Hin-mediated Inversion on Positively Supercoiled DNA*

Hin recombinase requires negatively supercoiled DNA for an efficient inversion. We have generated positively supercoiled plasmid DNA using reverse gyrase from Sulfolobus shibatae and subjected it to the Hin-mediated inversion reaction. Both Hin and Fis showed the same DNA binding activity regardless of the superhelical handedness of the substrate plasmid. However, inversion activity on positively supercoiled DNA was less than 1% of negatively supercoiled DNA. Assays designed to probe steps in inversion, showed that on positively supercoiled DNA, Hin was able to cleave the recombination sites with the same efficiency shown on negatively supercoiled DNA but was not able to exchange the cleaved DNA. Based on the theoretical differences between positive and negative supercoiling, our data may suggest that unwinding of the double helix at recombination sites is needed after DNA cleavage for strand exchange to occur.

The Hin recombinase from Salmonella typhimurium catalyzes a site-specific recombination that leads to inversion of the intervening DNA that is flanked by two recombination sites known as hix (1). Hin, a 21-kDa protein, exists as a homodimer in solution and binds to hix site that is composed of imperfect 13-bp 1 inverted repeats (2,3). For an efficient Hin-mediated inversion reaction, an additional protein factor, Fis is required (4). Fis, a 11-kDa protein, binds to two different sites in 64-bp DNA sequences called the recombination enhancer (see Fig. 1).
A current mechanistic view of the Hin-mediated inversion reaction begins with the interaction between hix-bound Hin dimers, which results in bringing the two hix sites in close proximity to form a paired-hix structure (5,6). This nucleoprotein complex should be assembled with the Fis-bound enhancer into a higher order nucleoprotein complex (a synaptic complex named invertasome). In this synaptic complex, Hin is able to cleave the middle of hix sites to which it is bound, generating 2-bp recessed 5Ј-phosphate ends (5,7). Each of the four 5Ј-ends of the cleaved DNA strand is covalently but transiently linked to the serine residue at position 10 of each protomer of Hin (7). Exchange of the cleaved DNA strands and religation now lead the intervening DNA to inverted configuration (8) (see Fig. 1).
One of the interesting properties of the Hin-mediated inversion system has been that it requires negative (Ϫ)supercoiling of substrate DNA for an efficient inversion reaction (1). Analysis of invertasome formation on a series of topoisomers (from fully relaxed to the physiological density) of a substrate plasmid showed that invertasome is not formed on fully relaxed plasmid (thus, no cleavage), and efficient invertasome formation requires high density of negative supercoiling (9). These authors suggested that negative supercoiling is required to form invertasome by promoting or stabilizing the physical contact between Hin and Fis. Topological studies of recombinant DNA after inversion suggested that negative supercoiling may provide specific geometry (a branch point in negatively supercoiled plasmid) of recombination sites for the formation of invertasome ( Fig. 1) (10). The dependence of (Ϫ)supercoiling of substrate plasmids has also been well established in many related site-specific recombination systems such as Gin invertase (11,12), ␥␦ resolvase (13), Tn3 resolvase (14), and integrase (15).
Negative supercoiling has been suggested to provide common roles in different site-specific recombination systems (for review, see Ref. 16). First, it provides specific intertwining or geometry of double helix DNA for each recombinase to make proper synaptic complexes required to perform their specific roles, inversion (Hin and Gin; 5, 10), deletion (␥␦ and Tn3 resolvase; 17), and integration ( integrase; 18). Second, the strand exchange (a major mechanism in DNA recombination) is governed by (Ϫ)supercoiling (8,10,19,20). Third, (Ϫ)supercoiling is needed to melt or unwind recombination sites after the productive synaptic complex is formed (21).
Not only the requirement of (Ϫ)supercoiling and its roles but also many biochemical activities of recombinases, such as the DNA cleavage activity in synaptic complexes and binding specificity, are remarkably similar in different site-specific recombination systems. The recombinases described above (except integrase) cleave the middle of recombination sites to which they are bound and are transiently attached to the cleaved DNA by a phosphodiester bond between a serine residue (at position 9, or 10) and the 5Ј-phosphate end of DNA (7,22,23). However, there has been two models to account for strand exchange after DNA cleavage. In one model, subunits of bound dimers exchange along with DNA strands (subunit exchange or simple rotation model; 8,19,24,25). In the other model, protein-guided strand exchange occurs between juxtaposed recombination sites without exchange of subunits (26).
