Clostridium difficile Isolates with High Linezolid MICs Harbor the Multiresistance Gene cfr

Abstract

We studied the molecular mechanisms of linezolid resistance in 9 isolates of toxigenic Clostridium difficile with high linezolid MICs. The activity of linezolid was determined against 891 clinical isolates of toxigenic C. difficile. The MIC₅₀ and MIC₉₀ of linezolid were 0.75 μg/ml and 1.5 μg/ml, respectively. Nine strains (1%) showed high linezolid MICs (6 μg/ml to 16 μg/ml) and also were resistant to clindamycin, erythromycin, and chloramphenicol. These strains were selected for molecular studies: sequencing of domain V of the 23 rRNA gene, detection of the cfr methyltransferase gene, and sequencing of the ribosomal protein genes rplC and rplD. Molecular relatedness between strains was assessed using PCR ribotyping and MLVA (multilocus variable-number tandem-repeat analysis) typing. The strains belonged to ribotypes 001 (2/9), 017 (6/9), and 078 (1/9). MLVA showed that strains of ribotype 001 and 017 belonged to the same clonal complex in each ribotype. We did not detect mutations in the 23S rRNA gene. The cfr gene was detected in 7 of 9 strains. Sequencing of cfr amplicons revealed a similarity of 100% to a fragment of transposon Tn6218 of C. difficile, which was annotated as a putative chloramphenicol/florfenicol resistance protein. We were unable to detect mechanisms of resistance to linezolid in the 2 strains belonging to ribotype 001. While the relevance of our results lies in the detection of the cfr gene as a possible mechanism of resistance to linezolid in C. difficile, our findings should be assessed by further investigations to characterize these possible cfr genes and their contribution to linezolid resistance.

INTRODUCTION

Linezolid is the first antibiotic in the oxazolidinone group and has been used for more than 10 years to treat Gram-positive infections (1). Resistance to linezolid in Gram-positive bacteria is very uncommon and has been reported mainly in clinical isolates of coagulase-negative staphylococcus, Staphylococcus aureus, Enterococcus faecalis, and Enterococcus faecium, as well as in nonhuman isolates of other Gram-positive bacteria and Gram-negative bacteria (2). Resistance is due to the presence of 1 or more of 3 mechanisms: point mutations in the 23S rRNA gene (central loop of domain V, mainly G2576T), mutations/deletions in ribosomal proteins L3 (rplC gene) and L4 (rplD gene), and methylation of position A2503 of the 23S rRNA gene mediated by an rRNA methyltransferase encoded by the multiresistance gene cfr (chloramphenicol-florfenicol resistance), which is harbored mainly in transferable plasmids (3). This last mechanism also has been associated with resistance to phenicols, lincosamides, oxazolidinones, pleuromutilins, and streptogramin A antibiotics (4).

Clostridium difficile is the main cause of antibiotic-associated diarrhea. Although linezolid is not used for the treatment of C. difficile infection (5, 6), some observations suggest that patients treated with linezolid for ventilator-associated pneumonia could be protected against this infection (7).

Resistance to linezolid occasionally has been described in clinical isolates of C. difficile (8–10), although possible mechanisms of resistance in this species have not been determined.

Our objective was to study the molecular mechanisms of linezolid resistance in clinical isolates of toxigenic C. difficile with high linezolid MICs.

MATERIALS AND METHODS

Microbiological methods.

The study sample comprised 891 human clinical isolates of toxigenic C. difficile recovered during 2010 to 2011 at our hospital. The strains were isolated from feces in CLO agar (bioMérieux, Marcy l'Etoile, France) under anaerobic conditions at 37°C for 48 h and identified by conventional methods. C. difficile toxin genes (tdcA conserved fragment, tcdA deleted fragment, tcdB, cdtA, and cdtB) were detected in isolates by multiplex PCR by following a method adapted from other authors (11, 12).

Susceptibility to linezolid, erythromycin, chloramphenicol, and clindamycin was determined after anaerobic subculture in Brucella agar at 37°C using Etest strips incubated for 48 h. The breakpoints were 8 μg/ml for erythromycin and clindamycin and 32 μg/ml for chloramphenicol (13). Strains with a linezolid MIC of ≥4 μg/ml were further characterized. The identification of linezolid-resistant strains was confirmed by 16SrRNA gene PCR and sequencing.

Molecular methods. (i) DNA extraction.

Chromosomal bacterial DNA was extracted using Chelex resin (InstaGene Matrix; Bio-Rad).

(i) PCR amplification and sequencing of domain V of the 23 rRNA gene.

