Polymorphisms of the pbp5 gene and correlation with ampicillin resistance in Enterococcus faecium isolates of animal origin

Enterococcus faecium is an important human pathogen whilst also being a commensal bacterium of the intestinal tract of humans and animals. E. faecium is intrinsically resistant to moderate levels of ampicillin, with MICs typically ranging from 1 to 16 µg ml¹ (Kak & Chow, 2002). This natural ampicillin resistance of E. faecium was demonstrated to be associated with expression of penicillin-binding protein 5 (PBP5), which has a low affinity for ß-lactams (Fontana et al., 1983, 1994; Rice et al., 2001). The pbp5 gene is located in the bacterial chromosome and is considered to be intrinsic to the species, although recently it has also been reported as a transferable determinant (Rice et al., 2005). Recently, an increase in ampicillin resistance has been observed in clinical E. faecium isolates, which could compromise therapy in life-threatening infections caused by this micro-organism (Klare et al., 2003). The emergence of high-level ampicillin resistance in E. faecium may be due to either increased production of PBP5 or mutations in the pbp5 gene resulting in lower affinities for ampicillin (Kak & Chow, 2002; Rice et al., 2001; Sifaoui et al., 2001; Williamson et al., 1983; Zorzi et al., 1996). Specific amino acid changes in the C-terminal region of PBP5 (especially at aa 466', 485, 496, 499, 525, 586 and 629, around the active-site region of PBP5) have been associated with ß-lactam resistance (Jureen et al., 2003; Ligozzi et al., 1996; Rice et al., 2001, 2004; Rybkine et al., 1998; Sauvage et al., 2002; Zorzi et al., 1996), and 12 different alleles of this gene have previously been demonstrated in clinical E. faecium isolates when this C-terminal region of pbp5 gene was analysed (Jureen et al., 2003).

As far as we know, all studies of the characterization and detection of mutations in pbp5 have been performed on E. faecium isolates of human origin and not on animal isolates. Thus, the pbp5 gene of ampicillin-resistant (Amp^R) and ampicillin-susceptible (Amp^S) E. faecium isolates recovered from healthy animals was sequenced in the present study. The deduced amino acid changes were found to correlate with their specific ampicillin MICs.

Bacterial isolates. Antimicrobial resistance was analysed by the disc-diffusion method in a previous study in enterococcal isolates recovered from faecal samples of healthy poultry and pets (Poeta et al., 2006). A total of 11 E. faecium isolates of that study showed an Amp^R phenotype, comprising 10 faecal samples from poultry and 1 faecal sample from a dog. These 11 Amp^R E. faecium isolates, as well as 5 Amp^S E. faecium isolates (4 from poultry and 1 from a dog), were included in the present study.

Ampicillin-susceptibility testing. The MIC of ampicillin (Eli Lilly) was determined in our E. faecium isolates using the agar-dilution method according to the Clinical and Laboratory Standards Institute (formerly the National Committee for Clinical Laboratory Standards, 2005). Enterococcus faecalis ATCC 29212 and Staphylococcus aureus ATCC 25923 were used as quality-control strains.

Analysis of ß-lactamase activity. The production of ß-lactamase was tested by streaking the bacterial colonies over nitrocefin discs (Cefinase; Becton Dickinson Microbiology Systems) (Marshall et al., 1995). After 510 min at room temperature, a change in colour from yellow to purple was indicative of a positive reaction.

PCR amplification of pbp5 and DNA sequence analysis. Total DNA from E. faecium isolates was obtained using an InstaGene matrix (Bio-Rad) and 10 µl DNA was used in each PCR, with 3.5 mM MgCl₂. The C-terminal region of the pbp5 gene was amplified by PCR in all 16 E. faecium isolates included in this study using primers PBP5F-1 (5'-AACAAAATGACAAACGGG-3') and PBP5R (5'-TATCCTTGGTTATCAGGG-3'), producing an amplicon of 779 bp. PCR conditions were as follows: 95 °C for 5 min, followed by 30 cycles of 94 °C for 30 s, 54 °C for 30 s and 72 °C for 2 min, with a final 7 min extension period at 72 °C (Jureen et al., 2003). PCR products were detected on ethidium-bromide-stained agarose gels and purified using a QIAquick gel extraction kit (Qiagen), according to the manufacturer's instructions. The purified products were sequenced on both strands using an Applied Biosystems 3730 DNA sequencer, using primers PBP5F-2 (5'-GACAAACGGGATCTCACAAG-3') and PBP5R. Duplicated sequences were obtained from independent PCRs and compared with that of the pbp5 gene reference sequence (GenBank accession no. X84860).

The complete pbp5 gene was also amplified by PCR in six of our E. faecium isolates (two Amp^S and four Amp^R) using primers PBP5F-3 (5'-AAAAATCGAACAACAGGCGCTTA-3') and PBP5R (amplicon size 1916 bp), using TaKaRa DNA polymerase (Takara Bio) under conditions recommended by the manufacturer. Amplicons were purified and sequenced by primer walking. The sequences obtained from both strands were also compared with that of the reference sequence.

