Interaction of the Gifsy-1 Xis Protein with the Gifsy-1 attP Sequence

Abstract

The Gifsy-1 phage integrates site specifically into the Salmonella chromosome via an integrase-mediated site-specific recombination mechanism. Initial genetic analysis suggests that Gifsy-1 integrase-mediated excision of the Gifsy-1 phage is influenced by proteins encoded by both the Gifsy-1 and the Gifsy-2 phages. Our studies show that the Gifsy-1 Xis protein regulates the directionality of integrase-mediated excision of the Gifsy-1 phage. Electrophoretic mobility shift assays, DNase I footprinting, dimethyl sulfate (DMS) interference assays, and DMS protection assays were used to identify a 31-base-pair sequence in the attP region to which the Gifsy-1 protein binds. The results suggest that this recombination directionality factor binds in vitro to three imperfect direct repeats, spaced 10 base pairs apart, in a sequential and cooperative manner in the absence of other phage-encoded proteins. Our studies suggest that, while the Gifsy-1 Xis does not require additional factors for specific and high-affinity binding, it may form a microfilament on DNA similar to that described for the phage lambda Xis protein.

The Salmonella chromosome is known to host many temperate phages that are inducible under specific conditions (8). A number of the Salmonella phages carry genes that affect the pathogenicity of Salmonella enterica serovar Typhimurium. The Gifsy-1 phage is approximately 50 kb in length, resides at 57 centisomes in the Salmonella enterica serovar Typhimurium chromosome, and carries at least one gene, gipA, that contributes to Salmonella pathogenicity (9). The gene product of gipA contributes to the survival of Salmonella serovar Typhimurium in the Peyer's patches of the mouse intestine (31). The Gifsy-2 phage located at 24 centisomes of the Salmonella chromosome is also approximately 50 kb in size. The Gifsy-2 phage carries at least two important genes that contribute to the pathogenicity of Salmonella serovar Typhimurium, gtgE and sodc1 (8, 18). Genetic analysis also suggests that the Gifsy-1 and Gifsy-2 phages may exert regulatory control over each other through two mechanisms. First, the Gifsy-1 phage encodes a gene product that regulates the frequency of excision of both Gifsy-1 and Gifsy-2. Second, the Gifsy-2 phage may encode a gene product that participates in the integrase-mediated excision of Gifsy-1 (8).

Gifsy-1 resembles phage lambda in its gene organization and is classified as a lambdoid phage (8). A hallmark of this class of phages includes the presence of genes encoding a tyrosine recombinase or integrase (Int) and an excisionase (Xis) near an att site of one arm of the integrated phage. Tyrosine recombinases are large, complex, multifunctional proteins. These recombinases, including the Gifsy-1 Int, all possess a C terminus containing the conserved amino acid residues RKHRHY (5, 11, 12, 20, 35) and domains that bind DNA sequences at the site of catalysis and flanking sites. Typically, little homology exists between Int proteins outside of the C-terminal region (5, 7, 11, 35).

Little is known concerning the molecular mechanism of site-specific recombination as it occurs in the Gifsy-1 phage system. The integration of the circular phage chromosome into the host chromosome is thought to occur by a process similar to that which occurs in the phage lambda recombination system. The integration occurs by Int-mediated recombination between identical 14-base-pair core att sites in the phage chromosome (attP) and an identical sequence in the host chromosome (attB). This site-specific recombination requires the 409-amino-acid Gifsy-1 Int protein. Deletion analysis showed that the excision of the phage requires the Gifsy-1 Int and Xis proteins (10). Integration assays performed using strains lacking IHF indicate that integration of the Gifsy-1 phage requires IHF. In agreement with this observation, an IHF binding site was discovered 14 base pairs downstream of the core attP sequence (23, 39). Subsequent electrophoretic mobility shift assays (EMSA) and DNase I footprinting studies showed that IHF was capable of binding to this site in vitro.

In this report, we have begun the study of the Gifsy-1 phage site-specific recombination system with an analysis of excisive recombination and the characterization of the Gifsy-1 Xis protein. Gifsy-1 Xis is a small (94 amino acids in length), basic (pI, 10.7) recombination directionality factor. At first glance, the Gifsy-1 Xis protein itself possesses primary amino acid sequence homology to the predicted gene product of orf2C of the Bacteriodes conjugative transposon CTnDOT only in its predicted N-terminal helix-rich domain, as determined by Clustal analysis (3, 13-16, 19, 36, 37) (Fig. 1). The expression of the orf2C gene product is essential for the excision and regulation of transfer of CTnDOT, though its specific function in the excision and transfer mechanism is not yet known (24, 34). Thus, primary amino acid sequence analysis offers few clues to the molecular mechanism of Gifsy-1 Xis-mediated site-specific recombination.

FIG. 1.

Clustal alignment of Gifys-1 Xis with the predicted gene product of orf2C from conjugative transposon CTnDOT. An asterisk indicates a single, fully conserved residue, a double dot indicates conservation of strong groups, and a single dot indicates conservation of weak groups.

In this report, we analyze the Gifsy-1 Xis protein with its cognate DNA target sequence by EMSA and footprinting analyses. Specifically, we show that the Gifsy-1 Xis binds to three discrete sequences within the Gifsy-1 phage attP site. The binding of Gifsy-1 Xis may produce a nucleoprotein filament similar to the one formed by lambda Xis (2, 25). In addition we demonstrate step-wise binding of the Gifsy-1 Xis protein to Gifsy-1 attP DNA.

MATERIALS AND METHODS

Dideoxy sequencing.

