Comparison of phenotypic and genotypic taxonomic methods for theidentification of dairy enterococci

Paula Isabel Alves1, Maria Paula Martins2, Teresa Semedo3, José Joaquim FigueiredoMarques1,2, Rogério Tenreiro3 and Maria Teresa Barreto Crespo1,*1Instituto de Biologia Experimental e Tecnológica (IBET) and Instituto de Tecnologia Química e Biológica(ITQB), Universidade Nova de Lisboa, Apartado 12, 2781-901 Oeiras, Portugal; 2Estação AgronómicaNacional, Instituto Nacional de Investigação Agrária e das Pescas, Quinta do Marquês, 2784-505 Oeiras,Portugal; 3Departamento de Biologia Vegetal e Centro de Genética e Biologia Molecular, Faculdade deCiências, Universidade de Lisboa, Campo Grande, 1749-016 Lisboa, Portugal; *Author for correspondence(e-mail: tcrespo@itqb.unl.pt; phone: 351 214469551; fax 351 214421161)

Received 15 May 2003; accepted in revised form 15 September 2003

Key words: Enterococcus, Ewes’milk and cheese, Genomic characterization, Identification, Phenotypic charac-terization, Polyphasic taxonomy

Abstract

The purpose of the study was to assess the phenotypic and genotypic taxonomic congruence in order to allowspecies allocation of dairy enterococci. A total of 364 enterococci isolated from ewes’milk and cheese from fourPortuguese Registered Designation of Origin areas and 25 type and reference strains of Enterococcus spp. werecharacterized by a polyphasic taxonomical approach involving 40 physiological and biochemical tests, whole-cell protein profiles, amplification of 16S-23S intergenic spacer regions �ITS-PCR� and subsequent restrictionanalysis �ARDRA�. Ribotyping was also performed with reference strains and a subset of 146 isolates. Numeri-cal hierarchic data analysis showed that single-technique identification levels increase from the physiological andbiochemical tests to the protein approach, being lower with ITS/ARDRA and ribotyping. Cross-analysis con-firmed a higher unmatching level in all pairwise combinations involving physiological and biochemical data.Whole-cell protein profiles followed by ITS/ARDRA identified 89% of the enterococci. Reliable identification ofenterococci from milk and cheese could be obtained by analysis of whole-cell protein profiles. ITS-PCR can beused to confirm E. durans and E. faecium and ARDRA further confirms E. faecalis. Results revealed E. faecalis,E. durans, E. hirae and E. faecium as the prevalent species, although species prevalence showed some degree ofvariation among the areas.

Introduction

Cheeses produced from ewes’ raw milk are part of thedaily diet in Portugal and they are still produced in atraditional way in some Portuguese areas. They allhave typical features, this specificity being related tothe adventitious microbiota locally associated withraw milk and traditional cheese-making environmentsand practices.

For many years, microbiota from cheeses havebeen of interest to researchers, as it is well known thatmicrobes are involved in the production of the typi-cal organoleptic characteristics of cheeses during rip-ening. In some particular cases enterococci are knownto be the predominant lactic acid bacteria �Macedo etal. 1995; Franz et al. 1999; Suzzi et al. 2000�.Although several strains of enterococci are used asprobiotics, the presence of enterococci in food has

237© 2004 Kluwer Academic Publishers. Printed in the Netherlands.Antonie van Leeuwenhoek 85: 237–252, 2004.

been a matter of some controversy since some strainsof enterococci, particularly Enterococcus faecalis andE. faecium, are important causes of nosocomial andcommunity-acquired diseases �Johnson 1994; Jett etal. 1994; Stiles and Holzapfel 1997; Manero andBlanch 1999; Franz et al. 1999�. In addition, produc-tion of biogenic amines and intrinsic antibiotic resis-tance have provided further arguments for theirpotential pathogenic activity �Johnson 1994; Giraffa2002�.

Enterococci occur and grow in a variety of cheeses,especially artisanal ones produced in Southern Eu-rope, notably in Portugal, Spain, Italy and Greece.The main reasons for this prevalence have been con-sidered to be poor hygienic conditions during collec-tion and processing of milk but other factors canaccount for the predominance of enterococci such astheir resistance to high temperatures, their adaptabil-ity to different substrates and wide range of habitats�including warm blooded animals, soil, surface waterand plants�, as well as their production of proteolyticenzymes involved in casein degradation �Ordoñez etal. 1978; Trovatelli and Schiesser 1987; Wessels et al.1990; Litopoulou-Ttzanetaki and Tzanetakis 1992;Freitas et al. 1995; Suzzi et al. 2000�. In Mediterra-nean countries the most frequently isolated Entero-coccus species from cheeses are E. faecium, E.faecalis and E. durans �Franz et al. 1999�.

The description of the genus Enterococcus by itsgrowth characteristics and biochemical activities isvague because there are no characters that can distin-guish them unequivocally from other gram-positive,catalase-negative, coccus-shaped bacteria �Devrieseet al. 1993; Devriese and Pot 1995�. Nevertheless,Enterococcus spp. have been identified by phenotypiccharacteristics using a large number of tests �Schlei-fer and Kilpper-Bolz 1987; Facklam and Collins1989; Knudtson and Hartman 1992; Devriese et al.1993�. A 12-character based biochemical key wasproposed for Enterococcus spp. identification �Man-ero and Blanch 1999�, but previously genus assess-ment with eight physiological tests was required andlow identification thresholds were obtained for E.hirae and E. durans.

There are more reliable and rapid methods to iden-tify Enterococcus species. The development of newmethods involving protein profiles and molecular bi-ology techniques has opened new perspectives foridentification. Such techniques are SDS-PAGE ofwhole-cell proteins �Devriese et al. 1995; Teixeira etal. 1995; Descheemaeker et al. 1997; de Angelis et

al. 2001; Vancanneyt et al. 2001� and molecular biol-ogy methods such as 16S rRNA sequencing �Patel etal. 1998�, broad-range PCR using universal primersto amplify the variable regions of 16S rDNA �Mon-stein et al. 1998�, ribosomal intergenic spacer PCR�Kostman et al. 1995; Naïmi et al. 1997; Naïmi et al.1999; Tyrrell et al. 1997�, tDNA-PCR �Baele et al.2000�, randomly amplified polymorphic DNA�RAPD� analysis �Descheemaeker et al. 1997�, PCRbased assays that target the tuf gene �Ke et al. 1999�,ribotyping �Pryce et al. 1999; Lang et al. 2001� andby reverse checkerboard hybridization to chaperonin60 gene sequences �Goh et al. 2000�.

The purpose of the present study was to devise apolyphasic taxonomical approach for the species al-location of enterococci isolated from four differentPortuguese Registered Designations of Origin �RDO�areas of cheese production. The approach consideredin the present work aimed at setting up methods thatcould be implemented in reference laboratories. Ac-cording to this strategy, in a first step physiologicaland biochemical and whole-cell protein characteriza-tion was performed, and in a second phase, PCR am-plification of rDNA intergenic spacer regions �ITS-PCR�, amplified ribosomal DNA restriction analysis�ARDRA� and ribotyping were performed to selectthe best performing procedure.

