Background: Members of the genus Bifidobacterium are anaerobic Gram-positive Actinobacteria, which are natural inhabitants of human and animal gastrointestinal tract. Certain bifidobacteria are frequently used as food additives and probiotic pharmaceuticals, because of their various health-promoting properties. Due to the enormous demand on probiotic bacteria, manufacture of high-quality products containing living microorganisms requires rapid and accurate identification of specific bacteria. Additionally, isolation of new industrial bacteria from various environments may lead to multiple isolations of the same strain, therefore, it is important to apply rapid, low-cost and effective procedures differentiating bifidobacteria at the intra-species level. The identification of new isolates using microbiological and biochemical methods is difficult, but the accurate characterization of isolated strains may be achieved using a polyphasic approach that includes classical phenotypic methods and molecular procedures. However, some of these procedures are time-consuming and cumbersome, particularly when a large group of new isolates is typed, while some other approaches may have too low discriminatory power to distinguish closely related isolates obtained from similar sources.

Results: This work presents the evaluation of the discriminatory power of four molecular methods (ARDRA, RAPD-PCR, rep-PCR and SDS-PAGE fingerprinting) that are extensively used for fast differentiation of bifidobacteria up to the strain level. Our experiments included 17 reference strains and showed that in comparison to ARDRA, genotypic fingerprinting procedures (RAPD and rep-PCR) seemed to be less reproducible, however, they allowed to differentiate the tested microorganisms even at the intra-species level. In general, RAPD and rep-PCR have similar discriminatory power, though, in some instances more than one oligonucleotide needs to be used in random amplified polymorphic DNA analysis. Moreover, the results also demonstrated a high discriminatory power of SDS-PAGE fingerprinting of whole-cell proteins. On the other hand, the protein profiles obtained were rather complex, and therefore, difficult to analyze.

Conclusions: Among the tested procedures, rep-PCR proved to be the most effective and reliable method allowing rapid differentiation of Bifidobacterium strains. Additionally, the use of the BOXA1R primer in the differentiation of 21 Bifidobacterium strains, newly isolated from infant feces, demonstrated slightly better discriminatory power in comparison to PCR reactions with the (GTG)5 oligonucleotide. Thus, BOX-PCR turned out to be the most appropriate and convenient molecular technique in differentiating Bifidobacterium strains at all taxonomic levels.