Heat Shock Treatment Increases the Frequency of Loss of an Erythromycin Resistance-Encoding Transposable Element from the Chromosome of Lactobacillus crispatus CHCC3692

Abstract

A 3,165-bp chromosomally integrated transposon, designatedTn3692, of the gram-positive strain Lactobacillus crispatus CHCC3692 contains an erm(B) gene conferring resistance to erythromycin at concentrations of up to 250 μg/ml. Loss of this resistance can occur spontaneously, but the rate is substantially increased by heat shock treatment. Heat shock treatment at 60°C resulted in an almost 40-fold increase in the frequency of erythromycin-sensitive cells (erythromycin MIC, 0.047 μg/ml). The phenotypic change was followed by a dramatic increase in transcription of the transposase gene and the concomitant loss of an approximately 2-kb DNA fragment carrying the erm(B) gene from the 3,165-bp erm transposon. In cells that were not subjected to heat shock, transcription of the transposase gene was not detectable. The upstream sequence of the transposase gene did not show any homology to known heat shock promoters in the gene data bank. Significant homology (>99%) was observed between the erythromycin resistance-encoding gene from L. crispatus CHCC3692 and the erm(B) genes from other gram-positive bacteria, such as Streptococcus agalactiae, Streptococcus pyogenes, Enterococcus faecium, and Lactobacillus reuteri, which strongly indicates a common origin of the erm(B) gene for these species. The transposed DNA element was not translocated to other parts of the genome of CHCC3692, as determining by Southern blotting, PCR analysis, and DNA sequencing. No other major aberrations were observed, as judged by colony morphology, growth performance of the strain, and pulsed-field gel electrophoresis. These observations suggest that heat shock treatment could be used as a tool for the removal of unwanted antibiotic resistance genes harbored in transposons flanked by insertion sequence elements or transposases in lactic acid bacteria used for animal and human food production.

Gram-positive bacteria are reported to carry multiple copies of mobile DNA sequences, called insertion sequences (IS) and transposons (14, 20, 28, 33, 34, 39). The transposable elements are occasionally activated, although our understanding of the molecular processes that trigger their movement is rather limited. Furthermore, they are capable of promoting their own transfer to other sites in the genome or may be translocated out of the cell and eventually end up in other bacteria (40). Translocation is catalyzed by specific recombinant enzymes, known as transposases, which are often encoded in the DNA of the element itself. In addition, these mobile elements often contain genes coding for enzymes conferring resistance to a wide range of antibiotics, such as aminoglycosides (7), tetracycline (25), erythromycin (21, 26, 32, 46), and chloramphenicol (22). Cells harboring an erythromycin resistance gene (erm) can overcome inhibition of their protein-synthesizing machinery (50S inhibitors) in the presence of erythromycin by posttranscriptional modification of the 23S rRNA by adenine-N₆ methyltransferase (36). Lactobacillus crispatus CHCC3692 was previously found to be resistant to erythromycin at concentrations of >250 μg/ml, as demonstrated by Etest susceptibility screening (9). The genetic determinant for the erythromycin resistance gene is located on the chromosome and is therefore probably associated with a nonconjugative transposon (36). Sequences similar to the erm gene studied here have previously been detected in the gram-positive bacteria streptococci, staphylococci, and enterobacteria (3). The fact that transposons are movable elements implies that they can be excised from the chromosome and that the antibiotic resistance can be either lost or transferred to other parts of the genome or, indeed, to other organisms. Unlike other IS elements or transposons which are bounded by terminal inverted repeats, the transposon described here, Tn3692, does not contain this type of flanking sequences. Instead, two homologous (94%) direct repeats of 50 nucleotides (nt) preceding the two transposase-encoding genes (tps) surround the movable element. This structure might imply that the insertion or excision event catalyzed by the encoded transposase occurs by a RecA-like mediated recombination event such as that in Escherichia coli between the directly repeated IS elements (13). This event is supported by the lack of a small duplication of the target DNA that is reported for other IS elements or transposons which is a result of a staggered break from insertion in the host DNA. The establishment of an erythromycin-sensitive variant (Erm^s) of L. crispatus CHCC3692 by heat shock treatment and the characterization of the erm transposable region itself, along with its integration sites and flanking sequences, are reported here.

MATERIALS AND METHODS

Strain and culture conditions.

