INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Enterococcus faecalis is an opportunistic human pathogen commonly harboring mobile genetic elements conferring resistant to multiple antibiotics including vancomycin (8). A significant percentage of clinical isolates are hemolytic, a trait commonly encoded on plasmids resembling the hemolysin/bacteriocin (cytolysin)-encoding element pAD1 (28, 31). The pAD1 hemolysin has been shown to contribute to virulence in animal models (7, 30, 33) and has been found to be a lantibiotic consisting of two components (27).

pAD1 (14) is highly conjugative and encodes a response to the octapeptide sex pheromone cAD1 (37) secreted by plasmid-free recipients. The pheromone-induced mating response results in synthesis of a plasmid-encoded surface protein (25, 26) designated aggregation substance (Asa1). The asa1-encoded protein plays an important role in initiating contact with potential recipients upon random collision. Plasmid-free strains of E. faecalis actually secrete multiple sex pheromones, each specific for a particular group of conjugative plasmids (17, 18), and production of only the related sex pheromone gets shut down upon plasmid entry. For recent reviews of sex of enterococcal sex pheromone, see references 9 to 11, 20, and 49.

When a given plasmid is acquired by the recipient, the corresponding pheromone becomes shut down and/or masked by the production of a plasmid-encoded peptide that is secreted and competitively inhibits any tendency of the cells to be induced by endogenously secreted pheromone (29). In the case of pAD1, the inhibitor is called iAD1 and represents the last eight residues (carboxyl terminal) of a plasmid-encoded 22-amino-acid precursor resembling a lone signal sequence (11, 13, 36). Reduction of cAD1 involves the plasmid-encoded TraB, an apparent membrane protein (1a). Interestingly, the degree of shutdown can differ significantly depending on the bacterial host. For example, in strain OG1X/pAD1, pheromone (cAD1) is expressed at 2% of that produced by plasmid free OG1X, whereas in the nonisogenic FA2-2/pAD1, expression of cAD1 is reduced by only 50% (38). In both cases, the production of iAD1 masks all cAD1 activity present in culture supernatants. It is noteworthy that expression of cAD1 is significantly (about eightfold) greater in the case of plasmid-free OG1X compared to FA2-2 (48). In both hosts, the production of cAD1 is increased about 8- to 16-fold under anaerobic conditions, a behavior that does not occur in the case of other pheromones tested (e.g., cPD1, cCF10, and cAM373) (reference 48 and unpublished observations).

As illustrated in Fig. 1, regulation of the pAD1 mating response involves the binding of imported cAD1 to the plasmid-encoded, negatively regulating TraA protein, causing the latter to release its binding to pAD1 DNA and facilitating expression of the positive regulator TraE1 (23, 24, 40, 44, 47). The mechanism by which TraA regulates TraE1 expression involves the control of readthrough of a pair of transcription terminators, t1 and t2, just upstream of traE1 (24, 41, 44); a small RNA transcript, mD, encoded upstream of t1 also negatively regulates transcriptional readthrough (3, 4).

View larger version (11K): [in this window] [in a new window]   FIG. 1.   Map of the pAD1 regulatory region with features relating to control of the pheromone response. Regulation is geared to controlling transcriptional readthrough of the two transcription terminator sites t1 and t2 from the iad promoter. The pheromone (cAD1) is imported into the cell via a host-encoded uptake system, a process aided by the plasmid-encoded surface protein TraC. The negatively regulating (neg. reg.) TraA binds directly to cAD1, which results in an upregulation of transcription from the iad promoter, some of which passes through the terminators to ultimately generate TraE1. The transcript mD is present at high levels in the uninduced state and plays a significant role in preventing transcriptional readthrough of t1; it becomes greatly reduced upon induction. The traB product is involved in shutdown of endogenous cAD1 in plasmid-containing cells. traC encodes a surface protein that binds to exogenous pheromone (43). Arrows below the map indicate relative degrees of transcription in various regions; horizontal arrows above the map represent orientations of the indicated determinants.

