Hin Recombinase Mutants Functionally Disrupted in Interactions with Fis

Abstract

A previous genetic screen was designed to separate Hin recombinase mutants into distinct classes based on the stage in the recombination reaction at which they are blocked (O. Nanassy, Zoltan, and K. T. Hughes, Genetics 149:1649–1663, 1998). One class of DNA binding-proficient, recombination-deficient mutants was predicted by genetic classification to be defective in the step prior to invertasome formation. Based on the genetic criteria, mutants from this class were also inferred to be defective in interactions with Fis. In order to understand how the genetic classification relates to individual biochemical steps in the recombination reaction these mutants, R123Q, T124I, and A126T, were purified and characterized for DNA cleavage and recombination activities. Both the T124I and A126T mutants were partially active, whereas the R123Q mutant was inactive. The A126T mutant was not as defective for recombination as the T124I allele and could be partially rescued for recombination both in vivo and in vitro by increasing the concentration of Fis protein. Rescue of the A126T allele required the Fis protein to be DNA binding proficient. A model for a postsynaptic role for Fis in the inversion reaction is presented.

The Hin recombinase of Salmonella spp., in conjunction with the Fis and HU accessory proteins, catalyzes a reversible site-specific recombination reaction that results in the alternate expression of flagellin antigens, a process known as flagellar phase variation. The Hin-Fis site-specific recombination reaction provides a model system for the investigation of the molecular mechanism of genetic recombination. The recombination reaction catalyzed by the Hin recombinase takes place between two chromosomal sites, hixL on the left and hixR on the right, that flank an invertible DNA segment (27). The underlying molecular mechanism of recombination has been shown to require three proteins, Hin, Fis, and HU (14), and three DNA sites, hixL, hixR, and a recombinational enhancer element (RE) (8, 13). Hin dimers bound to the hixL and hixR recombination sites are thought to act with Fis dimers bound to two sites within the RE during the inversion reaction. The role of HU is to bend DNA, facilitating the assembly of a complex between Hin and Fis. Immunoelectron microscopy has permitted visualization of the nucleoprotein intermediate, or invertasome, that is responsible for carrying out strand exchange (10). By analogy to the closely related Gin recombinase system described below, Hin-Fis-directed invertasome formation is thought to trap DNA supercoils such that after recombination four negative supercoils are lost, effectively driving the reaction to completion. The isolated invertasome structure included Hin and Fis, which were colocalized in the cross-linked complex (10). Fis aids the binding of Hin to the hixR site in vivo (23) and activates the Hin dimers within the synaptic complex in order to initiate concerted DNA cleavage (7, 21).

In the closely related Gin invertase system, two negative nodes are trapped as a result of invertasome formation (16–18). Strand exchange is initiated after cleavage of the DNA by the recombinase within the invertasome, and single right-handed (clockwise) rotation about the helix axis, followed by religation, results in a net loss of four negative supercoils (a +4 change in the linking number of negative supercoils) (9, 16, 28, 30). The energy which is initially trapped in the supercoils drives the inversion reaction. Fis may initially act as a topological filter by favoring the trapping of exactly two nodes in the DNA within the invertasome (2, 7). This putative filtering role of Fis facilitates Hin-mediated DNA strand exchange between hixL and hixR to generate a site-specific inversion event. These data from the invertase systems suggest that supercoiling and Fis are both required for invertasome assembly and in the later stages of the strand exchange reaction (2, 7, 19, 21).

