DNA Methods in Food Safety: Molecular Typing of Foodborne and Waterborne Bacterial Pathogens

By Omar A. Oyarzabal

Molecular typing of foodborne pathogens has become an indispensable tool in epidemiological studies. Thanks to these techniques, we now have a better understanding of the distribution and appearance of bacterial foodborne diseases and have a deeper knowledge of the type of food products associated with the major foodborne pathogens. Within the molecular techniques, DNA-based techniques have prospered for more than 40 years and have been incorporated in the first surveillance systems to monitor bacterial foodborne pathogens in the United States and other countries. However, DNA techniques vary widely and many microbiology laboratory personnel working with food and/or water face the dilemma of which method to incorporate.

DNA Methods in Food Safety: Molecular Typing of Foodborne and Waterborne Bacterial Pathogens succinctly reviews more than 25 years of data on a variety of DNA typing techniques, summarizing the different mathematical models for analysis and interpretation of results, and detailing their efficacy in typing different foodborne and waterborne bacterial pathogens, such as Campylobacter, Clostridium perfringens, Listeria, Salmonella, among others. Section I describes the different DNA techniques used in the typing of bacterial foodborne pathogens, whilst Section II deals with the application of these techniques to type the most important bacterial foodborne pathogens. In Section II the emphasis is placed on the pathogen, and each chapter describes some of the most appropriate techniques for typing each bacterial pathogen.

The techniques presented in this book are the most significant in the study of the molecular epidemiology of bacterial foodborne pathogens to date. It therefore provides a unique reference for students and professionals in the field of microbiology, food and water safety and epidemiology and molecular epidemiology.

• Language: English
• Publisher: Wiley
• Release date: Aug 21, 2014
• ISBN: 9781118278659

Book preview

Preface

Molecular typing of foodborne pathogens has become an indispensable tool in epidemiological studies. Thanks to these techniques, we can have a better understanding of the distribution and appearance of bacterial foodborne diseases and have a deeper knowledge of the type of food products associated with the major foodborne pathogens. Within the molecular techniques, DNA-based techniques have prospered for more than 40 years and have been incorporated in the first surveillance systems to monitor bacterial foodborne pathogens in the United States and other countries. However, DNA techniques vary widely, from techniques based on amplification of selected segments of the DNA to the latest whole genome sequencing analysis. Because of the wide array of available techniques and the different results they generate, we have compiled in Section I the different DNA techniques in use for the typing of bacterial foodborne pathogens. This section covers the following techniques: (i) pulsed-field gel electrophoresis, the main typing technique at the molecular subtyping network for foodborne bacterial disease surveillance (PulseNet) by the Centers for Disease Control and Prevention (CDC); (ii) multilocus sequence typing, a very powerful technique to study bacterial population structures and changes; and (iii) high-throughput sequencing techniques that are poised to be the predominant techniques in the near future. In Section I, we have also included chapters on the analysis of results obtained with band-migration techniques, the databases and internet applications available as repository of data produced by these techniques, and the application of these molecular techniques to outbreak detection and public heath surveillance.

Section II deals with the application of techniques to type the most important bacterial foodborne pathogens. Here the emphasis is placed on the pathogen, and each chapter describes some of the most appropriate techniques for typing each bacterial pathogen. As techniques progress and as we have better access to automated and robust techniques to study proteins, it is expected that DNA techniques will be used in association with other protein-based techniques or as first screening techniques. Until then, the techniques presented in this book are the most powerful techniques to study the molecular epidemiology of bacterial foodborne pathogens.

Omar A. Oyarzabal

Seattle, WA, USA

Sophia Kathariou

Raleigh, NC, USA

Section I

Typing Method, Analysis, and Applications

1

Polymerase Chain Reaction-Based Subtyping Methods

Yi Chen and Insook Son

Center for Food Safety and Applied Nutrition, Food and Drug Administration, College Park, MD, USA

Polymerase chain reaction (PCR)-based molecular subtyping methods have been developed and applied to study the population genetics and molecular epidemiology of foodborne pathogens for more than two decades. These methods are based on PCR reaction and subsequent analysis of the banding pattern in gel electrophoretic. Some methods involve restriction digestion and ligation. The principles and performance (discriminatory power, epidemiological concordance, ease of use, reproducibility, typeability, etc.) of some typical PCR-based molecular subtyping methods are discussed in the following text.

Randomly amplified polymorphic DNA

Randomly amplified polymorphic DNA (RAPD) technique is a PCR technique widely used for subtyping various bacterial pathogens. It was first described by Williams et al. (1990) and unlike conventional PCR arbitrary PCR primers are used and the target PCR products of RAPD are unknown. The primers are usually 9–10 bp long and are arbitrarily chosen by the researcher or can be randomly generated by computers. The arbitrary primer can simultaneously anneal to multiple sites in the whole genome under low stringent conditions. When the two primers anneal within a few kilobases of each other in the proper direction, a fragment is amplified. These products can then be separated by gel electrophoresis and the banding patterns of different isolates compared. The genomic locations where these primers anneal are usually specific to a genotype and thus RAPD patterns are used for subtyping purposes. The annealing of arbitrary primers can be affected by only a few nucleotide differences. Because of its marked sensitivity, RAPD PCR has proven useful for differentiating both Gram-positive and Gram-negative bacteria, especially for closely related species or epidemiologically related strains (Hadrys, Balick, and Schierwater, 1992; Power, 1996; Milch, 1998) (Figure 1.1).

c1-fig-0001

Figure 1.1 Randomly amplified polymorphic DNA analysis using arbitrary primers. Arbitrarily designed short primers (8–12 nucleotides) anneal to a large template of genomic DNA. When two primers anneal in the opposite direction to two genomic locations that are reasonably distant from each other, a fragment is amplified. These randomly amplified fragments are then analyzed by gel electrophoresis, resulting in a different pattern of amplified DNA fragments on the gel. To enhance priming with short primers, many primers are designed with a GC content between 10 and 70% and low annealing temperatures are used.