Previous attempts to clone and identify the chromosome-borne pheromone determinants have been unsuccessful due to extensive probe degeneracy and the absence of a good screening assays. Using a new approach originally designed to identify the cAD1 determinant or determinants important to pheromone biosynthesis, we have found a gene necessary for the production of the pheromone in strain OG1X. While the determinant does not correspond to cAD1 per se, it appears to be critical for the synthesis of normal levels of the pheromone. Here we describe the characterization of this determinant and show that it can influence the production of not only cAD1 but other pheromones as well.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, reagents, and assays. The bacterial strains and plasmids used in this study are listed in Table 1. Luria-Bertani broth (35) was used for the growth of Escherichia coli. E. faecalis strains were grown in Todd-Hewitt broth (Difco Laboratories, Detroit, Mich.) or brain heart infusion (Difco). Solid media used was Todd-Hewitt broth with 1.5% agar. Antibiotics were used at the following concentrations: chloramphenicol, 20 µg/ml; ampicillin, 100 µg/ml; erythromycin, 20 µg/ml for E. faecalis and 200 µg/ml for E. coli; streptomycin, 1 mg/ml; rifampin, 25 µg/ml; and fusidic acid, 25 µg/ml. X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside (GIBCO BRL, Inc., Gaithersburg, Md.) was used at a concentration of 100 µg/ml. Synthetic cAD1 was prepared at the University of Michigan peptide synthesis core facility. Pheromone response (clumping) assays including preparation of culture filtrates were as previously described (18). Restriction enzymes were purchased from GIBCO BRL, and reactions were carried out under the conditions recommended.

General DNA and RNA techniques. Routine screening of plasmid DNA was carried out by using a small-scale alkaline lysis procedure described previously (45). Plasmid DNA from E. coli was prepared with a Plasmid Midi kit (Qiagen, Inc., Chatsworth, Calif.) as recommended by the manufacturer. Plasmid isolated from E. faecalis made use of equilibrium centrifugation in CsCl-ethidium bromide buoyant density gradients as described elsewhere (16). Plasmid DNA was analyzed by digestion with restriction enzymes and electrophoretic separation of fragments in 0.7% agarose gels; for Southern analyses, 1.5% agarose was used. Standard recombinant DNA techniques were used in the construction of plasmid chimeras (2). Introduction of plasmid DNA into cells was by electrotransformation as previously described (22).

View larger version (41K): [in this window] [in a new window]   FIG. 2.   Nucleotide and protein sequences of eep. Segments corresponding to specific primers used for generating PCR products as well as for the primer extension study are underlined, as are regions corresponding to possible Shine-Dalgarno (SD), 10, and 35 sequences. The beginning of a downstream open reading frame corresponding to what is likely for a prolyl-tRNA synthetase is also shown.

Cloning of eep. Cloning involved the use of the E. coli-E. faecalis shuttle vector pAM401 (50). Chromosomal DNA was extracted from a 100-ml culture of plasmid-free E. faecalis OG1X cells grown in brain heart infusion medium: ~100 µg in 100 µl was partially digested with Sau3A and fractionated by centrifugation essentially as described elsewhere (2), using a sucrose-density step gradient consisting of 10, 15, 20, 25, 30, 35, and 40% concentrations of sucrose in 1 M NaCl-5.0 mM EDTA-20 mM Tris-Cl (pH 7.5). Centrifugation utilized an SW41 rotor run at 25,000 rpm for 24 h. The 12-ml gradient was fractionated into 750-µl portions; 10-µl samples from each fraction were examined by gel electrophoresis. A fraction corresponding to a size of 3 to 8 kb was used for ligation with pAM401 that had been digested with BamHI and dephosphorylated with calf intestinal alkaline phosphatase (GIBCO BRL). Ligated DNA was used to transform E. coli DH5 by electroporation, with selection for chloramphenicol resistance. Transformant colonies were pooled; the plasmid DNA mixture was extracted and used to transform E. faecalis OG1X/pAM2011E by electroporation and selection on X-Gal plates containing chloramphenicol and erythromycin. pAM2011E is a miniplasmid derivative of pAD1 containing the pheromone response regulatory region with a Tn917lac insertion just upstream of traE1 (46); the lacZ determinant is readily induced upon exposure of cells to pheromone. Plasmid DNA was isolated from cells deriving from blue colonies and introduced into E. coli DH5, selecting for resistance to chloramphenicol.