A genetic assay for detecting protein-protein interactions within various Hin-Fis-DNA complexes that effect the repression of the ant gene of phage P22 has been previously described (23). This system exploits the fact that Hin will bind hix sites near the ant promoter (P_ant-O_hix-ant) and repress transcription of the downstream ant structural gene of P22 (12). The ant gene encodes antirepressor, an inhibitor of P22 lysogenic repressor c2 (equivalent to cI of phage lambda) (31). Hin binds hixR and hixL recombination sites that are placed at the normal operator position within the ant promoter region and represses transcription. Hin also binds, and represses the transcription at a fully inversely symmetrical hix site, termed hixC, whose sequence is based on a hix half-site sequence present in both hixL and hixR (11). In the absence of Hin the ant gene of phage carrying the P_ant-O_hixC-ant construct is fully expressed, resulting in lytic growth. The frequency of lysogeny under these ant gene-derepressed conditions is less than 1/10⁸ (12). In the presence of Hin the ant gene of phage carrying the P_ant-O_hixC-ant construct is repressed, resulting in a mixture of lytic and lysogenic growth. The frequency of lysogeny under Hin-repressing conditions is on the order of 10 to 40% (11, 23). Mutations in the hix sites that decreased Hin binding result in increased transcription (derepression) of the ant reporter gene. A mutant hixC site, 10G, has symmetric base substitutions of T:A to G:C at position −10 and A:T to C:G at position +10 (11). The frequency of lysogeny of phage carrying P_ant-O_10G-ant is about 1/1,000 in strain LT2 compared to 1/10 for a hixC site (23). By adding a second wild-type copy of the hix site about 1 kb upstream of the hix binding-defective operator, it was possible to detect suppression of the binding defect of the hixC-10G symmetric mutant site by means of decreased transcription of the reporter gene; the frequency of lysogeny increases to 1/100 (23). It was surmised that the suppression of the defective hix operator by the second wild-type hix site upstream might be due to Hin tetramer formation (T⁺). The ability of a recombination-deficient Hin mutant protein to exhibit the phenotype of upstream suppression in binding to a mutant hix operator site near a second wild-type hix sequence was taken as evidence that the mutant Hin was essentially equivalent to wild-type Hin in the formation of Hin tetramers (T⁺). A higher level of suppression of the mutant hix operator was obtained when both a wild-type hix site and the RE sequence were placed at the same position upstream of the defective hix operator. The frequency of lysogeny was similar to that for a strain with a wild-type hixC site (1/10). It was hypothesized that this configuration of DNA sites (wild-type hix plus RE placed 1 kbp upstream of P_ant-O_hix-) allows this assay to determine if the invertasome complex is formed in vivo on the infecting phage and that the ability to form the invertasome (I⁺) enhanced binding and repression at the defective hix operator site. Since invertasome formation requires both Hin-hix interactions and Fis-RE interactions, these assays were postulated to detect interactions within various Hin-Fis-DNA complexes. Using this system, it was possible to classify hin recombination-deficient mutants as mutant types that form tetramers in vivo (T⁺) but that cannot carry out recombination (R⁻). Among the T⁺ mutants was a class that could also form invertasomes in vivo (T⁺ I⁺) and a class that could not (T⁺ I⁻). Another class of mutants that was unable to suppress the mutant hix operator with either the upstream hix site alone or the hix site and the RE (T⁻ I⁻) was obtained. It remains an important goal to understand how this genetic classification relates to the individual biochemical steps in the recombination reaction.

This study presents biochemical evidence that some of the mutants from the T⁺ I⁻ genetic class are impaired in interactions with Fis and addresses the functional role(s) of Fis in the strand exchange reaction. The distinctive localization of the point mutations that characterize this genetic class in a particular region of the putative dimerization helix of Hin initially suggested the hypothesis that this mutant class may help in understanding one specific step in the recombination reaction. The T⁺ I⁻ Hin mutants genetically classified as defective for interactions with Fis in vivo were characterized biochemically, along with mutants from other genetic classes. The results show that one of these mutants, A126T, requires higher concentrations of Fis both in vivo and in vitro for its own maximal catalytic activity than does wild-type Hin. This supports our previous genetic results that suggested that this mutant is defective in interactions with Fis.

MATERIALS AND METHODS

Bacterial strains and plasmids.

Plasmid pMS621 (15) was digested with PvuII, and a cassette containing the lacI^q gene flanked by two EcoRI sites that had been filled in with a fragment of DNA polymerase I (Klenow fragment) from vector pMS421 (6) was inserted to yield the pZN38 plasmid. This plasmid is a Hin expression plasmid containing the ColE1 origin of replication. Plasmid pRJ807 (24) was taken through the identical steps as pMS621 above to yield the pZN28 plasmid. This plasmid is a Fis expression plasmid containing the ColE1 origin of replication. The Hin and Fis wild-type and mutant proteins were overexpressed in the RJ1632 background and purified as described previously (1, 24). The coding sequences for the A131V, S70G, D65N, T124I, and A126T Hin mutants were cloned into the pZN38 vector by digesting this vector with ClaI and HindIII to excise the fragment of the wild-type hin gene encoding the C terminus and replacing it with the same fragment from the pKH66-derived hin mutant expression plasmids (23) to yield the pZN42, pZN56, pZN64, pZN66, pZN68, and pZN70 plasmids. All these expression constructs were sequenced in the strain background used for purification prior to the actual purification. The R123Q Hin mutant was purified using a pMS571-derived (15) expression plasmid which was cloned as described previously (1) with the appropriate pKH66 derivative containing the hin mutation resulting in the R123Q mutant (23). Strain RJ2843, kindly provided by Reid Johnson, was used for purification of the R85H Fis mutant (24).

