Routine Molecular Identification of Enterococci by Gene-Specific PCR and 16S Ribosomal DNA Sequencing

Abstract

For 279 clinically isolated specimens identified by commercial kits as enterococci, genotypic identification was performed by two multiplex PCRs, one with ddl_{E. faecalis} and ddl_{E. faecium} primers and another with vanC-1 and vanC-2/3 primers, and by 16S ribosomal DNA (rDNA) sequencing. For 253 strains, phenotypic and genotypic results were the same. Multiplex PCR allowed for the identification of 13 discordant results. Six strains were not enterococci and were identified by 16S rDNA sequencing. For 5 discordant and 10 concordant enterococcal strains, 16S rDNA sequencing was needed. Because many supplementary tests are frequently necessary for phenotypic identification, the molecular approach is a good alternative.

Identification at the species level of enterococci isolated from clinical specimens is considered necessary, as is quantitative evaluation of their resistance to penicillin, ampicillin, vancomycin, and teicoplanin and high-level resistance to gentamicin and streptomycin (11). It is also necessary to distinguish the low-virulence motile enterococcal species with constitutive low-level resistance to vancomycin from the species that are more frequently isolated from clinical specimens, such as Enterococcus faecalis and E. faecium, which in some countries can often show high-level inducible and transmissible resistance to glycopeptides (37).

Commercially available kits are often used by clinical laboratories as an alternative to the numerous physiological tests needed to identify enterococcal species (6, 9, 10, 11, 37); nevertheless, all commercial kits vary in their performance and persistently show many drawbacks, especially in cases of atypical strains, and at best need supplementary manual tests, which somewhat impair their usefulness (14, 17, 23, 33, 34, 36; P. A. d'Azevedo, C. G. Dias, A. L. S. Gonçalves, F. Rowe, and L. M. Teixeira, Abstr. 100th Gen. Meet. Am. Soc. Microbiol., abstr. C-242, 2000). In some cases identification of atypical E. faecium strains may be problematic, as is distinguishing E. gallinarum from E. faecium and also identifying E. durans, E. avium, E. raffinosus, E. hirae, and E. mundtii (1, 2, 6, 7, 35, 37).

In 1995 a multiplex PCR was devised for the identification of E. faecium and E. faecalis by primers targeted at specific sequences in the ddl (d-Ala–d-Ala ligase) chromosomal genes of the two species and in the glycopeptide resistance ligase genes vanA, vanB, and vanC (8). The vanC gene is present in the motile, low-level constitutive glycopeptide-resistant species E. gallinarum (vanC-1) and E. casseliflavus and E. flavescens (vanC-2/3); thus, demonstration of its presence indicates the presence of one of the aforementioned species (4, 24, 27, 29, 30).

The use of a PCR with primers for ddl_{E. faecalis} and ddl_{E. faecium}, together with primers for vanC-1, -2, and -3, may be the most simple molecular approach for both rapid and precise identification of enterococci while avoiding the drawbacks of commercial kits. It has been succesfully used to identify vancomycin-resistant enterococci (8, 13, 18, 21, 22, 25, 26, 30, 31), but it may also be used broadly to identify all enterococci; nevertheless, this approach certainly fails to identify some enterococcal species outside the reach of the primers, and for this reason it is also necessary to include among the proposed molecular methods for enterococcal species identification the amplification and sequencing of the 16S ribosomal DNA (rDNA) gene (25, 28, 38).

Two hundred seventy-nine clinical strains consecutively identified as enterococci by three hospital laboratories in Rome were studied (Table 1). The three laboratories mainly used the following commercial kits: one hospital used API 32 Strep (Bio-Merieux, Marcy l'Etoile, France), the second used API 20 Strep (Bio-Merieux) or BBL Sceptor (Beckton-Dickinson Microbiology Systems, Paramus, N.J.), and the third used API 20 Strep or BBL Crystal GP (Beckton-Dickinson Microbiology Systems).

