Molecular microbial analysis of Bifidobacterium isolates from different environments by the species-specific amplified ribosomal DNA restriction analysis (ARDRA)

Authors


*Corresponding author. Tel.: +41 (21) 7858901; Fax: +41 (21) 7858925; E-mail: ralf.zink@rdls.nestle.com

Abstract

One hundred and six isolates of the genus Bifidobacterium, isolated from different environments (mainly gastrointestinal), were identified and classified taxonomically to species level by amplified ribosomal DNA restriction analysis. Two restriction endonucleases (Sau3AI and BamHI) were chosen for aligning the 16S rRNA sequences of 16 bifidobacterial species retrieved from various databases, to obtain species-specific restriction patterns. A rapid and accurate identification scheme was obtained by comparing the resulting 16S rDNA digestion profiles of 16 Bifidobacterium type-strains and 90 strains of various origins. All of the investigated strains were previously confirmed at the species level as belonging to the genus Bifidobacterium by fluorescence in-situ hybridisation and by polymerase chain reaction amplification with genus- and species-specific primers. The present work demonstrates that species-specific detection of Bifidobacterium adolescentis, Bifidobacterium animalis, Bifidobacterium bifidum, Bifidobacterium breve, Bifidobacterium catenulatum, Bifidobacterium coryneforme, Bifidobacterium cuniculi, Bifidobacterium dentium, Bifidobacterium infantis, Bifidobacterium lactis, Bifidobacterium longum, Bifidobacterium suis, Bifidobacterium magnum, Bifidobacterium pseudolongum, Bifidobacterium pseudocatenulatum and Bifidobacterium pullorum present in different micro-ecological environments (e.g. gastrointestinal tract) can be accomplished in a reliable, rapid and accurate manner, circumventing the recognised deficiencies of traditional identification techniques.

1. Introduction

Members of the genus Bifidobacterium are Gram-positive, pleomorphic and strictly anaerobic bacteria, and major constituents of the human intestinal microflora as well as of other warm-blooded animals and even honeybees [1]. Bifidobacteria are numerous in the human colon (up to 109–1010 cfu g−1 faeces) and are therefore part of the most prevalent bacteria detectable in the faeces of human subjects consuming a ‘western’ diet [2,3]. Bifidobacterium longum, Bifidobacterium breve, Bifidobacterium bifidum, Bifidobacterium infantis, Bifidobacterium lactis and Bifidobacterium animalis are used in the production of fermented dairy products [4,5] and seem to be involved in the maintenance of the intestinal microflora balance and health [6]. The influence of the normal microflora on the host was well demonstrated by comparing germfree and conventional animal hosts [7]. Different authors [8–13] have reported that bifidobacteria, as an essential part of the normal intestinal microflora, produce antigenic compounds which are accessible to the immune defence mechanisms of the human body.

Historically, species-specific identification methods for the genus Bifidobacterium have been based on carbohydrate fermentation profiles, cell-wall analysis or on their key enzyme fructose-6-phosphate phosphoketolase involved in hexose degradation (fructose-6-phospate shunt) [14]. In addition to these methods for speciation, electrophoretic-profiling studies were carried out on transaldolases, 6-phosphogluconate dehydrogenase and various isoenzymes that are well characterised for each Bifidobacterium species [15,16]. Other biochemical analysis demonstrated that it was possible to use the electrophoretic patterns of β-galactosidases for differentiation of Bifidobacterium species, in particular to distinguish between dairy and non-dairy related bifidobacteria [17]. This tool seems to allow a reliable identification of B. animalis in dairy products, but is quite tedious because of protein extraction and subsequent gel analysis. The identification of bifidobacteria based on phenotypic characteristics does not always provide clear results and is sometimes unreliable since bifidobacterial cells can change their morphology depending on the respective media, growth and culture conditions [1].

In recent years several molecular tools have been proposed for bifidobacterial identification [18–20]. Various polymerase chain reaction (PCR) methods have been developed using species-specific primers designed on widespread genes (e.g. 16S rRNA or recA). Sequence analysis of the recA gene in B. animalis, Bifidobacterium adolescentis, B. bifidum, B. breve, B. infantis and B. longum demonstrated differences which have been proposed for the identification of these species [21]. Differentiation among Bifidobacterium species has also been performed using pulsed-field gel electrophoresis (PFGE) [22] or randomly amplified polymorphic DNA (RAPD) with presumed species-specific patterns [23,24]. The application of PFGE to the study of Bifidobacterium communities also permits the determination of the specific origin of individual strains and analysis of different molecular types, e.g. B. longum. In practice, the application of RAPD for species-level identification is often unreliable and type-strains must be included as positive controls in the same RAPD-PCR cycles [23,25].