RAPD has been widely used to subtype various foodborne pathogens such as Listeria monocytogenes, Salmonella, and Escherichia coli O157:H7. Nilsson et al. (1998) developed and optimized a RAPD subtyping method for Bacillus cereus that showed excellent reproducibility. Mazurier et al. (1992) investigated the epidemiologic relevance of RAPD using well-characterized outbreak isolates of L. monocytogenes and found that RAPD correctly classified 92 out of 102 isolates into corresponding epidemic groups. Aguado, Vitas, and García-Jaloń (2001) performed RAPD and serotyping analysis to study the cross-contamination of L. monocytogenes in processed food products. Using RAPD, the authors illustrated that the strains isolated from different meat type and brand on the same date had identical subtypings, suggesting cross-contamination. The authors also found that RAPD and serotyping results were concordant, but RAPD demonstrated higher discriminatory power. The authors finally concluded that RAPD was an easy method that could be used to identify cross-contamination in post-processing environment. Vogel et al. (2001) analyzed 148 L. monocytogenes strains from vacuum-packed cold-smoked salmon produced in 10 different smokehouses using RAPD with 4 different primers separately. The authors demonstrated that RAPD provided higher discriminatory power than ribotyping and serotyping for epidemiologic typing of L. monocytogenes; however, the discriminatory power of RAPD was not as good as pulsed field gel electrophoresis (PFGE) and amplified fragment length polymorphism (AFLP). The authors obtained 16 reproducible RAPD profiles and the clustering of isolates using the 4 primers was identical. They identified dominant RAPD types in products from each smokehouse but also found identical RAPD types in different smokehouses, and concluded that these were persistent strains in the smokehouse environment. This study was reported in 2001. It would be interesting to reanalyze those strains with identical RAPD types from different smokehouses using other discriminatory methods that were developed in the last decade. In an earlier study conducted in Japan (Yoshida et al., 1999), researchers analyzed 20 epidemiologically unrelated L. monocytogenes strains isolated from different animals and locations and on different dates, and identified 18 types by RAPD using 4 primers. They also analyzed seven epidemiologically related L. monocytogenes strains isolated from raw milk and a bulk tank on a dairy farm and showed that those strains had the same RAPD type. The results demonstrated that RAPD was epidemiologically concordant. O'Donoghue et al. (1995) used RAPD to study the diversity of L. monocytogenes of different serotypes and the authors reported that serogroup 1/2 of L. monocytogenes strains are genetically more diverse than serogroup 4, a finding that was confirmed by many other subtyping methods. Kim et al. (2005) studied a set of E. coli O157:H7 strains using RAPD and discovered that RAPD could not differentiate O157 strains that varied in the degree of virulence. Another study of E. coli O157:H7 (Vidovic, Germida, and Korber, 2007) demonstrated that RAPD yielded excellent discriminatory power for differentiating E. coli O157:H7 from animal sources.

Reproducibility is one of the biggest concerns of RAPD. Certain factors such as DNA quality and concentration, the type of Taq polymerase employed, and PCR reaction conditions can all affect the reproducibility of RAPD PCR. Therefore, it is critical to maintain the greatest consistency in DNA template quality, reagent selection, and experimental design for successful RAPD PCR. In addition, because the arbitrary primers are not specifically designed for certain genomic loci, the hybridization of the primers to the genome can be partial, which confound the PCR reaction. A RAPD protocol used by Nath, Maurya, and Gulati (2010) had only 40% reproducibility when subtyping Salmonella typhi strains isolated from typhoid patients between 1987 and 2006 in India. Penner et al. (1993) conducted an inter-laboratory reproducibility study of RAPD protocols with different primers and found two major variables with RAPD. One variable was that small and large polymorphic fragments were not always reproduced and therefore the size ranges of DNA fragments were different among the laboratories. The other major variable was that reproducible results were obtained with only four of the five primers using the same reaction conditions. These results highlight the importance of protocol optimization and the maintenance of consistent thermal cyclers among different laboratories when performing RAPD. Davin-Regli et al. (1995) demonstrated that variations in the concentration of template DNA could significantly affect the reproducibility of RAPD banding patterns. Bidet et al. (2000) evaluated three RAPD protocols using different single primers each for subtyping Clostridium difficile, and the reproducibility were only 88, 67, and 33% for the three primers. Due to the very low reproducibility of RAPD, the authors cautioned that the discriminatory power might be an overestimation.

Amplified fragment length polymorphism (AFLP)

AFLP is a highly discriminatory subtyping method used for molecular subtyping. With AFLP, genomic DNA is purified and digested with two restriction enzymes and then two different restriction-specific adaptors are ligated to ends of the restriction fragments (Figure 1.2). PCR primers, which are complementary to the adaptors, are designed to selectively amplify a subset of the ligated restriction fragments under stringent PCR conditions. In order to further select a subset of fragments to amplify, PCR primers are usually designed with a specific base or doublet or triplet of bases adjacent to either restriction site, and thus only the subset of genomic fragments that have matching bases adjacent to the restriction sites are amplified. PCR amplicons are then analyzed by gel electrophoresis, and gel patterns (polymorphisms between and within restriction sites) are used to assign subtypes (Savelkoul et al., 1999; Foley et al., 2004; Foley, Zhao, and Walker, 2007; Singh and Mohapatra, 2008).

c1-fig-0002

Figure 1.2 Amplified fragment length polymorphism analysis. A DNA template is first digested with two restriction enzymes, preferably a hexa-cutter and a tetra-cutter; and then the restriction fragments are ligated to the adaptors. Primers are designed to be complementary to the adapter and restriction site sequences, and their 3′ ends were added by a random nucleotide for selective amplification. Amplicons of selective amplification are visualized by gel electrophoresis.

