Disruption of an Enterococcus faecium Species-Specific Gene, a Homologue of Acquired Macrolide Resistance Genes of Staphylococci, Is Associated with an Increase in Macrolide Susceptibility

Abstract

The complete sequence (1,479 nucleotides) of msrC, part of which was recently reported by others using a different strain, was determined. This gene was found in 233 of 233 isolates of Enterococcus faecium but in none of 265 other enterococci. Disruption of msrC was associated with a two- to eightfold decrease in MICs of erythromycin azithromycin, tylosin, and quinupristin, suggesting that it may explain in part the apparent greater intrinsic resistance to macrolides of isolates of E. faecium relative to many streptococci. This endogenous, species-specific gene of E. faecium is 53% identical to msr(A), suggesting that it may be a remote progenitor of the acquired macrolide resistance gene found in some isolates of staphylococci.

We have recently screened a number of gram-positive cocci for macrolide susceptibility (12) and subsequently for some of the acquired resistance genes, erm(A), erm(B), erm(C), ere(A), ere(B), mef(A), and msr(A) (24, 29), that are known to effect macrolide, streptogramin, and/or lincosamide (MS/L) susceptibility (unpublished data). We noted that some of the Enterococcus faecium isolates for which the MICs (2 to 16 μg/ml) of erythromycin (ERY) were elevated failed to hybridize to any of the aforementioned resistance gene probes. We next performed PCR amplification using DNA from macrolide nonsusceptible, probe-negative E. faecium strains and primers for msr(A/B) (29); a fragment with homology to msr(A) and msr(B), the acquired macrolide resistance genes found in staphylococci, was recovered. In the current work, the complete sequence (1,479 nucleotides [nt]) of the gene encompassing this fragment along with ∼450 bp upstream was determined. This gene was found to contain the 405-bp fragment previously deposited in GenBank (accession no. AJ243209) and recently reported as msrC, a species-specific gene of E. faecium (22). An insertion disruption mutation of this gene has now been generated and the E. faecium mutant was found to be more susceptible to ERY, azithromycin, tylosin, and quinupristin, suggesting that this msr-like gene can confer some protection to isolates of E. faecium against these antimicrobials.

MATERIALS AND METHODS

Bacterial strains and MIC studies.

The microorganisms used in this study were obtained from the collection of our laboratory over the past several years. A total of 498 isolates of enterococci, 56 streptococcal isolates (some of which were previously described) (5, 12), and two staphylococcal isolates (as negative controls for the gene described) were used in the various studies. The majority of these clinical isolates came from the United States but some were from Thailand, Argentina, Belgium, and Spain. The enterococcal isolates included 246 Enterococcus faecalis, 233 E. faecium, 6 E. hirae, 5 E. durans, 2 E. casseliflavus, 2 E. mundtii, 2 Staphylococcus aureus, 1 E. gallinarum, 2 E. solitarius, and 1 E. raffinosus isolate. ERY MICs were also determined by agar dilution (19, 20) for a group of 90 E. faecalis, 64 E. faecium, 29 Streptococcus pyogenes, 10 group B streptococci, and 17 Streptococcus pneumoniae isolates. ERY, kanamycin (KAN), and tylosin were purchased from Sigma Chemical Co., St. Louis, Mo., and quinupristin was provided by Rhone-Poulenc Rorer Pharmaceuticals, Inc., Collegeville, Pa.

DNA extraction, PCR, sequencing, and cloning.

E. faecium isolate SE34 (TX1330) (MIC of ERY, 0.25 to 0.75 μg/ml) was used as a recipient strain; it was recovered from the feces of a healthy community volunteer (5) and has been used in our lab because it lacks resistance to most agents tested and is transformable by electroporation. DNA extraction (32) and PCR were done using the PCR Optimizer kit (Invitrogen, San Diego, Calif.); PCR products were analyzed by automated DNA sequencing at the Microbiology and Molecular Genetics core facility, University of Texas Medical School, Houston, Tex. Parts of the sequence described in this study were generated using msr(A/B) primers (29) (primer I, +5′-GCA AAT GGT GTA GGT AAG ACA ACT-3′ and primer II, −5′-ATC ATG TGA TGT AAA CAA AAT-3′) and other sequence parts were generated by inverse PCR and octamer primer of the Rad Prime labeling kit (Gibco BRL, Grand Island, N.Y.) using DNA from TX2465 (16), TX2597, and TX2046 (15). These three E. faecium clinical isolates (ERY MIC, 2 to 16 μg/ml) were chosen arbitrarily as examples of nonsusceptible isolates that were negative with erm(A), erm(B), erm(C), ere(A), ere(B), mef(A), and msr(A) probes (henceforth referred to as MS/L probe-negative isolates). The sequence of the msrC coding region using DNA from TX1330 was also determined in later experiments using specific primers designed from the other sequences. Sequence analysis was done using the BLAST network service of the National Center for Biotechnology Information. The GCG software package (Genetics Computer Group, Madison, Wis.) was used to compare similarities among other sequences. Filter matings were performed using E. faecium GE-1 (7), which is tetracycline (TET) resistant, as a recipient strain. Cloning was done with standard methods (26) by using Sau3A-digested genomic DNA from TX2465, TX2597, and TX2046 E. faecium isolates and using pBluescript vector and Escherichia coli DH5α cells.