The comparison of 16S rRNA sequences has attracted attention as a very reliable method for the classification and identification of bacterial species [26]. Since rRNAs are extremely conserved and are present in large copy-numbers within individual bacterial cells, their utilisation as a molecular target has increased in recent years.

In Bifidobacterium species a direct PCR amplification with 16S rRNA gene-targeted, species-specific primers has also been used to investigate the distribution of bifidobacteria in the human intestinal microflora [18,19]. The use of species-specific primers revealed that Bifidobacterium catenulatum and Bifidobacterium pseudocatenulatum seem to be the most common bifidobacteria in the human intestinal microflora [18]. Traditional methods based on carbohydrate fermentation patterns [15,27] were unable to differentiate the B. catenulatum group from that of B. adolescentis. It was also demonstrated that B. breve is usually present in the adult intestinal microflora and is one of the typical and dominant Bifidobacterium species in infants [28,29].

The aim of our study was to focus on the amplified ribosomal DNA restriction analysis (ARDRA) method to develop a rapid, reproducible and easy-to-handle molecular tool for the identification of all Bifidobacterium species isolated from various environments. We have investigated 106 strains from 16 different species of the genus Bifidobacterium: B. adolescentis, B. animalis, B. bifidum, B. breve, B. catenulatum, Bifidobacterium coryneforme, Bifidobacterium cuniculi, Bifidobacterium dentium, B. infantis, B. lactis, B. longum, Bifidobacterium suis, Bifidobacterium magnum, Bifidobacterium pseudolongum, B. pseudocatenulatum and Bifidobacterium pullorum, isolated from the intestinal tract of mainly adults and infants, and the dairy environment.

2. Materials and methods

2.1 Bacterial strains and culture conditions

The names and origins of all bacterial strains used in this study are summarised in Tables 1 and 2. The type-strains were obtained from: American Type Culture Collection (ATCC; Rockville, MD, USA); Deutsche Sammlung von Mikroorganismen (DSM; Göttingen, Germany) and the Bacteria Collection Universiteit Gent (Gent, Belgium). All strains were cultured anaerobically in MRS (Difco, Detroit, MI, USA) broth or agar supplemented with L-cysteine–hydrochloride (0.5 g l−1) at 37°C for 16 h.

Table 1.  Bacterial strains used originating from various international micro-organism collections
SpeciesStrainaOrigin
  1. a Source of cultures: ATCC, American Type Culture Collection; DSM, Deutsche Sammlung von Mikroorganismen; LMG, Bacteria Collection Universiteit Gent.

B. adolescentisATCC 15703Tintestine of adult
B. adolescentisATCC 15704intestine of adult
B. adolescentisATCC 15706intestine of adult
B. animalisATCC 25527Trat faeces
B. animalisATCC 27674rabbit faeces
B. animalisATCC 27673sewage
B. bifidumATCC 29521Tinfant faeces
B. bifidumATCC 15696intestine of adult
B. breveATCC 15700Tintestine of infant
B. breveATCC 15701intestine of infant
B. catenulatumATCC 27539Tintestine of adult
B. coryneformeDSM 20216Thindgut of honeybee
B. cuniculiATCC 27916Tfaeces of rabbit
B. dentiumATCC 27534Tdental caries
B. dentinumATCC 27680human dental caries
B. infantisATCC 15697Tintestine of infant
B. infantisATCC 15702intestine of infant
B. lactisDSM 10140Tdairy product (yoghurt)
B. longumLMG 13197Tintestine of adult
B. longumATCC 15708intestine of infant
B. magnumATCC 27540Trabbit faeces
B. magnumATCC 27681rabbit faeces
B. pseudocatenulatumDSM 20438Tfaeces of infant
B. pseudolongumDSM 20099Tswine faeces
B. pullorumDSM 20433Tfaeces of chicken
B. pullorumATCC 49617faeces of chicken
B. suisATCC 27533Tswine faeces
B. suisATCC 27532swine faeces
Table 2.  Identification of selected isolated strains of Bifidobacterium by the ARDRA approach, species-specific primers and by carbohydrate fermentation profiles (ID-32A)
ARDRA identificationStrain NCC CodeID-32A identificationIdentification with specific primersaOrigin
  1. nd: not determined; pseudo&catenulatum: belong to the group of B. pseudocatenulatum and B. catenulatum species; animalis, adolescentis, bifidum, breve, cuniculi, infantis, longum, magnum, pseudocatenulatum, pullorum: allocation of these species by using the different identification tools mentioned above. No bifidum, no breve, no infantis, no longum: no achievable PCR-amplicon using the corresponding species-specific PCR-primers [18,19].

