Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Present address: M. Torpdahl, Department of Bacteriology, Mycology and Parasitology, Statens Serum Institut, Artillerivej 5, 2300 Copenhagen, Denmark.
M. Torpdahl, Department of Bacteriology, Mycology and Parasitology, Statens Serum Institut, Artillerivej 5, 2300 Copenhagen, Denmark (e-mail: email@example.com).
Aims: This study was undertaken to investigate the usefulness of amplified fragment length polymorphism (AFLP) in determining the population structure of Salmonella.
Methods and Results: A total of 89 strains were subjected to AFLP analysis using the enzymes BglII and BspDI, a combination that is novel in Salmonella. Both species S. bongori and S. enterica and all subsp. of S. enterica were represented with emphasis on S. enterica subsp. enterica using a local strain collection and strains from the Salmonella Reference Collection B (SARB). The amplified fragments were used in a band-based cluster analysis. The tree resulting from the subgroup analysis clearly separated all subgroups with high bootstrap values with the species S. bongori being the most distantly related of the subgroups. The tree resulting from the analysis of the SARB collection showed that some serotypes are very clonal whereas others are highly divergent.
Conclusions: AFLP clearly clustered strains representing the subgroups of Salmonella together with high bootstrap values and the serotypes of subspecies enterica were divided into polyphyletic or monophyletic types corresponding well with multilocus enzyme electrophoresis (MLEE) and sequence-based studies of the population structure in Salmonella.
Significance and Impact of the Study: AFLP with the enzyme combination BglII and BspDI allows discrimination of individual strains and provides evidence for the usefulness of AFLP in studies of population structure in Salmonella.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The genus Salmonella has been subdivided into seven subgroups on the basis of DNA/DNA hybridizations in combination with biotyping (Le Minor et al. 1982, 1986). Six of the subgroups are subspecies (enterica, salamae, arizonae, diarizonae, houtenae and indica) of the major species Salmonella enterica, while the last group has been elevated to species rank S. bongori (Reeves et al. 1989). The H antigens are the flagellar proteins that form the filament for the phase 1 and 2 flagellin and are encoded by the fliC and fljB genes. Salmonella enterica subsp. enterica, salamae, diarizonae and indica are predominantly diphasic in flagellar expression whereas subspecies arizonae, houtenae and S. bongori are predominantly monophasic (Popoff 2001). On the basis of sequence comparison of a combination of genes an evolutionary model has been proposed which separates subspecies that are predominantly diphasic in flagellar expression from the predominantly monophasic subspecies (Selander et al. 1996). Other investigations based on comparison of 23S rRNA sequences (Christensen et al. 1998) and amplified fragment length polymorphism (AFLP) (Scott et al. 2002) also support this model.
Serotyping defined by the combination of surface antigens O, H and Vi according to the Kaufmann–White scheme, is used to classify members of the genus Salmonella. Currently over 2500 serotypes are defined in Salmonella, with the majority of these included in subspecies enterica (Popoff 2001). The discriminatory power of serotyping is typically low, and several phenotypical (e.g. phage- and biotyping) and DNA-based typing methods have been developed to discriminate within serotypes (Olsen et al. 1993). Some of these methods are useful in outbreak investigations but the discriminatory power and reproducibility are not always sufficient in studies of the population structure in Salmonella (Selander et al. 1986). The number of studies on the evolutionary relationship of Salmonella have increased significantly with the development of powerful tools like multilocus enzyme electrophoresis (MLEE) (Selander et al. 1986) and nucleotide sequence comparison of individual or a combination of genes (Selander et al. 1996). Data from MLEE, comparison of housekeeping and invasion genes support the proposed structure for Salmonella as well as further divide Salmonella enterica subsp. houtenae into two groups (Boyd et al. 1996). MLEE analysis of common serotypes in Salmonella has revealed a basically clonal population structure, with most isolates of the same serotype corresponding to one clonal lineage (Beltran et al. 1988; Selander et al. 1990; Boyd et al. 1993). A further development of MLEE and sequence analysis is multilocus sequence typing (MLST), which involves sequencing parts of several housekeeping genes and using them in a combined comparison (Enright and Spratt 1999; Spratt 1999). This technique has been developed for several bacterial species, i.e. Neisseria meningitidis (Maiden et al. 1998), Staphylococcus aureus (Enright et al. 2000), Streptococcus pneumoniae (Enright and Spratt 1998) and recently for Salmonella (Kidgell et al. 2002; Kotetishvili et al. 2002).