L. crispatus CHCC3692 was obtained from pars oesophageae of suckling pigs (Karlebo, Denmark; 1987) and deposited in the culture collection of C. Hansen A/S (accession no. CHCC3692). The strain, which is resistant to erythromycin (Erm^r), was grown routinely at 37°C in Difco-MRS broth (11) as well as on MRS agar (1.5% agar) containing 10 μg of erythromycin per ml. Screening for Erm^s isolates after the ampicillin enrichment procedure (30, 35) was carried out by plating aliquots of cells on MRS agar in an appropriate dilution to give approximately 100 to 150 colonies per plate after incubation at 37°C for 20 h. Erm^s colonies were identified by replica plating on MRS agar without additives and MRS agar containing 10 μg of erythromycin per ml, on which the Erm^s isolates were unable to grow. Finally, the Erm^s colonies were cultured in MRS broth. Total genomic DNA and RNA were isolated from both the Erm^s clones and the Erm^r strain for intensive characterization. Cloning of partial libraries of endonuclease size-fractionated lactobacillus DNA (42) and of genomic and PCR-amplified DNA fragments was carried out in pUC19 (29, 49). E. coli strain DH5α (19) was used to propagate pUC19 by the general calcium chloride procedure (37).

Enzymes and chemicals.

Restriction endonucleases, RNase A, RNase T1, and the nick translation kit were all from Roche (Mannheim, Germany). Taq-2000 DNA polymerase was from Stratagene (La Jolla, Calif.). All chemicals were reagent grade unless otherwise specified.

Heat shock and screening for erythromycin-sensitive variants of CHCC3692.

Exponentially growing cells (optical density [OD] at 600 nm, 0.2) in MRS broth (10 ml) without the addition of erythromycin were subjected to heat shock at 60°C for different times (ranging from 5 to 30 min). Prior to heating, the cells were placed on ice for at least 10 min. The individual cell samples (vials of 10 ml) were then incubated at 60°C for specified times, after which they were again transferred to ice for 10 min. The viability of the cells immediately after the different heat shock periods was determined by cell plating on MRS agar. The heat-treated cultures, including the reference strain, were then allowed to grow in the same medium for a period of 16 h at 37°C, at which point aliquots of the heat- and non-heat-shocked cells (the reference group) were reinoculated (1% [vol/vol]) into fresh MRS broth and allowed to grow for an additional 16 h. Isolation of temperature-induced Erm^s cells, including the spontaneous Erm^s isolates from the reference group, was performed by the ampicillin enrichment procedure (30, 35).

DNA extraction and PCR amplification.

Genomic DNA was prepared from CHCC3692 by the method of Gevers et al. (16). PCR was used to screen for and verify mutant transposon rearrangements and to generate DNA fragments employed in cloning and DNA sequencing. Oligonucleotide primers were synthesized by TAG Copenhagen (Copenhagen, Denmark). DNA amplifications were performed on a Trio-Thermoblock 48 PCR cycler from Biometra (Göttingen, Germany) with Taq-2000 polymerase from Stratagene according to the protocol of Innis and Gelfand (23), with some minor modifications. All reactions were preheated at 94°C for 3 min prior to thermocycling. Denaturation was at 94°C for 30 s, primer annealing was at 55°C for 30 s, and primer extension was at 72°C for 60 s, repeated 34 times, followed by an extension period at 72°C for 5 min. The PCR products were visualized by UV illumination of agarose gel electrophoresis gels stained with ethidium bromide. Extraction of PCR fragments from the separating gels was performed with the QIAquick gel extraction kit from Qiagen (Hilden, Germany). The primers used in this study for both DNA amplification and sequencing are listed in Table 1.

TABLE 1.

Oligonucleotides used for PCR analyses and DNA sequencing of the erythromycin resistance-encoding transposon Tn3692 of L. crispatus CHCC3692