AFLP generally yields excellent discriminatory power, which is comparable to PFGE, the current gold standard, except for a few cases described in the following paragraph. However, AFLP is more time consuming due to the extra ligation-mediated PCR procedure. The selective PCR step could generate some randomness and thus affects the reproducibility of AFLP. Internal variability due to incomplete digestion and/or ligation is also known to affect the final banding patterns.

Ripabelli, McLauchin, and Threlfall (2000) and Guerra, Bernardo, and McLauchlin (2002) developed AFLP schemes for subtyping L. monocytogenes. They found that although not discriminatory enough, AFLP results were congruent with serotyping, phage typing, and other subtyping methods, confirming the three genetic lineages of L. monocytogenes. Keto-Timonen et al. (2007) subsequently improved AFLP by careful selection of restriction enzymes. The discriminatory power of their AFLP scheme was over 0.99 when using Simpson's index of diversity and the results were congruent with PFGE. Lomonaco et al. (2011) compared 2 AFLP methods with PFGE for subtyping 103 unrelated L. monocytogenes strains isolated from different environmental and food sources in Italy. The authors found that the two AFLP methods and PFGE had similar discriminatory power. However, one AFLP method suffered from unsatisfactory typeability for certain strains from dairy products. This AFLP method uses restriction enzyme Sau3AI; therefore, it has been suggested that some L. monocytogenes strains from dairy products are not restricted with Sau3AI, which is possibly due to the methylation of cytosine at GATC. Therefore, careful selection of restriction enzyme is very critical for the typeability of AFLP. Herrera et al. (2002) reported a discrepancy in the relatedness of the Shigella flexneri strain typed by plasmid profiling, serotyping, and AFLP analysis, and no definitive conclusions were drawn about the epidemiologic concordance of AFLP with this work.

Since the invention of the original AFLP scheme, modifications of this technique have been developed. One example is the fluorescent amplified fragment length polymorphism (FAFLP). With this technique, the amplified restricted fragments are labeled with fluorescent molecules and therefore the detection of banding patterns is of higher resolution than the traditional gel electrophoretic patterns. Ross and Heuzenroeder (2005) compared FAFLP and PFGE using a set of Salmonella enterica serovar Typhimurium DT126 isolates from several foodborne outbreaks in Australia. The authors found that AFLP had slightly higher discriminatory power than PFGE. While both methods successfully clustered isolates within an outbreak, some unrelated isolates could not be differentiated from outbreak isolates by either method. In addition to FAFLP, Lan and Reeves (2007) also reported a radioactively labeled AFLP method for subtyping S. enterica.

Repetitive-sequence-based PCR

Repetitive-sequence-based PCR (rep-PCR) typing is another subtyping technique that utilizes PCR. In bacteria, there are dispersed chromosomal repetitive elements that are randomly distributed throughout the genomes (Versalovic, Koeuth, and Lupski, 1991a) (Figure 1.3). These sequences differ in size and do not encode proteins. With rep-PCR, primers that are complementary to these repeats are designed and used to amplify differently sized DNA fragments lying between the repeats. One type of repeat sequence is repetitive extragenic palindromes (REPs), which are regulatory sequences within untranslated regions of bacterial operons. REPs were initially discovered in Salmonella and E. coli (Gilson et al., 1984; Gilson et al., 1990; Sharples and Lloyd, 1990; Ridley, 1998). The family of REPs comprises short DNA segments, generally 30–40 bp, include inverted repeats, and there are around 500–1000 copies per genome (Stern et al., 1984). They can appear as a single copy or as multiple adjacent copies and occupy up to 1% of the genomes of Salmonella and E. coli. Jersek et al. (1996) employed previously developed primers and found that Listeria spp. also possess short REP elements.

c1-fig-0003

Figure 1.3 Repetitive-sequence-based PCR. Primers are designed to bind to the repetitive elements and regions between these repeats are amplified. These fragments are then analyzed by gel electrophoresis.

Another type of repetitive sequences is named as enterobacterial repetitive intergenic consensus (ERIC) sequence. ERIC sequences are widely distributed across a wide range of species and were originally described in Salmonella, E. coli, and other members of Enterobacteriaceae. The ERIC sequence is a palindrome of 127 bp that contains a conserved central inverted repeat. Variations of ERIC sequences include shorter sequences caused by internal deletions and longer sequences caused by insertions. They are mostly present in intergenic regions of the genome. The number of copies of ERIC sequences ranges from around 30 copies in E. coli to around 150 copies in Salmonella Typhimurium and over 700 copies in some other Enterobacteriaceae (Hulton, Higgins, and Sharp, 1991; Burr, Josephson, and Pepper, 1998). ERIC PCR was first described by Versalovic, Koeuth, and Lupski (1991b). Jersek et al. (1996) subsequently found ERIC sequences in Listeria spp. and found that rep-PCR showed a higher discriminative power than ERIC-PCR for subtyping closely related strains of L. monocytogenes.

Another class of repeats, BOX, was originally found in Streptococcus pneumonia (van Belkum and Hermans, 2001). BOX sequences are also intergenic regions that form stem-loop structures. They are mosaic repetitive elements that include combinations of three subunits, BOX-A (59 bp), BOX-B (45 bp), and BOX-C (50 bp). The evolutionary origin and functions of these BOX regions remain unclear, and they are not related to REP and ERIC sequences. These regions have proven useful for the differentiation of enteric species and the development of strain-specific subtyping methods (Weigel et al., 2004; Cesaris et al., 2007). Both REP and ERIC sequences contain conserved regions for primer targeting and variable regions for polymorphism detection. For example, REP primers usually target the left and right sides of a conserved palindromic sequence and are oriented in opposite directions so that the primer extends outwardly in a 3′ direction away from the palindrome. The regions between the repetitive palindromic islands were thus amplified. These regions range in size from 200 bp to 4 kbp and provide a unique chromosomal pattern for a given strain. In a rep-PCR design, multiple primers or one single oligoprimer can be used. A list of these primers is provided in Table 1.1.