One of the terms to define topological state of a covalently closed circular DNA is the linking number (Lk). Lk is the number of turns one strand of DNA makes around the other. The Lk of (Ϫ)supercoiled DNA is less than that of relaxed DNA (Lk 0 ). If Lk of a covalently closed circular DNA is greater than Lk 0 , then the DNA is positively (ϩ) supercoiled. As the result of the linking number difference (⌬Lk), which is equal to Lk Ϫ Lk 0 , (Ϫ)supercoiled DNA is said to be underwound and (ϩ)supercoiled DNA is overwound (27). Therefore, the handedness of supercoiling between the two different forms is opposite. Al-though there is evidence that positive supercoiling could be biologically relevant (28), covalently closed circular DNA isolated from most of organisms is negatively supercoiled. Thus, it is likely that Hin that requires (Ϫ)supercoiled DNA to perform its roles has evolved without facing (ϩ)supercoiled DNA.
A simple curiosity on what would happen if Hin is reacted with (ϩ)supercoiled DNA drove us to initiate this study. Would Hin and Fis bind to (ϩ)supercoiled DNA? Would they perform inversion as they have done on (Ϫ)supercoiled DNA? If they don't, then in which step is the inversion reaction blocked? Whatever the results (either positive or negative) may be, we contemplated that the results would be informative to understand the system in detail. Furthermore, we were particularly interested in strand exchange on (ϩ)supercoiled DNA because (ϩ)supercoiled DNA is different from (Ϫ)supercoiled DNA in handedness and state of double helical twist that are likely to matter in strand exchange. Therefore, we have generated a positively supercoiled plasmid in vitro using reverse gyrase isolated from a thermophilic Archaeon and subjected it to a Hin-mediated inversion system. Our results showed that on (ϩ)supercoiled plasmid, Hin, and Fis could bind to their binding sites, and Hin could pair hix sites. Detailed assays showed that Hin was able to cleave the hix sites on (ϩ)supercoiled DNA but was not able to exchange the cleaved strands. The cleavage activity of Hin on (ϩ)supercoiled DNA was Fis-dependent as it was on (Ϫ)supercoiled DNA. Therefore, Hin and Fis reacted with (ϩ)supercoiled DNA as they do with (Ϫ)supercoiled DNA except at the strand exchange step. Consequences of these results are discussed in detail.

EXPERIMENTAL PROCEDURES
Proteins and Plasmids-Hin was prepared from Escherichia coli strain DH1 harboring a plasmid pHL104 in which the wild-type hin gene is placed under the control of the tac promoter. Hin proteins were induced with 0.1 mM IPTG at 25°C. The essential steps for purification of Hin were performed as described by Johnson and Simon (2). At the end of purification, fractions containing Hin were pooled and dialyzed against a buffer containing 50% glycerol, and 1 M NaCl. Purified Hin was not contaminated with Fis and was 60% pure as estimated by scanning the Coomassie Blue-stained SDS-polyacrylamide gels (Hewlett Packard Scanjet IICX). Fis proteins were purified to homogeneity by the method of Johnson et al. (4). Restriction enzymes were purchased from New England Biolabs, Inc. Reverse gyrase was isolated from Sulfolobus shibatae as described by Nadal et al. (29). Negatively supercoiled plasmids were purified from E. coli strain DH10B by CsCl ultracentrifugation.
Generation of Positively Supercoiled pKH336 -Negatively supercoiled pKH336 (20 g) was incubated at 75°C for 45 min with 3,000 units of reverse gyrase (29) in an 80-l reaction mixture containing 50 mM Tris-HCl (pH 8.0), 0.55 mM dithiothreitol, 0.55 mM Na 2 EDTA, 10 mM MgCl 2 , 1.25 mM ATP, 7% polyethylene glycol 6000 (Microselect for molecular biology, Fluka), 2.5 mM NaH 2 PO 4 /Na 2 HPO 4 , 10 mM NaCl, 0.005% Triton X-100 (sulfact-Amps, Pierce), and 30 g/ml bovine serum albumin. After incubation, NaCl was added at a final concentration of 500 mM, and the incubation was continued for 2 min. SDS and Na 2 EDTA were added at final concentrations of 0.9% and 9 mM, respectively. After addition of 40 g of proteinase K (Merck), the reaction mixture was incubated at 65°C for 1.5 h. DNA was extracted three times with 2 volumes of chloroform/isoamyl alcohol (24:1) and precipitated with ethanol. The positively supercoiled form was separated from the open circular form on agarose gel, and an agarose block containing the positively supercoiled form was cut out. Positively supercoiled DNA was isolated again from the agarose block using Geneclean (BIO 101, Inc., Vista, CA). Care was taken to minimize nicking during the whole procedure.