Generation of eep mutants. A PCR product corresponding to an internal segment of eep (see above) was generated as described above to obtain pAM3328 and pAM3329 by using the E. coli vector pVA891, which carries an erm determinant that expresses in both E. coli and E. faecalis. The inserts corresponding to the two clones were in opposite orientation in the vector, based on the position of a ClaI site. The plasmids were introduced by electrotransformation into E. faecalis OG1X, where integrants occurring via homologous recombination were selected by using erythromycin. Colonies arose at a frequency of about 10⁸. The strains were designated FA3328 and FA3329. Revertants generated by growth for several (more than six) subcultures in the absence of erythromycin were obtained by screening for sensitivity to the drug. Depending on the original mutant, representative revertant strains were designated FA3328R and FA3329R.

Nucleotide sequence accession number. The GenBank accession no. for the sequence shown in Fig. 2 is AF152237.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Our strategy for cloning E. faecalis determinants relating to pheromone production was to make use of the E. coli-E. faecalis shuttle plasmid pAM401 to shotgun clone fragments of E. faecalis OG1X chromosomal DNA first in E. coli, where transformation efficiencies are high, followed by introduction of plasmid DNA from pooled transformants into E. faecalis OG1X/pAM2011E. (pAM2011E is a miniplasmid derivative of pAD1 with a Tn917lac insertion on the traE1-proximal side of t2; it includes the region shown in the map of Fig. 1 in addition to replication determinants located further to the right.) The enterococci are much less efficiently transformed, especially by nonsupercoiled plasmid DNA; thus, the supercoiled configuration of chimeric DNA coming out of E. coli was believed to facilitate uptake by the enterococci. It was anticipated that E. faecalis OG1X/pAM2011E cells acquiring a multicopy plasmid chimera with a gene(s) that increased pheromone production may result in blue colonies on selective (chloramphenicol and erythromycin) X-Gal plates. The rationale was that additional pheromone would exceed the shutdown capacity of pAM2011E, which contains an intact traB and encodes normal synthesis of iAD1. Using this approach, detailed in the Materials and Methods, we were able to identify four clones which resulted in light blue colonies. Plasmid DNA from each of these derivatives was extracted and introduced back into E. coli DH5, where it was then reisolated, digested with XbaI and SalI (sites for these enzymes flanked the original BamHI site used for cloning), and analyzed by agarose gel electrophoresis. This revealed that in one case the cloned DNA was of a size of 3.0 kb (not shown), whereas the other three were 5 kb or larger. Electing to focus our subsequent studies on the smaller fragment, the corresponding chimera, designated pAM3325, was introduced into plasmid-free E. faecalis OG1X, selecting for chloramphenicol resistance. This gave rise to transformants with an increased production of cAD1 by a factor of 8; culture filtrates exhibited a cAD1 titer of 512, compared to 64 in the case of cells carrying an empty vector (Table 2). It is important to note that neither pAM3325 nor the three larger chimeras gave rise to detectable cAD1 expression in E. coli.

Analysis of cloned DNA. The cloned DNA present in pAM3325 was subcloned to pBluescript II SK, generating pAM3326; sequence analysis was then conducted (see Materials and Methods). The sequence exhibited two open reading frames, designated ORF1 and ORF2; neither contained the cAD1 sequence. Both were accompanied by appropriately positioned ribosome binding sites. ORF1 corresponded to 422 amino acids and is shown in Fig. 2; ORF2 corresponded to 372+ amino acids (the absence of a translational stop site implied that it continued beyond the end of the cloned DNA segment; only the first 23 amino acids are shown in Fig. 2). ORF2 is probably part of a prolyl-tRNA synthetase (ProS), as it had very strong similarity to a number of such proteins in the database. If it is assumed that its actual size is close to the average size of these proteins (about 564 amino acids), then the cloned portion (ORF2) would correspond to a protein missing about 34% from its carboxyl terminus.

