Biochemical analysis of the yeast condensin Smc2/4 complex: an ATPase that promotes knotting of circular DNA.

To better understand the contributions that the structural maintenance of chromosome proteins (SMCs) make to condensin activity, we have tested a number of biochemical, biophysical, and DNA-associated attributes of the Smc2p-Smc4p pair from budding yeast. Smc2p and Smc4p form a stable heterodimer, the "Smc2/4 complex," which upon analysis by sedimentation equilibrium appears to reversibly self-associate to form heterotetramers. Individually, neither Smc2p nor Smc4p hydrolyzes ATP; however, ATPase activity is recovered by equal molar mixing of both purified proteins. Hydrolysis activity is unaffected by the presence of DNA. Smc2/4 binds both linearized and circular plasmids, and the binding appears to be independent of adenylate nucleotide. High mole ratios of Smc2/4 to plasmid promote a geometric change in circular DNA that can be trapped as knots by type II topoisomerases but not as supercoils by a type I topoisomerase. Binding titration analyses reveal that two Smc2/4-DNA-bound states exist, one disrupted by and one resistant to salt challenge. Competition-displacement experiments show that Smc2/4-DNA-bound species formed at even high protein to DNA mole ratios remain reversible. Surprisingly, only linear and supercoiled DNA, not nicked-circular DNA, can completely displace Smc2/4 prebound to a labeled, nicked-circular DNA. To explain this geometry-dependent competition, we present two models of DNA binding by SMCs in which two DNA duplexes are captured within the inter-coil space of an Smc2/4 heterodimer. Based on these models, we propose a DNA displacement mechanism to explain how differences in geometry could affect the competitive potential of DNA.


James E. Stray ‡ and Janet E. Lindsley §
From the Department of Biochemistry, University of Utah School of Medicine, Salt Lake City, Utah 84132-3201 To better understand the contributions that the structural maintenance of chromosome proteins (SMCs) make to condensin activity, we have tested a number of biochemical, biophysical, and DNA-associated attributes of the Smc2p-Smc4p pair from budding yeast. Smc2p and Smc4p form a stable heterodimer, the "Smc2/4 complex," which upon analysis by sedimentation equilibrium appears to reversibly self-associate to form heterotetramers. Individually, neither Smc2p nor Smc4p hydrolyzes ATP; however, ATPase activity is recovered by equal molar mixing of both purified proteins. Hydrolysis activity is unaffected by the presence of DNA. Smc2/4 binds both linearized and circular plasmids, and the binding appears to be independent of adenylate nucleotide. High mole ratios of Smc2/4 to plasmid promote a geometric change in circular DNA that can be trapped as knots by type II topoisomerases but not as supercoils by a type I topoisomerase. Binding titration analyses reveal that two Smc2/4-DNA-bound states exist, one disrupted by and one resistant to salt challenge. Competition-displacement experiments show that Smc2/4-DNA-bound species formed at even high protein to DNA mole ratios remain reversible. Surprisingly, only linear and supercoiled DNA, not nicked-circular DNA, can completely displace Smc2/4 prebound to a labeled, nicked-circular DNA. To explain this geometrydependent competition, we present two models of DNA binding by SMCs in which two DNA duplexes are captured within the inter-coil space of an Smc2/4 heterodimer. Based on these models, we propose a DNA displacement mechanism to explain how differences in geometry could affect the competitive potential of DNA.
Faithful segregation of the genetic material at cell division requires the regulated activities of SMC proteins and the complexes containing them. In Saccharomyces cerevisiae there are six SMC proteins, Smc1p-6p, which co-purify in pairs with either the cohesin complex (Smc1p and Smc3p), the condensin complex (Smc2p and Smc4p), or a recently described complex containing Smc5p and Smc6p (1)(2)(3)(4)(5). The SMCs comprise a ubiquitous family of proteins distinguished by the conservation of five secondary structure features. The N-and C-terminal globular domains contain, respectively, a Walker A and a Walker B nucleotide binding motif (6), which together make up a two-part ATP binding site. To each globular domain is connected a very long ␣ helix of high coiled-coil probability, and both of these helices join to a central globular "hinge" (7)(8)(9). Conservation of the coil motifs among SMCs led to early speculation that these proteins had the potential to assemble into dimers, tetramers, or polymeric networks of even higher order (10,11). It is now well established that the fundamental unit of SMC association is a dimer (12)(13)(14). Most bacterial genomes encode a single SMC, and consequently, only homodimers are formed. All eukaryotes have at least two SMCs, and both heterodimeric and homodimeric pairs have been observed among them (15,16). Electron microscopy analyses revealed that Bacillus subtilis SMC protein forms flexible V-shaped dimers, whose shape symmetry was unchanged by the addition of bulky terminal extensions (17), proving that SMCs form anti-parallel coiled-coils. Subsequent coil-interaction analyses of the Bacillus SMC (18) and S. cerevisiae Smc1p and Smc3p (19) has demonstrated that coiled-coils form by intramolecular pairing of helices within a single protomer and that hinge interactions alone are sufficient to stabilize the dimer. Taken together, these observations support the model of SMC dimer architecture that is depicted in Fig. 1A.
The crystal structure of the Pyrococcus furiosus Rad50 (an SMC family member) N-terminal and C-terminal domains has been solved in complex with ATP (16). Solution-phase analysis established that isolated Rad50 N and C termini form a stable heterodimeric NC domain. Two NC domains associate in the presence of ATP to form an active ATPase, and in full-length Rad50 the coiled-coils may help to bring the N and C termini of a single protomer together to form one-half of the four-part active site (20). Hinge-hinge interactions bring the second NC half-site into register to promote the formation of two active sites in trans and sandwich two ATP molecules within the N a C a /N b C b tetramer. Domain analysis suggests that the architecture of the heterodimeric condensin and cohesin SMCs is essentially the same as that predicted for Rad50 and their bacterial SMC counterparts (18,19). 1 As the details of SMC structure have come into focus, speculative models of DNA binding based on their architecture have emerged as well. The "embrace" model was proposed for chromatin binding by the cohesin (19) and depicts two chromatin 30-nm fibers, one from either sister chromatid, as bound between the coils of an Smc1/3 heterodimer. This binding mode is in part consistent with a model based on electron spectroscopic imaging of the Xenopus condensin XCAP-C/E heterodimer (Smc2/4 homologs) in complex with relaxed plasmid DNA (21). Appreciating the resolution limits of this technique, the images appear to show the DNA duplex wrapping twice in a right-handed sense around the NC/NC globular head domains. The image-based model places two gyres of DNA, one from each encircled NC dimer, between the coiled-coils. These and other models of DNA binding await experimental validation, and few mechanistic details of DNA/chromatin binding by either the cohesin or condensin SMCs exist.