AFLP is a whole genome fingerprinting technique that can be used without prior knowledge of the DNA sequence and can be adapted to different organisms and purposes by choice of enzymes, adaptors and primers (Vos et al. 1995). In Salmonella, AFLP has mostly been used for typing purposes (Lindstedt et al. 2000; Nair et al. 2000; Desai et al. 2001) and has only recently been shown to be a powerful tool for studies of population structure, when using the EcoRI and MseI enzyme combination (Scott et al. 2002).
In this study AFLP with the BglII–BspDI enzyme combination was performed on 89 Salmonella isolates. These represented both species and all subspecies with emphasis on subspecies enterica. The AFLP-generated fragments were used in a cluster analysis to evaluate the usefulness of the technique in studies of population structure in Salmonella. The Salmonella Reference Collection B (SARB) that includes subspecies enterica strains was chosen on the basis of high genetic diversity as investigated with MLEE (Boyd et al. 1993).
Materials and methods
Bacterial strains and DNA extractions
A total of 89 Salmonella strains were used in this study. The strains representing the subgroups of Salmonella are listed in Table 1 and include eight reference strains from the Culture Collection, University of Göteborg (CCUG; Sweden) and 11 strains from The Royal Veterinary and Agricultural University (Denmark). The SARB, consisting of 72 strains from the Salmonella Genetic Stock Centre (University of Calgary, Calgary, Canada) (Boyd et al. 1993) were also included. Two strains from the SARB collection were re-serotyped and the serotyping showed that SARB49 supposed to be S. Paratyphi C actually was S. Paratyphi B and identical to SARB47. Strain SARB58 S. Sendai was serotyped to be S. Saintpaul and identical to SARB56. As a result these two isolates were not considered for AFLP analysis.
Table 1. Strains representing the seven subgroups in Salmonella
CCUG, Culture Collection (University of Göteborg).
Strains were grown overnight in Luria–Bertani (LB) medium at 37°C and DNA was extracted using the EASY-DNA kitTM (Invitrogen, Carlsbad, CA, USA).
Approximately 1 μg DNA was incubated for 2 h at 37°C in a total volume of 20 μl with 10 U of BglII and BspDI and a restriction buffer containing 10 mmol l−1 Tris–acetate (pH 7.5), 10 mmol l−1 Mg-acetate, 50 mmol l−1 K-acetate, 5 mmol l−1 dithiothreitol and 50 ng ml−1 bovine serum albumin.
The BglII–BspDI adaptor was assembled from the oligonucleotides 5′-CGGACTAGAGTACACTGTC-3′ and 5′-GATCGACAGTGTACTCTAGTC-3′ (DNA Technology, Aarhus, Denmark) by mixing equal amounts, heating for 10 min at 65°C and cooling at room temperature for 15 min. Approximately 0.25 μg digested DNA was transferred to a new tube containing 40 pmol adaptor, 1 U T4 DNA ligase, 2 μl 10X ligase buffer (United States Biochemical, Cleveland, OH, USA) and 8 μl restriction buffer. The ligation mixture was incubated at room temperature overnight.
