Mutations in the DNA Mismatch Repair Proteins MutS and MutL of Oxazolidinone-Resistant or -Susceptible Enterococcus faecium

Abstract

Mutations in mutS and mutL, which encode DNA mismatch repair (MMR) proteins, can confer hypermutator phenotypes and may facilitate the emergence of mutational antibiotic resistance in bacteria. Linezolid-resistant enterococci (LRE) rarely emerge during therapy and contain mutations in 23S rRNA genes. As enterococci with defective MMR could be prone to the development of oxazolidinone resistance mutations, we investigated 13 clinical isolates of Enterococcus faecium, including 2 LRE, for mutations in mutSL. A 4,944-bp fragment spanning mutSL was sequenced from two pairs of linezolid-resistant (MICs, 64 μg/ml) and linezolid-susceptible (MICs, 2 μg/ml) E. faecium isolates (one pair from Austria and one pair from the United Kingdom) identical by pulsed-field gel electrophoresis. The pairs represented distinct strains in which linezolid resistance had emerged during therapy. The MutSL peptides of all four isolates had amino acid substitutions compared with the sequence of E. faecium strain DO (used for genome sequencing). These were Val352Ile (one pair of isolates only) and Met628Leu in MutS and Leu387Pro, Tyr406Phe, Thr415Ser, Phe427Leu, and Phe565Ile in MutL. The significance of these changes remains unknown; these isolates did not show a demonstrable hypermutator phenotype. The same substitutions were found in two of nine geographically diverse linezolid-susceptible enterococcal isolates; the other seven isolates had MutSL sequences identical to that of strain DO. Multilocus sequence typing revealed that all isolates with alternate MutSL peptides belonged to a distinct lineage of a prevalent E. faecium clonal complex, designated CC17. Further studies are needed to investigate the prevalence of these MutSL mutations and their possible roles in the emergence of E. faecium strains resistant to oxazolidinones and other antibiotic classes.

The MutS and MutL families of peptides are involved in DNA mismatch repair (MMR) and consequently help to replace nucleotides introduced erroneously into DNA during replication; they are also inhibitors of recombination between nonidentical DNA sequences (6, 25, 30). These processes are required for maintenance of genome stability, and homologs of these peptides are found in eukaryotic and prokaryotic cells. Some bacteria contain multiple mutS homologs encoding peptides belonging to two different subfamilies, MutS1 and MutS2, but only the MutS1 family is thought to be involved in MMR (8). Mutations in the mutS and mutL loci have been associated with hypermutability phenotypes in various bacterial species (10, 31) and with certain hereditary cancers in humans (30). Such mutations may play a significant role in the emergence of mutational resistance to various antibiotic classes in bacteria (19, 20).

Resistance to linezolid, the first oxazolidinone antibiotic to be licensed, is rare but has been selected, usually during therapy, in isolates of enterococci (11, 12, 22, 29; R. D. Gonzales, P. C. Schreckenberger, M. B. Graham, S. Kelkar, K. Den Besten, and J. P. Quinn, Letter, Lancet 357:1179, 2001; G. E. Zurenko, W. M. Todd, B. Hafkin, B. Myers, C. Kaufmann, J. Bock, J. Slightom, and D. Shinabarger, Abstr. 39th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 848, 1999) and staphylococci (S. Tsiodras, H. S. Gold, G. Sakoulas, G. M. Eliopoulos, C. Wennersten, L. Venkataraman, R. C. Moellering, and M. J. Ferraro, Letter, Lancet 358:207-208, 2001). In all clinical isolates reported to date, linezolid resistance has been mediated by G2576T mutations in the genes encoding 23S rRNA (29). Enterococci contain multiple copies of these 23S ribosomal DNA (rDNA) genes, and many linezolid-resistant enterococci (LRE) are heterozygous at position 2576, containing both wild-type (G2576) and mutant (T2576) alleles (29). Linezolid-susceptible enterococci (LSE) may also be heterozygous at this position (29), which implies that more than one 23S rDNA gene copy must carry the T2576 mutation before phenotypic resistance is expressed (16) (N. Woodford, unpublished data). It seems likely that LRE emerge via two discrete steps: first, a mutational event to introduce a T2576 mutation into one gene copy, followed by intragenomic events, such as homologous recombination, to ensure the presence of sufficient copies of the mutant form to confer phenotypic resistance (16). We hypothesized that enterococci with defective MMR might be more prone to undergo these mutations and recombination. Therefore, we have investigated 13 clinical isolates of enterococci, including 2 isolates of LRE, for mutations in mutS and mutL, as possible prerequisites for the emergence of the 23S rDNA mutation that confers oxazolidinone resistance.

