Pulsed Field-Gel-Electrophoresis PFGE

SCAMDM Meeting7 June 2012 Tel Aviv ILAgenda item 4 D03

For SCAMDM comments and approval for publication

Version including agreed PG comments

Guidance Document

Identification of probiotics at strain level

Knut J. Heller1, W. Bockelmann1, E. Brockmann2

1Max Rubner-Institut(Federeal Research Institute for Nutrition and Food), Department of Microbiology and Biotechnology, Hermann-Weigmann-Str. 1, D-24103 Kiel, Germany;

2Chr. Hansen, Bøge Allé 10-12, DK-2970 Hørsholm, Denmark

*Corresponding author: Email <>

Probiotic properties are those of strains, not of species or even higher taxa (Heller, 2003). This has been stressed by WHO/FAO in 2002 (N.N., 2006) by the following statement: “Strain typing has to be performed with a reproducible genetic method or using aunique phenotypic trait. Pulsed Field Gel Electrophoresis (PFGE) is the gold standard.Randomly Amplified Polymorphic DNA (RAPD) can also be used, but is lessreproducible. Determination of the presence of extrachromosomal genetic elements, such as plasmids can contribute to strain typing and characterization.” That probiotic properties are those of strains and not of higher taxa is reflected by the fact that species names of micro-organisms with established probiotic properties are extended by additional identifiers: e.g. a combination of letters and numbers or the name of the person who isolated it. Examples are: Lactobacillus rhamnosus GG, Bifidobacterium animalissubsp. lactisBB-12, Lactobacillus casei Shirota, etc.

In addition to strain-specific methods needed for identification of probiotic strains, correct assignment to species or sub-species remains indispensable. The latter is important for evaluating the genetic background probiotic micro-organisms are imbedded in: whether they belong to groups with established positive impacts on human health or whether they belong to groups which include spoilage or even pathogenic micro-organisms (Pot et al., 1997). In this communication we focus on strain identification.However, it has to be clear that this identification is just based on a typing method, which only allows to demonstrate identity or non-identity of a strain in question with that of a given probiotic strainin the context/frame of the typing scheme of the applied method.

Typing methods have been compared in several reviews (Vandamme et al., 1996; Salvelkoul et al., 1999; Domigk et al., 2003) and they have been discussed with respect to their taxonomic resolution powers. A consensus of the latter is represented in Fig. 1. In a rather recent review, Li et al. (2009) have focussed on genomic typing methods and have extended the overview by including methods based solely on DNA extracted from the environment.

The methods with the broadest range of application are “DNA hybridization probes” and “DNA sequencing”. They basically allow taxonomic discrimination from the highest down to the lowest level, the strain level. However, while many “DNA hybridization probes” have been developed and published for higher taxonomic levels (targeting rDNA), only few are available for strain levels (targeting genes specific for the strains). The reasons for this are rather simple: DNA probes for strain level identification have very limited applications but require intensive and laborious testing for their development. DNA sequencing, on the other hand, has the potential to become the standard method for all problems of taxonomic resolution, even when considering resolution below species level. At present, sequencing of ribosomal RNA genes (16S rDNA) represents the standard for species identification, since standard amplification primers binding to conserved regions and standard sequencing primers can be applied. For strain identification, however, 16S rDNA sequencing cannot be successfully applied, due to limited variability of this gene. One has to acknowledge that very powerful sequencing techniques have been developed during recent years, which allow sequencing of large numbers of nucleotides at rather low cost. This has led to the development of the “Multi Locus Sequencing Typing/ Multi Locus Sequence Analysis (MLST / MLSA)” technique (Maiden et al., 1998), in which several gene loci are sequenced and the sequence information generated is catenated and used for differentiation and typing purposes. During recent years this technique has gained momentum and several MLST databases have been developed and are accessible online.The most recently developed“next generation” sequencing techniques, the “454 sequencing” (Margulies et al., 2005) and the“Solexa/Illumina” and “SOLiD” sequencing,bear the potential of sequencing and analysing entire genomes of microorganisms within one or a few days. Thus, whole genome sequencing may eventually develop into the standard method of strain typing at least for industrially important strains. However, before whole genome sequencing can become a standard typing method, a clear concept has to be developed for defining the extent of allowable sequence deviation within one and the same strain.

For very obvious reasons, “phenotype determination by classical methods” is not an appropriate method for taxonomic resolution at strain level: it simply requires too much time and experimental efforts.

Many other methods listed in Fig. 1, e.g. rRNA sequencing, cell wall structure, % G+C, are not suitable for discriminating at strain level, since discrimination is limited to higher taxa. These methods will not be discussed any further.