A USB thermosequenase kit was used to sequence portions of the pNFB3 plasmid for the footprinting experiments. For the confirmation of vector constructs, a BigDye sequencing kit was used for fluorescence sequencing. The reactions were resolved at the University of Illinois in Urbana-Champaign Keck Biotechnology center.

Oligonucleotides and restriction endonucleases.

Restriction enzymes were obtained from Invitrogen. The plasmid pNFB3 was generously supplied by the Bossi laboratory, CNRS, Centre de Génétique Moléculaire. PCR products containing portions of the Gifsy-1 attP region were generated using the pNFB3 plasmid as a template. All primers were synthesized by Integrated DNA Technologies, Inc. (Table 1).

TABLE 1.

Single-stranded primers used for the PCR amplification of Gifsy-1 attP fragments from the pNFB3 vector

Primer	Sequence (5′ to 3′)
G1attP 350	ATGTGAGCAATGATGGAG
attP 350r	CTCCATCATTGCTCACAT
attP right150f	ACATCACAATAGAGGTCTATACG
attP right150r	CGGAGAAGCTCGATGGCCGC
attP left150f	GAAATGCGCACCCACAGACC
attP left150r	CTATGATGGACACAGAATCTTCC
attP ihfecoR1f	GCGCGCGCGCGAATTCGGATATGGACGCACTGGTGGAC
AttP Cla1f	ATAATTGATACCCCGCCACCC
AttP Cla1r	AACCATGACTCATGATGTCGTAGGTTGACTGTACATGCAACC
AttP45-85r	GGTTGCATGTACAGTCAACCTACGACATCATGAGTCATGGTT
G1attP 650sma1f	CGCGCGCCCCGGGTGATGAGAAATGCGCACCCACAGACC
G1attP 650sma1r	CGCGCGCCCCGGGCGGGAGAGCTCGATGGCCGC
AttP footprintexfa	GGTGGACACAAAAACCATGACTCATGATGTCGTAGGTTGACTGTACATGCAACCAGTATCATTTTTGTG
AttP footprintexr	CACAAAAATGATACTGGTTGCATGTACAGTCAACCTACGACATCATGAGTCATGGTTTTTGTGTCCACC

^aattP footprintexf and attP footprintexr are complementary oligonucleotides that were annealed together to generate the attP69 ligand.

Strains and cloning.

Salmonella serovar Typhimurium strain 14028s and derivative strains cured of the Gifsy-1, Gifsy-2, and Gifsy-3 phages were also generously supplied by the Bossi laboratory. PCR products containing the sequenced Gifsy-1 xis gene were cloned into the pET27b+ vector (Novagen). The primers were used to incorporate XbaI and BamHI restriction sequences into the ends of the xis gene to facilitate cloning into the expression vector.

The DNA encoding the Gifsy-1 Xis protein was amplified from the pSM13-1 vector using primers designed to add the XbaI and BamHI restriction enzymes’ recognition sites to the end of the xis gene. Using these restriction sites, the xis gene was cloned in frame to the start codon provided by the pET27b+ expression vector. This strategy results in the isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible expression of untagged Gifsy-1 Xis. The successful cloning was verified by DNA sequence analysis. The pET27b+ Xis vector was electroporated into Escherichia coli DH5α cells and stored as dimethyl sulfoxide stocks at −80°C.

Radiolabeling and purification of DNA ligands.

For the single-end-labeled DNA used in the footprinting reaction mixtures, DNA oligonucleotides were 5′-phosphorylated as described previously (17). Oligonucleotides were purified away from protein and ATP by using G-25 spin columns (GE Healthcare). The phosphorylated and purified oligonucleotides were then used directly in PCRs. The PCR products were then analyzed and purified in 5% polyacrylamide, 0.5× Tris-borate-EDTA (TBE) gels. DNA was isolated from the gels by using one of two methods. Gel slices containing DNA were minced using a sterile razor blade and allowed to incubate overnight in Elutip-D low-salt buffer (0.2 M NaCl, 20 mM Tris HCl, 1.0 mM EDTA, pH 7.4). After the elution, the DNA was concentrated and purified using an Elutip-D syringe column (Whatman). Alternatively, gel slices containing DNA were minced and loaded into a D-Tube electroelution cartridge (Novagen) and eluted at 120 V in 0.5× TBE buffer for 1 h. The DNA in solution was then diluted 1:5 in Elutip-D low-salt buffer and purified by using an Elutip-D column.

The double-stranded DNA ligands used in the EMSA reaction mixtures were prepared by dissolving complementary sets of single-stranded nucleotides in TE (10 mM Tris, 1 mM EDTA, pH 8) supplemented with 200 mM NaCl. The oligonucleotides were heated to 95°C for five minutes and allowed to cool to room temperature for 8 h. The annealed oligonucleotides were then concentrated by ethanol precipitation and purified in 1% agarose gels to remove unannealed single-stranded oligonucleotides. The DNA was eluted from gel slices using 5-kDa MWCO (molecular weight cutoff) D-Tube electroelution cartridges (Calbiochem/NovaBiochem/Novagen). The ligands were again concentrated and used in T4 polynucleotide kinase (Invitrogen) reaction mixtures as described for the reaction mixtures with single-stranded oligonucleotides.

l-[³⁵S]cysteine metabolic labeling.