Table 1.1 Primers and PCR conditions for common repetitive elements

Van Kessel et al. (2005) analyzed 61 L. monocytogenes strains from raw milk using an automated rep-PCR system. The results showed that rep-PCR clusters correlated with species and serotypes of Listeria spp. Jersek et al. (1999) developed a rep-PCR scheme targeting short repetitive extragenic palindromic (REP) elements and enterobacterial repetitive intergenic consensus (ERIC) sequences in L. monocytogenes and found that these techniques have a high discriminatory power (0.98). One advantage of rep-PCR is that it can be automated due to the simple PCR operation. However, rep-PCR suffered from low discriminatory power when subtyping L. monocytogenes strains from the same serotype, and it could not discriminate between serotypes 1/2b and 4b strains. Zunabovic et al. (2012) evaluated the potential of three rep-PCR methods (GTG₅ and REPI+II) for the typing of Listeria spp. including L. monocytogenes from a cold-smoked salmon production facility and compared rep-PCR methods with PFGE. The authors found that although rep-PCR yielded a lower discriminatory power than PFGE, it was still a useful tool for tracing contamination niches and transmission routes of Listeria spp. in the food processing environment. However, it is important to note that this study evaluated rep-PCR using a set of Listeria spp. that included six species. A close examination of their data showed that the discriminatory power of rep-PCR within L. monocytogenes was still limited. Hahm et al. (2003) suggested that rep-PCR and BOX-PCR can serve as first step screening prior to PFGE for subtyping E. coli O157. The authors also found differences between the strain relatedness identified by each method, but the subtype profiles of the E. coli O157:H7 isolates were virtually identical using rep-PCR and BOX-PCR. Nath, Maurya, and Gulati (2010) analyzed a collection of S. enterica serotype Typhi strains isolated from typhoid patients between 1987 and 2006 using ERIC PCR and concluded that ERIC-PCR was very efficient with excellent discriminatory power and reproducibility.

A modification of rep-PCR is the incorporation of fluorescently labeled primers where amplified products are visualized by a fluorescence-based DNA analyzer. Del Vecchio et al. (1995) described a fluorescence-enhanced rep-PCR for subtyping Staphylococcus aureus. This modification reduces the labor required for manual gel electrophoresis and simplifies visualization, comparison, and storage of DNA banding patterns. Because rep-PCR is very simple, and does not require extra restriction or ligation steps like other PCR-based subtyping methods, rep-PCR can be easily automated. Brusetti et al. (2008) developed a florescent-BOX-PCR for subtyping E. coli and B. cereus and the authors concluded that the increased resolution power by using florescence-labeled oligos detected up to 12 times more fragments than traditional BOX-PCR, and thus improved the discriminatory power. Healy et al. (2005) described a modification using a commercially available automated rep-PCR system. The automated system significantly improves the reproducibility of rep-PCR over the manual operations. The built-in software programs improved image analyses, but images must be captured and imported prior to analysis and subjectivity remains because optimization parameters can be modified. The automated system has obvious advantages over PFGE and MLST, both of which require skilled operators.

Despite the reports of the success of ERIC-PCR for subtyping various bacterial species, some scientists cast doubt on the performance of ERIC-PCR. One disadvantage of ERIC-PCR is that the distributions of some ERIC sequences limit the potential of ERIC-PCR as a subtyping tool. In order to understand the performance and mechanism of ERIC-PCR, Wilson and Sharp (2006) studied the distribution and evolution of ERIC sequences in several Enterobacteriaceae species. The authors found that the copy numbers and locations of ERIC sequences vary greatly among different strains within the same species, which serve as the base of discriminatory ability of ERIC-PCR. However, the authors observed that some E. coli strains do not have sufficient full length ERIC sequences and the number of amplified fragments is not sufficient to generate meaningful patterns for subtyping purposes. Another disadvantage of ERIC-PCR is that the reaction appeared to be not specific. In a detailed investigation, Gillings and Holley cautioned that ERIC-PCR would generate amplified products even from genomes that do not possess any ERIC sequences (Gillings and Holley, 1997). Other scientists reported similar findings. For example, Niemann et al. (1999) used ERIC-PCR to subtype strains of Sinorhizobium meliloti, a member of Proteobacteria that usually do not have ERIC sequences. The authors sequenced part of the amplified fragments and found that they did not match ERIC sequences; however, the resulting banding patterns were still able to serve subtyping purposes. Wei et al. (2004) also reported that the amplified fragments by their ERIC-PCR protocol showed no similarity to ERIC sequences. This indicated that the short primers might bind to nonspecific regions. Therefore, it appeared that the ERIC primers work as arbitrary primers just like those primers used in RAPD. Wilson and Sharp (2006) stated that we should revisit earlier conclusions that many bacteria species contain ERIC sequences based on positive ERIC-PCR amplifications. Deplano et al. (2000) described a multicenter evaluation of epidemiological typing of methicillin-resistant S. aureus strains by repetitive-element PCR analysis, a study conducted by the European Study Group on Epidemiological Markers of the European Society of Clinical Microbiology and Infectious Diseases. The study used PFGE as the reference method and showed that rep-PCR had lower discriminatory power insufficient interlaboratory reproducibility. The authors concluded that it was difficult to fully standardize rep-PCR assays.