Of the methods suitable for strain identification, those based on serology target the surfaces of micro-organisms. Serology has been very successfully applied for differentiation of pathogenic bacteria. However, so far it has not been shown whether serology is capable of identifying a probiotic strain among its non-probiotic relatives within the same species.

SDS-PAGE (sodium dodecylsulphate polyacrylamide gel electrophoresis) of total cellular protein separated basically according to size (molecular mass) has been successfully shown to be applicable in lactic acid bacteria for differentiation at strain level (Pot et al., 1997). However, due to the necessity of growing and harvesting the cells under strictly defined conditions for generation of absolutely reproducible protein patterns, this method is powerful when used as “in-house” method. However, for identification of strains in different laboratories the method may not be robust enough. Similar considerations regard DNA-amplification techniques, where amplification conditions are applied, which are based on primers and primer binding conditions allowing to target sequences not exactly matching the primer sequences. This is especially true for RAPD (random amplification of polymorphic DNA), where primers with arbitrary sequences are applied, and to a somewhat lesser extent for rep-PCR (Repetitive Element PCR) and other similar methods like ERIC-PCR (Enterobacterial Repetitive Intergenic Consensus PCR)(de Bruijn, 1992), Box-PCR (BOX-A1R-based repetitive extragenic palindromic PCR) (Louws et al., 1994) etc. In these cases small changes in the amplification procedures result in significant changes in the electrophoretic patterns generated. As a consequence, inter-laboratory identification at strain level becomes very difficult if not impossible. Somewhat reduced robustness is also the major argument against AFLP (amplified fragment length polymorphism) (Vos et al., 1995), which otherwise is a method of very high discriminating power and which allows high throughput of samples to be tested.In this method, genomic DNA is hydrolysed by two restriction enzymes of which at least one is a frequent cutter enzyme. To the ends generated, two different adapters are ligated, which differentially recognize the two different ends produced by the two enzymes. For subsequent PCR, primers basically corresponding to the adapter sequences but with extended selectivity are applied.

In contrast to AFLP, ribotyping (Snipes et al., 1989) and PFGE (Snell and Wilkins, 1986) do not involve any step of amplification by PCR. For ribotyping, chromosomal DNA is hydrolysed by restriction enzymes and separated by agarose gel electrophoresis. In the following Southern blot, a labelled (radioactive, fluorescence, digoxigenin etc.) probe specific for rRNA genes identifies those DNA fragments, which carry regions of those rRNA genes. Due to the larges heterogeneity of DNA regions flanking the rRNA genes, the banding patterns of hybridizing fragments show very high intra-species variation. In PFGE, chromosomal DNA is hydrolyzed by means of a rare cutting restriction enzyme. The very large DNA fragments obtained are separated by an electrophoresis technique, in which the electric field frequently changes between two different directions, which leads to clear separation of DNA fragments several hundred thousand basepairs in length. This leads to typical banding patterns , which are used for fingerprinting. Thus, both ribotyping and PFGE are very robust techniques, since exactly defined, reproducible electrophoresis patterns are generated by complete hydrolysis of DNA with restriction endonucleases: ambiguities resulting from varying amplification conditions are thus excluded. However, ambiguities with this technique may arise from poor hydrolysis of bacterial cell walls for liberation of DNA within the plug moulds, incomplete restriction hydrolysis, and variations in the electrophoresis conditions. The latter, however, can be compensated for by always using defined marker-DNA as control in the electrophoretic separation. Actually, for epidemiologic studies, PFGE has been demonstrated to be applicable as a standardized effective method for identification of strains in foodborne disease outbreaksby concerted actions of laboratories joined in the so-called Pulse-Net (Boxrud et al., 2010).

Some of the typing techniques have been compared with each other with respect to their discriminating powers. In a survey involving 35 isolates of Campylobacter jejuni, Männinen et al. (2001) discriminated 8 different strains by ribotyping, 10 by PFGE and 10 by AFLP. Tynkkynen et al. (1999) analysed 19 Lactobacillus rhamnosus isolates and were able to discriminate 7 different strains by RAPD, 10 by ribotyping, and 12 by PFGE. Finally, Mättö et al. (2004) analysed 18 Bifidobacterium longum and 10 Bifidobacterium adolescentis isolates. They were able to discriminate for B. longum 7 different strains by RAPD, 13 by ribotyping, and 14 by PFGE (only 14 of the 18 isolates were analysed by PFGE). For B. adolescentis, RAPD yielded 6, and ribotyping as well as PFGE 9 different strains each. Thus, PFGE is a robust method with the best discriminative power at strain level of all methods – except AFLP (Vogel et al., 2004) - described. It is more labour-intensive than e.g. AFLP, but less labour-intensive than ribotyping (a rapid PFGE method for analysis of bifidobacteria has been described some years ago) (Briczinski and Roberts, 2006)). For these reasons, PFGE is called the “gold-standard” of strain identification. This has been acknowledged by FAO/WHO as already indicated in the first paragraph of this document (N.N., 2006).

