Enterococcus faecium PBP5-S/R, the Missing Link between PBP5-S and PBP5-R

Abstract

During a study to investigate the evolution of ampicillin resistance in Enterococcus faecium, we observed that a number of E. faecium strains, mainly from the recently described subclade A2, showed PBP5 sequences in between PBP5-S and PBP5-R. These hybrid PBP5-S/R patterns reveal a progression of amino acid changes from the S form to the R form of this protein; however, these changes do not strictly correlate with changes in ampicillin MICs.

TEXT

The low-affinity class B penicillin-binding protein 5 (PBP5) is considered the major contributor to ampicillin resistance in Enterococcus faecium (1, 2, 3, 4). We previously reported that the gene encoding PBP5 exists in two distinct allelic forms, pbp5-S and pbp5-R, and that ampicillin-susceptible, community-associated (CA) strains found in healthy individuals carry pbp5-S (ampicillin MIC usually ≤2 μg/ml), while the pbp5-R allele is harbored by ampicillin-resistant hospital-associated (HA) strains (ampicillin MIC usually ≥16 μg/ml) (5, 6). The nucleotide sequences of the two dominant pbp5 alleles differ by about 5%, and we found that 20 to 21 (some PBP5 variants have an extra codon, 466′) amino acid positions were consistently different when we compared PBP5-S and PBP5-R (5). Moreover, each PBP5 type has a consensus sequence pattern for the 20 to 21 differing amino acids, and we refer to these as the PBP5-S consensus and the PBP5-R consensus (shown at the top and bottom in Table 2, respectively). We also observed a third type of PBP5 pattern (PBP5-S/R), described as a “hybrid-like” PBP5 because it falls between the two other sequences, and this was present in some strains with ampicillin MICs of 4 μg/ml (5). The observation of two distinct pbp5 alleles (5), as well as other studies (6), suggested the existence of two distinct E. faecium clades, estimated to have diverged approximately 3,000 years ago: the HA clade, now referred to as clade A, which includes most of the strains responsible for human infections, and the CA clade, or clade B, which contains primarily human commensal isolates (7–9). In addition, the analysis recently presented by Lebreton et al. revealed a further split within clade A into subclade A1 (which has the vast majority of clinical isolates, e.g., clonal complex 17 [CC17]) and subclade A2, associated with animals and sporadic human infections (8); this later split was estimated to have occurred 75 years ago. We have now analyzed the PBP5 sequences of 22 additional strains, determined their ampicillin MICs, and compared them with the PBP5-S and PBP5-R consensus. In addition, we included in our analysis 8 other strains, previously reported to belong to the PBP5-S group (Com15), PBP5-R group (TX16 and C68), and the intermediate PBP5-S/R hybrid-type group (1141733, TX2050, TX1401, TX2042, and D366) (5).

TABLE 2.

Amino acid changes in the PBP5 sequence of the E. faecium strains compared to the PBP5-S and PBP5-R consensus sequences^h

^aPreviously described strains (5) used as references for the PBP5-S (Com15), PBP5-R (TX16 and C68), and hybrid-like (1141733, TX2050, TX1401, TX2042, and D366) patterns (also highlighted in gray).

^bND, not determined.

^cClade assigned by Lebreton et al. (8).

^dStrains for which the ampicillin MIC was determined in this study.

^eNo clear consensus in this amino acid position; 586-Val is also commonly found in strains from the PBP5-R group.

^fAmp, ampicillin.

^gMLST, multilocus sequence type.

^hThe strains of each pattern are grouped together and are separated from other patterns by line spaces.

ⁱ—, absence of 466 in the consensus sequence.

The PBP5 sequences were obtained from the NCBI website (GenBank, National Center for Biotechnology Information, http://www.ncbi.nlm.nih.gov/) if available or after sequencing, using the primers previously described (5). The ampicillin MICs were determined using the broth microdilution method according to CLSI guidelines (10). For 13 strains for which the clade was unknown, the assignment into clade A or B was performed by PCR detection of eight genes, spread over different regions of the chromosome and determined by Lebreton et al. to be present in ≥98% of the genomes of one clade but absent from the genomes of the comparative clade (8) (Table 1).

TABLE 1.

Genes and primers used to predict the clade of the E. faecium strains in which the clade was unknown

Clade specificity	Gene identifier	Primer sequence (5′–3′)	Size (bp)
A	EFAU004_01075	01075-F: CATCCACTGCAGATTTTTTGG	243
		01075-R: GATTGTTCGCAAAGAAACAT
	EFAU004_01280	01280-F: CTTTAATGCTGGCCAATATT	297
		01280-R: AATCAAGGGGCGATTCCGAA
	EFAU004_02466	02466-F: GCATAAATCTCGGCTACTCA	274
		02466-R: CAAAAACAAGTAATCAAGGT
	EFAU004_02509	02509-F: GATCATACAAGCAATTTCTA	298
		02509-R: ACTCATTTGCTGTTTGAGGT
B	EfmE980_0162	0162-F: CGCAAAATTGAAAAATACA	354
		0162-R: CTCAAAAATGGGTGGGATACAG
	EfmE980_0607	0607-F: TTCCGCTTCCACTTCAATTT	314
		0607-R: TGGTCCACATCACTCACTT
	EfmE980_1863	1863-F: ACAAAGTACTACTTCTCTGCTC	208
		1863-R: GTACCGCTGGTGACAATCGCC
	EfmE980_2464	2464-F: ATCAACCGATAGTTCAAATC	278
		2464-R: CTTTCTTTTTATGAACAAGC

Our results showed that the main variability in the PBP5 sequence of the analyzed strains occurs in the 20 to 21 positions previously reported to consistently differ between PBP5-S and PBP5-R (5) (Table 2), with some occasional additional differences outside these 20 to 21 amino acids. For the patterns shown in Table 2, the designation PBP5-S_X/R_Y indicates that X positions match the PBP5-S consensus and that Y positions match the PBP5-R consensus. Distinctive variant patterns with differences in the 20 to 21 amino acids of the consensus sequences were identified among the 22 strains (Table 2); the strains of each pattern are grouped together and are separated from other patterns by line spaces. While one could imagine that the patterns shown in Table 2 may have arisen sequentially, we have no direct evidence that this is the case.