The PCR and sequencing technique was performed as previously described (14). PCR products were sequenced using the BigDye Terminator method and detected in an ABI Prism 3100Xl automatic DNA sequencer (Applied Biosystems Inc., Foster City, California, USA). The sequences were compared using BioEdit, version 7.0.0, with wild-type reference strains of C. difficile ribotype 001 (ATCC 9689), C. difficile ribotype 017 (NCTC 13287), and C. difficile ribotype 078 (our collection was typed in a reference laboratory in Leiden, The Netherlands). In order to rule out the presence of heterozygous G2576T mutants not detected by sequencing, amplicons were digested with NheI endonuclease as previously described (14).

(ii) Detection of the cfr gene by PCR.

The presence of the cfr gene was investigated by PCR using primers cfr-Cd-ex-F (5′-TCC TCT ACG GCA AAC AAA CC-3′) and cfr-Cd-ex-R (5′-GCT CCA CTT GAG TGA TGC CTA-3′), which were designed specifically for this study. The identity of amplicons (approximately 1,200 bp) was assessed by sequencing with the same primers and primers cfr-Cd-F (5′-TGA AAT ATA AAG CTG GTT GGG AGT CA-3′) and cfr-Cd-R (5′-TCC ATA CAA TTG ACC GCA AGC AGC-3′). Sequences were compared to those deposited in GenBank using the BLAST tool (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

(iii) PCR amplification and sequencing of rplC and rplD.

The ribosomal protein genes rplC (ribosomal protein L3) and rplD (ribosomal protein L4) were amplified and sequenced using primers L3-CdF (5′-AAAAGTTGGTATGACTCAAATATT CAC-3′), L3-CdR (5′-AGTTACTACTGAACCTTTAGCTCCTG-3′), L4-CdF (5′-GGAGGAATAACAATGCCAAAA-3′), and L4-CdR (5′-ATGCGTACACCTCCTCCACT-3′), which were designed specifically for this study. In order to find possible mutations, the sequences obtained were compared to those of linezolid-susceptible reference wild-type C. difficile strains of ribotypes 001, 017, and 078 analyzed in parallel to the C. difficile strains studied and with sequences of the rplC and rplD genes of C. difficile 630, which were deposited in GenBank (15) under accession numbers NC_009089 (region: 102884 to 103513) and NC_009089 (region: 103543 to 104166), respectively.

(iv) Molecular typing.

C. difficile strains were typed by PCR ribotyping (16), and the strains belonging to the same ribotype were subtyped by multilocus variable-number tandem-repeat analysis (MLVA) (17).

Nucleotide sequence accession numbers.

Sequences from this study have been deposited in GenBank under accession numbers KM359438 and KM359439.

RESULTS AND DISCUSSION

During a study of the activity of linezolid against 891 clinical isolates of toxigenic C. difficile, we detected 9 strains (1%) from 9 different patients against which the linezolid MIC was ≥4 μg/ml (Table 1). All of these strains also were resistant to other ribosome-inhibiting antimicrobials, namely, clindamycin, erythromycin, and chloramphenicol. The MIC₅₀ and MIC₉₀ of linezolid against the 891 isolates were 0.75 μg/ml and 1.5 μg/ml, respectively.

TABLE 1.

Genotypic and phenotypic characteristics of the isolates of Clostridium difficile studied

Strain	Ribotype	Date of isolation (day/mo/yr)	23S rRNA gene mutationa	Mutations in the ribosomal protein genesa		cfr detection by PCR	MIC,c μg/ml (Etest)				MLVA typingb
				rplC (L3)	rplD (L4)		Lin	Eryth	Chlo	Clin	A6	B7	C6	E7	F3	G8	H9	STRs
10047605	001	19/03/2010		A567G, G342T		−	16	>256	48	>256	29	9	15	5	6	6	1	71
10057673	001	07/04/2010		A567G, G342T		−	16	>256	32	>256	29	9	15	5	6	6	1	71
10107643	078/126	28/07/2011		G90A, G342T, T486C, A531G, C543T	C102T, C288T, G312A, G487A, C567T	+	16	>256	>256	>256
11088756	017	10/05/2011		T486C, G342T		+	6	>256	>256	>256	3	7	21	8	6	40	2	87
11093768	017	16/05/2011		T486C, G342T		+	8	>256	>256	>256	3	7	21	8	6	40	2	87
11140508	017	03/08/2011		T486C, G342T		+	12	>256	>256	>256	3	8	22	8	6	40	2	89
11150587	017	24/08/2011		T486C, G342T		+	12	>256	>256	>256	3	8	21	8	6	40	2	88
11172819	017	03/10/2011		T486C, G342T		+	12	>256	>256	>256	3	7	21	8	6	40	2	87
11223280	017	22/12/2011		T486C, G342T		+	6	>256	>256	>256	3	7	22	8	6	41	2	89
NCTC 13287	017			T486C, G342T		−	1.5	>256	3	>256
078 control	078			G90A, G342T, T486C, A531G, C543T	C102T, C288T, G312A, G487A, C567T	−	0.75	>256	4	>256
ATCC 9689	001			A567G, G342T		−	1.5	1.5	6	3

^aCompared with sequences of C. difficile 630.