PCR was performed with 2 μl of a 1 : 10 dilution of the ligation mixture in a total volume of 50 μl. The reaction mixture contained 2.5 μl 50 mmol l−1 MgCl2, 5 μl 10X PCR buffer and 0.3 μl Taq DNA polymerase, 0.2 mmol l−1 dNTPs (all from Gibco BRL, Gaithersburg, MD, USA) and 65 ng of each of the BglII and BspDI primers. The BglII primer FAM-5′-GAGTACACTGTCGATCT-3′ (Oswell DNA services, Southhampton, UK) was labelled with 6-carboxyfluorescein (FAM) in the 5′-end and the BspDI primer 5′-GTGTACTCTAGTCCGAT-3′ was unlabelled. PCR was performed using a T3 Thermocycler (Biometra, Goettingen, Germany) starting with 3 min at 94°C, followed by 24 cycles of 1 min at 94°C, 1 min 56°C and 1,5 min at 72°C and ending with an extension step for 10 min at 72°C.
Detection and analysis of AFLP fingerprints
PCR fragments were detected using an ABI Prism 377 sequencer (PE Biosystems, Foster City, CA, USA). One microlitre PCR product, 2 μl formamide, 0.5 μl internal size standard labelled with ROX dye (GeneScan 500 ROX; Applied Biosystems, Foster City, CA, USA) and 0.5 μl loading buffer (supplied with the size standard) were heated for 2 min at 94°C and cooled on ice. The mixture was loaded on a 5% denaturing polyacrylamide gel and the fragment separation was performed over 3.5 h.
The fragment collection and selection of fragments between 50 and 500 bp was performed using GENESCAN 3.1 software (Applied Biosystems). Fingerprints for each strain as well as the internal size standard were analysed by BioNumerics 2.0 software (Applied Maths, Kortrijk, Belgium) where bands were visually screened, normalized and matched. Neighbour-joining trees were created using the Dice similarity coefficient before creating a composite data set. The creation of a composite data set enabled bootstrap analysis, where the significance of the branches in the trees was tested with 500 computer-generated trees.
The BglII–BspDI enzyme combination was chosen in this study because of the number of bands as well as discriminatory power and reproducibility of these bands. These AFLP fingerprints generated a total of 80–100 bands per isolate in the size range 50–500 bp. A total of 89 strains were examined with the BglII–BspDI AFLP combination. All strains exhibited unique AFLP profiles and no single bands was common for all strains. All strains were run twice to ensure reproducibility and the average clustering of identical strains was found to be 98.2% (s.d. ± 1.6%).
Cluster analysis of subgroups
A total of 19 strains representing S. bongori and the subspecies of S. enterica were digested with BglII and BspDI. Amplification of the fragments generated unique fingerprints for each strain (Fig. 1). Cluster analysis of the fragments using the dice similarity coefficient and a neighbour-joining tree revealed groups corresponding to the seven subgroups of Salmonella (Fig. 1). A test of the significance of branches in the tree, resulted in high bootstrap values for each of the seven subgroups, except for subspecies indica where only one strain was analysed (Fig. 1). The bootstrap values were 100% for S. bongori (three strains), 100% for subspecies arizonae (three strains), 100% for subspecies diarizonae (four strains), 100% for subspecies houtenae (three strains), 97% for subspecies salamae (three strains) and 100% for subspecies enterica (two strains). The three S. bongori strains were the most divergent, clustering apart from the subspecies of S. enterica with a bootstrap value of 100%. Closest to the group of S. bongori were the subspecies arizonae, these two groups cluster apart from the rest of the strains with a bootstrap value of 64%. Salmonella bongori and subspecies arizonae belong to the predominantly monophasic groups, that also include subspecies houtenae. In this analysis, subspecies houtenae clustered with the predominantly diphasic groups.
Cluster analysis of SARB strains
The AFLP fragments generated from digestion of the 70 SARB strains were used in the cluster analysis (Fig. 2). A bootstrap analysis of the tree revealed several clusters with two or more strains that were supported with high bootstrap values (Fig. 2). Some serotypes that were represented by two or more strains clustered corresponding to their serotype with high bootstrap values: 100% for S. Typhi (SARB63 and 64), 100% for S. Pullorum (SARB51 and 52), 98% for S. Heidelberg (SARB23 and 24), 87% for S. Montevideo (SARB30 and 31) and 100% for S. Typhimurium (SARB65, 66, 67 and 68).
