Oligonucleotide microarray for identification of Enterococcus species


  • Edited by O.P. Kuipers

*Corresponding author. Tel.: +43 1 4277 54207; fax: +43 1 4277 54389, E-mail address: loy@microbial-ecology.net


For detection of most members of the Enterococcaceae, the specificity of a novel oligonucleotide microarray (ECC-PhyloChip) consisting of 41 hierarchically nested 16S or 23S rRNA gene-targeted probes was evaluated with 23 pure cultures (including 19 Enterococcus species). Target nucleic acids were prepared by PCR amplification of a 4.5-kb DNA fragment containing large parts of the 16S and 23S rRNA genes and were subsequently labeled fluorescently by random priming. Each tested member of the Enterococcaceae was correctly identified on the basis of its unique microarray hybridization pattern. The evaluated ECC-PhyloChip was successfully applied for identification of Enterococcus faecium and Enterococcus faecalis in artificially contaminated milk samples demonstrating the utility of the ECC-PhyloChip for parallel identification and differentiation of Enterococcus species in food samples.


Bacteria of the genus Enterococcus are found in a wide variety of habitats such as soil, water, plants, fermented food, and in the gastrointestinal tracts of animals and humans [1]. In addition, members of this genus have recently attracted attention in clinical microbiology as emerging nosocomial, antibiotic-resistant pathogens causing bacteraemia, endocarditis, urethritis and other infections [2,3]. Their ability to survive adverse environmental conditions also makes some gastrointestinal enterococci suitable as indicators for hygienic quality in food and drinking water [4]. Rapid and accurate identification of enterococci at the species level is therefore an essential task in both clinical microbiology and food hygiene. Identification of enterococci isolates based on classical phenotypic and biochemical characterization is often difficult to accomplish due to considerable similarities among some of the species [5]. Therefore, commercial systems such as API (bioMérieux, Marcy l'Etoile, France) or MicroScan (Dade International, MicroScan Int., West Sacramento, CA, USA) often fail to correctly identify Enterococcus species [6–8].

Rapidly increasing data sets of rRNA sequences of prokaryotes [9,10] allow the design of specific hybridization probes (so-called “phylogenetic probes”) for various taxa or phylogenetic entities at user-defined levels of resolution. Application of multiple probes targeting different sites on the rRNA (genes) significantly reduces the risk of misidentification and allows discrimination down to the species level [11]. This concept was applied previously to design a comprehensive rRNA-targeted oligonucleotide probe set of hierarchical and parallel specificity for most Enterococcaceae[12].

It was the aim of this study to extend and evaluate the previously developed nested phylogenetic probe set for enterococci [12] for reverse hybridization on microarrays. Although DNA microarrays are circulating for almost 10 years [13], they have only recently attracted attention as powerful diagnostic tools for the identification of microorganisms in complex environmental and clinical samples [14–24]. Here we show, by analyzing milk that was artificially contaminated with Enterococcus species, that the ECC-PhyloChip is a highly reliable tool to correctly identify and differentiate members of the Enterococcaceae.

2Materials and methods

2.1Reference strains

Reference organisms for evaluating the microarray are listed in Table 1 and were obtained either from the Deutsche Sammlung von Mikroorganismen und Zellkulturen, (DSMZ, Braunschweig, Germany), the Laboratorium voor Mikrobiologie Gent, (LMG, Gent, Belgium) or the Institut für Lebensmitteltechnologie, Universität Hohenheim (LTH, Stuttgart, Germany). Strains were grown overnight in Brain–Heart-Infusion medium (Difco, Liverpool, UK) at 37 °C. One milliliter of culture was harvested by centrifugation (5 min at 7150g) and washed in phosphate-buffered saline (PBS: 130 mM NaCl, 10 mM NaH2PO4, 10 mM Na2HPO4, pH 7.2) prior to DNA extraction.

