Gradual Opening of Smc Arms in Prokaryotic Condensin

1 Gradual opening of Smc arms in prokaryotic condensin

2 3 Roberto Vazquez Nunez1,#, Yevhen Polyhach2,#, Young-Min Soh1, Gunnar Jeschke2, Stephan 4 Gruber1* 5 6 1Department of Fundamental Microbiology, University of Lausanne, Bâtiment Biophore, 1015 7 Lausanne, Switzerland 8 2Laboratory of Physical Chemistry, ETH Zurich, Vladimir-Prelog-Weg 2, 8093 Zurich, 9 Switzerland 10 11 12 13 14 15 Running title: Shapes of B. subtilis SMC condensin. 16 17 Keywords: Chromosome condensation, Condensin, SMC, kleisin, Smc/ScpAB, Cohesin, DNA 18 loop extrusion, DNA segment capture. 19 20 #Equal contribution 21 22 *Corresponding author, Lead Contact, E-mail: S.G. [email protected];

1 Summary 2 3 Multi-subunit SMC ATPases control chromosome superstructure apparently by catalyzing a 4 DNA-loop-extrusion reaction. SMC proteins harbor an ABC-type ATPase ‘head’ and a ‘hinge’ 5 dimerization domain connected by a coiled coil ‘arm‘. Two arms in a SMC dimer can co-align, 6 thereby forming a rod-shaped particle. Upon ATP binding, SMC heads engage, and arms are 7 thought to separate. Here, we studied the shape of B. subtilis Smc-ScpAB by electron-spin 8 resonance spectroscopy. Arm separation was readily detected proximal to the heads in the 9 absence of ligands, while separation near the hinge largely depended on ATP and DNA. 10 Artificial blockage of arm opening eliminated DNA stimulation of ATP hydrolysis, but did not 11 prevent basal ATPase activity. We identified an arm-to-arm contact as being important for 12 controlling the molecular transformations. Point mutations at this arm interface eliminate 13 Smc function. We propose that partially open, intermediary conformations provide 14 directionality to SMC DNA translocation by binding suitable DNA substrates. 15

1 Introduction 2 3 SMC protein complexes are ancient enzymes with a unique architecture that organize 4 chromosomal DNA molecules, presumably by DNA loop extrusion (Yatskevich et al., 2019). In 5 eukaryotes, cohesin folds DNA into loop domains to regulate gene expression and to direct 6 DNA recombination (Ba et al., 2020; Merkenschlager and Nora, 2016). By a distinct 7 mechanism, cohesin also holds sister chromatids together (Yatskevich et al., 2019). In mitosis, 8 condensin folds DNA into a series of loops that are dynamically anchored along a chromatid 9 axis thus supporting chromosome condensation and sister chromatid resolution (Earnshaw 10 and Laemmli, 1983; Gibcus et al., 2018; Marsden and Laemmli, 1979; Naumova et al., 2013). 11 The essential functions of another relative, the Smc5/6 complex, are less well understood 12 (Aragón, 2018). 13 In many bacteria, Smc-ScpAB complexes initiate a loop-extrusion-type activity at one or few 14 selected starting points that are defined by 16 bp parS DNA sequences. parS sites are located 15 in the replication origin region. They recruit the clamp-like CTP-binding protein ParB (Jalal and 16 Le, 2020; Osorio-Valeriano et al., 2019; Soh et al., 2019), which in turn promotes the loading 17 of Smc-ScpAB complexes onto the chromosome (Gruber and Errington, 2009; Minnen et al., 18 2016; Sullivan et al., 2009; Wilhelm et al., 2015). Bidirectional translocation of Smc-ScpAB 19 away from a parS site brings together the flanking DNA sequences, thus co-aligning the left 20 and the right arm of the chromosome (Minnen et al., 2016; Tran et al., 2017; Wang et al., 21 2017; Wang et al., 2015). This translocation is thought to localize DNA entanglements on the 22 replicating chromosome (i.e. knots and catenanes) facilitating chromosome individualization 23 by DNA topoisomerase (Bürmann and Gruber, 2015; Gruber et al., 2014; Orlandini et al., 2019; 24 Wang et al., 2014). 25 Smc-ScpAB complexes translocate rapidly (~1 kb/sec) along chromosomal DNA in vivo. In vitro 26 they support only limited ATP hydrolysis activity (< 1/sec), implying a large motor step size 27 (~1 kb or ~600 nm) (Hirano and Hirano, 2006; Vazquez Nunez et al., 2019; Wang et al., 2017; 28 Wang et al., 2018). Several models for Smc translocation have been proposed (Diebold- 29 Durand et al., 2017; Hassler et al., 2018; Marko et al., 2019; Terakawa et al., 2017). Smc 30 proteins are comprised of an ABC-type ATP-binding head domain and a hinge domain 31 connected at a distance by a long antiparallel coiled coil arm. Hinge domains form homotypic 32 interactions in prokaryotic Smc complexes (Haering et al., 2002). Smc dimers furthermore

1 associate with a kleisin subunit, in bacteria named ScpA. Via its amino- and carboxy-terminal 2 domains, ScpA bridges the head of one Smc protein with the head-proximal arm of the other 3 (Bürmann et al., 2013). This generates tripartite SMC-kleisin rings that entrap chromosomal 4 DNA double helices (Gligoris et al., 2014; Wilhelm et al., 2015). The kite subunit ScpB also 5 forms dimers that associate with the central region of ScpA (Bürmann et al., 2013). 6 Smc arms contact one another. They co-align lengthwise thus collapsing the Smc-ScpAB 7 complex into a rod-shaped particle (Diebold-Durand et al., 2017; Minnen et al., 2016; Soh et 8 al., 2015). Eight distinct contacts are found between the two Smc arms, four of which involve 9 amino-terminal sequences (1N, 4N, 6N, and 8N; see Figure 2B) and the other four carboxy- 10 terminal sequences (2C, 3C, 5C, and 7C) (Diebold-Durand et al., 2017). In yeast condensin, 11 corresponding sequences are also found in juxtaposition (Lee et al., 2020) (Figure S4D). In 12 some SMC complexes (including cohesin and condensin), the arms fold at an ‘elbow’, thus 13 bringing the hinge into proximity of the heads (Bürmann et al., 2019) . Such folding has not 14 yet been observed for bacterial Smc and its role is unclear. 15 Smc heads engage with one another by sandwiching two ATP molecules using active site 16 residues provided by both heads thereby forming the catalytic center for ATP hydrolysis 17 (Hirano et al., 2001; Hopfner, 2016; Lammens et al., 2004). ATP-engagement of Smc heads is 18 thought to be incompatible with full arm alignment (Diebold-Durand et al., 2017; Kamada et 19 al., 2017; Lammens et al., 2004; Muir et al., 2020) thus delineating two mutually exclusive 20 conformations, one with ATP-engaged heads (O-shaped ‘open’ conformation) and one with 21 completely aligned arms (I-shaped ‘closed’ conformation). In the open conformation, ATP- 22 engaged heads divide the lumen of the SMC-kleisin ring into the S compartment that is 23 encircled by the long arms and the hinge, and the K compartment which is enclosed by ScpAB. 24 The open conformation of B. subtilis (Bsu) Smc-ScpAB harbors two sites for DNA binding in 25 the S compartment: a hinge/DNA interface and a head/DNA interface (Vazquez Nunez et al., 26 2019). In the closed conformation, the Smc arms are aligned. The S compartment is thus 27 closed and separated from the K compartment by juxtaposed Smc heads. Principally 28 consistent observations have recently been made by cryogenic electron microscopy for other 29 SMC complexes, whereas AFM studies indicated much higher flexibility in the SMC arms (see 30 Discussion). SMC arm dynamics appear crucial for DNA loop extrusion, but exactly how the 31 different conformations bind to DNA and contribute to DNA translocation remains to be 32 elucidated.

1 Here, we studied the shape of the Bsu Smc-ScpAB by EPR spectroscopy. We observed in 2 addition to the open and the closed conformation an intermediary one with partially open S 3 compartments. Separation of arms was detected near the heads even in the absence of 4 ligands. Opening at the hinge, however, is largely dependent on ATP and DNA binding. We 5 showed that partial opening of the S compartment is sufficient for ATP hydrolysis, but 6 intriguingly DNA stimulation of the ATPase activity requires complete opening. This implies 7 that an open S compartment is an integral part of the normal ATP hydrolysis cycle. We 8 identified one out of the eight arm contacts as particularly important for controlling the 9 dynamic Smc architecture. Mutating a single residue at this 4N contact interface eliminates 10 Smc function, presumably by preventing arm closure. The mutations cause defect in 11 chromosome loading or DNA translocation. Altogether, our results suggest that the S 12 compartment opens by a graded transition starting at the heads. Head-proximal arm contacts 13 dissociate relatively easily, while hinge-proximal contacts are more stably engaged. We 14 identified the 4N contact as critical for controlling the opening/closure reaction. It may provide 15 directionality to the DNA translocation motor.

1 Results 2

3 Conformations of Smc-ScpAB detected by EPR. 4 5 Here, we aimed to characterize the conformational ensemble of Smc-ScpAB by measuring 6 arm-to-arm distances. Based on published work, we expected the distances to be broadly 7 distributed covering a spectrum of closed and open conformations. Such dynamic structures 8 are difficult to assess by high resolution techniques such as X-ray crystallography and electron 9 microscopy. Therefore, we employed electron paramagnetic resonance (EPR-DEER) which 10 detects the coupling of electron spins over a larger distance range (from 1.5 to 10 nm) and 11 yields information on distance distributions for a population of particles (Jeschke, 2012; 12 Polyhach et al., 2012; Reginsson and Schiemann, 2011). We labelled purified cysteine-mutant 13 Smc protein at one out of four selected positions (D193C, E217C, R643C and R718C) with the 14 spin label MTSL and reconstituted holo-complexes by mixing with unmodified ScpA and ScpB 15 (Figure 1A). The EPR experiments were performed with the cysteine-lite Smc3S protein (C119S, 16 C437S and C826S) to minimize any off-target labelling. MTSL-labelling efficiencies were 17 around 90 % as calculated from continuous-wave EPR spectra (Figure S1A). Of note, protein 18 samples were prepared in deuterated environment to reduce electron spin relaxation 19 induced by proton nuclear spin diffusion (El Mkami and Norman, 2015). Deuterated buffers 20 had only minor impact on protein function, as judged by near-normal ATPase activity (Figure 21 1SB). For most experiments, the Smc protein also included the Walker B E1118Q (EQ) 22 mutation which hinders the ATP hydrolysis step. Protein preparations were pre-incubated

23 with or without ATP and linear 40 bp double-stranded DNA, designated as dsDNA40, flash- 24 frozen in liquid nitrogen and measured at a temperature of 50 K. 25 Most distances measured across the hinge domain dimer in Smc3S(R643C, E1118Q) were 26 narrowly distributed with a peak at ~1.5 nm, at the lower edge of the sensitive range of high- 27 power Q-band DEER, well-fitting with available crystal structures (Cα-Cα ~1.1 nm) (Figures 1A 28 and 1B) (Griese and Hopfner, 2011; Haering et al., 2002; Kamada et al., 2017; Soh et al., 2015). 29 A minor fraction with larger distances (~4-6 nm) was potentially also noticeable (Figure 1B). 30 Arm-to-arm distances near the heads showed a clear bimodal distribution in the absence of 31 ligand (the apo state) (Figure 1B, S1C, S1E-F). For Smc3S(D193C, E1118Q) (at contact 1N), a

1 population with narrowly distributed short distances represented about 55 % of all distances. 2 These were centered at ~1.7 nm and displayed a good fit with the Smc rod model built from 3 crystal structures (Diebold-Durand et al., 2017). The long-distance population showed a much 4 broader distribution spanning from about 3.0 to at least 6.0 nm, indicating separated Smc 5 arms near the heads in a substantial proportion of Smc complexes (Figure 1B). A similar 6 pattern was observed with Smc3S(E217C, E1118Q) (at contact 2C): ~69 % of the dipolar 7 couplings resided in a narrow short-distance population centered at ~3.9 nm, while other 8 distances ranged from ~6.0 nm to more than 8.5 nm (Figure 1B). Again, the short-distance 9 population is in good agreement with the rod model (Diebold-Durand et al., 2017). For the 10 hinge-proximal arm position, Smc3S(R718C, E1118Q) (at contact 7C), 71 % of the distances 11 accumulated in a population with short arm-to-arm distances at ~1.5 nm. This number likely 12 underestimates the size of the population due to partial suppression of the dipolar 13 modulation at such short distances (similar to R643C) (Figure 1B). These measurements imply 14 that in the absence of ligands one or more conformations exist in addition to the closed rod 15 structure. Smc arms appeared separated near the heads (especially at contact 1N) in a larger 16 fraction of complexes than near the hinge, implying the presence of partially open, 17 intermediary conformations (Figure 1A). This is consistent with the pattern of arm cross- 18 linking observed in vivo, where the levels of cysteine cross-linking were somewhat higher at 19 contacts 8N and 6N than at contacts 3C, 2N and 1C (Diebold-Durand et al., 2017). 20 Addition of ATP to Smc(E1118Q)-ScpAB did not noticeably change the distribution between 21 the short- and the long-distance population for S217C and R643C (Figure S1D). For D193C, the 22 fraction of long distances became slightly larger indicating a trend towards arm opening upon 23 ATP binding. For S217C, the long distances became somewhat shorter. The distance 24 distribution at the hinge (R643C) remained virtually unchanged also in the presence of ATP

25 and dsDNA40 (Figure 1D). Arm-to-arm distances, however, showed a pronounced shift from 26 the short-distance to the now predominant long-distance population upon pre-incubation

C 27 with ATP and dsDNA40. For example, close to the hinge (R718C) (at contact 7 ), a broad long- 28 distance population ranging from about 2.0 to 7.0 nm represented about 70 % of all

29 measurable distances in the presence of ATP and dsDNA40 (Figure 1C). The extent of shift to 30 larger distances varied slightly between the cysteine positions, possibly resulting from 31 uncertainty in quantifying short-distance (<1.5 nm) or long-distance (>10 nm) populations or 32 indicating that the cysteine residues or their chemical labelling mildly affected the stability of