Disruption mutation in msrC of E. faecium.

In order to construct the disruption mutation in the msrC gene, we generated a 628-bp intragenic DNA fragment (nt 1251 to 1879; see Fig. 1) by PCR from TX2465, one of the macrolide nonsusceptible, MS/L probe-negative E. faecium strains, and cloned it into the pCR2.1 vector of the TA Cloning kit (Invitrogen), resulting in pTEX5259. The fragment was recloned into the previously published pBluescript derivative pTEX4577, containing aph(3′)-IIIa (8, 28), resulting in pTEX5259.03. Plasmid pTEX5259.03 DNA was electroporated into electrocompetent cells of TX1330 (9, 11) and selection was made on Todd Hewitt agar (Becton Dickinson, Cockeysville, Md.) supplemented with 0.25 M sucrose and KAN at 6,000, 8,000, and 12,000 μg/ml. The resulting colonies were restreaked on KAN and analyzed by susceptibility testing to ERY, and quinupristin alone by broth microdilution method using twofold dilutions or smaller increments of antibiotic concentrations (19, 20). Susceptibility to ERY was also determined by E-test (PDM Epsilometer test; AB Biodisk North America, Inc., Piscataway, N.J.). A growth-curve study comparing the wild-type TX1330 and the msrC disruption mutant was done using Mueller-Hinton II broth (Becton Dickinson) and measuring optical density at 600 nm hourly and CFU at 0, 6, 12, and 24 h, on brain heart infusion (BHI) agar (Difco Laboratories, Detroit, Mich.) for TX1330 and on BHI and BHI with KAN (to detect possible revertant colonies) for the disruption mutant. Recombinant colonies were further analyzed by pulsed-field gel electrophoresis (18) of SmaI digestion products of the genomic DNA and by hybridization to confirm the expected disruption of the gene. The 498 enterococci and 2 staphylococcal isolates were tested for the presence of msrC by colony lysate hybridization under high stringency conditions with the 628-bp intragenic DNA probe, using previously published methods (27). To test for the presence of this gene on plasmid or total genomic DNA, DNA gels were Southern blotted and filters were hybridized with this probe under high stringency conditions.

FIG. 1.

Complete nucleotide sequence from strain TX2465 of msrC and its upstream region. Shown are the putative promoter sequences (−10 and −35, underlined), a possible polypeptide (15 aa) and five inverted repeat regions in the upstream region (indicated by arrows), the predicted amino acids (with one-letter code), SD sequence (in bold letters), and stop codons (indicated by asterisks). Possible WA and WB regions are shown in boxes with bold letters, and the possible Q-linker region is in brackets. Conserved SGG sequences are shown with bold letters and underlined.

Nucleotide sequence accession numbers.

The nucleotide sequence of the complete msrC of strain TX2465 (accession no. AY004350) and TX1330 (accession no. AF313494) were deposited in GenBank.