Table 1.  Reference strains
Enterococcus asiniDSM 11492T
Enterococcus aviumLMG 10744T
Enterococcus casseliflavusDSM 20680T
Enterococcus cecorumDSM 20682T
Enterococcus columbaeDSM 7374T
Enterococcus disparDSM 6630T
Enterococcus duransDSM 20633T
Enterococcus faecalisLMG 7937T
Enterococcus faeciumDSM 20477T
Enterococcus flavescensDSM 7330T
Enterococcus gallinarumDSM 20628T
Enterococcus hiraeDSM 20160T
Enterococcus malodoratusDSM 20681T
Enterococcus mundtiiDSM 4838T
Enterococcus pseudoaviumDSM 5632T
Enterococcus raffinosusDSM 5633T
Enterococcus saccharolyticusLMG 11427T
Enterococcus solitariusDSM 5634T
Enterococcus sulfurousDSM 6905T
Lactococcus lactisDSM 20481T
Melissococcus plutoniusLTH 3442
Staphylococcus aureusDSM 20232T
Tetragenococcus halophilusDSM 20339T

2.2Contaminated milk samples

Two different Enterococcus food isolates were obtained from the Bavarian State Institute for Food Surveillance (LUAS, Oberschleißheim, Germany) for artificial contamination of milk samples. One isolate was tentatively identified as Enterococcus faecium by selective plating and subsequent biochemical characterization of grown colonies by API 20 STREP (bioMérieux, Marcy l'Etoile, France) (LUAS, personal communication). The identity of the second isolate could not be determined to the species level by using this approach. For each Enterococcus isolate, one milliliter of ultra high temperature milk was inoculated with cells using a sterile loop. Five replicates each were prepared for E. faecium (S1–S5) and the unidentified Enterococcus species (S6–S10). An enrichment step was performed by incubating the milk aliquots for 16 h at 37 °C with 1 ml of Brain–Heart-Infusion medium. Subsequently, cells were harvested by centrifugation (5 min at 7150g), resuspended in 1 ml of digestion buffer (100 mM Na2HPO4, 150 mM NaCl, 10 mM EDTA, 40 mM NaOH), and incubated for 10 min at room temperature for protein denaturation [25]. After centrifugation at 7150g for 5 min at 4 °C, surface fat was removed by using a sterile swab and the supernatant was decanted. The protein denaturation step was repeated twice, and the retrieved cells resuspended in 200 μl PBS. Half of the cells were plated on oxolinic acid-esculin-azide enterococci selective agar [26], whereas the other half was used for extraction of nucleic acids.

2.3Isolation of genomic DNA

Genomic DNA was isolated by enzymatic lysis of the cells and subsequent extraction of nucleic acids with chloroform/isoamyl alcohol as described previously [27]. Extracted DNA was resuspended in 50 μl double-distilled water and stored at −20 °C.

2.4rRNA-targeted probes and microarray fabrication

Tables 2 and 3 list names, sequences, and intended specificities of the oligonucleotide probes used in this study. Further information (e.g., G+C content or molecular weight of each probe) can be accessed at the probeBase database (http://www.microbial-ecology.net/probebase/) [28]. The free hybridization energy, ΔG, of each probe to its perfectly matching target sequence was calculated with the 2-state hybridization server (concentration of Na+ and temperature were set to 0.829 M and 42 °C, respectively) at the mfold website (http://www.bioinfo.rpi.edu/applications/mfold/) [29]. Each oligonucleotide was tailed at the 5′ end with a 15 dTTP spacer element and synthesized with a 5′-terminal amino-modification (MWG-Biotech, Ebersberg, Germany). Spotting of the modified oligonucleotide probes (50 pmol/μl in 50% dimethylsulfoxide) onto aldehyde-group-coated CSS-100 glass slides (CEL Associates, Houston, USA) was performed using a GMS 417 (Affymetrix, Santa Clara, USA) contact printing device. All probes were immobilized on the microarray in duplicate. Microarrays were dried over night at room temperature for effective cross-linking. Reduction of free reactive aldehyde groups with sodium borohydride and washing of slides was performed as described previously [19].

