Detection of Mutations Conferring Resistance to Linezolid in Enterococcus spp. by Fluorescence In Situ Hybridization

Abstract

A fluorescence in situ hybridization (FISH) assay was established to detect linezolid resistance (conferred by the mutation 2576G>T in the gene coding for the 23 string of the ribosomal RNA) in enterococci. The assay was evaluated with 106 Enterococcus isolates; it showed a sensitivity of 100% for the detection of phenotypic resistance and was even able to identify a single mutated allele in phenotypically linezolid-susceptible isolates.

Linezolid is an oxazolidinone antimicrobial with activity against gram-positive cocci, including vancomycin-resistant enterococci (VRE) (3). It acts by inhibiting the initiation of ribosomal protein synthesis (22). Resistance to linezolid has repeatedly been described in enterococci, mainly VRE, that had previously been exposed to the drug (5, 6, 18, 21). There are also rare reports of resistance in enterococci that had not previously been exposed to the drug (4, 15). Whenever linezolid resistance was investigated in clinical isolates of enterococci, resistance was attributed to a point mutation of the 23S rRNA gene(s) (the 2576G>T mutation) (5, 6, 13, 16, 18, 19, 21, 24). Enterococci carry from four alleles (Enterococcus faecalis) to six alleles (E. faecium) of the 23S rRNA gene (13, 16, 24). The level of resistance may depend on the number of mutated alleles, but the mutation of one allele has been reported to be sufficient to confer linezolid resistance (13, 16, 24). A number of methods for the determination of phenotypic resistance are available, such as Etest, broth microdilution, agar dilution, and disk diffusion, but these are time-consuming and are not completely reliable (19). The detection of mutations by molecular methods, such as real-time PCR (24), DNA sequencing (16), and restriction digestion of PCR-amplified sequences, is fast and reliable but comparably laborious (19, 24). Fluorescence in situ hybridization (FISH) with probes complementary to the rRNA is a rapid, easy-to-perform method that has been implemented successfully for the detection of resistance to macrolides caused by ribosomal mutations (20). The use of FISH has also been described for the rapid identification of enterococci (23). In the present study, the reliability of FISH for the detection of linezolid resistance in enterococci was evaluated in a blinded manner with 23 “precharacterized” strains with known numbers of mutated alleles and known linezolid MICs (Table 1). The applicability of the assay in a routine setting was subsequently determined by using 83 clinical isolates (Table 2).

TABLE 1.

Results of FISH and MIC tests for precharacterized strains^a

Strain no.	Species	Linezolid MIC (mg/liter [interpretation])	No. of alleles		FISH results with probe for:
			Wild-type allele	Resistant allele	LZD-susceptible wild-type rRNA	LZD-resistant rRNA
3	E. faecium	≤2 (s)	ND	ND	+	−
4	E. faecium	≤2 (s)	ND	ND	+	−
5	E. faecium	≤2 (s)	ND	ND	+	−
6	E. faecium	≤2 (s)	ND	ND	+	−
8	E. faecium	≤2 (s)	ND	ND	+	−
9	E. faecium	≤2 (s)	ND	ND	+	−
14	E. faecium	≤2 (s)	ND	ND	+	−
16	E. faecium	≤2 (s)	ND	ND	+	−
18	E. faecium	≤2 (s)	ND	ND	+	−
19	E. faecium	≤2 (s)	ND	ND	+	−
20	E. faecium	≤2 (s)	ND	ND	+	−
22	E. faecium	≤2 (s)	ND	ND	+	−
23	E. faecium	≤2 (s)	ND	ND	+	−
1	E. faecium	8 (r)	4	2	+	+
2	E. faecalis	16 (r)	1	3	(+)	+
7	E. faecium	16 (r)	5	1	+	+
10	E. faecium	16 (r)	3	3	+	+
11	E. faecium	16 (r)	4	2	+	+
12	E. faecium	16 (r)	4	2	+	+
13	E. faecium	16 (r)	4	2	+	+
15	E. faecium	16 (r)	3	3	+	+
17	E. faecium	16 (r)	2	4	+	+
21	E. faecium	16 (r)	2	4	+	+

^aAbbreviations and symbols: LZD, linezolid; s, susceptible; r, resistant; ND, not determined; +, positive; (+), weakly positive; −, negative.