DNA sequence analysis of binding-proficient, recombination-deficient (B⁺ R⁻) hin mutants.

The sequencing of all constructs described was performed on purified plasmid DNA using the Applied Biosystems Inc. automated sequencing method in accordance with the manufacturer's instructions. The following primers, which were obtained commercially (Macromolecular Resources, were used: HinD, TGGAAATTAGACAGACTG; HinE, TTATATCCATCCTGTTGT; Hin2, TACTGGTATCAATACTAT; P_tacS, GCTCGTATAATGTGTGGAATTGTG.

Media.

Media conditions, concentrations of antibiotics and lactose indicators, transductional crosses, and transformations were as reported previously (1, 5, 23).

In vitro Hin activity assays with purified proteins.

DNA cleavage and inversion reaction conditions were described previously (7, 13). For the inversion assays, plasmid DNA was digested after the reaction was allowed to proceed with PstI and HindIII, which cleave outside and inside the inverted region, respectively. pMS551 (0.1 pmol) (15) was used as a substrate. Time course reactions were typically initiated by the addition of the purified proteins to the total reaction volume. Aliquots (25 μl) were removed at the respective times following initiation of the reaction. Cleavage reaction mixtures were incubated at 37°C, and reactions were stopped by the addition of 2 μl of 10% (wt/vol) sodium dodecyl sulfate (rapid quenching important) and 2 μl of 2-mg/ml proteinase K (Boehringer Mannheim) followed by incubation at 37°C for 30 min and 65°C for 10 min. Inversion reactions were quenched after incubation at 37°C by adding diethyl pyrocarbonate to a final concentration of 0.08% (vol/vol) and incubating for >30 min at 65°C in an open-capped ultracentrifuge tube. Distilled, deionized H₂O (ddH₂O) was added back to 25 μl, and the PstI and HindIII restriction enzymes were added for another 1 h of incubation at 37°C. Reaction products were analyzed on a 1% agarose gel following electrophoresis at 2 V/cm in a 1× Tris-acetate-EDTA buffer (26). The gels were stained in a 2-μg/ml solution of ethidium bromide in gel running buffer for 20 min and destained in ddH₂O for at least 30 min. Reaction products were visualized using a Foto Eclipse system (Fotodyne), and the tagged-image format files were subsequently analyzed using the Image Quant software package (Molecular Dynamics). In order to increase the accuracy of quantifications, mutant and wild-type proteins were run on the same gel in order to express mutant protein activities relative to wild-type protein activity.

In vitro Hin activity assays with proteins overexpressed in crude lysates.

In cases where the biochemical activity of the Hin protein overexpressed in crude lysates of cells was assayed, the methodology described above was followed except for the following modifications. Reactions were typically initiated by the addition of 20 μg of crude lysate in a 25-μl total reaction volume. Cleavage reaction mixtures were incubated at 37°C for 180 min and stopped by the addition of 2 μl of 10% (wt/vol) sodium dodecyl sulfate (rapid quenching important) and 2 μl of 2-mg ml⁻¹ proteinase K (Boehringer Mannheim) followed by incubation at 37°C for 30 min and 65°C for 10 min. Inversion reactions were quenched after 120 min of incubation at 37°C by phenol-chloroform (1:1) extraction, followed by two chloroform extractions and precipitation of the DNA.

In vitro DNA binding activities of Hin.

Binding studies were performed as described previously, with the following modifications (1, 23). Various amounts of purified Hin protein were used to determine the apparent K_d values. CHAPS {3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate} was used at 20 mM in the binding reactions, and a 120-bp fragment from plasmid pJK110 (J. E. Karlinsey and K. T. Hughes, unpublished data), which includes the hixR site, was used to make the hixR probe.

In vivo Hin activity assays.

Escherichia coli strains TH613 (23) and RJ1651 (7) were used to measure in vivo inversion as described in the figure legends on MacConkey lactose indicator plates (lactose concentration, 0.2%).

RESULTS

Biochemical activities of purified Hin mutant proteins reflect their genetic classification.