Table 2.  23S rRNA gene-targeted oligonucleotide probes for Enterococcaceae
NameaEscherichia coli positionSequence [5′–3′]GC content (%)ΔG (kcal mol−1)SpecificitybReference
  1. aSuffix “i” in the probe name indicates that this probe cannot be used to detect 23S rRNA. The reverse complementary version of this probe targets 23S rRNA.

  2. bTarget organisms having a perfectly matching probe target site.

  3. cThe inverse complementary version of the published probe was used.

  4. dProbe was excluded from the final ECC-PhyloChip because it gave either false-positive signals with many non-target reference strains or a false-negative signal with the target strain.

Enc38i847AGA ATG ATG GAG GTA GAG44.4−17.5Most Enterococcaceae[12]
Eamprs09142CAC TGA AAA GTA ACA TCC38.9−16.9E. avium[12]
     E. malodoratus 
     E. pseudoavium 
     E. raffinosus 
     E. sulfureus 
Eacdfg571447AGA CAT ATC CAT CAG TCT38.9−17.1E. asini[12]
     E. casseliflavus 
     E. dispar 
     E. flavescens 
     E. gallinarum 
Eampr18343GGT GCC AGT CAA ATT TTG44.4−18.8E. avium[12]
     E. malodoratus 
     E. pseudoavium 
     E. raffinosus 
Edfm571456CTG CTT GGA CAG ACA TTT44.4−18.8E. durans[12]
     E. faecium 
     E. mundtii 
Eduhi09142CAC GCA AAC GTA ACA TCC50.0−20.0E. durans[12]
     E. hirae 
Esa38835ATT CTC AAC TTC GAC GCT44.4−19.5E. asini[12]
     E saccharolyticus 
Ecafl09i142GGA TGT TAC GTC TGC GTG55.5−20.6E. casseliflavus[12]
     E. flavescens 
Ecoce581490AGT GAC AAG CAT TTG ACT38.9−18.4E. cecorum[12]
     E. columbae 
Esasu58i1487GAG AGT CAA ATG CTT TCA38.9−17.7E. saccharolyticus[12]
     E. sulfureus 
Esoha571452TGG ACA GAC CTT TCC ATT44.4−18.9E. solitarius[12]
     T. halophilus 
Eas09136CGT AAC ATC CTA TCA AAG38.9−16.6E. asini[12]
Eav58i1494AAA TGC TTA CAT CTC TAA27.8−15.7E. avium[12]
Ece09142CAC TTA AAG GTA ACA TCC38.9−16.6E. cecorum[12]
Eco09i142GGA TAT TAC CCT TAA GTG33.3−16.0E. columbae[12]
Eca581502AGC TTG TCC GTA CAG GTA50.0−20.4E. casseliflavusThis study
Edr581500CTT ACT CGT GTA GAC AGA44.4−18.0E. duransThis study
Efa541399CAA AAA CAA CTG GTA CAG38.9−17.3E. faecalis[12]
Efm09142CAC ACA ATC GTA ACA TCC44.4−18.3E. faecium[31]
Efl58i1500TTC TAC CTA TAC GGA CAA38.9−17.1E. flavescens[12]
Ega09142CAC AAC TGT GTA ACA TCC44.4−18.1E. gallinarum[12]
Ehr581500CTT GCT CGT ACA GAC AGA50.0−19.6E. hiraeThis study
Ema58i1497TGC TTG CAT CTC TAA GGA44.4−19.1E. malodoratus[12]
Emu581498GTC CTT AAA GTT AGA AGC38.9−16.6E. mundtii[12]
Eps581497TCC TTA TAG ACG TAA GCA38.9−17.6E. pseudoavium[12]
Era581499TGT CCT TAA AGA CGT AAG38.9−17.1E. raffinosus[12]
Esa09142CAC TAA TAA GTA ACA TCC33.3−15.2E. saccharolyticus[12]
Enc01aVd1AGG TTA AGT GAA TAA GGG38.9−16.8Enterococcus spp.,[12]
     Vagococcus spp., 
     notE. solitarius 
Enc01bVd1AGG TTA AGT AAG AAA GGG38.9−16.8E. solitarius, T. halophilus[12]
Enc01cVd1AGG TTA AGT GAA CAA GGG44.4−18.2M. plutonius[12]
Esasu58d1487TGA AAG CAT TTG ACT CTC38.9−17.7E. saccharolyticus[12]
     E. sulfureus 
Eso18id276ACA CGA TCT TTT AGA CGA38.9−18.3E. solitarius[12]
Eso58d1496GTG AAC AAG AAA AAG CCT38.9−18.1E. solitarius[12]
Eso58id1496AGG CTT TTT CTT GTT CAC38.9−18.1E. solitarius[12]
Edr58id1500TCT GTC TAC ACG GAT AAG44.4−18.0E. duransThis study
Eav58d1494TTA GAG ATG TAA GCA TTT27.8−15.7E. aviumThis study
Edi38d835ATT CTT CAC TTC CAA ATT44.4−16.2E. dispar[12]
Efs18id343CGA AAT GCT AAC AAC ACC44.4−18.7E. faecalisModified from [31]c
Efi58d1476TGA CTC CTC TCC AGA CTT44.4−19.1E. faecium[12]
Efm09id142GGA TGT TAC GAT TGT GTG44.4−18.3E. faeciumModified from [31]c
Esu18d346CTA GGT GCA TAC CAA ATT38.9−17.4E. sulfureus[12]
Mpl15id268AAA CCA ACG AGC ATG CTT44.4−20.2M. plutoniusThis study
Mpl58id1502ACT CTG TAA GGA TGA GTT38.9−17.3M. plutoniusThis study
Tha09d126GAT GAA AAA TGC GAG GTT38.9−18.3T. halophilus[12]
Table 3.  16S rRNA gene-targeted oligonucleotide probes for Enterococcaceae
NameEscherichia coli positionSequence [5′–3′]GC content (%)ΔG (kcal mol−1)SpecificityaReference
  1. aTarget organisms having a perfectly matching probe target site.