Generation of ORF1 mutation. A PCR product representing an internal segment of DNA (Fig. 2) was ligated to the EcoRI site of the plasmid vector pVA891 and introduced into E. coli as described in Materials and Methods. The resulting chimera was then isolated and introduced into E. faecalis OG1X, selecting for transformants resistant to erythromycin. Since the vector was incapable of replicating in E. faecalis, transformants were expected to result from integration into the chromosome via recombination of homologous DNA. This would give rise to a mutation in the related determinant as long as the cells were maintained in the presence of erythromycin. With pAM3329 as a probe, it can be seen in the Southern blot of Fig. 3 that representative transformants resulting from each of two clones, FA3328 and FA3329, contained the integrated plasmid. Lane 1 shows the wild-type OG1X to have a single 2.9-kb EcoRI band representing the clone-related segment, whereas this band is missing in the two transformants (lanes 2 and 3) and is replaced by a much larger band (6.1 kb) corresponding to the integrated plasmid (pVA891) and two bands of 1.8 and 1.9 kb presumed to represent the two adjacent clone-containing segments. (The sum of the sizes of the two clone-containing bands of lanes 2 and 3 corresponds to the size of the original band of lane 1 plus the size of the introduced 586 bp segment [lower band in lane 6].)

View larger version (43K): [in this window] [in a new window]   FIG. 3.   Southern blot hybridization of E. faecalis chromosomal DNA digested with EcoRI. The probe corresponded to pAM3329 carrying an internal segment of eep. Lane 1, OG1X; lane 2, FA3328 (eep mutant); lane 3, FA3329 (eep mutant); lane 4, FA3328R; lane 5, FA3329R; lane 6, pAM3329.

Effect on the production of different pheromones. The additional presence of the plasmid-encoded eep determinant in E. faecalis OG1X cells also resulted in a slight (no more than twofold) increase in production of the pheromones cPD1, cOB1, and cCF10 (Table 2). There was no detectable increase, however, in the case of pheromone cAM373. A more significant effect on cPD1, cOB1, and cCF10 was noted in the case of eep-defective mutants, where these pheromones were not detectable (Table 2); cAM373 production remained normal.

Transcriptional start site for eep. The results of a primer extension study are shown in Fig. 4. The data imply that the initial ribonucleotide is an A which is located 80 nucleotides upstream from the translational start site (Fig. 2). Likely 35 and 10 hexanucleotides can be seen in an appropriate location upstream.

View larger version (62K): [in this window] [in a new window]   FIG. 4.   Primer extension analysis. Lanes 1 and 2 represent two different extractions of RNA from E. faecalis OG1X cells. The sequence data relate to use of the same primer, using as a template pAM3327. The asterisks marks the thymine corresponding to the 3' end of the extended fragment; its complementary adenosine represents the 5' transcriptional start site that is noted in Fig. 2.