A genetic screen described in the introduction predicted that recombination-deficient hin mutants could be classified according to the stage of the recombination reaction these mutants were blocked in: tetramer formation (T⁻ I⁻), invertasome formation (T⁺ I⁻), or post-invertasome formation (I⁺) (23). In order to determine if the genetic classification corresponded to the ascribed biochemical step specified, representative mutant Hin proteins from three genetic classes (I⁺, T⁺ I⁻, and T⁻ I⁻) were purified using standard protocols and characterized (7). Purified Hin, Fis, and HU proteins catalyze DNA recombination on a supercoiled DNA plasmid substrate containing two hixL sites and the RE. The same assay conducted in the absence of Mg²⁺ and in the presence of EDTA and ethylene glycol allows for cleavage of DNA by Hin but not recombination (13). The in vitro cleavage and recombination rates, relative to those for wild-type Hin, were determined kinetically using purified proteins along with the apparent dissociation constant of each hixC site mutant (Table 1). These results demonstrate that two of the hin mutants that had been classified as defective in the step preceding invertasome formation (T⁺ I⁻), the T124I and A126T mutants, are only partially defective in their biochemical activities. The T124I mutant displayed a low rate of cleavage of the DNA substrate (4% of wild type). Other data obtained in the course of characterizing this mutant's biochemical activity in crude extracts suggested that the T124I mutant may also accumulate cleaved substrate molecules even under recombination conditions (data not shown). The A126T mutant displayed lower cleavage and recombination rates than the wild-type protein. In contrast, the R123Q mutant (T⁺ I⁻) was completely defective for in vitro cleavage and recombination activities and may have the strongest phenotype for mutants from this class.

TABLE 1.

In vitro activities of Hin mutant proteins^a

Form of protein	Relative rate of:		Apparent Kd (M)
	Cleavage	Recombination
WTc	1.0	1.0	10−9
D65N	<0.01	<0.01	10−9
A126T	0.5	0.1	10−9
T124I	0.04	<0.01	10−8
R123Q	<0.01	<0.01	10−9b
S70G	0.06	0.1	10−9
A131V	0.6	0.66	10−9

^aInversion and cleavage rates (double-strand cleavage events at both hix sites only) were determined as described in Materials and Methods. The rates for mutant and wild-type Hin proteins on the same gel were compared, and reaction rates for the mutant proteins were expressed relative to the higher activity of the wild-type protein. Dissociation constants were measured by gel shift assays of a fragment of DNA containing the hixC site. Apparent binding constants are only approximate due to the uncertainty in exactly measuring the amounts of Hin added to the reaction mixtures. 

^bEquivalent dissociation constants were measured by gel shift assays of fragments of DNA containing the hixL and hixR sites, but not the hixC site. 

^cWT, wild type. 

The D65N mutant from the T⁻ I⁻ genetic class was inactive for both cleavage and recombination. The lack of detectable cleavage or recombination activity for the D65N mutant in vitro is consistent with its inability to suppress poor binding at the defective hix site in either the tetramer or invertasome assays in vivo. In contrast, the S70G mutant from the I⁺ genetic class displayed 10% of the wild-type protein's cleavage and recombination activities. The A131V mutant, which was not classified genetically due to its in vivo binding defect on the 10G site used in the genetic assay for protein-protein interactions (23), had both cleavage and recombination activities decreased by less than twofold relative to those for the wild-type protein.

To further examine the properties exhibited by mutants within the various genetic classes, we also characterized the biochemical activities of these mutants using crude extracts prepared from the identical strains used for protein purification. These experiments were undertaken as a control for the possible rapid degradation of either the cleavage or recombination activities of the mutant proteins during the course of the 24-h purification protocol. Cell extracts were prepared with the same expression system used for purification of the proteins and immediately used for cleavage and recombination reactions (see Materials and Methods). Maximal cleavage activity, but the absence of any recombination activity, for the T124I mutant protein were consistently observed immediately after cell lysis (Fig. 1). Overall, the results obtained were qualitatively identical to those shown in Table 1 obtained using purified proteins.

FIG. 1.

Effect of amino acid substitutions on Hin recombination and cleavage. Standard recombination (lanes 1 to 7) and cleavage (lanes 8 to 14) reactions were performed as outlined previously (23). However, the Hin proteins were overexpressed from a higher-copy-number (pBR322 origin-based) plasmid in the cell extracts (see Materials and Methods). Reaction product bands visualized following agarose gel electrophoresis are labeled. For recombination assays (lanes 1 to 7), arrows 2 indicate the expected inverted (recombinant) products. Plasmids that have undergone inversion result in a different distance between a PstI restriction site outside the invertible region and a HindIII restriction site located within the invertible region. Arrows 1 indicate the nonrecombined parental plasmid substrate bands. For cleavage assays (lanes 8 to 14), arrow 3 indicates nicked substrate molecules in cleavage reactions, while arrow 4 indicates substrate molecules cut at only one hix site. The vector (arrow 5) and invertible segment (arrow 7) are the expected cleavage reaction products after removal of the proteins from the nucleoprotein complexes and analytical gel electrophoresis. Unreacted supercoiled substrate DNA in the cleavage reactions is also noted (arrow 6). The results shown were consistently obtained for at least two independent preparations of each mutant or wild-type (WT) protein.