(This work was presented at the 23rd International Congress of Chemotherapy, Durban, South Africa, June 2003.)

MATERIALS AND METHODS

Bacterial isolates.

Two epidemiologically unrelated pairs of linezolid-resistant and -susceptible isolates of Enterococcus faecium were studied: one pair from a patient in Austria (isolates A1-E1527 and A2-E1528) and the other pair from a patient in the United Kingdom (isolates C1-E1531 and C2-E1532) (1, 11, 29). Isolates A1-E1527 and C1-E1531 were both linezolid resistant (MICs, 64 μg/ml). Isolate A1-E1527 was homozygous for the T2576 23S rDNA mutation, while isolate C1-E1531 was heterozygous, with the mutation present in three of six 23S rRNA gene copies (A. Sinclair, C. Arnold, and N. Woodford, unpublished data). Their respective linezolid-susceptible parent strains, A2-E1528 and C2-E1532 (MICs, 2 μg/ml), were both homozygous for the wild-type G2576 allele (29). The pairs represented two distinct strains in which linezolid resistance had emerged during therapy, as judged by pulsed-field gel electrophoresis (PFGE) of SmaI-digested genomic DNA. Nine other geographically diverse, linezolid-susceptible (MICs, ≤2 μg/ml) E. faecium (LSE) strains were studied as comparators (Table 1).

TABLE 1.

Isolates of E. faecium used in this study and amino acid substitutions identified in MutS and MutL in comparison with the sequence of strain DO, which was used for the genome sequencing project

Strain	Country of origin	Linezolid MIC (μg/ml)a	Comment	Altered amino acid residue							Reference(s)
				MutS			MutL
				352	628	387	406	415	427	565
DOb	United States		Reference strain	Val	Met	Leu	Tyr	Thr	Phe	Phe
A1-E1527	Austria	64	Patient A		Leu	Pro	Phe	Ser	Leu	Ile	11, 29
A2-E1528	Austria	2	Patient A		Leu	Pro	Phe	Ser	Leu	Ile	11, 29
C1-E1531	United Kingdom	64	Patient C	Ile	Leu	Pro	Phe	Ser	Leu	Ile	1, 29
C2-E1532	United Kingdom	2	Patient C	Ile	Leu	Pro	Phe	Ser	Leu	Ile	1, 29
E13	United Kingdom	2	Outbreak strain								13
E155	United States	2	Outbreak strain								4
E300	United States	2	Outbreak strain								7
E470	The Netherlands	2	Outbreak strain		Leu	Pro	Phe	Ser	Leu	Ile	28
E510	Australia	2	Outbreak strain								28
E734	The Netherlands	1	Outbreak strain	Ile	Leu	Pro	Phe	Ser	Leu	Ile	28
E1168	Italy	2	SENTRYb
E1186	Germany	2	SENTRYb
E1193	Spain	2	SENTRYb

^aIsolate A1-E1527 was homozygous for the T2576 23S rDNA mutation; isolate C1-E1531 was heterozygous for this mutation; isolates A2-E1528 and C2-E1532 were homozygous for wild-type G2576 (29).

^bhttp://www.ewi.med.uu.nl/enare/sentry_antimicrobial_resistance_about.html.

Amplification and sequencing of mutSL.

The genome sequence of E. faecium strain DO (http://hgsc.bcm.tmc.edu/microbial/Efaecium/) was interrogated with the sequences of the MutS and MutL peptides of Staphylococcus aureus (14) and Enterococcus faecalis V583 (http://www.tigr.org). The E. faecium mutS1 and mutL loci were identified, and a 5,172-bp sequence was downloaded. The 5,172-bp sequence included 200 bp of flanking sequence upstream and downstream of the reading frames (Fig. 1). The mutSL genes of E. faecium lie adjacent to each other, as noted also for E. faecalis (http://www.tigr.org), S. aureus, and other gram-positive species (21); this suggests that enterococci have active MMR systems.

