Molecular techniques for the identification and detection of microorganisms relevant for the food industry

The research described in this thesis concerns the development and application in food microbiology of molecular identification and detection techniques based on 16S rRNA sequences. The technologies developed were applied to study the microbial ecology of two groups of bacteria, namely starter cultures and sporeforming spoilage bacteria, that are of importance to the food industry and in particular the dairy industry.

In this Chapter the results are summarized and discussed in relation to recent developments. Moreover, the impact of the results obtained on quality control systems and risk assessment of the use of genetically modified microorganisms in food will be discussed.

Role of ribosomal RNA sequences in the identification and detection of microorganisms

In Chapter 1 several identification and detection techniques based upon rRNA sequences are described. They are based upon the structural and sequence differences within the ribosomal operon, like ribotyping, analysis of the spacer regions, or restriction legnth polymorphism of amplified parts. Some of the techniques strictly involve specific sequence differences within the rRNA genes, which consist of alternating conserved and more variable regions. The rRNA genes are uniquely suited for the reliable identification of microorganisms and the rational probe design for the specific detection of microorganisms. In particular 16S rRNA sequences have become to play an important role in microbial taxonomy. Many microorganisms have been reclassified or renamed based on new insights in their phylogenetic relationships based on 16S rRNA sequence homology. Within the constraints of the current data-bases it is now possible to identify new isolates quickly and reliably by sequencing their 16S rRNA. The use of rRNA sequencing in the identification and detection of microorganisms has indicated that many organisms are still unknown and not characterized, comprising both culturable and nonculturable microorganisms.

16S rRNA-derived DNA probes can either be based on conserved regions containing groupand/or genus-specific sequences or more variable regions containing speciesand/or subspecies- specific sequences. These probes can be applied in hybridizations to DNA or RNA fixed to a membrane or present in in situ fixed cells. In order to increase the specificity and the sensitivity of a detection method, such probes can also be applied in specific DNA amplifications. Direct amplification of specific 16S rRNA sequences from, for instance, a food matrix like cheese, is the most straightforward and sensitive detection method. Unfortunately, most of these food matrices, and also other environmental samples like soil and faecal material, contain components that may inhibit the PCR reaction. This necessitates complex DNA extraction protocols or the use of enrichment cultures in order to obtain nucleic acid extracts that can be applied in PCR amplifications successfully.

It is possible to identify an isolate reliably at the species or subspecies level based on ribosomal sequences (Chapters 2 and 5). However, it is possible to distinguish isolates at the strain level with other techniques, like the frequently applied Random Amplified Polymorphic DNA (RAPD) technique, in which strain specific patterns are formed by the amplification with arbitrarily chosen primers of approximately ten nucleotides . This technique is often used in addition to 16S rRNA-based identification methods, as has been shown recently by the discrimination of different serotypes of Listeria monocytogenes. The combination of both approaches can be very useful, in particular in epidemiologic studies. Moreover, RAPD patterns can be used for the development of strain-specific DNA probes as well. This can also be a very successful approach for obtaining species-specific DNA probes for microorganisms, such as Campylobacter jejuni, Campylobacter coli and Campylobacter lari, that are difficult to design on the basis of ribosomal sequences. It has been described that some genera, like Rhizobium, include species that contain an almost identical 16S rRNA but share almost no DNA-DNA homology. In order to discriminate between these species, other techniques have to be applied, of which RAPD or other chromosomal fingerprinting methods are used most frequently.

Development and application of molecular, 16S rRNA based, identification methods for Lactococcus species

In the food industry in general, but specifically in the dairy industry, lactic acid bacteria are used as starter cultures to initiate specific fermentations. Both for basic research on lactic acid bacteria and for their application in industrial food fermentations, reliable and simple methods for the identification of such bacteria are required. Because little difference exists in phenotypic properties of especially the mesophilic lactic acid bacteria, reliable identification and detection techniques have been developed based on specific sequences in variable regions of 16S rRNA (Chapter 1). Species-specific DNA probes, based on the first variable region (VI) of the 16S rRNA , were designed for Lactococcus garvieae, L. plantarum and L. raffinolactis and subspecies-specific DNA probes for L. lactis subsp. lactis and L . lactis subsp. cremoris. The third subspecies L. lactis subsp. hordnieae was not distinguishable from L lactis subsp. lactis based on 16S rRNA sequences because these differed in only one base pair. In addition, speciesspecific probes for Leuconostoc were developed based on the third variable region (V3) of the 16S rRNA .