The serotypes S. Derby (SARB9, 10 and 11), S. Miami (SARB28 and 29) and S. Wien (SARB71 and 72) were found in the same lineages of the tree but with no bootstrap support. The same was the case for S. Panama although SARB40 and 41 clustered together with a bootstrap value of 100%, the third S. Panama strain, SARB39 clustered in the same part of the tree with no bootstrap support.
Other strains with the same serotype were found in different lineages of the tree. The S. Enteritidis strains (SARB16 and 18) clustered together with a bootstrap value of 98% whereas SARB17 and 19 (also S. Enteritidis) clustered together in another lineage but with no bootstrap support. The same was the case for S. Dublin (SARB12, 13 and 14), S. Muenchen (SARB32, 33, 34 and 35), S. Paratyphi B (SARB43, 44, 45, 46 and 47), and S. Cholerasuis (SARB4, 5, 6 and 7) having two or more strains closely related and one or more distantly related. The serotypes S. Paratyphi C (SARB48 and 50), S. Newport (SARB36, 37 and 38), S. Saintpaul (SARB55 and 56), S. Typhisuis (SARB69 and 70) and S. Infantis (SARB26 and 27) were all represented by two or more strains that were highly divergent and falling in different parts of the tree.
In several cases, different serotypes cluster together in larger groups and three of these were supported by a high bootstrap value. Seven strains including, S. Dublin (SARB12 and13), S. Enteritidis (SARB 16 and 18), S. Gallinarum (SARB32) and S. Pullorum (SARB 51 and 52) cluster together with a bootstrap value of 88%. Within this group all strains cluster corresponding to the serotype with high bootstrap support. The other bootstrap supported group of this kind contains four strains: S. Typhisuis (SARB69), S. Paratyphi C (SARB48) and S. Cholerasuis (SARB4 and 6) that cluster together with a bootstrap value of 99%. The last group includes six strains: S. Typhimurium (SARB65, 66, 67 and 68), S. Saintpaul (SARB55) and S. Haifa (SARB22) supported by a bootstrap value of 74%.
AFLP has previously been applied for typing and characterization of Salmonella (Lindstedt et al. 2000; Nair et al. 2000; Desai et al. 2001), but has also been used as a tool in investigations of the population structure (Scott et al. 2002). To our knowledge this is the first time the enzyme combination BglII and BspDI has been used in an AFLP investigation of the population structure in Salmonella.
The data clearly separated the subgroups with high bootstrap values (Fig. 1) and supported the elevation of S. bongori to species rank (Boyd et al. 1996), being the most divergent of the subgroups. The second most divergent subgroup was subspecies arizonae, which is in agreement with results obtained with sequence comparison of five housekeeping genes and seven invasion genes (Boyd et al. 1996), seqeuence comparison of the 23S rRNA (Christensen et al. 1998) and AFLP with EcoRI–MseI (Scott et al. 2002). A proposed evolutionary model suggesting that diphasic subspecies and the predominantly monophasic subspecies evolved in two directions (Selander et al. 1996), was not supported in this analysis as subspecies houtenae was found closer to the diphasic subspecies in the tree. Other investigations also failed to support this model, including results from a MLEE analysis (Boyd et al. 1996) as well as gene sequence comparison of mdh (Boyd et al. 1994) and atpD (Christensen and Olsen 1998).
