Xis Protein Binding to the Left Arm Stimulates Excision of Conjugative Transposon Tn916

Tn916 and related conjugative transposons are clinically significantvectors for the transfer of antibiotic resistance among humanpathogens, and they excise from their donor organisms usingthe transposon-encoded integrase (^Tn916Int) and excisionase(^Tn916Xis) proteins. In this study, we have investigated therole of the ^Tn916Xis protein in stimulating excisive recombination.The functional relevance of ^Tn916Xis binding sites on the armsof the transposon has been assessed in vivo using a transposonexcision assay. Our results indicate that in Escherichia colithe stimulatory effect of the ^Tn916Xis protein is mediated bysequence-specific binding to either of its two binding siteson the left arm of the transposon. These sites lie in betweenthe core and arm sites recognized by ^Tn916Int, suggesting thatthe ^Tn916Xis protein enhances excision in a manner similar tothe excisionase protein of bacteriophage ${lambda}$ , serving an architecturalrole in the stabilization of protein-nucleic acid structuresrequired for strand synapsis. However, our finding that excisionin E. coli is significantly enhanced by the host factor HU,but does not depend on the integration host factor or the factorfor inversion stimulation, defines clear mechanistic differencesbetween Tn916 and bacteriophage ${lambda}$ recombination.

INTRODUCTION

Top

Introduction

Conjugative transposons are extremely promiscuous genetic elementsthat disseminate antibiotic resistance among gram-positive andgram-negative bacteria (reviewed in references 14, 49, 52, and54). Most conjugative transposons confer tetracycline-minocyclineresistance, while resistance markers for macrolides-lincosamides-stretogramins,aminoglycosides, beta-lactams, chloramphenicol, erythromycin,mercuric chloride, and nisin have been found on larger elements(49). Their ability to move between unrelated bacterial cellsthrough conjugation and their prevalence in drug-resistant pathogensthat cause human disease (for example, cholera [61], pneumonia[4], endocarditis [47], and nosocomial [9] and suppurative [12]infections) make these elements a serious health threat. Oneof the most extensively studied conjugative transposons is Tn916,which is 18 kb in length and encodes resistance to tetracycline(11, 18). Excision of the element from its bacterial host occursthrough staggered cleavages at the ends of the transposon, producingvariable (5 to 6 bp in length) 5' overhangs (8, 34). The overhangs,which are derived from heterologous bacterial DNA flanking thetransposon, are then sealed to produce a circular transposonintermediate with a novel 5- to 6-bp heteroduplex (8, 35, 53).The circularization of Tn916 enables the expression of transfergenes (11) and its ultimate single-strand transfer to a recipientbacterial cell (53) via a functional origin of transfer (25).Tn916 integrates into the recipient chromosome with limitedtarget preference (33, 53) and does not duplicate the targetsequence (13).

Two transposon-encoded proteins, integrase (^Tn916Int) and excisionase(^Tn916Xis), mediate transposition. The ^Tn916Int protein is aheterobivalent DNA-binding recombinase (32). Its C-terminaldomain is related, based on primary sequence homology, to thetyrosine family of recombinases (reviewed in references 3 and17), and it performs the strand cleavage and joining reactionsthrough a phosphotyrosine intermediate. Footprinting studiesof the N- and C-terminal domains of ^Tn916Int fused to the maltosebinding protein have revealed their binding sites on the transposon(32). The C-terminal domain of ^Tn916Int interacts with core-typesites at the ends of the transposon (Fig. 1), while the N-terminaldomain binds to an unrelated set of distal sites (called DR2sites) through an unusual three-stranded beta-sheet-DNA interface(15, 62). The ^Tn916Xis protein enhances excisive recombination(37, 38, 48, 50, 56) and binds to sites that are proximal tothe DR2 sites on each arm of the transposon (51).

View larger version (12K):
[in this window]
[in a new window]
FIG. 1. Comparison of the protein binding sites on transposon Tn916 and bacteriophage ${lambda}$ . The protein binding sites on each element are shown and drawn to the same scale, which is indicated in base pairs. The attachment sites (attL and attR) are labeled. The genomic DNA is shown as a heavy line, and the transposon or phage DNA is shown as a thin line. Core sites (bound by the Int catalytic domain) are shown as diamonds and labeled according to the method of Scott and Churchward (54). P and DR2 sites (bound by the Int N terminus) are shown as arrows, with their relative orientations indicated. IHF, Xis, and FIS binding sites are shown as squares, hexagons, and a pentagon, respectively. The newly identified X2 site on the Tn916 attL is shaded gray.

The general mechanism of Tn916 transposition is presumably relatedto the site-specific reactions that excise and integrate bacteriophage ${lambda}$ (54), since each encodes functionally homologous Xis and Intproteins that bind to the arms of each element (Fig. 1). Inbacteriophage ${lambda}$ , its integrase (Int) and excisionase (Xis) proteins,as well as the host-encoded proteins integration host factor(IHF) and the factor for inversion stimulation (FIS), cooperativelyassemble onto the phage to control the directionality of recombination.Xis plays a key architectural role in recombination, dramaticallybending DNA (57) and cooperatively recruiting the Int and FISproteins to the phage arm (59). However, the relative importanceof host-encoded factors and specific protein-DNA and protein-proteininteractions in conjugative transposition remains unknown andcannot be inferred from studies of bacteriophage ${lambda}$ . This is becausethe Int and Xis proteins of each element share no significantprimary sequence homology outside of the integrase catalyticactive site, and because there are fundamental differences inthe arrangement of their binding sites on the phage and transposon(Fig. 1). In this study we have determined the functional significanceof specific ^Tn916Xis-transposon interactions and host-encodedprotein factors in the excisive recombination of the Tn916 transposon.