Materials and methods

Sample preparation

Isolation of cultivable enterococci from milk wasperformed as follows: decimal dilutions of milk wereprepared with sterile Ringer solution �Merck, Darm-stadt, Germany� and plated on KF Streptococcus agar�Oxoid, Basingstoke, UK�. Cheese samples for theisolation of cultivable enterococci were prepared asfollows: 10 g of each sample were homogenized with90 ml of 2% �w/v� sodium citrate sterile solution for1 min in a stomacher apparatus �Masticator, IUL In-struments, Barcelona, Spain�. Decimal dilutions ofthe homogenate were prepared with sterile Ringer so-lution and plated on KF Streptococcus agar.

Microorganisms

The colonies selected from KF Streptococcus agarwere further isolated into pure cultures by streakingon the same medium and incubating for 2 days at

238

37 ºC. Unless otherwise stated enterococcal isolates,type and reference strains were grown in Brain HeartInfusion �BHI, Oxoid�. All isolates used throughoutthis work were kept at � 80 ºC.

A total of 364 presumptive enterococci, obtainedfrom ewes’milk and cheese of four different Portu-guese Registered Designation of Origin areas, Serrada Estrela �78 isolates�, Castelo Branco �98 isolates�,Nisa �68 isolates� and Azeitão �120 isolates� wereused in this work, except in ribotyping assays whereonly 146 isolates were used �Serra da Estrela�.

Twenty-five reference strains obtained from DSMZ�Deutsche Sammlung von Mikroorganismen undZellkulturen GmbH, Braunschweig, Germany� andCECT �Colección Espanola de Cultivos Tipo, Spain�were used: Enterococcus asini DSMZ 11492T, E.avium DSMZ 20679T, E. casseliflavus DSMZ20680T, E. cecorum DSMZ 20682T, E. columbae�DSMZ 7374T or CECT 4798T� E. dispar DSMZ6630T, E. durans DSMZ 20633T, E. faecalis DSMZ20478T, E. faecalis DSMZ 20376, E. faecalis CECT795, E. faecalis CECT 184 �formerly Streptococcusfaecalis subsp. liquefaciens�, E. faecalis CECT 187�formerly S. faecalis subsp. zymogenes�, E. faeciumDSMZ 20477T, E. faecium DSMZ 2146, E. flavescensDSMZ 7370T, E. gallinarum DSMZ 20628T, E. hiraeDSMZ 20160T, E. malodoratus DSMZ 20681T, E.mundtii DSMZ 4838T, E. pseudoavium DSMZ 5632T,E. raffınosus DSMZ 5633T, E. saccharolyticusDSMZ 20726T, E. solitarius DSMZ 5634T and E.sulfureus DSMZ 6905T.

When this work was performed, the species E. vil-lorum �Vancanneyt et al. 2001�, E. haemoperoxidusand E. moraviensis �Svec et al. 2001� had not beenproposed and accepted as new species.

Physiological and biochemical activities

Observation of cell morphology by phase contrastmicroscopy and Gram reaction were performed for allisolates. Each isolate and reference strains weretested, according to Lopes et al. �1999�, for the fol-lowing: growth at 10 ºC and 45 ºC; growth in 6.5%NaCl; growth at pH 9.6; hydrolysis of arginine; ac-etoin production by the Voges-Proskauer test and acidproduction from hydrocarbons. The substrates usedfor the test of acid production from hydrocarbonsources were D-���-ribose, L-���-arabinose, D-���-xylose, �-L-rhamnose, D-���-mannitol, D-���-sorbi-tol, ribitol, glycogen, glycerol, D-���-fructose,

D-���-mannose, D-���-galactose, D-���-glucose,lactose, maltose, sucrose, D-���-trehalose, D-���-cellobiose, D-���-raffinose, melibiose, D-���-me-lezitose, salicin, D-gluconate, D-���-turanose, escu-lin, �-gentibiose, D-amygdalin, inulin, arbutin,D-���-fucose and dextrin. All the substrates werepurchased from Sigma �St. Louis, Mo, USA�.

The type of hemolysis reaction was tested withColumbia agar supplemented with 5% sheep blood�BioMérieux, Marcy l´Etoile, France� according toLányí �1987�. Agglutination by antiserum �detectionof group D� was performed with the StreptococcalGrouping Kit �Oxoid�.

Whole-cell protein profiles by SDS-PAGE

Strains were grown in 5 ml of BHI �Oxoid� at 37 ºCfor 16 h. 0.5 ml of this culture, were inoculated in 50ml of BHI and were incubated at 37 ºC for 48 h withgentle shaking. Cells were harvested by centrifuga-tion at 10 700g at 4 ºC and the resulting pellet waswashed three times with sodium phosphate buffer�0.01 mol l–1� containing 0.8% �w/v� NaCl �pH 7.3�.

Preparation and analysis of whole-cell protein pro-files by conventional one-dimensional SDS-PAGEwere performed as described by Laemmli �1970� andPot et al. �1994�. The whole-cell protein extract ofPsychrobacter immobilis LMG 1125 �Laboratoriumvoor Microbiologie Gent Culture Collection, Gent,Belgium� was used as reference profile.

DNA preparation

Strains were cultured in 20 ml of BHI broth �Oxoid�at 37 ºC for 16 h, and then harvested by centrifuga-tion at 10 700g at 4 ºC. Total DNA was extracted bythe guanidium thiocyanate method �Pitcher et al.1989�.

Amplification of Intergenic Spacer Region by PCR(ITS-PCR)

The method used for species identification based onITS-PCR was adapted from Tyrrell et al. �1997�. Theoligonucleotides used in this study for the amplifica-tion of the ITS region were primers pS1490, 5’-TGCGGC TGG ATC CCC TCC TT-3’ and primer pL132,5’- CCG GGT TTC CCC ATT CGG-3’ �Normand etal. 1996� purchased from Life Technologies �Paisley,UK�. PCR amplifications were performed in aThermo RoboCycler �Stratagene, La Jolla, CA, USA�

those from Nisa and

239

in 0.2 ml reaction tubes. 50 �l of PCR mixture con-tained 1x PCR buffer �pH 8.4� �Life Technologies�,2.5 mmol l–1 MgCl2 �Life Technologies�, 0.1 mmoll–1 deoxynucleoside triphosphates �dNTPs� �LifeTechnologies�, 1 �mol l–1 of each primer, 2 U of TaqDNA polymerase �Life Technologies� and 250 ng ofenterococcal DNA quantified with a photometer�UV1101, Biotech Photometer, WPA, Cambridge,UK�. Thermocycler reactions were as follows: initialcycle of 94 ºC for 3 min, 35 cycles of 94 ºC for 1min, 55 ºC for 1 min, 72 ºC for 2 min, a final exten-sion step of 72 ºC for 7 min and thereafter cooled to4 ºC.