An overnight culture of E. coli strain BL21(DE3) ihfA ihfB (32) containing the pET27b+xis plasmid was prepared. Five microliters of the overnight culture was subcultured into 100 ml of M9 medium (4) supplemented with 0.4% glucose and 0.3 mM l-cysteine. The culture was grown to an optical density at 600 nm of 0.6. The cells were centrifuged, washed, and aliquoted into two 50-ml cultures in M9 medium containing 0.4% glucose and lacking l-cysteine. Labeling was begun by removing 1 ml of culture and adding enough l-[³⁵S]cysteine (Perkin Elmer) to obtain 1 mCi/ml. Two hundred microliters was immediately withdrawn, lysed with sodium dodecyl sulfate (SDS) loading buffer, and frozen at −20°C for the first time point. The culture was allowed to incubate for 15 min, at which time IPTG was added to a concentration of 1 mM and another aliquot was withdrawn and processed as described above. The culture was allowed to incubate for another 15 min, at which time the antibiotic rifampin (200 μg/ml; Calbiochem) was added and another aliquot was withdrawn. After the addition of rifampin, three more time points were collected (45 min, 1 h, and overnight). Samples were boiled for 5 min and resolved by SDS-polyacrylamide gel electrophoresis (PAGE) (21). The gels were subjected to Coomassie staining and exposed to FUJI immunoprecipitation (IP) screens to visualize the protein expression and metabolic labeling of proteins.

Purification of Gifsy-1 Xis.

For overexpression, the pET27 vector was moved into a BL21(DE3) ihfA ihfB strain from DH5α. This step was repeated for each preparation of Gifsy-1 Xis-containing lysates. Overexpression was accomplished by growing cells to an optical density at 600 nm of 0.6 and adding IPTG to a final concentration of 0.1 mM with shaking at 37°C for 15 min before the addition of rifampin to a concentration of 200 μg/ml. The cultures were further incubated for 4 h and harvested by centrifugation for 15 min at 4°C. The cell pellets were then resuspended in 25 ml of HEPES purification buffer (50 mM NaCl, 50 mM HEPES, pH 7.0, 10 mM EDTA, pH 8.0, 5% glycerol, 1 mM dithiothreitol [DTT]). Phenylmethylsulfonyl fluoride (PMSF; Sigma) was added to a final concentration of 2.5 mM, and three tablets of Complete Mini EDTA-free protease inhibitor cocktail (Roche) were added. The cell suspension was then frozen at −80°C and subsequently thawed on ice. After the thawing, lysates were generated by sonication and were then centrifuged at 43,000 × g for 35 min at 4°C. An additional protease inhibitor tablet was added to the lysate. Poly(ethyleneimine) (PEI; Sigma) precipitation was used to precipitate DNA and acidic protein from the lysate (6). The amount of PEI (10% PEI, pH 7.9) added in microliters to the lysate was determined by dividing the total volume of lysate by 35. PEI was added drop-wise to the ice-bath-chilled lysate and gently stirred. The precipitate was then removed by centrifugation at 30,500 × g for 15 min at 4°C. The supernatant from the PEI precipitation was buffer exchanged into 25 ml of fresh purification buffer containing PMSF by using Amicon ultracentrifugation devices with a molecular-mass cutoff of 5 kDa. Three washes of 15 ml of buffer were used before the final volume was restored to 25 ml with purification buffer.

Twenty-five milliliters of Xis-enriched lysate was loaded into a fast protein liquid chromatography system consisting of two sequential columns. The first column consisted of two 5-ml Sepharose Q columns (GE Healthcare). The second column was placed directly in line with the first and consisted of two 5-ml Sepharose SP columns (GE Healthcare). Desalted PEI supernatant was loaded onto the column at a rate of 1.5 ml/minute. After the lysate was loaded, the columns were washed with purification buffer until the 280-nm UV monitor yielded a steady baseline. The Sepharose Q columns were then disconnected from the system. The protein was eluted from the Sepharose SP columns in HEPES buffer using a gradient of 50 mM to 500 mM NaCl over a volume of 145 ml at a flow rate of 2 ml/min. The fractions for which peaks were observed were collected and assayed for protein content on SDS-PAGE 10 to 20% gradient tricine gels. Fractions containing protein with an estimated mass of ∼11.2 kDa were pooled into aliquots of 5 fractions each and analyzed again by SDS-PAGE. The pooled aliquots were subsequently assayed by EMSA for their ability to shift attP DNA. Fractions containing the putative Xis protein were pooled, buffer exchanged into tricine purification buffer (50 mM tricine, pH 8.4, 10 mM EDTA, pH 8.0, 5% glycerol, 1 mM DTT), concentrated, and resolved on a Mono S fast protein liquid chromatography column (GE Healthcare) using an NaCl gradient of 20 to 500 mM. The peak fractions were collected and analyzed by SDS-PAGE. Putative Xis-containing aliquots were pooled together and concentrated to a volume of ∼2.5 ml in HEPES purification buffer. The enrichment of samples for the Gifsy-1 Xis protein was followed by SDS-PAGE gel analysis. All samples were resolved on 10 to 20% acrylamide tricine gradient gels.

EMSA.

Dilutions of Gifsy-1 Xis were assayed for binding by titration against DNA ligands in EMSA binding buffer (6.7 mM Tris, pH 7.0, 3.3 mM sodium acetate, pH 7.0, 1 mM EDTA, 65 mM NaCl, 150 mM KCl, 300 μg/ml bovine serum albumin, 12% glycerol, 1 mM DTT, 0.1 mg/ml sonicated herring sperm DNA [Promega]) containing labeled DNA at a concentration of 0.5 nM. The reaction mixtures were allowed to incubate for 30 min at 25°C and were loaded into polyacrylamide gels (5% acrylamide, 0.5× TBE). The gels were resolved using a constant current of 10 mA at 4°C to prevent overheating (the gels were maintained at 25°C). The reaction mixture volumes were ∼one-third of the total well volume. The results were visualized by drying acrylamide gels onto DE81 ion exchange filter paper (Whatman) followed by exposure to FUJI IP screens. The FLA-3000 FUJI apparatus and software were used to analyze the gels.

