| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Centre for Infectious Diseases and Microbiology – Public Health, Institute of Clinical Pathology and Medical Research (ICPMR), Westmead, New South Wales 2145, Australia
Correspondence
Gwendolyn L. Gilbert
l.gilbert{at}usyd.edu.au
Received 15 November 2007
Accepted 15 February 2008
Abbreviations: AFLP, amplified fragment length polymorphism; MLVA, multilocus variable-number tandem-repeat analysis; mPCR/RLB, multiplex PCR-based reverse line blot; PT, phage type; RDNC, reaction-does-not-conform.
 |
INTRODUCTION |
|---|
 TOP INTRODUCTION METHODSRESULTSDISCUSSIONREFERENCES |
|---|
S. Typhimurium is a common and diverse serovar that requires further subtyping for surveillance or outbreak investigations. Phage typing (Anderson et al., 1977) is widely used in Australia and Europe to differentiate individual strains of S. Typhimurium. However, it is restricted to a few reference laboratories, which causes delays and limits its epidemiological value. Interpretation of results, even by experienced staff, may vary between laboratories. Untypable or ‘reaction-does-not-conform’ (RDNC) isolates are not uncommon. Often a single phage type (PT) predominates for long periods in a geographical area, making recognition of outbreaks difficult.
Molecular typing methods have been used for further discrimination of Salmonella strains, including PFGE (Guerra et al., 2000), random amplified polymorphic DNA typing (De Cesare et al., 2001), multilocus sequence typing (Fakhr et al., 2005), IS200 typing (Jeoffreys et al., 2001; Soria et al., 1994), ribotyping (Nastasi & Mammina, 1995), plasmid profiling (Millemann et al., 1995) and amplified fragment length polymorphism (AFLP) analysis (Hu et al., 2002; Lindstedt et al., 2000). Some are relatively difficult to perform, slow and expensive whilst others have limited reproducibility and discriminatory powers. Multilocus variable-number tandem-repeat analysis (MLVA) has been developed for various Salmonella serovars (Cho et al., 2007; Lindstedt et al., 2004; Liu et al., 2003; Witonski et al., 2006). It has allowed a high level of resolution of variant strains of the relatively homogeneous S. Typhimurium definitive type 104 (Lindstedt et al., 2003).
A number of phage type-specific sequences have been identified in S. Typhimurium and its phages and used to distinguish phenotypically related definitive types of S. Typhimurium by PCR and sequencing (Hu et al., 2002; Lan et al., 2007; Mikasova et al., 2005; Ross & Heuzenroeder, 2005). As this involves multiple single PCRs and sequencing, it is time-consuming and unsuitable for routine identification and typing. In this study, we developed a multiplex PCR-based reverse line blot (mPCR/RLB) assay to rapidly identify most of the phage types currently prevalent in our region and compared the results with those of MLVA.
|
METHODS |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
).
|
All isolates were grown on horse blood agar overnight at 37 °C. Single colonies of each were suspended in 200 µl molecular biology grade water and boiled for 10 min. The suspension was centrifuged and the supernatant was used for PCR.
Primers.
Thirty-two sets of primers were designed, based on previously published primers and S. Typhimurium phage-type-related AFLP fragment sequences. Of these, 14 sets were based on bacteriophage ST64B (Ross & Heuzenroeder, 2005), 12 on bacteriophages P22, Gifsy-1 and Gifsy-2 (Mikasova et al., 2005), and six on the AFLP fragments (Hu et al., 2002). All primers were designed or modified so that they were 18–28 bp in length, with melting temperatures between 59 and 71 °C and product sizes in the range 90–249 bp (Table 2). The nucleotide–nucleotide BLAST program (National Center for Biotechnology Information) was used to compare primer sequences with sequences in the GenBank database, to minimize cross-priming with other regions of phage genomes or unrelated genes. All primers were 5' and 3' biotin labelled and supplied by Sigma-Genosys.
|
). BLAST searches of the GenBank database were performed to minimize cross-hybridization with other genes. Probes were 5' amino-labelled and supplied by Sigma-Genosys.
|
RLB hybridization. The RLB hybridization assay was performed based on a well-established protocol developed in this laboratory (Kong & Gilbert, 2006) with modifications. Probes were attached to a Biodyne Nylon 6.6 membrane (PALL; Gelman Laboratory). The hybridization temperature was 60 °C. Streptavidin–peroxidase (Roche Diagnostics) was used as conjugate and diluted 1 : 4000 in 2x SSPE (0.18 M NaCl, 10 mM NaH2PO4 and 1 mM EDTA, pH 7.7) containing 0.5 % SDS. The X-ray film (Hyperfilm; Amersham) exposure time was 7 min.