  2. a Identification using species-specific PCR-primers as described earlier [18,19].

animalis221animalisndfaeces of chicken
animalis239animalisNdfaeces of chicken
animalis257animalisndhuman faeces
animalis273animalisndrat faeces
animalis311bifidumndinfant faeces
animalis330infantisno infantisyoghurt
animalis363longumno longuminfant faeces
animalis383longumno longumyoghurt
animalis387infantisno infantisfaeces of infant
animalis391longumno longumyoghurt
animalis402longumno longumyoghurt
animalis414longumno longumyoghurt
animalis424longumno longumcheese starter
adolescentis251adolescentisadolescentisintestine of adult
bifidum246bifidumbifiduminfant faeces
bifidum268bifidumbifidumcalf faeces
bifidum373bifidumbifidumhuman faeces
bifidum390bifidumbifidumintestine of infant
bifidum409bifidumbifidumintestine of adult
bifidum420bifidumbifidumintestine of adult
bifidum453bifidumbifidumhuman faeces
bifidum464bifidumbifidumyoghurt
bifidum479bifidumbifidumOTC tablets
breve272brevebreveinfant faeces
breve281brevebreveinfant faeces
breve292brevebreveinfant faeces
breve306brevebreveinfant faeces
breve308longumbreveinfant faeces
breve313brevebreveinfant faeces
breve321brevebreveinfant faeces
breve322bifidumbreveinfant faeces
breve336brevebreveinfant faeces
breve348brevebreveinfant faeces
breve354brevebreveinfant faeces
breve364brevebreveinfant faeces
breve370brevebreveinfant faeces
breve378brevebreveinfant faeces
breve385brevebreveinfant faeces
breve393brevebreveinfant faeces
breve399brevebreveinfant faeces
breve407brevebreveinfant faeces
breve412brevebreveinfant faeces
breve419brevebreveinfant faeces
breve439brevebreveintestine of infant
breve452brevebreveinfant faeces
breve458brevebreveinfant faeces
breve466brevebreveinfant faeces
breve485brevebreveyoghurt
cuniculi297cuniculindfaeces of rabbit
infantis283infantisInfantisinfant faeces
infantis290bifiduminfantiscalf faeces
infantis294infantisinfantisinfant faeces
infantis318infantisinfantisinfant faeces
infantis356infantisinfantisintestine of infant
infantis365infantisinfantisintestine of infant
infantis376infantisinfantisintestine of infant
infantis428infantisinfantisinfant faeces
longum284longumlonguminfant faeces
longum293longumlonguminfant faeces
longum302longumlonguminfant faeces
longum305infantislonguminfant faeces
longum317longumlonguminfant faeces
longum324longumlonguminfant faeces
longum332bifidumlongumyoghurt
longum335longumlonguminfant faeces
longum344longumlonguminfant faeces
longum352bifidumlongumyoghurt
longum417infantislongumyoghurt
longum444longumlongumintestine of adult
longum450longumlongumintestine of infant
longum461longumlongumcalf faeces
longum469longumlonguminfant faeces
longum481longumlonguminfant faeces
longum490longumlonguminfant faeces
longum501longumlonguminfant faeces
longum510longumlonguminfant faeces
longum521longumlongumadult faeces
longum531longumlongumadult faeces
longum552longumlonguminfant faeces
longum572longumlonguminfant faeces
longum585longumlongumadult faeces
magnum296magnumndrabbit faeces
pseudocatenulatum274pseudo&catenulatumpseudo&catenulatumadult faeces
pseudocatenulatum277bifidumpseudo&catenulatumcalf faeces
pseudocatenulatum298breveno breveinfant faeces
pseudocatenulatum291pseudo&catenulatumpseudo&catenulatumadult faeces
pseudocatenulatum312pseudo&catenulatumpseudo&catenulatuminfant faeces
pseudocatenulatum371longumpseudo&catenulatumadult faeces
pullorum289pullorumndfaeces of chicken
pullorum310pullorumndfaeces of chicken