A 5 �l aliquot of the amplification product wascombined with 2 �l of loading buffer �bromophenolblue prepared according to Sambrook et al. 1989� andthe preparation was loaded on a 2% �w/v� agarose gel�SeaKem LE agarose, BioWhittaker Molecular Appli-cations, Rockland, ME, USA� and subjected to elec-trophoresis at 90 V for 2 h.

Amplified Ribosomal DNA Restriction Analysis(ARDRA)

Amplicons from ITS-PCR were recovered from theamplification mixture by extraction with chloroform/isoamyl alcohol �24:1�. DNA was precipitated at �20ºC with 3M sodium acetate �1:10� and 2 volumes ofethanol and re-suspended in 30 �l of sterile distilledwater �Sambrook et al. 1989�. For each restrictionendonuclease assayed �AluI, BfaI, HaeIII, HhaI,NlaIII, RsaI and Sau3AI�, a 10 �l aliquot wasdigested for 3 h with 1 U of each enzyme accordingto manufacturer’s instructions �Biolabs, New En-gland, MA, USA�. After digestion, 2 �l of loadingbuffer were added and the preparation was loaded ona 2% �w/v� agarose gel �Seakem LE agarose� andsubjected to electrophoresis at 80 V for 3 h.

Ribotyping

The genomic DNA of enterococcal isolates �ca 10�g� was digested by incubation with 30 U of BamHI,EcoRI, HindIII or PstI endonucleases according tomanufacturer’s instructions �Biolabs�. A 20 �l aliquotof the digestion mixture was combined with 5 �l ofloading buffer and the preparation was electrophore-sed on 0.8% �w/v� agarose gel at 35 V for 22 h. DNAfragments were subsequently transferred to a nylonmembrane �Hybond-N�, Amersham Biosciences,Bucks, UK� by Southern Blot �Sambrook et al. 1989�.

Hybridization was performed at 68 °C using the 16SrDNA of E. faecalis �DSMZ 20478T� as probe. Theprobe was produced using primers pA, 5’-AGAGTTTGATCCTGGCTCAG-3’ and pH, 5’-AAGGAGGTGATCCAGCCGCA-3’ �Massol-Deyaet al. 1995� and an amplification protocol identical tothe one described for the ITS region. Probe labellingwas performed with digoxigenin-11-dUTP �RandomPrimed Labelling Kit, Roche, Mannheim, Germany�and hybridization signals were detected with Dig-High Prime DNA Labelling and Detection Starter KitII �Boehringer Mannheim, Mannheim, Germany�, asrecommended by the manufacturer.

Data analysis

All the data obtained were analyzed using a doublestrategy: a� the use of type and reference strains forcomparison; b� assessment of the taxonomic congru-ence between the physiological and biochemical andgenotypic methods applied to the whole set ofenterococcal isolates, using pairwise comparisonsamong physiological and biochemical-based, protein-based and ITS/ARDRA-based methods.

Physiological and biochemical characteristics werecoded as 0 for negative and 1 for positive and analy-sed by hierarchic clustering using the software pack-age BioNumerics version 1.5 �Applied Maths, Kortr-ijk, Belgium, 1998�. Resemblance was computed withthe simple matching coefficient �SSM� and agglomer-ative clustering was performed with the unweightedaverage linkage �UPGMA�. Reproducibility was as-sessed by calculation of the average error probability�p�, as defined by Sneath and Johnson �1972�, andusing duplicates of strains randomly selected amongthe isolates from the four RDO areas.

BioNumerics 1.5 software was also used to regis-ter whole-cell protein electrophoretic patterns, nor-malising densitometric traces, calculating Pearsonproduct-moment correlation coefficient �r� betweenstrains and performing cluster analysis by theUPGMA algorithm. The reproducibility level of themethod was assessed by the minimum correlationvalue �r� observed in a random sample of 10% dupli-cate protein extracts.

ITS-PCR profiles and restriction patterns obtainedby Amplified Ribosomal DNA Restriction Analysis�ARDRA� were analysed by BioNumerics 1.5 soft-ware using the Dice correlation coefficient as similar-ity measure and UPGMA as clustering algorithm.

240

Ribotyping patterns were visually analysed and amatrix of Boolean data �presence/absence of hybrid-ization fragments� was constructed. Numerical analy-sis of ribopatterns was performed with NTSysprogram �Numerical Taxonomy and MultivariateAnalysis System; F.J. Rohlf, NY 11794, USA� usingDice coefficient and UPGMA.

Results

Identification based on physiological andbiochemical characteristics

Twenty-two type and reference strains from culturecollections �the ones mentioned in Materials andMethods with the exceptions of E. columbae DSMZ7374T, E. flavescens DSMZ 7370T and E. asiniDSMZ 11492T� and the 364 presumptive Enterococ-cus spp. isolates �gram-positive non-motile cocci,catalase-negative� were tested for the following:growth at 10 ºC and 45 ºC, growth in 6.5% NaCl,growth at pH 9.6, hydrolysis of arginine, VP test, typeof hemolysis, presence of group D antigen and acidi-fication of hydrocarbons as described in Materials andMethods.

The reproducibility level of the method wasassessed by the evaluation of the average error prob-ability and the value obtained was 99.5%.

The boolean data matrix of 40 characteristics and386 operational taxonomic units �OTUs� was ana-lyzed by hierarchic numerical methods and clustersfor species identification were delineated taking intoaccount the clustering pattern of type and referencestrains. Only clusters containing type and referencestrains from single species were assumed as leadingto identification of isolates, as depicted in the dendro-gram of Figure 1. When similarity levels of suchclusters were globally analyzed, a minimum of 80%S-level was chosen and eight clusters considered.

Cluster I �80.0% S-level�, which included two iso-lates and the type strain DSMZ 6630T, was identifiedas E. dispar. Cluster II �84.2% S-level� included thetype strain and 13 isolates of E. durans. Cluster III�83.1% S-level�, defined as E. hirae, grouped 56 iso-lates with the type strain of this species. Cluster IV�82.4% S-level�, E. faecium, and cluster V �82.5% S-level�, E. raffınosus, allowed the identification of 9and 11 isolates, respectively. Among the isolatesidentified as E. faecalis, 24 were grouped with tworeference strains in sub-cluster VI A �82.5% S-level�

and 177 clustered with the type strain and two otherreference strains in sub-cluster VI B �86.2% S-level�.As the type strain of E. solitarius showed 100% simi-larity with strain CECT 795 of E. faecalis, this strainremained included in sub-cluster VI B. Six isolateswere identified as E. mundtii �cluster VII; 87.9% S-level� and three as E. cecorum �cluster VIII; 84.6%S-level�. In total, 82.7% of the isolates were consid-ered to be identified by this approach. The majorityof isolates were identified as E. faecalis �55%, 201/364 isolates� and E. hirae �15%, 56/362 isolates�.