  2. bProbe was excluded from the final ECC-PhyloChip because it gave either false-positive signals with many non-target reference strains or a false-negative signal with the target strain.

EUB338338GCT GCC TCC CGT AGG AGT66.7−22.4Most Bacteria[32]
Enc131131CCC CTT CTG ATG GGC AGG66.7−21.8Most[12]
     Enterococcus spp., 
     M. plutonius 
Ecf459461GGG ATG AAC ATT TTA CTC38.9−16.8E. pseudoavium[12]
     E. casseliflavus 
     E. flavescens 
     E. dispar 
Ecg191193GCG CCT TTC AAC TTT CTT44.4−19.5E. gallinarum[12]
     E. casseliflavus 
     E. flavescens 
Ecc461462AGG GAT GAA CTT TCC ACT44.4−18.7E. cecorum[12]
     E. columbae 
Enc9393GCC ACT CCT CTT TTT CCG55.6−20.3E. hirae[12]
     E. faecium 
Edi131131CCC CCG CTT GAG GGC AGG77.8−24.4E. asini[12]
Ece9292CCA CTC ATT TTC TTC CGG50.0−19.2E. cecorum[12]
Edi137138ATG TTA TCC CCC GCT TGA50.0−20.3E. dispar[12]
Efs129129CCC TCT GAT GGG TAG GTT55.6−19.7E. faecalis[12]
Esa452453CAT TCT CTT CTC ATC CTT38.9−16.9E. saccharolyticus[12]
Eso193194ACG CAC AAA GCG CCT TTC55.6−22.2E. solitarius[12]
Esu9090CAC TCC TCT TAC TTG GTG50.0−18.4E. sulfureus[12]
Mplu464465GTC ACG AGG AAA ACA GTT44.4−18.9M. pluton[12]
Enc145b146GGG ATA ACA CTT GCA AAC44.4−18.4Enterococcus spp.,[12]
     notE. dispar, 
     E. asini, 
     E. solitarius, 
     E. columbae, 
     E. caecorum, and 
     E. faecalis 
Enc1259b1260GAA GTC GCG AGG CTA AGC61.1−21.7Enterococcus spp.,[12]
     notE. solitarius, 
     E. columbae, 
     E. caecorum, and 
     E. faecalis 