Sequencing. Sequencing was performed, using single PCR amplicons, to compare sequences of the corresponding targets from different isolates that gave strong or weak signals in RLB. The sequencing was conducted in both directions with BigDye Terminator version 3.7 (Applied Biosystems) on an ABI 3100 DNA sequencer.
MLVA. The MLVA typing method was used as described by Lindstedt et al. (2004) with the following modifications. A 30 µl multiplex PCR mix was prepared using the Qiagen HotStar Taq PCR system containing 3.0 µl of 10x PCR buffer, 2.5 mM MgCl2, 0.2 mM dNTP, 0.1 µM each forward and reverse primer, 2 µl template DNA, 0.2 µl Hotstar Taq polymerase (5 U µl–1) and water added to a total volume of 30 µl. PCR products were run on a 1.6 % normal agarose gel prior to the capillary electrophoresis. A single target PCR was run for any isolates with missing mPCR amplicons, to confirm absence of amplification. The capillary electrophoresis solution was prepared as follows: 0.5 µl of the PCR products was mixed with 0.5 µl of the ROX1000 (Bio-Ventures) internal size marker and 9 µl formamide. The mixed solution was then subjected to capillary electrophoresis on an ABI 3100 Genetic Analyzer (Applied Biosystems).
Data analysis. For each probe, RLB hybridization signal intensities on the membrane were scored as 2 for strong (gene present), 1 for weak (gene present with minor sequence variation) and 0 for absent (gene absent or present with major sequence variations) signals. These were combined into a 38-digit string for each isolate representing the RLB pattern. The string was divided into seven gene groups (G1–G7) each containing six gene markers, except for G7, which comprised only two. A single digit difference in the string was regarded as a different RLB type. The data were analysed using BioNumerics (Applied Maths) software and each unique pattern was given a number. A dendrogram was generated using the categorical coefficient and Ward's algorithm.
MLVA peaks were identified according to colours and sizes using GENESCAN software provided by Applied Biosystems and were binned into alleles. The allele numbers were entered into BioNumerics as character values. The number of repeats at each locus was calculated by subtracting the known length of the flanking sequence from the amplicon length and dividing the result by the repeat sequence length for all loci, except STTR3, which contains variable numbers of either 27 or 33 bp repeats. The lengths of flanking sequences were determined from the S. Typhimurium LT2 complete genome sequence from GenBank (accession no. NC_003197). An allele string for each isolate, in the order STTR9–STTR5–STTR6–STTR10pl–STTR3, was constructed by combining allele numbers, corresponding with the number of repeats (0 if there was no amplicon; 1 for flanking sequence only; 2 for one repeat; and 3 for two repeats and so on), for the first four loci. For STTR3, the allele numbers were based on eight amplicon size variants among all isolates tested ranging from 370 bp (allele 1) to 523 bp (allele 8). Different MLVA types were numbered consecutively as they were identified, based on differences in combinations of allele numbers.
Polymorphism indices were calculated using Simpson's index of diversity, as described by Hunter & Gaston (1988).
|
RESULTS |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
. The polymorphism index for RLB typing was 0.9866. Fig. 2 shows a representative RLB profile of selected phage types of S. Typhimurium. Of the 168 isolates, 99 were clustered into 33 RLB types, each of which contained multiple isolates with identical patterns and, in most cases, corresponded with different phage types. Each of the remaining 69 isolates gave a unique RLB pattern. All RLB types could be further clustered into five groups (A–E; Fig. 1).
|
).