2.2 DNA isolation

Two ml of each 16-h stationary-phase culture were collected by centrifugation at 12 000×g (10 min, 4°C), the pellet washed twice with 2 ml of water, and 1 g of glass beads (Sigma, St. Louis, MO, USA) added to the bacterial cell suspension. Mechanical lysis of the cells was achieved with the Mini-Beadbeater™ (Biospec, USA) for 3 min at maximum speed at 4°C. After mechanical rupture the resulting suspension was centrifuged at 12 000×g (2 min, 4°C) and 10 μl of the supernatant was directly added to the prepared PCR tube.

2.3 ARDRA-PCR

PCR was used to amplify the 16S rRNA gene of all Bifidobacterium isolates investigated. DNA fragments of approximately 1.5 kb (corresponding to the size of the 16S rRNA gene) were amplified using the primers P0 (5′-GAAGAGTTTGATCCTGGCTCAG-3′) and P6 (5′-CTACGGCTACCTTGTTACGA-3′). Each PCR mixture (50 μl) contained a reaction mix of 20 mM Tris–HCl, 50 mM KCl, 200 μM of each deoxynucleoside triphosphate, 0.5 μM of each primer, 1.5 mM MgCl2 and 1.5 U of the polymerase SuperTaq (Enzyme Technologies, UK). Each PCR cycling profile consisted of an initial denaturation time of 3 min at 95°C followed by an amplification for 30 cycles of denaturation (30 s at 94°C), annealing (30 s at 55°C) and extension steps (2 min at 72°C). The PCR was completed with an elongation period (7 min at 72°C). The resulting amplicons were separated by a 0.8% (w/v) agarose gel electrophoresis at 7 V cm−1 followed by ethidium bromide staining. Before restriction digestion of the PCR products, all amplicons were purified with the QIAquick PCR Purification Kit (Qiagen, Valencia, USA). In order to achieve complete digestion, all restriction digestions were carried out for 3 h at 37°C in 10-μl volumes of incubation buffer (Boehringer-Mannheim, Germany) containing 2 U of the respective restriction enzyme (Boehringer-Mannheim, Germany) and about 1.2 μg of the purified PCR product. The resulting digestion products were visualised under UV-light (260 nm) after agarose gel electrophoresis (3.0% (w/v); NuSieve 3:1 agarose, FMC Bioproducts, USA) at 7 V cm−1 followed by ethidium bromide staining. All restriction analysis was performed with the Vector nt (Informax) tool (on-line tool for restriction mapping of nucleotide sequences: http://www.medkem.gu.se/cutter).

2.4 Identification of Bifidobacterium isolates using genus-specific, species-specific bifidobacteria PCR primers and biochemical tests

In order to confirm our results for species identification, additional species-specific primers for Bifidobacterium spp. and their respective PCR conditions [18,19] were applied. As a basic confirmation that all isolates were members of the Bifidobacterium genus, we applied additional PCRs by using a specific bifidobacterial genus-specific primer pair (Lm26, Lm3). The sequences of these primers and the PCR conditions were retrieved from a previous report [30]. All achievable PCR products were analysed by electrophoresis on 0.8% (w/v) agarose gels and visualised by ethidium bromide staining. All bifidobacterial strains were additionally analysed by carbohydrate fermentation profiles based on ID-32A system (BioMérieux, France).

2.5 Fluorescence in-situ hybridisation (FISH) specific for the Bifidobacterium genus

To confirm that all investigated strains and species were in fact members of the genus Bifidobacterium, all faecal isolates were hybridised with a fluorescently labelled, genus-specific, 16S rRNA-targeted probe using a FISH Kit (RiboTechnologies Microscreen, The Netherlands). Ten ml of a 16-h stationary-phase bacterial culture of each isolate was collected by centrifugation at 12 000×g (10 min, 4°C) and bacterial cells were treated in the subsequent sample preparation according to the supplier's instructions. Analysis was also performed using 0.5 g of faecal sample by following the supplier's instructions.