1 the conformations. Regardless, these results provided strong support for the notion that Smc 2 arms separate from one another upon ATP and DNA addition. The Smc arms are presumably 3 fully detached in a significant fraction of complexes when bound to ATP and DNA. From the 4 distance distribution at R718C, we estimated that the arms are connected to the hinge at 5 narrow angles (0 to ~45°). More open angles—as seen in some crystal structures of Smc hinge 6 fragments (~180°)—were however not observed. The open conformation thus represents 7 elongated, oval-shaped particles. Of note, we obtained similar trends in the absence of ScpAB 8 and when using wild-type Smc instead of Smc(E1118Q) proteins (Figure S1E-F), although arm 9 dissociation (at D193C) with ATP and DNA was less pronounced with wt Smc when compared 10 to Smc(E1118Q). 11 Altogether, we conclude from the EPR measurements that apo Smc-ScpAB exists mainly in 12 the closed conformation and in smaller sub-populations of the open and very likely also of an 13 intermediary conformation. Upon ligand binding, the distributions shift towards larger arm- 14 to-arm distances generating a sizeable fraction of open conformations. The broad range of 15 measured distances indicates that the arms are somewhat flexible at least when arm 16 alignment is lost. 17

18 Partial arm opening is sufficient for ATP hydrolysis, but DNA stimulation 19 requires full opening. 20 21 We next wondered whether intermediary conformations—as observed by EPR–are able to 22 support ATP hydrolysis or whether a fully open conformation is prerequisite for ATP 23 hydrolysis. To test this, we blocked S compartment opening by engineering a covalent arm- 24 to-arm junction at selected arm positions: at contacts 1N, 3C, 4N, and 7C (Figure 2B). Stiff arms 25 would be expected to prevent ATP hydrolysis when conjoined, while flexible arms would 26 support normal ATP hydrolysis even when conjoined. A reduction in the ATPase rate varying 27 with the position of the engineered arm-to-arm junction would imply limited flexibility. 28 We again made use of site-directed chemical modification. We cross-linked cysteine residues 29 in purified preparations of Smc with the thiol-reactive compound 1,3-propanediyl- 30 bismethanethiosulfonate (M3M) to then reconstitute arm-linked holo-complexes and 31 measure their ATPase activity (Figure 2A). M3M cross-linking is efficient as well as reversible

1 thus allowing to break the engineered arm-to-arm connection to recover lost activity. These 2 experiments were performed with cys-free Smc3SV (C119S, C437S, C826S, C1114V) protein 3 (Figure 2C). However, comparable results were obtained with the Smc3S variant (Figure S2A). 4 Incubation of Smc3SV(D193C), Smc3SV(Q320C), Smc3SV(A715C) or Smc3SV(Y944C) with M3M 5 produced at least 70 % to 85 % of chemically cross-linked dimers as judged by non-reducing 6 SDS-PAGE analysis. Addition of up to 10 mM DTT was required to reverse the cross-linking 7 reaction (Figure 2C). 8 All cross-linked preparations yielded significant ATPase activity regardless of the position of 9 the engineered arm-to-arm junction. While a Cys-free control sample showed no discernible 10 effect of M3M treatment on ATP hydrolysis (Figure 2C), a substantial loss of ATPase activity 11 was observed with head-proximal cysteine residues (D193C at contact 1N and Y944C at 12 contact 3C). No such effect was seen for the more distant positions (Q320C and A715C at 13 contacts 4N and 7C, respectively). In all cases, normal ATPase activity was restored by pre- 14 incubation with reducing agent. These results strongly suggest that complete separation of 15 Smc arms is not required for productive head engagement and for ATP hydrolysis. 16 Intermediary conformations may support ATP hydrolysis. Cross-linking of Q320C (at contact 17 4N) even increased the basal ATPase rate (see below). 18 We found that the stimulation of the ATPase activity by addition of 40 bp double-stranded

19 DNA (dsDNA40) was eliminated by M3M treatment regardless of the position of the cysteine 20 residue on the arm and regardless of any dampening or stimulating effect of the arm-to-arm 21 junction on ATP hydrolysis (Figure 2B). DNA stimulation was however restored upon pre-

22 incubation with reducing agent. The stimulation of the Smc ATPase by dsDNA40 thus depends 23 on the separation of residues on opposite arms, being consistent with the idea that complete 24 arm opening is required for DNA stimulation. DNA binding at the hinge/DNA interface may 25 boost ATP hydrolysis by modifying the organization of the arms as previously proposed (Soh 26 et al., 2015). In contrast to the strong effect on ATP hydrolysis (Figure 2D), the arm-to-arm 27 junction at Y944C did not substantially hinder DNA binding by Smc-ScpAB (Figure S2B), 28 suggesting that ATP hydrolysis is uncoupled from DNA binding in the cross-linked material or 29 that DNA binding predominantly occurs at a site this is uncoupled from the ATPase even in 30 unmodified Smc-ScpAB. 31 Altogether, these findings imply that the Smc ATPase can operate in two modes: as a basal 32 ATPase in a partially open conformation and as a DNA-stimulated ATPase presumably with an

1 open conformation. The separation of head-proximal arms promotes ATP hydrolysis, while 2 hinge-proximal arms may remain juxtaposed during basal ATP hydrolysis cycles. Interestingly, 3 ScpAB becomes essential for ATP hydrolysis when arms are conjoined at contact 1N (D193C) 4 (Figure 2D), suggesting that it helps to overcome constraints imposed by the engineered 5 junction on the ATPase heads. Of note, the fact that DNA is unable to stimulate ATP hydrolysis 6 in arm-locked complexes is consistent with the prior observation that the hinge/DNA 7 interface but not the head/DNA interface is important for DNA stimulation of ATP hydrolysis 8 (Vazquez Nunez et al., 2019). 9

10 Stimulation of ATP hydrolysis by chemical modification of Smc arms. 11 12 During the above experiments, we noticed that another cysteine variant, Smc3SV(E295C), 13 showed aberrant enzyme kinetics. The protein supported robust cysteine cross-linking (87.8 14 ± 4.6 %) with the cross-linked protein preparation displaying a ~2.5-fold higher ATPase activity 15 (Figure 3C, S3A and S3B). This increase was reversed to normal levels by pre-incubation with 16 reducing agent. The E295C mutation largely eliminated DNA stimulation of the ATPase even 17 without cross-linking. This finding confirmed that robust ATP hydrolysis can be achieved when 18 arms are covalently conjoined (at contact 4N) and suggests that Smc arms can have a positive 19 effect on ATP hydrolysis even when they are artificially linked together. 20 In the crystal structure (PDB: 5NNV), E295 residues form inter-subunit hydrogen bonds with 21 S294 residues (Figure 3B), thus likely stabilizing the 4N arm contact (Diebold-Durand et al., 22 2017). The lack of these hydrogen bonds in the E295C mutant presumably interferes with 23 normal arm alignment. Cross-linking of E295C may further disrupt Smc rod alignment. 24 Intriguingly, we previously isolated point mutations at the 4N arm contact (i.e. D280G, Q320R, 25 or E323K) that suppressed the lethal phenotype of an arm length-modified Smc protein 26 (Bürmann et al., 2017), supporting the notion that the 4N arm contact is particularly important 27 for controlling conformational transitions during the ATP hydrolysis cycle. 28

29 A critical contact for arm alignment. 30

1 To discern the physiological relevance of the 4N arm contact, we next looked at the 2 conservation of Smc arm sequences at this interface. Residue G302 and neighboring residues 3 are well conserved in firmicute Smc proteins while arm sequences are otherwise relatively 4 poorly conserved (Figure 4A). The glycine residue is located directly at the interface (Figure 5 4B), separated by two α-helical turns from E295. To test whether G302 is important for Smc 6 function, we generated ten substitutions by allelic replacement at the endogenous smc locus 7 in B. subtilis (Figure 4C). Remarkably, substitution to glutamate (GE), tryptophan (GW), or 8 phenylalanine (GF) prevented growth of Bsu on nutrient rich medium—similar to smc deletion 9 mutants— despite the GE and GW proteins being expressed at normal levels (Figure 4D). 10 Three other mutants (GM, GQ, GK) were unable to support growth of a ΔparB mutant on this 11 growth medium, while the arginine mutant (GR) supported growth of the double mutant only 12 poorly (Figure S4A). Substitutions for residues with smaller side chains (alanine, valine and 13 serine) had no discernible effects on growth. Thus, residues with negatively charged or bulky 14 sidechains at position 302 hinder protein function, presumably by destabilizing the 4N arm 15 contact. To test this, we combined the G302 mutations with a sensor cysteine for arm 16 alignment (A715C, at contact 7C) and one for head juxtaposition (S152C). In these 17 experiments, we focused on two G302 mutants with a severe growth defect and sidechains 18 with distinct physicochemical properties (GE and GW). The two mutations indeed reduced 19 cysteine cross-linking at both sites in vivo, with GW having rather mild and GE quite dramatic 20 effects (Figure 4E). The GW mutation also strongly reduced the abundance of the closed 21 conformation (i.e. the short distance population) as measured by EPR with D193C in the 22 absence and presence of ATP and DNA (Figure S4B, Table S2). We conclude that residue G302 23 and the 4N arm contact are important for co-aligning arms in the closed conformation. 24

25 Defective chromosome loading of 4N arm contact mutants. 26 27 We next determined the chromosomal distribution of Smc(GE) and Smc(GW) proteins by 28 chromatin-immunoprecipitation (ChIP) using α-ScpB and α-Smc antisera. Both mutants had 29 close to normal enrichment at the parS-359 site but very low enrichment at the dnaA locus 30 as determined by ChIP-qPCR (Figure 4F). These mutants thus appear to target normally to 31 parS but then fail to load onto DNA or to translocate along DNA (Minnen et al., 2016).

1 Consistent with this notion, we found that their E1118Q variants showed elevated levels of 2 enrichment at parS, similar or even higher than the otherwise wild-type Smc(E1118Q) protein. 3 Formation of the 4N arm contact may thus be required for the loading reaction or for the 4 release of Smc protein from ParB/parS loading sites. Notably, a putative ParB/Smc interface 5 has been mapped onto the Smc arm in the vicinity of the 3C and 4N arm contacts (Minnen et 6 al., 2016). 7 We found that the GW and the GE mutation alleviated the requirement for ScpA in the 8 targeting of the Smc(E1118Q) protein to a parS site (Figure 4G). ScpA may thus contribute to 9 overcoming constraints imposed by the aligned Smc arms to facilitate the transition to the 10 open conformation for parS targeting (Minnen et al., 2016). The G302 mutations likely 11 alleviate these constraints and eliminate the requirement for ScpAB. The integrity of the 4N 12 arm contact is subsequently required for the closure of the S compartment and the clearance 13 of Smc-ScpAB from ParB/parS. Purified preparations of Smc(G302W) and to a weaker extent 14 also Smc(G302E) displayed increased ATP hydrolysis activity, which was curiously hindered 15 rather than enhanced by the addition of ScpAB and DNA (Figure S4C). These point mutations 16 in the Smc arms thus lead to mis-regulation of the ATPase by DNA.

17 Intermediary conformations in vivo. 18 19 We next focused on the organization of Smc heads in vivo by inferring their conformations 20 from patterns of cross-linking obtained with Smc(K1151C). The K1151C residue was previously 21 engineered to detect the ATP-engaged state (Cα-Cα distance: 11.6 Å) by BMOE cross-linking 22 (Minnen et al., 2016). Here, we made use of bis-maleimide cross-linkers with longer spacers— 23 which supported robust cross-linking in intact Bsu cells—to detect alternative conformations 24 (Figure 5A). We observed that cross-linking of Smc(K1151C)-HaloTag was much more efficient 25 with BM3 (17.8 Å) and BM11 (62.3 Å) than with BMOE (8 Å) or BM2 (14.7 Å) (Figure 5A and 26 S5A) (Minnen et al., 2016). The Walker A ATP-binding mutant Smc(K37I) mutant also showed 27 an upward trend in cross-linking with extending spacer length, but the cross-linking 28 efficiencies did not reach levels comparable to wild-type Smc (Figure 5A). The Smc(E1118Q) 29 protein displayed roughly equivalent cross-linking efficiencies regardless of spacer length 30 (Figure 5A and S5A). Finally, a clear downward trend in cross-linking with increasing spacer 31 length was observed when the E1118Q mutation was combined with three different Smc arm

1 alterations (Figure 5B). For example, the Smc(204Ω/996Ω, E1118Q) protein—harboring a 13 2 amino acid peptide insertion in each arm—showed most efficient cross-linking with BMOE 3 and decreasing efficiencies with increasing spacer length (Diebold-Durand et al., 2017). A 4 similar trend was observed with an artificially shortened ‘mini’-Smc(E1118Q) protein 5 (Bürmann et al., 2017) and with Smc(GE, E1118Q) (Figure 5B). This downward trend in cross- 6 linking was expected for the ATP-engaged state because the K1151C residues are closely 7 juxtaposed and thus ideally positioned for cross-linking by the smallest compound BMOE 8 (Minnen et al., 2016). 9 We conclude that in addition to two previously known states (disengaged and ATP-engaged) 10 (Figure 1A), Smc heads also occur in a third, hitherto uncharacterized state. We envision that 11 in this ‘ATP-pre-engaged’ state active site residues from both heads are aligned (thus 12 supporting efficient K1151C cross-linking by BM3 and BM11), but the closure of the catalytic 13 pocket remains incomplete due to a gap persisting between the signature motif residues from 14 one head and the ribonucleotide bound to the opposing head (thus reducing K1151C cross- 15 linking by BMOE). Taking all in vivo cross-linking data into account (Figure 5A and B), we next 16 estimated the occupancy of the different states in wild-type Smc. Assuming a 100 % 17 occupancy of the ATP-engaged state in Smc(204Ω/996Ω, E1118Q) (Figure 5B) and of the 18 disengaged state in the ATP-binding mutant Smc(K37I) (Figure 5A), we propose that wild-type 19 Smc-ScpAB complexes rarely populated the ATP-engaged state and distributed roughly 20 equally between the other states (Figure 5A). In Smc(E1118Q), about a quarter of complexes 21 appear to be ATP-engaged suggesting that head engagement is an inefficient and stepwise 22 process in vivo, unless arm integrity is affected. 23 Of note, while ScpA is crucial for forming the ATP-engaged state in Smc(E1118Q), the 24 requirement for ScpA was (partially) alleviated in the GW and GE variants (Figure S5E) 25 (Minnen et al., 2016). As mentioned above, ScpA was dispensable for the targeting of the GE 26 and GW Smc(E1118Q) proteins to parS sites (Figure 4G) being consistent with the notion that 27 ScpAB helps to disrupt arm co-alignment to facilitate head engagement and parS targeting. 28 This is in excellent agreement with the efficient and ScpA-independent parS targeting of a 29 hinge dimerization mutant of Smc(EQ), which also displays mis-aligned Smc arms and 30 increased levels of head engagement (Minnen et al., 2016). 31 Finally, we wanted to establish whether the ATP-pre-engaged state—detected by BM3 and 32 BM11 cross-linking of Smc(K1151C)—corresponded to the closed, the open or the

1 intermediary conformation. Using a combination of cysteine residues, we found that arm 2 cross-linking with Q320C (at contact 4N) appeared to occur largely independently of K1151C 3 cross-linking, implying that the 4N arm contact is mostly engaged when heads are cross-linked 4 at K1151C (Figure 6A). The efficient BM3 cross-linking of ATP-pre-engaged heads (in otherwise 5 wild-type Smc) thus must have occurred on a (largely) closed conformation, implying that the 6 closed conformation exists in a disengaged and an ATP-pre-engaged state (Figure 6B).

