Site-specific DNA invertible elements often control the production of bacterial surface proteins that are subject to phase variation (ON/OFF switching). Inversion of the DNA element occurs as a result of the reciprocal exchange of DNA catalysed by a specialized enzyme (recombinase) that acts at specific sites. By continually switching the orientation of the invertible element in the chromosome, and consequently the production of the variable protein(s), the cell population remains continually responsive to environmental change such as immunological challenge. In addition to phase-variable surface proteins, Mycoplasma pulmonis has a family of phase-variable restriction-modification enzymes. We report here that a single recombinase in M. pulmonis, HvsR, catalyses independent DNA inversions at non-homologous loci, causing variations in surface lipoproteins and in the DNA recognition sequence specificity of restriction enzymes. Thus, HvsR is a site-specific DNA recombinase with dual substrate specificity.
Most bacterial proteins exhibiting phase variation are localized to the cell surface, notable exceptions being the phase-variable restriction and modification (R-M) enzymes (Dybvig et al., 1998; Saunders et al., 1998; Ryan and Lo, 1999; De Bolle et al., 2000). Mycoplasma pulmonis has two paralogous hsd (host specificity determinant) operons that encode type I R-M enzymes. The hsd loci are organized as invertible elements capable of generating polymorphic hsdS gene sequences via site-specific DNA inversion. Because the HsdS subunit determines the DNA recognition sequence specificity of the restriction enzyme, the generation of hsdS variants results in the phase-variable production of a family of restriction enzymes of differing specificities (Dybvig et al., 1998). Another DNA inversion system in M. pulmonis is the vsa (variable surface antigen) locus that encodes a family of surface lipoproteins originally known as the V-1 antigens. Site-specific DNA inversion at the vsa locus results in the placement of one of several vsa genes within the vsa expression site and consequently, the phase-variable production of the Vsa proteins (Bhugra et al., 1995; Shen et al., 2000).
From the complete genome sequence of M. pulmonis (Chambaud et al., 2001), an open reading frame (ORF) (Mypu_5310) adjacent to the vsa locus was identified that is predicted to encode a site-specific DNA recombinase. The recombinase, designated HvsR (hsd and vsasite-specific recombinase), has significant homology with several XerD recombinases and has conserved motifs that place it firmly in the λ integrase family of site-specific DNA recombinases (Ron et al., 2002). In the current study, we find that HvsR catalyses inversions at both the hsd and vsa loci, causing variations in the production of R-M enzymes and surface proteins respectively.
Construction and characterization of hvsR mutants
To isolate mutants in hvsR and other genes of interest, M. pulmonis strain CT was subjected to transposon mutagenesis. Library member CT228 had transposon Tn4001T inserted at nucleotide position 656621 of the complete genome sequence (http:genolist.pasteur.frMypuList), disrupting hvsR such that it would encode a truncated protein of only 70 amino acids rather than the full-length protein of 248 amino acids. Another mutant, CTG36, has Tn4001T inserted at nucleotide position 295837, which disrupts the recA gene such that a truncated RecA protein of only 24 amino acids would be produced. In the experiments described below, CTG36 serves as a control to determine whether (i) the site-specific inversions at the hsd and vsa loci are RecA-independent and (ii) the presence of Tn4001T in the mycoplasmal chromosome inhibits DNA inversions.
Because of the close proximity of hvsR to the vsa locus, it was anticipated that HvsR might catalyse vsa inversions. The vsa locus of strain CT has a single vsa expression site. Site-specific DNA inversion alternately places one of seven different genes (vsaA, vsaC, vsaE, vsaF, vsaG, vsaH and vsaI ) at the expression site, whereas the other six genes are silent (Bhugra et al., 1995; Shen et al., 2000). The vsaA gene occupies the expression site in the majority of cells in CT228 and CTG36. Using a previously described experimental design (Shen et al., 2000; Gumulak-Smith et al., 2001), polymerase chain reaction (PCR) amplification of CT228 genomic DNA using a primer specific for the expression site paired with a vsaA-specific primer yielded a product of the expected size (Fig. 1A). No PCR product was obtained when the vsa expression site primer was paired with primers specific for the other vsa genes. Thus, CT228 lacks a subpopulation of cells with vsa genes other than vsaA occupying the expression site. Subpopulations of cells expressing alternative vsa genes are readily detected by PCR in wild-type M. pulmonis cells (Shen et al., 2000). In the current study, the presence of such subpopulations was verified for wild-type cells (data not shown) and also for the recA mutant CTG36 (Fig. 1A), demonstrating that Tn4001T does not inhibit the ability of M. pulmonis to undergo hsd and vsa inversions. Because the vsa and hsd recombination sites are not closely related, it was originally deemed unlikely that a single recombinase would catalyse both vsa and site-specific hsd inversions. Unexpectedly, using previously described primer pairs for PCR amplification of hsd inversion products (Sitaraman and Dybvig, 1997), inversions within the hsd loci were readily detectable in wild-type (data not shown) and CTG36 strains but not in strain CT228 (Fig. 1B).