In the Annex I to this Document, the most suitable methods for strain identification are listed together with short descriptions of the methods. The literature cited in Annex I is included in the list of references at the end of the body of this text.

The availability of techniques for fingerprinting below species level allows for strain differentiation. However, when applying such techniques, one always has to bear in mind that fingerprinting methods are just typing methods: they analyse one characteristic trait, which then is used for attributing the organisms to groups of identical or very similar organisms (strains) or non-identical ones. Except for methods based on DNA sequencing, typing methods are not suited for phylogenetic considerations.The results of the typing methodsfor different strains within one species have to be interpreted with care.

It has to be clear from the beginning of an experiment, whether any observed deviation from a given pattern is supposed to result in the denomination of a new strain or not. Since strains do not form a taxonomic unit but form a group of members, which have been assigned to the same strain on the basis of an arbitrary definition, the definition may be such that it either does not allow for any deviations in a banding pattern or that it allows for small deviations of up to 10% in a banding pattern of one and the same strain. The latterdefinition is often applied, when tracing back microorganisms in outbreaks to the original source of infection (Chiou et al., 2001).

When just one method is applied for assigning microorganisms to one group, one has to be aware, that the group identified by this method may in fact be somewhat heterogeneous. Recently, it has been described that Lactobacillus fermentum strains with identical PFGE patterns differed in up to four characteristic traits (ARDRA patterngenerated with a particular restriction enzyme, growth at a particular temperature, ability to metabolize two different sugars) (Njeru et al., 2010). This certainly raises the fundamental question of whether typing methods are really useful in attributing microorganisms to groups of microorganisms sharing one or few important functional traits, which are not tested by the typing method. The Joint FAO/WHO Working Group Report on Drafting Guidelines for the Evaluation of Probiotics in Food(N.N., 2006) states that “Strain identity is important to link a strain to a specific health effect as well as to enable accurate surveillance and epidemiological studies”, thereby indicating that strain identification serves different purposes. High resolving typing methods like PFGE will be adequate for the assessment of recovery in studies investigating probiotic functionality as well as safety, where the strain identity of the probiotic applied is typically not only secured by typing methods but also through a documented origin traceable to a reference material. Due to the described possible heterogeneity of strains belonging to one typing group, typing methods can however not be fully sufficient to generally make the link to probiotic functionality. The same argument will also limit their applicability for accurate surveillance as well as epidemiological studies where there is no documented link to the probiotic reference material. The conclusion for probiotic strains can only be that typing methods like PFGE are important as long as specific tests for those genes – or even for the activity of those genes - involved in making a strain probiotic are not available, due to lack of information of the genes involved. As soon as such information is available, typing methods together with methods testing the activities of relevant genes will show that certain functional traits are present within the correct strain, i.e. in a microorganism with the correct genetic background.

References

Bennasar, A., Mulet, M., Lalucat, J., García-Valdés, E. 2010. PseudoMLSA: a database for multigenic sequence analysis of Pseudomonas species. BMC Microbiol. 10:118.

Briczinski, E.P., Roberts, R.F. 2006. Technical Note: A rapid pulsed-field gel electrophoresis method for analysis of bifidobacteria. J. Dairy Sci. 89:2424-2427

Boxrud, D., Monson, T., Stiles, T., et al. 2010. The role, challenges, and support of Pulse-Net laboratories in detecting foodborne disease outbreaks. Publ.Health Rep. 125:57-62

Busconi, M., Reggi, S., Fogher, C. 2008.Evaluation of biodiversity of lactic acid bacteria microbiota in the calf intestinal tracts. Antonie Van Leeuwenhoek 94(2):145-155.

de Bruijn, F.J. 1998. Use of repetitive (Repetitive Extragenic Palindromic and Enterobacterial Repetitive Intergeneric Consensus) sequences and the polymerase chain reaction to fingerprint the genomes of Rhizobium melilotiisolates and other soil bacteria. Appl. Environ. Microbiol. 58:2180-2187

Chiou, C.S., Hsu, W.B., Wei, H.L., Chen, J.H. 2001. Molecular epidemiology of a Shigella flexneri outbreak in a mountainous township in Taiwan, Republic of China. J. Clin. Microbiol. 39:1048-1056

De Vuyst, L., Vancanneyt, M. 2007. Biodiversity and identification of sourdough lactic acid bacteria. Food Microbiol. 24(2):120-127.