1 Discussion 2 Chromosomes undergo continuous reorganization by DNA loop extrusion. The latter involves 3 a unique DNA motor which moves by apparently making large translocation steps. Elucidating 4 the underlying mechanism requires an understanding of the conformational transitions in 5 SMC complexes during the ATP hydrolysis cycle as well as the concomitant DNA binding and 6 unbinding events. Here we provide structural insights into three separate conformations and 7 corresponding state of the ATPase. The conformational dynamics appears largely governed 8 by a limited rigidity of the head-proximal Smc arms and by a finite stability of arm contacts. 9 While the architecture appears dynamic, principles of organization can be deduced. 10 The conformational landscape of SMC complexes. 11 Cryo-electron microscopy recently helped to reveal the molecular architecture of SMC-kleisin 12 complexes (in apo and ATP-engaged states) from eukaryotes (Collier et al., 2020; Higashi et 13 al., 2020; Lee et al., 2020; Shi et al., 2020). Solving structures of more dynamic or flexible 14 conformations, however, remains challenging even for cryo-EM. EPR is ideally suited to help 15 close these gaps by providing insights into the distributions of conformations (Jeschke, 2018; 16 Kazmier et al., 2014; Vercellino et al., 2017). 17 The closed conformation. 18 Independent lines of evidence point to the prevalence and conservation of a closed 19 conformation in SMC complexes. It is predominant in reconstituted Smc-ScpAB, as judged by 20 EPR (Figure 1), and in living Bsu cells according to site-specific cross-linking experiments 21 (Diebold-Durand et al., 2017). Its main characteristic features are disengaged heads, and 22 completely aligned arms. Point mutations that hinder arm alignment eliminate Smc function 23 in Bsu implying that the closed conformation is critically important (Figure 4). Yeast condensin 24 was found by cryo-electron microscopy as an elongated particle with ‘non-engaged’ heads 25 and completely aligned arms (Lee et al., 2020), presumably corresponding to the Smc-ScpAB 26 rod, albeit displaying folded Smc2 and Smc4 arms (see below). Intriguingly, the Smc2 and 27 Smc4 arms in apo condensin are most closely juxtaposed at the region that corresponds to 28 the 4N arm contact in Smc-ScpAB, being consistent with the idea that this contact has retained 29 its special function in yeast condensin, and possibly other SMC complexes, too (Figure S4D). 30 The condensin arms bend ~180° at an elbow allowing the hinge to fold back and contact the 31 middle of the Smc2 arm (Lee et al., 2020). This feature has not been observed for Smc-ScpAB

1 and may have evolved in eukaryotic SMC proteins (and independently also in MukBEF) 2 (Bürmann et al., 2019). The function of arm folding and of the hinge/arm contact is not well 3 understood. It is conceivable that the folding may be relevant to the process of stepwise 4 opening and closing of the S compartment, possibly stabilizing half-open states and thus 5 providing additional levels of control. Of note, the closed conformation is rare in AFM images 6 of yeast condensin and Bsu Smc-ScpAB, unless the material is cross-linked prior to sample 7 deposition (Fuentes-Perez et al., 2012; Ryu et al., 2019; Yoshimura et al., 2002). We suspect 8 that the nonspecific adsorption of samples onto the AFM mica may affect SMC rod folding 9 dynamics. 10 The open conformation. 11 The open conformation of Smc-ScpAB has previously been inferred indirectly from crystal 12 structures of isolated ATP-engaged heads and supported by low resolution images obtained 13 by rotary shadowing electron microscopy and by loss of cysteine cross-linking (Anderson et 14 al., 2002; Diebold-Durand et al., 2017; Kamada et al., 2017; Lammens et al., 2004). In this 15 study, we provide direct evidence for the existence of the open conformation. We 16 characterize its shape and show that it represents an integral part of the DNA-stimulated ATP 17 hydrolysis cycle. Smc dimers that are unable to completely open the S compartment retained 18 the ability to hydrolyze ATP but lost the stimulation upon DNA addition. Hinge/DNA binding 19 may open the arms at the hinge as previously proposed (Soh et al., 2015), thus facilitating the 20 transition to ATP-engaged heads and supporting efficient ATP hydrolysis. 21 Intermediary conformations. 22 Here, we have expanded our knowledge on the conformational landscape of Smc-ScpAB by 23 detecting and characterizing intermediary conformations using three complementary 24 approaches (EPR, ATPase measurements and cysteine cross-linking) (Figure 1, 2 and 5). 25 Altogether, our data suggest that Smc heads and the head-proximal arms undergo dynamic 26 structural changes. 27 Patterns of cross-linking demonstrate that the Smc heads can adopt at least three different 28 states in vivo, here denoted as disengaged, ATP-pre-engaged, and ATP-engaged state (Figure 29 6B). The ATP-engaged state occurs in the open and the intermediary conformation. We 30 propose that in the ATP-pre-engaged state, the active site residues from both heads are 31 aligned via ATP but the arms remain (largely or completely) aligned. The heads may transition

1 gradually from the disengaged, via the ATP-pre-engaged to the ATP-engaged state. Both steps 2 might be impeded by arm alignment. 3 Artificially engineered arm junctions reduce—but do not block—ATP hydrolysis (Figure 2 and 4 3). The ATP-engaged state—presumably a prerequisite for ATP hydrolysis—is thus compatible 5 with partially open arms. Whether such basal ATPase cycles are futile or contributing to DNA 6 translocation remains to be established. The fact that the essential head/DNA interfaces is 7 exposed in the intermediary conformation (while the non-essential hinge/DNA interface is 8 likely inaccessible) is consistent with the notion that this state is functionally relevant (Figure 9 6B), (Vazquez Nunez et al., 2019). 10 Cryo-EM structures of DNA-bound cohesin in the ATP-engaged state show a partially open 11 conformation with separated head-proximal arms (Collier et al., 2020; Higashi et al., 2020; Shi 12 et al., 2020). Most of the more distal arm sequences are less well resolved—probably due to 13 structural flexibility—but appear to remain aligned (and elbow-folded) (Collier et al., 2020; 14 Higashi et al., 2020). Whether conformations with completely dissociated arms are present in 15 cohesin (and condensin) is not fully established, but rotary-shadowing electron micrographs 16 would be consistent with this notion (Anderson et al., 2002). Upon arm dissociation, the arms 17 either remain folded at the elbow thus producing B-shaped structures as inferred from AFM 18 images of condensin (Eeftens et al., 2016) or they unfold at the elbow to generate O-shaped 19 complexes (as seen by EM and similar to the open conformation of Smc-ScpAB). 20 The intermediary conformations of Smc-ScpAB are structurally not well defined, probably due 21 to significant flexibility in the head-proximal arms. In condensin, however, a well-defined 22 partially open state was detected by cryo-EM. This apo bridged conformation of yeast 23 condensin showed partial separation of arms and heads (Lee et al., 2020). The separated 24 heads are however bridged—and likely stabilized—by one of two hawk subunits, which are 25 specific to condensin and cohesin. Our observations, together with the recent cryo-EM 26 structures, indicate that intermediary conformations (with partially opened arms) are widely 27 conserved in SMC complexes. 28 The DNA translocation cycle. 29 It is tempting to speculate that the intermediary conformations control the process of 30 compartment opening and closure and direct the binding and unbinding of suitable DNA 31 substrates thus driving directional DNA translocation. Here, we propose that one of the eight

1 arm contacts acts as a molecular switch for arm opening in response to initial DNA contact 2 (Figure 6B). Chromosomal DNA is entrapped in the K compartment of the closed 3 conformation as determined by site-specific crosslinking in Bsu Smc-ScpAB (Vazquez Nunez 4 et al., 2019) and in yeast cohesin (Chapard et al., 2019). The open conformation holds two 5 DNA binding interfaces in the S compartment, of which the head/DNA interface—but not the 6 hinge/DNA interface—is essential for Smc function and critical for Smc DNA translocation 7 (Hirano and Hirano, 2006; Vazquez Nunez et al., 2019). These observations suggest that DNA 8 is transiently occupying the S compartment. How DNA enters and exits the S compartment is 9 a crucial open question. 10 The loop-capture model proposes that DNA is inserted as a loop on top of the heads and upon 11 ATP hydrolysis exits the S compartment by transfer to the K compartment between the heads 12 (Diebold-Durand et al., 2017; Marko et al., 2019). Alternatively, the DNA enters the S 13 compartment by transfer between the heads as recently proposed for yeast cohesin (Collier 14 et al., 2020). How the latter may lead to DNA translocation and DNA loop extrusion is however 15 unclear. Unilateral and gradual opening of the arms—as described in this study—is likely 16 directly relevant to this key substrate selection process. Future experiments will have to 17 establish how the conformations (open, closed and intermediary) of SMC complexes 18 associate with chromosomal DNA double helices to promote DNA loop extrusion. 19 20

1 Acknowledgements 2 We thank Frank Bürmann for help in setting up initial EPR-DEER experiments and for helpful 3 comments on the manuscript and Daniel Klose for collecting EPR-DEER data. We are grateful 4 to all members of the Gruber lab for support and stimulating discussions. We thank the 5 Genome Technologies Facility (GTF) at the University of Lausanne for deep sequencing. This 6 work was supported by the European Research Council (Horizon 2020 ERC CoG 724482).

1 Methods 2 3 Experimental Model 4 Bacillus subtilis strains and growth. 5 Bsu strains used in this work are derived from the 1A700 isolate. Genotypes and strain 6 numbers are listed in Table S5. Strain usage is detailed in Table S1. Naturally competent Bsu 7 cells were transformed as in (Diebold-Durand et al., 2017) including a longer starvation 8 incubation time (2 hours) for high efficiency. The transformants were selected on SMG-agar 9 plates with appropriate antibiotics and single-colonies were isolated. The strains were 10 confirmed by PCR and Sanger sequencing as required. For dilution spot assays, cells were 11 grown for 8 hours at 37 °C in SMG medium and then were diluted to 81 and 59 000-fold. 12 Afterwards, dilutions were spotted onto ONA (16 h incubation) or SMG (24 h incubation) agar 13 plates at 37 °C (Vazquez Nunez et al., 2019). 14 15 Protein purification 16 Full-length Bsu Smc 17 Native Smc proteins were purified following the procedure described in (Bürmann et al., 18 2017). E. coli BL21-Gold (DE3) strain was transformed with pET-22 or pET-28 derived plasmids 19 containing the smc recombinant sequences. Proteins were expressed using ZYM-5052 20 autoinduction medium for 23 h at 24 °C. Cells were resuspended in lysis buffer (50 mM Tris- 21 HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM DTT, 10 % (w/v) sucrose) supplemented with 22 protease inhibitor cocktail (PIC) and sonicated. The lysate was centrifuged, the supernatant 23 was filtered with a 0.45 µM pore size membrane and then loaded onto two HiTrap Blue HP 5 24 mL columns connected in series and eluted with lysis buffer containing 1 M NaCl. The main 25 peak elution fractions where diluted in salt-less buffer (50 mM Tris-HCl pH 7.5, 1 mM EDTA, 1 26 mM DTT) to a conductivity equivalent of 50 mM NaCl (≈ 8 mS/cm). The diluted sample was 27 supplemented with PIC and loaded on a HiTrap Heparin HP 5 mL column. Elution was 28 performed by applying a linear gradient of buffer up to 2 M NaCl. The main peak fractions 29 (aprox. 5 mL) were collected and further purified by gel filtration on a XK 16/70 Superose 6 30 PG column in 50 mM Tris-HCl pH 7.5, 200 mM NaCl, 1 mM EDTA, 1 mM TCEP. Main peak 31 fractions where collected, concentrated with a Vivaspin 15 10K MWCO filter, flash frozen with

1 liquid nitrogen and stored at −80 °C. Protein concentration was calculated by absorbance 2 using theoretical molar absorption and molecular weight values. 3 4 ScpA 5 Native ScpA was purified following the procedure reported in (Vazquez Nunez et al., 2019). 6 The protein was expressed using E. coli BL21-Gold (DE3) transformed with a pET-28 derived 7 plasmid containing ScpA coding sequence. Cells were cultivated in ZYM-5052 autoinduction 8 medium at 16 °C for 28 h, harvested and resuspended in lysis buffer (50 mM Tris-HCl pH 7.5, 9 200 mM NaCl, 5 % glycerol) supplemented with PIC. After sonication and centrifugation, the 10 supernatant was applied to a 5 mL HiTrapQ ion exchange column and eluted with a gradient 11 up to 2 M NaCl. The peak fractions were mixed with 4 M NaCl buffer to reach a final 12 concentration of 3 M NaCl. The sample was injected into a HiTrap Butyl HP column and eluted 13 in a reverse gradient to 50 mM NaCl. Peak fractions were pooled and concentrated to 5 mL in 14 Vivaspin 15 10K MWCO filters and subsequently purified by gel filtration in a Hi Load 16/600 15 Superdex 75 pg column, equilibrated in 20 mM Tris-HCl pH 7.5, 200 mM NaCl. Protein was 16 concentrated, aliquoted, flash frozen and stored at -80 °C. 17 18 ScpB 19 Native ScpB was purified using the procedure described in (Vazquez Nunez et al., 2019). A 20 pET-22 derived plasmid with the coding sequence of ScpB was transformed in chemically 21 competent E. coli cells. Cells were cultivated I ZYM-5052 autoinduction medium at 24 °C for 22 23 h. Cells were harvested and resuspended in lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM 23 NaCl, 1 mM EDTA, 1 mM DTT) supplemented with PIC. After sonication and centrifugation, 24 the supernatant was diluted to 50 mM NaCl, loaded onto a 5 mL HiTrap Q HP column and 25 eluted with a gradient to 2 M NaCl. The sample was diluted in lysis buffer with 4 M NaCl 26 buffer in order to reach a final concentration of 3 M NaCl. The sample was applied on two 5 27 mL HiTrap Butyl column connected in series. Protein was eluted with a reverse gradient to 50 28 mM NaCl. A second run on the 5 mL HiTrap Q HP column in the same conditions as described 29 above was necessary in some cases when the sample still showed 260/280 ratios close to 1. 30 The peak fractions were concentrated and further purified by SEC using a Hi Load 16/600 31 Superdex 200 pg equilibrated in 50 mM Tris-HCl pH 7.5, 100 mM NaCl, 1 mM DTT. The

