Mechanisms of Linezolid Resistance among Staphylococci in a Tertiary Hospital

Abstract

The mechanisms of linezolid resistance among 86 staphylococcal isolates from two intensive care units were investigated. The most frequent was the G2576T mutation in the 23S rRNA (82%). The cfr gene was found in 17% of the isolates, seven S. aureus and eight S. epidermidis isolates. Four of the S. epidermidis isolates had the G2576T mutation and the cfr gene. In four S. haemolyticus isolates, the mechanism could not be identified.

TEXT

Linezolid is an oxazolidinone with antimicrobial activity against Gram-positive bacteria indicated for the treatment of infections by multiresistant Staphylococcus aureus and coagulase-negative staphylococci, as well as penicillin-resistant Streptococcus pneumoniae and vancomycin-resistant enterococci (VRE) (1). Activity results from binding to the 23S rRNA in the 50S ribosomal subunit (1, 2), and resistance arises most frequently from a G-to-T mutation at position 2576 of the 23S rRNA (Escherichia coli numbering) (1). Other mutations in the 23S rRNA (G2447T, T2500A, and C2534T) have been found in clinical and laboratory-derived staphylococcal isolates (3, 4), as well as mutations in the L3, L4, and L22 ribosomal proteins (5–7). Another mechanism involves the cfr gene, which codes for an adenine methyltransferase that modifies adenosine at position 2503 in the 23S rRNA (8).

Linezolid use started in our hospital in 2002, and in 2005–2006 a small outbreak of linezolid-resistant Enterococcus faecalis was detected in two intensive care units (9). In subsequent years, linezolid-resistant staphylococci appeared in the same units and later in other hospital areas. This study focused on isolates from two intensive care units, which are called REA and ICU here and are described in reference 9.

Between 2005 and 2009, 256 linezolid-resistant staphylococci were isolated from different hospital units (2.5% of the total number of staphylococcal isolates). Among the resistant isolates, the most abundant were Staphylococcus epidermidis (43.4%) and S. haemolyticus (35.4%). There were also sporadic isolates of S. hominis (7.3%), and S. aureus (2.8%) (Table 1). Eighty-six staphylococcal isolates were obtained from the two ICUs (18% of the ICU staphylococcal isolates). Samples included blood, catheters, and cerebrospinal fluids (four patients had spinal catheters). Of these, 66 were obtained from 46 ICU patients and 20 were from 13 REA patients. Only isolates considered clinically significant (grown in at least two out of three culture bottles) were included. One isolate per event was considered, with the exceptions of four pairs of isolates obtained from blood and catheter from the same event. Events were considered independent if they were separated by 1 month or more.

Table 1.

Numbers of linezolid-resistant staphylococci isolated in ICUs during the study period

Yr	S. aureus			S. epidermidis			S. haemolyticus			S. hominis
	No. of Linr isolates	%a	Total	No. of Linr isolates	%a	Total	No. of Linr isolates	%a	Total	No. of Linr isolates	%a	Total
2005	0		115	6	2.3	258	5	31.3	16	2	3.6	55
2006	1	0.7	139	2	0.7	268	11	42.3	26	0		90
2007	1	0.6	164	11	2.1	521	35	47.9	73	3	1.6	187
2008	6	4.0	150	34	5.6	606	26	49.1	53	6	3.8	158
2009	0		124	46	8.7	530	17	63.0	27	6	4.9	122
Total	8	1.0	692	125	4.0	2,183	102	46.0	195	21	3.0	612

^aPercentage of the total number of clinically significant blood and catheter isolates of that species.

Identification and susceptibility testing were done using the Wider (Francisco Soria Melguizo, S.A., Madrid, Spain) and Vitek 2 (bioMérieux Vitek, Marcy-l′Etoile, France) systems. MICs of linezolid were determined by Etest as concentrations resulting in 90% growth inhibition (bioMérieux Vitek, Marcy-l′Etoile, France) and interpreted according to the CLSI guidelines (10). Isolates with linezolid MICs of >4 μg/ml were considered resistant.

The G2576T mutation in the 23S rRNA was investigated by pyrosequencing (9). Nearby mutations were searched for by Sanger sequencing of domains II and V. A search for the cfr gene was done by PCR (11), and mutations in the L3, L4, and L22 ribosomal protein sequences, i.e., the rplC, rplD and rplV genes, were studied by sequencing and comparison with GenBank sequences (12).

Clonal analysis by repetitive extrapalindromic sequence (rep)-PCR DNA fingerprinting using DiversiLab (bioMérieux, Marcy-l′Etoile, France) did not find intraspecific variability. Consequently, different approaches were tested for each species to select methods with discriminatory capacity. SmaI pulsed-field gel electrophoresis (PFGE) was used for S. epidermidis and S. haemolyticus, and ApaI PFGE for S. hominis. Randomly amplified polymorphic DNA (RAPD) was used to study S. haemolyticus and S. hominis, using the primers OPA2 (5′-TGCCGAGCTG-3′), OPI12 (5′-AGAGGGCACA-3′), and OPA18 (5′-AGGTGACCGT-3′). S. aureus was analyzed by spa typing (13) and multilocus sequence typing (MLST) (14), and S. epidermidis was analyzed by PCR of the repeat region of the sdrF gene (15) and MLST (16). The staphylococcal chromosomal cassette mec (SCCmec) types were identified by multiplex PCR (17).

