins and detection by Coomassie staining. Note the presence of small amounts of ring species in two of the three control samples lacking one of the six cysteines (lanes 5 and 7), indicating minor off-target crosslinking. (b) ATPase activity assays with selected hexameric Smc5/6 complexes relevant for this study. Combinations of certain modifications lead to a reduction of ATPase activity. Error bars show standard deviations from the mean for three technical replicates. (c) Control experiments for entrapment experiments shown in Fig. 1 with preparations lacking a selected cysteine (5C). Note that low levels of co-entrapment detected with two of the three 5C samples are likely explained by weak off-target crosslinking. (d) Under conditions promoting DNA entrapment the ring species is visible in a protein gel in the presence of a linear (lane 1) but not a circular (lane 2) DNA substrate, presumably due to co-retention of the ring species with circular DNA in the loading well (see scheme on the right). The ring species re-appears after digestion of the circular substrate (lane 3). (e) Time-dependence of DNA entrapment in the SK ring (top panel) and the K compartment (bottom panel). Crosslinker was added to a sample aliquot at the indicated time points. Source data

Extended Data Fig. 3. Schematic overview of assays used to examine the nature of Smc5/6 association with its DNA substrate.

(a) In the topological loading assay, a crosslinkable version of the Smc5/6 hexamer (‘6C’) is incubated with a small supercoiled (‘sc’) plasmid substrate in the absence or presence of ATP and Nse5/6. Upon crosslinking with BMOE and protein denaturation (‘SDS’) only fully crosslinked rings that were topologically associated with the substrate are retained, leading to a characteristic laddering pattern in agarose gels. (b) The salt-stable binding assay does not involve crosslinking. Complexes are first incubated with the substrate under low-salt conditions before the salt concentration is increased to 1 M NaCl by buffer changes. Smc5/6 complexes are immobilized on beads using a TwinStrep tag on Smc6, and co-purified plasmid substrate is examined on an agarose gel.

Extended Data Fig. 4. Analysis of Nse3 crosslinking to SMC arms.

Crosslinking of purified Smc5/6 hexamers with cysteines at the Smc6/Nse3 interface (a) or at the Smc5/Nse3 interface (b) in the presence and absence of ligands. Detection of crosslinked species by SDS-Page and Coomassie staining. Numbers below the gel indicate the efficiency of the respective crosslink. Source data

Extended Data Fig. 5. Head-distal Smc5/6 coiled-coil arms do not detectably open upon addition of Nse5/6, ATP, and/or DNA.

a) Model of an Smc5/Smc6/Nse2 complex obtained with AlphaFold-Multimer. The positions of two engineered cysteine pairs for inter-arm crosslinking are denoted. (b, c) Results from crosslinking experiments showing efficient arm alignment in all tested conditions. As in Fig. 3 and Supplementary Fig. 6. Source data

Extended Data Fig. 6. Crosslinking of selected SMC/kleisin interfaces.

Similar to Fig. 3(a) The Smc5/Nse4 interface is efficiently crosslinked and does not detectably respond to the presence of ligands apart from weak off-target crosslinking between Smc5 and the loader subunit Nse5 [see (B)]. (b) Control reaction for (A) with protein samples lacking the engineered cysteine in Nse4 confirming off-target crosslinking of Smc5(G1054C) to Nse5. (c) Addition of the loader reduces abundance of the ring species due to gate opening. The pattern in the presence of ligands (ATP and plasmid DNA) mirrors the one obtained with the Smc6/Nse4 interface (Fig. 3), except when loader, ATP and plasmid are added, presumably due to co-retention of ring species with circular DNA in the loading well (see scheme on the right). (d) As in Fig. 3 but with Smc5 and Smc6 subunits carrying the signature motif head-engagement mutation (‘SR’). (e) As in Fig. 3 but with Smc5 and Smc6 subunits carrying the Walker B ATP hydrolysis mutation (‘EQ’). (f) Fusion of Smc6 to Nse4 with the linker used in our in vitro assays is lethal in yeast. Plasmid shuffling assay as in Supplementary Fig. 7f, but with a deletion mutant for both smc6 and nse4. Adding back both wildtype genes separately, but not either of them alone or a fused version, restores viability. (g) Effect of mutating the head DNA binding interfaces on Smc5 and/or Smc6 on formation of a closed SMC/Kleisin (SK) ring. (h) Effect of mutating the head DNA binding interfaces on Smc5 and Smc6 on Smc6/Nse4 gate closure in the presence of the loader and various ligands. Source data

Extended Data Fig. 7. Identification of putative DNA binding residues on Smc6 and Smc5.