Assays for Inversion and Protein Binding-Binding reaction for Hin (30 l) was set up in a buffer containing 50 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 100 mM NaCl. This buffer supports both Hin binding and NcoI cleavage. Different amounts of Hin protein (0, 150, 300, and 600 ng) were mixed with 0.25 g of pKH336 in 30 l of reaction mixture. After a 10-min incubation at 37°C, 7.5 units of NcoI were added to each sample. Incubation continued for an additional 1 h at 37°C. Phenol and chloroform extraction were performed to remove proteins, and the extent of NcoI cleavage was analyzed by agarose gel electrophoresis using TAE buffer (40 mM Tris acetate, pH 8.0, 1 mM Na 2 EDTA). Agarose gels were stained with ethidium bromide (0.5 g/ml) for 10 min followed by destaining in water for 30 min. Agarose gel electrophoresis in the presence of chloroquine was performed in TAE buffer containing chloroquine (4 g/ml) at room temperature. Chloroquine was removed by soaking the gel in 1 mM MgSO 4 for 1 h, and ethidium bromide staining was done as above. Polaroid 665 films were used to take pictures, and negatives were scanned by a densitometer (Hewlett Packard Scanjet IICX) when necessary. Fis-binding assay was performed in a buffer containing 20 mM Tris acetate (pH 7.9), 10 mM magnesium acetate, 1 mM dithiothreitol. Fis (56 ng) was incubated with 1.75 g of pKH336 at 37°C for 10 min in 175 l of reaction followed by addition of 10 units of the restriction enzyme MfeI. An aliquot (25 l) was removed at 0 (before the addition of MfeI) and 20 s and at 1, 2, 4, and 8 min. DNA cleavage activity of MfeI in each sample was immediately blocked by adding 2 l of 500 mM Na 2 EDTA followed by phenol and chloroform extraction. The extent of MfeI cleavage was analyzed by agarose gel electrophoresis as above. In vitro inversion reaction was performed as described by Lim et al. (6) in 25 l of inversion buffer (20 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 100 mM NaCl, 1 mM dithiothreitol, and 100 g/ml polycytidylic acid). Hin (300 ng) and Fis (8 ng) were reacted with 0.5 g of pKH336 at 37°C for 1 h. DNA precipitated by ethanol was digested with PstI and ClaI, and the extent of inversion was measured by agarose gel electrophoresis as above.
Assays for hix-pairing Activity of Hin-This assay was followed as described by Lim et al. (6). Hin (300 ng) was incubated with 0.5 g of pKH336 in 25 l of cleavage buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM dithiothreitol, 10 mM Na 2 EDTA, 100 g/ml polycytidylic acid, 30% ethylene glycol) at 37°C for 20 min followed by the addition of 1 l of 6% glutaraldehyde. Incubation was continued for an additional 20 min at room temperature. Glutaraldehyde was removed by dialysis against 250 ml of buffer containing 10 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , and 100 mM NaCl. Dialysis was performed for 45 min at room temperature using a Microdialyzer TM (Pierce). Dialysates were digested with PstI and ClaI (5 units each). The paired-hix structure was analyzed by agarose gel electrophoresis as above.
Assays for Hin-mediated DNA Cleavage and Strand Exchange-Hin cleavage reactions (7) were performed in 50 l of the cleavage buffer above. Hin (600 ng) and Fis (16 ng) were mixed with 1 g of pKH336.
After a 1-h incubation at 37°C, the reaction was divided in half, and 25 l of each half was transferred to a new tube. To analyze the extent of cleavage, 1 l each of proteinase K (10 mg/ml) and of 10% SDS were added to one tube, and incubation at 37°C was continued for an additional 30 min. This reaction was kept on ice until the other half (for chase) was finished. To the other half of the cleavage reaction, 125 l of the inversion buffer above were added (chasing). Two min after, phenol and chloroform extraction were performed. DNA recovered from ethanol precipitation was digested with PstI and ClaI (5 units each). The cleaved product (from the first half of the reaction) and inverted product (from the second half) were analyzed by agarose gel electrophoresis as described above.