A second hvsR mutant, CTG151, has been obtained that has Tn4001T inserted at nucleotide position 656303 and would produce a truncated HvsR protein of 176 amino acids. PCR analysis to identify hsd and vsa inversions in CTG151 indicated that this strain has a phenotype indistinguishable from that of CT228 (data not shown). The finding that two independent hvsR mutants, CT228 and CTG151, lack DNA inversions strongly indicates that HvsR is required to catalyse both vsa and hsd inversions.
Analysis of HvsR-catalysed vsa and hsd inversions in Escherichia coli
One explanation for how a single recombinase promotes inversions at the non-homologous vsa and hsd recombination sites could be that another mycoplasmal protein(s) interacts with HvsR and the recombination sites to confer sequence specificity to the recombination reaction. Therefore, experiments were performed to determine whether vsa and hsd inversions would occur in E. coli expressing the cloned hvsR gene. The absence of UGA codons, which code for tryptophan in mycoplasmas, in the hvsR gene obviated the necessity of using UGA suppressor strains of E. coli.
The hvsR gene was amplified by PCR and cloned into plasmid pGEM3Z-NE, a derivative of pGEM3Z that lacks lacZα sequences. The hvsR gene was cloned in two orientations (Fig. 2A). In pREC+, hvsR is oriented such that the gene would be transcribed from the lac promoter (plac). In pREC−, hvsR is oriented oppositely from the direction of transcription from plac and not expressed. The availability of pREC+ and pREC− allowed the discernment of rearrangements that are specifically linked to HvsR activity from those that might occur spontaneously.
Several reporter systems were devised to evaluate the ability of various vsa- and hsd-derived DNA sequences to undergo site-specific recombination catalysed by HvsR. The general principle was to generate reporter strains of E. coli harbouring two sequences of interest (recombination sites) in a predetermined orientation. This reporter strain was transformed with pREC+, pREC− or pGEM3Z-NE, and the transformants were directly screened for the occurrence of site-specific inversion by means of PCR using appropriately designed primers. The expectation is that two copies of a suitable substrate sequence, when oriented as inverted repeats, will undergo site-specific recombination (inversion) in the presence of HvsR but not otherwise. Therefore, suitable substrates in the correct orientation are expected to recombine when the reporter strain is transformed with pREC+, but not when transformed with pREC– or pGEM3Z-NE.
A lacZ-based reporter system was constructed for detecting hsdS inversions in E. coli. Two copies of hsdS-derived sequences (361 bp) that were previously shown to contain the hsdS recombination sites vip (5′-CAAAGT GCAATA-3′) and hrs (5′-TAATTAAGATTATTGAACCT-3′) (Dybvig et al., 1998) were cloned as inverted repeats in orientation I (Fig. 2B) and targeted to the E. coli chromosome (see Experimental procedures). Site-specific inversion involving the hsdS sequences results in the reporter alternating between orientations I and II (Fig. 2B). In orientation II, the hsdS-derived sequence, which is in-frame with the lacZα gene, is transcribed from the inducible promoter plac. The HsdS-LacZα fusion peptide exhibits β-galactosidase activity upon α-complementation and can be detected by the blue colour produced on Xgal plates, indicating the occurrence of site-specific DNA inversion.
The reporter strain containing the hsdS sequences in orientation I was transformed with one of three plasmids: pREC+, pREC– or pGEM3Z-NE as described (Chung et al., 1989). When monitored on Xgal plates, transformants containing pREC+ were blue (LacZ+) and transformants containing pREC− or pGEM3Z-NE were white (LacZ–). Changes in the orientation of the reporter construct due to DNA inversion were assayed by performing PCR on individual bacterial colonies. The PCR product corresponding to orientation II (865 bp, primer pair o.5 + o.10) was obtained only from colonies of the reporter strain transformed with plasmid pREC+ (Fig. 2C). No PCR product corresponding to orientation II was detectable in colonies transformed with plasmid pREC−. Both types of transformants tested positive for orientation I as expected, as evidenced by the presence of the corresponding PCR product of 665 bp (primer pair o.9 + o.5). Sequence analysis of the PCR products corresponding to the two orientations further verified the occurrence of the predicted site-specific inversion involving hsdS-derived sequences. Therefore, site-specific inversion of hsdS sequences required the HvsR recombinase.