Primer	Sequence (5′ to 3′)	Nucleotide positionsa
ermTnR4	CCAGAAGACTACCTGACATA	64-84
ermTnR5	CCGCGAATAGCTAGTTC	475-491
ermTnR6	GAAGGAGGGATTCGTCATGTTGG	889-901
ermTnR3	CGAAATTGGAACAGGTAAAGGGC	1204-1226
ermTnD2	GGTACGCTTGTAGAATCCTTC	1484-1504
ermTnD1	GTAACGTGTAACTTTCC	1866-1882
LA28ermend3	GATATACTACTGACAGCTTCC	1908-1928
ermTnD5	CGGAATAAGTCCTTAGTCGC	2231-2250
Tp28R	GGACTTATTCCGATATTCTATCGGAG	2239-2264
LA28erm4	CCGATATTCTATCGGAG	2248-2264
LA28endre3	CGCTAATGAAGATCTGCC	3041-3058
Tp28D	GGACGACGAATCCATGAGCAAG	3270-3291
ermTn25D2	GACATAGTAAAACCCTTG	3424-3441
ermTn25R	GTAGTGACCCTCGAAATTCGG	3452-3472

^aNumbering is based on the DNA sequence deposited in GenBank.

Southern blot analysis.

Restriction enzyme-digested DNA was subjected to electrophoresis (1.4% agarose) and transferred to GeneScreen Plus membranes (NEN Life Science Products, Inc., Boston, Mass.) by alkaline transfer (10× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate], pH 7.0) overnight. Prior to DNA transfer, the gel was soaked in 0.25 M HCl for 10 min with gentle shaking. DNA was denatured (0.5 M NaOH, 1.5 M NaCl) for 30 min and neutralized (0.5 M Tris-HCl, 3.0 M NaCl, pH 7.4) for an additional 30 min with gentle shaking. The DNA probes were radiolabeled with [α-³²P]dATP by random primed labeling, using a nick translation kit from Roche. Hybridization was performed at 42°C in formamide and dextran sulfate according to the manufacturer's instructions (NEN Life Science Products, Inc.). Membranes were prehybridized in this buffer system for at least 2 h, after which the denatured probe was added and hybridization was allowed to proceed for 16 to 17 h. Filters were then rinsed for 15 min in 2× SSC-1% sodium dodecyl sulfate at 42°C and rinsed for a further 15 min at 54°C. The dry membranes were exposed to Kodak X-Omat AR films at room temperature using an intensifying screen. Films were developed after 16 h of exposure.

DNA sequencing.

Sequences were obtained by using restriction enzyme-generated cloning sites and the universal M13 primer (44). Sequencing was done by the dideoxynucleotide chain termination method (38) and was performed on an ABI-310 automatic sequencer from Applied Biosystems (Foster City, Calif.). Big dye chain terminators and all reagents were supplied by Applied Biosystems. Cycle sequencing and handling of the labeled oligonucleotides were done according to the recommendations of the manufacturer.

Extraction of RNA and transcriptional analyses.

RNA was prepared from heat-shocked and non-heat-shocked CHCC3692 by use of the RNeasy mini protocol for isolation of total RNA from gram-positive bacteria (200 ml; OD₆₀₀ = 0.2) according to the manufacturer's (Qiagen) instructions, except for some minor modifications. Lysis of the cells was performed in TEX buffer (20 mM Tris-HCl, 2 mM EDTA, 1.2% Triton X-100, pH 8.0), with lysozyme (10 mg/ml), and incubated at 37°C for 60 min. To remove any residual traces of DNA from the RNA extraction, a DNase step was introduced using an on-column DNase digestion method (Qiagen). Equal quantities (approximately 1 μg; OD₂₆₀/OD₂₈₀ = 2.0) of total RNA, as determined by A₂₆₀ measurements, were obtained from the individual extraction reactions. Transcriptional analyses were performed by reverse transcription-PCR (RT-PCR) using the SuperScript one-step RT-PCR with Platinum Taq kit from Invitrogen (Carlsbad, Calif.). cDNA synthesis followed by PCR amplification was performed on a RoboCycler gradient 96 instrument from Stratagene. Transposase cDNA synthesis was performed with the primers Tp28D and Tp28R (Table 1) spanning most of the transposase open reading frame (ORF). One cycle was performed at 50°C for 30 min (0.25 μg of RNA), followed by predenaturation at 94°C for 2 min. PCR amplification conditions were as follows: 40 cycles were performed with denaturation at 94°C for 30 s, primer annealing at 50°C for 30 s, and primer extension at 70°C for 60 s. A final extension step was done at 70°C for 10 min. Topoisomerase IV cDNA synthesis was chosen as an internal standard relative to the transposase. A set of degenerate primers (parE-f1, 5′-CARTTYGARGGXCARACXAARG-3′; parE-r1, 5′-CCRTCXGTRTCXGCRTCXGTCAT-3′) was designed to amplify a 500-bp region of the topoisomerase IV gene (10). The RT-PCR conditions were the same as those for the transposase, with the modification that cDNA synthesis and primer annealing were performed at 42°C. The PCR products were visualized by UV illumination of agarose gel electrophoresis gels stained with ethidium bromide.