Data from the present investigation of the SARB collection showed that serotypes are either monophyletic or polyphyletic which supports findings from a MLEE investigation (Boyd et al. 1993). The monophyletic serotypes includes S. Typhi (two strains), S. Pullorum (two strains), S. Heidelberg (two strains), S. Montevideo (two strains) and S. Typhimurium (four strains) that all clustered together with high bootstrap values corresponding to serotype. The serotypes S. Wien (two strains), S. Miami (two strains) and S. Panama (three strains) also clustered in the same lineage of the tree but with no bootstrap support. The polyphyletic serotypes include S. Newport (three strains), S. Saintpaul (two strains), S. Paratyphi C (two strains), S. Infantis (two strains) and S. Typhisuis (two strains) that are all highly divergent falling in different lineages of the tree. Also included in the polyphyletic serotypes are S. Enteritidis (four strains), S. Dublin (three strains), S. Cholerasuis (four strains), S. Muenchen (four strains) and S. Paratyphi B (five strains). They all have some strains falling in the same lineages and others in a different part of the tree.
Results from this investigation also clustered the three S. Derby together with no bootstrap support. This is supported by the EcoRI–MseI AFLP investigation (Scott et al. 2002), but in contradiction to data obtained with MLEE that indicated that this serotype is highly divergent (Boyd et al. 1993). The present data indicates that S. Haifa (SARB22) and S. Saintpaul (SARB55) are closely related to the S. Typhimurium group of strains. The same is the case in the MLEE analysis (Boyd et al. 1993) but where MLEE failed to discriminate between S. Haifa and S. Saintpaul, AFLP clearly separated these two. The cluster containing S. Paratyphi C (SARB48), S. Typhisuis (SARB9) and S. Cholerasuis (SARB4 and 6) was supported by a bootstrap value of 99% and supports the MLEE and EcoRI–MseI AFLP results (Boyd et al. 1993; Scott et al. 2002). MLEE analysis and sequencing of the fliC gene (Li et al. 1993) have revealed that S. Gallinarum and S. Pullorum have a close relationship to S. Enteritidis (SARB16). This is also indicated in this investigation by a cluster supported by a high bootstrap value containing S. Gallinarum (SARB21), S. Pullorum (SARB51 and 52) and S. Enteritidis (SARB16 and 18). Close to these are S. Dublin (SARB12 and 13) that has also been shown to be highly similar to SARB16 by MLEE analysis and sequencing of the fliC gene (Selander et al. 1992). The last S. Dublin (SARB14) was distantly related to this group but closely related to S. Paratyphi B (SARB46), which again is in accordance with the MLEE results (Boyd et al. 1993).
Although the results of this investigation support results obtained with sequence comparison (Boyd et al. 1996; Christensen et al. 1998) and MLEE (Boyd et al. 1993) several differences were shown. This might be explained by the different approaches the techniques are based on. AFLP is a whole genome technique analysing both conserved and variable regions whereas MLEE and sequencing of housekeeping genes is based on analysis of small and conserved regions of the genome. Comparing the present AFLP data and the EcoRI–MseI AFLP data (Scott et al. 2002) gives a high degree of similarity but differences are seen and there are several possible explanations for this. One is that AFLP analysis is highly dependant on visual screening and therefore the person analysing the data. Another explanation could be the parameters (including the cut off values) used when analysing data. Finally, AFLP is dependant on the recognition sites of the preferred combination of restriction enzymes. The choice of restriction enzymes affects the number of bands; many bands might be highly discriminatory but make the analysis more difficult. Fewer bands, although still discriminatory to a certain level might be better to reveal clonal lineages within S. enterica subsp. enterica. The complete procedure for 48 strains including laboratory work as well as computer analysis can be performed in 2 days. The analysis part can be performed rapidly when using the Pearson correlation for cluster analysis. We found that the average clustering of identical strains improved when using the Dice similarity coefficient where bands are assigned manually and therefore more time consuming.
In conclusion, AFLP is a rapid and simple technique providing results which correlate well with those obtained in other investigations by different methods. This indicates that AFLP has the potential as a valuable tool in the investigation of population structure within Salmonella.
We thank Katja Kristensen for her excellent technical assistance, Karen A. Krogfelt for critically reading the manuscript and Henrik Christensen for providing strains from The Royal Veterinary and Agricultural University (Denmark).