Protein based identification

An analysis of the electrophoretic whole-cell proteinprofiles of the 364 ewes’milk and cheese enterococ-cal isolates and 22 type and reference strains �thesame used in physiological and biochemical identifi-cation� was performed and the profiles obtained werecompared by a densitometric-based numerical analy-sis, with those obtained for the type and referencestrains of Enterococcus spp. The observed reproduc-ibility level of the method and the cut-off value fordiscrimination of isolates was 95%.

The relationships among isolates and type and ref-erence strains were analyzed in the dendrogram de-rived from unweighted pair group average linkage ofr-values for protein electropherograms using a strat-egy similar to the one described for physiological andbiochemical data. Seven clusters were assumed as al-lowing identification at species level �Figure 2�, lead-ing to a minimum of 50% r-value on protein basedidentification.

Cluster I �69.9% r-value� grouped three isolatesand the type strain of E. casseliflavus. Cluster II�60.4% r-value�, including 21 isolates and strainDSMZ 2146, was identified as E. faecium; althoughthe type strain of this species, DSMZ 20477T, did notcluster with this group of strains. Clusters III �60.1%r-value�, IV �58.7% r-value� and V �52.8% r-value�were identified as E. durans �90 isolates�, E. hirae �52isolates� and E. dispar �2 isolates�, respectively.Cluster IV was identified as E. faecalis �52.8%r-value�. It contained 139 isolates, the type strain andthe other four reference strains of this species; thetype strain CECT 4798T of E. columbae and the typestrain of E. solitarius showed r-values of 97.0% and95.5% with E. faecalis CECT 184, respectively, andwere also included in this cluster. Cluster VII �72.5%r-value� grouped the type strain and two isolates ofE. raffınosus. Protein profiles allowed the identifica-

241

Figure 1. Dendrogram, obtained from physiological and biochemical characteristics, of 364 enterococcal isolates from ewes’cheese and milkand 22 reference strains. Clustering was performed with simple matching and UPGMA.

242

Figure 2. Dendrogram derived from unweighted pair group average linkage of r-values for protein electropherograms of 364 enterococcalisolates from ewes’cheese and milk and 22 reference strains.

243

tion of 309 isolates, representing 84.9% of the totalisolates. The majority of isolates were identified as E.faecalis �38%, 139/364 isolates�, E. durans �25%, 90/364 isolates� and E. hirae �14%, 52/364 isolates�.

ITS/ARDRA-based identification

The 16S-23S intergenic spacer region was amplifiedby PCR for the 364 Enterococcus isolates and 24 typeand reference strains �all but E. columbae CECT4798T�. The size of the ITS amplicons ranged from280 bp to 620 bp, with E. hirae DSMZ 20160T

showing three main bands and all the other type andreference strains only two �Figure 3A�. Among theisolates, 90% had also two amplicons, 9% had threeand the remainder displayed four main bands.

In the dendrogram derived from the computer-as-sisted analysis of ITS-PCR profiles �data not shown�,56% of the isolates could not be identified as theyclustered at 100% similarity with the representativesof 12 of the 19 species under study. The type strainsof E. mundtii and E. dispar also clustered together atmaximum similarity. Besides E. hirae �3 amplicons�,only E. durans, E. faecium and E. cecorum were dis-

tinguishable, allowing the identification of 83 isolatesas E. durans and 17 isolates as E. faecium, taking inaccount their unique clustering at 100% similaritylevel with the type and reference strains of these spe-cies. Therefore, the analysis of ITS-PCR resultsshowed an identification level of 27.5%.

As most of the species have very similar ITS-PCRprofiles, making their distinction very difficult,restriction of the amplicons in all type and referencestrains was performed with endonucleases AluI, BfaI,HaeIII, HhaI, NlaIII, RsaI and Sau3AI to check fortheir discriminating ability �data not shown�. SinceSau3AI showed the highest discriminating ability, thisenzyme was selected �Figure 3B� and ARDRA wasperformed in the author’s laboratories �Instituto deBiologia Experimental e Tecnológica and Faculdadede Ciências da Universidade de Lisboa� after halvingthe set of isolates. Although restriction patterns withineach laboratory were constant, inter-laboratory repro-ducibility was reduced by differences in amplificationyield and electrophoresis resolution, preventing thecombined analysis by automatic computer-assistedtool. In each laboratory, a dendrogram was producedfrom restriction patterns of isolates and a representa-

Figure 3. ITS-PCR profiles �A� and Sau3AI ARDRA profiles �B� obtained for the type and reference strains of Enterococcus spp. Lanes: M,100 bp DNA ladder �Gibco BRL�; 1, E. faecalis �DSMZ 20478T�; 2, E. faecalis �DSMZ 20376�; 3, E. faecalis �CECT 184�; 4, E. faecalis�CECT 187�; 5, E. faecalis �CECT 795�; 6, E. faecium �DSMZ 20477T�; 7, E. faecium �DSMZ 2146�; 8, E. avium �DSMZ 20679T�; 9, E.pseudoavium �DSMZ 5633T�; 10, E. raffınosus �DSMZ 5633T�; 11, E. saccharolyticus �DSMZ 20726T�; 12, E. casseliflavus �DSMZ 20680T�;13, E. durans �DSMZ 20633T�; 14, E. gallinarum �DSMZ 20628T�; 15, E. hirae �DSMZ 20160T�; 16, E. malodoratus �DSMZ 20680T�; 17,E. mundtii �DSMZ 4838T�; 18, E. sulfureus �DSMZ 6905T�; 19, E. dispar �DSMZ 6630T�; 20, E. cecorum �DSMZ 20682T�; 21, E. columbae�DSMZ 7370T�; 22, E. flavescens �DSMZ 7374T�; 23, E. asini �DSMZ 11492T�.

244

tive member of each profile was compared individu-ally to the digestion profiles of all type and referencestrains. Species allocation of isolates was possible af-ter the production of new dendogram. The results ob-tained not only confirmed ITS-PCR results, but alsoallowed the identification of an extra 86 isolates, in-creasing the identification level to 52%. According toITS/ARDRA-based identification the most prevalentspecies were E. durans �25%, 90/364 isolates�, E.faecalis �22%, 79/364 isolates� and E. faecium �5%of total, 17/364 isolates�.

Ribotype-based identification

Selection of the restriction endonuclease was per-formed using BamHI, EcoRI, HindIII and PstI to di-gest the DNA of the type and reference strains �resultsnot shown�. The enzyme HindIII was selected takinginto account the number of hybridization fragmentsobtained, their relative separation and the discrimina-tive ability for enterococcal species.

Ribotyping with HindIII was performed for the 146isolates from Nisa and Serra da Estrela, and thetwenty-two type and reference strains �Figure 4� used

in the physiological, biochemical and protein-basedmethods. Visual comparison of the hybridization pro-files of the isolates and type and reference strains fol-lowed by dendrogram analysis led to the identifica-tion of 43 isolates as E. faecalis and two isolates asE. faecium. 80% of the isolates that had been identi-fied by protein profiles as being E. faecalis gave asimilar result by ribotyping, pointing for a good cor-relation between the two methods. Nevertheless, ri-botyping showed an extremely low identificationlevel �30.8%� in this group of ewes’ cheese and milkenterococcal isolates.

