Identification of the cAD1 Sex Pheromone Precursor in Enterococcus faecalis

Abstract

The Enterococcus faecalis virulence plasmid pAD1 encodes a mating response induced by exposure to an octapeptide sex pheromone, cAD1, secreted by plasmid-free enterococci. The determinant for the pheromone in E. faecalis FA2-2, designated cad, was found to encode a 309-amino-acid lipoprotein precursor with the last 8 residues of its 22-amino acid signal sequence representing the cAD1 moiety. The lipoprotein moiety contained two 77-amino-acid repeats (70% identity) separated by 45 residues. The nonisogenic E. faecalis strain V583 determinant encodes a homologous precursor protein, but it differs at two amino acid positions, both of which are located within the pheromone peptide moiety (positions 2 and 8). Construction of a variant of strain FA2-2 containing the differences present in V583 resulted in cells that did not produce detectable cAD1. The mutant appeared normal under laboratory growth conditions, and while significantly reduced in recipient potential, when carrying pAD1 it exhibited a normal mating response. A mutant of FA2-2 with a truncated lipoprotein moiety appeared normal with respect to recipient potential and, when carrying plasmid DNA, donor potential. A gene encoding a protein designated Eep, believed to be a zinc metalloprotease, had been previously identified as required for pheromone biosynthesis and was believed to be involved in the processing of a pheromone precursor. Our new observation that the pAD1-encoded inhibitor peptide, iAD1, whose precursor is itself a signal sequence, is also dependent on Eep is consistent with the likelihood that such processing occurs at the amino terminus of the cAD1 moiety.

Certain conjugative plasmids in Enterococcus faecalis encode a mating response to sex pheromones secreted by plasmid-free enterococci (19, 20). The response is characterized by the synthesis of a surface protein aggregation substance which is able to bind to enterococcal binding substance on the surfaces of both recipient and donor cells. The response of plasmid-containing donors to nearby recipient (plasmid-free) cells results in the initiation of mating aggregate formation. However, if donors are exposed to a culture supernatant of recipients, a self-aggregation (clumping) is observed, a phenomenon that serves as the basis for quantitative assay for pheromone activity. For recent reviews of the enterococcal pheromone systems, see references 12 and 15.

pAD1 (11, 17, 50) is a highly conjugative pheromone-responding plasmid that has been studied extensively; its nucleotide sequence has recently been completed (24). pAD1 is a member of a widely disseminated family of mobile enterococcal elements that encode a hemolysin/bacteriocin (cytolysin) and resistance to UV light (29, 32, 35, 44); its hemolysin and aggregation substance have been shown to contribute to virulence (10, 30, 34, 36, 37, 47). The cognate sex pheromone cAD1 is an octapeptide with the sequence LFSLVLAG (42).

When a plasmid-free, pheromone-producing bacterium acquires a plasmid by conjugation, pheromone activity in culture supernatants of the transconjugant can no longer be detected. This is because of plasmid-encoded functions that involve “masking” and, in some cases, a “shutdown” of endogenous pheromone. Masking relates to the secretion of specific octa- or heptapeptides that desensitize the cells to exogenous pheromone (16, 33, 43). They act as competitive inhibitors of the pheromones and serve to prevent self-induction of conjugation functions in the absence of recipient cells. While a given plasmid-bearing cell does not emit the cognate pheromone, it continues to produce unrelated pheromones specific for other families of plasmids. In the case of pAD1, the inhibitor iAD1 has the structure LFVVTLVG, which is 50% identical to cAD1 (41).

Exogenous cAD1 is believed to bind to a plasmid-encoded surface lipoprotein, TraC (48), that enhances donor sensitivity and participates in uptake of the peptide via a host-encoded ABC peptide transport system (38). There is evidence that once inside the cell the peptide binds directly to a DNA-binding, negative regulator protein, TraA, which in turn releases its binding to DNA, allowing induction of the mating response (26). The inhibitor iAD1 probably competes with cAD1 for binding to TraC; there is currently no evidence that secreted inhibitor reenters the cell.