Multiple-locus variable-number tandem repeat analysis

Multiple-locus variable-number tandem repeat analysis (MLVA) is one of CDC's candidates for complimenting PFGE for epidemiological subtyping. MLVA targets tandem repeat (TR) polymorphisms in the genomes of different bacterial pathogens (Hyytia-Trees et al., 2010). PCR primers are designed to amplify all possible tandem repeats (TRs) in the chromosome based on whole genome sequences. The size and number of repeats at each loci are then analyzed by computer and combinations of these repeats define MLVA types (Table 1.2). Tandem repeats are well recognized as containing phylogenetic signals. The repeats sometimes are targets of evolutionary events such as mutation and recombination and these evolutionary events may change the size and number of the repeats. The number of such repeats at a specific locus is similar among isolates that are closely related and varies between unrelated isolates. TRs correlate with many genomic changes essential for bacterial survival under stress conditions. Such changes include deletions; insertions and mutations that affect gene regulation; antigenic shifts; and inactivation of mismatch repair systems (Ramazanzadeh and McNerney, 2007). TRs actually play an important role in the adaptation of bacteria, especially those with small genomes. Therefore, MLVA is expected to provide relatively accurate information about the genetic relatedness of different bacterial strains. Unlike PFGE, the targets of MLVA are specific TRs that can be PCR amplified using primers designed based on whole genome sequences. Thus, MLVA is easier to interpret than PFGE, because the fragments generated by MLVA are of known size and sequence. In addition, the essential steps in MLVA are multiplex PCR and capillary electrophoresis, which are very easy to perform, standardize, and automate, making MLVA a potentially high-throughput subtyping method (Lindstedt, 2005; Lindstedt et al., 2008). The final results in MLVA are sizes of each TR loci and thus it is easier to compare than gel-banding patterns generated by other fragment-based methods. The key to the development of reliable and accurate MLVA schemes is the identification of TRs. For instance, one of the limitations associated with the development of MLVA for Salmonella is that different serovars differ slightly in their genomic organizations and thus some well-characterized TRs may not be present in all serovars. In this case, serovar-specific MLVA typing schemes have been developed (Ross and Heuzenroeder, 2005; Ross et al., 2011). Large amounts of complete genome sequence data are essential for developing MLVA schemes and PFGE is superior to MLVA in this aspect since no prior knowledge of the whole genome sequence data is required with PFGE (Karama and Gyles, 2010; Kruy, van Cuyck, and Koeck, 2011; Sobral et al., 2012). A list of TRs used to develop MLVA strategies for L. monocytogenes are listed in Table 1.3.

Table 1.2 Select VNTR loci in L. monocytogenes identified in the literature

Table 1.3 Comparison of major strain typing methods in terms of performance on various criteria

Another important feature for the development of reproducible and epidemiologically relevant MLVA schemes is the stability of the targets. Some TRs can be very unstable and potentially separate isolates within the same outbreak clone, which would confound the study of long-term epidemiology. Some extremely unstable TRs may even change during regular laboratory culturing and affect the reproducibility of MLVA. Another potential drawback of MLVA is that the primers for amplifying the TRs are designed based on the relatively small number of whole genome sequences currently available. Consequently, typeability may become a limiting factor because not all TRs from strains of the same species may be successfully amplified. For example, an insertion within a TR would confound the analysis of the size of the TR. Therefore, the selection of TRs for MLVA typing and design of PCR primers are critical to an epidemiologically relevant MLVA scheme. Intensive evaluation and validation are needed for each MLVA scheme.

MLVA has been applied to many foodborne pathogens such as E. coli, Salmonella spp., and L. monocytogenes and has been proven to yield very high discriminatory power. Hyytia-Trees et al. (2006) evaluated the epidemiologic relevance of a MLVA scheme for E. coli O157:H7 and claimed MLVA had promising epidemiologic relevance by correctly clustering isolates belonging to eight well-characterized outbreaks. In 2006, a MLVA scheme for subtyping L. monocytogenes was described by Murphy et al. (2007). In that study, MLVA was shown to discriminate isolates of the same serotype and correlate with PFGE data from the same set of isolates. Kawamori et al. (2008) compared MLVA and PFGE for subtyping E. coli O157:H7 and found that there was a good correlation between MLVA and PFGE.

Although MLVA is a fragment-based method, the utilization of meaningful molecular markers, PCR, and capillary electrophoresis generate a more phylogenetically meaningful and nonambiguous output, which provide a major evolution over other fragment-based subtyping methods. Ross et al. (2009) evaluated MLVA and PFGE for subtyping Salmonella Enteritidis, and the discriminatory indexes were 0.968 and 0.873, respectively. These studies demonstrated the great potential of MLVA for reliable and rapid subtyping of foodborne pathogens.

PCR-restriction fragment length polymorphism (PCR-RFLP)

RFLP is a subtyping technique that targets the polymorphisms within and between restriction sites. Briefly, genomic DNA from cell cultures is cut into fragments using restriction enzymes, and then the restriction fragments are separated using gel electrophoresis (Figure 1.4). Different strains can differ in the distances between restriction sites or in the sequences between the restriction sites (insertions, deletions, etc.), and thus yield different gel patterns. The number of restriction sites in the whole genome varies from 10 to 1000 depending on the type of restriction enzyme used. Some frequent cutter restriction enzymes can produce >1000 fragments with different sizes. Some rare cutters can produce around 10 fragments with sizes ranging from 500 to 800 000 bp (Zheng and Kathariou, 1995; Simpson, Santo Domingo, and Reasoner, 2002; Rousseaux et al., 2004; Foley, Lynne, and Nayak, 2009).

c1-fig-0004

Figure 1.4 PCR-RFLP. Primers are designed to amplify a specific genomic region and PCR amplicons are then digested with select restriction enzymes to generate fragments of various lengths. These fragments are then analyzed by gel electrophoresis.

There are three categories of RFLP analysis. First, PCR can be used to amplify a specific region of the whole genome and this region is analyzed by RFLP using frequent cutting restriction enzymes. Second, the whole genome can be analyzed using frequent-cutting restriction enzymes followed by gel electrophoresis and southern blotting using probes specific to certain genes. When the probes target rDNA genes (16S, 23S, and 16S–23S interspacer region), the method is known as ribotyping (Dolzani et al., 1994; Sontakke and Farber, 1995; Lagatolla et al., 1996). In this method, ribosome DNA is amplified, digested with a restriction enzyme, and visualized by gel electrophoresis. This procedure basically serves the same purpose as the traditional ribotyping with southern blotting. Third, the whole genome can be analyzed using rare cutting restriction enzymes, yielding fragments up to 800 000 bp (macrorestriction). Traditional gel electrophoresis is not able to analyze these large fragments, therefore, a special technique, called pulsed-field gel electrophoresis (PFGE), is required to accurately separate these large fragments (Schwartz, 1986; Schwartz and Cantor, 1984).