There is a growing need for new production strains for the innovation of dairy products. These can be obtained either by genetic modification of known production strains or by isolation of new strains from natural ecological niches. Both for the application of genetically modified starter strains and to allow for an efficient search for strains from natural ecosystems, it is important to know if and where a Lactococcus strain survives outside the dairy environment. Lactose-utilizing Lactococcus isolates from environmental samples taken on cattle farms and in the waste flow of a cheese production plant were identified up to the species level, using amplified variable regions (VI) of the 16S rRNA in combination with species-specific DNA probes (Chapter 3). These isolates were further characterized by using specific PCR amplification of sequences related to cit P , prt P and nis A, coding for citrate permease, protease and prenisin, respectively. It was possible to isolate Lactococcus spp. from various environments, indicating that lactococci can survive outside the dairy plant and that some are able to persist in soil, effluent water, on vegetation and on cattle. During the characterization of the environmental lactococcal isolates discrepancies were observed between the 16S rRNA based identification of L . lactis strains and identification based on their phenotypical properties. Similar findings were obtained with isolates from spontaneous milk fermentations, collected from several european countries . The classical differentiation between L. lactis subsp. lactis and L. lactis subsp. cremoris is based on phenotypical differences. Llactis subsp. cremoris strains are characterized by their inability to hydrolyse arginine, to metabolize a number of sugars, and to grow at 37 °C and in the presence 4% NaCI . Based on SDS-PAGE of whole-cell proteins both phenotypes are distinguishable and form two seperate clusters. However, within the group of environmental isolates, identified as L.lactis subsp. lactis on the basis of their phenotype, 16S rRNA sequences belonging to both L. lactis subsp. lactis and L lactis subsp. cremoris were encountered. Detailed characterization of a large collection of lactococcal isolates has led to the conclusion that within the species L. lactis two ribotypes are present, each having a specific 16S rRNA sequence. However, for each ribotype different phenotypes can be found. The ribotype of the strain NCDO 712, the parental strain of MG1363, which is frequently used in genetic studies of L. lactis , shows that it belongs to L. lactis subsp. cremoris , confirming the conclusion based on the mapping of its chromosome. It is quite remarkable that this strain shows the L. lactis subsp. cremoris ribotype, while phenotypically it resembles L. lactis subsp. lactis .

The phenotype described for L. lactis subsp. cremoris is only observed with isolates that have been obtained from industrial starter cultures or traditional fermented milk products like villi. Such strains may belong to either ribotype. This may suggest that the continuous culturing in milk has resulted in a differentiation of phenotypic properties between "starter" strains and strains originating from environmental sources. Detailed characterisation of L. lactis isolates from natural environments indicated that they differ from strains commonly present in starter cultures. This does not only suggest that starter isolates do not survive outside the dairy environment, but also that L. lactis strains present within the natural population have potential to be applied in product innovation and differentiation. This applies in particular to isolates that were obtained from spontaneous milk fermentations, which are still used in the southern part of Europe for the production of artisanel cheeses. An initial inventory of the biodiversity of such isolates illustrates the large diversity of properties, like aroma formation, acidification and bacteriocin production, which can be exploited to obtain differentiation in fermented milk products.

Molecular detection techniques were not only used for the identification of environmental isolates but also for the monitoring of the survival of L lactis in the human gastrointestinal tract. For the potential application of L lactis as probiotic, as genetically modified starter culture, or as live vaccine, it is important to determine whether these bacteria survive in the gastrointestinal tract after consumption by humans. In Chapter 4, a human feeding study is described in which the fate was monitored of L. lactis strain TC165.5, which was genetically marked by insertion of the naturally occurring sucrose-nisin conjugative transposon Tn 5276 and spontaneous chromosomal resistance to rifampicin and streptomycin. A method was developed for the efficient extraction of microbial DNA from human faeces. The passage of strain TC165.5 through the intestinal tract was monitored by selective plating and specific detection of the nis A gene by PCR amplification. The study showed that up to 0.1-1% of L. lactis cells, consumed in a dairy product, may survive passage of the gastrointestinal tract, provided that they pass within 3 days after consumption. The partial survival of lactococci provides a positive prospective for the use of Lactococcus strains as probiotic or in the development of live vaccines.