FIG. 1. A schematic representation of Hin-mediated inversion and pKH336. a, Hin (rectangles) and Fis (circles) bind to their binding sites (hix and enhancer, respectively) residing on the plasmid pKH336. Supercoiling is not drawn for clarity. b, at a branch point of (Ϫ)supercoiling, the two hix sites and the enhancer are aligned as shown in the picture with bound proteins to form the synaptic complex (invertasome) (5, 10). Proteins are not drawn for clarity. c, the plasmid pKH336 after inversion. Note that the distance between PstI and ClaI are further apart due to the inversion.

RESULTS
Making Positively Supercoiled Substrate Plasmid pKH336 -Plasmid pKH336 was used as a substrate plasmid in our previous report that dissected the role of negative supercoiling in Hin-mediated inversion reaction (9). We have generated positively supercoiled pKH336 ((ϩ)pKH336) by using reverse gyrase (30, 31) isolated from a thermophilic Archaeon S. shibatae (29) (see "Experimental Procedures"). This enzyme converted negatively supercoiled pKH336 ((Ϫ)pKH336) into a highly (ϩ)supercoiled form in the presence of ATP and polyethylene glycol (Fig. 2). To confirm the integrity of the reverse gyrasegenerated (ϩ)pKH336, agarose gel electrophoresis was performed with and without chloroquine (Fig. 2). Chloroquine, a DNA-intercalating agent, unwinds the DNA double helix upon binding. Because the Lk is the sum of writhe (superhelical turns) and twist, the unwinding effect of chloroquine-binding will lower twist of the plasmid. This, in turn, increases writhe because Lk does not change unless the backbone of DNA is broken and rejoined. Therefore, a certain amount of chloroquine will remove supercoiling (relaxation) from (Ϫ)pKH336 and add more supercoiling to (ϩ)pKH336 (32,33).
In the presence of 4 g/ml concentration of chloroquine (Fig.  2, lane 3), the single band of (Ϫ)pKH336 in Fig. 2, lane 1 was resolved in many topoisomers relaxed by chloroquine (lane 3). However, (ϩ)pKH336 moved as a fast moving single band without (lane 2) and with (lane 4) chloroquine demonstrating that (ϩ)pKH336 generated by reverse gyrase is highly (ϩ)supercoiled. Scanning of negatives showed that supercoiled form in (Ϫ)pKH336 preparation was about 90% and that in (ϩ)pKH336 preparation was 55%. Positively supercoiled DNA seemed to be stained less by ethidium bromide than (Ϫ)supercoiled or open-circular DNA. Thus, the amount of (ϩ)pKH336 in lane 2 of Fig. 2 was underestimated when negatives were scanned. We considered that the amount of (ϩ) topoisomers in (ϩ)pKH336 preparation would be about 60% rather than 55%. Thus, the amount of (ϩ)pKH336 in lane 2 is 67% of that of (Ϫ)pKH336 in lane 1.
Most of the open circular forms in both preparations seemed rather to be nicked than relaxed because they did not change mobility in agarose gel with chloroquine.
To know the minimum linking number difference (⌬Lk) of (ϩ)pKH336 in Fig. 2, lane 2, we have generated a series of topoisomers of (ϩ)pKH336 with a lesser amount of reverse gyrase and have counted the number of topoisomers from the totally relaxed to the point where normal agarose gel cannot resolve. The number was 14 (data not shown). Therefore, (ϩ)pKH336 shown in lane 2 of Fig. 2 is composed of a group of topoisomers with ⌬Lk greater than ϩ14. Because our previous experiments with topoisomers of (Ϫ)pKH336 showed that topoisomers with ⌬Lk of less than Ϫ14 were competent in inversion (9), we assumed that (ϩ)pKH336 with ⌬Lk more than ϩ14 would be competent for assays designed to observe the intermediates in Hin-mediated reactions.