All phage types with multiple isolates, except PT 99 and PT U290, produced two or more RLB patterns. For some phage types, the majority of isolates had the same RLB patterns. For example, four of five PT 101 isolates shared RLB 87, four of six PT 126 isolates belonged to RLB 3 and three of four untypable isolates shared RLB 61. Most other phage types showed significant differences in RLB patterns between isolates and some RLB patterns were shared between phage types. For example, six PT 1 isolates had five different RLB patterns spread among cluster groups B, C and E, namely RLB 25, 26, 33, 47 and 95 (the latter was shared with PT 5 and 136). Six PT 6 var1 isolates produced five RLB patterns spread between groups A, C and E, namely RLB 11, 42, 81, 82 and 83 (the latter was shared with PT 140 var1). Four PT 141 isolates each belonged to a different RLB type, namely RLB 18, 24, 92 and 94 (RLB 92 was shared with PT 4, 193 and 195). Similarly, each of five PT U302 isolates gave a different RLB pattern, namely RLB 12, 62, 64, 68 and 69 (RLB 69 was shared with PT 186) (Fig. 1).
Of the 46 phage types tested, 21 gave distinct RLB patterns, i.e. not shared by any other phage types, including 14 with multiple isolates (PT 12a, 20 var1, 30, 43, 44, 64, 104L, 126, 126 var4, 160, 197, 208, U290 and untypable) and seven with single isolates (PT 1 var, 4 var2, 29, 164, 165, 177 and 194). The number of distinct RLB patterns for each of these phage types varied between one and five. For two phage types, PT 126 var4 and PT 197, each of four and five isolates, respectively, gave different RLB patterns.
The numbers of gene target differences, between different RLB patterns for the same phage type, varied between 1 and 26 of the 38 probe targets. Of the isolates tested, PT 1 (six isolates) and PT 195 (four isolates) were the most variable, with 26 and 23 differences, respectively. Among phage types with distinct RLB patterns, PT 208 and PT 197 had 12 and 10 target differences between isolates, respectively.
The probes CA4-P2, CC2-P2 and gtgB-P1 produced strong positive signals for all phage types and therefore could be used as control markers in future work. Amplicons which produced different signal intensities with the same probe were sequenced. The results showed that those producing strong signals (score 2) were identical to corresponding probe sequences; those that gave weak signals (scored as 1) had one (isolates 146, 158 and 187) or more than one (isolates 127 and 128) nucleotide base difference. Isolates that produced amplicons (as shown by single PCR and gel electrophoresis), but gave no signal with the corresponding probe (scored as 0), differed by more than two nucleotides (isolates 100, 123, 146, 158 and 187) (Table 4).
|
The 168 S. Typhimurium isolates were separated into 97 MLVA allele strings which were numbered from 1 to 97 (Fig. 1). Twenty-six of the 97 MLVA types were represented by multiple isolates, and most of these corresponded with a single phage type. For example, MLVA types 1, 6 and 25 corresponded with PT 160, PT 44 and PT U307, respectively, and MLVA 12 corresponded with PT 135 (six of 15 PT 135 isolates belonged to other MLVA types).
Some MLVA types contained two or more phage types, suggesting close relationships between them. For example, MLVA 2 contained five phage types, MLVA 84 contained four and MLVA 35 and 69 each contained two. On the other hand, even small numbers of some phage types were represented by several MLVA types; for example, six PT 1 isolates were represented by four MLVA types, 15 PT 135 isolates were represented by seven MLVA types and five PT 101 isolates were represented by five MLVA types (Fig. 1).
In many cases, individual RLB patterns included multiple MLVA types and vice versa; for example, RLB 41 contained five MLVA types and MLVA 12 contained five RLB types (Fig. 1), indicating that RLB and MLVA types are generally complementary and together provide a higher level of discrimination than either alone. However, there were some exceptions, where the RLB, MLVA and PT types were consistent. For example, four PT 99 isolates shared the same RLB 41 and MLVA 49 types, three PT 44 isolates shared RLB 22 and MLVA 6 types, and three untypable isolates had the same RLB 61 and MLVA2 types (Fig. 1). Further information is required to determine whether isolates with consistent RLB, MLVA and phage types are epidemiologically related.
Outbreak investigations
Outbreak 1.