^bNumber of tandem repeats for each locus. STRs, summed tandem repeats. Differences in summed tandem repeats of ≤2 indicate the same clonal complex.

^cLin, linezolid; Eryth, erythromycin; Chlo, chloramphenicol; Clin, clindamycin.

Linezolid resistance is not frequent in clinical Gram-positive isolates, and in large-scale studies, mainly in Europe and the United States, linezolid-resistant strains account for <1% of all isolates (18, 19). Mutations in the 23S rRNA gene and in ribosomal proteins traditionally have been considered the main mechanism of resistance to linezolid (3). In our study, we did not detect mutations in the 23S rRNA gene of C. difficile isolates; however, since we did not sequence each copy of this gene separately, mutations in a few copies of the 23S rRNA gene may have gone undetected. With respect to ribosomal proteins L3 and L4, the C. difficile strains studied differed from those of C. difficile strain 630, although these sequence changes were considered characteristic of each ribotype and not related to linezolid resistance, since, as previously described (8), they were present in wild-type reference strains belonging to ribotypes 001, 017, and 078 and not only in the linezolid-resistant strains studied (Table 1).

The molecular investigation of other possible mechanisms of resistance revealed the presence of the cfr gene in 7 of the 9 strains (6 from ribotype 017 and 2 from ribotype 078) (Table 1). The sequences of the cfr fragments amplified showed 2 different profiles with a 2-bp difference (GenBank sequence accession numbers KM359438 and KM359439). The alignment of these sequences (approximately 1,050 bp) showed 100% similarity to a fragment of transposon Tn6218 of 2 different strains of C. difficile (GenBank accession numbers HG002396.1 and HG002389.1) coding for a putative chloramphenicol/florfenicol resistance protein (20) (GenBank sequence accession numbers CDF47162.1 and CDF47263.1) and 89% similarity to the rRNA large subunit methyltransferase cfr of Bacillus amyloliquefaciens (GenBank accession number HG328254.1). These results indicate that most of our linezolid-resistant strains (7/9 strains) harbor a cfr methyltransferase gene located in mobile genetic elements. In this sense, transposons harboring resistance genes have been detected in C. difficile (8, 15, 20), although their relationship with linezolid resistance has not yet been reported in this pathogen.

In recent years, cfr methyltransferases have been described in clinical isolates of linezolid-resistant coagulase-negative staphylococci and S. aureus, and some even have been associated with outbreaks in the intensive care unit (21, 22). The cfr gene was first described in S. sciuri of animal origin and since then has been described in several Gram-positive bacterial species, including Enterococcus species and Bacillus species (2), although never in C. difficile.

In order to know whether other C. difficile strains belonging to ribotypes 017 and 078 could be cfr positive without linezolid resistance, we analyzed 10 isolates of each ribotype (078 and 017) with a linezolid MIC of <4 μg/ml recovered in our hospital during the same period. None of these linezolid-susceptible strains was cfr positive, and all of them were chloramphenicol susceptible (MIC, <4 μg/ml). The MICs for clindamycin and erythromycin were variable, ranging from 1.5 to >256 μg/ml and from 0.5 to >256 μg/ml, respectively.

Although linezolid-resistant C. difficile strains are not frequent, their selection and transmission could lead to an increase in C. difficile infection mainly in clinical settings where linezolid is used frequently. In our study, MLVA demonstrated that all cfr-positive ribotype 017 C. difficile isolates belonged to the same clonal cluster (summed tandem repeat differences, ≤2), as was the case for the strains of ribotype 001 (cfr negative) (Table 1), revealing possible horizontal transmission of these strains among patients in our hospital. A chart review did not reveal previous treatments with linezolid or other drugs acting in the ribosome or a clear contact between those patients during admission.

Linezolid is not used to treat C. difficile infection, although the new oxazolidinone cadazolid is being evaluated as a promising therapeutic option to treat C. difficile infection (23, 24). Cadazolid has shown very good activity against C. difficile strains (including some linezolid-resistant strains) in an in vitro gut model (23); however, in the studies cited, the molecular mechanisms of resistance to linezolid were not reported. It would be interesting to test cadazolid against the C. difficile isolates harboring the cfr gene that we studied.

In conclusion, we studied possible molecular mechanisms of linezolid resistance in C. difficile and describe the presence of the cfr gene in linezolid-resistant isolates. The relevance of our results should be confirmed by further investigations to characterize the molecular environment of these cfr genes of C. difficile, mainly their role in mediating linezolid resistance, as well as in resistance to other antimicrobials acting in the ribosome, such as chloramphenicol and clindamycin. Furthermore, it would be interesting to investigate other possible mechanisms of linezolid resistance, since, with the methods used, we were unable to detect any mechanisms of linezolid resistance in 2 strains from ribotype 001.