In this chapter, the first type of RFLP, PCR-RFLP will be discussed. PCR-RFLP was used for molecular subtyping of foodborne pathogens in early studies. PCR-RFLP targets the polymorphisms of the genomes that cause creation or abolishment of restriction enzyme, which results in different sizes of restriction fragments. The entire procedure contains PCR, restriction, and gel electrophoresis. PCR-RFLP is relatively easy to perform and does not need sophisticated and PFGE equipment (Mikasova et al., 2005). Often times, in order to increase the discriminatory power, more than one restriction enzyme is employed, and separate restriction reactions may be needed if different enzymes require different reaction conditions.

The performance of PCR-RFLP is determined by the genes and enzymes used. Genes that are selected for this analysis include housekeeping genes, virulence genes, and those genes encoding important surface proteins. PCR-RFLP analysis using single gene and enzyme usually provides limited discriminatory power and is often used for species identification. Kärenlampi, Tolvanen, and Hänninen (2004) used AluI to digest the PCR products of partial groEL and found that the PCR-RFLP essay performed better than 16S rRNA sequencing for the identification of Campylobacter spp. By targeting ial gene, Kingombe, Cerqueira-Campos, and Farber (2005) developed a PCR-RFLP essay for the differentiation between enteroinvasive E. coli and Shigella spp.

Multiple restriction enzymes and multiple genomic loci are used to enhance the discriminatory power of PCR-RFLP and generate data containing phylogenetic signals. For example, PCR-RFLP analysis of four virulence genes classified L. monocytogenes into two subdivisions, with division I containing serotypes 1/2a, 1/2c, and 3c and division II containing serotypes 1/2b, 3b, and 4b. This classification expanded previous findings concerning the population structure of L. monocytogenes by multilocus enzyme electrophoresis (MEE) and was subsequently confirmed by many other molecular subtyping methods (Vines et al., 1992). De Baets et al. (2004) described a PCR-RFLP protocol targeting stx genes and using restriction enzymes, HincII, AccI, HaeII, PvuII for subtyping shiga-toxin E. coli (STEC) strains. The authors found that the PCR-RFLP profiles statistically correlated with severe symptoms of the patients and therefore could be used to evaluate the virulence of STEC strains. Lukinmaa et al. (2004) described a PCR-RFLP method targeting lipooligosaccharide (LOS) biosynthesis genes of Campylobacter jejuni and found the method to be superior to serotyping for subtyping purposes. A study of PCR-RFLP targeting fla gene of C. jejuni suggested that fla is a good marker for PCR-RFLP (Owen et al., 1993). Bidet et al. (1999, 2000) evaluated PCR-ribotyping for the outbreak investigations of C. difficile and compared it with RAPD and PFGE. The authors found that PCR-ribotyping offers the best combination of typeability and discriminatory power for the entire strain collection, and with a clustering consistent with PFGE clusters. The typeability of PFGE is not as good as PCR-ribotyping for some serotypes, but for some other serotypes, ribotyping yielded much lower discriminatory power than PFGE.

PCR melting profile analysis

PCR melting profile analysis (PCR-MP) is another molecular subtyping technique based on PCR. This technique was first described by Masny and Plucienniczak (2003) and uses low denaturation temperatures (80–88°C) during ligation-mediated PCR. Under low denaturation temperatures, less stable DNA fragments are amplified. With this technique, genomic DNA is first digested with a restriction enzyme and then amplified by ligation-mediated PCR, the small number of less stable DNA fragments are then visualized by gel electrophoresis (Masny and Plucienniczak, 2003). In order to increase the discriminatory power, multiple denaturation temperatures can be used in separate PCR and the electrophoretic patterns generated under different denaturation temperatures can be combined for analysis. PCR-MP patterns have been found to be very specific to individual genomes. Krawczyk et al. (2006) used this technique for the epidemiological investigation of a group of E. coli strains and concluded that PCR-MP is a rapid, discriminatory, and reproducible technique. The clustering of PCR-MP correlated with the clustering generated by PFGE. Krawczyk et al. (2007) used this method to study the intraspecies genetic relatedness of S. aureus. The authors only used one denaturation temperature to save time and labor and found that PCR-MP generated clusters of isolates that are consistent with PFGE results. PCR-MP was shown to be reproducible, with variations in the intensity of certain gel electrophoretic bands but not in the presence/absence of bands. The authors cautioned that the precise denaturation temperature is crucial to the reproducibility of the method and found that when identical reaction mixes were analyzed in different models of thermal cyclers, the resulting banding patterns were slightly different.