To date, the Xenopus complexes are the biochemically bestcharacterized eukaryotic condensins (9,12,22,23). Two separable SMC-containing complexes have been purified by immuno affinity; they are the five-protein 13 S condensin of XCAP-C, -E, -D2, -G, and -H and the 8 S XCAP-C/E heterodimer. In add-back experiments only the 13 S condensin could rescue in vitro condensation of sperm chromatin in SMCdepleted egg extracts. The 13 S condensin is required for both the establishment and maintenance of chromosome condensation (9), and both activities require ATP and mitotic activation of the complex. Condensin-dependent changes to the geometry of plasmid DNAs were probed in topology-trapping experiments (22)(23)(24), and two ATP-dependent activities unique to the mitotic 13 S condensin were discovered. At high mole ratios the 13 S condensin induced geometries in circular DNAs, which favored the formation of positive knots in the presence of a type II topoisomerase, and trapped positive supercoils in the presence of a type I topoisomerase (23). The positive loop solenoid model was developed to explain how trapping of positive supercoils in relaxed circular DNA would promote positive knotting in nicked circular substrates. Moreover, these results suggested that the condensin must act to change the global writhe of a DNA substrate rather than inducing changes in DNA twist or locally wrapping DNA. The authors envisaged that the 13 S condensin preferentially traps positive loops by nonplanar bending and sequesters them into a radial solenoidal stack (23).
A detailed understanding of the interaction dynamics of the Smc2/4 complex is essential if we are to correctly interpret the requirement of these proteins in chromosome condensation and segregation. This sort of analysis has been applied only to the study of the Xenopus condensin. In this paper we describe the expression, building, and testing of wild-type and mutant Smc2/4 complexes from the budding yeast and present our findings for the DNA-associated activities of this condensin SMC heterodimer pair.
The Strep-tag II TM affinity tag consists of the eight-amino acid sequence N-WSHPQFEK-C (35), which binds with high specificity to the biotin binding site of a streptavidin derivative (36) Strep-Tactin TM (IBA GmbH, Gottinger, Germany). All plasmid constructs were sequenced to verify that inadvertent mutations due to PCR or cloning were not introduced.
Protein Expression and Purification-High level expression of proteins after single or double transformation of BCY123 was performed essentially as described (37). Cells were grown in 0.8 liters of selective medium (1.5% glycerol, 1% lactic acid) and supplemented 1 h before galactose induction with 40 ml of YENB (2% yeast extract, 4% nutrient broth). Plasmids pJES37 and pJES88 were used for individual expression of wild-type SMC2-His 6 and SMC2-Strep-tag fusions, and pJES72 and pJES103 were used for independent expression of the untagged mutant alleles smc2-P and smc2-Sig, respectively. Untagged wild-type SMC4 was expressed from pJES19, and intein/CBD fusions of SMC4, wild type, and smc4 mutants were expressed from the plasmids pSJ50, pJES46, and pAEH23, respectively. The wild-type Smc2/4 complex was routinely purified from cells expressing both SMC2 and the SMC4intein/CBD fusion under maintenance selection for the plasmids pJES85 (Trp ϩ ) and pSJ50 (Ura ϩ ). Because it proved difficult to maintain balanced expression of smc mutant alleles, the Smc2/4 mutant complexes were assembled in vitro by co-disruption of cells expressing each protein separately. Cell harvesting and disruption were carried out as previously described (37) with the following changes. Cells were washed 1ϫ in Buffer A containing 25 mM Na-HEPES (pH 8.1), 500 mM NaCl, 10% glycerol, 1 mM EDTA, 1 mM EGTA and then diluted at 1 g of cells/ml in Buffer B (Buffer A reduced to 0.25 mM EDTA and EGTA) and frozen at Ϫ70°C. Cell slurry (16 ml) was thawed and spiked with phenylmethylsulfonyl fluoride (1 mM), leupeptin (0.5 g/ml), pepstatin (0.5 g/ml), and benzamidine (10 mM) before cell disruption. Cells were broken to ϳ80% completion by glass bead lysis and diluted in 30 ml of ice-cold Buffer B plus protease inhibitors. Debris was cleared by centrifugation at 15,000 rpm for 20 min at 0°C, and soluble extracts were then exposed to the appropriate affinity resins as described below. Intein/CBD fusion proteins were purified essentially as described (33) over chitin resin (New England Biolabs), washed with 10 volumes of Buffer B containing 0.25% Tween 20 and 5 volumes of Buffer B without detergent, and eluted in Buffer B plus protease inhibitors containing 40 mM dithiothreitol. Smc2-His 6 was purified over nickel nitrilotriaceticagarose (Qiagen, Chatsworth, CA), washed in Buffer B, and eluted in the same buffer by imidazole challenge. Strep-tag fusion proteins were captured on Strep-Tactin resin, washed with Buffer B, and eluted by competition with 2 mM desthiobiotin.
Irrespective of which affinity approach was used in the first step, the second purification step for Smc2p, Smc4p, and Smc2/4 was ion exchange over Q-Sepharose. Proteins used in ATPase assays were purified using phosphocellulose (P11, Whatman; 0.25 ml of resin/mg of protein). For proteins used in other analyses, high concentration fractions eluting from Q-Sepharose (ϳ2 mg/ml; 300 l) were separated by size over a column of high resolution (Amersham Biosciences high resolution) Superose 6 (1.0 ϫ 30 cm; exclusion limit ϳ4 ϫ 10 7 Da) in B:300 (Buffer B with 300 mM NaCl) running at a flow rate of 0.5 ml/min (4°C). The column was calibrated in the same buffer using 20 S proteasome (1.2 MDa, a gift from the laboratory of C. P. Hill, University of Utah) in addition to thyroglobulin (669 kDa), apoferritin (440 kDa), catalase (240 kDa), IgG (158 kDa), and BSA 2 (67 kDa). The column void volume was established with blue dextran Ͼ2 MDa. Protein concentrations were determined using the Bio-Rad protein assay reagent (Bradford) by measurement of absorbance at 280 nm (⑀ 280 ϭ 112,430 M Ϫ1 cm Ϫ1 for Smc2/4 (38)) and/or by comparison of Coomassie-stained BSA in SDS-polyacrylamide gels. Proteins were frozen in liquid N 2 and stored at Ϫ70°C.