Hin-mediated Inversion Reaction on (ϩ)pKH336 Is Less Than 1% of That on (Ϫ)pKH336 -Next, we have performed a standard Hin-mediated inversion reaction with (ϩ)pKH336. After the reaction, DNA was precipitated with ethanol followed by PstI and ClaI restriction enzyme digestion to observe inversion. All the restriction enzymes used in this work were able to cleave their recognition sites on (ϩ)pKH336 as well as those on (Ϫ)pKH336. Data shown in Fig. 3a demonstrated that the extent of inversion in 1 h with (ϩ)pKH336 is less than 1% of that with (Ϫ)pKH336. When inversion reactions were loaded onto an agarose gel without restriction digestion of substrate plasmids, (Ϫ)pKH336 showed a typical pattern of Hin-mediated relaxed topoisomers due to multiple rounds of inversion (1), but no relaxation was observed in (ϩ)pKH336 (Fig. 3b). These data suggest that Hin could not change the topology of (ϩ)pKH336. To see which step of inversion reaction was blocked on (ϩ)pKH336, a series of biochemical assays that could analyze each step of inversion reaction (binding, pairing of hix, DNA cleavage, and strand rotation) were performed with (ϩ)pKH336.
Hin Binds to hix sites on (ϩ)pKH336 as Well as on (Ϫ)pKH336 -Binding activity of Hin to hix and Fis to the enhancer on (ϩ)pKH336 was analyzed by using restriction enzymes that recognize and cut each binding site. The binding assay is based on the rationale that once the restriction enzyme site is occupied by DNA binding proteins such as Hin or Fis, the site is no longer cleavable by the restriction enzyme (6). Because this binding assay method utilizes supercoiled DNA, it was adequate to measure DNA binding activity of Hin and Fis on (ϩ) and (Ϫ)pKH336. pKH336 has two identical hix sites (hixL-AT; see Ref. 8) containing 5Ј-CCATGG-3Ј sequence that is the cleavage site of restriction enzyme NcoI. The reaction buffer for NcoI cleavage and that for Hin binding were almost  1 and 2). In the presence of chloroquine (4 g/ml), topoisomers of (Ϫ)pKH336 are resolved (lane 3) because chloroquine relaxes (Ϫ)supercoiling. However, (ϩ)pKH336 (lane 4) is still moving fast as a single band because chloroquine makes (ϩ)pKH336 more positively supercoiled. A small amount of linearized pKH336 was generated during preparation of (ϩ)pKH336 using reverse gyrase (lanes 2 and 4). identical (see "Experimental Procedures"). Hin-binding reactions (with an increasing amount of Hin) were incubated at 37°C for 10 min followed by NcoI digestion for 1 h. The NcoI cleavage pattern in the presence of different amounts of Hin between (Ϫ) and (ϩ)pKH336 were identical (Fig. 4), suggesting that Hin binds to hix sites regardless of the handedness of supercoiling.
Fis Binds to the Enhancer on (ϩ)pKH336 -One of the Fisbinding sites in the enhancer (proximal domain) has 5Ј-CAATTG * -3Ј sequence that contains the guanine residue (G * ) critical in Fis binding (34) and can be cleaved by MfeI restriction enzyme. pKH336 has an enhancer sequence between the hix sites. Thus, there is a unique MfeI site in pKH336. In Fis-binding assay with MfeI, we measured the time of MfeI cleavage with a fixed amount of Fis. When Fis was not present, MfeI (1.5 units) was able to digest 0.25 g of (ϩ)pKH336 completely within 4 min. However, in the presence of 8 ng of Fis (the amount used in a standard inversion reaction), about 10% of (ϩ)pKH336 was left uncleaved even after 8 min (Fig. 5a), indicating that the rate of restriction digestion slows down by a factor of little more than 2. The slower cleavage rate of MfeI suggests that Fis binds to the enhancer sequence on (ϩ)pKH336. The same binding experiment was performed with (Ϫ)pKH336, and the results were exactly the same as (ϩ)pKH336 shown in Fig. 5b. As a control, we performed MfeI cleavage reaction on a (Ϫ)supercoiled plasmid, which has a unique MfeI site and no enhancer sequences, with and without the presence of Fis. The cleavage activity of MfeI was not affected by the presence of Fis (data not shown), suggesting that Fis alone does not inhibit the activity of MfeI. These suggest that Fis also binds quantitatively and qualitatively well to the enhancer regardless of superhelical handedness. These data also suggest that the binding of Fis to enhancer is not as stable as shown in gel mobility shift assays (34). The reason for this is, we believe, that the MfeI reaction buffer in which the Fis-binding assay was performed was radically different from that used in gel mobility shift assays.