1 fractions containing the protein were concentrated, aliquoted, flash frozen and stored at -80 2 °C. 3 4 Method details 5 ATPase assay 6 ATPase activity measurements were done using the coupled reaction of pyruvate 7 kinase/lactate dehydrogenase (Bürmann et al., 2017). ADP production was monitored for 1 8 h by oxidation of NADH absorbance changes at 340 nm. The values were collected using a 9 Synergy Neo Hybrid Multi-Mode Microplate reader. The reaction mixture consisted in 1 mM 10 NADH, 3 mM Phosphoenol pyruvic acid, 100 U Pyruvate kinase, 20 U Lactate dehydrogenase 11 and the appropriate ATP concentrations. Double-stranded oligonucleotides (40 bp (5’- 12 TTAGTTGTTC GTAGTGCTCG TCTGGCTCTG GATTACCCGC)) were added for measurements that 13 required it to a final concentration of 3 µM, unless indicated differently. The final protein 14 concentration in the assay was 0.15 μM Smc dimers in ATPase assay buffer (50 mM HEPES- 15 KOH pH 7.5, 50 mM NaCl, 2 mM MgCl2). Measurements were carried out at 25 °C. 16 17 M3M crosslinking-ATPase assay 18 For this assay, a fresh full-length Smc purification (as described above) was done for each 19 experiment. The last SEC purification step, however, was done in 50 mM Tris-HCl pH 7.5, 200 20 mM NaCl in order to remove any trace of reducing agents. The peak fractions were 21 concentrated to ~35 µM Smc dimer in a Vivaspin 15 10K MWCO filters. The crosslinked 22 reaction was carried on by mixing 10-fold thiol molar excess of M3M (diluted in DMSO). A 23 typical reaction was done in 500 µL with 30 µM Smc dimers and 600 µM M3M in 50 mM Tris- 24 HCl pH 7.5, 200 mM NaCl at 4 °C overnight, protected from light. The final concentration 25 sometimes differed slightly, depending on the purification yield of the single cysteine mutant, 26 but the M3M:thiol ratio was always maintained. A negative control was always included by 27 mixing the same amount of protein with an equivalent volume of DMSO as in the 28 experimental M3M reaction. 29 After crosslinking the sample was centrifuged to remove big aggregates and the excess of 30 crosslinker was removed with a Zeba spin 7K MWCO 0.5 mL desalting column, preequilibrated 31 with 50 mM Tris-HCl pH 7.5, 200 mM NaCl. An additional gel filtration purification was done 32 to remove small aggregates and traces of crosslinker in a Superose 6 increase 10/300 column

1 equilibrated in 50 mM Tris-HCl pH 7.5, 200 mM NaCl. Afterwards, the fractions at the 2 maximum of the peak were recovered. The concentration of the protein was typically ~4 µM 3 Smc dimer. The sample was mixed with ScpA and ScpB and the ATPase assay was performed 4 as described above. 10 mM DTT final concentration was added for the samples that required 5 it. 6 7 Fluorescence anisotropy measurements. 8 Fluorescence anisotropy was measured using a 40 bp dsDNA (as described in ATPase assay) 9 modified at the 3′ end with fluorescein. Measurements were recorded using a Synergy Neo 10 Hybrid Multi-Mode Microplate reader (BioTek) with the appropriate filters in black 96-well 11 flat bottom plates at 25°C. Buffer conditions were 50 mM Tris-HCl pH 7.5, 50 mM NaCl, 2 mM

12 MgCl2 50 nM dsDNA and 1 mM ATP for all measurements. Anisotropy measurements where 13 exported from the BioTek Synergy Neo software and subsequently fit to a binding polynomial 14 using non-linear regression in GraphPad Prism 8 (Vazquez Nunez et al., 2019) 15 16 Double Electron-Electron Resonance (EPR-DEER). 17 Smc3C, ScpA and ScpB purifications were performed as described above, except for the last 18 gel filtration step of Smc purification which was performed in absence of reducing agent. In 19 order to avoid premature thiol oxidation, the sample was mixed with 10-fold molar excess of 20 MTSL per thiol, immediately after the last purification step. The reaction volume was 500 µL 21 and consisted in ~40 µM Smc dimer and 800 µM MTSL in 50 mM Tris-HCl pH 7.5 at 4 °C and 22 200 mM NaCl. The reaction was incubated over-night at 4 °C protected from light. After the 23 incubation, the sample was centrifuged to remove big aggregates and the MTSL excess was 24 removed using a Zeba Spin desalting column, 7K MWCO, 0.5 mL. Afterwards the sample was 25 further purified by gel filtration in a Superose 6 10/300 increase, equilibrated with 50 mM 26 Tris-HCl pH 7.5 at 4 °C, 200 mM NaCl and 90 % deuterium oxide (heavy water). The sample 27 was concentrated to ~20 µM Smc dimer and the label efficiency was determined by 28 calculating the double integral of the continuous wave (CW) EPR spectrum. Room 29 temperature CW EPR measurements were done at X band (~ 9.7 GHz) using a commercial X- 30 band Magnettech MiniScope MS 400. Spectra were taken at 13 dB attenuation, 31 corresponding to 5 mW incident microwave power and 0.15 mT magnetic field modulation 32 amplitude. Samples were placed into 1 mm o.d. glass capillaries. Spin labelling efficiency for

1 all mutants was estimated by comparing double integrals of EPR signals from the Smc mutants 2 and the concentration standard (200 µM water solution of TEMPOL). 3 Samples were flash frozen and stored at -80 °C until use. Before measurement, the protein 4 samples (Smc, ScpA and ScpB) were thawed and centrifuged to remove big aggregates. The 5 final concentration per condition were: 5 µM SmcScpAB, 10 µM dsDNA 40 bp and 1 mM ATP

6 in 50 mM Tris-HCl pH 7.5 at 4 °C, 50 mM NaCl, 2 mM MgCl2, 40 % Ethylene glycol-d6 (as 7 cryoprotectant) and 90 % deuterium oxide. Samples were incubated 5 minutes in 1.5 mL 8 centrifuge tubes at 25 °C with gentile (700 rpm) shaking. Distance measurements were 9 performed at Q-band frequencies (~ 34.4 GHz) using a standard double electron-electron 10 resonance (DEER) sequence (Pannier et al., 2000). DEER traces were acquired at 50 K. A 11 commercial X/Q-band Elexsys E580 spectrometer (Bruker) power-upgraded to 200 W 12 equipped with the homebuilt TE102 rectangular resonator was used (Polyhach et al., 2012). 13 The protein solution ready for measurement was filled into the 3 mm o.d. quartz tubes, shock- 14 frozen by immersion into liquid nitrogen and inserted into the pre-cooled resonator. All 15 observer pulses were 16 ns long, the pump pulse was 12 ns long, the pump frequency was set 16 at the global maximum of the nitroxide EPR spectrum and the observer frequency was set 17 100 MHz lower. 18 19 In vivo cross-linking

20 Bsu cultures in 200 mL SMG were grown to exponential phase (OD600 = 0.02) at 37 °C. Cells 21 were collected by filtration harvesting and washed in cold PBS + 0.1 % (v/v) glycerol (‘PBSG’). 22 A biomass equivalent to 0.85 OD units was sorted. Cells samples were centrifuged 2 min at 23 10,000 g, resuspended in fresh PBSG (200 µL). The cross-linking reaction was started by 24 adding 0.5 mM BMOE and it was incubated for 10 min on ice. The reaction was quenched by 25 the addition of 1.4 mM 2-mercaptoethanol. Cells were centrifuged and resuspended in 30 μL 26 of lysis mix (75 U/mL ReadyLyse Lysozyme, 750 U/mL Sm DNase, 5 μM HaloTag TMR Substrate 27 and protease inhibitor cocktail ‘PIC’ in PBSG). Lysis reaction was incubated at 37 °C for 30 min. 28 Afterwards, the material was mixed with 10 μL of 4X LDS-PAGE buffer, samples were 29 incubated for 5 min at 95 °C and resolved by SDS-PAGE. Gels were imaged on an Amersham 30 Typhoon scanner with Cy3 DIGE filter setup. 31 32 Chromatin Immunoprecipitation

1 The procedure followed the one described in (Vazquez Nunez et al., 2019). Bsu strains were 2 cultured in 200 mL SMG medium from OD600 = 0.004 initial OD to OD600 = 0.02 at 37 °C. 3 Cells were crosslinked with 20 mL of buffer F (50 mM Tris-HCl pH 7.4, 100 mM NaCl, 0.5 mM 4 EGTA pH 8.0, 1 mM EDTA pH 8.0, 10 % (w/v) formaldehyde) and incubated for 30 min at room 5 temperature with occasional manual shaking. Cells were harvested by filtration and washed

6 in PBS at 4 ºC. The cell biomass equivalent to 2 OD600 units was resuspended in 1 mL TSEMS 7 (50 mM Tris pH 7.4, 50 mM NaCl, 10 mM EDTA pH 8.0, 0.5 M sucrose and protease inhibitor 8 cocktail) containing 6 mg/mL lysozyme from chicken egg white. Primary lysis was done at 37 9 °C for 30 min with shaking at 1400 rpm. Protoplasts were washed twice and resuspended in 10 1 mL TSEMS. The sample then was split into 3 aliquots and pelleted. Pellets were flash frozen 11 and stored at −80 °C until used. 12 Each pellet was thawed and resuspended in 2 mL buffer L (50 mM HEPES-KOH pH 7.5, 140 13 mM NaCl, 1 mM EDTA pH 8.0, 1 % (v/v) Triton X-100, 0.1 % (w/v) Na-deoxycholate) containing 14 0.1 mg/mL RNase A and PIC. The suspension was sonicated during 1 min in 3 cycles of 20 15 seconds and 40 % amplitude at 4 °C using a MS73 probe. The sonicated material was 16 centrifuged at 4 °C and 20,000 × g and 200 μL of the supernatant were collected as input 17 reference. The immunoprecipitation was carried out by mixing 750 μL of the extract with 50 18 μL of Dynabeads Protein-G suspension freshly charged with 50 μL α-ScpB antiserum and 19 incubated for 2 h on a wheel at 4 °C. Beads were washed with 1 mL each of buffer L, buffer L5 20 (buffer L containing 500 mM NaCl), buffer W (10 mM Tris-HCl pH 8.0, 250 mM LiCl, 0.5 % (v/v) 21 NP-40, 0.5 % (w/v) Na-Deoxycholate, 1 mM EDTA pH 8.0) and buffer TE (10 mM Tris-HCl pH 22 8.0, 1 mM EDTA pH 8.0). All wash steps were done at 25 °C during 5 min shaking (1400 rpm). 23 Beads were resuspended in 520 μL buffer TES (50 mM Tris-HCl pH 8.0, 10 mM EDTA pH 8.0, 1 24 % (w/v) SDS). The reference sample was mixed with, 300 μL buffer TES and 20 μL 10 % SDS. 25 Formaldehyde cross-links were reversed over-night at 65 °C with vigorous shaking. 26 For phenol/chloroform DNA extraction, samples were cooled to room temperature, mixed 27 with 500 μL phenol equilibrated with buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA), emulsified 28 and centrifuged for 10 min at 20,000 × g. Subsequently, 450 μL of the supernatant was 29 emulsified with 450 μL chloroform and centrifuged for 10 min at 20,000 × g. DNA precipitation 30 was done by taking 400 μL of the supernatant and mixing with 1.2 μL GlycoBlue, 40 μL of 3 M 31 Na-Acetate pH 5.2 and 1 mL ethanol and incubated for 20 min at −20 °C. Samples were 32 centrifuged at room temperature and 20,000 × g for 10 min, and the precipitate was dissolved

1 in 100 μL buffer PB (QIAGEN) for 10 min at 55 °C, purified with a PCR purification kit, and 2 eluted in 50 μL buffer EB. 3 For qPCR, samples were diluted in DNase-free water (1:10 for IP and 1:1000 for input) and 10 4 μL reactions (5 μL Takyon SYBR MasterMix, 1 μL of 3 μM primer mix, 4 μL sample) were run 5 in duplicates in a Rotor-Gene Q machine (QIAGEN) using the appropriate primer pairs. 6 7 Quantification and statistical analysis 8 Analysis of cross-linking efficiencies. 9 TMR fluorescence or Coomassie stained bands were quantified using ImageQuant TL 1D V8.1. 10 Lanes were defined manually. Bands were detected automatically, and the band intensities 11 corrected for background signal using the Rolling Ball algorithm with a ball radius set to 129. 12 Values from three replicate experiments were exported to Microsoft Excel for calculation of 13 average fractions and standard deviation. 14 15 Steady-state kinetics 16 The absorbance values were exported and fit to a straight-line equation in GraphPad prism 8. 17 The slope values were transformed to rate values using the molar absorption coefficient of 18 NADH. The rates were expressed into absolute values by correcting for the protein 19 concentration. When an ATP titration was done, the rates were fit using non-linear regression 20 to the Hill equation: [ ] 21 + [ ℎ] 𝑉𝑉𝑚𝑚𝑚𝑚𝑚𝑚. 𝐴𝐴𝐴𝐴𝐴𝐴 ℎ ℎ 22 Where Vmax is the maximal rate at a given𝐾𝐾0 5 protein𝐴𝐴𝐴𝐴𝐴𝐴 concentration, [ATP] is the variable ATP