Sedimentation Equilibrium Analysis-Equilibrium sedimentation was performed after size separation of the wild-type Smc2/4 in buffer containing 10 mM Na-HEPES (pH 7.6), 300 mM NaCl, and 0.1 mM EDTA. The inclusion of EDTA was necessary for Smc2/4 stability, and the absorbance (A 235 ) of this buffer was still relatively low (0.04 absorbance units). Analytical ultracentrifugation experiments were performed at 20°C using a Beckman XL-A centrifuge equipped with absorbance optics. Sedimentation was performed at 5000, 6500, 7000, and 7500 rpm, and the data were collected at three wavelengths (360, 280, and 235 nm) as a function of radial distance by averaging 10 absorbance measurements at steps of 0.001 cm. Scans were begun 12 h after the start of the run or increase in speed, and successive scans were repeated every 4 h thereafter. Equilibrium was judged to have been reached when the overlay of UV traces from two or more successive scans showed no variation across the radial distribution. Radial absorbance data were plotted using XLGraph, 3 and the data in the optical window were selected using Xlaedit. 4 Scans were analyzed individually and globally using the winNONLIN3 4 version of the fitting program NON-LIN (39). Equilibrium absorbance distributions were first fit to an ideal, non-dissociating single species model in which the absorbance at a particular radial distance A r equals A r0 exp[(r 2 Ϫ r 0 2 )] ϩ Y, where A r0 is the absorbance at the reference radius r 0 . The effective reduced molecular weight sigma () is defined as M 1 (1 Ϫ ) 2 /2RT, where M 1 is the molecular weight of the smallest species, is the partial specific volume of the protein (ml/g), is the density of the solvent (g/ml), is the angular velocity (radians/s), and R and T are the universal gas constant and absolute temperature, respectively. The base-line offset term Y serves as a normalization factor for A r0 values. The program SEDNTERP (version 1.01) 5 was used to derive estimates of the partial specific volume for proteins based on amino acid composition (0.7401 ml/mg for Smc2p and 0.7353 ml/mg for Smc4p) and the density of the sedimentation solvent (1.0194 g/ml) at 20°C with data from the International Critical Tables and standard references (40). The equilibrium association constants obtained for fits to a reversible Smc2p plus Smc4p heterodimerization model, assuming a theoretical average monomer M r of 148,059, were high and variable, well outside a meaningful range of values, suggesting that the Smc2/4 complex behaves essentially as a non-dissociating complex. Hence, the Smc2/4 complex was treated as a monomer for all subsequent fits. Smc2/4 association and other equilibrium models were fit to the general where the absorbance of each n-mer species, A n , equals K n A r0 n exp[n(r 2 Ϫ r 0 2 )] ϩ Y, and K n is the equilibrium association constant (in absorbance units) for the formation of the nth species (second, third, fourth, etc.). The best fits were obtained using a monomer º dimer equilibrium model, in which the monomer molecular weight was equal to that of the Smc2/4 heterodimer. The value for the smallest species in association models was calculated ( calc ) using the theoretical heterodimer M r based on the sequence for Smc2/4 (296,118 g/mol), with av ϭ 0.7377 ml/g taken as the average of the estimates for Smc2p and Smc4p. Calculated association constants were converted to molar units using the equation K a ϭ (K a Abs Ϫ1 (1.2 cm ϫ ⑀) nϪ1 /(n) (41). The theoretical native molar extinction coefficients at 280 nm for Smc2p (56,730 M Ϫ1 cm Ϫ1 ) and Smc4p (55,700 M Ϫ1 cm Ϫ1 ) were converted to values appropriate for 235 nm using the 235/280 conversion ratio 4.05 to yield the Smc2/4 extinction coefficient ⑀ 235 ϭ 463,442 M Ϫ1 cm Ϫ1 . Base-line offset terms (Y) were floated, and was then globally fit to a single species model without compensatory adjustment for non-ideality. More complex models were fit globally to a common association constant by holding calc for the theoretical Smc2/4 heterodimeric M r and n constant. The fit was judged to be in good agreement with the model if separate K n values for each data set converged with the global K n . For a more complete description of fitting strategies using NONLIN see Hansen et al. (42) or Laue and co-workers (43).
DNA Substrate Preparation, Binding Reactions, and Electrophoresis-Topoisomerase I was purified from raw wheat germ (WG topo I) as described (46) and stored at Ϫ70°C at 400 g/ml. The various plasmids used in analyses, pBSKS(ϩ) (Stratagene, 2961 bp), pRS424 (47) (5616 bp), or pJES7 (14,700 bp), were relaxed at 23°C with WG topo I, singly nicked in the presence of ethidium bromide with DNase I as described (48), or linearized by restriction digestion. Circular pBSKS(ϩ) was radioactively labeled by nick translation in the presence of [␣-32 P]dCTP using standard procedures. The plasmid pBSKS(ϩ) was linearized with XmaI and end-labeled by incorporation of [␣-32 P]dCTP with T4 DNA polymerase. Radiolabeled DNA was quantitated by ethidium bromide staining in agarose gels by comparison to non-labeled DNA standards. All binding reactions (50 l) contained 1 mM ATP unless indicated otherwise and either 10 or 100 ng of the various topologic forms of DNA. Other details can be found in the figure legends. Reaction mixtures were separated at ambient temperature in 0.8% agarose gels in 1ϫ TAE at 3 V/cm for 12-18 h. To prevent the loss of Smc2/4-DNA-bound species that could not enter the gel, samples were embedded in the wells with molten agarose before gel submersion. Molten agarose (20 l at 55°C of 1% low melting temp agarose in 1ϫ TAE plus 0.01% bromphenol blue) was rapidly mixed with each 50-l reaction and quickly transferred to dry wells. Gels were stained with ethidium bromide for image-capture under UV illumination or vacuum-dried for autoradiography.