RESULTS AND DISCUSSION

PCR amplification using the msr(A/B) primers and DNA from TX2465, TX2597, and TX2046 generated an ∼350-bp DNA fragment from each of these three strains. Then, using inverse PCR and also the octamer primer, we generated an ∼2.4-kb sequence (Fig. 1) from the TX2465 E. faecium isolate. When this sequence was used to search GenBank, it showed the highest homology score with a 405-bp fragment (accession no. AJ243209), recently named msrC (22). The 405-bp fragment is 95% identical to nt 1488 to 1890 of the coding region of the sequence shown in Fig. 1; based on this identity, we consider the current sequence to be the complete sequence of an msrC gene. Since this gene appears to be an endogenous chromosomally encoded gene, we have maintained the format of the gene name as msrC rather than adopt the recent recommendations for acquired macrolide resistance genes [e.g., msr(A)] (24). Analysis of the 2.4-kb sequence of strain TX2465 (Fig. 1) revealed an open reading frame (1,479 bp) with an ATG potential start codon at nt 496, preceded by a putative Shine-Dalgarno (SD) sequence and a TAA stop codon at nt 1972 to 1974. The coding sequence of this msrC gene (1,479 bp) showed 53% identity to msr(A) (1,467 bp), 57% identity to msr(B) (531 bp) over the corresponding region, and 47% identity to vga(A) (1,569 bp) and vga(B) (1,659 bp) (1, 2). The predicted MsrC protein (492 amino acids [aa]) showed similarities to ABC proteins of other gram-positive bacteria [54% similarity to Msr(A) (488 aa) of Staphylococcus epidermidis; 59% similarity over aa 301 to 492 of MsrC, compared to the 176-aa C-terminal region of Msr(B) of Staphylococcus xylosus; 50% similarity to Vga(A) (522 aa); and 46% similarity to Vga(B) (552 aa)] (2, 13, 14, 25). As reported in these references, Msr(A), Vga(A), and Vga(B) contain two ATP-binding domains, each of which in turn contains the two ATP-binding motifs, WA and WB, described by Walker et al. (2, 13, 14, 25, 30). The predicted amino acid sequence of MsrC also contains two homologous ATP-binding domains and, in the region corresponding to these domains of Msr(A), Vga(A), and Vga(B), we detected the presence of the highly conserved SGG sequence found between the WA and WB ATP-binding motifs of the previously investigated proteins (2, 3, 10). The interdomain sequence, called the Q-linker, which separates the two ATP-binding domains of Msr(A) and Vga(B), has been described as being rich in glutamine (2, 25); the corresponding amino acid region of MsrC was also found to be richer in glutamine (14 Q in 138 aa) than the rest of the gene sequence (17 Q in 288 aa). The sequence upstream of msrC in TX2465 showed a short open reading frame encoding a potential polypeptide of 15 aa, 6 of which are identical to 6 of the amino acids of the 8-aa leader peptide of Msr(A). The putative msrC leader peptide was initiated by a potential ATG start codon at nt 222, preceded by an SD sequence with a TAA termination codon at nt 267. The region of nt 1 to 494 (Fig. 1) also contains five possible inverted repeat sequences, one of which surrounds the ribosome binding site immediately preceding the msrC gene, as had been described for msr(A) and suggested to be responsible for the inducible nature of resistance (25). The 8-aa peptide preceding msr(A) and the longer leader peptides for erm(A), erm(B), and erm(C) have been shown to be involved in regulating the expression of these resistance genes (4, 6, 13, 17, 23, 25, 31), although their exact function is not known.

The coding region of msrC from the recipient strain TX1330 was also amplified by PCR and sequenced and, when compared with the msrC nucleotide sequence shown in Fig. 1, showed 95% identity, which resulted in 10 aa changes, 4 of which are in the Q-linker region while 3 and 2 aa changes were found between the two ATP-binding motifs in both of the ATP-binding domains.

We were unable to recover ERY-resistant clones either by cloning of Sau3AI DNA fragments from TX2465, TX2597, and TX2046 or by cloning the whole PCR-amplified gene and ∼500 nt upstream from TX2465 into E. coli; this is consistent with observations that the msr(A) gene from S. aureus does not seem to express resistance in E. coli (13). None of these three E. faecium strains transferred ERY resistance in mating experiments, and msrC appears to be on their chromosomes, as plasmid DNA did not show any hybridization but genomic DNA showed hybridizing bands (data not shown). Hybridization of lysates of 498 enterococcal isolates showed that this gene was present in all 233 E. faecium isolates tested but not in the 265 other enterococcal species or in either of the staphylococci which were used as negative controls.

We next constructed a mutant, following electroporation of pTEX5259.03, which cannot replicate in enterococci, into TX1330. Seven Kan^r colonies were recovered and 5 of these were shown to have aph(3′)-IIIa; all 5 were interrupted in msrC, which was confirmed by pulsed-field gel electrophoresis and hybridization (data not shown). TX1330 and one of the mutant colonies showed almost identical growth curves by hourly determinations of optical density at 600 nm and CFU. There was little to no loss of Kan^r by the cultures, indicating that the insertion was stable during the 24-h incubation period. Mutant colonies showed a decrease in broth microdilution MICs (Table 1) of ERY (from 0.5 to 0.75 μg/ml for TX1330 to 0.06 to 0.09 μg/ml for mutants), of tylosin (from 16 μg/ml for TX1330 to 8 μg/ml for mutants), and of quinupristin (from 96 μg/ml for TX1330 to 48 to 64 μg/ml for mutants), suggesting that this gene provides some protection against these agents. Because of the small difference for quinupristin, we also tested a lower inoculum of 10³ CFU/ml. TX1330 (tested in triplicate) grew in medium with 32 μg/ml but not at 48 μg/ml (MIC, 48 μg/ml), while mutants 1 (in triplicate) and 2 (in duplicate) all grew on 12 μg/ml but not on 16 μg/ml (MIC, 16 μg/ml), further verifying that there is a small but true difference. Mutant colonies also showed a decrease in E-test MICs (Table 1) of azithromycin (from 1.56 μg/ml for TX1330 to 0.38 μg/ml for mutants). E-test MICs of clindamycin and norfloxacin for TX1330 and mutant colonies were almost identical (data not shown).