To determine if vsa-derived sequences could undergo site-specific inversion in E. coli, another reporter system was devised by cloning a section of the vsa locus into plasmid pACYC184 giving the reporter plasmid pVrsA-D (Fig. 3A). The cloned vsa fragment contained oppositely oriented vsaA and vsaD gene sequences, including the vrs box recombination site vrsA and vrsD respectively (Bhugra et al., 1995; Shen et al., 2000). The orientation of this fragment is such that an amplicon (380 bp) is obtained when using the primer pair o.6666 + o.6859 but not with the primer pair o.6666 + o.9590. Inversion between the vrsA and vrsD recombination sites se-quences results in a PCR amplicon (411 bp) when using the primer pair o.6666 + o.9590. The PCR product indicative of site-specific inversion was obtained when the reporter strain was transformed with pREC+, but not upon transformation with pREC− or pGEM3Z-NE (Fig. 3B). Sequence analysis of the PCR products demonstrated that inversion had indeed occurred as expected. The vrs sequences are conserved between M. pulmonis strains, even though strains differ in the overall number and sequences of the vsa genes that are present (Shen et al. 2000). This result is therefore applicable to other pairwise combinations of vsa genes with intact vrs boxes. To summarize, the hsdS- and vsa-based reporter systems indicate that HvsR is capable of carrying out inversions at both the hsd and vsa loci in the absence of accessory mycoplasmal proteins.
The following series of experiments were devised to examine the hsd sequence recognized by HvsR. Suitable reporter systems were constructed and screened for site-specific inversion by PCR using methods similar to those described above. As stated earlier, two kinds of recombination sites have been identified at the hsd loci: vip and hrs. We cloned pair-wise combinations of the vip-hrs, vip and hrs sequences, represented by Seq1 and Seq2 in Fig. 4, into the plasmid pACYC184. E. coli strains harbouring the resulting reporter plasmid were transformed with pREC+, pREC− or pGEM3Z-NE. Two transformant colonies were selected randomly from each of the three transformations and screened for the presence of inversions by PCR. Inversions involving any given sequence pair (Seq1 and Seq2) were detected by PCR. The primer combination o.6 + o.7 would result in a PCR product of 1.2 kb. Should inversion occur between Seq1 and Seq2, a PCR product of 1.0 kb would be obtained using the primer pair o.6 + o.8. HvsR catalysed inversions between two viphrs sequences (Table 1, row 1, column 3) or between a viphrs (5′-CAAAGTGCAATATAATTAAGATTATTGAACCT-3′) and an hrs (5′-TAATTAAGATTATTGAACCT-3′) sequence (Table 1, row 2, column 3) oriented as inverted repeats. However, no inversion was observed when a viphrs and a vip (5′-CAAAGTGCAATA-3′) sequence (Table 1, row 3, column 3) were provided as substrates. The results of these experiments indicate that the 20 bp hrs sequence is probably the minimal sequence recognized by the recombinase at the hsd loci. The 12 bp vip sequence did not recombine with the vip-hrs se-quence, indicating conclusively that the vip sequence is insufficient for recombination by HvsR.
Table 1. . Testing specific hsd linker combinations ( Fig. 4 ) as substrates for site-specific DNA inversion in E. coli.
Inversion occurrence when transformed with
The vrs box recombination sequence (5′-CATCAAATA ATGAACAAAGTGGAAATAATTC-3′) of the vsa genes (Bhugra et al., 1995; Shen et al., 2000) and the h recombination sequence (5′-TAATTAAGATTATTGAACCT-3′) of the hsd genes do not exhibit significant sequence similarity. The finding that HvsR catalyses both vsa and hsd inversions is therefore an unexpected result. However, it corroborates previous studies of the mycoplasmal populations that emerge in the respiratory tract of infected rats. M. pulmonis isolated from the trachea and lung exhibited high levels of variation in both the Vsa proteins and in R-M enzymes as compared with organisms isolated from the nose (Gumulak-Smith et al., 2001). Having a single recombinase catalyse inversions at vsa and hsd provides a degree of linkage between surface antigen and restriction enzyme variation in M. pulmonis. Recent work has revealed similar examples of phenotypic switching involving R-M enzymes. Streptococcus pneumoniae has a DNA invertible element that is predicted to encode a phase-variable restriction system (Tettelin et al., 2001). The transcription of putative R-M genes is induced when Helicobacter pylori contacts human gastric cells (Donahue et al., 2000). Although antigenic variation is a common and understandable strategy used by bacteria to combat host defences, the significance of R-M enzyme variation, especially during pathogenesis, is unknown.