Knotting and Supercoiling Assays-All topology trapping and DNA binding assays were performed in complete BKS, BKS containing 100 g/ml BSA (fraction V, Sigma), 1 mM dithiothreitol, and 2.5 mM MgCl 2 . Phage T4 topoisomerase II was kindly provided by Wai-Mun Huang (University of Utah), and yeast topoisomerase II was purified as described previously (34). Supercoiling reactions (50 l) were performed by combining 37 l of complete BKS, with or without 0.5 l of 100 mM ATP (1 mM final), 100 ng of supercoiled pRS424 (0.55 nM plasmid) or relaxed pBSKS(ϩ) (1 nM plasmid), and variable amounts of Smc2/4 (10 l in Buffer B:300) as detailed in Fig. 4. This mixture was incubated at 23°C for 15 min, and 100 ng of WG topo I was then added for an additional 30 min to relax the DNA. Reactions were quenched with 20 l of stop solution (50 mM Tris-HCl (pH 7.5), 1% SDS, 10 mM EDTA, and 100 g/ml proteinase K) and placed at 42°C for 1 h to digest proteins. DNA was recovered by ethanol precipitation after phenol:chloroform extraction, separated on 0.8% TAE-agarose gels (with or without 0.2 g/ml chloroquine), and recorded after ethidium bromide staining. Knotting reactions were performed under the same conditions except that nicked pRS424 was used, and incubation in the presence of 100 ng of T4 or yeast topo II was limited to 15 min.

Complementation Analysis of SMC Expression Constructs-
Our investigation of the S. cerevisiae condensin SMCs Smc2p and Smc4p began with the creation of a set of plasmids for the expression of each in the budding yeast. To assess the effect of ATPase mutations and/or the placement of affinity tags, all expression constructs were tested for their ability to rescue the appropriate smc⌬. Heterozygous diploid stains (SMC/ smc⌬::KanMX4) were transformed with the plasmids, sporulated, and dissected onto rich medium containing glucose. Expression levels were sufficiently high under these conditions to allow constructs carrying untagged wild-type SMC2 or SMC4 alleles to rescue deletions. The same was not true for wild-type alleles carrying certain affinity tags. We found that Smc4p could tolerate large C-terminal fusions (up to at least 52 kDa) without effect; in contrast, tags longer than 11 residues placed at the C terminus of Smc2p prevented rescue. N-terminal fusions proved deleterious to both SMCs, and neither His 6 -SMC2 nor His 6 -SMC4 could complement deletions (data not shown). None of the targeted ATPase mutations (smc2-P, smc2-Sig, smc4-Pϩ, and smc4-Sig) was able to complement their respective smc deletions nor did they appear to act in a dominantnegative manner.
Assembly and Purification of the Smc2/4 Complex-When Smc2p and Smc4-Int/CBD were co-expressed, a stable 1:1 complex of the proteins could be purified based on the Smc4p affinity tag (Fig. 1B). All highly purified preparations of wildtype Smc2/4 used in the present study were obtained in this manner. One of the co-purifying proteins near 70 kDa, present in SMC preparations at variable levels (see Fig. 2B), was confirmed by antibody staining to be Hsp70p (data not shown).
Smc2p and Smc4p were purified individually, and we found that each protein in isolation was more prone to nonspecific absorption, degradation, or aggregation than the Smc2/4 complex. Smc4p was particularly sensitive to degradation, a phenomenon that may be similar in nature to the time-dependent cleavage of the Schizosaccharomyces pombe Smc4p homolog, Cut3, previously observed (49). Smc2p, although resistant to degradation, tended to aggregate.
We assessed whether isolated Smc2p or Smc4p could assemble into stable homodimers by performing affinity separation with mixtures of differentially size-tagged Smc2p or Smc4p (i.e. tags large enough to detect size shifts on SDS-gels). No untagged-Smc2p co-purified with the tagged Smc2p; the same was true for Smc4p (data not presented). These pull-down results do not prove a lack of self-association, but under these conditions, SMC homodimerization appears minimal.
When purified Smc2p and Smc4p were mixed, they assembled rapidly in vitro to form stable 1:1 complexes. Fig. 1C displays the results of one such a mixing experiment that employed Smc2-His 6 and untagged Smc4p. This, in combination with the results of affinity purification presented in Fig.  1B, showed that placement of a single affinity tag at the C terminus of Smc2p or Smc4p permits isolation of the Smc2/4 complex. Assembly was independent of nucleotide, and Smc2 and -4 assembled after purification behaved similarly to in vivo assembled (co-expressed) Smc2/4 (i.e. stability, solution behav-ior, DNA binding, and ATP hydrolysis).
Solution Properties of the Smc2/4 Complex-In vivo assembled Smc2/4 complex was subjected to a number of analyses to determine the oligomeric state of the 1:1 hetero-complex. Size separation of the SMC complex over Superose 6 ( Fig. 2A) showed that Smc2/4 (296 kDa) could distribute across a range of exclusion volumes predicted for globular proteins of ϳ200 kDa to 1.2 MDa. This column has a very large exclusion limit, and the largest Smc2/4 species eluting in the UV trace shown in Fig. 2A were within the separation range of the column. The broadly eluting distribution could represent the fractionation of stable "shape conformers" of individual heterodimeric Smc2/4 complexes with identical mass (50) or could arise by oligomerization of the Smc2/4 complexes. To discriminate between these two possibilities, proteins eluting at different positions in the distribution were chosen for shape-independent mass determination by sedimentation equilibrium. We found that the data from the protein complex in fraction 2 (Fig. 2, A and B) did not fit any scheme for a single molecular weight species, suggesting that the larger species from the gel filtration column were not simply shape conformers of an Smc2/4 heterodimer. We noted that the Smc2/4 complex was also prone to aggregation, and the species in fraction 2 may have represented an oligomeric mixture. This tendency, coupled with the low absorbance of Smc2/4 at 280 nm, restricted the concentration range that could be used and prevented further analysis of fraction 2. For fraction 8 (Fig. 2, A and B) we performed two independent sedimentation experiments and globally fit the data from each (Experi- ment 1, 9 data sets, 3 speeds, and three concentrations; Experiment 2, 6 data sets, two speeds, and three concentrations) to models describing single species and association behavior. The data from both experiments were also combined (15 data sets) and fit globally. Neither of these data sets was adequately described by a single Smc2/4 species or by models involving dissociation of the heterodimer. However, the data were fit well by a monomer-dimer equilibrium model (2(2/4) º (2/4) 2 ), where the monomer molecular mass was equal to that of the Smc2/4 heterodimer. Examples of fits obtained for this model are presented in Fig. 2C, for which the estimated K d was 3.2 Ϯ 1.3 M. Residuals were small and random, indicating a close correspondence between the experimental data and the absorbance distributions predicted by the association model. Good agreement was also obtained for the fitted molecular size (295 Ϯ 23 kDa) when its value was held constant during fitting ln K d to a monomer-dimer equilibrium. Moreover, association constants obtained by simultaneous fitting of separate ln K a values for individual data sets converged with the common association constant obtained by global fitting.