23 concentration, K0.5 is the semi-saturation concentration and h is the Hill coefficient. 24 Fit parameters obtained from the average of duplicates summarized in Table S4. 25 26 DEER traces analysis. 27 Experimental DEER traces were processed with the open-source DeerAnalysis package 28 (Jeschke et al., 2006), available at www.epr.ethz.ch/software/index. For background 29 correction, a homogeneous distribution of spins in 3D space was assumed in all cases. 30 Distance distributions for all mutants were calculated using double Gaussian model, whereas

1 preliminary (calibrating) analysis for one of the mutants (D193) was performed model-free 2 with the Tikhonov regularization. 3 4 Analysis of qPCR Data 5 The threshold cycle (TC) was obtained by analyzing the fluorescence raw data in the Real- 6 Time PCR Miner server (http://ewindup.info/miner/) (Zhao and Fernald, 2005). 7 IP/input ratios were calculated as α 2ΔCT, where ΔCT = CT (Input) – CT and α is a constant 8 determined by extraction volumes and sample dilutions. Data are presented as the mean of 9 duplicates. 10 11 12 13 14 15 16 17 18 19 20 21 22

bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. ABClosed Intermediary Smc3S(E1118Q)-ScpAB, apo R643 hinge 1.5 nm 0.05 0.030 R643C R718C 0.025 R718 0.04 1.5 nm 0.020 0.03 >82 % 0.015 >71 % 0.02 0.010

0.01 0.005

P(r) 0.00 0.00 D193C E21E217C7C ~69 % 0.025 ~55 % 0.025

0.020 0.020 ~31 % ScpBcpBB ~45 % 0.015 0.015 joint L 8.5 ScScpAccppAA 3.5 nm E217 nm 0.010 0.010

0.7 nm D193 4.9 0.005 0.005 nm 0.000 0.000 2345678910 2345678910

Disengaged Distance (nm) Distance (nm) heads R643 C 3S 1.5 nm Smc (E1118Q)-ScpAB, ATP, dsDNA40 R718 0.05 0.030 2.8 nm R643C R718C 0.025 0.04 >28 % 0.020 0.03 >85 % 0.015 0.02 0.010

Open 0.01 0.005

P(r) 0.00 00.00.00000 D193C ~57 % E217CE21 0.0255 00020.00.025.0.02025 ~18 % 0.02020 ~82 % 00.020.020 ~43 % E217 8.5 nm 0.0150115 00.015.0015

0.0100110 00.010.010

0.0050005 0.005 D193 5.1 nm 0.00000 0.000 2345678910 ATP-engaged heads 2345678910 Distance (nm) Distance (nm)

Figure 1 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

1 Figure 1. Distance distributions in Smc-ScpAB determined by electron spin resonance (EPR- 2 DEER). 3 A. Structural models of Smc dimers in the closed (top left panel), an intermediary (top right 4 panel), and the open conformation (bottom panel). Models were built with the help of 5 available crystal structures (Diebold-Durand et al., 2017; Soh et al., 2015). Dashed lines 6 indicate putative bending of the Smc arms. Concentric circles indicate the positions of the 7 cysteines used for MTSL labelling. Distances are maxima for sub-populations as observed by 8 EPR. 9 B. Probability distributions as measured by EPR-DEER with reconstituted Smc-ScpAB in the 10 absence of ligands (apo). Schematics indicate the conformation for sub-populations with 11 estimated relative abundance given. 12 C. Probability distributions of Smc-ScpAB measured in the presence of 1 mM ATP and 10 µM

13 double-stranded 40 bp DNA (dsDNA40). Display as in (B). 14

28 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in A Smcperpetuity.3SV It is madeGGelel filtrfiltration availableaattion XL under-Smc a3SVCC-BY 4.0XL International-Smc3SVScpp licenseAB ATPaseATP. ase

ScpASScpA ScpB

+ M3MM +

BCB hingehinge SmcSmS c3SV SmcSmmc3SV33SSV(D193C)(DD193C) - dsdsDNADNA

) 404

-1 188 ++d dsDNAsDNA 18 8N 40 16 16

(min 14 7C A715C 14 12 12 10 10 8 8 6N 6 6 4 4 2 2 5C 0 ATP hydrolysis rate 0 Q320C Arm contacts 4N XL-Smc

3C Y944C 2C

1N D193C Smc Smc Nt DMSO M3M M3M DMSO M3M M3M Ct DTT DTT Heads Smc3SV(Y944C) Smc3SV(Q320C) Smc3SV(A715C) ) -1 20 20 18 18 18

(min 16 16 16 14 14 14 12 12 12 10 10 10 8 8 8 6 6 6 4 4 4 2 2 2 ATP hydrolysis rate 0 0 0 XL-Smc XL-Smc XL-Smc

Smc Smc Smc DMSO M3M M3M DMSO M3M M3M DMSO M3M M3M DTT DTT DTT D Smc3SV(D193C) Smc3SV(D193C), M3M 14 ) 14 Smc -1 12 12 Smc, dsDNA (min 40 10 10 Smc-ScpAB 8 8 Smc-ScpAB, dsDNA 6 6 40 4 4

ATP hydrolysis rate 2 2 0 0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 ATP (mM) ATP (mM) Figure 2 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

1 Figure 2. ATP hydrolysis by Smc-ScpAB with cross-linked arms. 2 A. Schematic for the preparation of cross-linked protein for ATPase measurements. Single 3 cysteine Smc3SV proteins were purified and incubated with M3M cross-linker. Excess of M3M 4 and any protein aggregates were removed by gel filtration. Eluate fractions were mixed with 5 stoichiometric amounts of purified ScpA and ScpB to reconstitute holo-complexes. ATPase 6 activity was determined. 7 B. Positions of arm contacts and engineered cysteine residues in the Bsu Smc dimer in the 8 closed conformation (Diebold-Durand et al., 2017). Amino-terminal sequences (Nt) are shown 9 in light blue colours and carboxy-terminal sequences (Ct) in dark blue colours. Inter-subunit 10 arm contacts are numbered 1 to 8 from the heads towards the hinge. Superscripts (N or C) 11 indicate the involvement of amino- or carboxy-terminal sequences at the contact, 12 respectively. Black circles denote the positions of cysteine residues engineered for site- 13 specific crosslinking. 14 C. ATP hydrolysis rates for preparations of Smc-ScpAB with cross-linked arms. Top panels: bar 15 graphs of ATP hydrolysis rates per Smc protein measured with 1 mM ATP, 0.15 µM Smc-ScpAB

16 complex in the absence (black bars) and presence (blue bars) of 3 µM dsDNA40. Error bars 17 correspond to the standard deviation calculated from 3 technical replicates. Bottom panels: 18 Coomassie-stained SDS-PAGE gel of cross-linked Smc protein. Solvent only (‘DMSO’), cross- 19 linker only (‘M3M’), cross-linker with subsequent treatment with reducing agent (‘DTT’). 20 D. ScpAB dependence of the ATP hydrolysis rates of cross-linked Smc3SV(D193C). Hydrolysis 21 rates were measured at 0.15 µM protein concentration in the presence and absence of

22 stoichiometric amounts of ScpAB and 3 µM dsDNA40. Non-crosslinked control (left panel), 23 M3M cross-linked proteins (right panel). Lines represent non-linear regression fits to the Hill- 24 model. 25

29 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

ABClosed Intermediary PDB: 5NNV Nt 1.9 Å E295 S294

N Nt 4

E295

2.4 Å M3M E295 S294 ATP 8787.8.8 % XXL C Smc3SV(E295C) 44.6060 SD

) 30 -1

N E295 4 25 XL-Smc (min 20 15 10 Nt 5 Smc

ATP hydrolysis rate 0 Ct DMSO M3M M3M DMSO M3M M3M DTT DTT

Figure 3 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

1 Figure 3. Cross-linking of E295C at the 4N arm contact enhances ATP hydrolysis. 2 A. Schematic of arm contacts in the Smc dimer in closed and intermediary conformations. The 3 position of E295 is indicated as black dot and the 4N arm contact is denoted by a red bracket. 4 ATP molecules are represented as purple dots. 5 B. Structural organization of the 4N arm contact in Bsu Smc (PDB: 5NNV) in cartoon 6 representation. Hydrogen bonds are indicated as dotted lines in yellow colours. 7 C. ATP hydrolysis rates and SDS-PAGE of cross linked Smc(E295C). Display as in Figure 2C.

A 270 280 290 300 G302 310 320 Bsu Gst Spn Sau B hinge C Ct Nt Nt Ct NtNt G302* smc wt ∆ A V S M Q E R K W F

G302 ONAO CBB 90 º N 4 SSMG

heads Smc4S(S152C)-HaloTag Smc4S(A715C)-HaloTag D E wt SR KI GW GE wt SR EQ GW GE G302* A715C smc wt ∆ GE GW Smc-HT XL

Smc-HT α-Smc G302* 60

50 40 40 S152C 30 30 20 20 CBB 10 10 Crosslinking efficency (%) Crosslinking efficency 0 (%) Crosslinking efficency 0 FG 0.4 α-Smc 2.0 α-Smc parS-359 parS-359 dnaA dnaA ter ter 0.3 1.5 input) input) 0.2 1.0

0.1 0.5 ChIP q-PCR (% ChIP q-PCR (%

0.0 0.0 WT ∆smc GW GE Smc(EQ) + + GW GW GE GE ScpA + ∆ + ∆ + ∆

Figure 4 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

1 Figure 4. Residue G302 at the 4N arm contact is essential for Smc function. 2 A. Alignment of 4N arm contact sequences from selected firmicute species (Bsu, Bacillus 3 subtilis; Gst, Geobacillus stearothermophilus; Spn, Streptococcus pneumoniae; Sau, 4 Staphylococcus aureus). Residues G302 is indicated. 5 B. The position of residue G302 at the 4N arm contact (PDB: 5NNV). Left panel: schematic of 6 arm contact zones as in Figure 2B; right panel: structural view of the 4N arm contact from Bsu 7 Smc in cartoon representation in front and side views. Colours as in Figure 3B. G302 is shown 8 in red colours. 9 C. Colony formation by Bsu strains with single-amino acid substitution in G302. 92-fold and 95- 10 fold dilutions of stationary phase cultures were spotted on nutrient-rich medium (ONA) and 11 nutrient-poor minimal glucose-glutamate medium (SMG) and grown at 37°C for 16 and 24 h, 12 respectively. 13 D. Smc protein levels in mutants G302E (GE) and G302W (GW) grown in SMG medium 14 determined by immunoblotting with polyclonal antibodies raised against Bsu Smc protein. 15 Coomassie staining (CBB) of cell extracts on separate gels shown as control for uniform 16 protein extraction. 17 E. Quantification of cross-linked species of Smc4S-HaloTag (C119S, C437S, C826S, C1114S) 18 harbouring G302 mutations (labelled with TMR-HaloTag ligand). Cysteine pairs were 19 introduced at reporter positions S152 (left panels) and A715 (right panels). Cross-linking 20 efficiencies were calculated by in-gel fluorescence quantification of band intensities of non- 21 crosslinked (Smc-HT) and crosslinked (Smc-HT XL) species. Efficiencies were compared 22 between wild type (wt), the ABC signature motif mutant S1090R (SR), the ATP-binding mutant 23 K37I (KI), G302W (GW), and G302E (GE) mutants. 24 F. Chromatin-immunoprecipitation coupled to quantitative PCR (ChIP-qPCR) in Bsu cells using 25 α-Smc antiserum. Selected loci near the replication origin and at the terminus regions were 26 analysed. Dots represent data points from one out of two biological replicates. Mean values 27 are indicated as bars. 28 G. Chromatin-immunoprecipitation coupled to quantitative PCR (ChIP-qPCR) using α-Smc 29 antiserum. As in (F) for strains carrying the E1118Q (EQ) mutation and lacking or having scpA. 30

A C BMOE BM2 BM3 BM11 Disengaged (1)

8 Å 14.7 Å 17.8 Å 62.3 Å Smc4S(K1151C)-HaloTag D193 7 Å 90 º wt K37I E1118Q K1151 43.0 Å

ATP-pre-engaged (2) D193 11 Å

90 º

30 30 30 25 25 25 K1151 20 20 20 20.0 Å 15 15 15 10 10 10 ATP-engaged (3)

o 5 5 5 r r D193 Crosslinking efficency (%) Crosslinking efficency (%) Crosslinking efficency Cr Crosslinking efficency (%) Crosslinking efficency Cro C 0 0 0 2 3 11 2 3 11 2 3 111 77 Å OOE M BBM BMB BMB M BM BM BM M BMB BM BMB BMOE BMOEBBM BMOEBMB

(1) (2) (1) (1) (2) (3) 90 º K1151 11.6 Å

~50 % ~50 %100 % ~50 % ~25 % ~25 % (estimated) (assumed) (estimated)

B Smc4S(K1151C)-HaloTag GE, EQ 204Ω/996Ω, EQ mini-Smc(EQ) G302* 204Ω/996Ω

mini-Smc

4455 (estimated) ~80 % 50 50 4400 (assumed) 100 % (estimated) ~75 % 45 45 3355 40 40 3300 35 35

ce 2255 30 30

ffic 996Ω ef 2200 25 25 1155 20 20 1100 204Ω 15 15 5 10 10 5 5 0 Crosslinking efficency (%) Crosslinking efficency effic e Crosslinking efficency (%) Crosslinking efficency Crosslinking efficency (%) Crosslinking efficency 2 3 (%) Crosslinking efficency 0 0 11 2 3 2 3 11 11 BM BM BM BMOE BM BM BM BM BM BM BMOE BMOE Figure 5 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