DNA fingerprinting analysis.

Pulsed-field gel electrophoresis of genomic CspI- or SmaI-digested DNA was performed according to standard methods (L. Hung and R. Bandziulis, Promega Notes 24:1-2, 1990, Promega, Madison, Wis.).

Nucleotide accession number.

The sequence of the complete 3,165-bp erm transposon, Tn3692, with 5′- and 3′-flanking sequences of 190 and 118 nt, respectively, is deposited in GenBank under accession no. AY262353.

RESULTS

Removal of the erythromycin resistance determinant from CHCC3692.

Exponentially growing cells were subjected to heat shock for different times at 60°C. The viability of the cells after the different heat shock periods was determined by cell plating, and the results are presented in Table 2. Heating at 60°C over a period of 5 min killed approximately 90% of the cells compared to the control group, which was not subjected to the heat shock treatment. Cells were subjected to heat shock at 60°C for 10, 15, and 20 min and were subsequently screened for sensitivity to erythromycin, using ampicillin enrichment for the Erm^s phenotype as reported previously (30, 35). The results from the replica screening of Erm^s isolates are shown in Table 3. Eighteen selected Erm^s derivatives of CHCC3692 grew like the mother strain in MRS broth but were incapable of growing in the presence of erythromycin at a concentration of 1 μg/ml. The heat shock treatment substantially increased the frequency of cell sensitivity to erythromycin, almost 40-fold, after exposure to 60°C for 10 min. The lower frequency of sensitivity after heat shock for 15 or 20 min is probably a sign of random fluctuation in smaller cell populations due to the extra stress imposed on the strain under these conditions.

TABLE 2.

Viability of L. crispatus CHCC3692 after heat shock for various times at 60°C

Time (min) at 60°C	Cell no./ml
0a	1.5 × 106
5	1.6 × 105
10	8.5 × 104
15	2.5 × 102
20	0.9 × 102
25	0
30	0

^aReference group.

TABLE 3.

L. crispatus CHCC3692 erythromycin-sensitive isolates recovered after heat shock at 60°C

Time (min) of heat shock	Cell no. (104)/mlb	No. of Erms isolates	Frequency (%)
0a	288	3	1.0c
10	64	24	38.0
15	60	18	30.0
20	47	9	19.0

^aReference group.

^bCalculated for living cells after ampicillin enrichment.

^cSpontaneous mutation rate relative to heat shock-induced mutations.

Molecular characterization of the erythromycin-sensitive derivatives of CHCC3692.

The Erm^s variants of CHCC3692 were characterized by PCR analyses and Southern blotting to test for the possible removal or deletion of the erm transposable element in these mutants. PCR was performed with sense primer LA28ermend3, derived from a sequence (erm) upstream of the termination codon of the tps gene, and antisense primer LA28endre3, derived from the ORF of tps (Table 1). As expected, a fragment of approximately 1,150 bp was amplified when we used template DNA from the wild-type strain, CHCC3692 (Fig. 1, lane 3). However, no amplified DNA fragments were observed for any of the Erm^s variants, suggesting excision and loss of the transposase and the erm genes from these isolates (Fig. 1, lanes 5 to 7). An internal PCR positive control (a 16S ribosomal gene fragment) specific for Lactobacillus was run in parallel and gave an expected amplified fragment of approximately 800 bp in all of the reactions (Fig. 1, lanes 9 to 12). Southern blots of EcoRI- and PstI-digested genomic DNA from CHCC3692 and three erythromycin-sensitive isolates (a spontaneous one from the reference group and two isolates from heat shock at 10 and 20 min), with the plasmid pLEB22, harboring the erm(B) gene (M. Skaugen, personal communication), as a hybridization probe, resulted in the identification of a 2.0-kb EcoRI fragment and a 6.0-kb PstI fragment for CHCC3692. However, no hybridization to this probe was observed for the EcoRI- and PstI-digested DNA from the Erm^s isolates, providing further evidence of the loss of the erythromycin resistance genetic determinant from these isolates (results not shown).