Biosafety assessment of the use of genetically modified Lactococcus spp. infermented food products

The studies presented in Chapters 3 and 4 were part of an assessment of the biosafety aspects of the use of genetically modified starter cultures, specifically addressing general ecological parameters like survival, dissemination and transfer of genetic information. In order to quantify the survival of starter lactococci, careful monitoring of the waste flow of a cheese production plant was performed. This indicated the absence of typical industrial strains (Chapter 3) suggesting that most starter organisms are not able to persist in non-dairy environments, although natural niches are available, In addition, the numbers of lactococci found in the non-dairy environments were considerably lower than those that are daily released into the environment via the industrial production of fermented milk products. It was therefore concluded that there is no environmental release of viable starter bacteria resulting from the waste flow of the production process of fermented foods.

Another avenue for the release of lactic acid bacteria into the environment is by means of the consumption of fermented milk products by humans. The results of the human feeding trial (Chapter 4) showed that only a small fraction of viable cells of the marked L.lactis strain survived passage through the human gastrointestinal tract.

It is well known that L. lactis strains possess efficient systems for transfer of genes via plasmids, transposons and phages. A number of studies have been published on gene transfer in lactococci under natural conditions, including large-scale fermentation, cheese manufacturing, and passage through the gastro-intestinal tract of mice. The results showed that the transfer rates decreased rapidly under natural conditions where cell-to-cell contact and growth are limited. On the other hand, under conditions favourable for cell-to-cell contact, such as on agar plates and intestinal mucosal surfaces, the transfer frequencies are relatively high, up to 10 ^-4.

The transfer of genetic information from one strain to another per se is not to be regarded as a potential hazard, since it was demonstrated that lactococci are already capable of transferring genetic elements. In some cases, however, the properties encoded by new genetic traits may be potentially hazardous in combination with specific strains or in specific ecosystems. If the encoded properties are already present in the ecosystem, no specific new hazard may be expected, since in the past it has not resulted in hazardous biological consequences. By relating the potential hazards of the application of genetically modified lactic acid bacteria to the regular hazards associated with the, consumption of fermented dairy products, the potential hazard may be normalized. So both for legislation purposes and for the acceptance by the consumers it is important to identify and normalize the biosafety aspects by relating them to current practice. In this way the potential hazards are more comprehensible and acceptable.

Development and application of molecular detection and identification methods for Clostridium tyrobutyricum

Major spoilage problems in the food industry are caused by sporeforming bacteria, belonging to the genera Clostridium and Bacillus, which survive heat-treatments that are applied to prolong the shelf-life of the food product. Butyric acid fermentation in cheese (late blowing), caused by the germination of clostridial spores which survive the heat treatment of cheese milk, is still causing considerable loss in the cheese producing industry. For the production of semi- hard cheeses, like Gouda cheese, it is very important to limit the number of spores in the cheese milk of bacteria capable of causing late blowing. Although Clostridium tyrobutyricum is regarded as the causative agent of butyric acid fermentation, also spores of other clostridia, such as C. beijerinckii and C.sporogenes are frequently isolated from late-blown cheeses. The current detection method for C.tyrobutyricum is not specific, since also other clostridial species are able to form butyric acid and hydrogen. In Chapter 5 and 6 the development of specific detection methods is described for the Clostridium spp. most frequently encountered in dairy environments. Based on specific sequences in the V2 and V6 region of the 16S rRNA, species- specific DNA probes were developed for C. tyrobutyricum, C. acetobutylicum, C. beijerinckii, C. butyricum and C.sporogenes (Chapter 5).

In Chapter 6 the causative relationship between C.tyrobutyricum and late blowing in cheese is demonstrated. Cheese experiments were performed to provoke this defect by using spores from different strains of several dairy-related clostridia. To overcome problems associated with isolation of Clostridium spp., specific clostridial sequences were directly detected in DNA isolated from cheese by a two-step specific PCR amplification. Only specific sequences of C.tyrobutyricum were detected in both the experimental and in commercially obtained cheeses showing signs of late blowing. This clearly identified C.tyrobutyricum as the causative agent of late blowing in cheese.