The known enterococcal sex pheromones (cAD1, cPD1, cCF10, cAM373, and cOB1) (15) and related inhibitors are all relatively hydrophobic, linear octa- or heptapeptides that are active at nanomolar concentrations. Interestingly, some of them have relatively strong neutrophil chemotaxis activity (22, 46). With the recent availability of enterococcal genome sequence data, it was noted that they correspond to part of the signal sequences of precursors of certain lipoproteins (13). In most cases, the signal sequences correspond to 21- or 22-amino-acid segments, with the last 7 or 8 residues representing the specific pheromone. Typical lipoprotein signal peptidase target sites are appropriately located such that cleavage results in separation of the signal sequence, which in turn needs only to be processed at a second location seven or eight residues from the other processing site to generate a mature pheromone peptide. It is not known if there is a functional relationship between the activity of the putative lipoproteins and the pheromone component of their precursor structures or whether the lipoprotein connection is simply fortuitous.

Interestingly, the plasmid-encoded inhibitors are synthesized as 20- to 23-amino-acid precursors which resemble a signal sequence. Such precursors must be processed to generate the mature inhibitor peptide, perhaps by a mechanism resembling the processing system utilized by pheromone precursors. We have recently characterized a gene on the E. faecalis chromosome that is necessary for the production of cAD1 as well as certain other sex pheromones (3). This gene (eep) encodes a 46.5-kDa membrane protein initially identified based on its ability to enhance expression of cAD1 (eightfold) when present on a plasmid vector (pAM401) in E. faecalis. In addition, enterococcal eep mutants did not produce detectable pheromone. It was therefore suggested that it might be involved in processing of pheromone precursors.

In this communication we present data relating to the gene for the cAD1 precursor (cad) in E. faecalis FA2-2 and other strains of E. faecalis, as well as a strain which differs significantly only in the pheromone component of the precursor. The effects of mutations within cad are also presented, as well as data relating to the likely involvement of Eep in processing.

MATERIALS AND METHODS

Bacterial strains, plasmids, and reagents.

The bacterial strains and plasmids used in this study are listed in Table 1. pAM211 and pAM714 are derivatives of pAD1 containing insertions of Tn916 (28) and Tn917 (2, 31), respectively. The transposon insertions are in locations that are not involved in transfer and are believed to exhibit wild-type conjugation. pAM4110 is a derivative of pAM373 with a Tn917-lac insertion (18); it is readily transferable but produces an elevated level of the inhibitor iAM373.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Description	Reference or source
E. faecalis
FA2-2	rif fus	17
OG1X	str gel	33
OG1SS	str spc	25
JH2SS	str spc	49
V583	Clinical isolate (resistant to vancomycin, gentamicin, erythromycin, and tetracycline)	45
FA373	FA2-2 carrying pAM373; Tn 918 (tet) on chromosome	14
FA3328	Eep-negative mutant of OG1X	3
FA3333	FA2-2 defective in cAD1	This study
FA3335	FA2-2 with pAM3334 integrated in chromosome	This study
DS16	Clinical isolate carrying pAD1 and pAD2	17
39-5	Clinical isolate carrying pPD1	53
E. coli DH5α	F−φ80 lacZΔM15 endA1 recA1 hsdR17 (rK− mk+) supE44 thi-1 ΔgyrA96 (ΔlacZYA-argF) U169	Promega Research Laboratories
Plasmids
pAM211	pAD1::Tn 916 (tet)	27
pAM714	pAD1::Tn 917	31
pAM4110	pAM373::Tn917-lac	18
pCF10	Carries Tn 925 (tet); responds to cCF10	21
pAM88	pSU18 with additional (gram-positive) cat determinant	Victoria Francia
pAM401	E. coli-E. faecalis shuttle vector; cat tet	51
pVA891	E. coli plasmid vector; erm cat	39
pAM3332	PCR fragment with part of cad (V583) in pVA891	This study
pAM3334	PCR fragment of cad (internal) in pAM88	This study

Media, reagents, and assays.