ATPase Activity of Wild-type and Mutant SMC Proteins-As members of the ABC family of ATPases, the SMC proteins are predicted to hydrolyze ATP. Although a number of results for the eukaryotic heterodimeric complexes (13, 22, 51) and prokaryotic homodimeric complexes have been reported (18,52,53), no steady state parameters have yet been established for any SMC-containing complex. We examined the ATPase activity of the wild-type Smc2/4 complex and found that it obeyed simple Michaelis-Menten kinetics. Hydrolysis rates determined under initial velocity conditions using a range of ATP concentrations (50 -5000 M) were fit to a k cat of 0.21 Ϯ 0.005 s Ϫ1 and a K m of 420 Ϯ 30 M (Fig. 3A). These data provide the first steady state parameters describing ATP hydrolysis by an SMC protein from any species.
We also determined that the relatively low level of ATPase activity displayed by the Smc2/4 complex is in fact intrinsic to the heterodimer and not due to a contaminating source. ATPase activity could be reconstituted when separately purified Smc2p and Smc4p, neither of which could hydrolyze ATP on its own (see Fig. 3), were mixed at a 1:1 stoichiometry. The ATPase activity of the in vitro assembled complex equaled that of the complex assembled in vivo (Fig. 3B). In contrast, when the Smc4-P-loop mutant or the ABC Signature motif mutant Smc4-Sig was mixed with wild-type Smc2p, no ATPase activity was recovered. These results confirm that the condensin Smc2/4 sub-complex can hydrolyze ATP and, furthermore, that targeted inhibitory mutations in the ATPase domain of one SMC can act in trans to inhibit hydrolysis activity of the opposite ABC catalytic site. As expected, the Smc2-P/4-P double P-loop mutant complex could not hydrolyze ATP (Fig. 3B) but versus radial distance and are overlaid with curves generated from global fitting of the data (see "Experimental Procedures") for a reversible heterodimer-heterotetramer model using a common calculated for the size of the smallest species of 296,118 Da (Smc2/4 theoretical molecular size). The fits shown correspond to an estimated K d of 3.2 M, and the residuals are displayed above each distribution. retained its ability to bind DNA as well as wild-type Smc2/4 (data not shown). Finally, neither linear (sheared salmon sperm) nor a mixture of supercoiled and relaxed plasmid DNA (pRS424) affected the ATPase activity of in vitro or in vivo assembled Smc2/4 complex (Fig. 3B) under conditions that favored DNA binding.
Binding Induced Geometry Change; DNA Knotting and Supercoiling Assays-We next tested whether exposure to the Smc2/4 complex could introduce geometric changes in closed circular DNA. For historical reference, topologic trapping experiments utilizing topoisomerases and naked plasmid DNA provided the first biochemical readout of the effects of DNA binding and ATP hydrolysis by the mitotically active Xenopus 13 S condensin complex (22)(23)(24). Knotting reactions utilize nicked circular DNA substrates that should not accumulate twist to detect changes in net writhe as reflected by formation of knots after reaction with a type II topoisomerase. Supercoiling assays utilize a relaxed plasmid substrate and a type I topoisomerase to trap compensatory changes in twist and writhe generated upon binding and/or energy-dependent reconfiguration of the DNA by the condensin.
DNA templates were relaxed after exposure to a high mole ratio of Smc2/4:DNA (64:1), in the presence of ATP. SMC-dependent geometry changes were assessed after gel separation of reaction products and are presented in Fig. 4, A and B. The Smc2/4 complex isolated from asynchronous yeast cultures induced the formation of a broad distribution of DNA knots in the presence of either phage T4 or yeast topo II (Fig. 4A). Knot formation was strictly dependent on both topo II (lane 1) and the Smc2/4 complex (lane 2). Smc2/4-dependent knots increased in direct proportion to increases in the Smc2/4:plasmid ratio up to 256:1 (the highest ratio tested), and no knotting was detected at mole ratios below 32:1 (data not shown). These results demonstrate that the Smc2/4 complex can induce topo II-dependent DNA knotting in the absence of the Ycs4p-Ycg1p-Brn1p regulatory complex and without mitotic activation. In the supercoiling assays only minor changes in net writhe were detected after exposure of DNA substrates to the Smc2/4 complex (Fig. 4B, lane 3) and topo I.
To further analyze this lack of supercoiling, we tested for concentration-dependent positive supercoiling using the smaller 2.96-kb relaxed plasmid pBSKS(ϩ) as a template (Fig.  4C). The products of the reactions were resolved in the presence of 0.2 g/ml chloroquine to allow the magnitude and direction of changes in superhelicity to be assessed. A slight positive shift in the resultant topoisomer distribution was observed after simultaneous plasmid exposure to Smc2/4 and topo I (lane 5, 128:1). A minor ATP dependence was also seen (compare lanes 5 and 6), but this ATP effect on positive supercoiling was not reproducible. As a control, the relaxation activity of WG topo I was shown not to be limited by plasmid binding at the high mole ratios of Smc2/4:plasmid used in this study (see Fig.  4C, lane 8). Additionally, supercoiling reactions performed at the highest mole ratios without topo I caused no changes in the topoisomer distribution of relaxed or supercoiled DNA substrates (Fig. 4C, lanes 7 and 10), indicating that Smc2/4 preparations were free of topoisomerase or nuclease contamination.
Stoichiometry and Stability of DNA Binding by the 2/4 Complex- Fig. 5 presents the results of a series of gel shift experiments that summarize the DNA binding studies of the Smc2/4 complex under a number of conditions. Earliest attempts to resolve DNA-bound species electrophoretically failed due to their lack of mobility in agarose gels (see Fig. 5D). The formation of low mobility complexes after binding of Smc2/4 to DNA appears to be a general characteristic of many SMCs (16,22,51,52,54,55). This limitation likely results from one or more of the following factors: 1) the large size of multiple Smc2/4 complexes bound to the same plasmid, 2) the even larger size of networks formed by intermolecular protein-DNA cross-linking, 3) the precipitation of protein-DNA aggregates by exposure to low salt electrophoresis buffer, and 4) the neutralization of DNA charge at high protein occupancy. The molten agarose/cosolidification loading protocol described in the procedures was adopted to avoid buffer-induced precipitation and facilitated entry of Smc2/4-bound species into the gel.