Pairwise comparisons of identification methods

In order to assess the agreement between the identi-fication methods applied to the whole set of entero-coccal isolates, pairwise comparisons were performedamong physiological and biochemical-based, protein-based and ITS/ARDRA-based identifications.

The results of this analysis are displayed in Figure5. For each pairwise comparison, a matrix output wasused where the central diagonal represents thematched identifications between the two methods un-

Figure 4. Enterococcal ribotyping profiles obtained with HindIII for type and reference strains and some Serra de Estrela cheese isolates�QSE�. Lanes: M, 1 kbp DNA ladder �Gibco BRL�; 1, E. faecalis �DSMZ 20376�; 2, E. sulfureus �DSMZ 6905T�; 3, E. faecium �DSMZ2146�; 4, E. casseliflavus �DSMZ 20680T�; 5, E. hirae �DSMZ 20160T�; 6, E. dispar �DSMZ 6630T�; 7, E. cecorum �DSMZ 20682T�; 8, E.mundtii �DSMZ 4838T�; 9, E. gallinarum �DSMZ 20628T�; 10, E. durans �DSMZ 20633T�; 11, E. faecium �DSMZ 20477T�; 12, E.pseudoavium �DSMZ 5632T�; 13, E. malodoratus �DSMZ 20681T�; 14, E. faecalis �DSMZ 20478T�; 15, E. avium �DSMZ 20679T�; 16, QSE74; 17, QSE 27G; 18, QSE 29A; 19, QSE 29B; 20, QSE 29C; 21, QSE 110; 22, QSE 117; 23, QSE 2; 24, QSE 32; 25, QSE 65; 26, QSE115; 27, QSE 118; 28, QSE 119; 29, QSE 120; 30, QSE 121.

245

der analysis and delineates two triangles containingthe unmatched identifications; the isolates identifiedonly by one of the methods are located in the outerrows and those without any identification at the bot-tom right corner. A more scattered distribution is ob-tained whenever physiological and biochemical basedidentification is involved �Figure 5A and Figure 5B�,in contrast with the gathering of isolates, both in thecentral diagonal and the outer rows, observed in thecombination of other methods �Figure 5C�. The un-matching level of the protein-ITS/ARDRA combina-tion is 8.8%, much lower than the 20-30% observedfor the physiological and biochemical-protein andphysiological and biochemical-ITS/ARDRA combi-nations.

Although the number of isolates identified simulta-neously by both methods �matches� is higher for thephysiological and biochemical-protein combination�44.5%� than for the protein-ITS/ARDRA combina-tion �38.5%�, the additional number of isolates iden-tified only by one of the methods �assumed assuccessful identifications� varies inversely. As a con-sequence, minimal identification levels vary between64.9% for the physiological and biochemical-proteincombination and 79.9% for the protein-ITS/ARDRA.This last dual method thus allies the highest identifi-cation and the lowest unmatching levels.

Assuming ITS/ARDRA as a subsequent confirma-tive step for the whole-cell protein analysis on a two-step identification procedure, 143 isolates were thenidentified as E. faecalis, 106 isolates as E. durans, 39as E. hirae, 28 as E. faecium, 3 as E. casseliflavus, 2as E. dispar and 2 as E. raffınosus, leading to a glo-bal identification level of 89%. Overall diversity ofresults per group of isolates, per approach and pergeographical area are given in Table 1. Analysis ofprevalence among isolates from RDO areas showedthat E. faecalis varies from 16% �Nisa� to 57%�Castelo Branco�, E. durans from 24% �Azeitão� to40% �Serra da Estrela�, E. faecium from 3% �Castelo

Figure 5. Pairwise comparisons of the identification �ID� methodsapplied to ewes’ milk and cheese enterococcal isolates. �A� Physi-ological and biochemical-based ID versus protein-based ID; �B�Physiological and biochemical-based ID versus ITS/ARDRA-based ID; �C� Protein-based ID versus ITS/ARDRA-based ID. Thenumber of isolates is indicated for each crossed situation and theshading is proportional to its relative prevalence �from 1 to 35%�.Letters refer to the outcome of the identification process for eachmethod: C – E. cecorum; D – E. durans; E – E. faecium; F – E.faecalis; H – E. hirae; L – E. casseliflavus; M – E. mundtii; P – E.dispar; R – E. raffınosus; U – unidentified.

246

Table 1. Distribution of enterococci from different RDO areas according to species allocation and ID profiles obtained with different ap-proaches

Species ID profilea RDO area

Ph P I A R Azeitão Castelo Branco Nisa Serra da Estrela

E. faecalis�143� F F U F F 6 9 7 32

F F U U F 16 34 0 0F F U U U 6 3 1 0F F U F U 3 1 2 3M F U U F 1 2 0 0U F U F F 1 0 0 0M F U F F 1 0 0 0H F U U F 1 0 0 0U F U F U 0 1 0 0H H U F U 1 1 0 0F H U F F 0 1 1 1U H U F U 0 1 0 0F D U F U 0 2 0 0F D U F F 0 0 0 1F E U F U 0 1 0 0F U U F U 2 0 0 0U U U F U 1 0 0 0Total 39 56 11 37

E. durans�106� U D D U U 7 0 1 14

H D D U U 1 1 7 8F D D U U 2 7 5 5D D D U U 0 5 1 0E D D U U 2 0 0 0U D D U U 0 0 1 0H D U U U 4 0 1 1U D U U U 4 0 0 1F D U D U 0 3 0 0F D U U U 0 1 1 1D D U U U 0 0 1 0C D U U U 0 1 0 0E E D U U 2 0 0 0H E D U U 1 0 0 0F E D U U 0 0 2 0E H D U U 1 0 0 0H H U D U 0 1 0 0F H D U U 0 1 0 0M F D U F 0 1 0 0F F D U F 1 2 0 0H F D U U 0 0 0 1F U D U U 2 0 0 0E U D U U 1 0 0 0H U D U U 1 0 0 0H U U D U 2 0 0 0P U U D U 0 1 0 0Total 31 24 20 31

E. faecium�28� F E E U U 1 1 0 0

E E E U E 0 0 0 1H E E U U 0 0 0 1F E U U U 4 2 0 1R E U U U 1 0 0 0

247

Branco� to 11% �Azeitão� and E. hirae from 0%�Serra da Estrela� to 17% �Azeitão�. Castelo Branco,Serra da Estrela and Azeitão have similar distributionpatterns, with E. faecalis predominating over E. du-rans, whereas the reverse happens in Nisa. Regardingthe prevalence of unidentified isolates, a distinct pat-

tern was observed with 1% in Castelo Branco, 5% inSerra da Estrela, 14% in Azeitão and 28% in Nisa,pointing further to the distinctiveness of this last RDOarea.