3. Results

3.1 Specificity 16S DNA restriction fragment length polymorphism (RFLP) patterns

All available 16S rDNA sequences of the Bifidobacterium species were retrieved from different databases (GenBank or EMBL). Theoretical restriction profiles with different restriction enzymes were obtained from the 16S rDNA sequences by using the Webcutter analysis protocol. The restriction enzyme Sau3AI was found to give the clearest and most reliable distinction in theoretical ARDRA patterns, in order to differentiate the majority of Bifidobacterium type-strains. Only B. adolescentis/coryneforme, B. animalis/pseudolongum and B. suis/longum could not be discriminated with a Sau3AI digestion of the 16S rRNA gene. Fig. 1 depicts the electrophoretic patterns obtained for 16 Bifidobacterium type-strains (Table 1) after Sau3AI digestion of the 16S rRNA gene. We could obtain clearly distinguishable RFLP patterns of the 16S rRNA gene in repeated experiments, confirming the theoretically expected restriction profiles. All strains of B. adolescentis and B. coryneforme resulted in identical Sau3AI patterns of their 16S rDNA, in agreement with their theoretical restriction profiles, necessitating an additional digestion with a second restriction enzyme, e.g. BamHI (Fig. 2). Similarly BamHI was also used to differentiate B. animalis and B. pseudolongum (Fig. 2). To date, B. suis cannot be separated from B. longum using ARDRA patterns.

Figure 1.

Agarose gel electrophoresis of digested 16S rDNA with Sau3A1 (negative image). Lane 1, B. adolescentis ATCC 15703T; lane 2, B. coryneforme DSM 20216T; lane 3, B. catenulatum ATCC 27539T; lane 4, B. pseudocatenulatum DSM 20438T; lane 5, B. infantis ATCC 15697T; lane 6, B. longum LMG 13197T; lane 7, B. suis ATCC 27533T; lane 8, B. breve ATCC 15700T; lane 9, B. bifidum ATCC 29521T; lane 10, B. magnum ATCC 27540T; lane 11, B. dentium ATCC 27534T; lane 12, B. cuniculi ATCC 27916T; lane 13, B. animalis ATCC 25527T; lane 14, B. pseudolongum DSM 20099T; lane 15, B. pullorum DSM 20433T; lane 16, B. lactis DSM 10140T; lane MS, 50-bp DNA molecular marker (Sigma, USA); lane M, 50-bp DNA molecular marker (Gibco BRL, USA); lane m, DNA molecular mass marker VI (Boehringer-Mannheim, Germany).

Figure 2.

Restriction profiles of 16S rDNA sequences obtained using the restriction enzyme BamH1 (negative image of an agarose gel electrophoresis): lane 1, B. adolescentis ATCC 15703T; lane 2, B. coryneforme DSM 20216T; lane 3, B. pseudolongum DSM 20099T; lane 4, B. animalis ATCC 25527T; lane 5, B. lactis DSM 10140T; lane M, 50-bp DNA ladder molecular marker (Gibco BRL, USA); lane m, DNA molecular mass marker VI (Boehringer-Mannheim, Germany).

3.2 Allocation of all the isolates within the genus Bifidobacterium

All faecal and intestinal isolates were subjected to FISH with a single probe specific for the genus Bifidobacterium. After hybridisation of bacterial cells, any occurring fluorescence inside the cells could be clearly distinguished from the dark background (data not shown). All strains originally isolated from yoghurt, cheeses, probiotic tablets and from mucus were also assayed in PCR amplification trials using the genus-specific primers Lm26 and Lm3. Four of the strains (NCC223, NCC426, NCC432 and NCC319) investigated were identified as not belonging to the genus Bifidobacterium by their general absence of PCR products using these genus-specific primers. All remaining isolates demonstrated a distinct PCR amplicon of about 1.35 kb, in agreement with literature reports [31].

3.3 Identification of Bifidobacterium strains with species-specific primers and biochemical assay

The ARDRA identification method was compared with results achieved with classical identification tools (e.g. carbohydrate fermentation profiles) and with species-specific PCR primers. We tested six pairs of recently published bifidobacteria species-specific primers, all targeting the 16S rDNA sequences [18,19]. These primers cover all of the bifidobacterial species that have been identified in human intestinal tracts. However, the identification of B. catenulatum and B. pseudocatenulatum by this method was not reliable. Furthermore, identification of some bifidobacterial isolates by carbohydrate fermentation profiles was rather doubtful or unacceptable and a few strains belonging to B. pseudocatenulatum, B. infantis, B. longum and B. animalis were initially misidentified as B. bifidum (Table 2).