Some apparent properties of Eep. A hydrophilicity plot of Eep (not shown) revealed an apparent signal sequence as well as at least three additional membrane-spanning regions, one of which is located close to the carboxyl terminus. This suggests that Eep is a membrane protein. Using the GenBank BLASTP program, we found that Eep exhibited the strongest similarity with a hypothetical protein in B. subtilis (accession no. Z99112); regions over the entire 422 amino acids ranged from 33 to 62% identity and 48 to 75% similarity. Similarities with proteins of unknown function in H. influenzae (U60831) and Helicobacter pylori (AF016039) were also evident. The highest similarity with proteins having a known function were to metalloproteases from Treponema denticola (AE001235) and Chlamydia pneumoniae (AE001618).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Using a screening protocol devised to detect production of an increased amount of pheromone that might exceed the shutdown capacity of pAD1, we were successful in cloning a chromosomal determinant (eep) whose product indeed led to increased cAD1 secretion by the E. faecalis OG1X host. Eep itself does not contain the pheromone sequence but was necessary for pheromone expression. A knockout of the determinant resulted in a decrease in cAD1 titer from a titer of 64 to 2 or <2 (undetectable). Production of the pheromones cPD1, cOB1, and cCF10 was also reduced in the mutant; however, cAM373 was not affected. Deduced hydrophilicity properties suggested that Eep is a membrane protein, and comparisons of Eep to the GenBank database suggested some similarities to certain known bacterial proteases. Thus, it is conceivable that Eep corresponds to a membrane protein that may be involved in processing and/or secretion of the pheromone. The fact that when additional Eep was provided on a multicopy plasmid in trans an enhancement of pheromone expression was observed in wild-type cells suggests that there may be more cAD1 precursor than is normally processed or that Eep is able to enhance expression of the pheromone determinant.

Although the cAD1 pheromone determinant was not identified by using the approach described here, it is noteworthy that pheromone determinants are now beginning to reveal themselves via their appearance in the databases of complete genome sequencing projects. For example The Institute for Genomic Research (TIGR) has recently posted on their public domain web site (31a) a major portion of the genome sequence of E. faecalis V583 (42). Within this database are the pheromone sequences of cPD1 and cOB1, both of which appear to represent the final eight amino acid residues of the signal sequence of apparent lipoprotein precursors, and a sequence corresponding to cCF10 is present within two different determinants, one of which exhibits a possible relationship to the signal sequence of a lipoprotein precursor. The cAD1 determinant has not yet been identified in the TIGR database, which is still incomplete; however, very recently it has been identified in the database of another E. faecalis strain (2a). Preliminary PCR studies in our laboratory using primers based on the related sequence kindly provided by P. Barth have now revealed a similar determinant is present in E. faecalis OG1X as well as in the TIGR strain V893 (1). The cAD1 determinant (cad) corresponds to the final 8 residues in the signal sequence of an apparent lipoprotein precursor with a size corresponding to 143 amino acids and without any known homologues in the GenBank database. Thus pheromone secretion may be coupled to the processing and/or export of lipoprotein precursor structures. It would not be surprising if Eep is involved in such events; indeed, since at least some other pheromones (e.g., cPD1 and cCF10) were reduced in the eep mutant, perhaps their related precursors also make use of the protein for processing. The fact that cAM373 was not affected implies that production of at least some pheromones do not require Eep.

Firth et al. (21) have made the interesting observation that the traH gene encoded by the staphylococcal plasmid pSK41 encodes a lipoprotein precursor bearing a signal sequence containing seven of eight contiguous amino acids identical to those of cAD1 (a threonine is substituted for a serine). Indeed, staphylococcal strains carrying this plasmid actually secreted a detectable cAD1 activity. When intact traH was cloned in E. coli, clumping inducing activity was also detectable but not from a host deficient in SPase II (lipoprotein signal peptidase) (5). The pSK41 plasmid also encodes a protein TraJ which shares amino acid similarity with SPase II but which was not able to substitute for SPase II in E. coli. Although Eep could conceivably function in a similar way, it does not exhibit homology with SPase II or the pSK41 TraJ.

Finally, there is a point worth noting regarding the absence of any effect by Eep on expression of the cAM373 pheromone. This pheromone induces a response encoded by the conjugative plasmid pAM373 (12), which is smaller (size of 36 kb) than the other known pheromone plasmids, most of which are significantly larger (10, 49). In addition pAM373 differs from the others in that it does not have a homologous determinant for aggregation substance; recent sequencing data, however, have indicated that homologues of traA and traC of pAD1 are present (16a). A particularly interesting characteristic of pAM373 is the fact that peptides with activity similar to cAM373 are also produced by Staphylococcus aureus and Streptococcus gordonii (12, 15).