2.5PCR amplification and fluorescent labeling

For subsequent microarray hybridization, ?4.5-kb DNA fragments containing large parts of the 16S and the 23S ribosomal RNA genes were PCR-amplified from DNA of reference organisms or contaminated milk samples by using the primer pair 616V–985R [12]. PCR mixtures were prepared in 100 μl volume containing 50 pmol of each primer, 200 μM of dNTPs, 10 μl of 10× Ex Taq™ reaction buffer, and 2.5 U of Ex Taq polymerase (Takara, Biomedicals, Japan). Thermal cycling was performed by using an initial denaturation step at 94 °C for 2 min, followed by 32 cycles of denaturation at 94 °C for 1 min, 52 °C annealing for 1 min 30 s, and elongation at 72 °C for 2 min 30 s. Cycling was completed by a final elongation step at 72 °C for 5 min. For the milk samples, PCR were run in duplicates. One reaction contained 10 μl of the sample DNA while the second reaction additionally contained 1 ng of E. faecium pure culture DNA, serving as a control for successful amplification (absence of PCR inhibitors). Negative controls with no template DNA were also included in all PCR amplification experiments. Presence and size of amplification products were analyzed by 1% agarose gel electrophoresis. Purified PCR amplicons were fluorescently labeled with Cy5 by random priming according to an established protocol [19].

2.6Microarray hybridization

Vacuum-dried Cy5-labeled PCR products (400 ng) and 0.5 pmol of the Cy5-labeled control oligonucleotide CONT-COMP [19] were resuspended in 20 μl of hybridization buffer (5× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate], 1% blocking reagent [Roche, Mannheim, Germany], 0.1%n-lauryl sarcosine, 0.02% SDS, 5% formamide), denatured for 10 min at 95 °C, and immediately placed on ice. Then the solution was pipetted onto the microarray, covered with a cover slip, and inserted into a watertight custom-made hybridization chamber containing 100 μl of hybridization buffer for subsequent equilibration. Hybridization was performed overnight at 42 °C in a water bath. After hybridization, the slides were washed immediately for 5 min in 50 ml washing buffer [containing 3 M tetramethylammoniumchloride (TMAC), 50 mM Tris–HCl, 2 mM EDTA, and 0.1% SDS]. For optimization of the washing conditions, separate microarrays were washed at 46, 48, 49, 50, and 52 °C, respectively. Subsequent microarray evaluation experiments were performed at the optimal washing temperature of 49 °C. After the stringent washing, slides were washed twice with ice-cold double-distilled water, air dried, and stored in the dark at room temperature. Fluorescent images were recorded with a GMS 418 fluorescent scanner (Affymetrix, Santa Clara, USA) and quantitatively analyzed by using the ImaGene 4.0 software (BioDiscovery, Inc., Los Angeles, CA). Signal-to-noise ratios (SNRs) were determined for each probe as outlined previously [19]. Probe spots with SNRs equal to or greater than 2.0 were considered as positive.

2.716S rRNA sequence retrieval from contaminated milk

For confirmation of microarray results, almost-complete 16S rRNA gene fragments were amplified from contaminated milk DNA (sample S6) by using the primer pair 616V–630R and cloned with the TOPO TA cloning kit (Invitrogen Corp., San Diego, USA) as described previously [19]. Insert sequences were partially sequenced and phylogenetically analyzed by using the ARB program package [9] as outlined previously [30].