Hin Can Make Paired-hix Structure on (ϩ)pKH336 -To test whether Hin bound at distant hix sites can interact with each other to bring the hix sites in close proximity on (ϩ)pKH336, Hin was reacted with (ϩ) and (Ϫ)pKH336, and interacting Hin dimers were cross-linked with glutaraldehyde followed by restriction enzyme digestion. The resulting nucleoprotein structures were resolved on agarose gel (9). Bands labeled as T in Fig. 6 represent the paired-hix structure. The amount of T band on (ϩ)pKH336 (lane 4) was 90% of that of (Ϫ)pKH336 (lane 3), suggesting that paired-hix can be formed on (ϩ)pKH336. Con-sidering that the amount of supercoiled population in (ϩ)pKH336 was 67% of that in (Ϫ)pKH336 (Fig. 2) and that (Ϫ)supercoiling promotes formation of the paired-hix structure (9), Fig. 6 demonstrated that Hin makes the paired-hix structure on (ϩ)pKH336 as efficiently as on (Ϫ)pKH336.
hix Sites on (ϩ)pKH336 Are Cleavable by Hin and the Cleavage Activity Needs Fis-Next, we investigated the Hin-mediated DNA cleavage on (ϩ)pKH336. In a buffer containing no Mg 2ϩ and 30% ethylene glycol and 10 mM EDTA, Hin, Fis, and negatively supercoiled plasmid DNA are needed to form the synaptic complex (invertasome) in which Hin cleaves the middle of hix sites and remains attached to the cleaved DNA strand via a phosphoserine linkage (5,7). Thus, to observe the resulting cleaved DNA bands (between the two hix sites and the rest of the vector) in agarose gel, the reaction should be treated with SDS and proteinase K to remove Hin proteins covalently linked to DNA.
The cleaved DNA bands labeled as Cleaved in Fig. 7a (lanes  1 and 2) showed that the extent of Hin-mediated cleavage on (ϩ)pKH336 is about 60% of (Ϫ)pKH336. Again, considering the amount of supercoiled population in preparation of (ϩ)pKH336 is 67% of that in (Ϫ)pKH336 and that the cleavage activity of Hin requires supercoiled DNA, it is likely that the overall  2). In the presence of Hin (ϩ Hin), however, PstI-ClaI digestion of both forms of pKH336 yields an additional band representing the paired-hix structure (lanes 3 and 4). DNA bands labeled as Vector indicate PstI-and ClaI-digested pKH336 that was not engaged in hixpairing. Vector (L), large fragment; Vector (S), small fragment. efficiency of Hin-mediated DNA cleavage on (ϩ)pKH336 is almost the same as that on (Ϫ)pKH336. To resolve this point clearly, we have prepared (ϩ)pKH336 with a lesser amount of open circular form from the preparation of (ϩ)pKH336 shown in Fig. 2 (see "Experimental Procedures"). In this newly prepared (ϩ)pKH336, the open circular form constituted less than 10%, which was the same value of (Ϫ)pKH336 preparation (data not shown). When 0.25 g (standard reaction has 0.5 g of DNA) of both (Ϫ) and newly made (ϩ)pKH336 were subjected to the cleavage assay, Hin showed the same efficiency of DNA cleavage activity (Fig. 7c). Thus, these results demonstrated that Hin was able to cleave the middle of hix regardless of supercoiling handedness.
To see if Fis is required in cleavage reaction with (ϩ)pKH336 as it is with (Ϫ)pKH336, Fis was deleted from the cleavage reaction. As shown in Fig. 7b (lanes 1 and 2), without Fis, the cleavage activity of Hin on both pKH336 dropped. In reactions with (Ϫ)pKH336, the cleaved DNA in the absence of Fis (Fig.  7b, lane 1) was 14% of that in the presence of Fis (Fig. 7a, lane  1), and in reactions with (ϩ)pKH336, the cleaved band formed without Fis was 25% of that with Fis (lanes 2 of Fig. 7, a and b,  respectively). These suggest that Fis is also required for an efficient DNA cleavage on (ϩ)pKH336.