Comparison of different methods

All the methods described earlier have been extensively used to study the population structures and epidemiology of various foodborne pathogens. Table 1.3 lists a summary the performance criteria of major PCR-based subtyping methods. For each method, there are different versions (i.e., same technique but different selection of genomic loci and restriction enzymes). Therefore, many studies have focused on the comparison of various molecular subtyping methods for specific pathogens. Melles et al. (2009) analyzed 994 S. aureus strains by MLVA and high throughput-AFLP, and found these methods to have similar discriminatory power. Turki et al. (2013) compared PFGE, ribotyping, ERIC, and RAPD for the differentiation of a set of 57 Salmonella Kentucky strains in Tunisia. RAPD was conducted using two different primers (RAPD1 and RAPD2). The authors found that RAPD2 and ERIC were the most discriminatory. Delgado and Mayo (2004) compared PFGE and RAPD when analyzing 147 strains of drug-resistant Salmonella Typhimurium in Spain. In this study, 36 RAPD profiles and 38 PFGE were generated, indicating that PFGE was slightly more discriminatory than RAPD. Lim et al. (2005) compared RAPD with three different primers, ERIC, and ribotyping for the typing of 57 strains of Salmonella in Korea. These 5 methods generated 42, 51, 54 (RAPD), 50 (ERIC), and 4 (ribotyping) patterns. RAPD and ERIC yielded similar discriminatory power and ribotyping yielded the lowest performance. A combination of RAPD and ERIC would differentiate all 57 strains of Salmonella. Torpdahl et al. (2005) evaluated AFLP, PFGE, and MLST for the subtyping Salmonella. These authors found a consistent clustering of strains using all three methods. AFLP and PFGE were found to be more discriminatory than MLST; AFLP and PFGE had similar discriminatory power, but AFLP was less reproducible and more time consuming than PFGE. The authors concluded that PFGE is the preferred method for surveillance and outbreak investigations, and AFLP is still very useful for local outbreak investigations. Iyoda et al. (1999) compared AFLP and PFGE using a set of E. coli serotype O157:H7 isolates and found that both methods had similar discriminatory power. Zhou et al. (2011) compared AFLP with several restriction enzymes versus PFGE for subtyping of Vibrio cholerae serogroups O1 and O139. The authors found that the applicability of AFLP in V. cholerae subtyping and outbreak investigations is limited due to the lower discriminatory power compared to PFGE. Hahm et al. (2003) compared multiplex-PCR, rep-PCR, BOX-PCR, PFGE, ribotyping, and AFLP for the subtyping of foodborne and environmental isolates of E. coli. The authors demonstrated that all methods yielded satisfactory differentiation of O157 from other serotypes of E. coli. Except for PFGE, other methods clustered some O55 strains together with O157 strains. In addition, PFGE was the only method that correctly clustered O157 strains according to their source, while PFGE appeared to be the most effective methods for subtyping O157 strains for outbreak investigation. Foley et al. (2004) compared PFGE, MLST, and rep-PCR for the subtyping of 128 strains of S. enterica serovar Typhimurium isolated from food animals. The results showed that the three methods had similar discriminatory power. However, the authors found that the clusters generated by PFGE, MLST, and REP did not have any correlations and none of the methods yielded accurate clustering of isolates according to their sources. REP was the least reproducible method among the three. Eriksson et al. (2005) compared PFGE, ribotyping, and RAPD in a study of Salmonella strains associated with an outbreak in Sweden and Norway. The authors found that PFGE and ribotyping had similar reproducibility, while RAPD had significantly lower reproducibility. All three methods have equally satisfactory typeability.

Concluding remarks

Based on the preceding discussions, readers can see that fragment-based subtyping methods that utilize PCR technology sometimes suffer from poor reproducibility due to the internal variability of PCR and restriction analyses. The same PCR on the same culture in different operations may generate slightly different patterns due to the variability of primers, enzymes, buffer, thermocycler, or DNA template. Gel electrophoresis is another factor of internal variability such as uneven lane-to-lane migration of DNA fragments, and variations in intensity of bands in separate runs. The presence of multiple bands with similar sizes makes it difficult to analyze the gel pattern. Therefore, many commercial software packages have been developed to aid gel-banding pattern recognition and analysis. Gerner-Smidt et al. (1998) concluded that the computer software were robust and performed well after evaluating two commercial software packages. However, the analysis of gel-banding patterns using commercial software is not always reliable and the usefulness and reliability of currently available software is still under debate. Eriksson et al. (2005) compared three analytical systems for DNA banding patterns of Cryptococcus neoformans and found that different algorithms provided slightly different topologies using the same set of isolates. In another study, Rementeria et al. (2001) conducted a thorough comparison and evaluation of three commercial software packages for the analysis of gel-banding patterns for RAPD and PFGE. The authors found general agreement between different software and visual observation, but slight discrepancies still existed. The authors finally concluded that computerized analyses based on gel-banding patterns do not provide an indisputably correct analysis in genotype definition. The computerized analysis of different gel images must go through a normalization process that needs to be supervised by operators to make decisions at some steps, and thus the final results are subjective. Singer, Sischo, and Carpenter (2004) found that subjectivity can possibly influence the divergence between gel-banding patterns and the true genetic relationship of isolates. A commonly used algorithm for the analysis of banding pattern data is the UPGMA (unweighted pair group method with arithmetic mean) analysis, which is based on the number of different bands and the number of common bands. However, UPGMA itself is not a good algorithm for inferring the genetic relationship among different bacterial strains; therefore, it is difficult to accurately infer the relatedness of isolates. van Belkum (2000) suggested that a binary output (numbers or characters) are preferred over gel-banding patterns for molecular subtyping strategies and Duck et al. (2003) showed that parameters of computer software need to be optimized for each species to compensate for the various intra- and inter-gel variations in PFGE libraries, and that the algorithms used for gel analysis still need to be improved.

Another limitation of fragment-based methods is that the mechanisms of the variations detected by these methods are poorly understood, which makes it difficult to infer genomic changes between different isolates from gel-banding patterns. This limitation was recently overcome by the application of DNA sequence-based typing approaches, which can be used for direct analysis of an evolutionary relationship between different bacterial isolates (Chen, Zhang, and Knabel, 2005, 2007). Overall, electrophoretic banding patterns used in various fragment-based methods are not phylogenetically meaningful and have proven not to be epidemiologically meaningful under some circumstances.

Despite of the disadvantages discussed earlier, PCR fragment-based subtyping methods, have facilitated the epidemiologic investigation of foodborne outbreaks. As discussed earlier, many PCR-based methods generate clusters that are consistent with the general population structures of various foodborne pathogens even though the further inference of strain relatedness within a type using these methods can be problematic. Before sequencing technology is widely available to the scientific community, these PCR fragment-based subtyping methods have greatly contributed to our understanding of the epidemiology of various foodborne pathogens and served as basis for the development of more advanced technologies.