TABLE 1.

Phenotypic and genotypic identification of enterococcal species examined

Species	No. of isolates identified by phenotype	No. of isolates testing positive by multiplex PCR for:
		ddlE. faecalis	ddlE. faecium	vanA	vanB	vanC-1	vanC-2	vanC-3
Enterococcus faecalis	197	201	0	1	1	0	0	0
Enterococcus faecium	63	0	48	1	0	0	0	0
Enterococcus casseliflavus	6	1	1	0	0	0	4	0
Enterococcus duransa	7	0	0	0	0	0	0	0
Enterococcus aviuma	2	0	0	0	0	0	0	0
Enterococcus raffinosusa	1	0	0	0	0	0	0	0
Not identifiedb	3	0	1	0	0	1	0	0

^aConfirmed by 16S rDNA sequencing. 

^bA multiplex PCR-negative strain which was identified as E. raffinosus by 16S rDNA sequencing. 

All strains submitted as glycopeptide resistant were verified with the E-test method (AB BIODISK, Solna, Sweden) in our laboratory.

Genotypic identification was performed by first applying a multiplex PCR with ddl_{E. faecalis}, ddl_{E. faecium}, vanA, and vanB primers as previously described (8); in the case of a negative reaction, a second multiplex PCR was applied with vanC-1, -2, and -3 primers (4). We preferred to do two separate multiplex PCRs to avoid interferences in the amplifications by the excessive number of primers, as already noted (13). If the second reaction was also negative, 16S rDNA amplification was carried out with primers 27f and 685R (20), the PCR products were directly sequenced by dye-terminator sequencing, and the sequences were analyzed by the BLAST software of the National Center for Biotechnology Information. A second set of vanB primers was used (31) when the first vanB primers did not amplify phenotypically VanB strains. Of the 279 strains, 13 gave no multiplex amplification at all; of these, six strains (four submitted as E. faecalis and two submitted as E. faecium) were not enterococci. They were identified by 16S rDNA sequencing as two Streptococcus spp., one Leuconostoc sp., one Aerococcus sp., and one Streptococcus bovis strain. Among the other multiplex PCR-negative strains, four were identified by 16S rDNA sequencing as E. avium (three strains) and E. durans (one strain) (all phenotypically identified as E. faecium); one strain was submitted as an Enterococcus sp. because of the uncertainty of the phenotypical results and was identified as E. raffinosus by 16S rDNA sequencing; two strains submitted as E. faecalis but negative by the two multiplex PCRs were lost before 16S rDNA sequencing (Table 2).

TABLE 2.

Genotypic and phenotypic identification in discordant strains

Phenotypic identification	Multiplex PCR identification	16S rDNA PCR and sequencing identification
E. faecalis	None	Lactobacillus
E. faecalis	None	Aerococcus sp.
E. faecalis	None	S. bovis
E. faecalis	None	Streptococcus sp.
E. faecium	None	Leuconostoc sp.
E. faecium	None	Streptococcus sp.
E. faecalis	None	NAa
E. faecalis	None	NA
E. faecium	None	E. avium
E. faecium	None	E. avium
E. faecium	None	E. avium
E. faecium	None	E. durans
Not identified	None	E. raffinosus
E. faecium	E. faecalis	NA
E. faecium	E. faecalis	NA
E. faecium	E. faecalis	NA
E. faecium	E. faecalis	E. faecalis
E. faecium	E. faecalis	E. faecalis
E. faecium	E. faecalis	E. faecalis
E. faecium	E. faecalis	E. faecalis
E. faecium	E. faecalis	E. faecalis
E. faecium	E. faecalis	E. faecalis
E. casseliflavus	E. faecium	E. faecium
E. casseliflavus	E. faecalis	E. faecalis
Not identified	E. gallinarum	E. gallinarum
Not identified	E. faecium	E. faecium

^aNA, specimen was not available. 