MATERIALS AND METHODS

Top

Materials and Methods

Cell strains, plasmids, and reagents. The plasmids pUC18::Tn1545- ${Delta}$ 4 (48) and pAM120 (21) were giftsfrom Gordon Churchward. pAM120 contains the entire Tn916 transposoncloned into pGL101. pUC18::Tn1545- ${Delta}$ 4 contains the left and rightarms of Tn1545 and the aphA-3 gene for kanamycin resistance,cloned into pUC18. Escherichia coli DH5 was purchased from LifeTechnologies (Grand Island, N.Y.). E. coli strains RJ1796 (MG1655lac ${Delta}$ X74), RJ1976 (RJ1796 hupA::Cm hupB::Km), RJ1989 (RJ1796 ihfB::Cm),and RJ1802 (RJ1796 fis::Km) were gifts from Reid Johnson. Antibioticsfor excision assays were obtained from Sigma (St. Louis, Mo.)and used at the following concentrations: ampicillin, 50 µg/ml;chloramphenicol, 34 µg/ml; kanamycin, 25 µg/ml;tetracycline, 25 µg/ml. Oligonucleotides (Table 1) weresynthesized on a Beckmann Oligo 1000 M synthesizer, and DNAsequencing was performed at the Davis Sequencing Facility (Davis,Calif.).

View this table:
[in this window]
[in a new window]
TABLE 1. Oligonucleotides used in this study

trans-complementation vectors for excision assays. Complementation plasmids for production of ^Tn916Xis and ^Tn916Intwere constructed as follows: an EcoRI-HindIII fragment containingopen reading frames (ORFs) 1 and 2 of Tn916 was generated byPCR from a pAM120 template, using primers CORF1N and CORF2C(Table 1). This 1.5-kb product was ligated into similarly digestedpUC18 to produce pKC304, which expressed ^Tn916Xis and ^Tn916Int.Plasmid pKC310, which expressed ^Tn916Int alone, was constructedin a similar manner, using primers CORF2N and CORF2C to amplifya 1.2-kb EcoRI-HindIII fragment that was ligated into pUC18.The primers CORF1N and CORF2N contained a consensus ribosomebinding site and a 9-bp spacer before the initial start codonto allow protein expression from the lac promoter of pUC18 (51).Expression constructs were verified by DNA sequencing, and proteinexpression was verified by the appearance of 8- and 47-kDa bandson a sodium dodecyl sulfate-20% polyacrylamide gel electrophoresis(PAGE) gel, which corresponded to the expected gene productsof ORF1 and ORF2, respectively.

^Tn916Xis purification for DNase I footprinting. To produce the ^Tn916Xis protein for DNase I footprinting assays,ORF1 of Tn916 was cloned into a pET11A expression vector andexpressed from the T7 promoter. A NdeI-BamHI fragment containingORF1 was generated from pAM120 template DNA, using primers EORF1Nand EORF1C (Table 1). The fragment was ligated into similarlydigested pET11A DNA and transformed into BL21(DE3). Purificationof ^Tn916Xis protein was performed as previously described (51).

Mutant minitransposon plasmids for footprinting and excision assays. The 3.25-kb PvuII fragment of pUC18::Tn1545- ${Delta}$ 4 containing theTn1545- ${Delta}$ 4 minitransposon was ligated into the EcoRV site of pACYC184to form pKC405, a low-copy-number reporter plasmid compatiblewith the pUC-derived trans-complementation plasmids. PKC405carried resistance determinants for kanamycin within the Tn1545- ${Delta}$ 4minitransposon and for chloramphenicol within the pACYC184 backbone.

pKC502, the ${Delta}$ X' derivative, was constructed by PCR amplificationof a BamHI-BssHII fragment of pKC405, using primers 5021 and5022 (Table 1). This fragment, which is missing 172 bp upstreamof the BssHII site of Tn1545- ${Delta}$ 4, disrupts the X' binding siteand was ligated into similarly digested pKC405. PKC600 ( ${Delta}$ X'/ ${Delta}$ X1)and pKC601 ( ${Delta}$ X'/ ${Delta}$ X2) were constructed by overlap extension PCRmutagenesis (31, 36) of pKC502. For pKC600 and pKC601, a KpnI-XbaIfragment was generated by PCR using primers 600N and 600C; X1and X2 site disruptions were introduced using mutagenic primerspairs 6011 and 6012, 6021 and 6022, respectively (Table 1).The mutant fragments were then ligated into similarly digestedpKC502. pKC603 ( ${Delta}$ X'/ ${Delta}$ X1/ ${Delta}$ X2) was constructed in a similar fashionby overlap extension mutagenesis of pKC600 ( ${Delta}$ X'/ ${Delta}$ X1) using mutagenicprimers 6021 and 6022. The resulting sequences of the modifiedXis binding sites are shown below in Fig. 3B.