Increased Fis concentration rescues both the in vivo and in vitro recombination defect of the A126T mutant.

Having established that the A126T and T124I mutants from the T⁺ I⁻ class exhibit residual biochemical activity (cleavage and/or recombination), we wanted to better understand the molecular causes of their defect(s). Genetically, the larger T⁺ I⁻ class was inferred to be defective in formation of the invertasome, possibly due to a loss in interactions with Fis; thus the possibility that the mutations may influence potential Hin-Fis interactions during the reaction was examined. We wanted to determine whether it was possible to increase the in vivo recombination activity of the hin mutants by providing more Fis protein in trans. The in vivo recombination assay qualitatively measures the ability of Hin expressed in trans from a plasmid to invert a DNA fragment containing a promoter on a lambda prophage in E. coli (3). Recombination activity (R⁺) due to Hin expressed from the plasmid in the tester strain inverts the promoter in a direction where it will transcribe the lac operon. Thus, recombination results in a Lac⁺ phenotype on MacConkey lactose plates following growth at 37°C for at least 36 h, whereas the R⁻ hin mutants yield only Lac⁻ colonies (tested up to 72 h). By eliminating the chromosomal copy of the fis gene and providing wild-type fis on a plasmid (pZN28), the Lac⁺ phenotype appeared more quickly (<36 h) in the otherwise isogenic tester strain.

When hin mutants were subcloned to a higher-copy-number (pBR322) expression system with only the chromosomal copy of fis present in the tester strain, the A131V and S70G mutants exhibited nearly wild-type levels of recombination activity in vivo while A126T did not (Table 2). Recombination activities for wild-type and mutant hin proteins were measured in strains harboring either a single copy of fis or fis expressed from a higher-copy-number plasmid (ColE1 origin) on MacConkey lactose plates (Table 2). With the hin mutants expressed from the same low-copy-number (pSC101) expression vector used in the original mutant screen, we observed a slight-gain-of-function phenotype for the recombination activity of the A126T, A131V, and S70G mutants when fis was expressed in trans from higher-copy-number plasmid pZN28. In contrast, the recombination activities of the D65N, R123Q, and T124I mutants were not rescued by increasing the fis gene dosage. Thus, the recombination defect in the A126T mutant cannot be rescued by simply increasing the concentration of the Hin protein in vivo, and instead higher concentrations of Fis protein are required for the observed complementation. These data are consistent with the hypothesis that the A126T mutant has lost the ability to respond to Fis properly in order to carry out recombination. An alternative hypothesis is that A126T is disrupted directly in Hin-Fis interactions in vivo. Given that this effect was observed in vivo, it was important to determine whether it is also possible to rescue the recombination activity of A126T by increasing Fis concentrations in vitro.

TABLE 2.

In vivo activities of Hin mutant proteins^a

Form of protein	Low-copy-no. Hin expression in:		High-copy-no. Hin expression in strain with chromosomal fis gene
	Strain with Chromosomal fis gene	Fis-overexpressing strain
WTb	+++	+++	+++
D65N	−	−	−
A126T	−	+	−
T124I	−	−	−
R123Q	−	−	−
S70G	−	+	+++
A131V	−	+	+++

^aIn vivo Hin-mediated recombination phenotypes were scored in the indicated strain backgrounds (see Materials and Methods). +++, strong stimulation (red colonies within 36 h); +, weak stimulation (red colonies within 72 to 96 h; −, no stimulation (colonies remain white after 96 h). 

^bWT, wild type. 

Fixed-time-point recombination reactions were conducted with A126T and wild-type Hin at increasing Fis concentrations (Fig. 2). Increasing amounts of recombination product were observed for the A126T mutant at higher Fis concentrations. This suggested that the recombination defect of the A126T mutant may be rescued by Fis (see also below). Since the T124I mutant did not exhibit any residual recombination activity compared to A126T, we wanted to test whether the in vitro recombination activity of this mutant could also be rescued by increasing the Fis concentration in the reaction mixture. No detectable recombination activity was observed for the T124I mutant protein in reaction mixtures with 0- to 200-ng Fis amounts tested up to 450 min (data not shown). Similarly, Fis was not able to restore either cleavage or recombination activity for the R123Q mutant protein in reaction mixtures with 0- to 200-ng Fis amounts tested up to 450 min (data not shown). Thus, the in vitro rescue of the recombination defect of A126T by Fis is a distinctive property of this mutant from the genetic T⁺ I⁻ class.