Estimation of dissociation constants by EMSA.

Estimations of the dissociation constants describing the affinities of Gifsy-1 Xis for attP69 (a 69-base-pair attP DNA ligand) and attP left 306 bp were determined by titrating serial dilutions of Gifsy-1 Xis (dilution factor, 1.5) against 50 pM of each DNA ligand. The DNA binding reactions were resolved on 5% nondenaturing polyacrylamide gels, and the K_d values were estimated as the concentrations of Gifsy-1 Xis at which 50% of each DNA ligand migrated as free DNA.

DNase I protection assays.

Ten-microliter dilutions of purified Gifsy-1 Xis protein were assayed for binding to 0.5 nM DNA ligand in 100 μl footprinting buffer (25 mM HEPES, 50 mM potassium acetate, 5 mM magnesium acetate, 2 mM CaCl₂, 1 mM DTT, pH 7.0). The reaction mixtures were allowed to incubate at room temperature for 30 min. Q DNase I (1.5 μl; Promega) was added, and the total reaction mixture was allowed to incubate for 2 min before the reaction was terminated with the addition of 90 μl of 30 mM EDTA (26). The reaction mixture was deproteinized with buffered phenol chloroform (Sigma), precipitated with 95% ethanol, and washed with 70% ethanol. After being dried, the reaction mixtures were dissolved in 12 μl formamide loading buffer (22). Just prior to being resolved in an 8% denaturing sequencing gel (National Diagnostics), the samples were heated to 90°C for 3 min and immediately cooled in an ice bath. The samples were normalized for radioactivity by scintillation counting to ensure that equal amounts of DNA were loaded into each lane. The sequencing gels were run at a constant current of 85 W.

DMS protection assays.

The binding reactions were carried out in the same manner as the DNase I assays. Immediately after the reaction mixtures were incubated at room temperature, 1 μl of dimethyl sulfate (DMS) was added to the reaction mixtures and gently mixed. The reaction mixtures were allowed to incubate at room temperature for 1.5 min before they were terminated by the addition of 10 μl of a 250 mM DTT solution. The DNA was then phenol extracted, precipitated, and washed. Samples were then dissolved in 10 μl water and placed on ice. One hundred fifty microliters of 1 M piperidine was added. Preferential cleavage of guanine residues was accomplished by heating the samples at 90°C for 30 min. The DNA was concentrated using isobutanol and washed with 1% SDS to remove excess piperidine. The DNA pellets were dried, and the samples were resuspended in formamide loading buffer.

DMS interference assays.

An amount of DNA sufficient for three 400-μl EMSA reactions was diluted with TE to a volume of 100 μl. The DNA was methylated by the addition of 1 μl DMS for 1.5 min. The methylation reactions were stopped by the addition of 25 μl 1 M β-mercaptoethanol. The DNA was then ethanol precipitated and dried. The DNA was resuspended in 1,200 μl of an EMSA reaction mixture and allowed to incubate at room temperature for 45 min. The reaction mixture was divided into 400-μl aliquots and resolved as for a standard EMSA. The protein-DNA complexes were visualized using FUJI IP screens. Protein-DNA complexes were excised from the gel and the DNA was purified using an Elutip-D column (Whatman) and ethanol precipitated. The DNA was resuspended in water and cleaved using hot piperidine as described for the DMS protection assays. The cleavage patterns were visualized by resolution of samples on an 8% urea denaturing sequencing gel and exposure to FUJI IP screens. For quantitative DMS interference analysis, the protocol was altered as follows: methylated DNA ligands singly labeled on either the top or the bottom strand were incubated with Gifsy-1 Xis protein in standard EMSA reaction mixtures and resolved in 8% acrylamide, 0.5× TBE native gels. An aliquot of DNA from the initial DMS methylation reaction mixture was set aside to use as a DNA-only control sample. The final resolution of the piperidine-digested DNA was performed on 8, 12, or 20% acrylamide urea sequencing gels. Visual assessment of the Gifsy-1 Xis binding to methylated DNA was corroborated by quantitative analysis using a FUJI phosphorimage apparatus. Quantitative analysis of the DMS interference assays was accomplished by using FUJI ImageGauge V3.0 software. In brief, the amount of radioactive probe present in a band representing cleavage at a specific residue was counted. The signal in each lane was normalized according to the rate of cleavage of the internal standard residue, and the background value, determined by automated graphical analysis, was subtracted from each value. For each residue analyzed, an equivalent residue in the methylated-DNA-only lane was used as a standard. The frequency of cleavage at a specific residue was then represented as the ratio of the signal contained by the sample band divided by the signal of the standard band and multiplied by 100.

RESULTS

Expression and purification of Gifsy-1 Xis.

The Gifsy-1 xis gene was cloned and expressed as described in Materials and Methods. The expression of Gifsy-1 Xis protein was either masked by other proteins or so weak it was not detected on Coomassie-stained gels. In order to identify and monitor Gifsy-1 Xis expression, we radiolabeled proteins with [³⁵S] cysteine to detect the expression of Gifsy-1 Xis protein in cells induced with IPTG. After a short period of induction with IPTG to allow the expression of the T7 RNA polymerase, the antibiotic rifampin was added to the cultures to limit transcription to genes under the control of the T7 promoter. Under these conditions, the increased expression over time of an approximately 11-kDa protein was observed in the lysates from induced and radiolabeled cells. Autoradiography of gels revealed that the 11-kDa protein showed enhanced labeling in comparison to that of other proteins the size of Gifsy-1 Xis (Fig. 2). Using methods further outlined in Materials and Methods, the protein was purified to homogeneity at a final concentration of 1.5 mg/ml (13.4 mM) (Fig. 3A).