TABLE 2.

Evaluation of FISH with clinical isolates^a

Species	Linezolid MIC (mg/liter [interpretation])	No. (%) of positive FISH results with probe for:
		Total	LZD-susceptible wild-type rRNA	LZD-resistant rRNAb
E. faecium	0.5 (s)	11	11 (100)	0 (0)
	1 (s)	26	26 (100)	1 (4)
	2 (s)	6	6 (100)	1 (16)
	8 (r)	1	1 (100)	1 (100)
E. faecium	0.5 (s)	7	7 (100)	0 (0)
(VRE)	1 (s)	7	7 (100)	1 (14)
E. faecalis	0.5 (s)	4	4 (100)	0 (0)
	1 (s)	15	15 (100)	0 (0)
	2 (s)	5	5 (100)	0 (0)
E. avium	1 (s)	1	1 (100)	0 (0)

^aThe VRE strains were isolated from rectal smears or stool. All other isolates were isolated from blood cultures. Abbreviations: LZD, linezolid; s, susceptible; r, resistant.

^bThe results for the three linezolid-susceptible E. faecium isolates that showed a signal with the LZD resistant probe are highlighted in boldface.

Resistance to linezolid was defined by a MIC ≥8 mg/liter, according to DIN 58940, EUCAST 2006-06-20 V1.3, and CLSI Norm M100-S15 (8, 9). For the set of precharacterized test strains, the linezolid MIC was determined by broth microdilution with Iso-Sensitest broth as the nutrient medium. For the clinical isolates, the MIC of linezolid was determined by broth dilution with a Micronaut system (Merlin, Bornheim-Hesel, Germany). Single clinical isolates that were suspected of being linezolid resistant were additionally tested by microbroth dilution with Iso-Sensitest nutrient medium and by Etest (AB Biodisk, Solna, Sweden) on Mueller-Hinton agar (Heipha, eppelheim, Germany). For all resistant isolates, the number of mutated 23S rRNA gene alleles was determined by a previously described method with the LabChip technology and a BioAnalyzer 2100 apparatus (24).

For FISH, isolates were cultured on sheep blood agar for 18 to 24 h and suspended in 0.9% saline (0.5 McFarland standard). Ten microliters of the suspension was applied to a glass slide, air dried, and fixed for 10 min in methanol. DNA probes for FISH containing so-called locked nucleic acids (LNAs) at the site of point mutation (for the linezolid-susceptible wild type, fluorescein isothiocyanate-5′-CCCAGCTCGCGTGC-3′; for the linezolid-resistant resistant strains, Cy3-5′-CCCAGCTAGCGTGC-3′, where the boldface and underlining indicate the LNA bases; Thermo Hybaid, Ulm, Germany) were used, because these allow sharper discrimination of single-nucleotide alterations, as described recently (12). FISH was performed at 46°C with 30% formamide in the hybridization buffer and the corresponding salt concentrations in washing buffer, as described elsewhere (2, 14). Hybridization was performed by use of a microwave (Sharp R208; SeaPro) approach, as recently published (17). Bacterial probe EUB338, which targets nearly all bacteria, was used as a positive control in all experiments (1). All FISH experiments were conducted in duplicate.

In preliminary experiments with eight previously investigated strains (24), the presence of a single mutated allele led to a strong FISH signal with the respective probes. The signal was only marginally stronger when multiple alleles were mutated. FISH thus allowed the detection of a single mutated allele but not assessment of the number of mutated alleles (data not shown).

Twenty-three precharacterized Enterococcus strains, 13 linezolid-sensitive and 10 linezolid-resistant strains, were then investigated in a blinded manner by FISH (Table 1). In both independent experiments, FISH showed the correct results for all resistant and all susceptible strains (Table 1).