FIG. 1.

Agarose gel electrophoresis of PCR-amplified DNA fragments from L. crispatus CHCC3692. Lanes 3 and 5 to 7, DNA amplified with transposase primers and templates derived from wild-type (Erm^r) and erythromycin-sensitive (Erm^s) isolates, respectively; lanes 9 and 10 to 12, DNA amplified with specific 16S Lactobacillus species primers and templates isolated from the Erm^r and Erm^s isolates, respectively. Lanes 5 and 10, template DNA was isolated from a spontaneous Erm^s isolate; lanes 6 and 11, template DNA was derived from cells that were heat shocked at 60°C for 10 min; lanes 7 and 12, template DNA was isolated from cells that were heat shocked at 60°C for 20 min; lane 1, reference standard marker (λ DNA digested with EcoRI and HindIII).

Characterization of the transposase-encoding genes in CHCC3692.

An almost complete transposase-encoding gene was previously isolated from CHCC3692 by PCR and characterized by DNA sequencing (unpublished data). A linkage to the erm gene was present approximately 350 nt upstream from the transposase stop codon. In order to investigate the excision point(s) of the transposase gene relative to the erm gene, Southern blotting of EcoRI-digested genomic DNA from strain CHCC3692 (Erm^r) was performed. An 800-bp probe (generated by using primers LA28erm4 and LA28endre3 [Table 1]) corresponding to two-thirds of the transposase-encoding gene resulted in the hybridization of three fragments, of 2.0, 2.2, and 5.0 kb (Fig. 2, lane 2). Genomic DNA digested with EcoRI and PstI resulted in the identification of three fragments, of 1.5, 2.0, and 2.5 kb (Fig. 2, lane 4). Restriction with PstI alone gave a hybridizing fragment of 6.0 kb (Fig. 2, lane 6). The hybridization pattern obtained from Southern blots of the Erm^s isolates after 0 (a spontaneous mutation), 10, and 20 min of heat shock at 60°C was almost identical to the pattern of CHCC3692, except that the 2.0-kb band was missing for the EcoRI digestion and the double digestion with EcoRI-PstI (Fig. 2, lanes 1 and 3). The hybridizing band of the PstI-digested DNA was 4.0 kb instead of the 6.0 kb obtained with CHCC3692 (Fig. 2, lane 5; Fig. 3). Judging from Southern blotting with probes derived from the transposase gene, it is clear that an EcoRI fragment of 2.0 kb, which probably harbors the entire coding region of the erm gene, was deleted from the Erm^s strain. Verification of this was obtained by isolating and sequencing the 2.0-kb fragment from a partial library of EcoRI-restricted, size-fractionated DNA of CHCC3692. Upstream it contained the coding sequence for 17 amino acids (aa) of a truncated part (aa 207 to 223) of the C-terminal transposase molecule (encoded by the antisense strand) followed by a noncoding region of approximately 350 nt. Downstream of this region, an ORF of 738 nt contained the coding region (encoded by the sense strand) for the erm gene (molecular mass of the product, ∼27 kDa). Immediately following the translational stop codon of this gene, a small ORF of 129 nt was observed. The DNA sequence further downstream (approximately 250 nt) was followed by the C-terminal antisense transposase-encoding gene (aa 224 to 374) (Fig. 3 and sequence deposited in GenBank). The two other EcoRI- and EcoRI-PstI fragments, of 2.2 and 2.5 kb, respectively, also carrying a part(s) of the tps gene(s) (Fig. 2 and 3), were cloned from partial and size-fractionated genomic DNA of CHCC3692 and characterized by DNA sequencing. The 2.2-kb EcoRI fragment from CHCC3692 (Fig. 2, lanes 1 and 2) contained an (upstream) apparently noncoding region of 1.6 kb followed by a coding sequence of 151 aa (positions 224 to 374) of another C-terminal transposase molecule encoded by the antisense DNA strand and was identical to the sequence depicted for the 2.0-kb EcoRI fragment (Fig. 3). A PstI restriction site was identified approximately 1.5 kb upstream of the EcoRI site in the transposase gene. This PstI-EcoRI DNA fragment is most likely identical to the ones of similar size in the Southern blots (Fig. 2, lanes 3 and 4). The EcoRI-PstI fragment of 2.5 kb (Fig. 2, lanes 3 and 4) was isolated and cloned, as it was believed to harbor the remaining (amino-terminal) coding sequence of the tps gene from the 2.0-kb EcoRI fragment (Fig. 3). It contained, as expected, the N-terminal part of the transposase gene of 223 aa (positions 1 to 223) and a downstream region of 1.83 kb from the tps start codon with no obvious coding capacity. In order to obtain a consensus sequence for the three subcloned fragments, PCR analyses were performed to obtain overlapping sequences of the 2.0-kb EcoRI fragment, the 2.2-kb EcoRI fragment, and the 2.5-kb EcoRI-PstI fragment. Template DNAs from Erm^r and Erm^s strains were used with a primer derived from the upstream sequence of the transposase gene, ermTnR4, and a primer derived from a sequence within the transposase gene, LA28endre3 (Table 1 and Fig. 3). An expected fragment of approximately 1 kb was amplified from the erythromycin-sensitive templates. A fragment of the same size was also observed for the CHCC3692 template, in addition to another fragment of approximately 3.0 kb (Fig. 4). The 1.0-kb fragment is most likely a result of generated erythromycin-sensitive templates in the wild-type DNA pool. From these results, it is clear that a fragment of 2.0 kb lies between the primers ermTnR4 and LA28endre3 in the erythromycin-resistant strain. For verification purposes, the 3.0-kb PCR fragment was subjected to direct sequencing after extraction from the gel. Downstream from the ermTnR4 primer, it contained the C-terminal coding region (506 bp) for a truncated transposase of 168 aa. This was followed by a region of approximately 355 bp to the start codon of the erm gene. Immediately following the translational stop codon of the erm gene, a small ORF of 129 nt was located. The sequence further downstream was followed by the transposase-encoding gene and was identical to the one in the 2.0-kb EcoRI fragment previously sequenced. The full sequence of the erm transposable element with flanking sequences was deposited in GenBank as described above.