The PBP5 sequence analysis and ampicillin susceptibility testing showed that PBP5-S/R proteins occur in strains with variable MICs that range from 0.5 to 128 μg/ml (Table 2) (median, 2 μg/ml); in comparison, we previously found that PBP5-S and PBP5-R were associated with MICs of ≤2 μg/ml and ≥16 μg/ml, respectively (5). We also found that different amino acid variations were found in strains with the same ampicillin MIC (e.g., TX0338 versus EnGen0009, MIC = 1 μg/ml) and the same sequence type (ST) (e.g., EnGen0010 versus EnGen0011, ST 26), and the same PBP5 sequence was found in strains with different MICs and in multiple STs (e.g., EnGen0021 versus EnGen0011). Despite the lack of an absolute correlation between the sequence of the PBP5 proteins and the increase in ampicillin resistance, the previously predicted role (11, 12, 13) of certain amino acid changes in elevating the ampicillin MIC was also observed in the most resistant strains, discussed below.

We observed that the majority of the analyzed strains that clustered in subclade A2, based on the classification of Lebreton et al. (8), have mixed PBP5-S/R amino acid sequences with 13 or more amino acid positions matching the PBP5-R consensus (Table 2). In addition, we found one A2 strain, EnGen0025, which had the 21 amino acids corresponding to the PBP5-R consensus, including the methionine-to-alanine alteration at position 485 (Table 2); this position is in close proximity to the active site of the enzyme (12) and has been shown to be important in increasing ampicillin MICs (5, 11). As previously demonstrated (5, 11), the addition of a serine after position 466 (Ser-466′) or aspartic acid (Asp-466′), located in a loop that forms the outer edge of the active site, was found in strains with the highest levels of resistance to ampicillin (EnGen0043 and EnGen0025, respectively). In contrast, and in accordance with the work of Rice et al. (11), a valine substitution at position 629 (Val-629) when present by itself (not combined with Ser-466′) appears to have a low impact on increasing the MIC of ampicillin and was observed in susceptible strains (e.g., TX2050 from clade B) and in low-level ampicillin-resistant strains (e.g., EnGen0024 and TX0034 from clade A). Of note, when we extended our analysis to all the sequenced E. faecium strains, we found that Ser-466′ was found only in strains with Val-629.

Interestingly, when analyzing the PBP5 sequence of the strain EnGen0052 (MIC of 128 μg/ml), we did not observe any of the amino acid changes usually associated with high ampicillin MICs (11, 12, 13). This strain showed PBP5 amino acid variations in five additional positions, outside the 20 to 21 consensus positions: three of these (Asn-39, Ile-314, and Ala-406) were also found in other PBP5-S/R variants, but two amino acid variations (Glu-509 and Phe-606) are unique to this strain. This raises the possibility that these changes, alone or in combination with the PBP5-S/R backbone, could also play a role in elevating the MIC of ampicillin. We also observed in some strains from the PBP5-S/R group (e.g., EnGen0048 and EnGen0011) the combination of a serine at position 401 and an isoleucine at position 499. Even though this combination was not associated with specific ampicillin MICs, it was observed only in PBP5-S/R variants.

We found it intriguing that the well-characterized strain TX16 (DO, MIC of 16 μg/ml), which belongs to CC17 of the A1 subclade (14), had a mixed PBP5 sequence, PBP5-S₄/R₁₇, and is the only A1 strain of more than 70 A1 strains examined that had Met-485 (found in the PBP5-S consensus) rather than Ala- or Thr-485, found in all other A1 strains. This strain was isolated in 1991, and its PBP5 pattern appears to be a late transition form in the PBP5-S/R evolution.

As can be inferred from these observations, our ability to state which amino acid changes in PBP5 are essential for resistance is still limited. Other factors besides PBP5, including the upstream genes, have also been implicated as playing a role in the level of resistance (11, 15). The fact that one of the A2 strains that showed the highest level of resistance (EnGen0025) is closely related to the A1 subclade (8) underlines the potential contribution of the host background toward increasing the ampicillin MICs. In a framework where the species E. faecium is responsible for close to 40% of HA enterococcal infections in the United States (16), further investigations are needed to fully understand the role of changes within PBP5 variants and their linkage to β-lactam resistance.

Nucleotide sequence accession numbers.

pbp5 sequences sequenced in this study have been deposited at GenBank (GenBank, National Center for Biotechnology Information, http://www.ncbi.nlm.nih.gov/) under the accession numbers KJ742831 for E. faecium TX1310, KJ742832 for E. faecium TX1441a, KJ742833 for E. faecium TX0338, KJ742834 for E. faecium TX1333, and KJ742835 for E. faecium TX0334.