The salt stability and electrophoretic mobility of Smc2/4-DNA bound species were examined in binding reactions employing a 2961-bp linear or nicked circular DNA at varying degrees of Smc2/4 binding saturation. Fig. 5, A and B, demonstrate that Smc2/4-DNA-bound species began to resist salt disruption as the level of protein:DNA increased. We tested mole ratios that bracketed those used in knotting and supercoiling reactions and found that DNA binding at a mole ratio of 16:1 could induce a complete shift of both linear and circular DNA. These shifted species were disrupted by modest salt concentrations of between 100 and 300 mM NaCl (Fig. 5, A and  B); however, as the mole ratio of Smc2/4:DNA in binding reactions was increased (64:1 linear; 256:1 circular), the sensitivity to salt challenge disappeared, and the resistant species remained stable to 0.5 M NaCl challenge. Salt-refractory complexes appeared to form at a lower mole ratio for linear DNA than for circular DNA. This result was not anticipated, and it might have been reasonable to expect that proteins bound to linear DNA having free ends would dissociate more readily than from those bound to closed circular DNA. Despite this paradox, it is clear that DNA binding of Smc2/4 at protein:DNA mole ratios less than or equal to 16:1 is reversible.
The DNA binding experiments in Fig. 5, A and B, were performed in the presence of 1 mM ATP to mimic knotting and supercoiling reaction conditions. It has been suggested that DNA binding by the SMCs may be modulated by ATP binding/ hydrolysis (16,52,53,55). To test the effect of nucleotide on DNA binding, reactions were performed in the absence of nucleotide (No Nuc) or the presence of ATP, ADP ϩ sodium orthovanadate (Na 2 VO 4 ), ATP ϩ Na 2 VO 4 , or the non-hydrolyzable analog AMP-PNP (Fig. 5C). In these experiments, the Smc2/4 complex was exposed to each nucleotide or nucleotide analog for 30 min at 30°C before adding nicked circular DNA. We found no apparent nucleotide-based differences in the degree or magnitude of gel shifts for the bound species formed.
To determine the effect of template size on binding, two plasmids of different lengths (3 and 14.7 kb) were compared in parallel reactions (Fig. 5D). Binding titrations were performed by holding plasmid concentrations constant at 2 nM while varying the concentration of the Smc2/4 complex. Bound species were resolved, and the concentrations at which 100% of the DNA shifted were compared. In these experiments the gels were loaded without molten agarose (see "Experimental Procedures"), and consequently the bound species failed to enter the gel. These results demonstrate that binding occurs over a very narrow range of protein concentrations that correlates with the Smc2/4 to base pair ratio, not with the Smc2/4 to plasmid ratio. The point at which substrate DNA was shifted completely occurred at an Smc2/4:bp ratio of between 1:50 and 1:100 for both plasmids. This corresponds to an Smc2/4:plasmid ratio of between 64:1 and 128:1 for the small template and 256:1 and 512:1 for the large template (Fig. 5D, see the arrows above the gels). Estimates of the percent of total protein bound were made by differential centrifugation of the protein-DNA complexes (18,52,53). By this measure we estimated that ϳ50% of the total protein in these assays was bound to the DNA. Moreover, we noted that the percent protein bound increased with increasing mole ratios of Smc2/4:plasmid.
DNA Competition Studies-The DNA knotting and supercoiling activities ascribed to the Xenopus condensin (13,22,23) and the yeast Smc2/4 complex presented in this work are manifest only at high protein to DNA ratios. Therefore, we tested whether the salt-stable species assembled at higher molar ratios of protein to DNA (Fig. 5, A and B) remained dynamic and reversible or became static protein-DNA aggregates. Fig. 6A shows the results of DNA challenge experiments in which Smc2/4 pre-bound to 32 P-labeled nicked circular DNA at a mole ratio of 128:1 was challenged with unlabeled DNA of an equivalent length (linear, nicked circular, or supercoiled). It is important to note that a single Smc2/4-32 P-labeled DNA master mix was used for the entire set of reactions shown; the only difference between lanes is the form and concentration of FIG. 4. Smc2/4-induced geometry changes in closed circular DNA. A, knotting assay showing the products of reactions performed with nicked pRS424 plasmid DNA (5616 bp) incubated with a 64:1 ratio of Smc2/4:plasmid and treated with either phage T4 or yeast topo II. B, supercoiling assay showing the relaxed topoisomer distribution obtained using supercoiled pRS424 mixed with either Smc2/4 (64:1), WG topo I, or both. C, supercoiling assay performed with 100 ng relaxed pBSKS(ϩ) (2961 bp) and variable amounts of Smc2/4 with or without 1 mM ATP. Reaction products were resolved on a 1% TAE-agarose gel containing 0.2 g/ml chloroquine diphosphate in the gel and running buffer. Images were recorded digitally after ethidium bromide staining. the competitor DNA added. Both unlabeled linear and supercoiled DNA efficiently displaced Smc2/4-bound DNA species ( Fig. 6A; lanes 2-6 and 17-18) as judged by the reappearance of free, labeled DNA. In contrast, nicked circular DNA was not an efficient competitor and failed to cause release of labeled DNA at the highest concentrations used (lanes 9 -12). These results demonstrate two facts about DNA binding by the Smc2/4 complex: 1) Smc2/4-DNA complexes formed at high mole ratios remain dynamic and reversible, and 2) competitor DNAs with different geometries appear to engage the Smc2/4 complex in fundamentally different ways.
The models in Fig. 6B depict DNA binding by "coil-confinement" as envisaged for the Smc2/4 complex. These models bear resemblance to the embrace model proposed for sister chromatid cohesion by the cohesin SMCs (19). The two binding modes, "threading" and "entrapment," depict two ways in which a single DNA duplex could potentially gain access to the internal space bounded by the coiled-coils of a single Smc2/4 complex. According to these models, one would predict that binding of an open circular DNA duplex (nicked or relaxed plasmid) would occur preferentially by entrapment and rarely by threading. This prediction of the coil-confinement model was used to develop the "threading-displacement" and "first-in-last-out" mechanisms illustrated in Fig. 7 to explain how DNAs of equal length but differing shape/topologies could vary in their ability to act as cold DNA competitors. DISCUSSION Here we present the first thorough biochemical characterization of the SMC core of the S. cerevisiae condensin complex. Several of the experimental results are surprising. First, we found that the Smc2/4 complex self-associates in solution to form multimers. The simplest model that adequately describes the data suggests that stable Smc2/4 heterodimers associate to form heterotetramers with a K d of ϳ3 M. Second, the Smc2/4 complex alone has ATPase activity that is unaffected by the presence of dsDNA. Third, Smc2/4 can reconfigure DNA in such a way that topo II creates knots, but the geometry does not seem to introduce a net writhe into the circular DNA substrate. Fourth, the binding of Smc2/4 to DNA is complicated. A molar ratio of ϳ1:100 Smc2/4 to DNA base pairs is required to see a gel shift regardless of plasmid size. The binding is independent of ATP, ADP, or ATP analogs, and complexes can be disrupted by high salt until the Smc2/4:plasmid ratio becomes very high. These salt-stable complexes are not simply irreversible aggregates, because the addition of an excess of unlabeled DNA can disrupt them. Unexpectedly, the ability of competitor DNA to dislodge bound Smc2/4 depends on its shape. DNA with a long, narrow shape (linear or supercoiled) fully disrupts complexes, whereas open circular DNA does not. This finding led us to propose potential models for how the Smc2/4 complex could bind and trap DNA.