Species ID profilea RDO area

Ph P I A R Azeitão Castelo Branco Nisa Serra da Estrela

M E U U U 1 0 0 0U E U U U 1 0 0 0H 1 0 0 0F F E U F 0 0 1 0U F E U U 0 0 1 0H F E U U 0 0 1 0F F E U F 1 0 0 0H H E U U 0 2 0 0U H E U U 0 1 0 0F H E U E 0 0 0 1U D E U E 0 0 0 1U U E U U 1 0 1 0F U E U U 2 0 0 0Total 13 6 4 5

E. hirae�39� H H U U U 13 2 1 0

F H U U U 2 6 1 0D H U U U 0 1 4 0U H U U U 2 1 2 0R H U U U 3 0 0 0H H U U F 0 0 1 0Total 20 10 9 0

E. raffınosus�2� R R U U U 1 0 1 0

Total 1 0 1 0E. casseliflavus�3� E L U U U 1 0 0 0

U L U U U 0 1 0 0C L U U U 0 0 1 0Total 1 1 1 0

E. dispar�2� D P U U U 0 1 0 0

C P U U U 0 0 0 1Total 0 1 0 1

Unidentified�41� U U U U U 6 0 12 2

F U U U U 7 0 3 1R U U U U 1 0 4 0H U U U U 2 0 0 0E U U U U 1 0 0 0P U U U U 0 0 1 0F U U U F 0 1 0 0Total 17 1 20 3

a �Ph�, protein profiles �P�, ITS-PCR �I�, ARDRA �A�and ribotyping �R�. ID codes : F – E. faecalis; E – E. faecium; D – E. durans ;H – E. hirae; L – E. casseliflavus; C – E. cecorum; P – E.dispar; R – E. raffınosus; M – E. mundtii; U – unidentified.

Table 1. Continued

E U U U

ID profiles refer to the identification obtained with physiological and biochemical data

248

Discussion

Analysis of results from physiological and biochemi-cal tests was performed to determine characteristicsthat are usually considered as typical for the genusEnterococcus �such as growth at 45 ºC, 10 ºC, pH 9.6and with 6.5% NaCl� and to allow a preliminarycharacterization of the isolates. Besides being verytime consuming, this type of work yielded resultsthat, in terms of a taxonomic identification, did notalways match the results obtained by other methods.

Various factors can be considered to explain thismismatch in identification results. Notably, that aninadequate choice of physiological and biochemicaltests was used, as some of the tests described byManero and Blanch �1999� to discriminate enterococ-cal species have not been used. Nevertheless, distinc-tion between species based on one character, asproposed in the latter identification scheme for E.faecalis and E. solitarius and for E. durans and E.hirae, was apparently no more powerful a strategy, asdemonstrated by the low identification thresholds ob-tained by Manero and Blanch �1999� for these lasttwo species.

Several authors �Pot et al. 1994; Teixeira et al.1995; Teixeira et al. 1996� have shown that analysisof whole-cell protein profiles is a reliable tool for dif-ferentiation and identification of enterococcal species.Therefore we examined whether a strain isolated forthe first time from a traditional/fermented product,and that did not cluster well with the type and refer-ence strains from culture collections when physiolog-ical and biochemical characteristics were studied,could be differentiated by whole cell protein-profiles.The basis for the use of whole-cell protein finger-prints in microbial taxonomy is that bacterial strainswith 90 to 100% DNA relatedness display only mi-nor differences in their protein fingerprints and thatstrains with at least 70% DNA relatedness have simi-larities in their electrophoretic patterns. Therefore,this technique is very sensitive and provides informa-tion on the similarity of strains within the same spe-cies or subspecies �Vandamme et al. 1996�. Theresults obtained show that whole-cell protein analysisallowed the identification of about 85% of theisolates. The remaining 15% did not cluster with anyof the Enterococcus type strains.

The use of the spacer region between the 16S andthe 23S rDNA genes was evaluated as a rapid tool toseparate strains and confirm results obtained previ-ously. The genus Enterococcus is characterized by a

high variability of the ITS zone, as isolates can fre-quently possess intergenic regions of different dimen-sions. Most of the isolates had ITS profiles of two orthree different sizes. The most frequent amplicons hada size of approximately 400 and 500 bp. The ampli-con of ca. 500 bp can be explained by the presenceof the gene coding for the alanine tRNA of the inter-genic region, while the amplicon of ca. 400bp can bedue to absence of this gene �Tyrrell et al. 1997�. It isalso important to note that the alanine tRNA gene canhave differences in its size that would originate ITSregions of different sizes. Previous studies by Naïmiet al. �1997; 1999� had already detected the presenceof insertions of 114 and 107 bp in strains of E.faecium and E. hirae, respectively; this could explainthe appearance of fragments with larger sizesobtained for both species �ca. 600 bp�.

ITS-PCR allowed a clear identification of E.faecium, E. hirae and. E. durans, species displayingdistinct ITS profiles. As the other type strains groupedtogether, no further identification was allowed. Theseparation could eventually be increased by the useof acrylamide gels �Tyrrell et al. 1997� or by capil-lary electrophoresis as performed on tDNA-PCRproducts by Baele et al. �2000�. When considering thenumber of strains that were identified as E. durans byITS-PCR profiles and also as such by SDS-PAGE,only 81% of the ITS-PCR profiles were simulta-neously identified by protein profiling. Regarding E.faecium, only 23.5% were identified by proteinanalysis. ITS-PCR seems therefore to be useful forconfirmation of protein-based identification.

Restriction analysis of ITS amplicons with Sau3AIconfirmed some of the E. faecalis, E. durans and E.faecium identifications already obtained by physio-logical and biochemical characteristics and proteinmethods. The most interesting development was thepossibility of detection by ARDRA of isolates thatformerly could not be identified by ITS-PCR. Amongthese isolates, 84% had been identified as E. faecalisby SDS-PAGE.

The number of fragments obtained after restrictionof total DNA with an endonuclease is generally veryhigh. As this analysis can be very troublesome it wasdecided to perform ribotyping i.e. typing of the strainsusing as probe the rDNA 16S of the type strain of thegenus Enterococcus – E. faecalis DSMZ 20478T. Thismethod allows the visualisation of a high number ofgenomic restriction fragments containing ribosomalgenes. As the number of rrn operons, their localiza-tion and sequence �both internal and in neighbour re-

249

gions� are variable, it is possible to obtain differentrestriction profiles �ribotypes�. Species genotypicallyhomogeneous can present just one or a few ribotypes,while heterogeneous ones can present several �Gri-mont and Grimont 1986�.