The media used were Todd-Hewitt broth (THB) (Difco Laboratories, Detroit, Mich.) and N2GT (nutrient broth no. 2 [Oxoid Ltd., London, United Kingdom] supplemented with 0.2% glucose and 0.1 M Tris-HCl [pH 7.5]) for E. faecalis and Luria-Bertani broth (40) for Escherichia coli. Cell density was determined using a Spectronic colorimeter at a wavelength of 600 nm. The solid medium used was Todd-Hewitt broth with 1.5% agar. General reagents were essentially as previously described (6, 7). Matings were performed as detailed elsewhere (14). 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; fusidic acid, 25 μg/ml; and spectinomycin, 500 μg/ml. Synthetic cAD1 peptide was prepared at the University of Michigan peptide synthesis core facility. The synthetic vAD1, peptides used for the cAD1 alanine substitution experiments, and cAD1 derivatives containing additional residues at the N terminus were purchased from PeptidoGenic Research & Co. (Livermore, Calif.). Pheromone response (clumping) assays, including preparation of culture fluids, were performed as previously described (20). Restriction enzymes were purchased from Invitrogen Life Technologies (Grand Island, N.Y.), and reactions were carried out under the conditions recommended by the manufacturers.

DNA and RNA methodology.

Cloning, electrophoretic analyses, plasmid isolation, and electroporation techniques were as previously described (1, 3, 4). PCR was performed with a Perkin-Elmer Cetus apparatus under conditions recommended by the manufacturer. Specific primers, in some cases containing specifically added restriction sites, were synthesized at the Biomedical Research DNA Core Facility of the University of Michigan or Invitrogen Life Technologies. PCR products were handled as previously described (3). The primers used for the generation of the reverse PCR product were 5′ GCCACAGCCTGCTAATACTAAACTAAATAA (rev-pcr-1) and 5′ TGGTAACACAGACACAATCGAAATCGACAAT (rev-pcr-2). The primers used to generate the mutated (V583) sequence were 5′ ggaattccGTCATTGACAATCCGCTCCTTAA (891-top; contains an incorporated EcoRI site [lowercase]) and 5′ ggaattccTTCTTTTGGTCCAACACCAGA (891-btm; contains an incorporated EcoRI site [lowercase]). The primers used to generate the segment representing an internal portion of cad were 5′ acatgcatgcatgtGGTTACCGTGCAGTCTTTGAA (su-top; contains an incorporated SphI site [lowercase]) and 5′ gctctagagcTTCAACACCTGCAGGTGAGCC) (su-btm; contains an incorporated XbaI site [lowercase]).

DNA sequencing made use of the Biomedical Research DNA Core Facility of the University of Michigan, and sequences were analyzed using a MacVector software package from Eastman Kodak. Primer extension was performed as described elsewhere (7) and made use of the primer rev-pcr-1 described above.

Identification of cad.

Information that played a role in the identification of cad came initially from a partial enterococcal sequence database from AstraZeneca (13). However, sequence information utilized in the current study made use of that available for E. faecalis strain V583 located on the public domain website of The Institute for Genomic Research (TIGR; http://www.tigr.org/tdb/mdb/mdbcomplete.html). The latter differed from the AstraZeneca data at two of the eight amino acid residues of cAD1, but downstream sequences were essentially the same. Oligonucleotides based on the TIGR (V583) sequence were designed to amplify (via PCR) various segments using template DNA from E. faecalis FA2-2. The sequences upstream and downstream of cad were determined using inverse PCR with outwardly reading primers, rev-pcr-1 and rev-pcr-2 (Fig. 1); chromosomal DNA that had been cleaved with HindIII and self-ligated was used as a template. (Note that the inverse PCR made use of a different E. faecalis strain [OG1X], but sequences adjacent to cad ultimately proved to be the same for FA2-2 and OG1X)

FIG. 1.

Key regions of the nucleotide sequence of cad along with related amino acids in E. faecalis FA2-2. The region representing vAD1 of E. faecalis V583 is also marked (box). The arrows approximate the locations representing specific primers used for PCR and for the primer extension experiment shown in Fig. 3. The locations of the −10 and −35 hexamers of the promoter, as well as the transcription start site (bent arrow) and the ribosome binding site (S. D.), are noted.

Generation of cad mutants.