(a) Model of ATP-engaged Smc6/Smc5 heads with the Nse4 N-terminus bound to the Smc6 neck. AlphaFold-Multimer models of Smc6/Nse4 and Smc5 were superimposed on their counterparts of the cohesin complex (PDB: 6ZZ6, see panel B). The DNA molecule from the cohesin cryo-EM structure contacts several putative DNA binding residues on Smc6 and Smc5. (b) Similar view on top of engaged heads in the cohesin cryo-EM structure (PDB: 6ZZ6) with DNA-interacting residues on Smc3 and Smc1 indicated. Note that the Scc2 molecule is not shown for simplicity. (c) Sequence alignment showing strong evolutionary conservation of examined residues in Smc6. Smc5 residues show weaker conservation consistent with results of functional assays shown in Fig. 4. (d) Positively charged residues on the Smc5 head were mutated to alanine (‘A’) and the mutant alleles were tested for function by plasmid shuffling. As in Fig. 4b but with residues selected based on a recent cryo-EM structure of Smc5/6 (Yu et al.). (e) As in (D) but for residues on the Smc6 head. (f) Positively charged residues on Smc5 and Smc6 heads were mutated to alanines in combinations. As in Fig. 4b and c. Source data

Extended Data Fig. 8. Entrapment assays involving Smc6-Nse4 fusion and BMOE crosslinking of interfaces.

(a) Schematic representation of complexes used for experiments in Fig. 5 and Supplementary Fig. 8d. The ‘Nse4^open’ complex has a split kleisin and thus has its SK ring permanently opened. The ‘Smc6-Nse4 fusion’ complex contains a peptide linker fusing the C-terminus of Smc6 with the N-terminus of Nse4. The linker contains a recognition site for the HRV 3C protease and can thus be opened by cleavage. (b) Schematic overview of entrapment in the absence and presence of the Smc6-Nse4 fusion. Without the fusion (top) during initial DNA segment capture the Smc6/Nse4 gate is opened by Nse5/6 (Smc5 and Smc6 arm distance is exaggerated for clarity). Upon gate closure and ATP hydrolysis, one of the loop strands is lost after escape between the disengaged heads, and the other strand becomes topologically entrapped in the SK ring. The green-filled area indicates the lumen of the ring compartment (SK) which is maintained even after protein denaturation due to crosslinking. (bottom) Scenarios for DNA entrapment by complexes with Smc6-Nse4 fusion (top row, middle panel) in combination with different cysteine pairs for crosslinking (other panels). Top row, left panel: Upon crosslinking of Smc5/Smc6 and Smc5/Nse4 interfaces an SK-plus-linker joint compartment becomes denaturation-resistant (orange-filled area) leading to DNA loop entrapment (not detected in this assay) rather than DNA entrapment. Top row, right panel: Crosslinking of only the Smc6/Nse4 interface leads to a denaturation-resistant linker compartment (blue-filled area) that topologically entraps a DNA strand. The SK compartment is also formed and entraps the other DNA strand, but it is sensitive to denaturation (green-dashed area). Bottom row: Crosslinking of all three interfaces entraps one DNA strand in a denaturation-resistant SK compartment (green-filled area) and another in the linker compartment (blue-filled area), only the latter of which can be released by incubation with 3C protease. Incompletely crosslinked rings lacking the Smc5/Smc6 or Smc5/Nse4 crosslinks (or both) entrap only the DNA strand in the linker compartment (in square bracket). (c) As in Fig. 5c but with cleavage of the linker prior to mixing of samples for DNA entrapment. (d) Salt-stable DNA binding of Smc5/6 complexes with a split Nse4 protein (‘Nse4 open’) or linked Smc6 and Nse4 proteins (‘Smc6-Nse4’ fusion protein). As in Fig. 3d. Pre-treatment with 3C protease cleaves the fusion linker peptide but does not alter DNA binding. Asterisks denote unspecific degradation products. (e) Insertion of a Spy-tag and Spy-Catcher into the Smc5 and Smc6 hinge-domains (see scheme on the left), respectively, leads to permanent, covalent fusion of the hinge domains as shown in the protein gel on the right. Source data