FIG. 2.

Effect of Fis on the recombination activities of Hin mutant A126T and the wild-type (WT). Standard recombination reactions were carried out with 200 ng of purified Hin mutant or wild-type proteins and various Fis amounts. Reaction products were analyzed and labeled as for Fig. 1 after a 90-min reaction time course. Plasmids that have undergone inversion result in a different distance between a PstI restriction site outside the invertible region and a HindIII restriction site located within the invertible region. Inverted (recombinant) products (arrows (2) are the expected inversion reaction products after analytical digests and gel electrophoresis. Arrows 1, parental (nonrecombinant) vector bands.

To show complementation of the A126T mutant in vitro recombination activity by Fis, recombination activities of A126T and wild-type Hin proteins were measured at varying Fis concentrations in time course experiments. Since wild-type Hin is more active than the A126T mutant, experiments with wild-type Hin went to completion at earlier time points. Two different Fis preparations and two different preparations of the A126T mutant were used in these experiments, yielding similar results. The reaction products were quantified for comparing the kinetics of the appearance of recombination products for the A126T Hin mutant and wild-type proteins (Fig. 3). The recombination activity of A126T at 0- to 12-ng Fis amounts is very low compared to that of wild-type Hin. Even at the maximal Fis amount tested (100 ng), the recombination rate for A126T only started to approach that for wild-type Hin at 3- to 12-ng Fis amounts.

FIG. 3.

Kinetics of appearance of recombination products for the A126T Hin mutant at different Fis concentrations. Graphs depict recombinant product bands (smaller-sized fragment) of standard kinetic recombination reactions carried out with 200 ng of mutant or wild-type (WT) Hin protein and various Fis amounts (in nanograms; brackets). Time points (minutes) at which reactions were stopped are indicated. Reaction products visualized following agarose gel electrophoresis were quantified using the Image Quant software package (see Materials and Methods).

Based on these data it is possible to estimate that the A126T mutant requires about 10-fold more Fis, stoichiometrically, in order to carry out recombination at its intrinsic maximal rate than wild-type Hin. These results confirmed those of the fixed-time-point experiments in demonstrating that the A126T mutant protein only exhibits efficient recombination activity at higher Fis concentrations. Taken together with the specific in vivo functional complementation by Fis, these in vitro data for the A126T mutant support the hypothesis that this mutant is disrupted in interactions with Fis. In addition, these interactions with Fis are required for efficient progress through the recombination reaction by the A126T mutant.

Complementation of the A126T recombination defect by Fis requires the DNA binding activity of the Fis protein.

We wanted to examine whether the function of Fis as a positive regulator of recombination activity for the A126T Hin mutant required the DNA binding activity of the Fis protein. There was the possibility that direct Hin-Fis protein-protein interactions that positively affect recombination in an enhancer-independent fashion occur during this postsynaptic step. The A126T mutant does not recombine efficiently at Fis concentrations of 3 ng after a 15-min reaction time point. We tried complementing its weak recombination activity by incubation with 3 ng of wild-type Fis for 5 min, followed by 27 ng of previously characterized Fis mutant R85H, which is stably expressed but which does not bind DNA, for an additional 10 min (25). The recombination activity of the A126T Hin mutant in the presence of 3 ng of wild-type Fis was comparable to its recombination activity in the presence of 3 ng of wild-type Fis mixed with 27 ng of an R85H Fis mutant after 15 min (Fig. 4). Control experiments indicated no positive effect on the efficiency of recombination by wild-type Hin with 3 ng of wild-type Fis alone versus 3 ng wild-type Fis plus 27 ng of R85H Fis. These results suggest that the positive effect that Fis has on the recombination activity of the A126T Hin mutant requires the DNA binding activity of Fis. Alternatively, the R85H mutant may exert a dominant-negative effect on the molecular step in recombination that is defective for the A126T Hin mutant protein. However, we view this as less likely since if such a dominant-negative effect existed for R85H Fis, it would not be readily observable under our assay conditions with wild-type Hin.

FIG. 4.