Using chromosomal DNA from E. faecalis V583 as a template, a 655-bp PCR product was generated with primers that flanked the cAD1 component of cad by more than 300 nucleotides on each side. (The approximate locations of the PCR primers, 891-top and 891-btm, are shown in Fig. 1.) The segment corresponding to cAD1 of strain V583 is designated vAD1 because it differs by two amino acids from cAD1. EcoRI sites introduced by the primers were used to clone the PCR fragment into the E. coli vector pVA891. A resulting chimera in E. coli strain DH5α, designated pAM3332, was introduced into E. faecalis strain FA2-2 by electroporation and selection on plates containing erythromycin. Because the plasmid cannot replicate in E. faecalis, transformation resulted in integration of the element via homologous recombination. Several transformants were subcultured in the absence of erythromycin for five or more passages in broth at 42°C, resulting in several derivatives that had become sensitive to erythromycin. Of eight strains tested, four no longer produced detectable cAD1 while four produced a normal level of cAD1. DNAs from two of each type were sequenced and confirmed to be the desired mutant or the wild type. One of the cAD1-negative strains was designated FA3333.

The generation of a mutation affecting the mature portion of cad involved the PCR amplification of a product representing a 588-bp internal segment of the gene, which was then cloned into the E. coli vector pAM88. (pAM88 was kindly provided by V. Francia and is a derivative of pSU18 [5] with a cat determinant that expresses in gram-positive bacteria [M. V. Francia and D. B. Clewell, unpublished data].) The primers used, su-top and su-btm, are indicated in Fig. 1 and were designed to add SphI and XbaI restriction sites, respectively, for cloning. A plasmid chimera, pAM3334, was generated in DH5α and used to transform E. faecalis FA2-2 by electroporation and selection with chloramphenicol. A strain confirmed to have the plasmid inserted into cad via PCR was designated FA3335.

Nucleotide sequence accession number.

The GenBank accession number for the sequence shown in Fig. 1 is AF421355.

RESULTS AND DISCUSSION

Nucleotide sequence of cad in E. faecalis FA2-2.

The cad coding sequence in the chromosome of E. faecalis FA2-2 was identified as described in Materials and Methods; key regions are shown in Fig. 1 and 2. It encodes a 309-amino-acid protein representing an apparent lipoprotein precursor (Cad) with a typical“lipobox” associated with a cysteine residue (52) at the signal sequence junction. (This polypeptide is larger than that originally found in the preliminary sequence data provided earlier by AstraZeneca [13].) The only homologue noted in the database is a lipoprotein of Treponema pallidum, and the resemblance is not particularly strong. The signal sequence corresponds to 22 amino acids, the last 8 of which constitute cAD1. The resemblance of the region to the iAD1 precursor sequence (16) is striking in that the latter appeared to represent a “lone” signal sequence, with the last 8 residues corresponding to the mature peptide. Although the regions upstream of the cAD1 and iAD1 moieties differ from each other, both can be represented as amphipathic helices (not shown). The hydrophilicity plot (Fig. 2) shows an absence of transmembrane regions in the presumed mature or processed lipoprotein, suggesting that the latter is an extracellular or surface molecule. It has a net negative charge (pI, 4.9); the precursor has a pI of 5.1. A 77-amino-acid segment is repeated within the sequence with a 45-amino-acid separation; the repeated regions exhibit 70% amino acid identity (Fig. 2).

FIG. 2.

Amino acid sequence of Cad and hydrophilicity plot. (A and B) Sequences and comparison, respectively, of the 77-residue repeats. (C) Hydrophilicity plot. In panel A, large letters represent cAD1 and repeated sequences are underlined. In panel B, asterisks indicate where the amino acids differ.

The nonisogenic E. faecalis strain V583 contains an almost identical lipoprotein (TIGR database) precursor differing in only three amino acids, two of which are located within the pheromone-related segment of the signal sequence (shown in bold type) (LSSLVLAA [designated vAD1 here] rather than LFSLVLAG [cAD1]). The related codons in the pheromone segment involve a change in 2 nucleotides in each case (Fig. 1), and we were not able to detect a cAD1 activity in supernatants of the V583 strain. The third amino acid difference is at position 37, where V583 has an alanine but FA2-2 has a valine. A synthetic peptide representing vAD1 of V583 (i.e., LSSLVLAA) had no cAD1 activity (using E. faecalis DS16 as a responder) at concentrations as high as 10 μg/ml (more than 10,000 times the concentration at which cAD1 activity is generally detectable), nor was it able to induce a clumping response by V583. Furthermore, synthetic cAD1 was not able to induce clumping of V583 at concentrations as high as 100 μg/ml. Interestingly, an iAD1 activity with a titer of 16 was detectable in V583 culture supernatants. This is believed to relate to a pAD1-like element in V583.