Self-association of Smc2/4 Heterodimers-Sedimentation equilibrium results showed that Smc2/4 heterodimers interact FIG. 5. Smc2/4 DNA binding assessed by gel mobility shift. Sodium chloride challenge experiments were performed using either linear (A) or nicked circular (B) pBSKS(ϩ) DNA. Radiographic exposures of dried agarose gels showing Smc2/4-DNA-bound complexes formed at various Smc2/4 to DNA mole ratios after challenge with increasing concentrations of NaCl. 32 P-Labeled DNA (10 ng) was mixed at the ratios indicated in complete BKS for 30 min, and NaCl was then added to the final concentrations listed (100 -500 mM). The products were resolved electrophoretically as described under "Experimental Procedures." C, the affects of various nucleotides or nucleotide analogs (2 mM final) on DNA binding by Smc2/4. Smc2/4 was incubated for 30 min in complete BKS supplemented with ADP ϩ 4 mM sodium orthovanadate, ATP ϩ 4 mM sodium orthovanadate, ATP, AMP-PNP, or no nucleotide (No Nuc). Labeled nicked circular DNA (10 ng) was then added, and the products of binding were separated on agarose gels, dried, and exposed to film. D, the stoichiometry of DNA binding and binding saturation were examined using nicked plasmids of different lengths. Plasmid DNA concentrations were held constant at 2 nM, and increasing amounts of Smc2/4 were added as indicated. Arrows above the gel indicate the point at which all free DNA is shifted into the wells. Total DNA binding occurred at a base pair to Smc2/4 ratio ranging between 50:1 and 100: 1. reversibly. The free energy change for Smc2/4 self-association (⌬G 2/4-(2/4)2 ) is within a physiologically moderate range (ϳ8 kcal/mole). Absorbance interference prevented analysis of the effects of ATP binding/hydrolysis on the heterodimer-heterotetramer transition. However, inclusion of Mg-ATP did not alter the sedimentation profiles of Smc2/4 through sucrose gradients or the elution profiles from sizing columns (data not presented). These results raise the possibility that interaction between SMC pairs could occur in vivo and that perhaps these integral condensin proteins could contribute to a framework essential for the proper management of chromatin and/or the mainte-nance of chromosome structure. For proteins that may have a scaffolding or architectural role in chromosomes, it is useful to know that they have the intrinsic ability to self-associate at relevant concentrations.
Hydrolysis of ATP by the Smc2/4 Complex-Because of the large number of potentially contaminating ATPases in cells, it is difficult to prove that a low level of ATPase activity in a protein preparation is intrinsic to the protein of interest. We have proven that the Smc2/4 complex is an ATPase by mixing the separately purified Smc2p and Smc4p. Although before mixing there is essentially no ATPase activity, an equimolar combination of the proteins increased this activity by at least 6-fold. This ATPase activity is independent of, and not stimulated by double-stranded DNA.
A mutation in the P-loop or the signature motif of only one of the SMCs abolishes the ATPase activity of the complex, and each of these mutations fails to rescue deletions of the respective SMC gene. Based upon the Rad50 structure (16) one might predict that mutation of a single P-loop or signature motif might not affect the second active site. Nevertheless, the failure of wild-type/mutant Smc2/4 heterodimers to hydrolyze ATP suggests that the ATPase mechanism may involve active site cooperation. Mutation of one ATPase active site of P-glycoprotein (Pgp), another member of the ABC family, was found to have a similar effect (56), and this enzyme is proposed to operate by alternating ATP hydrolysis between two active sites (57,58). A similar mechanism has been predicted for other ABC ATPases including the cystic fibrosis transmembrane conductance factor, the histidine permease HisP, and the maltose transporter MalK, as reviewed by Holland et al. (59). The results presented here are consistent with a similar ATPase mechanism for Smc2/4. Smc2/4 Promotes Knotting without Appreciably Changing Writhe-The Smc2/4 complex promotes knotting of circular DNA in the presence of topo II but does not cause significant trapping of positive supercoils in the presence of topo I. Topoisomer distributions were shifted only slightly in the presence of Smc2/4:DNA at ratios that promoted robust knotting (Fig. 4C). Unlike the activated Xenopus 13 S condensin, which showed a strong correlation between knotting and the level of positive supercoiling (22,24), DNA knotting promoted by the Smc2/4 complex does not appear to introduce significant superhelical strain. The chirality of the knots formed in the presence of Smc2/4 has been analyzed, and the results will be presented elsewhere. 6 We conclude that knotting and supercoiling can represent two distinct and separable activities of the condensin. Although both Smc2/4 and the Xenopus 13 S condensin induce shape changes in DNA, their exact geometries clearly differ.
DNA Binding by the Smc2/4 Complex-Gel shift analyses were performed using a 3-kb circular or linear DNA to examine the stability and reversibility of Smc2/4-DNA bound species as a function of protein:DNA ratio. Reversibility testing of DNAbound species was necessary since Smc2/4-dependent DNA knotting, supercoiling, and re-annealing are manifest only at high protein:DNA plasmid ratios (Refs. 22 and 49 and this study). Smc2/4-DNA bound species resisted salt disruption as the ratio of Smc2/4:DNA in binding reactions was increased. Mole ratios that were too low to support efficient knotting (i.e. 16:1) remained sensitive to salt disruption; however, the knotting-proficient ratio of 64:1 caused DNA-bound species to be more salt-resistant (compare 16:1 and 64:1 ratios in Fig. 5B). This correlation between DNA knotting and salt stability may suggest that both DNA binding and interactions between het-6 J. E. Stray and J. E. Lindsley, manuscript in preparation.