3.1Evaluation of the ECC-PhyloChip

A total of 52 previously published [12,31] and 7 newly designed rRNA-targeted oligonucleotide probes for members of the family Enterococcaceae (Tables 2 and 3) was spotted together with probe EUB338 that targets most bacteria including Enterococcaceae[32,33]. The microarray additionally included probes NONSENSE and CONT which served as controls for unspecific binding and hybridization efficiency, respectively [19].

Initially, the optimal washing temperature was determined experimentally as the best compromise between signal intensity and stringency for some of the probes by hybridizing the ECC-PhyloChip with fluorescently labeled target DNA of E. faecium and E. faecalis under increasing stringencies (data not shown). All following experiments were performed at the optimized washing temperature of 49 °C. Subsequently, specificities of all probes were evaluated by hybridizing fluorescently labeled target DNA from each of the 21 Enterococcaceae reference organisms and 2 control organisms (Lactococcus lactis, Staphylococcus aureus) to a separate ECC-PhyloChip. Based on the hybridization results, 19 probes had to be excluded from the final microarray (listed separately in Tables 2 and 3) because they did not show a positive signal with target organisms or exhibited non-specific binding to many non-target organisms. The remaining 41 probes showed a positive signal with the respective perfectly matching DNA and had calculated free hybridization energies in the range of −15.2 to −24.4 kcal mol−1 (Tables 2 and 3). Thirty-five of the fourty-one probes (85%) of the final version of the ECC-PhyloChip hybridized exclusively to their perfectly matching target organisms while six probes also cross-hybridized with a few non-target organisms having up to 3 mismatches in the respective probe target sites (Fig. 1). Nevertheless, for each of the 21 Enterococcaceae reference organisms, including Melissococcus plutonius and Tetragenococcus halophilus, a characteristic hybridization pattern was obtained with the final version of the ECC-PhyloChip (Fig. 1).

Figure 1.

Evaluation of the ECC-PhyloChip by hybridization with fluorescently labeled target DNA of reference organisms. Positive signals of perfectly matching and mismatching probe-target hybrids are shown by black and grey boxes, respectively. For each cross-hybridization event the number of mismatching base pairs (MM) and the free energy, ΔG (kcal mol−1), are indicated.

3.2Identification of Enterococcus species in contaminated milk by ECC-PhyloChip analysis

The developed microarray was tested with two artificially contaminated milk samples (each sample consisted of five replicates, S1–S5 and S6–S10, respectively). Milk samples were either inoculated with E. faecium or an Enterococcus species which could not be identified with the API 20 STREP test. ECC-PhyloChip analysis was performed after a pre-cultivation step of the milk samples in enrichment media. Microarray hybridization patterns were identical for replicates S1 to S5. Signal intensities for probes EUB338, Enc38i, Enc131, Edfm57, Enc93 targeting Enterococcus species at higher hierarchical levels and the species-specific probe Efm09 were above the threshold value. This hybridization pattern is indicative for E. faecium (data not shown). Hybridization patterns of replicates S6 to S10 were also identical to each other but differed from samples S1 to S5. Positive signals for the hierarchically nested probes EUB338, Enc38i, Enc131, Edfm57, Enc93, and for the species-specific probes Efm09 (targeting E. faecium), Efs129 (targeting E. faecalis), and Efa54 (targeting E. faecalis) unexpectedly indicated co-contamination of the milk with E. faecium and E. faecalis (Fig. 2). These microarray data were confirmed by comparatively analyzing cloned 16S rRNA gene sequences from replicate S6. A total of 10 clones was analyzed; eight of them were affiliated with E. faecium and two with E. faecalis (data not shown).

Figure 2.