Polymorphisms of the pbp5 gene were studied in a series of E. faecium isolates of animal origin that showed different susceptibilities to ampicillin. None showed ß-lactamase activity when nitrocefin discs were used for screening purposes.

C-terminal region of the pbp5 gene
The sequence of the C-terminal region of pbp5 was analysed in our 11 Amp^R and 5 Amp^S E. faecium isolates and was correlated with their specific ampicillin MIC values. The deduced amino acid changes detected between aa 461 and 634 of these enterococci are shown in Table 1, with amino acid positions considered to be important for ampicillin resistance indicated in bold.

Table 1. Polymorphisms in the C-terminal region of pbp5 in 16 E. faecium isolates of animal origin and correlation with their specific ampicillin MIC values

Eight different pbp5 alleles, designated here as AH, were detected when amino acid changes in this region were taken into account, none of which showed 100% identity with the reference sequence (GenBank accession no. X84860). Our five Amp^S isolates (MICs 18 µg ml¹) were classified as only three of these alleles (AC) and included changes at positions not previously considered to be important for ampicillin resistance (allele A 470H→Q, 581I→V, 595E→A; allele B 470H→Q, 595E→A; allele C 470H→Q, 471V→I, 487Q→L, 622E→D) (Table 1). The remaining five alleles (DH) were found in our 11 Amp^R isolates. It is important to stress that three of these alleles (FH) presented a serine insertion at position 466' in addition to other important amino acid changes: alleles F and H (485M→A, 496N→K, 499A→T, 525E→D and 629E→V) and allele G (485M→A, 496N→K, 499A→T, 525E→D, 586V→L and 629E→V). Allele F, identified in seven Amp^R isolates (MICs 32256 µg ml¹), was the most frequent allele in our study. Allele D was found in one Amp^R isolate (MIC 32 µg ml¹) and showed amino acid changes at two principal positions (496N→K and 629E→V). The remaining allele (E) was detected in one Amp^R isolate (MIC 64 µg ml¹), recovered from a dog, and showed only one amino acid change (470H→Q) with respect to the reference sequence. This specific substitution was also found in all of our isolates, both Amp^R and Amp^S, and has not previously been associated with increased ampicillin MICs by other authors (Jureen et al., 2003).

It is important to note that the insertion of serine at position 466' was detected in most of our Amp^R animal E. faecium isolates, and was generally associated with other amino acid changes. Insertions of aspartic acid or serine at position 466' have also been detected previously in strains with an increased level of ampicillin resistance (Jureen et al., 2003; Rybkine et al., 1998; Zorzi et al., 1996). According to the previously reported PBP5 crystal structure (Sauvage et al., 2002), when PBP5 protein of E. faecium (PBP5fm) forms a complex with benzylpenicillin, the active-site cavity of this protein is delimited by the ß_C3 strand on one side, by residues 461465 on the opposite side and by residues 537541 on the bottom of the cavity. Residues 461465 form part of a loop (aa 451465) that is well conserved in the B1 subgroup of class B high-molecular-mass PBPs. Thus, the substitution 461Q→K found in most of our Amp^R strains could be implicated in lower affinity for the antibiotic. The residue V465 points into the active site, being close to the ß-lactam ring; therefore, the insertion of a residue at position 466' may slightly displace V465 inside the active site, reducing its accessibility for ampicillin (Rice et al., 2004; Sauvage et al., 2002). The serine insertion was found in the strains with high-level ampicillin resistance and seems to be an essential determinant of resistance, affecting antibiotic recognition.

As in the findings of other authors (Rybkine et al., 1998; Zorzi et al., 1996), the change of methionine to alanine at aa 485 was observed in isolates with higher MICs (128256 µg ml¹), although two isolates showed a MIC of 32 µg ml¹. The bulky side chain of M485 located behind the K425 side chain is substituted by the small alanine molecule, obtaining more conformational space for lysine and resulting in a lower affinity for the ß-lactam antibiotic. Although the substitution 485M→A seems to be important for ampicillin resistance, it has been reported that this change by itself does not confer high-level ampicillin resistance (Sifaoui et al., 2001), as observed in our study.

We identified only one Amp^R isolate (MIC 128 µg ml¹) that showed the 586V→L substitution. However, Ligozzi et al. (1996) suggested that the region from aa 558 to 586 might play an important role in the ß-lactam-binding site of PBP5, as changes in this region occurred in the highly resistant strains. Residue 629 is close to one edge of the active site and the 629E→V substitution causes a change of a hydrophilic charged residue for a hydrophobic one, which is unfavourable for the overall stability of the protein and may decrease the ability of the protein to bind to ß-lactam by restricting mobility (Leszczynski & Rose, 1986; Rice et al., 2004).

In a comparison of the 8 alleles of the pbp5 gene detected in our animal isolates with the 12 alleles reported by Jureen et al. (2003) in human isolates, it is important to note that only one of our alleles (allele E with the single change 470H→Q) corresponded to one of those determined by Jureen and co-workers (designated allele 3). Allele 3 of pbp5 was detected particularly among Amp^S human isolates and also in a few Amp^R isolates (Jureen et al., 2003). In our study, allele E was found in one Amp^R E. faecium isolate (P42, MIC 64 µg ml¹) recovered from a dog faecal sample.