View larger version (94K):
[in this window]
[in a new window]
FIG. 3. (A) Sequence alignment of the X1 (attL), X2 (attL), and X' (attR) footprints. Nucleotides conserved in at least two of the three footprints are shaded gray. A box encloses the consensus nucleotide sequence within each binding site. (B) DNA sequences that disrupt the ^Tn916Xis binding sites. DNA fragments attR ${Delta}$ X', attL ${Delta}$ X1, and attL ${Delta}$ X2 were generated by PCR from pKC502, pKC600, and pKC601, respectively, and radiolabeled at one end (indicated by the asterisk). Triangles and diamonds indicate the locations of the DR2 and core binding sites, respectively. Transposon DNA is indicated by the thin line, and flanking plasmid DNA is indicated by the thick line. The sequences of the modified X', X1, and X2 sites are indicated, with modified bases shown in bold. (C) DNase I footprints of the Tn916 left arm with disrupted X1 (attL ${Delta}$ X1), X2 (attL ${Delta}$ X2), or X1 and X2 (attL ${Delta}$ X1 ${Delta}$ X2) sites. Positions of the X1 and X2 footprints are indicated by arrows. Lanes 1 to 5, labeled attL ${Delta}$ X1 incubated with 0.2, 0.5, 0.75, 1.0, and 2.0 µM purified ^Tn916Xis, respectively; lane 6, labeled attL ${Delta}$ X1 with no protein; lane 7, Maxam-Gilbert A-G ladder (A+G); lane 8, labeled attL ${Delta}$ X2 incubated with 2.0 µM purified ^Tn916Xis; lane 9, labeled attL ${Delta}$ X2 with no protein; lane 10, labeled attL ${Delta}$ X1 ${Delta}$ X2 incubated with 2.0 µM purified ^Tn916Xis; lane 9, labeled attL ${Delta}$ X1 ${Delta}$ X2 with no protein.

Footprinting and electrophoretic mobility shift assays. Single-end-labeled DNA fragments for DNase I footprinting wereprepared as follows: fragments attLWT and attRWT were generatedby PCR from a pAM120 template, using primer pairs FattL5 andFattL3, FattR5 and FattR3, respectively (Table 1). FragmentattR ${Delta}$ X' was generated by PCR from pKC502, using primers FattR5and FattR3. Fragments attL ${Delta}$ X1, attL ${Delta}$ X2, and attL ${Delta}$ X1 ${Delta}$ X2 were generatedby PCR from pKC600, pKC601, and pKC603 templates, respectively,using primers Fatt ${Delta}$ X and FattL5. The PCR products were digestedat both ends with appropriate restriction enzymes (Table 1),treated with calf intestinal phosphatase (New England Biolabs,Beverly, Mass.), and ³²P-labeled with T4 polynucleotide kinase(New England Biolabs). The end-labeled fragments were againdigested with the appropriate restriction enzyme (Table 1) toremove the radiolabel at one end. Finally, the fragments werepurified with a PCR purification kit (Qiagen).

Binding reactions were performed in binding buffer [10 mM Tris-HCl(pH 8.0), 5 mM MgCl₂, 1 mM CaCl₂, 2 mM dithiothreitol, 100 mMKCl, 1 µg of poly(dI-dC)] in a 200-µl total volume,using 100 kcpm of labeled DNA and ^Tn916Xis (0 to 4 µM).Following a 30-min incubation at 25°C, 0.2 U of DNase I(Life Technologies) was added. The DNase I digestion was stoppedafter 1 min by the addition of 50 µL of stop solution(0.2 M EDTA, 1.5 M CH₃COONH₄, 1 µg of calf thymus DNA,and 2 µg of glycogen). The digested DNA was phenol extractedand ethanol precipitated, diluted to 50 kcpm/µl with sampleloading buffer, and separated on a urea-PAGE sequencing gel.Footprints were visualized by phosphorimaging.

In vivo excision assays of modified minitransposons. For the excision assays of the modified Tn1545- ${Delta}$ 4 minitransposons,chemically competent E. coli DH5 cells were transformed withthe appropriate complementation and reporter plasmids. Followinga 20-min cell recovery, the transformation reaction mixturewas used to inoculate 5 ml of Luria-Bertani (LB) cultures containingchloramphenicol and ampicillin and incubated at 37°C ona roller platform. At each time point, aliquots were removedfrom the culture, serially diluted, and plated on LB ampicillin-chloramphenicolplates (to determine the total CFU), and LB ampicillin-chloramphenicol-kanamycinplates (to determine the unexcised CFU). Expression from thelac promoter was not tightly regulated, and it was not necessaryto induce protein expression with isopropyl-ß-D-thiogalactopyranoside(IPTG). Repeating the excision assays with and without IPTGshowed no increase in excision frequency for the wild-type transposon(data not shown). The ratio of the CFU on both plates was usedto determine excision frequency. For each assay, at least twoserial dilutions were used to determine the excision frequency,and at least three assays were performed on each transposonconstruct.

In vivo excision assays of IHF^-, HU^-, and FIS^- mutant E. coli. Excision assays for the IHF^- (RJ1989), HU^- (RJ1976), and FIS^-(RJ1802) mutant strains used plasmid pAM120. pAM120 containedintact ORFs for ^Tn916Int and ^Tn916Xis; therefore, no complementationplasmid was needed. The assays were performed in a similar fashionto the minitransposon excision assays, using ampicillin to selectfor the plasmid backbone and tetracycline to select for theintegrated transposon.

Detection of attT by PCR. Plasmid DNA was purified from cultures that had excised for72 h using a Qiaprep miniprep kit (Qiagen Inc., Valencia, Calif.).The purified DNA was diluted to 10 nM in distilled H₂O and subjectedto 30 rounds of PCR (94°C denaturation, 30 s; 55°C annealing,30 s; 75°C extension, 45 s) in Taq Master mix (Qiagen Inc.).The PCR was primed using primers specific for the left and rightends of the excised circular intermediate (PL1 and PR1; Table1), each at a concentration of 2.5 µM. The PCR productswere sequenced directly in both directions using primers PL1and PR1.