FIG. 1.

Strategy for sequencing mutSL of E. faecium. A 5,172-bp sequence was identified on contig 79 of E. faecium strain DO, including 200 bp upstream and downstream of the reading frames: mutS spanned nucleotides 201 to 2846, and mutL spanned nucleotides 2863 to 4972. The 4,944-bp region spanning nucleotides 92 to 5035 of the isolates included in this study was sequenced.

A 4,944-bp region, spanning nucleotides 92 to 5035 of the downloaded sequence and including the mutSL genes, was amplified as two overlapping fragments with primers 1F and 1R (5′-CCG CTT GTT GTA GCT TAT AGA AAA C and 5′-ACG TTC AGA TAA TTC TTG GAT TTT G, respectively) and primers 6F and 2R (5′-CTA AAA CAG CCA AAC AGG AG and 5′-CTC TTT TCT TCT TGG TGA TTG, respectively) (Fig. 1). Additional PCRs with primers 2F and 6R (5′-GCT GTA GTT AGT CAA GGA AAT G and 5′-CCT TCT CCG TTA TCA ATC AC, respectively) and primers 5F and 3R (5′-CAG TTT TAC CTA TTT TTG ATC G and 5′-AGA TCA TCA GAT ACT TCC CC, respectively) were used for some isolates. All PCRs were performed in 25-μl volumes consisting of 12.5 μl of HotStar Taq Master mix (Qiagen Inc., Hilden, Germany), 2 μl of the forward primer (10 pmol), 2 μl of the reverse primer (10 pmol), 8.5 μl of Milli-Q water, and 0.5 μl (20 ng) of chromosomal DNA. Cycling conditions included an initial denaturation at 95°C for 15 min, followed by 30 cycles of 94°C for 30 s, 52°C for 30 s, and 72°C for 60 s, with a final extension at 72°C for 7 min.

The mutSL PCR products were purified for sequencing with the Qiaquick PCR Purification kit or the Qiaquick 96 PCR Purification kit (Qiagen Inc.), used in accordance with the instructions of the manufacturer. Sequencing was performed with the primers described above and with others whose sequences were specific for regions internal to the amplified products (data not shown). The reaction mixtures consisted of 1 μl (5 to 20 ng) of PCR product, 1 μl of the mixture from the BigDye Terminator reaction kit (Applied Biosystems, Foster City, Calif.), 7 μl of reaction dilution buffer, 1 μl (5 pmol) of the sequence primer, 4 μl of Q-solution, and 6 μl of Milli-Q water. Cycle sequencing conditions were 25 cycles of 96°C for 10 s, 50°C for 5 s, and 60°C for 4 min. The sequencing reaction mixtures were purified with Sephadex G50 (ICN International, Birsfelden, Switzerland) in 96-well microtiter plates, injected directly out of water, loaded onto an ABI PRISM 3700 DNA analyzer (Applied Biosystems), and analyzed in accordance with the instructions of the manufacturer.

Frequency of mutation to linezolid resistance.

Bacteria were grown overnight in Iso-Sensitest broth (Oxoid, Basingstoke, United Kingdom) with or without 0.5 μg of linezolid per ml. These cultures were again diluted in fresh broth with or without 0.5 μg of linezolid per ml and were grown with shaking at 37°C to late exponential phase. Viable counts were determined, and aliquots were then spread over agar plates containing 6 μg of linezolid per ml, incubated at 37°C, and monitored daily for the appearance of mutant colonies. S. aureus strain RN4220 and its hypermutable derivative, RN4220mutS (with an insertionally inactivated mutS gene) (21, 24), were used as controls. These staphylococci were also investigated for the emergence of resistance to rifampin (50 μg/ml). Experiments were performed on three occasions.

MLST of E. faecium isolates.