Sequence analysis of DNA from the nonisogenic strain OG1X revealed a Cad precursor structure identical to that of strain FA2-2 except that it had an alanine rather than a valine at position 37. Another nonisogenic strain, DS16, also has an alanine at this position but in addition has an asparagine rather than a glycine at position 101.

A reasonable ribosome binding site is appropriately located upstream of the translational start site, and a putative promoter sequence consisting of consensus −10 (TATAAT) and −35 (TTGACA) sequences is present upstream. Results of primer extension experiments (Fig. 3) with extracted RNA showed that this site was the cad promoter. The initial nucleotide is an A, which is located 10 residues downstream of the −10 hexamer. An intrinsic termination site is not evident within 90 bp downstream of cad, but a 23-nucleotide sequence that is 91.3% AT spans the translational stop site.

FIG. 3.

Primer extension analysis of RNA representing cad. The primer used was rev-per-1. The asterisk 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. 1.

Located 66 nucleotides upstream of the cad translational start site is an oppositely oriented 33-amino-acid reading frame, which is probably nonfunctional, as it does not have a good ribosome binding site. (The open reading frame does not resemble anything in the database.) Similarly, the 90-nucleotide region immediately downstream of cad contains only a similarly oriented 27-amino-acid frame (no homologues) without a significant ribosome binding site.

Generation of mutants.

The differences between the cAD1 and vAD1 Cad sequences prompted a test of the effect substitution of vAD1 for cAD1 in the FA2-2 host would have with regard to not only pheromone production but also the ability to act as a donor when harboring pAD1. This was done by generating a PCR product containing the signal sequence of the V583 homologue (Fig. 1), cloning it onto a plasmid vector (pVA891) in E. coli, and then introducing it into FA2-2 by allelic exchange (see Materials and Methods). FA3333 was one of two candidates that no longer produced cAD1 and was confirmed by sequencing to have the intended modification (the entire allele was sequenced). The sequences upstream of the cad reading frame were identical for more than 200 nucleotides in both FA2-2 and V583, and this region was present intact in the variant. While the variant included the alanine at position 37 in Cad (i.e., from V583), this is not likely to be significant, since the pheromone-encoding strains OG1X and DS16 both have an alanine at this location. Culture filtrates were devoid of cAD1 activity, but normal amounts of cCF10, cPD1, and cAM373 were produced (Table 2). FA3333 had a normal growth rate and appearance of colonies on THB plates; however, they were poor recipients (>2 orders of magnitude fewer transconjugants in 90-min broth matings than FA2-2), as expected (data not shown). When a pAD1::Tn917 derivative, pAM714 (with wild-type conjugation properties), was introduced into the mutant (filter mating), the transconjugants exhibited a typical clumping response to exogenous cAD1. The sensitivity of these transconjugants to the pheromone was essentially the same as that of the wild-type host when carrying pAM714; when donors were compared in short (10-min) matings after a 45-min exposure to synthetic pheromone, the transfer frequencies were essentially the same for both.

TABLE 2.

Pheromone activity of culture filtrates and synthetic peptides

Strain (filtrate)	Synthetic peptidea	Activity (titer)
		cAD1	cPD1	cCF10	cAM373
FA2-2		32	64	16	16
FA3333		<2	64	16	16
FA3335		32	64	16	16
V583		<2	<2	<2	8
	cAD1	>512	2	<2	<2
	vAD1	<2	<2	<2	<2
	A-cAD1	32
	IA-cAD1	32
	AIA-cAD1	32

^aOne microgram per milliliter in each case; A, IA, and AIA residues added to amino terminus of cAD1.