If only the potentially significant amino acid positions (466', 485, 496, 499, 525, 586 and 629) of the C-terminal region of PBP5 were considered, all of our Amp^S isolates presented the same amino acid pattern as the reference sequence. In the case of Amp^R isolates, four different patterns could be identified: (i) amino acid changes at the above seven positions in one isolate (MIC 128 µg ml¹, allele G); (ii) changes at six positions in eight isolates (MICs 32256 µg ml¹, alleles F and H); (iii) changes at two positions in one isolate (MIC 32 µg ml¹, allele D); and (iv) no changes at important positions in one isolate (P42, MIC 64 µg ml¹, allele E) (Table 1). Thus, it seems that specific alterations in the C-terminal part of the pbp5 gene alone could not entirely be correlated with the different levels of ampicillin resistance among our strains and other factors are probably implicated (Jureen et al., 2004; Rice et al., 2001; Sifaoui et al., 2001).

N-terminal region of the pbp5 gene
In order to determine whether the ampicillin resistance phenotype detected in E. faecium P42 (allele E) was due to amino acid changes in positions outside the C-terminal region of pbp5, the complete pbp5 gene was amplified and sequenced for this isolate, as well as for five additional Amp^R and Amp^S E. faecium isolates (P44, P54, P13, P2 and P11, belonging to alleles B, C, F, G and H, respectively). The amino acid changes detected in the N-terminal region of the pbp5 gene of these isolates with respect to the reference sequence are indicated in Table 2. It was found that E. faecium P42 and one of the Amp^S isolates showed no amino acid changes with respect to the reference sequence, whilst four changes were detected in the other Amp^S isolate. Nevertheless, a large number (up to 12) and variety of amino acid changes were demonstrated in this region in the three Amp^R isolates (P2, P11 and P13). Zorzi et al. (1996) described most of these amino acid changes in the N-terminal domain of their studied strains, with the 68A→T and 85E→D substitutions only detected in the highly resistant penicillin strains (EFM-1 and H80721, with MICs of 90 and 512 µg ml¹, respectively). These two changes were also observed in our high-level Amp^R isolates. The function of the N-terminal domain is still unclear. It has been hypothesized that it is necessary for correct folding of the C-terminal module and both domains seem to be interdependent (Sauvage et al., 2002). In fact, deletion of some segments of the N-terminal domain of PBP5 from Enterococcus hirae resulted in proteins that were unable to bind penicillin (Mollerach et al., 1996). Future studies should be carried out to determine whether the amino acid changes in the N-terminal region of PBP5 could be associated with changes in ß-lactam susceptibility in E. faecium species.

Table 2. Amino acid changes detected in the N-terminal region of pbp5 in six of the E. faecium isolates of this study

Conclusions
The specific alterations in the C-terminal part of the pbp5 gene could be used as markers for ampicillin resistance (at least the insertion at aa 466' and the changes at position 485), as they are present in most of the Amp^R but not in the Amp^S strains, although, as observed by other researchers (Jureen et al., 2004; Rice et al., 2001; Sifaoui et al., 2001), it seems that other factors could be necessary for a complete explanation of the differences in levels of ampicillin resistance. High levels of resistance could be observed in strains that either overexpress PBP5 and/or present multiple mutations that decrease the affinity of PBP5 for the antibiotic. This possibility could explain the different levels of ampicillin resistance among strains classified as the same allele (allele F) or the P42 strains whose MIC values could be due simply to protein overexpression. Zorzi et al. (1996) suggested that only a single copy of the pbp5 gene was present in their analysed strains. Thus, the variations seen in the amount of PBP5 could not be due to a gene dosage effect, although the detection of pbp5 located within transferable elements could change this suggestion (Rice et al., 2005). More recently, it has been reported that expression of ampicillin resistance was higher when the pbp5 gene was located downstream of the ftsW_Efm and psr genes (Rice et al., 2001). The role of the putative ftsW_Efm gene product is unknown, although it has been suggested that its homologue in Escherichia coli may serve as a chaperone protein for PBP3 (Eberhardt et al., 2003). In addition, it has been proposed that overproduction of PBP5 could result from either an alteration in the psr promoter or a modification of the target site of Psr, probably located between the psr and pbp5 genes (Zorzi et al., 1996).

In conclusion, the C-terminal region of the pbp5 gene was found to be highly polymorphic among our Amp^R and Amp^S E. faecium isolates of animal origin and there seemed to be a correlation between specific changes located at positions close to the active site of PBP5 and the ampicillin MIC, although further studies of the mechanisms of ampicillin resistance in E. faecium are necessary.

P. P. was supported by a grant with the reference SFRH/BD/17424/2004 from Fundaçao para a Ciência e a Tecnologia (FCT) of Portugal. This work has been financed in part by the Integrated Action SpainPortugal from the Ministry of Education and Science (HP2005-0052).