In order to prevent late blowing in cheese, the number of spores has to be limited to less than 1 per 10 ml. of cheese milk. To improve the currently used methods for the routine detection of clostridial spores in milk, a specific and very sensitive method has to be developed. Although the current methods are very aspecific and often give false positive results mainly due to the presence of C.beijerinckii spores, it is possible to detect up to 1 spore per 100 nil by MPN-methods. Recently, Herman et al. have shown that it should be possible to detect one C.tyrobutyricum spore per 100 nil raw milk by concentrating the spores by centrifugation, followed by DNA extraction and a nested PCR amplification. However, all reports on DNA extraction from spores describe experiments with artificially obtained spore- suspensions. It is not clear if these methods can also be applied on natural spore populations. Since it is known that during the production of artificial spore suspensions residues of cell material, including DNA, can adhere to the outside of the spores, proper control experiments should be included. Such control experiments have not been described, so far making it impossible to evaluate the efficiency of the extraction methods described.

Relevance of molecular identification and detection techniques based on ribosomalsequences for the food microbiology in general and in particular for the dairyindustry

The availability of molecular methods for the detection and identification of microorganisms has several advantages. In many cases the identity of both desired and unwanted microorganisms present in food products is of major importance. Enormous efforts have been invested in the reproducible identification of isolates based upon phenotypic properties such as carbohydrate fermentation, formation of specific metabolites, enzyme activities, and lipid composition. Unfortunately, most of these methods appeared to be irreproducible or not sufficiently discriminative for proper identification. The 16S rRNA-based methods are far more suited for the quick and reliable identification of isolates. However, the 16S rRNA sequencing share one particular limitation with many other identification methods, like those of API or Biolog, namely that it is only possible to obtain a correct identification within the limitations of the database. As many other databases, the RPD-database is not complete and contains only the 16S rRNA sequences of approximately 3000 bacteria. Most of the widely distributed species are present, but the 16S rRNA sequences of many bacteria frequently encountered in foods have not yet been determined or deposited. The absence of most Brevibacterium spp. and many other relatives of the Arthrobacter group, including Corynebacterium spp., is very problematic since they are important in the dairy industry as part of the surface microflora of surface-ripened cheeses and as spoilage organisms of pasteurized milk and cream. It appears that the biodiversity within this group is enormous, complicating clear systematic descriptions for these bacteria.

Molecular identification methods based on ribosomal sequences have revolutionized the classification and systematics for many bacteria, but also the ability to specifically detect microorganisms has made a great impact on food microbiology. This is particularly the case with respect to the detection and specific quantification of those bacteria that are unculturable, like Candidatus arthromitus , or those for which no suitable selective media are available, like Bifidobacterium spp.. In addition, molecular techniques will allow to obtain more detailed information on critical points in the production processes of foods. Such information is essential for safeguarding the product quality. For instance, by the use of strain- specific RAPD patterns to monitor the population dynamics of mixed-strain starter bacteria during the fermentation process, better insights could be obtained in which factors are important for the quality of the resulting fermented food product.

Although the potential sensitivity of molecular methods should make them suitable for the reliable detection of pathogens and spoilage organisms in food products and raw materials, there are some major practical problems that still have to be solved. One of the largest problems is the presence of PCR-inhibiting components in several food products, like cheese and meat. This inhibition can be circumvented by applying enrichment procedures before the actual detection with a specific PCR reaction. Even if these procedures result in a quicker and more reliable detection of the target organism there are some important limitations to such procedures. It is known that a significant problem with the detection of some pathogens, like Salmonella, is the unreliability of a successful pre-enrichment in buffered peptone. This failure to obtain growth in the pre- enrichment step can be caused by competing flora or the physiological state of the target organisms. Still, the detection kits that are currently available, based on DNADNA hybridization, can be successfully used as verification methods for Listeria and Salmonella after a successful pre-enrichment.

Besides the practical problems, the implementation of molecular methods is also complicated by legislation, codes of practice, and regulations for processing and product control. Each adapted detection protocol for bacteria like Salmonella, Listeria and other important food pathogens requires extensive tests and validations before it is accepted by regulatory authorities. In addition, the routine application of these methods is still not possible because of the lack of automation for routine analysis of food products. Automated equipment developed for the medical market, is so far not suitable for this purpose and only a limited number of PCR detection kits have been specifically designed for application in the food industry.

In this thesis the usefulness has been demonstrated of molecular methods for the identification and detection of microorganisms relevant for the food industry, These methods have increased the insight in the microbiology of spoilage by sporeforming bacteria and in the ecological aspects of the use of lactic acid bacteria as starter cultures. Results obtained with molecular techniques, like 16S rRNA sequencing, quantification of specific DNA's, and in situ hybridisation, are expected to create new insights in the dynamics of complex microbial populations such as exist in starters, ripening flora of cheeses and other fermented foods.