FIG. 6. DNA competition analysis. Smc2/4-DNA-bound species formed at high mole ratios were challenged with cold competitor DNA. A, autoradiograph showing binding reaction products after DNA competition. Smc2/4 was pre-bound to labeled nicked circular DNA (2.96 kb) at a protein:DNA mole ratio of 128:1 and then challenged with the same size (2.96 kb) linear, nicked, or supercoiled plasmid at an unlabeled to labeled DNA ratio of 10:1, 50:1, 100:1, 300:1, or 500:1. Linear DNA (lanes 1-6) and supercoiled DNA (lanes 13-18) caused release of free, 32 P-labeled, nicked DNA from pre-bound complexes (lanes 6, 17, and 18). Nicked circular DNA was unable to fully displace labeled DNA (lanes 7-12). To control for distortions in the migration of free DNA caused by molecular crowding in challenge lanes, free 32 P-labeled DNA was mixed with a 500-fold excess of cold competitor of the appropriate topology (lanes 1, 7, and 13). B, models by which Smc2/4 may capture and trap DNA by coil-confinement. The entrance and exit of duplex DNA by threading or gated-entrapment is depicted. erodimers participate in the formation of knotting geometries.
The presence or type of nucleotide did not affect binding of Smc2/4 to circular DNA (see Fig. 5C). The crystal structure of the P. furiosus Rad50 catalytic domain revealed that binding of a non-hydrolyzable ATP analogue, AMP-PNP, trapped the NC head domains into a (NC) 2 tetrameric configuration (16). Thus, one might expect that binding of AMP-PNP or ADP ϩ orthovanadate would lock the ATPase domains of Smc2/4 in a closed state and block access of circular DNA to the inter-coil space (Fig. 7A). However, neither nucleotide prevented binding of circular DNA. These results indicate that either 1) these nucleotides do not lock the ATPase domains, 2) binding does not occur by coil-confinement, or 3) open circular DNA is capable of threading between the coils.
Apparent Cooperativity in Smc2/4 DNA Binding-The interaction of Smc2/4 with DNA is more complex than expected. No binding to plasmid DNA was detected until the Smc2/4 to base pair ratio reached ϳ1:100. This ratio held over a wide range of protein/DNA concentrations and plasmid lengths. Binding occurred in an all-or-nothing fashion, and additional protein caused the shifted species to migrate even more slowly. These observations suggest that stable DNA binding begins at a critical level of Smc2/4 saturation along the DNA lattice (nucleation), after which the loading density of Smc2/4 increases in proportion to the protein:DNA ratio. The free energy of Smc2/4 self-association may stabilize the nucleation. When DNAbound Smc2/4 complexes are forced into close proximity with each other, protein oligomerization may be favored. This would bring distant segments of DNA together and could lead to a knotting-proficient geometry. Atomic force images of the S. pombe condensin assembled with DNA at high protein ratios showed the formation of clustered condensin-DNA aggregates (51), and it is possible that these aggregates represent reversible species relevant to in vitro condensin function.
Reversibility of DNA Binding-Because knotting-proficient Smc2/4-DNA species resisted salt challenge, excess unlabeled DNA was used to drive apart the complexes to show that they indeed remained reversible. Both linear and supercoiled DNA readily dissociated labeled, circular DNA prebound to Smc2/4, demonstrating that Smc2/4-DNA bound complexes formed at ratios as high as 128:1 are reversible ( Fig. 6A; lanes 2-6 and  17-18). Under the same conditions, open circular DNA could not displace prebound circular DNA (Fig. 6A, lanes 8 -12). This was not expected, since the only difference between these competitor DNAs was essentially their shape.
We propose that this shape-dependent DNA competition can be explained by topologic trapping of DNA between the Smc2/4 coiled-coils. This explanation is consistent with the unique geometry of SMC pairs (19,50) and the recent findings for DNA binding by isolated domains of the Bacillus SMC (18). Bacillus or supercoiled plasmids through a mechanism of threading-displacement. Labeled circular DNAs are shown in red, cold competitor is shown in gray, Smc4p is shown blue, and Smc2p is shown in green. C, the first-in-last-out mechanism in which steric impasse (caused by high concentrations of competitor DNA) could prevent the exit of 32 P-labeled circular DNA held by Smc2/4 by the coil-confinement model. SMC deleted for both the N-and C-terminal globular domains forms dimers capable of binding DNA (18). Mutations that disrupt dimerization of this construct block DNA binding, and dimers of just the hinge regions also do not bind DNA. These results suggest that DNA binding requires both coiled-coils in correct proximity to the hinge dimer and is consistent with DNA binding in the cleft formed by the coiled-coil of a folded SMC pair. The cohesin SMCs have been proposed to embrace DNA segments (19), and this general mode of DNA binding can be used to explain the present results.
DNA Competition Supports a Mechanism of Gated DNA Binding and Release- Fig. 7A illustrates how capture of an open circular DNA could occur by gated-entrapment. This model predicts that interaction of the globular NC (ATPase) domains is transient, and that reversible NC-NC interaction could form a "gate" through which DNA duplexes diffuse or traverse by active transport. Operating in this fashion, the Smc2/4 complex would mimic gated ligand transport promoted by other ATPases of the ABC transporter family (59). Gate dynamics, possibly mediated by ATP binding/hydrolysis, could lead to transient trapping of a DNA duplex in the Smc2/4 inter-coil space. DNA could also bind by threading between the coils (Fig. 6B), but only linear or supercoiled DNA is envisioned to do so easily.
Based on these ideas two mechanisms of shape-dependent competition become apparent (Fig. 7, B and C). Threadingdisplacement can be promoted by linear or supercoiled DNA (colored gray), which may insert itself between the coils of Smc2/4 to displace prebound DNA circles (colored red). Open circular DNA would not be expected to readily thread between the coils and would, thus, have to enter between the head domains. The first-in-last-out mechanism explains how unlabeled circular DNA (gray) could sterically inhibit dissociation of prebound DNA circles (red).
The mechanisms presented are based on the prediction that plectonemically interwound branches of supercoiled DNA would thread more efficiently than nicked circular DNA. This is consistent with experimentally derived physical measurements of the average superhelical radius of interwound plasmids (60). For a 3.5-kb plasmid with a specific linking difference close to that of a plasmid directly isolated from Escherichia coli, the average radius was found to be ϳ4 nm, nearly 14 times less than that of an open circular plasmid of the same length (57.5 nm). The threading-displacement mechanism can be tested since the competitive potential of circular DNA should decrease as the superhelical radius increases (i.e. as the linking difference nears zero).