1 Figure 5. Smc head cross-linking in vivo. 2 A. Top panel: Chemical structures of bis-maleimido crosslinkers with variable spacer arms 3 (BMOE, bismaleimidoethane; BM2, 1,8-bismaleimido-diethyleneglycol; BM3, 1,11- 4 bismaleimido-triethyleneglycol; BM11, Bis-MAL-dPEG11). Bottom panels: Quantification of 5 cross-linked species of Smc4S harbouring the K1151C cysteine pair reporter using different 6 crosslinkers. Crosslinking efficiencies were calculated as described for Figure 4E and 7 compared between wt (left panel), K37I (middle panel) and E1118Q (right panel) variants. 8 Error bars were calculated as standard deviation from 3 biological replicates. The values and 9 schematics shown below the graphs indicate the approximately estimated relative abundance 10 of different conformations, based on assumed abundances for the K37I mutant and the 11 Smc(204Ω/996Ω, EQ) protein (panel B). 12 B. Smc(K1151C) crosslinking in strains harbouring arm modifications. G302E (GE) (left panel), 13 unstructured peptide insertions 204Ω/996Ω (middle panel) (Diebold-Durand et al., 2017) and 14 mini-Smc CC298 (Bürmann et al., 2017) (right panel). Representation and quantification as in 15 (A). 16 C. Structural models of different ATPase states. Model for the disengaged state (1) as 17 described in (Diebold-Durand et al., 2017). The ATP-engaged state (3) is based on a crystal 18 structure (PDB: 5XG3). The ATP-pre-engaged state (2) is manually built to aligning ABC motifs 19 without opening the arms. Individual subunits are coloured in light or dark blue colours, 20 respectively. D193 and K1151 inter-subunit Cα-Cα distances are shown as a reference for the 21 different conformations. 22

32 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

A Closed Smc4S-HaloTag, BM3

K1151C ++- K1151C + - + Q320C - ++Q320C - + +

K1151C XL (measured) 50 Q320C XL (measured) Double XL (measured) Q320C Double XL (expected) 40

Crosslinking efficency (%) Crosslinking efficency 10 K1151C Disengaged ATP-pre- state engaged 0 state

B IntermediaryIntermediary

Basal ATPase DNA-stimulated ATPase Route 1

ClosedClosed IIntIntermediaryermedidiary OpeOpenp n

Opening ATPATA bbibindinginnding

4N contactc t t detachment 4N contactcocon t ApoAp

ClosedClosed Route 2

Apo, ATPATATPTP binding Opening disengageddisdi enggageg d ATPATP-engaged-enengaged gagedd ATPATP-engaged-engageden gagedd

ATP-pre-engagedAATP-pre-engaged

Figure 6 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

1 Figure 6. 2 A. Simultaneous cross-linking of Bsu Smc heads and arms in vivo. Left panel: Schematics of 3 Smc-ScpAB in closed conformations with disengaged or ATP-pre-engaged heads. Cysteine 4 reporters Q320C and K1151C are indicated as dots in black and gray colours, respectively. 5 Right panel: Quantification of single and double cross-linking efficiencies in Smc4S-HaloTag 6 with Q320C and/or K1151C. The grey arrowhead denotes the species cross-linked only at 7 Q320C; species only cross-linked at K1151C (black colours); species cross-linked at Q320C and 8 K1151C (blue colours). Individual data points are shown as white circles from two biological 9 replicates. Average values are shown as bars (same colour coding as for arrowheads). Data is 10 also shown in Figure S6. 11 B. Model of the conformational landscapes and multi-step opening of Smc-ScpAB. Two 12 alternative pathways in the basal ATPase. Route 1: Spontaneous separation of the juxtaposed 13 heads leads to an intermediary conformation with constraint arms, which is then stabilized 14 by ATP binding and head engagement. Route 2: ATP binding in the closed conformation 15 generates ATP-pre-engaged heads, which are constraint by the arms. Partial arm opening and 16 head engagement result in the intermediary conformation. Optional: Full opening of the arms 17 produces the DNA-stimulated ATPase.

1 References 2 Anderson, D.E., Losada, A., Erickson, H.P., and Hirano, T. (2002). Condensin and cohesin 3 display different arm conformations with characteristic hinge angles. The Journal of cell 4 biology 156, 419-424. 5 Aragón, L. (2018). The Smc5/6 Complex: New and Old Functions of the Enigmatic Long- 6 Distance Relative. Annual review of genetics 52, 89-107. 7 Ba, Z., Lou, J., Ye, A.Y., Dai, H.Q., Dring, E.W., Lin, S.G., Jain, S., Kyritsis, N., Kieffer- 8 Kwon, K.R., Casellas, R., et al. (2020). CTCF orchestrates long-range cohesin-driven V(D)J 9 recombinational scanning. Nature. 10 Bürmann, F., Basfeld, A., Vazquez Nunez, R., Diebold-Durand, M.L., Wilhelm, L., and 11 Gruber, S. (2017). Tuned SMC Arms Drive Chromosomal Loading of Prokaryotic Condensin. 12 Molecular cell 65, 861-872.e869. 13 Bürmann, F., and Gruber, S. (2015). SMC condensin: promoting cohesion of replicon arms. 14 Nature structural & molecular biology 22, 653-655. 15 Bürmann, F., Lee, B.G., Than, T., Sinn, L., O'Reilly, F.J., Yatskevich, S., Rappsilber, J., Hu, 16 B., Nasmyth, K., and Löwe, J. (2019). A folded conformation of MukBEF and cohesin. Nature 17 structural & molecular biology 26, 227-236. 18 Bürmann, F., Shin, H.C., Basquin, J., Soh, Y.M., Giménez-Oya, V., Kim, Y.G., Oh, B.H., and 19 Gruber, S. (2013). An asymmetric SMC-kleisin bridge in prokaryotic condensin. Nature 20 structural & molecular biology 20, 371-379. 21 Chapard, C., Jones, R., van Oepen, T., Scheinost, J.C., and Nasmyth, K. (2019). Sister DNA 22 Entrapment between Juxtaposed Smc Heads and Kleisin of the Cohesin Complex. Molecular 23 cell 75, 224-237.e225. 24 Collier, J.E., Lee, B.G., Roig, M.B., Yatskevich, S., Petela, N.J., Metson, J., Voulgaris, M., 25 Gonzalez Llamazares, A., Löwe, J., and Nasmyth, K.A. (2020). Transport of DNA within 26 cohesin involves clamping on top of engaged heads by Scc2 and entrapment within the ring by 27 Scc3. eLife 9. 28 Diebold-Durand, M.L., Lee, H., Ruiz Avila, L.B., Noh, H., Shin, H.C., Im, H., Bock, F.P., 29 Bürmann, F., Durand, A., Basfeld, A., et al. (2017). Structure of Full-Length SMC and 30 Rearrangements Required for Chromosome Organization. Molecular cell 67, 334-347.e335. 31 Earnshaw, W.C., and Laemmli, U.K. (1983). Architecture of metaphase chromosomes and 32 chromosome scaffolds. The Journal of cell biology 96, 84-93. 33 Eeftens, J.M., Katan, A.J., Kschonsak, M., Hassler, M., de Wilde, L., Dief, E.M., Haering, 34 C.H., and Dekker, C. (2016). Condensin Smc2-Smc4 Dimers Are Flexible and Dynamic. Cell 35 reports 14, 1813-1818. 36 El Mkami, H., and Norman, D.G. (2015). EPR Distance Measurements in Deuterated Proteins. 37 Methods in enzymology 564, 125-152. 38 Fuentes-Perez, M.E., Gwynn, E.J., Dillingham, M.S., and Moreno-Herrero, F. (2012). Using 39 DNA as a fiducial marker to study SMC complex interactions with the atomic force 40 microscope. Biophysical journal 102, 839-848. 41 Gibcus, J.H., Samejima, K., Goloborodko, A., Samejima, I., Naumova, N., Nuebler, J., 42 Kanemaki, M.T., Xie, L., Paulson, J.R., Earnshaw, W.C., et al. (2018). A pathway for mitotic 43 chromosome formation. Science (New York, N.Y.) 359. 44 Gligoris, T.G., Scheinost, J.C., Bürmann, F., Petela, N., Chan, K.L., Uluocak, P., Beckouët, F., 45 Gruber, S., Nasmyth, K., and Löwe, J. (2014). Closing the cohesin ring: structure and function 46 of its Smc3-kleisin interface. Science (New York, N.Y.) 346, 963-967. 47 Griese, J.J., and Hopfner, K.P. (2011). Structure and DNA-binding activity of the Pyrococcus 48 furiosus SMC protein hinge domain. Proteins 79, 558-568.

1 Gruber, S., and Errington, J. (2009). Recruitment of condensin to replication origin regions by 2 ParB/SpoOJ promotes chromosome segregation in B. subtilis. Cell 137, 685-696. 3 Gruber, S., Veening, J.W., Bach, J., Blettinger, M., Bramkamp, M., and Errington, J. (2014). 4 Interlinked sister chromosomes arise in the absence of condensin during fast replication in B. 5 subtilis. Current biology : CB 24, 293-298. 6 Haering, C.H., Löwe, J., Hochwagen, A., and Nasmyth, K. (2002). Molecular architecture of 7 SMC proteins and the yeast cohesin complex. Molecular cell 9, 773-788. 8 Hassler, M., Shaltiel, I.A., and Haering, C.H. (2018). Towards a Unified Model of SMC 9 Complex Function. Current biology : CB 28, R1266-r1281. 10 Higashi, T.L., Eickhoff, P., Sousa, J.S., Locke, J., Nans, A., Flynn, H.R., Snijders, A.P., 11 Papageorgiou, G., O'Reilly, N., Chen, Z.A., et al. (2020). A Structure-Based Mechanism for 12 DNA Entry into the Cohesin Ring. Molecular cell 79, 917-933.e919. 13 Hirano, M., Anderson, D.E., Erickson, H.P., and Hirano, T. (2001). Bimodal activation of SMC 14 ATPase by intra- and inter-molecular interactions. The EMBO journal 20, 3238-3250. 15 Hirano, M., and Hirano, T. (2006). Opening closed arms: long-distance activation of SMC 16 ATPase by hinge-DNA interactions. Molecular cell 21, 175-186. 17 Hopfner, K.P. (2016). Invited review: Architectures and mechanisms of ATP binding cassette 18 proteins. Biopolymers 105, 492-504. 19 Jalal, A.S.B., and Le, T.B.K. (2020). Bacterial chromosome segregation by the ParABS system. 20 Open biology 10, 200097. 21 Jeschke, G. (2012). DEER distance measurements on proteins. Annual review of physical 22 chemistry 63, 419-446. 23 Jeschke, G. (2018). The contribution of modern EPR to structural biology. Emerging Topics in 24 Life Sciences 2, 9-18. 25 Jeschke, G., Chechik, V., Ionita, P., Godt, A., Zimmermann, H., Banham, J., Timmel, C., 26 Hilger, D., and Jung, H. (2006). DeerAnalysis2006—a comprehensive software package for 27 analyzing pulsed ELDOR data. Applied Magnetic Resonance 30, 473-498. 28 Kamada, K., Su'etsugu, M., Takada, H., Miyata, M., and Hirano, T. (2017). Overall Shapes of 29 the SMC-ScpAB Complex Are Determined by Balance between Constraint and Relaxation of 30 Its Structural Parts. Structure (London, England : 1993) 25, 603-616.e604. 31 Kazmier, K., Sharma, S., Islam, S.M., Roux, B., and McHaourab, H.S. (2014). Conformational 32 cycle and ion-coupling mechanism of the Na+/hydantoin transporter Mhp1. Proceedings of the 33 National Academy of Sciences of the United States of America 111, 14752-14757. 34 Lammens, A., Schele, A., and Hopfner, K.P. (2004). Structural biochemistry of ATP-driven 35 dimerization and DNA-stimulated activation of SMC ATPases. Current biology : CB 14, 1778- 36 1782. 37 Lee, B.G., Merkel, F., Allegretti, M., Hassler, M., Cawood, C., Lecomte, L., O'Reilly, F.J., 38 Sinn, L.R., Gutierrez-Escribano, P., Kschonsak, M., et al. (2020). Cryo-EM structures of holo 39 condensin reveal a subunit flip-flop mechanism. Nature structural & molecular biology 27, 40 743-751. 41 Marko, J.F., De Los Rios, P., Barducci, A., and Gruber, S. (2019). DNA-segment-capture 42 model for loop extrusion by structural maintenance of chromosome (SMC) protein complexes. 43 Nucleic acids research 47, 6956-6972. 44 Marsden, M.P., and Laemmli, U.K. (1979). Metaphase chromosome structure: evidence for a 45 radial loop model. Cell 17, 849-858. 46 Merkenschlager, M., and Nora, E.P. (2016). CTCF and Cohesin in Genome Folding and 47 Transcriptional Gene Regulation. Annual review of genomics and human genetics 17, 17-43. 48 Minnen, A., Bürmann, F., Wilhelm, L., Anchimiuk, A., Diebold-Durand, M.L., and Gruber, S. 49 (2016). Control of Smc Coiled Coil Architecture by the ATPase Heads Facilitates Targeting to 50 Chromosomal ParB/parS and Release onto Flanking DNA. Cell reports 14, 2003-2016.