E. faecalis and E. faecium present characteristic ri-botypes �Figure 4� allowing the identification of theisolates. The results of this technique were in closeagreement with the results obtained by the pairwisecombination of whole cell protein analysis and ITS/ARDRA. When these two species are considered, ri-botyping is a very promising methodology. Noidentification was obtained for the majority of isolatesfrom ewes’ milk and cheese tentatively considered asE. durans and E. hirae by protein profiles and ITS-ARDRA. As the ribotypes of both type strains areclearly distinct from each other and other species, theproblem may be rooted in a greater genomic diver-sity of ewes’ milk and cheese isolates such that theirribotypes did not match that of the reference strains.Indeed, high levels of genomic diversity were alsofound in other non-starter ewes’ cheese bacteria be-longing to the genus Lactobacillus �de Angelis et al.2001�.

Polyphasic taxonomy was a concept created byColwell �1970� that has evolved to the present daydefinition of a consensus type of taxonomy, with theaim of utilizing all the available data to delineateconsensus groups �Gillis et al. 2001�. The proposedapproach is laborious and is not suitable for imple-mentation in routine identification. Nevertheless theresults obtained in this work allowed the proposal ofa polyphasic taxonomic approach for the identifica-tion of isolates from dairy products. The sequentialpolyphasic identification proposed here will certainlyhelp to gather data suitable for an approach based onthe construction of ‘virtual’ taxa by aggregation oftaxonomic units in a multidimensional space of eco-logically significant characters that would be moreflexible. The need for such kind of approaches has al-ready been felt outside microbiology when dealingwith large numbers of samples from regionally vari-able populations �DuPraw 1964�. Moreover, localmicrobial populations constitute privileged environ-ments for gene transfer and speciation, and the above‘virtual’ taxa could accept more easily new isolatesfrom the same or from analogous environments.

Acknowledgements

This work was supported by Fundação para a Ciênciae Tecnologia Program PRAXIS XXI projects 2/2.1/BIA/309/94 and 2/2.1/BIO/1121/95 and Program SA-PIENS project POCTI/AGR/36165/99.

References

Baele M., Baele P., Vaneechoutte M., Storms V., Butaye P.,Devriese L.A., Verschraegen G., Gillis M. and Haesebrouck F.2000. Application of tRNA intergenic spacer PCR for identifi-cation of Enterococcus species. J. Clin. Microbiol. 38: 4201–4207.

Colwell R.R. 1970. Polyphasic taxonomy of the genus Vibrio: nu-merical taxonomy of Vibrio cholerae, Vibrio parahaemolyticusand related Vibrio species. J. Bacteriol. 104: 410–433.

de Angelis M., Corsetti A., Tosti N., Rossi J., Corbo M.R. andGobbetti M. 2001. Characterization of non-starter lactic acidbacteria from Italian ewe cheeses based on phenotypic,genotypic, and cell wall protein analysis. Appl. Environ. Microb.67: 2021–2020.

Descheemaeker P., Lammens C., Pot B., Vandamme P. and Goos-sens H. 1997. Evaluation of arbitrarily primed PCR analysis andpulsed-field gel electrophoresis of large Genomic DNA frag-ments for identification of Enterococci important in humanmedicine. Int. J. Syst. Bacteriol. 47: 555–561.

Devriese L.A., Pot B. and Collins M.D. 1993. Phenotypic identi-fication of the genus Enterococcus and differentiation of phylo-genetically distinct enterococcal species and species groups. J.Appl. Bacteriol. 75: 399–408.

Devriese L.A. and Pot B. 1995. The genus Enterococcus. In: WoodB.J. and Holzapfel W.H. �eds�, The Genera of Lactic Acid Bac-teria. Blackie Academic and Professional, London, UK, pp.327–367.

Devriese L.A., Pot B., Van Damme L., Kersters K. andHaesebrouck F. 1995. Identification of Enterococcus speciesisolated from foods of animal origin. Int. J. Food Microbiol. 26:187–197.

DuPraw E.J. 1964. Non-linnean taxonomy. Nature 202: 849–852.Facklam R.R. and Collins M.D. 1989. Identification of Enterococ-

cus species isolated from human infections by a conventionaltest scheme. J. Clin. Microbiol. 27: 731–734.

Franz C.M., Holzapfel W.H. and Stiles M.E. 1999. Enterococci atthe crossroads of food safety? Int. J. Food Microbiol. 47: 1–24.

Freitas A.C., Pais C., Malcata F.X. and Hogg T.A. 1995. Microbio-logical characterization of Picante de Beira Baixa cheese. J.Food Protect. 59: 155–160.

Gillis M., Vandamme P., De Vos P., Swings J. and Kersters K. 2001.Polyphasic taxonomy. In: Boone D.R. and Castenholz R.W.�eds�, Garrity G.M. �Ed-in-chief� Bergey’s Manual of System-atic Bacteriology, Vol 1. Springer-Verlag, New York, USA, pp.43–48.

Giraffa G. 2002. Enterococci in foods. FEMS Microbiol. Rev. 26:163–171.

Goh S.H., Facklam R.R., Chang M., Hill J.E., Tyrrell G.J., BurnsE.C.M., Chan D., He C., Rathim T., Shaw C. and HemmingsenS.M. 2000. Identification of Enterococcus species and phenotyp-

250

ically similar Lactococcus and Vagococcus species by reversecheckerboard hybridization to chaperonin 60 gene sequences. J.Clin. Microbiol. 38: 3953-3959.

Grimont F. and Grimont P.A.D. 1986. Ribosomal ribonucleic acidgene restriction patterns as potential taxonomic tools. Ann. Inst.Pasteur/Microbiol. 137: 165–175.

Jett B., Huycke M.M. and Gilmore M.S. 1994. Virulence of en-terococci. Clin. Microbiol. Rev. 7: 162–478.

Johnson A.P. 1994. The pathogenicity of enterococci. J. Antimi-crob. Chemoth. 33: 1083–1089.

Ke D., Picard F.J., Martineau F., Ménard C., Roy P.H., OuelletteM. and Bergeron M.G. 1999. Development of a PCR assay forrapid detection of enterococci. J. Clin. Microbiol. 37: 3497–3503.

Knudtson L.M. and Hartman P.A. 1992. Routine procedures forisolation and identification of enterococci and fecal streptococci.Appl. Environ. Microb. 58: 3027–3031.

Kostman J.R., Alden M.B., Mair M., Edlind T.D., LiPuma J.J. andStull T.L. 1995. A universal approach to bacterial molecular epi-demiology by polymerase chain reaction ribotyping. J. Infect.Dis. 171: 204–208.

Laemmli U.K. 1970. Cleavage of stuctural proteins during the as-sembly of the head of bacteriophage T4. Nature 227: 680–685.

Lang M.M., Ingham S.C. and Ingham B.H. 2001. Differentiationof Enterococcus spp. by cell membrane fatty acid methyl esterprofiling, biotyping and ribotyping. Lett. Appl. Microbiol. 33:65–70.

Lányí B. 1987. Classical and rapid identification methods formedically important bacteria. Method Microbiol. 19: 1–67.

Litopoulou-Tzanetaki E. and Tzanetakis N. 1992. Microbiology ofwhite brined cheese made from raw goat milk. Food Microbiol.9: 13–19.

