Genomic Relationships between Enterococcus faecium Strains from Different Sources and with Different Antibiotic Resistance Profiles Evaluated by Restriction Endonuclease Analysis of Total Chromosomal DNA Using EcoRI and PvuII

Abstract

Forty-seven Enterococcus faecium strains from different sources were evaluated by restriction endonuclease analysis (REA) of total chromosomal DNA. Strains from chicken, pork, and humans were clearly divided into separate clusters, whereas strains from different countries, strains with different antibiotic resistance profiles, or clinical and healthy-subject strains were not.

Infections by Enterococcus spp. have become increasingly important during the last decade (14, 24), and there has been much interest in Enterococcus faecium since it is prone to take up antibiotic resistance genes (5, 6, 16). Furthermore, enterococci are infamous for their ability to rapidly transfer their resistance genes to other enterococci (14), as well as to bacteria belonging to other genera (12, 17).

It has been proposed that strains are spread from animals to humans (1–3), and since the use of growth promoters and antimicrobial agents in animal husbandry often selects for E. faecium (9, 10, 15), it is important to clarify the role of E. faecium strains of animal origin in human infections.

E. faecium is a homogeneous species (19), and separation of strains belonging to this species requires methods with high discriminatory powers. It has been proposed that restriction endonuclease analysis (REA) is the best method to separate E. faecium strains (25), and this method has been used by other workers (7, 11). In these studies, however, only the ability of REA to separate strains and not the ability to reveal relationships between strains was investigated. In the present study, two separate restriction endonuclease digestions were used, compared to the one digestion used in the previous studies (7, 11). Also, use of a computer-based interpretation system allowed us to analyze gels with higher complexity, which meant that a broader range of DNA fragments could be included in the analysis compared to other studies (7, 11).

The aim of this study was to clarify the genomic relationships among E. faecium strains from different sources and with different levels of antibiotic resistance. REA of total chromosomal DNA with frequently cutting endonucleases was used due to its great capacity to resolve organisms at the strain level (8).

The strains included in this study are listed in Table 1. All of the strains were previously identified as E. faecium strains with methods other than REA (18, 19; data not shown). The test strains were obtained from different specimens and at different times.

TABLE 1.

Clusters of E. faecium strains obtained at the 45% similarity level: sources of strains, identification method(s) used, and antibiotic resistance

Cluster	Isolate	Sourcea	Identification method(s)b	Antibiotic resistancec
I	SE ch 4c	Swedish chicken	phen + RAPD	TC
I	SE ch 17d	Swedish chicken	phen + RAPD	TC
I	DK ch 5a	Danish chicken	phen + RAPD	EM, VA
I	DK ch 6b	Danish chicken	phen + RAPD	EM, PV, VA
II	NCC 218d	The Netherlands, human, hosp	RAPD	EM
II	SE hum B42d	Sweden, human blood, hosp	phen + RAPD	PV
II	DK hum 5/3bd	Denmark, human feces, healthy	RAPD	EM
II	DK hum 27bd	Denmark, human feces, healthy	RAPD	NX
II	DK hum 24ad	Denmark, human feces, healthy	RAPD	EM, NX
II	CCUG 542Te	Type strain, human feces	phen + RAPD
II	SE hum B50d	Sweden, human feces, hosp	phen + RAPD	TC
II	G	Danish probiotic yoghurt	phen + RAPD
II	DK hum 19ad	Denmark, human feces, healthy	RAPD	NX, VA
III	SE hum B36d	Sweden, human urine, comm	phen + RAPD	AM, EM, NX, PV, TC
III	SE hum ET45d	Sweden, human urine, comm	phen + RAPD	AM, EM, NX, PV, TC
III	SE hum B49d	Sweden, human wound, comm	phen + RAPD	AM, EM, PV, TC
III	SE hum B60d	Sweden, human feces, hosp	phen + RAPD
III	SE hum ET33d	Sweden, human urine, hosp	phen + RAPD	AM, EM, PV
III	SE hum ET47d	Sweden, human urine, comm	phen + RAPD	AM, EM, PV, TC
III	SE hum multid	Sweden, human blood, hosp	phen + RAPD	AM, CL, EM, PV, PP, TC, TS, VA
IV	SE hum ET144d	Sweden, human urine, hosp	phen + RAPD	AM, EM, NX, PV, TC
IV	SE hum 19ad	Sweden, human normal flora	phen + RAPD
V	SE ch 14a	Swedish chicken	phen + RAPD	TC
V	SE ch 2a	Swedish chicken	phen + RAPD	NX, TC
VI	NCC 212d	Probiotic strain	RAPD	CL, EM, PV
VI	DK hum 30bd	Denmark, human feces, healthy	RAPD	EM
VI	MC74d	Probiotic strain	RAPD	EM
VI	SF68d	Probiotic strain	RAPD	EM
VI	DK hum 26/2ad	Denmark, human feces, healthy	RAPD	EM
VI	NCC 270d	Probiotic strain	RAPD	EM
VI	DK hum 23ad	Denmark, human feces, healthy	RAPD	EM, PV
VII	DK hum 12ad	Denmark, human feces, healthy	RAPD	EM
VII	DK hum 20/2bd	Denmark, human feces, healthy	RAPD	NX
VIII	DK hum 1ad	Denmark, human feces, healthy	RAPD	EM
VIII	DK hum 17ad	Denmark, human feces, healthy	RAPD	EM
IX	DK po 4d	Danish pork	phen + RAPD
IX	DK po 24a	Danish pork	phen + RAPD	EM
X	DK hum 22/2ad	Denmark, human feces, healthy	RAPD	CL, EM, TC
X	DK hum 11ad	Denmark, human feces, healthy	RAPD	NX
XI	DK hum 3bd	Denmark, human feces, healthy	RAPD	TC
XI	DK hum 8bd	Denmark, human feces, healthy	RAPD	NX
Straggler	DK hum 25ad	Denmark, human feces, healthy	RAPD	EM, PV, TC
Straggler	DK ch 6d	Danish chicken	phen + RAPD	EM, PV, VA
Straggler	DK ch 7c	Danish chicken	phen + RAPD	EM, VA
Straggler	DK po 5d	Danish pork	phen + RAPD
Straggler	SE po 1c	Swedish chicken	phen + RAPD
Straggler	DK po 24b	Danish pork	phen + RAPD
Straggler	DK hum 18bd	Denmark, human feces, healthy	RAPD