3.4 Amplification and restriction analysis of 16S rRNA gene

The study was extended to include isolates from various sources. Their amplified 16S rRNA gene products were obtained after rapid extraction of chromosomal DNA. RFLP patterns were generated following digestion of the 16S rRNA gene of all of the Bifidobacterium isolates with Sau3AI and, if necessary, with BamHI. In order to identify all 106 isolates, their RFLP patterns were compared with those obtained for the type-strains. All bifidobacterial strains had been previously assayed with RAPD-PCR, which produced an individual profile for each isolate. The heterogeneity among the RAPD-PCR patterns clearly demonstrated that all bifidobacterial strains investigated by ARDRA identification were different (data not shown).

3.5 Analysis of the Bifidobacterium species among isolates

Among all isolates investigated only those belonging to the genus Bifidobacterium are listed in Table 2. All strains isolated from faecal environments were identified as being members of the species B. pseudocatenulatum, B. pullorum, B. magnum, B. cuniculi, B. bifidum,B. longum, B. breve, and B. animalis. All strains identified as B. breve (with the exception of NCC485) and B. infantis (with the exception of NCC290) were isolated from human infant faeces, while B. longum and B. pseudocatenulatum isolates were detected in adult and from infant faeces (Tables 1 and 3).

Table 3.  Distribution of all bifidobacterial isolates from different isolation environments
OriginBifidobacterium speciesStrain(s)
Intestine of adultB. bifidumNCC409, NCC420
 B. longumNCC444
 B. adolescentisNCC251
Intestine of infantB. bifidumNCC390
 B. breveNCC439
 B. infantisNCC356, NCC365, NCC376
 B. longumNCC450
Human faecesB. animalisNCC257
 B. bifidumNCC373, NCC453
Adult faecesB. pseudocatenulatumNCC274, NCC371, NCC291
 B. longumNCC521, NCC531, NCC585
Infant faecesB. breveNCC322, NCC272, NCC281, NCC292, NCC306, NCC308, NCC313, NCC321, NCC336, NCC348, NCC354, NCC364, NCC370, NCC378, NCC385, NCC393, NCC399, NCC407, NCC412, NCC419, NCC452, NCC458, NCC466,
 B. infantisNCC283, NCC294, NCC318, NCC428
 B. bifidumNCC246
 B. longumNCC284, NCC293, NCC305, NCC302, NCC317, NCC324, NCC335, NCC344, NCC469, NCC481, NCC490, NCC501, NCC510, NCC552, NCC572
 B. animalisNCC387, NCC363, NCC311
 B. pseudocatenulatumNCC312, NCC298
Faeces of rabbitB. cuniculiNCC297
 B. magnumNCC296
Calf faecesB. longumNCC461
 B. pseudocatenulatumNCC277
 B. bifidumNCC268
 B. infantisNCC290
Faeces of chickenB. animalisNCC221, NCC239
 B. pullorumNCC310, NCC 289
Rat faecesB. animalisNCC273
YoghurtB. longumNCC332, NCC352, NCC417
 B. breveNCC485
 B. bifidumNCC464
 B. animalisNCC330, NCC383, NCC391, NCC402, NCC414
Cheese starterB. animalisNCC424
Probiotic tabletsB. bifidumNCC479

B. magnum, B. cuniculi and B. pullorum were only detected in animal faeces (chicken, rabbit or rat). Other isolates belonging to the species B. animalis, B. longum and B. bifidum came from different environments: faecal/intestine (human and animal) and dairy products (yoghurts and cheese starter), as summarised in Tables 1 and 3. Only four of the investigated strains (NCC223, NCC426, NCC432 and NCC319) and previously classified by using carbohydrate fermentation profiles as B. adolescentis and B. breve, do not belong to the genus Bifidobacterium and their origin (mucus) might be considered as an atypical environment for bifidobacteria species already described [1].