RESULTS

Top

Results

Identification of a new ^Tn916Xis binding site on the left arm of the transposon (attL). In this study, we sought to determine the importance of specific^Tn916Xis-transposon interactions in the in vivo excision oftransposon Tn916. Previous work has shown that the ^Tn916Xisprotein stimulates transposon excision in vitro (50) and invivo (37, 38, 48, 56) and has localized its binding to singlesites on the left and right arms of the transposon, sites X1and X', respectively (51) (Fig. 1). Initially, we investigated^Tn916Xis protein binding to the transposon using DNase I footprintingto confirm previous study findings and to further delineatespecific ^Tn916Xis binding sites to be studied in vivo. A DNAfragment containing the nucleotide sequence of the left armof the transposon (attLWT) (Fig. 2A) was radiolabeled at its5' end and assayed for its ability to bind the ^Tn916Xis proteinusing DNase I footprinting. The results of this assay are displayedin Fig. 2B. Titration of attLWT with increasing amounts of purified^Tn916Xis resulted in the appearance of two protected regionsof similar size, between +28 to +74 bases and between +99 to+145 bases from the left end of the transposon (Fig. 2B, lanes3 to 6). These regions are denoted X1 and X2, respectively,and are located between the DR2 and core sites at the left endof the transposon (Fig. 1). The X1 and X2 footprints exhibithypersensitive cleavage to DNase I at nucleotides spaced approximatelyevery 10 bp (Fig. 2B), suggesting that the DNA may be wrappedaround the protein (51). A similar pattern of protection isseen in Xis binding to the phage P arm (63). Previous work hasrevealed that the ^Tn916Xis protein binds to the X1 site withinthe attL and a similar pattern of DNase I hypersensitivity (51).Our results confirm these observations and reveal a second,previously unidentified ^Tn916Xis binding site on the left arm,site X2.

View larger version (31K):
[in this window]
[in a new window]
FIG. 2. Footprinting reveals two ^Tn916Xis binding sites on the left arm of the transposon (attL). (A) DNA fragment used for footprinting studies. The DNA fragment attLWT contains the wild-type attL transposon sequence and was generated by PCR from pAM120 and radiolabeled at one end (indicated by the asterisk). Triangles and diamonds indicate the locations of the DR2 and core binding sites, respectively. Transposon DNA is indicated by the thin line, and flanking plasmid DNA is indicated by the thick line. The bases at the fragment ends are labeled (positive numbers run from the core to the interior of the transposon; negative numbers run from the core to the flanking DNA). The locations of the X1 and X2 footprints are indicated by dashed boxes. (B) DNase I footprint of the Tn916 left arm (fragment attLWT) revealing two ^Tn916Xis binding sites. The scale at the left indicates the distance in base pairs from the left core sequence (see panel A). The footprints of X1 and X2 are labeled. Lane 1, Maxam-Gilbert A-G ladder; lane 2, labeled attLWT with no protein; lanes 3 to 6, labeled attLWT incubated with 0.5, 0.75, 1.0, and 2.0 µM purified ^Tn916Xis, respectively; lane 7, labeled attLWT with no DNase I.

Binding of the ^Tn916Xis protein to the right arm of the transposon(attR) was also investigated by DNase I footprinting and waslargely consistent with previous work (51); the ^Tn916Xis proteinprotects site X' located +110 to +158 bases from the right endof the transposon and exhibits a similar pattern of DNase Ihypersensitivity as observed for sites X1 and X2. Interestingly,the X1 and X2 sites are footprinted in the presence of 1 µgof poly(dI-dC) DNA, while the footprint of the X' site doesnot occur when dI-dC DNA is present. This result suggests that^Tn916Xis binding to the X' site is less sequence specific thanits binding to the X1 and X2 sites. In each footprint, the ^Tn916Xisprotein protects a large 44- to 48-bp region of DNA. Given thesmall size of the ^Tn916Xis protein (68 amino acids), it is likelythat a protein multimer protects each site. This conclusionwas substantiated by a gel shift assay with ^Tn916Xis bindingto a 50-bp fragment containing the X2 site, which showed thatmultiple but poorly resolved ^Tn916Xis-DNA complexes are formedat the protein concentrations used in the DNase I footprintingstudies (data not shown).

^Tn916Xis protein binding can be disrupted by nucleotide mutagenesis. Comparison of the nucleotide sequences protected by the ^Tn916Xisprotein suggests that it recognizes a consensus nucleotide sequencelocated within the 3' half of each footprint (Fig. 3A). Eachprotected site is of a similar size and consists of two distincthalves. The 5' half is A-T rich but poorly conserved (only 6of 20 bases are conserved in at least two of the three sites).The 3' half of each site contains the completely conserved sequence(A/T)(A/T)GAAA, which is located within a stretch of nucleotidesin which 17 out of 20 bases are conserved in at least two ofthe three sites. In order to enable the construction of reporterplasmids that selectively disrupt ^Tn916Xis protein binding,we created the mutant binding sites attL ${Delta}$ X1, attL ${Delta}$ X2, and attR ${Delta}$ X',which contain multiple nucleotide substitutions within the 3'region of sites X1, X2, and X', respectively (Fig. 3B). Titrationof attL ${Delta}$ X1 with purified ^Tn916Xis (up to 2 µM) failed toprotect the X1 site but still footprinted site X2 (Fig. 3C,lanes 2 to 5). In contrast, selective mutation of the X2 sitein the attL ${Delta}$ X2 construct disrupts binding to the X2 site, butthe ^Tn916Xis protein can still bind to site X1 (Fig. 3C, lanes8 to 9). We have also shown that protein binding to the X' sitecan be eliminated by the introduction of several nucleotidesubstitutions within this site (Fig. 3B, attR ${Delta}$ X', and data notshown). When nucleotide substitutions are introduced into boththe X1 and X2 sites (construct attL ${Delta}$ X1 ${Delta}$ X2 contains the substitutionsfound in attL ${Delta}$ X1 and attL ${Delta}$ X2), ^Tn916Xis binding to the attL armis completely eliminated (Fig. 3C, lanes 10 to 11). Our resultsindicate that ^Tn916Xis binding to the X1 and X2 sites in attLrequires specific recognition of all or a subset of at least11 bp in the 3' half of the footprint. The data also indicatethat the ^Tn916Xis protein can bind to the X1 and X2 sites independently,because disruption of either the X1 or X2 sites does not affectprotein binding to the adjacent site (Fig. 3C, lanes 1 to 5,8).