Most of the Cleaved Products Assembled on (ϩ)pKH336 Did Not Lead to Inversion-The invertasome, accumulated in a buffer with 30% ethylene glycol and no Mg 2ϩ , is able to finish inversion within 30 s if MgCl 2 is added and the concentration of ethylene glycol becomes Ͻ5% by dilution (7). When the cleavage reaction with (Ϫ)pKH336 was chased to produce inverted products, more than 95% of the cleaved products (Fig. 7a, lane 1) became inverted (Fig. 7a, lane 3), but less than 5% the cleaved products assembled on (ϩ)pKH336 (Fig. 7a, lane 2) gave rise to inversion (Fig. 7a, lane 4), suggesting that most of the cleaved products on (ϩ)pKH336 were religated back to the original configuration.

Binding Properties of Hin and Fis on (ϩ)Supercoiled DNA
Are Not Changed-Hin was able to bind to hix sites on (ϩ)pKH336. This suggests that Hin protein is flexible enough to accommodate the structural changes of hix sites, such as changes in the helical repeats and the average rotation per nucleotide reside, due to the overwinding of the double helix in (ϩ)supercoiling. This was not a surprising result because Hin has been shown to bind to hix sites residing either on linear DNA or on (Ϫ)supercoiled DNA (underwound) with the same affinity (3). Once Hin binds to hix sites, it stays there more than 2 h on (Ϫ)supercoiled DNA. 2 This was also true on (ϩ)supercoiled DNA used in this study (data not shown).
Unlike Hin, Fis showed unstable binding at the enhancer on (Ϫ)supercoiled DNA (Fig. 5). The comparably unstable binding property of Fis was the same on (ϩ)pKH336 (Fig. 5). The reason for the unstable binding of Fis was because the Fisbinding buffer used in this study was quite different from that used in gel mobility shift assays. However, we believed that the Fis-binding assays we performed are enough to show that Fis binds to the enhancer on (ϩ)pKH336 as well as on (Ϫ)pKH336.
Hin could even make the paired-hix structure on (ϩ)pKH336 as well as on (Ϫ)pKH336, suggesting that Hin dimers bound to each hix site on (ϩ)pKH336 already have acquired conformational changes necessary to physically associate with each other and that (ϩ)supercoiling could also promote the hixpairing activity. It was demonstrated that (Ϫ)supercoiling promotes formation of the paired-hix structure (9). These results draw a general conclusion that binding and some other functional properties (DNA cleavage activity of restriction enzymes and hix-pairing activity of Hin) of the proteins tested in this study are not changed on (ϩ)supercoiled plasmid.
Formation of the Productive Synaptic Complex (Invertasome) on (ϩ)pKH336 -The efficiency of inversion on (ϩ)pKH336 in 1 h was less than 1% of that on (Ϫ)pKH336 (Fig. 3a), suggesting that a step of inversion is blocked or extremely slow on (ϩ)pKH336. Binding, and hix-pairing assays all showed that steps before formation of the synaptic nucleoprotein complex are identical on the two different topological states of pKH336. Three lines of evidence support that Hin is able to make a synaptic nucleoprotein complex on (ϩ)supercoiled plasmid, which are functionally identical to that assembled on (Ϫ)supercoiled plasmid. First, under the invertasome formation condition (30% ethylene glycol without Mg 2ϩ ), Hin was able to cleave the middle of hix sites on (ϩ)pKH336 as it was on (Ϫ)super-  1 and 2). The minor band labeled with an asterisk denotes a cleaved band between the hix site (closer to the enhancer) and a pseudo-hix site (58 bp away from the hix site) (6). If the nucleoprotein complex where the cleavage has completed (as shown in lanes 1 and 2) is chased, Hin is now able to exchange strands and complete the inversion (lanes 3 and 4). coiling. Second, the cleavage activity of Hin on (ϩ)pKH336 also required Fis (Fig. 7). Without Fis, invertasome structures were not observed under electron microscopy, and cleavage activity of Hin dropped from 45 to 11% on a (Ϫ)supercoiled substrate plasmid (5,7). Third, to observe the cleaved DNA fragment between the two hix sites on (ϩ)pKH336 on agarose gel, SDSproteinase K treatment of the cleavage reaction was necessary, suggesting that Hin is covalently attached to the cleaved DNA ends in the synaptic complex formed in (ϩ)pKH336.