Recombination sites recognized by members of the integrase family of site-specific recombinases exhibit imperfect dyad symmetry (Grainge and Jayaram, 1999). In the case of the vsa recombination site vrs, the dyad symmetry is weak (Fig. 5A). In contrast, the dyad symmetry of the various hsd recombination sites is impressive (Fig. 5B). Several nucleotides in the region of the hsd dyad symmetry are present in vrs. This low level of nucleotide similarity is presumably sufficient to account for the ability of HvsR to promote both vsa and hsd inversions.
Bacterial strains and plasmids
Mycoplasma pulmonis strains CT and KD735-15 have been described previously ( Shen et al., 2000 ) and were propagated in mycoplasma medium containing whole horse serum ( Dybvig et al., 2000 ). All reporter systems were constructed and analysed in the Sure™ strain of Escherichia coli (Stratagene) . Modification of plasmid pGEM3Z (Promega) to pGEM3Z-NE as well as cloning of pREC+ and pREC− was carried out in E. coli JM109 (Promega). E. coli was propagated on Luria–Bertani (LB) medium ( Sambrook et al., 1989 ). Xgal plates to monitor LacZ activity in E. coli consisted of agar supplemented with 160 µg ml −1 of Xgal (5-bromo-4-chloro-3-indoyl-β- d -galactopyranoside) and 1 mM isopropylthiogalactoside (IPTG). Plasmid pPH630, used for targeting the hsdS -based reporter system to the E. coli chromosome, was a gift from Dr Dipankar Manna at the University of Alabama at Birmingham. Plasmid pACYC184 was obtained from New England Biolabs.
Production of a collection of M. pulmonis mutants by transposon mutagenesis
A library of M. pulmonis mutants was constructed by transforming M. pulmonis strain CT with the transposon Tn4001T-containing plasmid pIVT-1, as described (Dybvig et al., 2000). Inverse PCR was used to determine the location of Tn4001T in the genome of each of the transformants using previously described primers and methods (Teachman et al., 2002), leading to the identification of mutants CT228, CTG151 and CTG36.
Cloning of the hvsR gene to generate pREC+ and pREC−
The vector pGEM3Z was digested with NdeI and EcoRI to eliminate the major portion of the resident lacZα sequences. The ends of the linearized plasmid were rendered blunt by using T4 DNA polymerase, and the plasmid was recircularized to obtain the cloning vector designated pGEM3Z-NE. The truncated LacZα peptide encoded by this plasmid consists of only the 34 N-terminal amino acids of the intact peptide and is incapable of causing host colonies to turn blue on Xgal plates.
For amplification of hvsR with high fidelity, Pfx DNA polymerase (Life Technologies) was used. PCR reagents provided with the enzyme were used to formulate the PCR reaction mix as recommended by the manufacturer. The number of cycles of amplification was limited to 25 to further minimize the occurrence of point mutations in the PCR product. The primers used for amplification of hvsR were oligonucleotides Pst1-CTRec-For and BamHI-CT-RecRev, which bind to the 3′ downstream region and the 5′ upstream region of hvsR respectively, The sequences of these and other primers used in this study are listed in Table 2. Incorporated into the primers were restriction sites, Pst I or BamHI, to facilitate the directional cloning of hvsR into pGEM3Z-NE, such that it could be transcribed from plac. The plasmid obtained as a result of this procedure is termed pREC+, in which hvsR would be transcribed from the plac promoter. Likewise, analogous primers that had alternative restriction sites, BamHI in place of Pst I and Sal I in place of BamHI, were used for directional cloning to generate plasmid pREC−, wherein hvsR would not be expressed from plac (Fig. 2A).
Table 2. . List of oligonucleotide sequences used for PCR and linker construction.