Xis protein binding to either the X1 or X2 sites is required for efficient excision. In order to determine if site-specific binding to the transposonis responsible for mediating the stimulatory effect of the ^Tn916Xisprotein, we employed an in vivo excision assay. The assay monitorstransposon excision from reporter plasmid pKC405 (Fig. 4A) andits derivatives that contain mutations in each of the ^Tn916Xisbinding sites. The transposon portion of the plasmid was derivedfrom minitransposon Tn1545- ${Delta}$ 4, which contains the left and rightends of the Tn1545 transposon surrounding the aphA-3 gene encodingresistance to kanamycin (48). Tn1545 is closely related to Tn916;their left and right arms are identical for 186 and 108 bp,respectively. The pKC405 plasmid is based on the low-copy-numberplasmid pACYC184 and confers chloramphenicol resistance on itsplasmid backbone. Therefore, excision of the Tn1545- ${Delta}$ 4 elementfrom pKC405 can be monitored by the appearance of kanamycinsensitivity when the ^Tn916Int protein alone or both the ^Tn916Intand ^Tn916Xis proteins are expressed from the trans-complementationplasmids pKC310 or pKC304, respectively. It should be notedthat for a cell to become kanamycin sensitive, all copies ofthe transposon must excise from the reporter plasmids in thecell. The results of the in vivo excision assay for plasmidscontaining the wild-type and altered forms of the minitransposonare shown in Fig. 4B. The plot shows the extent of transposonexcision over a 72-h period following transformation with thereporter and complementation plasmids. When both the ^Tn916Intand ^Tn916Xis are provided in trans, near-complete excision ofthe minitransposon is observed from plasmid pKC405 within 24h. In contrast, when just the ^Tn916Int protein is provided intrans (pKC310), only 12.3% ± 1.4% (mean ± standarddeviation) of the cells show compete excision of the transposonafter 72 h, and no excision occurs in the absence of ^Tn916Int(data not shown). These results demonstrate that the ^Tn916Intprotein is required for excision and that the ^Tn916Xis proteinstimulates excision approximately eightfold in E. coli.

View larger version (62K):
[in this window]
[in a new window]
FIG. 4. (A) Schematic of pKC405, the pACYC184 derivative containing the Tn1545- ${Delta}$ 4 minitransposon. The plasmid backbone is shown as a solid line, and the transposon is shown as a white box. The plasmid-borne CAT gene and transposon-borne aphA-3 gene are shown as arrows. p15ORI is shown as a gray box. The left and right transposon attachment points are labeled (attL and attR). N-terminal ^Tn916Int and ^Tn916Xis protein binding sites are shown as triangles and hexagons, respectively. (B) Excision assay time courses of the ^Tn916Xis mutant minitransposons. X1, X2, and X' refer to wild-type binding site sequences; ${Delta}$ X1, ${Delta}$ X2, and ${Delta}$ X' are disrupted binding sites. The sequences of the disrupted sites are presented in Fig. 3B. The trans-complementation plasmids pKC304 and pKC310 express ^Tn916Xis and ^Tn916Int, and ^Tn916Int only, respectively. Excision frequencies are expressed as 100 - (100) (unexcised CFU/total CFU). Error bars indicate the standard deviations from three independent trials.

The results of the excision assay indicate that the presenceof only one ^Tn916Xis binding site on the left arm is requiredfor full stimulation of excision by the ^Tn916Xis protein (Fig.4B). A series of reporter plasmids were constructed in whichthe ^Tn916Xis binding sites X', X1, and X2 were disrupted. Thesets of nucleotide substitutions used to disrupt each site areidentical to those used in the footprinting studies (Fig. 3B,mutant sites attL ${Delta}$ X1, attL ${Delta}$ X2, attL ${Delta}$ X1 ${Delta}$ X2, and attR ${Delta}$ X') and thereforeenable the impact of specific ^Tn916Xis-transposon interactionson excision to be assessed. The elimination of the X' site (pKC502; ${Delta}$ X'), results in near-wild-type levels of excision (99.0% ±0.8% excision after 24 h; Fig. 4A). In addition, plasmids thateliminate protein binding to both the X' site and either theX1 or X2 sites show wild-type levels of excision (pKC600, ${Delta}$ X'/ ${Delta}$ X1;and pKC601, ${Delta}$ X'/ ${Delta}$ X2). However, when all three sites are removed(pKC603, ${Delta}$ X'/ ${Delta}$ X1/ ${Delta}$ X2), a ${approx}$ 75% reduction in excision is observedcompared to that in the wild-type plasmid (pKC405) after 24h, and only 70.0% ± 1.3% of the cells show complete excisionof the transposon after 72 h. The results of the excision assaysuggest that ^Tn916Xis protein binding to at least one of theleft arm binding sites is necessary for wild-type levels ofexcision, while binding to the X' site appears to be dispensableunder physiological growth conditions. The results also suggestthat the ^Tn916Xis protein can exert its stimulatory effectsthrough nonspecific DNA or protein-protein interactions, sinceplasmids that disrupt all three of its binding sites still undergoexcisive recombination to a greater extent than when the ^Tn916Xisprotein is absent.