Lopes M.F.S., Pereira C.I., Rodrigues F.M.S., Martins M.P.,Mimoso M.C., Barros T.C., Figueiredo Marques J.J., TenreiroR.P., Almeida J.S. and Barreto Crespo M.T. 1999. Registereddesignation of origin areas of fermented food products definedby microbial phenotypes and artificial neural networks. Appl.Environ. Microb. 65: 4484–4489.

Macedo A.C., Malcata F.X. and Hogg T.A. 1995. Microbiological

79: 1–11.Manero A. and Blanch A.R. 1999. Identification of Enterococcus

spp. with a Biochemical Key. Appl. Environ. Microb. 65: 4425–4430.

Massol-Deya A., Odelson D.A., Hickey R.F. and Tiedje J.M. 1995.Bacterial community fingerprinting of amplified 16S and 16-23Sribosomal DNA gene sequences and restriction endonucleaseanalysis �ARDRA�. In: Akkermans A.D.L., van Elsas J.D. andde Bruijn F.J. �eds�, Molecular Microbial Ecology Manual, 3.3.2.Kluver Academic Publishers, Dordrecht, The Netherlands, pp.1–8.

Monstein H.-J., Quednau M., Samuelsson A., Ahrné S., IsakssonB. and Jonasson J. 1998. Division of the genus Enterococcus intospecies groups using PCR-based molecular typing methods. Mi-

Naïmi A., Beck G. and Branlant C. 1997. Primary and secondarystructures of rRNA spacer regions of enterococci. Microbiology143: 823–824.

Naïmi A., Beck G., Monique M., Levébvre G. and Branlanti C.1999. Determination of the nucleotide sequence of the 23S ribo-somal RNA and flanking spacers of an Enterococcus faecium

strain, reveals insertion-deletion events in the ribosomal spacer1 of enterococci. Syst. Appl. Microbiol. 22: 9–21.

Normand P., Ponsonnet C., Nesme X., Neyra M. and Simonet P.1996. ITS analysis of prokaryotes. In: Akkermans A.D.L., vanElsas J.D. and de Bruijn F.J. �eds�, Molecular Microbial Ecol-ogy Manual, 3.4.5. Kluver Academic Publishers, Dordrecht, TheNetherlands, pp. 1-12.

Ordoñez J.A., Barneto R. and Ramos M. 1978. Studies onManchego cheese ripened in olive oil. Milchwissenschaft 33:609–612.

Patel R., Piper K.E., Rouse M.S., Steckelberg J.M., Uhl J.R.,Kohner P., Hopkings M.K., Cockerill III F.R. and Kline B.C.1998. Determination of 16S rRNA sequences of enterococci andaplication to species identification of nonmotile Enterococcusgallinarum isolates. J. Clin. Microbiol. 36: 3399–3407.

Pitcher D.G., Saunders N.A. and Owen R.J. 1989. Rapid extrac-tion of bacterial genomic DNA with guanidium thiocyanate.Lett. Appl. Microbiol. 8: 151–156.

Pot B., Vandamme P. and Kersters K. 1994. Analysis of electro-phoretic whole-organism protein fingerprints. In: Goodfellow M.

�eds�, Chemical Methods in ProkaryoticSystematics. Wiley, Chichester, UK, pp. 493–521.

Pryce T.M., Wilson R.D. and Kulski J.K. 1999. Identification ofenterococci by ribotyping with horseradish-peroxidase labelled16S rDNA probes. J. Microbiol. Methods 36: 147–155.

Sambrook J., Fritsch E.F. and Maniatis T. 1989. Molecular Clon-ing: a Laboratory Manual. Cold Spring Harbor Laboratory Press,New York, USA.

Schleifer K.M. and Kilpper-Bolz R. 1987. Molecular and chemot-axonomic approaches to the classification of streptococci,enterococci and lactococci: a review. Syst. Appl. Microbiol. 10:1–19.

Sneath, P.H.A. and Johnson R. 1972. The influence on numericaltaxonomic similarities of errors in microbiological tests. J. Gen.Microbiol. 73: 377–392.

Stiles M.E. and Holzapfel W.H. 1997. Lactic acid bacteria of foodsand their current taxonomy. Int. J. Food Microbiol. 36: 1–29.

Suzzi G., Caruso M., Gardini F., Lombardi A., Vannini L., Guer-zoni M.E., Andrighetto C. and Lanorte M.T. 2000. A survey ofthe enterococci isolated from an artisanal italian goat’s cheese�semicotto caprino�. J. Appl. Microbiol. 89: 267–274.

Svec P., Devriese, L.A., Sedlacek I., Baele M., Vancanneyt M.,Haesebrouck F., Swings J. and Doskar J. 2001. Enterococcushaemoperoxidus sp. nov. and Enterococcus moraviensis sp. nov.,isolated from water. Int. J. Syst. Evol. Microbiol. 51: 1567–1574.

Teixeira L.M., Facklam R.R., Steigerwalt A.G., Pigott N.E., Mer-quior V.L.C. and Brenner D.J. 1995. Correlation between phe-notypic characteristics and DNA relatedness within Enterococ-cus faecium strains. J. Clin. Microbiol. 33: 1520–1523.

Teixeira L.M., Merquior V.L.C., Vianni M.C.E., Carvalho M.G.S.,Fracalanzza S.E.L., Steigerwalt A.G., Brenner D.J. and FacklamR.R. 1996. Phenotypic and Genotypic Characterization of atyp-ical Lactococcus garviae strains isolated from water buffaloswith subclinical mastitis and confirmation of L. garviae as a se-nior subjective synonym of Enterococcus seriolicida. Int. J. Syst.Bacteriol. 46: 664–668.

Trovatelli L.D. and Schiesser A. 1987. Identification and signifi-cance of enterococci in hard cheese made from raw cow andsheep milk. Milchwissenschaft 42: 717–719.

crobiol. UK 144: 1171–1179.

and O,Donnell A.G.

profile in Serra ewes,cheese during ripening. J. Appl. Bacteriol.

251

Tyrrell G.J. Bethune R.N., Willey B. and Low D.E. 1997. Speciesidentification of enterococci via intergenic ribosomal PCR. J.Clin. Microbiol. 35: 1054–1060.

Vancanneyt M., Snauwaert C., Cleenwerck I., Baele M., De-scheemaeker P., Goossens H., Pot B., Vandamme P., Swings J.,Haesebrouck F. and Devriese L.A. 2001. Enterococcus villorumsp. nov., an enteroadherent bacterium associated with diarrhoeain piglets. Int. J. Sys. Evol. Microbiol. 51: 393–400.

Vandamme P., Pot B., Gillis M., de Vos P., Kersters K. and SwingsJ. 1996. Polyphasic taxonomy, a consensus approach to bacterialsystematics. Microbiol. Rev. 60: 407–438.

Wessels D., Jooste P.J. and Mustert J.F. 1990. Technologically im-portant characteristics of Enterococcus isolates from milk anddairy products. Int. J. Food Microbiol. 10: 349–352.

252