Nine strains identified by commercial devices as E. faecium were identified as E. faecalis by multiplex PCR, six of which were also confirmed by sequencing of amplified 16S rDNA (three were lost before sequencing). Two strains identified as E. casseliflavus by commercial kits were identified as one E. faecium strain and one E. faecalis strain by multiplex PCR (and confirmed by 16S rDNA sequencing); another two strains labeled Enterococcus sp. because of lack of identification by a commercial device were identified as one E. faecium strain and one E. gallinarum strain by multiplex PCR (confirmed by 16S rDNA sequencing) (Table 2).

The other 253 isolates had the same identification by molecular methods as that by commercial kits. Most phenotypic identifications were done with API 32 Strep and API 20 Strep (189 of 279 strains), which had a 9.5% overall error rate; 81 strains were phenotypically identified by Sceptor, with an overall error rate of 8.6%, and 9 strains were identified by Crystal GP (one misidentified).

Overall, the two routine multiplex PCRs allowed for the identification of 263 of 279 strains: for only 15 definitive enterococcal species (10 concordant and 5 discordant results from phenotypic tests) was it necessary to use 16S rDNA amplification and sequencing. The latter technique was needed for the genotypic identification of five E. avium, two E. raffinosus, and eight E. durans strains. Moreover, 16S rDNA sequencing confirmed all multiplex PCR identifications for available strains. In some cases 16S rDNA sequencing was repeated, and in all cases the result was confirmed. It must be noted that the primers used for 16S rDNA amplification amplify from the V1, V2, V6, and V7 variable regions of 16S rDNA, which contain most base variations among enterococci. Moreover, the sequencing results were visually aligned with published enterococcal sequences (28).

The two phenotypic E. faecalis strains which were not identified by the two multiplex PCRs but were not available for 16S rDNA sequencing may be explained, in the light of our overall experience, as being enterococcal species different from E. faecium, E. faecalis, E. gallinarum, E. casseliflavus, and E. flavescens or as being subjects to an error in the two multiplex amplifications.

The difficulty in phenotypic identification of unusual enterococcal strains is not unexpected, particularly by manual commercial kits (14, 19, 23, 33), but incorrect identification at the genus level and the misidentification of strains that are frequent in clinical specimens, such as E. faecalis and E. faecium, point to the need for busy clinical laboratories not to rely only on commercial kits but also to do the fundamental additional tests which are necessary for genus and species identification of enterococci. In fact, only catalase and l-pyrollidonyl-β-naphthylamide hydrolysis (PYR) tests have been used routinely by the three laboratories, and none of the laboratories used additional tests; moreover, it is also evident that in some cases the serological grouping was misleading or incorrectly done. For more than 50% of discordant cases the commercial kits gave a low probability profile: this is a prompt to proceed to additional tests or to genotypic tests. In order to identify enterococcus species reliably it is necessary to do preliminary tests, including catalase, PYR, bile-esculin, and 6.5% NaCl tests, and then a commercial manual or automated device test, but it also often seems necessary to do at least the following additional tests: motility, pigment production, d-xylose fermentation or methyl-α-d-glucopyranoside fermentation, Litmus milk reduction, and pyruvate utilization (1–3, 6, 7, 9–11, 35, 37). It seems that the described molecular approach offers a good alternative to this array of physiological tests.

Until now in Italy glycopeptide-resistant enterococci have been rare (12); in other countries the identification of VanA and VanB resistance by genotypic tests is worthwhile (5). We have found two VanA- and three VanB-positive isolates by phenotypic susceptibility tests; by multiplex PCR the two vanA-positive and only 1 of 3 vanB-positive strains were identified (Table 1). Also, the second set of primers (31) failed to amplify the same two vanB-positive strains. Notwithstanding the fact that the alignment of the chosen primers shows no mismatch with the published sequences, the vanB primers may fail to amplify some vanB genes because of variability of the sequences in the primer regions (16, 29).