(A) ECC-PhyloChip hybridization pattern of contaminated milk sample S6. Each probe was spotted in duplicate. Probe names are located next to each probe pair and indicate the position of the probe spots on the microarray. Perfectly matching target organisms of each probe are listed in Tables 2 and 3. Positive probes with an SNR above two are indicated in boldface type. (B) Translation of the microarray hybridization pattern indicating the presence of E. faecium and E. faecalis in milk samples S6–S10.


4.1Specificity of the ECC-PhyloChip

A previously developed set of oligonucleotide probes for detection and identification of members of the Enterococcaceae[12] was extended and spotted as microarray. One advantage of microarrays compared to more conventional hybridization formats is that miniaturized microarrays require lower amounts of labeled target nucleic acids for successful hybridization. Thus, the final version of the ECC-PhyloChip could be hybridized with 400 ng of labeled PCR product in total, whereas 4920 ng (120 ng per probe and cavity) would be needed for hybridization of the same probe set in microwell plates [12].

Because of its specific nucleotide composition and the number, position, and types of mismatches to non-target organisms, theoretically each individual probe on the microarray would require specific hybridization conditions to ensure its optimal specificity [12,18]. However, such flexibility can neither be achieved with the microarray format used in this study nor with most commercially available microarray systems which only allow performing the hybridization and/or washing step under constant (monostringent) conditions. Therefore, the design of the microarray probes and the experimental conditions were adapted for this setup using the approach of Loy et al. [19]. All probes had the same length and TMAC was added to the wash buffer to minimize the influence of GC-content differences between probes on their melting behavior. Furthermore, the optimal wash temperature was determined experimentally. Applying these optimized conditions, some of the 60 probes still showed cross-hybridizations with many non-target species and were thus removed from the microarray. Of the remaining 41 probes (14 16S and 27 23S rRNA gene-targeted), 85% hybridized exclusively to their perfectly matching target species. Only six probes of the final version of the ECC-PhyloChip hybridized with mismatching DNA from a few non-target organisms (Fig. 1). Unspecific hybridizations of some microarray probes to not fully complementary target DNA are not unexpected under monostringent conditions and can at least partly be predicted by analyzing the thermodynamic properties of a given probe-target duplex [24]. One such property, the free hybridization energy ΔG can be calculated according to the nearest neighbor model, which takes into account base pairing and base stacking interactions of probe and target molecules [29,34]. We observed that for most of the positive probe-non-target combinations on the final version of the ECC-PhyloChip, the calculated theoretical ΔG values (−20.0 to −15.5 kcal mol−1) (Fig. 1) were in the range of ΔG values of all perfectly matched probe-target hybrids (−24.4 to −15.2 kcal mol−1) (Tables 2 and 3). The only exceptions were probe Ecf459 with Enterococcus malodoratus and probe Esoha57 with Enterococcus raffinosus having ΔG values of −13.3 and −13.0 kcal mol−1, respectively (Fig. 1). Thus, although not all unspecific hybridization events can be explained by high theoretical hybridization energies, our results confirm that theoretical ΔG values are useful indicators of the actual association/dissociation behavior of a given probe-(non)target combination [24,35]. It should be noted that the nearest neighbor algorithms for calculating thermodynamic properties of probe-target duplexes were established based on hybridizations in solution. It is thus likely that the prediction of microarray hybridization events will improve further when optimized algorithms for probes immobilized on solid supports become available.