FIG. 2.

Detection of sequences homologous to the transposase gene in the CHCC3692 Erm^s isolate (lanes 1, 3, and 5) and in the wild-type strain (lanes 2, 4, and 6). The Southern blot was probed with a ³²P-labeled 800-bp fragment of transposase. Lanes 1 and 2, genomic DNA digested with EcoRI; lanes 3 and 4, DNA digested with EcoRI and PstI; lanes 5 and 6, DNA digested with PstI. Size markers are shown on the left and are based on an EcoRI/HindIII digest of the λ phage.

FIG. 3.

Schematic outline of the erm transposon, Tn3692, in L. crispatus CHCC3692. The upper part depicts the relative map of the transposon in the wild type (Erm^r), with restriction sites for the endonucleases EcoRI and PstI. The lower part depicts the structure after removal of the erm transposable element. tps, transposase-encoding gene, tps′, truncated transposase gene; erm(B), erythromycin resistance-encoding gene. The gray shaded box (130 nt) encodes an unknown protein. The arrows protruding from the tps genes indicate the direct repeats of the 50-nt fragment (not drawn to scale). The chromosomal structure in Erm^s isolates is most likely a result of intragenic recombination between the direct repeats of the C-terminal transposase-encoding regions. The arrows indicate the direction of transcription. See text for further information.

FIG. 4.

Agarose gel electrophoresis of PCR-amplified DNA fragments from L. crispatus CHCC3692. Lane 2, amplified DNA from the wild type (Erm^r); lanes 3 to 5, amplified DNA fragments from Erm^s isolates undergoing 0, 10, and 20 min of heat shock at 60°C, respectively; lane 1, reference standard marker based on an EcoRI/HindIII digest of phage λ.

The response of the transposase gene to heat shock treatment.

To investigate the expression or induction of the transposase gene during heat shock, exponentially growing cells were exposed to heat shock for 10 min at 60°C. Six cultures, two Erm^r cultures exposed to erythromycin (15 μg/ml) and two cultures without erythromycin, plus two Erm^s isolates, were set up. Total RNA was isolated from each of the cultures that were subjected to heat shock and from the reference group, which was not subjected to heat shock. mRNA was analyzed by RT-PCR using primers Tp28D and Tp28R (Table 1), designed to amplify approximately 95% of the transposase ORF. A unique approximately 1.1-kb band corresponding to the transposase transcript was detected in isolates subjected to heat shock, irrespective of whether erythromycin was used in the culture medium. No transcripts were detected in cell isolates without heat shock (Fig. 5). This result indicates that the transposase promoter is strongly induced under stress conditions and might possibly be involved in transposition of the erm element. Interestingly, tps transcription seems to be induced at the same level in both the Erm^r and Erm^s isolates.