FIG. 2.

Autoradiography of lysates of l-[³⁵S]cysteine-labeled E. coli BL21 ihfA ihfB cells harboring the pET27b+ plasmid with Gifsy-1 xis cloned in frame to the Ptac promoter. The lanes of the 10 to 20% gradient tricine SDS-PAGE gel contain lysates of cells at an optical density at 600 nm of 0.6 (lanes 1), incubated in the absence of IPTG (A) or with 1 mM IPTG (B) for 15 min (lanes 2), incubated for 15 min after addition of rifampin (lanes 3), after incubation was continued for 1 h (lanes 4), and after incubation was continued overnight (lanes 5).

FIG. 3.

(A) Purification of Gifsy-1 Xis. First lane, molecular weight markers; second lane, crude extract; third lane, cleared lysate; fourth lane, PEI-precipitated extract; fifth lane, Sepharose SP XL-purified Xis; sixth lane, Mono S fractions. (B) Schematic of DNA ligands. The lengths in base pairs of the attP region and subregions are given beneath the diagram. (C) Partially purified (post-Sepharose SP XL column) Gifsy-1 Xis bound to radiolabeled PCR products (final concentration, 0.5 nM) amplified from the Gifsy-1 attP region. The putO ligand is a PCR-amplified portion of the unrelated put operator region used as a negative control. (D) Mono S column-purified Gifsy-1 Xis titrated against attP160. The specific-DNA-ligand final concentration is 0.5 nM for all reactions. Lane 1, DNA only; lanes 2 to 11, indicated dilutions of Gifsy-1 Xis.

Identification of the attP region required for Gifsy-1 Xis binding by EMSA.

To examine the Xis-DNA interactions, EMSA and footprinting assays were performed using partially purified Gifsy-1 Xis and several truncated attP DNA ligands (Fig. 3B and C). Fractions eluted from the Sepharose SP XL column containing the Gifsy-1 Xis protein were incubated with the PCR products of several regions of DNA surrounding the Gifsy-1 attP core sequence. The fractions containing Gifsy-1 Xis expressed specific binding activity to the attP left 306 bp DNA (Fig. 3C). The specific binding activity was further isolated to DNA ligand attP160 (a 160-base-pair sequence of Gifsy-1 attP DNA, here within a 171-base-pair sequence with a flanking primer-encoded BamHI restriction site) (Fig. 3D). Gifsy-1 Xis did not bind to the attP 150-bp left end. This indicated that the binding sequence of the Gifsy-1 Xis is contained within attP160, and not the additional sequences contained in the attP306L ligand (Fig. 3C). Gel-shift assays revealed a complex binding pattern for the Gifsy-1 Xis protein when challenged with attP160. At least three major complexes containing Gifsy-1 Xis and attP DNA are detected by EMSA with 5% polyacrylamide gels (Fig. 3D).

DNase I footprint analysis of Gifsy-1-attP interactions.

To further dissect the interaction of the Xis protein with its specific DNA ligand, nuclease protection assays were performed. Double-stranded DNA was labeled with ³²P on either the top or the bottom strand of attP160, and the footprints were analyzed by phosphorimage analysis (Fig. 4). We observed the smallest protected region, a large, ∼31-base-pair sequence, at a concentration of 400 nM Gifsy-1 Xis when the bottom strand of the attP160 ligand was used in DNase I protection assays with Gifsy-1 Xis (Fig. 4B). This specific protected region spanned residues 47 to 78. Several regions of enhanced cleavage bracketed the central, protected 31 base pairs at both the 5′ and the 3′ ends of the footprint. These enhanced DNase I cutting sites occurred at residues 30, 33, 37, and 39. These residues are 5′ to A/T tracts that occur at positions 41 to 45 and 93 to 97. DNase I footprint analysis also showed further evidence of DNA distortions mediated by Gifsy-1 Xis-DNA interactions, observed in the protection from and then hypersensitivity to DNase I digestion of 2 base-pairs, 90 and 91, 12 residues 3′ of the large footprint. It is possible that the reproducible protection shown for these two bases at a lower concentration of Xis protein might result from distortions in the DNA double helix induced by protein-DNA interactions at nearby sites instead of direct protection caused by Xis-DNA interactions. As the concentration of Gifsy-1 Xis was increased, these bases became hypersensitive to DNase I digestion. A smaller footprint was observed when the top strand of DNA was labeled and subjected to DNase I protection analysis. This footprint spanned bases 57 to 73 (Fig. 4A). This 16-base-pair sequence is divided roughly in half by 3 unprotected bases at residues 64 to 66. To identify additional sequences with which Gifsy-1 Xis might interact, sequences 5′ to the attP160 ligand were also analyzed by a DNase I protection assay. Because of the small size of the footprint observed on the top strand of the attP160 ligand, a longer strand of DNA extending beyond the 5′-end attP160 ligand was used. Additional footprints were not identified by DNase I protection analysis of longer DNA ligands (data not shown). Furthermore, EMSA analysis using attP fragments suggested that additional binding sites are not present beyond the footprinted region on the attP160 DNA ligand (Fig. 3B).

FIG. 4.