In an effort to mutate the lipoprotein component of Cad, we generated a PCR product corresponding to an internal part of cad (see Materials and Methods) (Fig. 1) and introduced it into pAM88 in E. coli DH5α. The chimeric DNA (pAM3334) was then introduced into the plasmid-free FA2-2 with selection (chloramphenicol resistance) for integration. The mutant, FA3335, was examined by PCR (data not shown) to confirm that the integration was correctly positioned (see Materials and Methods). The derivative exhibited a normal growth rate and colony morphology and gave rise to an amount of cAD1 in culture supernatants that was essentially identical to that of wild-type cells (Table 2). The mutant would be expected to have a truncated Cad missing 39 of its own amino acid residues from its carboxyl terminus (about 14% of the mature lipoprotein) but containing some fused missense residues. Because the normal cAD1 component is still present, it was not surprising to observe a normal amount of pheromone. The data also imply that if Cad has a self-regulatory function, the absence of the carboxyl-terminal region does not affect it. Introduction of pAM714 into strain FA3335 resulted in cells that responded normally to exogenous pheromone with respect to both clumping and mating.

Activities of modified synthetic forms of cAD1.

The difference between the lipoprotein precursors of strains FA2-2 and V583 with respect to the pheromone moiety raised the question of the contribution of each amino acid in the mature peptide to pheromone-specific activity. To address this point, we generated a series of synthetic peptides with an alanine substituted at different positions and determined their abilities to induce the clumping and mating responses at concentrations of 1 μg/ml and 50 ng/ml, respectively. The results, shown in Fig. 4, indicated that all but one of the substitutions caused a significant decrease in activity. (Position 7 was not tested, since it is already an alanine). The A1 substitution (alanine substituted at the no. 1 position) had an order of magnitude less activity with respect to induction of clumping and close to 2 orders of magnitude less regarding transfer of plasmid DNA. The others, with the exception of A5, had activities more than 3 orders of magnitude lower than that of cAD1. A5 had an activity that was indistinguishable from that of cAD1, a result consistent with the use of a common “face” of the peptide for induction of both clumping and conjugative transfer. Because peptides A2 and A8 correspond to the two positions in V583 which differ from cAD1, it was not surprising that the synthetic peptides differing at both positions in V583 (i.e., LSSLVLAA [differences in boldface]) had undetectable cAD1 activity (see above). Overall, the induction of clumping and induction of transfer potential were closely parallel for all of the substitutions.

FIG. 4.

Effects of the substitution of alanine in various positions within cAD1 on the clumping and mating response. E. faecalis strain DS16 was used as the responder for the clumping assays. The concentration of synthetic peptide used in each case was 1μg/ml. For the mating induction assays, the peptide concentration used was 50 ng/ml. The donor strain was FA2-2/pAM714, and the recipient was JH2SS. In each case, donors were exposed to the peptide for 45 min and then mixed with recipients for 10 min before being plated on selective medium. Note that a substitution at position 7 was not done because an alanine is already present at that position in cAD1. Solid bars, clumping titer; shaded bars, frequency per donor with induction; open bars, frequency per donor without induction.

We also determined the activities of synthetic forms of cAD1 that contained one, two, or three amino acid residues corresponding to those in the precursor added onto the amino terminus of the pheromone. Peptides ALFSLVLAG, IALFSLVLAG, and AIALFSVLAG (added residues in boldface) were all reduced about 16-fold in ability to induce clumping (Table 2) compared to cAD1. Thus, while the first residue (A) adjacent to the amino terminus affected activity, an extension of two additional residues (AIA) had no further effect. The data are consistent with the importance of processing occurring precisely at the amino terminus of the cAD1 moiety during biosynthesis.

Requirement for Eep in iAD1 in addition to cAD1 production.

The resemblance of the signal sequence of the cAD1 precursor to that of the iAD1 precursor suggests how processing occurs. Since cleavage of Cad at the carboxyl terminus of the cAD1 moiety most likely involves lipoprotein signal peptidase (SPase II), processing at the amino terminus of the segment involves a different proteolytic activity. We have speculated that the product of the recently identified eep determinant may play such a role (3, 13). Since the amino acid residues are similar around this region for both Cad and the iAD1 precursor (i.e., IA-LF and IT-LF, respectively), we determined whether detection of the inhibitor peptide (iAD1) was affected in supernatants of an eep mutant host, FA3328, carrying pAM211 (pAD1::Tn916). Table 3 shows that iAD1 activity was not detectable. This is consistent with both precursors being processed at the same site, although it does not prove unambiguously that Eep acts specifically at this site. Eep is a member of a recently recognized family of zinc metalloproteases that are able to process within membranes of both bacteria and higher systems (9).