. Restriction sites introduced into sequences are given in lower case letters; overhangs are in bold-faced lower case letters.
vsaI -specific primer
For amplification of chosen sections of the hsdS or vsa genes, 10–100 ng of mycoplasmal genomic DNA was used as a template in 50 µl PCR reactions. The reaction contained 5 µl of 10×Taq polymerase buffer (Promega), 200 µM of each dNTP, 20 pmol of each primer and 2.5 units of Taq DNA polymerase. Amplification conditions were as described previously (Gumulak-Smith et al., 2001), except that 5% dimethylsulphoxide was included in the reaction mixtures, the extension times were increased to 5 min, and the number of cycles was increased to 35. To perform PCR on E. coli colonies, the colony was lifted off the agar surface using a pipette tip and suspended thoroughly in 5 µl of phosphate-buffered saline (PBS); 1 µl of this suspension was used as a PCR template.
Construction of in-frame hsdS–lacZα fusions
A lacZ-based reporter system was devised by cloning two copies of sequences that are shared by several hsdS genes at their 5′-ends. Cloning was performed such that the hsdS-derived sequences formed inverted repeats in the final construct. The construction of this reporter system is depicted in Fig. 6. The hsdS sequences were amplified using the primers o.1 (CAATTCATATGACTTTAATGGTG) and o.2 (GTTGAA GACTTTTTTGCTCAC) or with primers o.1′ (gcgtccatggCAA TTCATATGACTTTAATG) and o.2′ (gcaagatctGTTGAA GACT TTTTTGC). The 361 bp amplicon resulting from PCR using primers o.1 and o.2 (hsdS1) was cloned into the vector pCR2.1 (Invitrogen) in register with lacZα sequences, so that a fusion peptide would be produced when the corresponding sequences are positioned downstream of the lac promoter (plac). The amplicon obtained using primers o.1′ and o.2′ (hsdS2) is identical to hsdS1 except for the presence of a Bgl II and NcoI site at the 5′ primer termini to facilitate directional cloning into the NcoI–Bgl II sites of pCR2.1 (Invitrogen). E. coli clones harbouring this construct are designated pVIP-Blue and form blue colonies on Xgal plates. This arrangement of hsdS-derived sequences corresponds to orientation II in Fig. 2B. pVIP-Blue was digested with SspI, cleaving sites within hsdS1 and hsdS2. The digest was precipitated, suspended in TE buffer, and treated with DNA ligase. There were two outcomes of this treatment: regeneration of pVIP-Blue or the ligation of the SspI fragment in the opposite orientation, resulting in the plasmid pVIP-White (see Fig. 6), which has the reporter gene corresponding to orientation I in Fig. 2. E. coli colonies possessing pVIP-White should not express the hsdS–lacZα fusion because of the incorrect orientation of plac relative to the hsdS–lacZα sequences. When the ligation mixture containing these two plasmids (pVIP-Blue and pVIP-White) was transformed into E. coli, the type of plasmid present in each transformant colony was readily inferred by screening for β-galactosidase activity on Xgal plates. E. coli colonies harbouring pVIP-Blue were blue and those with pVIP-White were white. The orientation was also verified by PCR using primer pairs o.10 + o.5 (for orientation II or pVIP-Blue) or o.9 + o.5 (for orientation I or pVIP-White).
Targeting the hsdS–lacZα reporter system to the E. coli nucleoid and elimination of the plasmid used for targeting (pPH630R)
The reporter system was amplified by PCR using primers o.9 and o.10 and pVIP-White as the template (see Fig. 6 and Table 2). Primer o.9 binds to bases 2008–1994 (minus strand) and primer o.10 binds to bases 3816–3834 (plus strand) of pCR2.1. Both primers had XmaI sites built into their 5′-ends to facilitate further cloning. The PCR product so obtained was cloned into the XmaI site of pPH630, a derivative of pNK474 (Way et al., 1984) that contains a modified Tn10 element such that the gene for the Tn10 transposase, IS10R, is not located within the transposon and is transcribed from the IPTG-inducible ptac promoter. The resulting plasmid, pPH630R, was transformed into Sure™E. coli cells by electroporation. Immediately after electroporation, the cells were transferred into broth containing 1 mM IPTG to enable transposase induction and incubated with shaking (225 r.p.m) at 37°C for 30 min. This was done to induce transcription of the transposase gene from the ptac promoter in the vector. Transposase activity targeted the reporter construct, contained within the Tn10 region, to the nucleoid. The cells were then harvested by centrifugation, washed with 1 mM HEPES buffer, suspended in fresh broth, incubated at 37°C for 1 h with continuous shaking, and assayed for colony-forming units (cfu) on agar containing gentamicin (15 µg ml−1).