(A) DNase I protection assay of attP160 5′ labeled on the top strand. Residues hypersensitive to DNase I digestion are indicated by arrowheads. Lanes 1 and 2, indicated dilutions of Gifsy-1 Xis incubated with DNA ligand and digested with DNase I; lane 3, DNA-only control digested with DNase I; lane 4, G+A sequencing ladder. The protected sequence lies between the residues indicated to the left of the gel. (B) DNase I footprint of attP160 5′ labeled on the bottom strand. Lane 1, G+A sequencing ladder; lane 2, DNA only; lane 3, DNA ligand plus DNase I; lanes 4 to 10, indicated dilutions of Gifsy-1 Xis, DNA ligand, and DNase I; lane 11, G+A ladder. Samples were resolved on an 8% sequencing gel. The protected bases are indicated to the left and right of the gel. Bases next to a vertical line correspond to the DNase I footprint of the top strand. (C) Sequence of the attP160 ligand. Sequences corresponding to the Gifsy-1 Xis footprint are underlined. The Gifsy-1 Xis attP core sequence is boxed. Arrowheads indicate bases hypersensitive to DNase I digestion in the presence of Gifsy-1 Xis protein. Horizontal arrows indicate the sequence limits of the attP69 DNA ligand used in later experiments.

attP69 (Fig. 4C) was used for further EMSA analysis. It includes residues that were hypersensitive to DNase I digestion in the presence of Gifsy-1 Xis. Gifsy-1 Xis bound to the attP69 ligand with approximately the same affinity as that with which it bound to the attP306L ligand. The K_d for the association of purified Gifsy-1 Xis with attP69 ligand was estimated to be approximately 42 nM (Fig. 5A). The K_d for specific binding to the attP306L ligand was estimated to be approximately 24 nM of Gifsy-1 Xis (Fig. 5B). These data indicated that the Gifsy-1 Xis bound to the longer attP ligand with approximately two-times-greater affinity. In addition, the multiple species observed in EMSA when the attP69 ligand was used, though present, were less apparent when the attP306L ligand was used (Fig. 5A and B). This result suggests that binding to the attP69 fragment occurs sequentially, with reduced cooperativity compared to the binding observed with the attP306L fragment. This result is similar to the EMSA results observed for the lambda Xis protein with short and long DNA fragments containing the X1, X1.5, and X2 sites (2).

FIG. 5.

(A) Titration of Gifsy-1 Xis against 5′-end-labeled attP69 (50 pM). Lane 1, no protein; lanes 2 to 17, indicated dilutions of purified Gifsy-1 Xis were incubated with labeled DNA ligand. (B) Titration of Gifsy-1 Xis against attP306L. Lanes 1 to 17 contain 50 pM attP306L. Lane 1, DNA only; lanes 2 to 17, indicated dilutions of Gifsy-1 Xis. The amount of labeled probe in each band was quantified by phosphorimage analysis.

DMS footprinting of Gifsy-1 Xis-attP interactions.

Footprinting of Gifsy-1 Xis interactions with attP DNA were carried out using DMS methylation assays to identify specific bases at which close protein-DNA interactions might occur. DMS methylates double-stranded DNA at N-7 at guanine residues in the major groove and methylates adenine weakly at N-3 in the minor groove (22). Two types of methylation analysis were performed: DMS protection and DMS interference. During DMS protection assays, Gifsy-1 Xis is allowed to incubate with DNA ligands. DMS is added after the reaction mixtures are allowed to equilibrate. In this process, methylation occurs at guanine and adenine residues over the entire length of the DNA ligand, with the exception of those bases that are specifically protected by interaction with Gifsy-1 Xis.

DNA fragments that were 5′ end-labeled on either the top or the bottom strand were analyzed in DMS protection assays. Methylation protection analysis of the top-strand DNA after exposure to DMS in the presence of Gifsy-1 Xis revealed that 5 residues are protected by specific interactions: residues A71, G70, A65, G60, and G50 (Fig. 6A). All but the G50 residue occur within the Gifsy-1 Xis footprint outlined by the DNase I protection assays. Four guanine residues on the bottom strand within the DNase I-footprinted sequences were protected from methylation when Gifsy-1 Xis was incubated with labeled DNA (Fig. 6B). These residues include G52, G54, G62, and G72. All residues revealed by DMS protection on the bottom strand correlated to the DNase I footprint on the bottom strand. Protection of adenine bases was not observed on the bottom strand.

FIG. 6.

(A) DMS protection assay of attP160 5′ labeled on the top strand and resolved on an 8% polyacrylamide denaturing sequencing gel. Lane 1, G+A sequencing ladder; lane 2, DNA-only control; lanes 3 and 4, indicated dilutions of Gifsy-1 Xis. (B) DMS protection assay of attP160 5′ labeled on the bottom strand. Lane 1, G+A sequencing ladder; lane 2, DNA-only control; lanes 3 and 4, indicated dilutions of Gifsy-1 Xis. Residues protected from methylation by DMS are indicated on the DNA sequences to the right of the gel images.

To gain more information on Xis-DNA interactions for the formation of Gifsy-1-attP complexes, DMS methylation interference assays were performed using the attP69 ligands. The DMS interference assays were performed by methylating DNA before incubating it with Gifsy-1 Xis in EMSA reaction mixtures. The Gifsy-1 Xis-DNA complexes formed in solution were then resolved by electrophoresis on nondenaturing gels. The Gifsy-1 Xis-attP complexes were eluted and cleaved using piperidine. The DNA species methylated at residues essential for Gifsy-1 Xis-attP interactions migrated as low-molecular-weight unbound DNA, while the DNA methylated at nonessential residues migrated in gel-shifted complexes with Gifsy-1 Xis. The residues with which Gifsy-1 Xis specifically interacted in each complex were discerned by the absence of cleavage at specific residues visualized in a denaturing gel compared to a methylated and cleaved residue at the equivalent position in the control lane (Fig. 7A and B). Thus, the nucleotides that participated in the formation of specific Gifsy-1 Xis-DNA complexes were identified through comparison of the cleavage pattern of DNA isolated from each complex with the cleavage pattern produced from freely migrating control DNA.