1 Muir, K.W., Li, Y., Weis, F., and Panne, D. (2020). The structure of the cohesin ATPase 2 elucidates the mechanism of SMC-kleisin ring opening. Nature structural & molecular biology 3 27, 233-239. 4 Naumova, N., Imakaev, M., Fudenberg, G., Zhan, Y., Lajoie, B.R., Mirny, L.A., and Dekker, 5 J. (2013). Organization of the mitotic chromosome. Science (New York, N.Y.) 342, 948-953. 6 Orlandini, E., Marenduzzo, D., and Michieletto, D. (2019). Synergy of topoisomerase and 7 structural-maintenance-of-chromosomes proteins creates a universal pathway to simplify 8 genome topology. Proceedings of the National Academy of Sciences of the United States of 9 America 116, 8149-8154. 10 Osorio-Valeriano, M., Altegoer, F., Steinchen, W., Urban, S., Liu, Y., Bange, G., and 11 Thanbichler, M. (2019). ParB-type DNA Segregation Proteins Are CTP-Dependent Molecular 12 Switches. Cell 179, 1512-1524.e1515. 13 Pannier, M., Veit, S., Godt, A., Jeschke, G., and Spiess, H.W. (2000). Dead-time free 14 measurement of dipole-dipole interactions between electron spins. Journal of magnetic 15 resonance (San Diego, Calif. : 1997) 142, 331-340. 16 Polyhach, Y., Bordignon, E., Tschaggelar, R., Gandra, S., Godt, A., and Jeschke, G. (2012). 17 High sensitivity and versatility of the DEER experiment on nitroxide radical pairs at Q-band 18 frequencies. Physical chemistry chemical physics : PCCP 14, 10762-10773. 19 Reginsson, G.W., and Schiemann, O. (2011). Pulsed electron-electron double resonance: 20 beyond nanometre distance measurements on biomacromolecules. The Biochemical journal 21 434, 353-363. 22 Ryu, J.-K., Katan, A.J., van der Sluis, E.O., Wisse, T., de Groot, R., Haering, C., and Dekker, 23 C. (2019). AFM images of open and collapsed states of yeast condensin suggest a scrunching 24 model for DNA loop extrusion. bioRxiv, 2019.2012.2013.867358. 25 Shi, Z., Gao, H., Bai, X.C., and Yu, H. (2020). Cryo-EM structure of the human cohesin- 26 NIPBL-DNA complex. Science (New York, N.Y.) 368, 1454-1459. 27 Soh, Y.M., Bürmann, F., Shin, H.C., Oda, T., Jin, K.S., Toseland, C.P., Kim, C., Lee, H., Kim, 28 S.J., Kong, M.S., et al. (2015). Molecular basis for SMC rod formation and its dissolution upon 29 DNA binding. Molecular cell 57, 290-303. 30 Soh, Y.M., Davidson, I.F., Zamuner, S., Basquin, J., Bock, F.P., Taschner, M., Veening, J.W., 31 De Los Rios, P., Peters, J.M., and Gruber, S. (2019). Self-organization of parS centromeres by 32 the ParB CTP hydrolase. Science (New York, N.Y.) 366, 1129-1133. 33 Sullivan, N.L., Marquis, K.A., and Rudner, D.Z. (2009). Recruitment of SMC by ParB-parS 34 organizes the origin region and promotes efficient chromosome segregation. Cell 137, 697- 35 707. 36 Terakawa, T., Bisht, S., Eeftens, J.M., Dekker, C., Haering, C.H., and Greene, E.C. (2017). 37 The condensin complex is a mechanochemical motor that translocates along DNA. Science 38 (New York, N.Y.) 358, 672-676. 39 Tran, N.T., Laub, M.T., and Le, T.B.K. (2017). SMC Progressively Aligns Chromosomal Arms 40 in Caulobacter crescentus but Is Antagonized by Convergent Transcription. Cell reports 20, 41 2057-2071. 42 Vazquez Nunez, R., Ruiz Avila, L.B., and Gruber, S. (2019). Transient DNA Occupancy of 43 the SMC Interarm Space in Prokaryotic Condensin. Molecular cell 75, 209-223.e206. 44 Vercellino, I., Rezabkova, L., Olieric, V., Polyhach, Y., Weinert, T., Kammerer, R.A., Jeschke, 45 G., and Korkhov, V.M. (2017). Role of the nucleotidyl cyclase helical domain in catalytically 46 active dimer formation. Proceedings of the National Academy of Sciences of the United States 47 of America 114, E9821-e9828. 48 Wang, X., Brandão, H.B., Le, T.B., Laub, M.T., and Rudner, D.Z. (2017). Bacillus subtilis 49 SMC complexes juxtapose chromosome arms as they travel from origin to terminus. Science 50 (New York, N.Y.) 355, 524-527.

1 Wang, X., Hughes, A.C., Brandão, H.B., Walker, B., Lierz, C., Cochran, J.C., Oakley, M.G., 2 Kruse, A.C., and Rudner, D.Z. (2018). In Vivo Evidence for ATPase-Dependent DNA 3 Translocation by the Bacillus subtilis SMC Condensin Complex. Molecular cell 71, 841- 4 847.e845. 5 Wang, X., Le, T.B., Lajoie, B.R., Dekker, J., Laub, M.T., and Rudner, D.Z. (2015). Condensin 6 promotes the juxtaposition of DNA flanking its loading site in Bacillus subtilis. Genes & 7 development 29, 1661-1675. 8 Wang, X., Tang, O.W., Riley, E.P., and Rudner, D.Z. (2014). The SMC condensin complex is 9 required for origin segregation in Bacillus subtilis. Current biology : CB 24, 287-292. 10 Wilhelm, L., Bürmann, F., Minnen, A., Shin, H.C., Toseland, C.P., Oh, B.H., and Gruber, S. 11 (2015). SMC condensin entraps chromosomal DNA by an ATP hydrolysis dependent loading 12 mechanism in Bacillus subtilis. eLife 4. 13 Yatskevich, S., Rhodes, J., and Nasmyth, K. (2019). Organization of Chromosomal DNA by 14 SMC Complexes. Annual review of genetics 53, 445-482. 15 Yoshimura, S.H., Hizume, K., Murakami, A., Sutani, T., Takeyasu, K., and Yanagida, M. 16 (2002). Condensin architecture and interaction with DNA: regulatory non-SMC subunits bind 17 to the head of SMC heterodimer. Current biology : CB 12, 508-513. 18 Zhao, S., and Fernald, R.D. (2005). Comprehensive algorithm for quantitative real-time 19 polymerase chain reaction. Journal of computational biology : a journal of computational 20 molecular cell biology 12, 1047-1064. 21

37 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Supplemental data

bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in A 4000 1500 B Smc3S(D193C, EQ) perpetuity.Smc3S(R718C, It is EQ) made available under aCC-BY 4.0 International licenseD2O. buffer 3000 1000 400 2000 15 - DNA 300 500 1000 ) + dsDNA40 200

0 -1 0 100

-1000 0 -500

(min -100

-2000 Intensity (A.U.)

-200 -3000 -1000 10 5000 Smc3S(E217C, EQ) Smc3S(R643C, EQ) -300 4000 3300 3325 3350 3375 3400 3425 3000 Magnetic field (G) 2000 500 1000 0 -1000 0 5 -2000 -3000

Intensity (A.U.) -4000 -500 -5000 ATP rate hydrolysis -6000 1500 3S 6000 TEMPOL Smc (EQ)

1000 4000 0 3S 3S 2000 Smc (D193C) Smc (D193C) +MTSL 500 D 0 3S 0 Smc (EQ)-ScpAB -2000 0.025 0.030 0.050

-4000 -500 D193C E217C R643C 0.020 0.025 -6000 0.040 -1000 ~55 % 3300 3325 3350 3375 3400 3425 3300 3325 3350 3375 3400 3425 0.020 0.015 ~69 % 0.030 Magnetic field (G) Magnetic field (G) ~46 % >82 % apo 0.015 ~70 % P(r) >85 % 0.010 apo 0.020 apo ATP C Smc3S(EQ)-ScpAB 0.010 ATP ATP 0.005 0.010 1.0 0.025 0.005 0.000 D193C 0.000 0.000 0.020 23456789 2345678910 23456789 Distance (nm) Distance (nm) Distance (nm) 0.8 apo apo 0.015 Smc3S(EQ) (without ScpAB) ATP dsDNA40 ATP dsDNA40 E 0.06 0.030 P(r) 0.010 R643C R718C V(t)/V(0) 0.6 0.025 0.005 0.04 0.020 ~83 % ~83 % 0.015 ~74 % 0.000 0.4 0.02 0.010 ~20 % 01234567 2345678910 apo Time (μs) Distance (nm) ATP dsDNA40 0.005 0.030 1.0 0.030 P(r) E217C D193C P(r) 0.025 0.030 0.025 E217C 0.020 0.8 0.020 0.020 ~66 % ~53 % 0.015 ~44 % 0.015 ~18 % 0.010 0.010 P(r) 0.010 0.005

V(t)/V(0) 0.6 0.000 0.000 0.005 23456789 2345678910 Distance (nm) Distance (nm) 0.000 F 0.4 2345678910 Smc3S-ScpAB 01234567 1.0 0.020 Time (μs) Distance (nm) D193C apo ATP dsDNA 1.0 0.005 0.015 40 R718C 0.9 0.9 0.004 0.010 P(r) ~64 % 0.8 V(t)/V(0) 0.005 ~53 % 0.8 0.003

P(r) 0.7 0.000 0123456 V(t)/V(0) 0.7 0.002 2345678910 Time (μs) Distance (nm)

0.6 0.001 G Smc3S(D193C, EQ) apo (no ScpAB) 0.000 1.0 1.0 1.0 0.5 Free fit Two Gausian fit 012345 2345678910 Time (μs) 0.8 1.0 0.05 Distance (nm) 0.8 0.8 R643C 0.6 0.6 F(t)/F(0)

0.6 F(t)/F(0) 0.9 0.04 V(t)/V(0)

0.03 0.4 0.4 0.8 0.401234567 01234567 01234567 Time (μs) Time (μs) 0.0100 Time (μs) 0.02 0.035 0.7 P(r) 0.030

V(t)/V(0) 0.0075 0.025 0.01 0.6 0.020 0.0050 P(r) P(r) 0.015 0.00 0.5 0.0025 0.010 012345 2345678910 0.005 Time (μs) Distance (nm) 0.0000 0.000 23456789 23456789 Figure S1 Distance (nm) Distance (nm) bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Figure S1. EPR-DEER analysis of Smc3S variants. A. Continuous wave (CW) EPR spectra for Smc3S(EQ) single cysteine mutant proteins. Spin labelling efficiency for all mutants was estimated by comparing double integrals of EPR signals with the concentration standard (200 µM water solution of TEMPOL). Traces were obtained directly from the main peak elution fraction after MTSL excess removal (see methods). B. ATP hydrolysis rate of Smc3S(D193C) at 150 nM protein concentration and stoichiometric

amounts of ScpAB2, in the presence of 1 mM ATP and with or without 3 µM dsDNA40.

Measurements were done in deuterated environment (D2O) and with 40 % dEthylene glycol in order to mimic the DEER measurement conditions. Inset shows the CW-EPR spectrum as in (A). C. Primary DEER traces obtained from Q band measurements (V(t)/V(0)) in the apo (black colours) and in the ATP-, DNA-bound state (gray colours). Straight black lines correspond to the estimated background decay. Right panels correspond the two-Gaussian fit probability distributions as shown in Figures 1B and 1D. D. Probability distributions of spin pair distances in single cysteine mutants of Smc3S(EQ) in the presence of ScpAB and the presence (blue colors) and absence (black colors) of 1 mM ATP. Shown as in Figure 1B and 1D. E. Probability distributions of spin pair distances in single cysteine mutants in the absence of ScpAB and the presence (blue colors) and absence (black colors) of 1 mM ATP and 3 µM

dsDNA40. Shown as in Figure 1B and 1D. F. Primary DEER traces (left panel) and probability distributions (right panel) of spin pair distances in Smc3S(D193C) in the apo state (black colours) and ATP- and DNA-bound state (grey colours). Shown as in (C). G. Primary DEER data from Smc3S(D193C-EQ) apo in the absence of ScpAB. The primary trace (left panel) was corrected for background (F(t)/F(0)) and model-free fitted with the Tikhonov regularization to provide the distance distribution shown in blue colours in the lower panel. Additionally, a two-Gaussian fit provided the distance distribution shown as a gray area in the lower panel. Given the good general agreement and the difficult of Tikhonov regularization in dealing with very narrow and very broad features simultaneously, two-Gaussian fits were used for all other mutants and conditions.

bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review)50.6 is the author/funder,% XL who has granted bioRxiv a license to display the preprint in A Smc3S(A715C) % XL perpetuity. It is made1.77 availableSD under aCC-BY 4.0 International license. 63.3 Smc3S(Q320C) SD 22 12 0.31 ) -1 20 ) * 18 -1 10 (min

16 (min 14 8 12 10 6 8 6 4 4 ATP hydrolysis rate 2

2 ATP hydrolysis rate 0 DMSO M3M M3M DMSO M3M M3M 0 DMSO M3M M3M DMSO M3M M3M DTT DTT DTT DTT 94.3 % XL 55.8 % XL Smc3S(Y944C) 0.17 SD 3S 16 Smc (D193C) 0.92 SD ) -1 14 - dsDNA * ) 12 -1

(min 12 + dsDNA * 10 10 (min 8 8 6 6 4 4

ATP hydrolysis rate 2 2 0 ATP hydrolysis rate DMSO M3M M3M DMSO 0 M3M M3M DMSO M3M M3M DMSO M3M M3M DTT DTT DTT DTT B Smc4C-ScpAB Smc3SV(Y944C)-ScpAB + M3M Smc3SV(Y944C)-ScpAB + DMSO 0.08 0.08 0.08

0.06 0.06 0.06

0.04 0.04 0.04 Anisotropy Anisotropy Anisotropy - DTT + DTT - DTT + DTT - DTT + DTT 0.02 0.02 0.02

0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 [Smc-ScpAB] (μM) [Smc-ScpAB] (μM) [Smc-ScpAB] (μM)

Figure S2 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Figure S2. ATPase activity and DNA binding of M3M-cross-linked Smc3S variants. A. Samples were prepared as described for Figure 2A. ATPase activity was determined with 1 mM ATP and measured in absence (black colours) and presence (green colours) of 3 µM

dsDNA40. Error bars show the standard deviation from 3 technical replicates. Band intensities of CBB gel were quantified and M3M crosslink efficiency was determined. Mean and standard deviations are shown above each well calculated from 3 technical replicates. Bands derived from off-target cysteine cross-linking are labelled with asterisks. B. DNA binding of M3M-cross-linked Smc3S variants as measured by fluorescence anisotropy with increasing protein concentration using fluorescein-labeled 40-bp dsDNA at 50 nM concentration. Data points from three experiments are shown as dots. The lines correspond to a nonlinear regression fit of the experimental data. Smc(Y944C) protein as in Figure 2C and also including wt Smc-ScpAB. Reduction of the M3M cross-link by addition of DTT had little effect on DNA binding.

bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in A perpetuity. It is madeB available under aCC-BY 4.0 International license. 3SV 3SV Smc(E295C) 30 Smc (E295C) 30 Smc (E295C), +M3M 35 Smc

) ) 25 )

-1 -1 -1 25 30 Smc, dsDNA40 (min (min (min 25 * 20 Smc-ScpAB 20

XL Smc-ScpAB, dsDNA40 20 * 15 15 15 10 10 10 5 5 ATP hydrolysis rate ATP hydrolysis rate ATP hydrolysis rate 5 Smc 0 0 0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 DMSO M3M M3M DMSO M3M M3M DTT DTT ATP (mM) ATP (mM)