Finally, the applicability of the assay was assessed by using 83 clinical isolates grown in blood cultures or on VRE screening agar plates (Table 2). Seventy-nine isolates were found to be susceptible by FISH and broth microdilution with the Micronaut system. One E. faecium blood culture isolate was determined to be resistant by broth dilution and Etest, with a MIC of 8 mg/liter. FISH showed a positive result with the probe for mutated rRNA as well as with the probe for wild-type rRNA, indicating the presence of mutated and nonmutated alleles, which was confirmed by use of the LabChip technology (24). The strain was vancomycin sensitive and was isolated from the blood culture of a patient who had received piperacillin-tazobactam (Tazobac) but not linezolid.

Three E. faecium isolates gave a positive result with the probe for linezolid resistance and were repeatedly determined to be phenotypically susceptible by broth microdilution with the Micronaut system as well as the Iso-Sensitest system and by Etest (Table 2, data in boldface). All three isolates showed a single mutated 23S rRNA gene allele by use of the LabChip technology (24). One strain was vancomycin resistant and was isolated from the stool of a known carrier of a vanA-type VRE. The patient had received linezolid for 10 days 5 months before isolation of the strain. Two vancomycin-susceptible strains were isolated from blood cultures from two patients treated in different intensive care units at different times. Both patients had received various antimicrobials, such as imipenem, vancomycin, and piperacillin, but not linezolid.

Considering the primary phenotypic resistance testing results, linezolid would have been a treatment option for the three patients. In view of previous reports on the exchange of mutated alleles by recombination in vitro under selective pressure (13), such strains might, however, be prone to the rapid development of resistance. In order to test this hypothesis, we performed in vitro mutagenesis experiments with the three strains and compared the results to those for two linezolid-susceptible isolates possessing only wild-type 23S alleles. Exponentially grown bacteria were streaked onto agar plates supplemented with linezolid (2, 4, and 6 mg/liter). Two of the three test strains but none of the susceptible reference strains showed single colonies that grew on plates with linezolid after 48 h of incubation at 37°C. All four representative colonies investigated from plates with 4 and 6 mg/liter linezolid showed linezolid MICs of 8 mg/liter, indicating spontaneous resistance development. The LabChip technology revealed that the number of resistant 23S rRNA gene alleles increased from one in the original strain to two in the mutants. Infections caused by such isolates should therefore preferably not be treated with linezolid. If such infections are treated with linezolid, close monitoring for the development of resistance is warranted by assessment of clinical improvement, repeated sampling, and, if applicable, testing of further isolates by phenotypic methods.

It is a crucial observation that the presence of a single mutated 23S rRNA gene allele leads to phenotypically detectable linezolid resistance in some (Table 1) but not all (Table 2) isolates. It may be speculated that 23S rRNA gene alleles could be expressed heterogeneously, for instance, depending on the distance of the mutated 23S rRNA gene allele from the origin of replication, which might influence the amount of replicated mutated rRNA available for formation of mutated/linezolid-resistant ribosomes (7, 11). As long as genome data for E. faecium are incomplete in this regard, this hypothesis remains speculative.

The occurrence of such strains even in patients who had never received linezolid stresses the need for the molecular determination of the linezolid resistance of enterococci, particularly in cases of elevated linezolid MICs or an unsatisfactory response to therapy with linezolid (19). FISH with LNA-containing probes proved to be a suitable and reliable method for this purpose. It showed 100% sensitivity for the detection of phenotypic linezolid resistance and even allowed detection of a single mutated 23S rRNA gene allele in phenotypically linezolid-susceptible enterococci. As a rapid and easily performed method, FISH may also be implemented for the initial determination of linezolid resistance or the screening of sample collections. Additional advantages of FISH are the ability to simultaneously differentiate enterococci to the species level by using previously described FISH probes and use of the test directly with clinical samples, such as blood culture samples (10, 23). FISH thus has the potential to significantly improve and accelerate the determination of linezolid resistance in E. faecium.