References

Aguado, V., Vitas, A.I. and García-Jaloń, I. (2001) Random amplified polymorphic DNA typing applied to the study of cross-contamination by Listeria monocytogenes in processed food products. Journal of Food Protection, 64, 716–720.

van Belkum, A. (2000) Molecular epidemiology of methicillin-resistant Staphylococcus aureus strains: state of affairs and tomorrow' s possibilities. Microbial Drug Resistance, 6, 173–188.

van Belkum, A. and Hermans, P.W. (2001) BOX PCR fingerprinting for molecular typing of Streptococcus pneumoniae. Methods in Molecular Medicine, 48, 159–168.

Bidet, P., Barbut, F., Lalande, V. et al. (1999) Development of a new PCR-ribotyping method for Clostridium difficile based on ribosomal RNA gene sequencing. FEMS Microbiology Letters, 175, 261–266.

Bidet, P., Lalande, V., Salauze, B. et al. (2000) Comparison of PCR-ribotyping, arbitrarily primed PCR, and pulsed-field gel electrophoresis for typing Clostridium difficile. Journal of Clinical Microbiology, 38, 2484–2487.

Brusetti, L., Malkhazova, I., Gtari, M. et al. (2008) Fluorescent-BOX-PCR for resolving bacterial genetic diversity, endemism and biogeography. BMC Microbiology, 8, 220.

Burr, M.D., Josephson, K.L. and Pepper, I.L. (1998) An evaluation of ERIC PCR and AP PCR fingerprinting for discriminating Salmonella serotypes. Letters in Applied Microbiology, 27, 24–30.

Cesaris, L., Gillespie, B.E., Srinivasan, V. et al. (2007) Discriminating between strains of Escherichia coli using pulsed-field gel electrophoresis and BOX-PCR. Foodborne Pathogens and Disease, 4, 473–480.

Chen, Y., Zhang, W. and Knabel, S.J. (2005) Multi-virulence-locus sequence typing clarifies epidemiology of recent listeriosis outbreaks in the United States. Journal of Clinical Microbiology, 43, 5291–5294.

Chen, Y., Zhang, W. and Knabel, S.J. (2007) Multi-virulence-locus sequence typing identifies single nucleotide polymorphisms which differentiate epidemic clones and outbreak strains of Listeria monocytogenes. Journal of Clinical Microbiology, 45, 835–846.

Davin-Regli, A., Abed, Y., Charrel, R.N. et al. (1995) Variations in DNA concentrations significantly affect the reproducibility of RAPD fingerprint patterns. Research in Microbiology, 146, 561–568.

De Baets, B.L., Van der Taelen, I., De Filette, M. et al. (2004) Genetic typing of shiga toxin 2 variants of Escherichia coli by PCR-restriction fragment length polymorphism analysis. Applied and Environmental Microbiology, 70, 6309–6314.

Delgado, S. and Mayo, B. (2004) Phenotypic and genetic diversity of Lactococcus lactis and Enterococcus spp. strains isolated from Northern Spain starter-free farmhouse cheeses. International Journal of Food Microbiology, 90, 309–319.

Deplano, A., Schuermans, A., Van Eldere, J. et al. (2000) Multicenter evaluation of epidemiological typing of methicillin-resistant Staphylococcus aureus strains by repetitive-element PCR analysis.The European Study Group on Epidemiological Markers of the ESCMID. Journal of Clinical Microbiology, 38, 3527–3533.

Del Vecchio, V.G., Petroziello, J.M., Gress, M.J. et al. (1995) Molecular genotyping of methicillin-resistant Staphylococcus aureus via fluorophore-enhanced repetitive-sequence PCR. Journal of Clinical Microbiology, 33, 2141–2144.

Dolzani, L., Tonin, E., Lagatolla, C. and Monti-Bragadin, C. (1994) Typing of Staphylococcus aureus by amplification of the 16S-23S rRNA intergenic spacer sequences. FEMS Microbiology Letters, 119, 167–173.

Duck, W.M., Steward, C.D., Banerjee, S.N. et al. (2003) Optimization of computer software settings improves accuracy of pulsed-field gel electrophoresis macrorestriction fragment pattern analysis. Journal of Clinical Microbiology, 41, 3035–3042.

Enger, L., Joly, S., Pujol, C. et al. (2001) Cloning and characterization of a complex DNA fingerprinting probe for Candida parapsilosis. Journal of Clinical Microbiology, 39, 658–669.

Eriksson, J., Löfström, C., Aspán, A. et al. (2005) Comparison of genotyping methods by application to Salmonella livingstone strains associated with an outbreak of human salmonellosis. International Journal of Food Microbiology, 104, 93–103.

Foley, S.L., Simjee, S., Meng, J. et al. (2004) Evaluation of molecular typing methods for Escherichia coli O157:H7 isolates from cattle, food, and humans. Journal of Food Protection, 67, 651–657.

Foley, S.L., Zhao, S. and Walker, R.D. (2007) Comparison of molecular typing methods for the differentiation of Salmonella foodborne pathogens. Foodborne Pathogens and Disease, 4, 253–276.

Foley, S.L., Lynne, A.M. and Nayak, R. (2009) Molecular typing methodologies for microbial source tracking and epidemiological investigations of Gram-negative bacterial foodborne pathogens. Infection, Genetics and Evolution, 9, 430–440.

Gerner-Smidt, P., Graves, L.M., Hunter, S. and Swaminathan, B. (1998) Computerized analysis of restriction fragment length polymorphism patterns: comparative evaluation of two commercial software packages. Journal of Clinical Microbiology, 36, 1318–1323.

Gillings, M. and Holley, M. (1997) Repetitive element PCR fingerprinting (rep-PCR) using enterobacterial repetitive intergenic consensus (ERIC) primers is not necessarily directed at ERIC