Figure S3 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Figure S3. M3M crosslinking and ATPase activity of the Smc3S(E295C) variant. A. M3M crosslinking and ATPase activity of Smc3S(E295C) as in Figure S2A. B. ATP hydrolysis rates for Smc3SV(E295C) at variable ATP concentrations. As in Figure 2D.

bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. ABG302 mutation Smc3S(GW, EQ)-ScpAB smc wt ∆ wt A V S M Q R K D193C parB wt wt ∆∆ ∆∆ ∆ ∆∆ ∆ 1.0 0.010 apo

ATP dsDNA40 0.008 0.8 ONA 0.006

P(r) ~100 % 0.004

V(t)/V(0) 0.6 ~85 % 0.002

0.4 0.000 SMG 0123456 2345678910 Distance (nm) Distance (nm)

C wtG302W G302E

25 25 25

20 Smc Smc-ScpAB 20 20 Smc, dsDNA40 Smc-ScpAB, dsDNA40 15 15 15

10 10 10

5 5 5 ATP hydrolysis rate (min-1) ATP hydrolysis rate (min-1) ATP hydrolysis rate (min-1) 0 0 0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 ATP (mM) ATP (mM) ATP (mM)

D Cα-Cα distances S. cerevisiae condensin B. subtilis Smc A404-G559 A408-L563 PDB:6YVU PDB:5NNV N411-E567 L415-S570 6N K418-L574 M422-E578 L426-L581

0 2 4 6 8 10 12 Cα-Cα distance (Å) S812-V922 R816-L995 D819-K999 A737-L909 N823-K1003 5C R733-E906 C E827-G1006 L731-S902 7 Q830-V1010 H726-M898 E834-Q1013 L723-M898 0 2 4 6 8 10 12 N313 0 2 4 6 8 Cα-Cα distance (Å) 10 12 E309 Cα-Cα distance (Å) D328-K486 E305 E324-S482 S320-K479 G302 N K317-E475 4 E298 N S313-V471 E295 4 L309-E467 L291 N305-R464 Q287 0 2 4 6 8 D284 10 12 Cα-Cα distance (Å) N 0 2 4 6 8 8 10 12 T896-M1075 Cα-Cα distance (Å) K900-S1078 D903-F1082 C S907-E1086 3 L910-M1089 K914-L1093 D918-L1097 0 2 4 6 8 10 12 Cα-Cα distance (Å) K974-S1212 D977-D1216 2C K980-D1219

0 2 4 6 8 10 12 Cα-Cα distance (Å) L199-F361 N195-K357 N K191-V353 1 K188-L350 T184-Q346 0 2 4 6 8 Cα-Cα distance10 (Å)12

Figure S4 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Figure S4. A. Dilution series spotting for Bsu strains carrying G302 single amino acid substitutions as well as a deletion of the parB gene. As in Figure 4C. B. Primary DEER traces (left panel) and probability distributions (right panel) of spin pair distances in Smc3S(D193C, G302W, E1118Q) in the apo (black colours) and ATP- and DNA- bound state. Shown as in Figure S1C. C. Measurement of ATP hydrolysis rates for Smc protein (wt), Smc(G302W) and Smc(G302E)

with increasing concentrations of ATP and in the absence or presence of dsDNA40 and ScpAB. D. Cα-Cα distances measured at the inter-subunit interfaces. For S. cerevisiae condensin (PDB: 5YVU) (left panels). Smc2 (in blue colours), Smc4 (in pink colours) and Brn1 (in green colours) are shown in cartoon representation reported in the cryo-EM structure (Lee et al., 2020). Distances are displayed as bars in gray colours except for the 4N arm contact, which are shown in blue colours. Dashed lines indicate the shortest distance observed as a reference. For the Bsu Smc arm crystal structure (PDB: 5NNV) (right panels).

bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this A preprint (which4S was not certified by peer review) is the author/funder,E who has granted bioRxiv a4S license to display the preprint in Smc (K1151C)-HaloTagperpetuity. It is made available under aCC-BY 4.0 InternationalSmc license(K1151C,. E1118Q)-HaloTag wt E1118Q K37I S1090R Smc wt GW GE ScpA + + +

30 30 25 25 20 20 45 45 45 15 15 40 40 40 10 10 35 35 35 5 5 30 30 0 0 30 Crosslinking efficency (%) Crosslinking efficency Crosslinking efficency (%) Crosslinking efficency 25 2 3 2 3 25 25 2 3 11 2 3 11 11 11 20 20 20 BM BM BM BM BM BM BM BM BM BM BM BM DMSOBMOE DMSOBMOE DMSOBMOE DMSOBMOE 15 15 15 10 4S 10 10 B Smc (K1151C)-HaloTag 5 5 5

Crosslinking efficency (%) Crosslinking efficency 0 Mini-Smc Mini-Smc(EQ) 0 0 Crosslinking efficency (%) Crosslinking efficency 204Ω/996Ω 204Ω/996Ω, EQ (%) Crosslinking efficency 2 3 11 2 3 11 2 3 11 BM BM BM BM BM BM BMOE BM BMOE BM BMOE BM Smc wt GW GE ScpA ∆ ∆ ∆

50 50 45 45 40 40 35 35 30 30 25 25 45 45 45 20 20 40 40 40 15 15 35 35 35 10 10 5 5 30 30 30 25 25 25

Crosslinking efficency (%) Crosslinking efficency 0 0 Crosslinking efficency (%) Crosslinking efficency 2 3 11 2 3 11 2 3 11 2 3 11 20 20 20 BM BM BM BM BM BM BM BM 15 15 15 BMOE BM BMOE BM BMOE BM BMOE BM Smc4S(K1151C)-HaloTag 10 10 10 C 5 5 5 Smc wt GW GE 0 0 0 Crosslinking efficency (%) Crosslinking efficency Crosslinking efficency (%) Crosslinking efficency Crosslinking efficency (%) Crosslinking efficency 2 3 11 2 3 11 2 3 11 BM BM BM BM BM BM BMOE BM BMOE BM BMOE BM

25 25 25 20 20 20 15 15 15 10 10 10 5 5 5 Crosslinking efficency (%) Crosslinking efficency (%) Crosslinking efficency Crosslinking efficency (%) Crosslinking efficency 0 0 0 2 3 11 2 3 11 2 3 11 BM BM BM BM BM BM BMOE BM BMOE BM BMOE BM

D BMH Smc4S(K1151C)-HaloTag 13 Å Smc4S(K1151C)-HaloTag 10 wt EQKI SR

wt EQKI SR 8

2 Crosslinking efficency (%) Crosslinking efficency 0

Figure S5 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Figure S5. Quantification of in vivo cross-linking efficiencies. A. Cross-linking of Smc4S(K1151C)-HT and ATPase mutants thereof using different maleimide crosslinkers. Same data as in Figure 5A. B. Cross-linking of Smc4S(K1151C)-HT variants harboring arm modifications. Data also shown in Figure 5B. C. Cross-linking of Smc4S(K1151C)-HT harboring G302W or G302E mutations. Displayed as Figure 5A. D. Species of Smc4S(K1151C)-HT cross-linked with bis-maleimido-hexane (BMH). Displayed as Figure 5A. E. Cross-linking of Smc4S(E1118Q, K1151C)-HT variants in the presence (top panels) and absence (bottom panels) of the scpA gene. Displayed as Figure 5A.

Smc4S-HaloTag wt EQ wt EQ BMOE BM3 BMOE BM3 BMOE BM3 BMOE BM3 K1151C + -++ -++ -++ - + K1151C + - + + -++ -++ - + Q320C - ++- ++- ++- ++Q320C - ++ - ++- ++- ++ 60

10 Smc-HaloTag Crosslinking efficency (%) Crosslinking efficency 0 K1151C XL (measured) Double XL (measured) Q320C XL (measured) Double XL (expected)

Figure S6 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.21.427566; this version posted January 21, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

Figure S6. Double cysteine cross-linking in vivo. A. BMOE and BM3 cross-linking of Smc4S-HT and Smc4S(EQ)-HT variants harboring single cysteine reporters K1151C or Q320C or a combination of the two cysteines. Quantifications as in Figure 6B.

Table S1. Related to Figures 4-6. Strain usage. Genotypes are detailed in Table S5. Figure Strains 4C BSG1002, BSG1007, BSG4097, BSG4099, BSG4095, BSG4098, BSG4096, BSG4094, BSG4092, BSG4093, BSG4037, BSG4038 4D BSG1002, BSG1007, BSG4094, BSG4037 4E BSG2264, BSG2589, BSG2588, BSG4153, BSG4154, BSG1921, BSG2607, BSG1922, BSG4082, BSG4161 4F BSG1002, BSG1007, BSG4037, BSG4094 4G BSG1008, BSG1892, BSG4075, BSG4351, BSG4159, BSG4352 5A BSG1457, BSG1607, BSG1488 5B BSG4160, BSG2693, BSG2531 6A BSG1457, BSG4516, BSG4514, BSG1488, BSG4701, BSG4515 S4A BSG1002, BSG1007, BSG1050, BSG4110, BSG4112, BSG4108, BSG4111, BSG4109, BSG4106, BSG4107 S5A BSG1457, BSG1488, BSG1607, BSG1600 S5B BSG2694, BSG2693, BSG2511, BSG2531, BSG1457, BSG4076, BSG4158 S5C BSG1457, BSG1488, BSG1607, BSG1600 S5D BSG1488, BSG4077, BSG4160, BSG1509, BSG4168, BSG4170 S6A BSG1457, BSG1921, BSG4512, BSG1488, BSG1922, BSG4513 S6B BSG1457, BSG2300, BSG4510, BSG1488, BSG2313, BSG4511

Table S2. Related to Figure 1, S1 and S4. Mean (µ) and standard deviation (σ) of double gaussian fit to DEER traces. P(r)1 corresponds to the fitted fraction of population 1. µ1 (nm) σ1 P(r)1 µ2 σ2 r.m.s. (nm) Smc-ScpAB, apo D193C, E1118Q 1.76 0.24 0.55 4.94 1.43 0.013 D217C, E1118Q 3.97 0.19 0.69 8.53 0.80 0.006 R718C, E1118Q 1.5 0.21 0.71 3.06 1.93 0.007 R643C, E1118Q 1.5 0.18 0.82 5.3 1.00 0.008 D193C 1.5 0.40 0.64 5.40 1.45 0.016 D193C, G302W 1.93 0.095 0.15 5.57 1.22 0.009 Smc-ScpAB , ATP, dsDNA40 D193C, E1118Q 1.76 0.19 0.18 5.10 1.51 0.013 D217C, E1118Q 3.94 0.20 0.57 8.51 1.29 0.008 R718C, E1118Q 1.50 0.12 0.28 2.80 2.45 0.006 R643C, E1118Q 1.50 0.20 0.85 5.58 0.83 0.007 D193C 1.5 0.38 0.53 4.58 2.40 0.009 D193C, G302W 2 (fixed) 0.69 N/A 5.37 1.45 0.014 Smc, apo D193C, E1118Q 1.70 0.22 0.53 4.85 1.50 0.016 D217C, E1118Q 4.03 0.23 0.66 8.66 1.09 0.007 R718C, E1118Q 1.5 0.19 0.74 4.01 1.56 0.007 R643C, E1118Q 1.5 0.17 0.83 4.92 1.21 0.007 Smc, ATP, dsDNA40 D193C, E1118Q 1.66 0.20 0.18 4.60 1.86 0.016 D217C, E1118Q 3.99 0.25 0.44 8.45 1.74 0.009 R718C, E1118Q 1.5 0.20 0.20 3.95 2.16 0.005 R643C, E1118Q 1.5 0.17 0.83 5.03 1.21 0.007

Table S3. Related to Figure 2. Crosslinking efficiency of purified Smc3SV single cysteine mutants. Smc3SV Crosslink efficiency (%) SD cysless N/A N/A D193C 85.40 0.40 Y944C 86.47 1.57 Q320C 70.22 0.43 A715C 59.69 0.97

Table S4. Related to Figures 2 and 3. Kinetic parameters of the Smc ATPase.

-1 Construct Kcat (min ) K0.5 (mM) h Smc3SV(D193C) 3.9 ± 0.8 0.39 ± 0.016 1.3 ± 0.3 Smc3SV(D193C)-ScpAB 7.2 ± 0.1 0.13 ± 0.007 1.1 ± 0.05 3SV Smc (D193C) + 3 µM dsDNA40 7.0 ± 0.7 0.22 ± 0.05 0.9 ± 0.1 3SV Smc (D193C)-ScpAB + 3 µM dsDNA40 14.0 ± 0.1 0.12 ± 0.003 1.3 ± 0.03 Smc3SV(D193C) M3M 0.4 ± 0.4 0.36 ± 0.4 3.6 ± 10.0 Smc3SV(D193C)-ScpAB M3M 4.7 ± 0.1 0.21 ± 0.01 1.4 ± 0.08 3SV Smc (D193C) M3M + 3 µM dsDNA40 0.7 ± 0.08 0.16 ± 0.03 2.5 ± 1 3SV Smc (D193C)-ScpAB M3M + 3 µM dsDNA40 4.8 ± 0.1 0.18 ± 0.01 1.4 ± 0.1 Smc3SV(E295C) 6.5 ± 1.4 0.13 ± 0.08 0.9 ± 0.3 Smc3SV(E295C)-ScpAB 14.4 ± 4.3 0.38 ± 0.31 0.8 ± 0.2 3SV Smc (E295C) + 3 µM dsDNA40 6.8 ± 0.8 0.08 ± 0.02 1.1 ± 0.3 3SV Smc (E295C)-ScpAB + 3 µM dsDNA40 39.2 ± 35.5 2.4 ± 5.4 0.6 ± 0.1 Smc3SV(E295C) M3M 10.5 ± 0.5 0.01 ± 0.007 1.5 ± 0.9 Smc3SV(E295C)-ScpAB M3M 30.0 ± 1.4 0.14 ± 0.018 1.0 ± 0.09 3SV Smc (E295C) M3M + 3 µM dsDNA40 11.4 ± 0.4 0.02 ± 0.005 2.1 ± 1.1 3SV Smc (E295C)-ScpAB M3M + 3 µM dsDNA40 29.6 ± 3.2 0.14 ± 0.04 1.0 ± 0.1