Novobiocin treatment is known to eliminate plasmids from several bacterial species (McHugh and Swartz, 1977) and, in this case, was used to eliminate pPH630R. Individual colonies were transferred to LB broth containing gentamicin (15 µg ml−1) and novobiocin (250 µg ml−1) and cultured overnight. The cultures were diluted and spread on LB/gentamicin plates. The β-lactamase gene (Amp) of pPH630 (and pPH630R) is located outside the Tn10 region (Fig. 7). Therefore, colonies lacking pPH630R, i.e. not producing β-lactamase, were identified using a filter-based acidimetric assay (Slack et al., 1977). This assay involves the hydrolysis of benzylpenicillin to penicilloic acid by β-lactamase, resulting in a decrease of pH and a change in the colour of the indicator (Bromocresol purple) to yellow. Extraction of total DNA and its electrophoretic analysis from these colonies further proved that the cells lacked pPH630R. These cells were also unable to grow in medium containing ampicillin (60 µg ml−1). The presence and orientation of the reporter construct in the nucleoid was verified again by PCR. One of the clones so obtained, W31, has the hsdS-based reporter stably integrated into its genome in orientation I (Fig. 2B) and was used to assay HvsR activity.
Construction of pVrsA–D
A section of the vsa locus containing vrsA and vrsD was amplified by PCR using the primers BamHI-o.6859 and Sal I-o.9590. Genomic DNA isolated from strain M. pulmonis KD735-15 was used as the template. The amplification products were gel-purified, digested sequentially with Sal I and BamHI, and directionally cloned into the corresponding sites in pACYC184.
Creation of reporter plasmids with vip/hrs sequences in three different combinations
Pairs of oligonucleotides were designed such that they could be annealed to generate linkers, symbolized by Seq1 and Seq2 in Fig. 4, that had the appropriate overhang for convenient cloning into the HindIII–EcoRV and Sal I–SphI sites of pACYC184. As an example, the reporter plasmid containing two viphrs sequences oriented as inverted repeats was constructed as follows. The phosphorylated oligonucleotides Hind3viphrsEcoRV and srhpiv are complementary and used to generate a linker that was cloned into the HindIII–EcoRV site of pACYC184. When annealed, the oligonucleotide pair (linker) had a HindIII overhang (AGCT) at the 5′-end of Hind3viphrsEcoRV. Because EcoRV produces blunt ends, the oligonucleotides were designed such that no overhang was present at the other end of the linker. For annealing, an equimolar mixture of the two oligonucleotides was heated to 98°C for 2 min and slowly cooled, allowing complementary sequences to anneal. This resulted in a double-stranded linker that was phosphorylated at both 5′ termini and had a HindIII overhang at one end and was blunt at the other. This linker was ligated to gel-purified, HindIII-EcoRV-treated pACYC184 and transformed into E. coli as per standard cloning procedures. Plasmid from these transformants was isolated and digested with SalI and SphI. The next phosphorylated oligonucleotide of choice, Sal1viphrsSph1, was annealed to srhpiv, resulting in a double-stranded phosphorylated linker with Sal I (TCGA) and SphI (CATG) overhangs that was ligated to the Sal I-SphI-digested, gel-purified plasmid. The resulting construct was transformed into E. coli. When PCR was performed on the transformants, using Hind3viphrsEcoRV and Sal1viphrsSph1 as primers, the production of a 449 bp amplicon indicated the correctness of orientation and positioning of the two linkers. The resulting plasmid had one copy of vip-h (Seq1) cloned between the HindIII and EcoRV sites and another copy (Seq2) between the Sal I and SphI sites (Fig. 4) such that the two vip-hrs sequences formed inverted repeats. Similar procedures were used to generate the other two linker combinations (viphrs-hrs and viphrs-vip) mentioned in the text. The sequences of the oligonucleotides used for linker construction are provided in Table 2.
Nucleotide sequence analysis
clustalw analysis was performed using the macvector software package.
Strains CT228, CTG151, and CTG36 were generated as part of a M. pulmonis global transposon mutagenesis project by Huilan Yu, Li Liu and K. Dybvig (unpublished data). We thank P. Caldwell and T. Johnson for technical assistance, and D. Manna for providing the targeting plasmid pPH630. This work was supported by NIH grants GM51126 and AI41113.