In order to verify that the decrease in kanamycin sensitivityresulted from the excision of the Tn1545- ${Delta}$ 4 element from thereporter plasmids, we confirmed the existence of the Tn1545- ${Delta}$ 4excision products (attB and attT) through restriction enzymemapping and PCR. Transposon excision from the reporter plasmidsshould regenerate the attB site (Fig. 5A) in a replicating plasmidthat confers chloramphenicol resistance. The presence of attBwas confirmed by isolating cellular DNA after a 72-h in vivoexcision assay, digesting the DNA with restriction enzyme PvuII,and separating the resultant fragments on an agarose gel (Fig.5B). Two closely spaced PvuII sites are located in the reporterplasmids, and PvuII digestion is expected to generate linearfragments of 6.5 and 4.3 kb for plasmids containing the Tn1545- ${Delta}$ 4transposon and plasmids that have lost the transposon, respectively(Fig. 5A). As shown in Fig. 5B, the observed excision frequenciesmeasured by the appearance of kanamycin resistance correlatewell with the appearance of the excisant product attB. For example,plasmids exhibiting high levels of excision in the assay (pKC405,pKC502, pKC600, and pKC601) all showed high levels of productformation, while reporter plasmids with diminished excisionfrequencies (pKC603) also showed the presence of the unexcisedreporter plasmid. Transposon excision from the reporter plasmidshould also generate a nonreplicative circular intermediatethat contains the attT site (8). To verify the presence of aregenerated attT site in the circular intermediate product ofexcision, we used PCR on the purified excision products. Figure5C shows the results of PCR using primers specific for the leftand right arms of the circular intermediate. The 337-bp bandrepresents the intact attT. Purification and sequencing of thisband for each of the PCRs confirmed the expected sequence forattT in each of the excision reactions. We therefore concludethat all of the reporter plasmids used to monitor transposonexcision faithfully excised the transposon, producing the nonreplicativetransposon circular intermediate and a regenerated attB siteon the reporter plasmid.

View larger version (32K):
[in this window]
[in a new window]
FIG. 5. Detection of excision products by restriction digestion and PCR. (A) Restriction digestion maps of reporter and complementation plasmids. Fragment A was produced by PvuII digestion of an unexcised reporter plasmid; fragment B was produced by PvuII digestion of the reporter plasmid following Tn1545- ${Delta}$ 4 excision. Fragments C and D resulted from PvuII digestion of the complementation plasmid. The size of fragment D was either 1.8 or 1.5 kb, depending on whether pKC304 or pKC310 was used in the excision reaction. (B) Gel electrophoresis of 72-h excision reaction products following PvuII digestion. The components of each excision reaction are indicated above each lane. The identities of the restriction fragments are indicated at the right and are illustrated in panel A. pKC60X::Tn1545- ${Delta}$ 4 (fragment A) is a fragment containing the unexcised transposon. pKC60X (fragment B) contains the plasmid backbone and restored attB and is indicative of an excision event. Fragments C and D resulted from digestion of the complementation plasmids. (C) PCR detection of attT. PCR primers complementary to the transposon ends (see Materials and Methods) were used to amplify the 337-bp DNA fragment containing the regenerated attT. In each PCR, the template DNA was purified plasmid DNA from a 72-h excision reaction. Lane 1, molecular weight markers; lanes 2 to 6, excision products for each of the reporter plasmids complemented with ^Tn916Int and ^Tn916Xis in trans (pKC304); lane 7, excision products from the wild-type minitransposon complemented with ^Tn916Int only; lanes 8 to 10, negative control reactions using only reporter or complementation plasmids as PCR template. The expected PCR product for attT is indicated by the arrow. The band labeled n.s. is an artifact from nonspecific amplification of the complementation plasmid.