Domigk, K.J., Mayer, H.K., Kneifel, W. 2003. Methods used for the isolation, enumeration, characterisation and identification of Enterococcus spp. Int. J. Food Microbiol. 88:165-188

Engel, G., Roesch, N., Heller, K.J. 2003. Typing by pulsed-field gel electrophoresis (PFGE)of bifidobacteria from fermented dairy products. Kieler Milchw. Forsch. 55:225–232 (in German)

Hanage, W.P., Fraser, C., Spratt, B.G. 2006. Sequences, sequence clusters and bacterial species. Philos. Trans. R. Soc. Lond. B Biol. Sci. 361(1475):1917-1927.

Hänninen, M.-L., Perko-Mäkelä, P., Rautelin, H., Duim, B., Wagenaar, J. A. (2001). Genomic Relatedness within Five Common Finnish Campylobacter jejuni Pulsed-Field Gel Electrophoresis Genotypes Studied by Amplified Fragment Length Polymorphism Analysis, Ribotyping, and Serotyping. Appl. Environ. Microbiol. 67:1581-1586.

Heller, K.J. 2003. Inclusion of Probiotics in Beverages: Can it Lead to Improved Health? pp. 247-257. In T. Wilson and N. J. Temple (Eds.), Beverages in Nutrition and Health. Humana Press Inc., Totowa, New Jersey.

Hudson,M.E. 2008. Sequencing breakthroughs for genomic ecology and evolutionary biology. Mol. Ecol. Resour. 8(1):3-17.

Klein, G., Pack, A., Bonaparte, C., et al. 1998.Taxonomy and physiology of probiotic lactic acid bacteria.Int. J. Food Microbiol.41:103-125

Li, W., Raoult, D. Fournier, P.-E. 2009. Bacterial strain typing in the genomic era. FEMS Microbiol. Rev. 33:892–916

Lortal, S., Rouault, A., Guezenec, S., Gautier, M.1997. Lactobacillus helveticus: strain typing and genome size estimation by pulsed field gel electrophoresis. Curr. Microbiol. 34:180-185

Louws,F.J., Fulbright,D.W., Taylor-Stephens, C., de Bruijn, F.J. 1994. Specific Genomic Fingerprints of Phytopathogenic Xanthomonas and Pseudomonas Pathovars and Strains Generated with Repetitive Sequences and PCR. Appl. Environ. Microbiol. 60:2286-2295

Maiden, M.C.J., Bygraves, J.A., Feil, E. et al. 1998. Multilocus sequence typing: A portable approach to the identification of clones within populations of pathogenic microorganisms Proc. Natl. Acad. Sci. U.S.A. 95: 3140–3145

Margulies, M. Egholm, M. Altman, W.E. et al., 2005. Genome sequencing in microfabricated high-density picolitre reactors. Nature 437:376-380

Mättö, J., Malinen, E., Suihko, M.-L., Alander, M., Palva, A., Saarela, M. 2004. Genetic heterogeneity and functional properties of intestinal bifidobacteria. J. Appl. Microbiol. 97:459-462

Möller, C. 2002. Application of new selection strategies for yoghurt starter cultures suitable for yielding yoghurt with mild taste. Ph.D. thesis, University at Kiel, Germany (in German)

N.N. 2006. Joint FAO/WHO Working Group Report on Drafting Guidelines for the Evaluation of Probiotics in Food. FAO Food and Nutrition Paper 85

Njeru, P.N., Roesch, N., Ghadimi, D.,et al., 2010. Identification and characterization of lactobacilli isolated from 1 Kimere, a spontaneously 2 fermented pearl millet dough from Mbeere, Kenya (East Africa). Beneficial Microbes (in press)

Pennacchia, C., Vaughan, E.E., Villani, F. 2006. Potential probiotic Lactobacillus strains from fermented sausages: Further investigations on their probiotic properties. Meat Sci.73:90-101

Picozzi, C, Bonacina G, Vigentini I, et al. 2010. Genetic diversity in Italian Lactobacillus sanfranciscensis strains assessed by multilocus sequence typing and pulsed-field gel electrophoresis analyses. Microbiology-SGM156: 2035-2045