Despite the few cross-hybridizations of some probes under monostringent experimental conditions, the ECC-PhyloChip allowed unambiguous identification of all target strains (if analyzed as pure cultures) because each tested member of the Enterococcaceae is targeted by at least three ECC-PhyloChip probes having nested or parallel specificities (Fig. 1). For example, the hierarchical probe set allows differentiation of T. halophilus from Enterococcus species, although the species-specific probe Tha09 needed to be removed due to lacking specificity from the final version of the microarray (Table 2). T. halophilus is unambiguously identified by positive signals of probes EUB338 (targeting most bacteria), Enc38i (targeting most Enterococcaceae), and Esoha57 (targeting E. solitarius and T. halophilus) if presence of E. solitarius can concurrently be excluded by a negative signal of probe Eso193 (targeting E. solitarius). Furthermore, the hybridization patterns of the reference strains (Fig. 1) also demonstrated that all 19 Enterococcus species tested on the ECC-PhyloChip can be differentiated and identified even if they occur in any mixtures in the analyzed samples. The only exception is if a sample is co-contaminated with Enterococcus asini and Enterococcus dispar in the presence of some other enterococci (e.g., Enterococcus casseliflavus). In this situation E. dispar cannot be unambiguously identified. Furthermore, specific identification of T. halophilus in a complex sample is not possible with this array if the sample also contains E. raffinosus.

4.2ECC-PhyloChip analyses of food samples

The ability to correctly identify Enterococcus species in selected food samples by ECC-PhyloChip hybridization was proven using artificially contaminated milk samples. An enrichment step was included prior to DNA isolation in order to increase the number of target organisms and thus the detection sensitivity of the assay. As expected, the microarray fingerprints of the milk replicates S1 to S5 were identical to the pure culture fingerprint of the inoculum E. faecium (Fig. 1). As the milk samples were artificially contaminated with single enterococcal isolate, the identification of two distinct Enterococcus species, E. faecium and E. faecalis, in replicates S6 to S10 by ECC-PhyloChip analysis came as a surprise (Fig. 2). However, this result was confirmed by 16S rRNA gene sequencing and demonstrated that (i) the developed ECC-PhyloChip is well suited to analyze samples contaminated with more than one Enterococcus species and that (ii) the culture used for contamination of milk replicates S6 to S10 consisted of two Enterococcus species.

Similar to a recently developed multiplex PCR method for the genus- and species-specific amplification of superoxide dismutase genes (sodA) of enterococci [36], the unambiguous identification of novel Enterococcus species by ECC-PhyloChip hybridization of isolates or environmental samples is not possible. Nevertheless, positive signals for probes targeting enterococci at broader specificity (e.g., EUB338, Enc38i, and Enc131), combined with negative signals for species-specific probes targeting already recognized enterococci, are strongly indicative for the presence of yet unknown (or not targeted) Enterococcus species. If such a result is obtained, comparative 23S or 16S rRNA gene sequence analysis is recommended for phylogenetic assignment of the novel Enterococcus species.

4.3Conclusions and outlook

Routine identification of enterococci is a laborious and time-consuming process involving cultivation and subsequent phenotypic characterization of isolates. The ECC-PhyloChip described here is suitable for rapid monitoring of most recognized Enterococcus species (n= 19) at high resolution, allows large numbers of samples to be analyzed in a short time period, and has the potential for full automation. We have not systematically tested sensitivity (i.e., the lowest absolute and/or relative abundance of target organisms that are detectable) of the ECC-PhyloChip, due to the inclusion of a pre-enrichment step in the protocol. If one would attempt rendering the ECC-PhyloChip assay completely independent from cultivation one could expect a detection limit for the relative abundance of the target organisms of about 5% of the total bacterial cells in the sample [17,37]. If required, several strategies are available to further increase the sensitivity of a diagnostic microarray approach [38–40]. For example, the use of target group-selective primers (instead of general bacterial primers) allows the detection of organisms representing less than 1% of all bacteria in a complex sample [24]. In this context, it is important to note that Enterococcaceae-specific primer pairs suitable for the amplification of large 16S and 23S rRNA gene fragments are already available [12] and could be used for cultivation-independent ECC-PhyloChip-based detection of low-abundant Enterococcus species in complex food and clinical samples.


This research was supported by grants of the Bayerische Forschungsstiftung (Development of Oligonucleotide-DNA-Chips, project 368/99) to K.H.S., W.L., and M.W., and a Marie Curie Intra-European Fellowship within the 6th European Community Framework Programme to A.L. Support of the microarray excellence centre by the University of Vienna is acknowledged.