To investigate the occurrence of internationally disseminated E. faecium clones, the 13 isolates were compared by multilocus sequence typing (MLST), based on seven housekeeping genes (atpA, ddl, gdh, purK, gyd, pstS, and adk), by previously described methods (9) (http://www.mlst.net). Isolates were clustered by the unweighted pair-group method with arithmetic averages from the matrix of pairwise similarities between the allelic profiles with START software (version 1.0.5; University of Oxford, Oxford, United Kingdom). The BURST program (START software, version 1.0.5) was used in phylogenetic analyses to identify the most likely ancestral type and derivatives that differed by only one or two loci, i.e., single-locus variants and double-locus variants, respectively.

Comparative modeling of E. faecium MutS.

A comparative model of E. faecium MutS was prepared by using the structures of MutS of Escherichia coli (PDB accession number 1E3 M) and Thermus aquaticus (PDB accession number 1EWQ) as templates. The structures were visualized by using the QUANTA modeling package (ACCELRYS, San Diego, Calif.). MODELLER6 software (23) was then used to calculate 10 models of the E. faecium MutS dimer, and the best model was subjected to energy minimization refinement by using the CHARMM program (5).

Nucleotide sequence accession numbers.

The 4,944-bp fragments spanning mutSL of E. faecium strains E13 (whose sequence is identical to that of strain DO) and E510 and linezolid-resistant strains A1-E1527 and C1-E1531 have been assigned GenBank accession numbers AY150295, AY150296, AY150297, and AY150298, respectively.

RESULTS

Sequencing of mutSL.

The mutSL genes of E. faecium DO (http://hgsc.bcm.tmc.edu/microbial/Efaecium/) were identified. The reading frame of mutS was 2,546 bp in length, sufficient to encode 881 amino acids; that of mutL was 2,109 bp, sufficient to encode 702 amino acids (Fig. 1). A 4,994-bp fragment spanning these genes was amplified from 13 isolates and sequenced.

There were no differences between the mutSL sequences obtained within the pairs of linezolid-resistant and -susceptible isolates from Austria (A1-E1527 and A2-E1528) and the United Kingdom (C1-E1531 and C2-E1532), and none of their genes was disrupted nor was predicted to encode a truncated product. However, the MutS and MutL peptides of all four isolates were predicted to have amino acid substitutions compared with the sequence of E. faecium strain DO. Both pairs had a Met628→Leu change in MutS and five identical substitutions in MutL; these were Leu387→Pro, Tyr406→Phe, Thr415→Ser, Phe427→Leu, and Phe565→Ile (Table 1). Isolates C1-E1531 and C2-E1532 additionally had a Val352→Ile substitution in MutS that was not present in isolates A1-E1527 and A2-E1528. The four isolates also shared 7 silent nucleotide changes within mutS and 21 silent changes in mutL (Table 2).

TABLE 2.

Nucleotide changes identified in a 4,994-bp fragment spanning mutSL of E. faecium isolates

Strain	Base at the following nucleotide positiona:
	mutS									mutL
	693	987	1072	1302	1536	1900	2142	2316	2400	2945	2960	2963	2969	2996	3023	3584	3719	3809	3821	3830	3841	3863	3896	3898	3925	3960	4118	4166	4176	4196	4211	4265	4268	4280	4374
E13b	C	C	G	C	C	A	A	C	C	T	G	C	A	T	G	G	T	T	T	C	T	A	C	A	C	T	T	C	T	A	G	T	G	T	T
E155	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
E300	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
E1168	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
E1186	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
E1193	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
E510	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	T	.	.	.	.	.	.	.	.	.	.	.	.
E470	T	T	.	T	A	T	G	T	T	C	T	A	C	G	A	T	C	C	A	T	C	G	.	T	G	C	C	G	C	T	A	C	A	C	A
A1-E1527	T	T	.	T	A	T	G	T	T	C	T	A	C	G	A	T	C	C	A	T	C	G	.	T	G	C	C	G	C	T	A	C	A	C	A
A2-E1528	T	T	.	T	A	T	G	T	T	C	T	A	C	G	A	T	C	C	A	T	C	G	.	T	G	C	C	G	C	T	A	C	A	C	A
E734	T	T	A	T	A	T	G	T	T	C	T	A	C	G	A	T	C	C	A	T	C	G	.	T	G	C	C	G	C	T	A	C	A	C	A
C1-E1531	T	T	A	T	A	T	G	T	T	C	T	A	C	G	A	T	C	C	A	T	C	G	.	T	G	C	C	G	C	T	A	C	A	C	A
C2-E1532	T	T	A	T	A	T	G	T	T	C	T	A	C	G	A	T	C	C	A	T	C	G	.	T	G	C	C	G	C	T	A	C	A	C	A

^aThe nucleotide position is expressed relative to the 4,994-bp fragment sequenced. Boldface indicates the changes resulting in the amino acid substitutions shown in Table 1; dots, wild-type sequence.