^ahosp, hospitalized patient; comm, community-based patient; healthy, healthy individual. 

^bIdentification procedures were performed by workers in our laboratory. phen, phenotypic identification; RAPD, randomly amplified polymorphic DNA identification. 

^cAM, ampicillin; CL, chloramphenicol; EM, erythromycin; NX, norfloxacin; PP, piperacillin; PV, penicillin V; TS, trimethoprim-sulfamethoxazole; TC, tetracycline; VA, vancomycin. 

^dReceived as E. faecium (phenotypically identified). 

^eIdentified as E. faecium by workers at the Culture Collection, University of Gothenburg, Gothenburg, Sweden. 

The strains were grown overnight in 50 ml of All Purpose Tween broth (Difco Laboratories, Detroit, Mich.). The cells were washed, resuspended in Tris-EDTA buffer (10 mmol of Tris liter⁻¹, 1 mmol of EDTA liter⁻¹; pH 8.3), and stored at −20°C.

The chromosomal DNA was prepared by enzymatic cell lysis, phenol-chloroform extraction, and dye-buoyant density centrifugation gradient, as previously described (23). This method removed most of the plasmid DNA but left the chromosomal DNA intact.

Restriction endonuclease digestion was performed as described previously for Lactobacillus digestion (23), except that two rather than three restriction endonucleases were used to digest the DNA. Two separate digestions were performed, one with EcoRI and one with PvuII (Boeringer-Mannheim Scandinavia, Bromma, Sweden). The DNA fragments were visualized by electrophoresis on a submerged 0.9% agarose gel, stained with ethidium bromide, and photographed as previously described for an analysis of Lactobacillus plantarum (8).

The gel images were scanned into a computer. Combined gel lanes consisting of the EcoRI digestion results and the PvuII digestion results in sequence were constructed, the lanes were compared by using the pattern recognition technique (Pearson coefficient), and dendrograms based on the unpaired group method using arithmetric averages (UPGMA) were constructed. All this was done by using GelCompar 4.0 software (Applied Maths, Kortrijk, Belgium).

Eleven clusters and seven single strains were identified at a similarity level of 45% in the UPGMA dendrogram based on the combined REA profiles obtained from digestion with EcoRI and digestion with PvuII (Fig. 1). All but two of the strains tested could be clearly separated. The method exhibited good reproducibility (typically, 90 to 95% similarity) (data not shown); thus, the reproducibility was about the same as that previously obtained with lactobacilli (8). The source of isolation (human, chicken, or pig) could be determined in the clusters obtained from the REA profiles, whereas the specific source of isolation (e.g., feces, blood, or urine) could not be determined (Table 1 and Fig. 1). Furthermore, the antibiotic resistance profiles of the strains (Table 1) were not reflected in the dendrogram, and human strains from healthy persons and clinical specimens did not form separate clusters.

FIG. 1.

UPGMA dendrogram based on the REA profiles of E. faecium strains from different sources.

The method used to prepare DNA for the REA included a large number of steps for obtaining pure, unfragmented, essentially plasmid-free DNA. The efficiency of plasmid DNA removal by this DNA preparation method was demonstrated by the fact that the existence of antibiotic resistance genes in many of the strains tested did not affect the clustering (Table 1 and Fig. 1).

The question of whether Enterococcus strains and/or their antibiotic resistance genes are spread between animals and humans has not been clearly answered. Strong evidence that animal enterococci are capable of infecting humans has been presented by Das et al. (3). Several other workers have also suggested that resistant strains are spread from animals to humans (1, 2, 13). Seyed-Akhavani et al. (21) have suggested the possibility that resistance is spread via plasmid transfer from resistant animal strains to previously susceptible human strains. Indeed, the fact that human E. faecium strains are able to receive resistance genes from donor strains from chickens has been demonstrated. For instance, strains DK ch 6d and DK ch 5a, which were used in this investigation, were found to transfer both erythromycin resistance and vancomycin resistance to strain CCUG 542 (18) and to strain G (19a). There are also workers who claim that they have not been able to find any valid evidence for the spread of strains or resistance from animals to humans (4, 20).

The present finding that strains cluster according to their hosts suggests that the strains are host specific. Such specificity has also been observed in Lactobacillus reuteri strains by Ståhl and Molin, who found that strains from humans or pigs could be separated from strains from rats by REA (22). The fact that no strains from animals were found to cluster together with human strains in the present study does not necessarily mean that animal-to-human spread does not occur. In the present study, 48 isolates representing only a minute part of the immense number of strains occurring in nature were investigated.