FIG. 5.

Effect of heat shock (60°C, 5 min) of L. crispatus on transcription of the transposase gene by RT-PCR. (A) Agarose gel electrophoresis of PCR-amplified cDNA. Lanes 2, 4, and 6, cDNAs amplified from heat-shocked cultures with transposase primers; lanes 3, 5, and 7, cDNA amplification of non-heat-shocked cultures. Lanes 2 and 3, Erm^r cells grown in the presence of erythromycin (15 μg/ml); lanes 4 and 5, Erm^r cells grown without erythromycin; lanes 6 and 7, Erm^s cells; lane 8, PCR fragment amplified from Erm^r cells with the same transposase primers and template DNA. (B) PCR-amplified topoisomerase IV cDNA, used to standardize the signal levels for panel A. Lane 8, PCR fragment amplified with topoisomerase IV primers and template DNA isolated from Erm^r cells; lane 1, standard reference marker based on a DNA ladder (Promega).

DISCUSSION

A mild mutagenic treatment was used to trigger excision and movement of the erythromycin resistance determinant from the genome of L. crispatus CHCC3692. This excluded treatment of the cells with chemical mutagens or irradiation with UV light, which might have resulted in retardation of many of the treated cells, even at a low dose of the mutagen. In order to eliminate the introduction of multiple mutations into the strain, heat shock treatment at 60°C was used to induce stress in the cells, with the resultant concomitant excision and loss of the erm transposable element (3,165 bp) from the chromosome of CHCC3692. The frequency was recorded to be 40-fold higher than the spontaneous excision rate recorded for a reference group which was not subjected to heat shock. The excised fragment (1,987 bp) contained a truncated ORF of 506 bp for a transposase that was approximately one-half the size of the wild-type transposase, which is not lost during transposition, and an ORF coding for erythromycin resistance (Fig. 3). The transposase gene linked to the erm transposon of CHCC3692 reported here showed an overall similarity of >70% to other transposase genes listed in the EMBL data bank. A positional identity of 66% (in a 674-nt overlap) was observed to a transposase gene from Lactobacillus delbrueckii subsp. lactis ISL6 (27) and an identity of 63% was observed in an overlap of 685 nt to a transposase from Streptococcus mutans UA15 (1). Comparison with the sequence upstream of the tps initiation codon did not reveal any homology to known heat shock promoters deposited in the gene data bank. The only significant homology (70% identity) was observed in a 72-nt overlap with the promoter from L. delbrueckii subsp. lactis ISL6 (15). Significant homology, exhibiting an average of approximately 99.4% positional identity, was observed between the erm gene of L. crispatus CHCC3692 and the erm(B) genes of the following species: Streptococcus agalactiae resolvase gene (4), Streptococcus lentus erm(B) ORF (45), Streptococcus pyogenes plasmid pBT233 (5), Streptococcus pneumoniae Tn1545 (43), Staphylococcus aureus Tn551 (47), Enterococcus faecium (24), Lactobacillus reuteri plasmid pTE80 (C. F. Lin and T. C. Chung, unpublished data), and Lactobacillus fermentum (12). All sequence data information were derived from GenBank/EMBL.

The lactobacilli can be divided into three evolutionarily related groups based on polymorphisms of their 16S rRNA genes: the L. delbrueckii group, the Lactobacillus casei-Pediococcus group, and the Leuconostoc group (6, 17, 18, 48). The erm gene, now designated erm(B) for L. crispatus, is the first one reported for the L. delbrueckii group, to which CHCC3692 belongs. The other strains harboring the erm(B) genes listed above belong to the L. casei-Pediococcus group, thus adding further evidence for wide horizontal transfer of this gene (M. Danielsen, unpublished data).