^bThe sequence of the 4,994-bp fragment from strain E13 was identical to that of E. faecium strain DO.

The mutSL loci of seven of nine comparator LSE isolates encoded peptides identical to those of E. faecium strain DO, although isolate E510 had a single silent nucleotide change within mutL that was not seen in any of the other isolates (Table 2). However, two isolates, E470 and E734, representing outbreak strains from hospitals in two different cities in The Netherlands, contained amino acid substitutions in both MutS and MutL. The predicted changes in isolate E470 were identical to those in isolates A1-E1527 and A2-E1528, whereas those in E734 were identical to those in C1-E1531 and C2-E1532 (Table 1). Isolates E470 and E734 also had all of the 28 silent nucleotide changes noted within the mutSL genes of isolates A1-E1527, A2-E1528, C1-E1531, and C2-E1532 (Table 2).

Determination of mutation frequencies.

Linezolid-resistant mutants were not obtained (in three experiments) from E. faecium isolates with either wild-type MutSL peptides (isolates E155, E510, E1168, and E1193) or alternate MutSL peptides (isolates A2-E1528, C2-E1532, E470, and E734) or from S. aureus RN4220 and RN4220mutS, indicating mutation frequencies of <10⁻⁹. The frequencies of mutation to rifampin resistance were 10⁻⁷ for S. aureus RN4220 and 10⁻⁶ for RN4220mutS.

MLST of E. faecium isolates.

Isolates C1-E1531 and C2-E1532 (LRE and LSE, respectively, from the United Kingdom) belonged to the same sequence type (ST) as isolates E470 and E734 from The Netherlands, ST16 (Fig. 2A). Isolates A1-E1527 and A2-E1528 (LRE and LSE from Austria) were double-locus variants of ST16, designated ST65. Both ST16 and ST65 belonged to clonal complex 17 (CC-17) (Fig. 2B). CC-17 also included (i) the comparator isolates that belonged to ST17 (isolates E155, E510, E1168, and E1193), which is the putative ancestral ST of this complex; (ii) ST78 (isolate E1186), a single-locus variant of ST17; and (iii) ST18 (isolate E13), a double-locus variant of ST17.

FIG. 2.

MLST analysis of E. faecium isolates. (A) Traditional dendrogram showing similarity between STs. The numbers in parentheses indicate the allele numbers for atpA, ddl, gdh, purK, gyd, pstS, and adk, respectively. (B) Schematic resulting from a BURST program analysis of the data showing STs allocated to CC-17 (represented by the box). Open circles, STs with no amino acid substitutions in MutSL; shaded circles, STs with amino acid substitutions in MutSL. STs differed by a single locus (heavy solid line), two loci (thin solid line), or three loci (dashed line).

DISCUSSION

This is, to our knowledge, the first report of mutations in the mutSL loci of enterococci. Amino acid substitutions were detected in the MutSL proteins of two LRE isolates, the corresponding progenitor LSE isolates, and two epidemiologically unrelated LSE isolates from The Netherlands. The isolates also possessed 28 identical silent nucleotide polymorphisms (7 in mutS and 21 in mutL). These isolates were unrelated, as judged by PFGE (data not shown). However, the pair of LRE and LSE isolates from the United Kingdom (C1-E1531 and C2-E1532, respectively) and both LSE isolates from The Netherlands belonged to a prevalent MLST ST, ST16, while the pair of LRE and LSE isolates from Austria (A1-E1527 and A2-E1528, respectively) belonged to ST65, which is closely related to ST16. ST16 is widely distributed geographically and includes strains resistant and susceptible to glycopeptides (9). At present, 19 different STs are contained within E. faecium CC-17 and are represented by isolates from clinical sites or isolates associated with hospital outbreaks (R. J. Willems, unpublished data). Indeed, 12 of the 13 isolates studied here (excluding isolate E300) belonged to CC-17. However, to date, alternate MutSL proteins have been found only in a distinct lineage in CC-17, represented by ST16 and its descendant, ST65.