4. Discussion

In modern bacterial taxonomy, ribosomal gene sequences are considered to have the potential to provide a powerful approach to the investigation of phylogenetic relationships. Therefore, they are often used for the design of species-specific primers for a rapid identification of lactic acid bacteria [20] and bifidobacteria [18,19,32,33]. In eubacterial DNA these rRNA loci include 16S, 23S and 5S rRNA genes. While the 16S rRNA gene is a highly appropriate molecular target for the analysis of species-specific relationships [34], the intergenetic 16S–23S spacer region allows excellent analysis at the subspecies and strain level [35]. The overall phylogenetic information content of the 23S rRNA molecule is higher than that of the 16S rRNA molecule, but the number of currently available complete 23S rDNA bifidobacterial sequences is rather poor in comparison to those of the 16S rRNA. Only a few methods for species-level identification are so far described for the genus Bifidobacterium[17–22,26,33,36]. Other typing or identification methods are rather time-consuming and do not always assure a reliable characterisation (e.g. plasmid profiles analyses; DNA–DNA hybridisations).

Although significant progress has been achieved with the application of ARDRA methods, some factors still remain to be solved. For example, B. longum and B. suis are difficult to distinguish, although plasmid-profiling for strains belonging to these species appears promising [36]. Nevertheless, we agree with Matsuki [18] that B. suis could be taxonomically combined with B. longum, taking their high DNA–DNA homology values (75–78%; [37]) and the level of 16S rDNA similarities of more than 99%[38] into consideration. In contrast to other reports [18], we were able to distinguish B. catenulatum and B. pseudocatenulatum, despite their close DNA–DNA homology, murein type and their high 16S rDNA sequence homologies. Consequently, the ARDRA system described here can be considered as a very reliable and reproducible molecular identification tool for at least the 16 Bifidobacterium species investigated in this study (Table 2).

Applying the species-specific ARDRA method, we observed that the majority of human gut isolates belonged to the species B. longum, B. bifidum or B. pseudocatenulatum. This is in contrast to Matsuki [18], who reported mainly B. adolescentis and B. catenulatum isolates as being the predominant members of the human adult intestinal microflora [18]. Furthermore these workers reported that their specific primers were unable to distinguish between B. longum and B. infantis[19]. Although we were able to identify B. adolescentis as part of the intestinal microflora of human adults, this species represented only less than 10% of all identified isolates. Insufficient data exist to attribute differences in the natural distribution of B. longum, B. bifidum, B. pseudocatenulatum, B. catenulatum and B. adolescentis to differing diets, various cultural or genetical backgrounds or simply inadequate sample size [9,13].

From both a scientific and technological point-of-view, the micro-ecology of bifidobacteria might help us to understand if and how a potential dietary supplementation of, for example, predominant species/strains of bifidobacteria, or the administration of specific functional carbohydrates as, for example, prebiotics, will help to favour and/or establish specific target bifidobacteria. In order to achieve such a basic understanding, we have to investigate and identify, in a reliable and taxonomically generate manner, the composition and stability of the human micro-flora in the gastrointestinal tract [7,13]. This identification and characterisation could provide a more rational approach with which to select appropriate strains of bifidobacteria as probiotic strains and could improve our understanding of the underlying traits such as survival and/or colonisation [21].

The ARDRA method described here proposes one solution for the inadequate identification methods for determining Bifidobacterium species and could be of help to evaluate the differences between simple and complex, dynamic or static, bifidobacterial microflora in the human gastrointestinal tract [6]. Furthermore Bifidobacterium species composition in the gastrointestinal tract seems to provide a good marker for the evaluation of human intestinal microflora stability [39]. In man, the ARDRA technique would enable investigators to monitor changes in the intestinal bifidobacterial microflora with respect to age, ethno-cultural and/or genetic background (shift in population balances) or diet (more carbohydrate- versus protein-based). This could allow specific insights into the effect of consumption habits and other environmental changes for humans and animal health. Thus, species identification with ARDRA should enable monitoring of Bifidobacterium populations belonging to the dominant intestinal flora over a defined period of time at the micro-ecological level. This would reduce the need for labour-intensive and time-consuming identification techniques (e.g. single-colony isolation from selective media; multiple physiological and biochemical testing), which remain frequently unreliable, slow and expensive, and are not accurate enough for the analysis of intestinal bifidobacteria [19].

Acknowledgements

We thank D. Carey Walker of the Nestlè Research Center, Lausanne, Switzerland for his critical review of the manuscript and for stimulating discussions. We would also like to acknowledge the excellent technical assistance of Valérie Meylan.

Ancillary