TABLE 3.

Inhibitor activity secreted by Eep-negative plasmid-carrying hosts^a

Strain	Eep	Inhibitor	Titer
O1X/pAM211	+	iAD1	8
FA3328/pAM211	−	iAD1	<2
OG1X/pAM4110	+	iAM373	16
FA3328/pAM4110	−	iAM373	8
OG1X/pAM351	+	iPD1	4
FA3328/pAM351	−	iPD1	<2
OG1X/pCF10	+	iCF10	2
FA3328/pCF10	−	iCF10	<2

^aFA3328 is Eep negative. +, present; −, absent.

Eep has been noted previously not to play a role in the production of a different sex pheromone, cAM373 (3). We therefore examined whether an effect could be detected on the corresponding inhibitor, iAM373, encoded by the pAM373 derivative pAM4110. Table 3 shows that there was only a twofold difference (one well in the microtiter dilution assay), which we do not consider significant. This is not surprising, considering that the related processing sites for the cAM373 and iAM373 precursors would be LG-AI and GL-SI, respectively, both of which differ significantly from the respective regions of the cAD1 and iAD1 precursors. Efforts to examine the production of inhibitors encoded by pPD1 and pCF10 were not conclusive, since the normal level of detectable inhibitor in both cases was too low to be able to resolve significant differences using the Eep-negative host.

Concluding remarks.

It is now clear that the determinant of cAD1 corresponds to part of the signal sequence of a lipoprotein precursor. Alterations designed to eliminate expression of cAD1 activity, but presumed to allow continued synthesis of the lipoprotein, did not affect cell growth or colony morphology under the laboratory conditions we normally use for bacterial cultivation. Also, the ability to establish a pAD1 derivative in such a host, although with reduced efficiency, implies that cAD1 is not required for maintenance of the plasmid. Furthermore, the presence of such a mutation in cells carrying the plasmid did not affect the ability to respond and transfer to a recipient strain with normal efficiency. Although deletion of the carboxyl-terminal 14% of the lipoprotein also did not result in a distinguishable phenotype, we are attempting to determine whether a more drastic modification of the lipoprotein might do so.

It is intriguing that E. faecalis V583 has a cad determinant differing from those of FA2-2, OG1X, and DS16 with respect to the two amino acid residues that eliminate cAD1 activity. This would seem to be more than a chance variation, considering that the two corresponding codons each have 2-nucleotide differences. V583 is a clinical isolate resistant to multiple antibiotics, including vancomycin and gentamicin. Among the other three strains, only DS16 carries antibiotic resistance, although this does not include resistance to vancomycin or gentamicin. Whether a connection exists between the peptide difference and the ability to acquire or express certain resistance traits, or other characteristics, remains an open question.

The determinant traH carried on the Staphylococcus aureus plasmid pSK41 has been reported by Firth et al. (23) to encode a lipoprotein precursor with a signal sequence containing a cAD1-like component. It differed in having a threonine rather than a serine at the number 3 position of the peptide, and a low cAD1 activity could be detected in supernatants of a pSK41-bearing strain of S. aureus. A plasmid clone containing the traH determinant was found to express cAD1 activity from E. coli; however, activity was not detected when the construct was present in a lipoprotein signal peptidase (SPase II)-negative host (8). Our efforts to clone a PCR product containing cad in E. coli and to observe expression of cAD1 have not been successful, suggesting that expression may be detrimental to the host. It is noteworthy, however, that we have been successful with regard to a different pheromone determinant, cam373 (camE) (23a). Production of the inhibitor peptide iAD1 has been readily detected in E. coli clones (16), and the fact that appropriate processing occurs suggests that an Eep-like activity, or something able to substitute for such an activity, exists in E. coli. In the E. faecalis host, one can envision an export system that involves a close association of SPase II (a homologue is present in the V583 database) and Eep, with the two membrane proteins working together in the processing of Cad (Fig. 5).

FIG. 5.

View of how Eep and SPase II may act within the cytoplasmic membrane in the processing of the precursor (Cad) protein to generate cAD1 and the related lipoprotein. Note that the processing of the 21-residue inhibitor precursor pAD1-containing cells resembles that of Cad after removal of the lipoprotein moiety from the latter. Arrows point to cleavage sites.