FIG. 7.

DMS interference assay of Gifsy-1 Xis bound to attP69. (A [top strand] and B [bottom strand]) Piperidine-cleaved methylated DNA eluted from corresponding bands from preparative EMSA. The sequence of the 69-base-pair ligand is aligned next to the gel. A vertical line indicates the Gifsy-1 Xis footprint identified by DNase I protection analysis of the attP160 ligand. Residues whose methylation interfered with Gifsy-1 Xis binding are indicated by diagonal lines. Imaged results of preparative EMSA of Gifsy-1 Xis bound to methylated attP69 ligand are included next to sequencing gels. 1, high-molecular-weight species; 2, intermediate-weight species; 3, low-weight species; 4, free DNA. (C) Quantitative analysis of cleavage at specific residues from each lane of top-strand DNA (first graph) and bottom-strand DNA (second graph). Frequency of cleavage was analyzed as described in Materials and Methods. Error bars were generated by repeating assays on top (n = 5) and bottom strands (n = 2) and calculating a standard deviation for the mean frequency of cleavage at each residue analyzed.

EMSA analysis of Gifsy-1 Xis incubated with specific DNA ligands indicated the occurrence of three species of protein-DNA complexes in three discrete gel-shift bands, a high-molecular-weight, fully occupied species, an intermediate-molecular-weight species, and a low-molecular-weight species (Fig. 3D and 5A). It was expected that the three different protein-DNA complexes isolated from EMSA should involve different specific contacts between Gifsy-1 Xis and DNA ligands. Analysis of the labeled and methylated attP69 top strand showed methylation interference at 3 residues, G70, G60, and G50 (Fig. 7A and C). Methylation at G60 resulted in a dramatic decrease in the formation of all complexes (bands 1, 2, and 3). Methylation at G70 severely interfered with the formation of bands 1 and 2 but allowed the formation of the lowest-molecular-weight complex present in band 3. Only the formation of the Gifsy-1 Xis-attP69 band 1 complex was significantly affected by methylation at residue G50. Interference assays using attP69 labeled on the bottom strand showed similar results (Fig. 7B and C). Methylation at G62 interfered strongly with the formation of all complexes, while methylation at G52 had little effect on the formation of Gifsy-1 Xis-attP69 complexes. The formation of the complexes in bands 2 and 3 was tolerant of methylation at G72, though methylation at this site severely inhibited the formation of band 1 complexes.

Examination of the Gifsy-1 attP DNA sequence identified three imperfect direct repeats (DRs) that span 9 base pairs and are spaced 1 base pair apart. DR1 contains residues 47 to 55, DR2 contains residues 57 to 65, and DR3 contains residues 67 to 75 (Fig. 8). This repeat pattern possessed the consensus sequence taaGtCAtc (perfectly conserved residues are capitalized).

FIG. 8.

Diagram of the attP160 DNA ligand. Pattern of Gifsy-1 Xis-DNA contacts as suggested by several footprinting techniques. Horizontal lines indicate sequences protected by Gifsy-1 in DNase I protection assays. Vertical arrowheads indicate sites rendered hypersensitive to DNase I digestion by Gifsy-1 Xis binding. Single dots indicate residues protected by Gifsy-1 Xis binding from DMS methylation in DMS protection assays. Double dots indicate residues whose methylation interfered with Gifsy-1 Xis-DNA binding and which were protected in DMS protection assays. Sequences in brackets delineate DRs (DR1, DR2, DR3). Horizontal arrows indicate the limits of the 5′-end-labeled oligonucleotides used in footprinting experiments. Boxed sequences indicate the 14-base-pair attP core sequence required for the site-specific integration of the Gifsy-1 phage into the Salmonella serovar Typhimurium chromosome.

DISCUSSION

We identified and characterized the DNA sequence necessary for efficient and specific Gifsy-1 Xis interactions with its attP site by using EMSA and footprinting techniques. EMSA analysis of the Gifsy-1 Xis protein bound to several attP DNA ligands revealed a complex binding pattern. Multiple Gifsy-1 Xis-DNA species are detected in EMSA with both the attP306L ligand and the attP69 ligand, though these species are more apparent when the shorter ligands are used (Fig. 5). Detailed analysis revealed that Gifsy-1 Xis bound cooperatively to a 306-base-pair sequence in the attP region with high affinity (K_d, ∼24 nM). This highly cooperative binding occurred in contrast to the less cooperative manner in which multiple subunits of Gifsy-1 Xis bound to shorter 160- and 69-base-pair concentric fragments with a lower affinity (K_d, ∼42 nM for attP69). The analysis of several DNA fragments from the attP site through EMSA did not identify additional sequences outside of the 69-base-pair sequence to which the Gifsy-1 Xis specifically binds. Additional nonspecific interactions between Gifsy-1 Xis and DNA lying outside of the 69-base-pair sequence may account for the increased affinity for the longer, 306-base-pair ligand. The affinity of Gifsy-1 Xis for ligands containing attP DNA in the presence of nonspecific DNA competitor for both the long and small DNA ligands is indicative of specific DNA binding activity.