There were seven linezolid-resistant S. aureus isolates. All of them were methicillin-resistant nosocomial isolates (HA-MRSA). In addition, they were resistant to fluoroquinolones, erythromycin, and clindamycin. They belonged to sequence type ST125 and carried the cfr gene. Their linezolid MICs ranged from 8 to 16 μg/ml. One isolate obtained in 2007 had spa type t109 and SCCmec type I. Six clustered isolates obtained in 2008 had spa type t067 and SCCmec type IV. None of them had mutations in the 23S rRNA, although a single methicillin-sensitive S. aureus isolate carrying the G2576T mutation was found later in an unrelated hospital area (18). Linezolid resistance could be transferred by conjugation to a rifampin-resistant S. aureus recipient strain (ATCC 29213) that acquired resistance to clindamycin and linezolid.

Among the coagulase-negative staphylococci, the mutation G2576T was detected in 67 of 79 isolates, and the cfr gene was detected in eight S. epidermidis isolates, four of which also had the G2576T mutation. Most of the isolates were multiresistant, with only two S. haemolyticus isolates being susceptible to levofloxacin and three S. hominis isolates being susceptible to methicillin. All the isolates were susceptible to vancomycin, daptomycin, teicoplanin, and quinupristin-dalfopristin.

Both PFGE and sdrF typing identified a major group of S. epidermidis that included 28 SCCmec type III isolates that had the G2576T mutation in all the alleles of the 23S rRNA gene. Four isolates from this group also had the cfr gene. Besides PFGE, sdrF and SCCmec typing identified six types, five of them represented by a single isolate each (Table 2). The sixth type included three isolates with pulsotype 2, sdrF type IV, and SCCmec type IV, two of them having the G2576T mutation in a 3:2 wild-type-to-mutant ratio and the third having the mutation in all the alleles. MLST analysis of selected isolates showed that the major group belongs to ST2 (CC2), while the pulsotype 2 group IV isolates belong to ST23 (CC23). Linezolid resistance has been observed previously in isolates from these sequence types (19–21).

Table 2.

Numbers of linezolid-resistant staphylococci isolated in ICUs during the study period, according to molecular typing and linezolid resistance mechanisms

Species	PFGE pulsotype	sdrF type	RAPD pattern	spa type	SCCmec type	No. of isolates with linezolid resistance mechanism
						G2576T	cfr	G2576T + cfr	NTa
S. aureus				t109	I		1
				t067	IV		6
S. epidermidis	1	I			III	24		4
	1	II			III	1
	1	III			I		1
	1	V			NT		1
	2	II			III		1
	2	IV			IV	3
	3	III			I		1
S. haemolyticus	1		1		I	21			2
	1		1		V	6
	1		3		V				1
	1		4		I	1
	2		1		I	1
	3		1		I				1
	4		1		I	1
	4		4		V	1
			2		V	1
S. hominis	1		1		NT	5
	1		1			3

^aNT, nontypeable.

Similarly, PFGE and RAPD analyses of S. haemolyticus isolates showed a major group divided into two subgroups by their SCCmec type: 26 had SCCmec type I and nine had SCCmec type V. Among them, 31 isolates had the G2576T mutation in all five alleles and four had neither the G2576T mutation nor the cfr gene (Table 2). PFGE, RAPD, and SCCmec typing identified six additional types, all of them represented by a single isolate each. Four of these had the G2576T mutation in all the alleles, while in the other two a mechanism for linezolid resistance could not be identified. The four isolates with no known resistance mechanism were further investigated by sequencing domains II and V of the 23S rRNA genes and the L3, L4, or L22 ribosomal protein gene, but no mutations were found.

All of the eight S. hominis isolates were identical by both PFGE and RAPD; three of them were methicillin sensitive, and five were resistant, but their SSC could not be typed. Six of the isolates had the G2576T mutation in a 1:4 ratio of wild type to mutant, while the other two had the mutation in all the alleles.

The MICs of these isolates span the full range of the standard assays, from 8 to >256 μg/ml for S. epidermidis and 8 to 64 for S. haemolyticus and S. hominis. No relation between the MIC and the number of mutant alleles was observed, and even among the group of S. epidermidis isolates having the G2576T mutation in all five alleles of the 23S rRNA gene, the MICs span the full range of values. Four S. epidermidis isolates from group I with MICs of >256 were further investigated by sequencing, but no additional mutations in the 23S rRNA gene or the L3, L4, and L22 genes were found. A correlation between MICs and numbers of mutant alleles has been found after selection of isogenic lines under laboratory conditions (22, 23), but the relation is less clear when independent clinical isolates are analyzed (24). The heterogeneity of MICs even among isolates with the same number of mutant alleles suggests that there might be additional unidentified factors affecting linezolid susceptibility.

The cfr methyltransferase gene was associated with different S. aureus and S. epidermidis clones and was shown to be transferable at least among S. aureus, suggesting that cfr might have spread by transfer between different clones. It is known that resistance plasmids can be transferred among S. aureus and S. epidermidis in the context of hospital outbreaks (25), and the spread of the cfr gene from S. aureus to S. epidermidis in a patient was recently reported (26).

The diversity of species and clones found suggests that linezolid resistance among Gram-positive cocci in our hospital resulted from several independent selection events followed by the expansion of a few clones. This suggests that both antibiotic selection pressure and cross transmission have played a role in the local emergence and spread of the resistant clones. After reaching a maximum in 2009, the number of resistant isolates decreased in 2010 and 2011. The reasons for this rapid decrease are unclear, but improved linezolid use might have been an important factor (27). Global linezolid consumption in the hospital has stayed stable at around 6,000 defined daily doses (DDD) per year, but there might be significant differences among different wards (27). Other factors, like the reinforcement in infection control measures, increased awareness of the problem among clinicians, and the development of a program to prevent catheter-related infections in ICUs, might have contributed.