Effect of DNA binding activity of Fis on the ability to suppress the Hin A126T recombination defect. Standard recombination reactions were carried out with 200 ng of Hin mutant or wild-type (WT) proteins and various Fis amounts. Reaction products were analyzed and labeled as for Fig. 1 after a 15-min reaction time course. For experiments where two types of Fis proteins were added (or the same type added at different times as a control, i.e., 3 ng of Fis plus 27 ng of R85H Fis), the first Fis amount represents the amount of Fis (wild-type or mutant) added during the first 5 min of the reaction while the second represents the additional Fis (wild-type or mutant) added after 5 min for the remaining 10 min of the 15-min incubation.

DISCUSSION

The catalytic steps in Hin-mediated DNA recombination have been postulated to involve both Hin-Hin interactions between Hin dimers and Hin-Fis interactions within the invertasome. A specific function for Hin-Fis interactions in the coordinate activation of Hin for cleavage has also been described (21). Cleavage is defined here as the concerted nucleophilic attack of the individual Hin subunits at the hix sites. However, there has been little progress in describing any intermediate steps in the inversion reaction that follow cleavage. By devising a genetic approach that dissects the inversion reaction into intermediate steps, we postulated that it would be possible to detect the pairing of two hix sites by Hin, as well as the formation of the invertasome in vivo. The isolation of mutants specific to each of these steps allowed the in vivo identification of intermediate steps in the reaction. In this paper, we have characterized Hin mutants from among all the genetic classes identified and find that their biochemical activities when purified are consistent with their earlier genetic classification and preliminary biochemical characterization (23). Moreover, the T124I and A126T mutants from the T⁺ I⁻ genetic class retain some residual cleavage and/or recombination activities which are suggestive of some potential roles for Fis in the inversion reaction.

We predicted that the T⁺ I⁻ genetic class would be proficient in Hin tetramer formation but defective in the step between Hin tetramer formation and the interaction with Fis to form the invertasome (23). For this reason we were not surprised to find that one T⁺ I⁻ mutant, A126T, could be rescued both in vivo and in vitro by excess Fis. How excess Fis can rescue the A126T mutant is not known. It is certainly possible that the effect is allosteric. It is also possible that the T⁺ I⁻ hin mutations at positions 123, 124, and 126 define a region of Hin that interacts with Fis to form the invertasome. A predicted model of the invertasome places the DNA strands between Hin dimers and locates Hin-Fis interactions at the opposite end of the E helix of Hin in the vicinity of position 99G (21). It is possible that the mutations at positions 123, 124, and 126 identify a Hin-Fis interaction that occurs during the process of invertasome formation. Another possibility is that the identification of the T⁺ I⁻ genetic class at positions 123, 124, and 126 supports an alternative invertasome structure that positions Hin dimers between the DNA strands in the invertasome structure. This would place the C-terminal end of helix E in sufficient proximity to allow contacts between Fis and the A126 region. This would be in agreement with the proposed structure of the related γδ resolvase synaptosome structure proposed by Murley and Grindley (22).

Mutants that separate the cleavage and recombination functions in the resolvase and invertase families have been previously described (7). These mutants, F105V and F105V/M109I, were localized in the putative dimerization helix of Hin and were blocked at the strand exchange step to a greater extent than at the cleavage step. These mutants retained about 3% cleavage activity relative to the wild type and had no detectable recombination activity. This in vitro phenotype is very similar to that exhibited by our T124I mutant. Indeed, the T124I, F105V, and F105V/M109I mutant protein activity profiles under normal reaction conditions may represent a clean-cut separation of Hin's cleavage and recombination activities. The T⁺ I⁻ A126T mutant also showed reasonably good Fis-activated cleavage, which implies that the invertasome is assembled. It is possible that the invertasome in this case is sufficiently stable to activate cleavage with ethylene glycol but too unstable in vivo to score in the repression assay.

Since the A126T Hin mutant is impaired in its ability to recombine in vitro, our earlier genetic classification suggested the hypothesis that Hin-Fis interactions are disrupted for this mutant. However, it appears that Hin-Fis interactions are not permanently disrupted for the A126T mutant because increasing the Fis concentrations both in vitro and in vivo complemented its activity. Stoichiometric concentrations of Fis are required in the normal inversion reaction (10), and all our biochemical assays described in this paper except those for the minus Fis controls provided adequate concentrations of Fis to saturate the enhancer. Since the DNA binding activity of Fis is required for complementation of the A126T mutant, the simplest interpretation of our data is that Fis bound at other, perhaps nonspecific, sites on the DNA is able to at least partially suppress the A126T mutant's recombination defect. However, our data do not characterize the Hin-Fis interaction for the A126T mutant beyond its functional aspects.