The possible significance of the amino acid substitutions observed in E. faecium MutS and MutL must remain speculative. No structures have yet been determined for MutS or MutL homologs of gram-positive bacteria, although structures have been determined for the MutS enzymes of E. coli (15) and T. aquaticus (18) and for a 40-kDa N-terminal fragment (designated LN40) of E. coli MutL (2). As part of this work, a comparative model of E. faecium MutS was prepared. MutS peptides consist of a number of domains, but their ATPase activity, which is critical for the MMR functions, is located in domain V (15, 18). Examination of a published structure-based alignment (18) suggests that E. faecium residue Met628 is necessary for the structural integrity of domain V. However, the significance of the replacement of this residue by leucine (as found in our LRE) is uncertain because leucine occurs at the corresponding position of T. aquaticus MutS and some yeast paralogs (18). The second MutS substitution found in our E. faecium strains, Val352Ile, appears to lie in domain III (18), and its structural and functional significance remains to be determined. So, too, does the significance of the substitutions observed in E. faecium MutL, as, to the best of our knowledge, the relevant structural information for correlating sequence and function does not exist. The substitutions were all located in the structurally unknown C-terminal part of the peptide. This is in contrast to the locations of most of the MutL mutations known to confer hypermutator phenotypes in E. coli and humans, which map in the more conserved N-terminal fragment (2). Undefined residues in the C-terminal part of MutL are thought to contribute to its ATPase activity (2), while others facilitate the formation of MutL dimers, which “lock” MutS into position at a DNA mismatch prior to repair (25, 30). Hence, the substitutions noted in the C-terminal part of E. faecium MutL might affect MMR efficiency if they affect residues crucial for dimerization or ATPase activity. Interestingly, no amino acid changes were observed in the MutL (also known as HexB) peptides of 13 pneumococci; only synonymous nucleotide changes were found at 10 polymorphic sites within mutL (17). It is tempting to speculate that the mutations found in E. faecium MutL may have biological (i.e., functional) significance, but this remains unproven.

Our attempts to generate linezolid-resistant mutants of selected LSE isolates were unsuccessful, even with alternative protocols that have previously generated E. faecalis mutants resistant to the oxazolidinone AZD2563 (27). Linezolid resistance emerges only at low frequencies (26). Therefore, it is possible that those isolates with alternate MutSL peptides (isolates A2-E1528, C2-E1532, E470, and E734) did have heightened mutability but that the frequency of emergence of linezolid resistance still remained below the detection threshold of our experiments. Unfortunately, the E. faecium isolates were multiresistant, which prevented us from investigating other agents to which resistance typically arises via chromosomal mutations, e.g., rifampin, ciprofloxacin, and streptomycin.

In conclusion, even though we identified mutations in MutSL, we found no evidence of hypermutable phenotypes in our E. faecium isolates. A recent study concluded that mutations in the hexAB loci (equivalent to mutSL) of pneumococci did not contribute to hypermutability in the emergence of rifampin resistance (17). There are conflicting data on the role of hypermutability in the emergence of low-level vancomycin resistance in staphylococci (21, 24). Further work is needed to investigate the prevalence of MutS and MutL diversity in enterococci and, in particular, whether the changes found here occur in all E. faecium isolates of ST16 and its descendant, ST65. If so, screening for specific mutations in mutSL by PCR might serve as a simple and fast molecular epidemiological tool with which to study the spread of clinically relevant E. faecium isolates belonging to ST16 and ST65. Although we found no evidence of hypermutability, this lineage may have the capacity to become transiently hypermutable (3) under adverse conditions. Structural studies are therefore needed to determine whether the amino acid substitutions observed have any functional significance and to explore further their possible role in the emergence of E. faecium strains resistant to oxazolidinones and other antibiotic classes.