In the cleavage reaction with (Ϫ)pKH336, more than 95% of the cleaved DNA readily (in 2 min at most) gave rise to inverted product when the reaction condition was changed for inversion (chasing). However, less than 5% of the cleaved DNA assembled on (ϩ)pKH336 was chased to inversion (Fig. 7a), suggesting that the step after the DNA cleavage, strand exchange, was either inefficient or blocked in the synaptic complex formed on (ϩ)pKH336. However, that most of the cleaved DNA on (ϩ)pKH336 were readily religated back to the original configuration instead of being in the progression of strand exchange in the chase assay (Fig. 7a) suggests that the strand exchange on (ϩ)pKH336 is blocked rather than inefficient. One h of incubation of the standard inversion reaction with (ϩ)pKH336 did not produce more inversion products (Fig. 3a) than did the 2-min chase assay with (ϩ)pKH336 (Fig. 7a). Thus, we believe that the small amount of inversion observed on (ϩ)pKH336, either in the standard inversion or in the chase experiment, could have come from the nicked population of (ϩ)pKH336 or unknown, but inversion competent, topoisomers of pKH336.
Why was Strand Exchange on (ϩ)pKH336 Blocked?-Certain structural changes in hix-bound Hin due to the local alterations in DNA structure, such as different helical repeats or altered average rotation per residue in (ϩ)pKH336, could account for this result. However, Hin and Fis were all involved in DNA cleavage and in religation of the cleaved DNA on (ϩ)pKH336. Furthermore, restriction enzymes were equally functional on both (Ϫ) and (ϩ)pKH336. It is hard to imagine that on (ϩ)pKH336, Hin protein that was inactive in strand rotation became active in religation. Therefore, we eliminated the possibility of structural changes in Hin on (ϩ)supercoiled DNA during strand exchange.
Based on the topological features of the synaptic complex of Gin formed on (Ϫ)supercoiled plasmid DNA, it was suggested that (Ϫ)supercoiling dictates strand exchange and the result from this is the right-handed rotation (clockwise) of the cleaved DNA (20). The (Ϫ)supercoiling-dictated right-handed rotation of the cleaved DNA is energetically favorable because it releases the torsional tension stored in (Ϫ)supercoiling (21). However, under certain conditions, strand exchange can occur through either orientation (counterclockwise or clockwise) in ␥␦ and Tn3-mediated resolution and Gin-mediated inversion reactions (19,20,35), suggesting that the direction of strand exchange on (ϩ)supercoiled DNA that is energetically favorable (if it occurs) would be left-handed (counterclockwise) rotation. Taken together, these considerations suggest that in the synaptic complex formed on (ϩ)pKH336, features other than the superhelical handedness of (ϩ)supercoiling block the strand rotation.
The major different property of (ϩ)supercoiling from that of (Ϫ) is that it is overwound. It is a well known fact that (Ϫ)supercoiling promotes transient unwinding of the double helix. The energy of (Ϫ)supercoiling can induce, by its linking number deficit, spontaneous unwinding of peculiar DNA regions even in the absence of protein. The binding of a protein to this unwound DNA region displaces the equilibrium to the unwound state of the DNA (36). Transient unwinding could be seen in the initiation of transcription during open complex formation (37), and in the beginning of DNA replication at the origin of DNA replication (38). By contrast, since (ϩ)supercoiled DNA has an excess of linking number (overwound), this unwinding reaction would be greatly inhibited.
The difference between (Ϫ) and (ϩ)supercoiling led us to suggest that the synaptic complex formed on (ϩ)pKH336 may be blocked in the process of unwinding probably at the center of hix sites and, thus, indicating that melting at the center of hix could be critical for strand exchange. Recently, Benjamin et al. (21) also proposed that in the postsynaptic complex of Tn3 resolvase and at the Gin recombination sites (gix), (Ϫ)supercoiling might be needed to unwind the double helix. The authors came to this conclusion from the results of the experiments with catenated plasmids. It is interesting that our data with (ϩ)supercoiled DNA also suggests the same conclusion. However, results presented in this work might indicate another important point in understanding the cleavage and strand exchange mechanism in invertases and probably in resolvases. Our data may suggest that the unwinding process in recombination sites is needed after DNA cleavage prior to strand exchange.