TABLE 1.

MICs of macrolides (14-, 15-, and 16-membered) for wild-type E. faecium TX1330 and two msrC disruption mutants

Organismc	Erythromycin		Azithromycin	Tylosin	Quinupristin
	MICa	E-test MICb	E-test MICb	MICa	MICa
TX1330	0.75	0.25	1.56	16	96
TX1330	0.5	0.38	1.56	16	96
TX1330	0.75	0.38	1.56	16	96
msrC mutant 1	0.09	0.094	0.38	8	64
msrC mutant 1	0.09	0.094	0.38	8	64
msrC mutant 1	0.06	0.094	0.38	8	48
msrC mutant 2	0.06	0.094	0.38	8	64
msrC mutant 2	0.06	0.094	0.38	8	64

^aBy broth microdilution method. The ERY concentrations tested were 0.04, 0.06, 0.09, 0.125, 0.187, 0.25, 0.37, 0.5, and 0.75 μg/ml; the quinupristin concentrations tested were 32, 48, 64, and 96 μg/ml. 

^bPerformed using inocula as recommended by the manufacturer. 

^cTX1330 and msrC mutant 1 were evaluated in triplicate, and mutant 2 was evaluated in duplicate. 

Neu (21) previously pointed out that isolates of E. faecium often tend to be more resistant to 14- and 16-membered ring macrolides with MICs at which 50% of strains are inhibited of 8 to 16 μg/ml. Among our clinical isolates, many of the E. faecalis (61 of 90) and most of the E. faecium (55 of 64) isolates hybridized with one of the macrolide resistance gene probes tested (unpublished data). Table 2 shows the distribution of ERY MICs found among MS/L probe-negative clinical isolates of E. faecalis, E. faecium, and streptococci. While 20 of 29 MS/L probe-negative E. faecalis isolates required MICs of ERY of ≤1 μg/ml, none of the 9 clinical isolates of MS/L probe-negative E. faecium required MICs of <1 μg/ml (MICs ranged from 2 to 16 μg/ml). Almost all S. pyogenes and group B streptococcal isolates required MICs of ERY of ≤0.125 μg/ml. The 17 S. pneumoniae isolates negative for MS/L probes showed MICs of ≤0.125 μg/ml. While more susceptible isolates of E. faecium do exist, such as the recipient strain used in this study, which was isolated from the feces of a healthy nonhospitalized volunteer (5), the above results indicate that clinical isolates of E. faecium are less susceptible to ERY than are isolates of E. faecalis or of streptococcal species. Whether the higher ERY MICs for MS/L probe-negative clinical isolates of E. faecium are related to changes in the structure or expression of MsrC has not been determined, in part due to the difficulty in generating and selecting targeted mutations in these organisms. We did not determine the role of amino acid changes in the MsrC of TX1330 relative to other E. faecium isolates.

TABLE 2.

Distribution of erythromycin MICs (μg/ml) among MS/L probe^a-negative enterococci and streptococci isolated from clinical sources

Organism (n)	Total no. of strains inhibited at each concn (μg/ml)
	≤0.03	0.06	0.125	0.25	0.5	1	2	4	8	16	32
E. faecalis (29)		2	1	3	5	9	8	1
E. faecium (9)							3	2	2	2
S. pyogenes (29)	7	4	17							1
Group B streptococci (10)			10
S. pneumoniae (17)	1	8	8

^aProbes for erm(A), erm(B), erm(C), ere(A), ere(B), mef(A), and msr(A). 

In conclusion, we have determined the complete sequence of msrC, a species-specific gene of E. faecium, from two strains and have shown that the presence of msrC, or possibly a downstream gene, results in some protection of an isolate of E. faecium against ERY, azithromycin, tylosin, and quinupristin. While in staphylococci the acquired gene msr(A) has been shown to confer resistance to ERY by increasing efflux (2), we have not determined the exact function encoded by the endogenous msrC gene in E. faecium; however, the similarity of MsrC to Msr(A) (54%) suggests that it also mediates efflux. Based on the hybridization results showing species specificity, msrC also appears useful as a means of identifying E. faecium isolates.