The several footprinting techniques employed suggest that the specific binding site of Gifsy-1 Xis is centered about a 31-base-pair sequence spanning residues 47 to 78, identified by DNase I protection assays (Fig. 8). The large area of protection identified by the DNase I footprint, in agreement with EMSA data, suggests that several Xis subunits may interact specifically with identified regions of attP DNA or may possibly bend or loop the DNA as has been observed for the IHF protein (28, 29), the Tn916 Xis protein (1, 30), and the lambda Xis protein (2, 25, 38).

The potential sites of interaction of the Gifsy-1 attP region identified outside of the large central DNase I footprint of Gifsy-1 Xis include not only the small dinucleotide sequence (attP160 residues 90 and 91) but also one major sequence possessing enhanced sensitivity to DNase I cleavage (attP160 residues 30, 33, 37, and 39). Both sequences reside on either side of the central DNase I footprint of Gifsy-1 Xis. Gel-shift analysis of a 69-base-pair sequence (attP69) that included these additional sequences reproduced the EMSA binding pattern with an estimated K_d of approximately 42 nM. These congruent data suggest that the sequences included in the 69-base-pair ligand are required and sufficient for efficient Gifsy-1 Xis binding.

Two independent methods, DMS protection and DMS interference, confirm Gifsy-1 Xis-DNA interactions with 3 guanine bases in close proximity on the top (G50, G60, and G70) and the bottom (G52, G62, and G72) strands of attP DNA. The analysis of labeled top-strand and bottom-strand assays showed a similar pattern of protein-DNA interactions occurring in three imperfect DRs, DR1, DR2, and DR3, 9 base pairs in length and spaced 1 base pair apart. Only methylation at residues 62 (bottom strand) and 60 (top strand) strongly interfered with the formation of all species of Gifsy-1 Xis-DNA complexes, indicating that this set of residues is essential for the formation of all Gifsy-1 Xis-DNA complexes. Methylation of residues G70 (top strand) and G72 (bottom strand) disrupted the formation of high- and intermediate-weight complexes but allowed the formation of the low-molecular-weight complex, indicating that protein-DNA interactions at this pair of residues are essential for the formation of higher order species (Fig. 9). Residues G52 (bottom strand) and G50 (top strand) appear to contribute to the formation of the high-molecular-weight species, as methylation at G52 and G50 only disrupted the formation of this complex but did not severely affect the formation of low- and intermediate-weight complexes. Each set of residues, G50-G52, G60-G62, and G70-G72, is spaced 10 bases apart, and they appear to neatly outline three imperfect DRs (DR1, DR2, and DR3) (Fig. 8). While nucleic acid residue G50 on the top strand was reproducibly protected in the DMS protection assays and shown to participate in direct interactions with Gifys-1 Xis in the DMS interference assays, this residue was not protected in DNase I footprinting studies. The top-strand G50 residue was found to provide a minor contribution to the formation of higher order Gifsy-1-attP DNA complexes. It is possible that weak interactions at this site are difficult to detect by a DNase I protection assay.

FIG. 9.

Composition of Gifsy-1 Xis-attP69 complexes as implied by quantitative DMS interference analysis. Circles indicate occupancy of residues by Gifsy-1 Xis. Binding sites are indicated by vertical lines with residue numbers appearing at the bottom of the diagram. “CH3” indicates a methylated binding site. Band numbers corresponding to DMS interference assay complexes are indicated by the number to the left of each model.

Though it is likely that the interactions reported here represent only a subset of the actual Gifsy-1 Xis-attP interactions, our results provide strong evidence that specific Gifsy-1 Xis-attP interactions result in the formation of three discrete complexes and that the 6 residues residing within the 31-base-pair footprint are important for the formation of specific Gifsy-1 Xis-attP complexes. Our analysis of the DNA sequence to which Gifsy-1 Xis binds highlights similarities to and differences from the well-studied lambda Xis protein. The ordered sequence of binding events is similar to the mode of protein-DNA interactions recently used to describe the binding of the lambda Xis protein to X1-X1.5-X2 sites. Three subunits of lambda Xis bind to the X1-X1.5-X2 sites when they are nested within a 263-base-pair sequence and bind in a stepwise manner when the same sites are within a 50-base-pair sequence (2, 27, 33). It was demonstrated recently that lambda Xis binds nonspecifically to DNA with a low affinity and that the specific binding of three monomers of lambda Xis to the X1-X1.5-X2 sites occurs in a cooperative manner (2).

Likewise, at least three Gifsy-1 Xis molecules bind to the attP region. As suggested by in vitro analysis, the three Gifsy-1 Xis subunits may bind to DR1, DR2, and DR3 in the Gifsy-1 attP sequence to form a “micronucleoprotein filament” similar to that formed by the lambda Xis protein (2). However, the Gifsy-1 Xis binds to DNA containing DR1, DR2, and DR3 with a high affinity and without other known protein factors. DMS analyses show that the binding of Gifsy-1 Xis to DR2 may coordinate the cooperative binding of Gifsy-1 Xis to the complete DR1-DR2-DR3 site. The Gifsy-1 Xis also binds to attP DNA with a much higher affinity than is observed for the lambda Xis protein. It is interesting to note that the Gifsy-1 Xis possessed a higher affinity for the longer attP306L DNA ligand that included the central footprinted sequence. The reason for this differential binding is unknown as yet.

Our studies detail the binding of Gifsy-1 Xis to the attP sequence in vitro in the absence of Gifsy-1 Int. Additional studies are necessary to determine how the binding of Gifsy-1 Int and IHF may promote protein-protein interactions with the Gifsy-1 Xis protein.