Downstream of the erm(B) gene in strain L. crispatus CHCC3692, a small ORF encoding a putative protein of 43 aa of unknown function was observed. This particular ORF often follows erythromycin resistance-encoding genes. The intervening sequence (250 nt) between the small ORF and the translational stop codon of the downstream transposase gene harbors several inverted repeats which show substantial homology (98%) to inverted repeats found in many other bacteria (41). This strongly indicates that this fragment is part of a transposable element or the resistance gene. The two transposase genes harbored in CHCC3692 have 100% identity in their nucleotide sequences. Two direct repeats of 50 nt (94% homology) flank the transposase carboxy-terminal encoding parts (Fig. 3). The upstream repeat of the truncated tps gene, however, has been recorded for all of the erythromycin-sensitive isolates in this study, which suggests a mechanism for the excision of the erm transposable element. The specific trigger for the excision of this element is at present unknown, although stress imposed on the strain, such as heat shock, clearly facilitates the excision event. However, the role of the strongly transcribed transposase gene, as demonstrated here, is likely to be essential; otherwise, heat shock might not have been expected to increase the excision rate. It should be emphasized, though, that an active transposase is not a prerequisite for excision, as demonstrated in this study. Heat shock in the absence of a transposase might also be catalyzed by DNA repair enzymes facilitating intragenic recombination between the two transposase genes. Whether the transposase is directly involved in the excision event during heat shock in strain CHCC3692 could be tested for in a strain with a mutation in the intact tps gene.

Unlike several retro-transposons, many transposable elements seem to never exist free of the host genome. Transposases that catalyze their movement act on the DNA of the element while they are still integrated in the host genome (2). In L. crispatus CHCC3692, the mechanism is suggested to involve binding of the transposase to a shorter sequence of 50 nt downstream of the C-terminal transposase-encoding sequences which is repeated in the same orientation flanking the element to be excised. The protein complex holding the homologous repeats together catalyzes the subsequent intragenic recombination between the tps genes in a conservative site-specific recombination event. This results in movement or loss of a DNA fragment of 1,987 bp harboring the truncated tps gene, including the erm(B) gene and its flanking sequences. The 50-bp repeat downstream of the truncated tps gene is always present in the erythromycin-sensitive strains, thus supporting this mode of excision. In general, the mechanism of DNA transposition involves only the breaking and rejoining of DNA, with the two ends of the element being inserted into a staggered break elsewhere in the chromosome (8, 31). This way of excision can be ruled out here, since this would result in a small duplication on each side of the transposable DNA fragment which was not observed in this investigation.

Figure 3 illustrates a map of the erm transposon, Tn3692, in the wild-type strain of L. crispatus, CHCC3692, and the organization of the locus upon excision of the erm transposable element. The Erm^s strain harbors an ORF of 1,128 bp encoding a (complete) transposase molecule (molecular mass, approximately 41 kDa) which is flanked downstream by a sequence of 50 bp. An almost identical sequence (repeat), except for three nucleotide substitutions, also terminates the excised truncated transposase gene fragment. The structure strongly implies that excision of the transposable element from the genome is catalyzed by the transposase and brought about by intragenic recombination between the homologous parts of the transposase genes. Removal of the erm element (1,987 bp) has no impact on the phenotype except for sensitivity to erythromycin, which is clearly demonstrated by the inhibited growth of the Erm^s isolates in the presence of this antibiotic.

In Fig. 6, the growth of the two phenotypes, Erm^r and Erm^s, is compared under similar conditions in MRS broth with or without erythromycin. The addition of erythromycin (20 μg/ml) delays the growth rate of the Erm^r phenotype, suggesting that the erm(B) gene is induced in the presence of erythromycin and might be transcriptionally regulated.

FIG. 6.

Growth curves of CHCC3692 Erm^r and CHCC3692 Erm^s in MRS broth in the presence (+) (20 μg/ml) or absence (−) of erythromycin (erm). Zero indicates the start of growth (1%) from an overnight culture of CHCC3692 in MRS broth.

Pulsed-field gel electrophoresis did not reveal any major rearrangement of the SmaI- or CspI-digested chromosomal pattern upon excision of the erm element. In fact, no difference at all was observed in the restriction enzyme-digested chromosomal DNA pattern for Erm^r and Erm^s cells (data not shown). Transposition of the erm element to other parts of the genome could also be ruled out based on Southern blot and PCR analyses (Fig. 1 and 2). One can speculate whether the tps gene in the erythromycin-sensitive cells acts as a target sequence and is responsible for (re)insertion of the erm element into CHCC3692 or other strains harboring a stress-induced transposase gene.