Multiple postsynaptic roles for Fis in Hin-mediated DNA recombination: analogies to other site-specific recombination systems.

The first postsynaptic role of Hin-Fis interactions in the activation of cleavage by Hin (21) was consistent with an earlier model for Hin-mediated DNA strand exchange. According to this model, after invertasome formation individual Hin monomers within each dimer undergo a conformational change about the dimerization helices in the presence of Fis, catalyzing cleavage of the DNA phosphodiester backbone (4, 20). This model also posits dynamic interactions between the dimerization helices of the Hin monomers that were corroborated by the structure of the γδ resolvase-DNA cocrystal (32). However, previous studies have been unable to address questions regarding the mechanism of catalysis following cleavage. The data presented here and those from previous studies of the Hin invertase system suggest a model for at least two types of postsynaptic Hin-Fis interactions during the catalytic cycle.

The second type of postsynaptic Hin-Fis interaction proposed in this model suggests that Fis facilitates recombination following cleavage of the DNA substrate, possibly by its presence at other nonspecific DNA sites on the recombination substrate. This model is suggested by our in vitro and in vivo complementation data for the A126T mutant. We hypothesize that Fis may facilitate the dynamic motions about, or of, the dimerization helices during strand exchange by direct interactions with Hin. What clues can a positive role for Fis in the transition from cleavage to strand exchange provide regarding strand exchange mechanisms of the larger resolvase and invertase recombinase families?

In the γδ resolvase-DNA cocrystal structure, a 26° kink at the γδ resolvase Asn-127 (corresponding to the Hin Leu-125) in the structure of one of the two dimerization helices was observed (32) (Fig. 5). The variable structure of this helix suggested that it could serve as a pivot point to swing DNA relative to the amino-terminal catalytic domain during strand exchange (32). We modeled the location of the T⁺ I⁻ mutants onto a hypothetical Hin structure that is based on the structure of the homologous γδ resolvase-DNA cocrystal (Fig. 5). These mutants localized in relative proximity to the kink in the dimerization helix. One interpretation of the biochemical activities of the T124I and A126T Hin mutants is that the associated mutations interfere with motions of the dimerization helices during strand exchange. Such large-scale molecular motions are consistent with the subunit exchange model for DNA recombination that was initially proposed for the resolvase family of recombinases (28, 29). This model postulates that two resolvase monomers dissociate, exchange partners, and reassociate during the course of the strand exchange reaction. Interference with subunit exchange may lead to the spectrum of activities and phenotypes observed for the T⁺ I⁻ mutants. For the A126T mutant, the observable result may have been to simply generate a slower recombinase.

FIG. 5.

Location of Hin residues 123, 124, and 126 within the Hin monomer. Shown is a side-by-side view of the two hypothetical conformations of the Hin monomers, which were modeled based on the homologous γδ resolvase-DNA cocrystal structure (32). The R123Q, T124I, and A126T mutants are labeled along the longer α helices (E and E′) that make up the putative dimer interface in Hin.

An alternative mechanism by which such interference may occur is by the Hin dimerization interactions becoming tighter as a result of these mutations. Assuming that Fis facilitates the release of dimeric contacts in Hin (which is consistent with the subunit exchange model discussed above), this release would be required to initiate the rotation of the top pair of recombinase subunits (9). The A126T mutation may alter the dimerization of the Hin protein in a general or an invertasome-specific fashion. Indeed, the location of alanine-126 in the proposed structure of the protein is consistent with this residue being a part of the proposed dimer interface (7) (Fig. 5). However, for the T124I and A126T mutants, which form putative tighter dimers, progress past this step is blocked. A defect in overcoming this rate-limiting step would make the A126T Hin mutant more Fis dependent than wild-type Hin. This hypothesis will be further tested in future studies. In summary, this model presents a speculative second role for Fis in facilitating a postsynaptic molecular step in the Hin-mediated recombination reaction, the dissolution of the Hin dimer en route to subunit exchange.

Interestingly, Fis-independent mutants identified in the Hin system have also been located within the dimer interface of the molecule (7, 21). In contrast with the stronger Fis dependence for recombination (but not cleavage) by A126T, the Fis-independent mutants may have weaker dimer interactions that allow for easier energetic activation past this rate-limiting postsynaptic step. It is the transition through this step that we infer is a consequence of the second postsynaptic role of Hin-Fis interactions in the catalytic mechanism of this recombination system.