Requirement of host-encoded factors (IHF, HU, and FIS) for transposon excision in E. coli. The host-encoded factor IHF is required for both the excisionand integration of bacteriophage ${lambda}$ (6, 40, 42, 60), while FISenhances bacteriophage excision at suboptimal Xis concentrations(59). Since Tn916-related conjugative transposons and bacteriophage ${lambda}$ appear to be mechanistically related, we tested the dependencyof transposon excision on IHF, FIS, and the histone-like factorHU. This work made use of a plasmid-borne copy of the full-lengthTn916 transposon (pAM120 [21]). As pAM120 contained the intactORFs for ^Tn916Xis and ^Tn916Int, it was not necessary to supplythese proteins in trans from a complementation plasmid. Excisionassays were performed using isogenic wild-type (RJ1796) andIHF^-, FIS^-, and HU^- mutant E. coli strains (RJ1989, RJ1976,and RJ1802, respectively). PAM120 was transformed into thesestrains and assayed for loss of tetracycline resistance carriedby the Tn916 transposon. The rates of transposon excision forthe wild-type, IHF^-, and HU^- strains are shown in Fig. 6A; growthrates are presented in Fig. 6B. The rates of growth and transposonexcision for the FIS^- strain RJ1802 were nearly identical tothe wild-type strain RJ1796 (data not shown). All four strainsreached saturation within 8 h, although the IHF mutant had alonger initial lag phase and did not reach the saturation densityof the wild-type strain. In the wild-type, FIS^-, and IHF^- strains,the maximum amount of excision occurred by the end of the exponentialphase of cell growth, as previously reported (10). Excisionin the wild-type and FIS^- strains was complete before entryinto the stationary phase. In the IHF^- and HU^- strains, transposonexcision continued well into stationary phase. After 24 h, 87.7%± 12.3% of the IHF^- cells had completely excised thetransposon, with complete excision occurring by 48 h. In theabsence of HU, Tn916 excision was severely attenuated. At 24h, the rate of Tn916 excision from RJ1976 (HU^-) was barely significant(7.3% ± 4.5%), compared to the complete excision in wild-typeand FIS^- strains and nearly complete excision in the IHF^- E.coli strain. Excision in RJ1976 reached a maximum of 11.4% ±5.9% over the course of the assay. The excision assays thereforesuggest that neither FIS nor IHF are essential for completeexcision of Tn916, although the IHF^- strain had a slightly reducedrate of excision. The HU protein, however, appears to significantlyenhance Tn916 excision.

View larger version (23K):
[in this window]
[in a new window]
FIG. 6. (A) Excision assay time courses of Tn916 excision in RJ1976 (HU^-; solid circles), RJ1976 (IHF^-; solid triangles), and RJ1796 (wild type; solid squares) E. coli strains. The excision frequency of Tn916 from plasmid pAM120 [100 - (100)(unexcised CFU/total CFU)] versus time (in hours) is displayed. (B) Growth rate of bacterial strains RJ1976 (HU^-; open circles), RJ1976 (IHF^-; open triangles), and RJ1796 (wild type; open squares). Growth rates are expressed as the absorbance of the cell culture at 600 nm versus time in hours.

DISCUSSION

Top

Discussion

Conjugative transposons spread antibiotic resistance determinantsthrough a novel mechanism mediated by the transposon-encoded^Tn916Int and ^Tn916Xis proteins. The ^Tn916Xis protein is a memberof the large and diverse family of recombination directionalityfactors (RDFs) found in numerous bacteriophage, transposon,conjugative plasmid, and bacterial systems (reviewed in reference30). The Xis protein from bacteriophage ${lambda}$ is the best-characterizedRDF, binding sequence-specifically to sites X1 and X2 on thephage (63) (Fig. 1) and stimulating excision by facilitatingthe formation of a recombinogenic nucleoprotein complex (44,45, 58). Because the mechanism of site-specific recombinationof bacteriophage ${lambda}$ is presumed to be similar to the Tn916-liketransposons (54), we investigated how the ^Tn916Xis protein interactswith the transposon and the importance of these interactionsin transposon excision. Previous studies have shown that the^Tn916Xis protein interacts with the X1 and X' sites on the leftand right arms of the transposon, respectively (51). Our ^Tn916Xisfootprinting data confirm the presence of these sites and reveala third and previously unidentified binding site on the leftarm of the transposon (site X2). This discrepancy may be a resultof the different binding buffers used in each study. Sequencealignment of X', X1, and X2 reveals two distinct features ofthe Xis binding site: an A-T rich 5' half and a 3' half containinga conserved sequence ((A/T)(A/T)GAAA). A similar organizationof features is observed in IHF binding sites (23, 24). The extensivefootprint (40 to 50 bp) and the observation of multiple Xis-DNAcomplexes by electrophoretic mobility shift analysis suggeststhat a multimer of the ^Tn916Xis protein is required to footprinteach site.

Reasoning that ^Tn916Xis binding to all or a subset of sitesis responsible for mediating ^Tn916Xis-dependent enhancementof transposon excision, we investigated the significance ofthese interactions using an in vivo transposon excision assay.Previous studies have demonstrated the ability of the ^Tn916Xisprotein to enhance transposon excision in E. coli (48, 56),Lactococcus lacti (38), Bacillus subtilis (37), and in vitro(50). However, the role of specific ^Tn916Xis-DNA interactionsin transposon excision has not been investigated. Systematicdisruption of the three ^Tn916Xis binding sites on the transposonreveals that sites X', X' plus X1, and X' plus X2 can be disruptedwithout affecting transposon excision. However, ^Tn916Xis bindingto at least one of the left arm sites is important, since theremoval of both of these sites causes a ${approx}$ 75% reduction in theability of ^Tn916Xis to enhance excision. These results parallelthose of bacteriophage ${lambda}$ , where Xis-DNA interactions on one armof the phage (the P arm) are required for the stimulatory effectof this protein (63). In bacteriophage ${lambda}$ , the Xis, FIS, and IHFproteins are believed to stabilize distinct nucleoprotein architecturesthat appropriately position Int for the initial top-strand transferstep (44, 45, 58) and for the resolution of subsequent Hollidayjunction intermediates (20). These accessory factors dramaticallydistort DNA (57) and may stabilize looped DNA structures inwhich the Int protein is simultaneously bound to both the coreand arm sites (28, 40, 41). Since the separation of the DR2and core sites on each arm of the transposon is shorter thanthe persistence length of DNA, their approximation through a^Tn916Int bridge may be stabilized in a manner similar to thatwith bacteriophage ${lambda}$ . This conclusion is substantiated by theobserved need for at least one functional ^Tn916Xis binding site(either X1 or X2) between the DR2 and core sites and by theobserved DNase I hypersensitivity caused by ^Tn916Xis proteinbinding, which suggests that this protein distorts the duplex(51, 63). However, our finding that the IHF protein is not essentialfor excision, and the fact that ${approx}$ 12% of the cells show completetransposon excision even in the absence of ^Tn916Xis, indicatethat these proteins do not play an essential role in stabilizingdistinct protein-nucleic acid structures during recombinationin E. coli. Furthermore, the independence of excision on FISis not surprising, considering the absence of a FIS bindingsite on the transposon and the lack of fis in genera which supportTn916 recombination (29). Interestingly, the ^Tn916Xis proteinhas a much more profound effect on the rates of transposon excisionin bacterial species other than E. coli (37, 38). This suggeststhat ^Tn916Xis may be dispensable in E. coli because of the presenceof a host-encoded accessory factor other than IHF or FIS, whichis absent in these other species.

Our data suggest a significant role for HU in Tn916 excisiverecombination, as substantiated by the ${approx}$ 90% attenuation of excisionefficiency in the HU^- E. coli strain. HU may serve as an architecturalfactor in the Tn916 recombinogenic complex, directly bindingto and bending the transposon ends or stabilizing particularlooped structures in a manner analogous to IHF in bacteriophage ${lambda}$ recombination. The relative importance of IHF and HU in excisiverecombination in vivo differs significantly between Tn916 andbacteriophage ${lambda}$ . IHF stimulates in vivo intramolecular recombinationof ${lambda}$ attL and attR (22) or two ${lambda}$ attL sites (27, 55) 5- to 10-fold,while HU confers only a modest (25%) increase in efficiency(22). In contrast, Tn916 excised from the HU⁺ E. coli strainover 10 times more efficiently than in the HU^- strain, whileexcision frequency in IHF⁺ and IHF^- strains differed by lessthan 13%. In light of the differing biological contexts of bacteriophage ${lambda}$ and transposon Tn916, it is not surprising that the lattergenetic element would evolve different host factor requirementsfor efficient excision. HU would appear to be better suitedthan IHF for enhancing recombination in a variety of geneticbackgrounds, since homo- and heterodimeric HU-like proteinsare ubiquitous in prokaryotes (16, 46, 64), while IHF appearsto be restricted to gram-negative bacteria (16). For example,neither B. subtilis nor Streptococcus thermophilus encode anIHF homologue (16, 29), but both encode HU homologues (HBsu[39] and HSth [16], respectively) and support Tn916 transposition(49). Interestingly, these HU homologues from gram-positivebacteria have been shown to be functionally interchangeablewith E. coli HU in promoting other recombination reactions (2,16), suggesting that they may be interchangeable in Tn916 recombinationas well.

The Xis and ^Tn916Xis proteins may differ in their abilitiesto recruit proteins to the transposon arm. The Xis protein,through cooperative protein-protein interactions, can directthe loading of proteins to the appropriate phage sites requiredfor synapsis (19, 26). In particular, the Xis protein bindscooperatively to sites X1 and X2 on the phage, facilitatingthe binding of the FIS (43) and Int proteins to the phage arm(7, 43). Some of these elaborate cooperative interactions appearbe absent in Tn916, since ^Tn916Xis does not cooperatively assembleonto sites X1 and X2. The lack of cooperativity in the Tn916system may result from the greater spacing between the X1 andX2 sites in Tn916 (27 bp) compared to the binding sites forthe Xis protein in bacteriophage ${lambda}$ (7 bp) and is consistent withthe lack of amino acid sequence similarity between the ^Tn916Xisand Xis proteins. It thus appears that the role of the Xis and^Tn916Xis proteins are fundamentally similar in E. coli, as bothpromote excision by binding between the core and arms sitesand presumably stabilize distorted nucleoprotein architecturesrequired for excision. However, in E. coli the stimulatory effectof ^Tn916Xis is significantly reduced compared to that of Xis,a presumable result of its weaker affinity for specific siteson the transposon and its inability to cooperatively assemblewith itself. The Xis protein has been shown in vitro to inhibitintegrative recombination in bacteriophage ${lambda}$ (1). Similarly,it is conceivable that the primary function of ^Tn916Xis is notto stimulate excision, but to inhibit integration. The transcriptionof the ^Tn916Xis and ^Tn916Int proteins is coordinately regulated(11), suggesting that within the donor, equal amounts of ^Tn916Xisand ^Tn916Int could function to promote excision, while the presenceof ^Tn916Xis could prevent reintegration. Since the ^Tn916Intprotein is only required to be produced in the donor cell (5)and is presumably transferred to the new recipient cell alongwith the transposon, the absence of ^Tn916Xis in the recipientwould then enable the integrative pathway to dominate.

ACKNOWLEDGMENTS

We thank Reid Johnson and Gordon Churchward for useful discussions.

This work was supported by a grant from the National Institutesof Health (GM57487).

K.M.C. and M.I. contributed equally to this work.

FOOTNOTES

* Corresponding author. Mailing address: Dept. of Chemistry and Biochemistry, DOE Laboratory of Structural Biology and Molecular Medicine, 405 Hilgard Ave., Los Angeles, CA 90095-1570. Phone: (310) 206-2334. Fax: (310) 206-4749. E-mail: rclubb{at}mbi.ucla.edu.

REFERENCES

Top