The drug and alcohol rehabilitation centre outbreak in 2004 involved 23 residents from whom isolates were obtained. Nineteen residents had a diarrhoeal illness during one week in July; three other cases occurred in September and one isolate was collected from a chef at the centre in October. The outbreak was believed to be either food-borne or person-to-person spread. All 23 isolates belonged to PT 135. The mPCR/RLB clearly distinguished PT 126 from PT 135 isolates. All of the latter had the same RLB type, which was the same as one of two RLB types observed among the sporadic isolates (Table 1). Twenty PT 135 outbreak isolates (16 from July, 3 from September and one from the chef in October) belonged to MLVA 12 and each of the other three isolates was a different type, as were the two PT 126 isolates. By contrast, there were eight MLVA types (including MLVA 12) among 15 sporadic PT 135 isolates (Table 1). Epidemiological investigation, before the molecular typing results were available, failed to identify a common source. The MLVA results suggest that there may have been more than one source but this could not be confirmed.
|
All 15 ‘outbreak’ isolates produced the same RLB pattern (RLB 103), which was also the most common (represented by 17 of 22 isolates) of five patterns observed among the sporadic isolates. The 14 human isolates were divided into two MLVA types (MLVA 2 and 5), which differed by a single allele. The prawn isolate and two isolates from cluster 2 belonged to MLVA 2 and the remaining 12 belonged to MLVA 5, the most common of the seven MLVA types found among the sporadic isolates (Table 1).
Outbreak 3. This outbreak occurred over a short period among students attending a residential catering college, who cooked and ate together, which suggested a single common food source. All 17 human isolates from the students had the same RLB 58 type, which was the same as that of the PT 170 controls. DNA from four food isolates hybridized with only one to three of 38 RLB targets, suggesting that they were not S. Typhimurium. The other two, from chocolate mousse, belonged to the same RLB 58 type as the human isolates.
All human isolates and the two from chocolate mousse also had the same MLVA type (MLVA 107), which differed at only one locus from MLVA 2, which had previously been shown to be associated with PT 170. The other four food isolates did not produce amplicons from two to five of the loci.
Outbreak isolates from the affected students and the chocolate mousse were later identified as S. Typhimurium PT 170. The other food isolates were identified as two Salmonella Sofia isolates, from chicken, and two Salmonella Java isolates, from fish. Epidemiological investigation implicated chocolate mousse, which contained raw egg, as the most likely common food source (Table 1).
|
DISCUSSION |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Based on published sequence data, we designed or modified 32 primer sets targeting sequences from phages ST64B (Ross & Heuzenroeder, 2005), P22, Gifsy-1 and Gifsy-2 (Mikasova et al., 2005) and S. Typhimurium phage type-related AFLP fragments (Hu et al., 2002). They were designed to produce short PCR products with similar physical and biochemical characteristics to allow simultaneous amplification in a multiplex reaction. Similarly, probes were designed so that their optimal hybridization conditions were similar. We designed two primer sets and probes for phage-related targets in order to increase the number of nucleotide polymorphisms identified and therefore the discriminatory ability of the assay. Each member of the pair serves as an internal control for the other. Because of their short length (100–150 bp), only single primer pairs and probes were used for targets based on phage type-related AFLP fragments.
This system identified and discriminated subtypes among many of the phage types tested. The RLB patterns were defined by the presence or absence and intensities of hybridization signals; differences in intensity between paired probes were confirmed by sequencing results. The mPCR results were validated by performing single PCRs for each primer set on one representative of each of the 46 phage types studied. The method reliably differentiated 21 of 46 individual phage types including some that are closely related, such as PT12 and PT12a. Some common phage types shared RLB types with one other, e.g. PT 8 with 9, PT 12 and 102 with 170, and PT 99 and RDNC with 135, suggesting a close relationship between members of the pair or group. Isolates of most of the common phage types exhibited two or more RLB patterns, indicating the potential for more discriminatory strain typing of common phage types. Multiple RLB patterns were found among the untypable and RDNC isolates, indicating that they represented diverse groups of largely unrelated strains.
The application of this novel mPCR/RLB system to the investigation of outbreaks confirmed its potential value. The first outbreak was suspected when phage-typing results demonstrated an increased number of cases due to the common phage type 135 within a short period of time. The combination of mPCR/RLB and MLVA helped to define the clusters and identify related cases some time after they had occurred. In the second outbreak, all 15 isolates from three apparently related restaurant outbreaks had the same RLB pattern, which was also one of five patterns exhibited by sporadic isolates of S. Typhimurium PT 170. These results support epidemiological data and are compatible with the supposition that the three outbreaks were related to each other and to the food sample. In the third outbreak, investigation began immediately after salmonellae were isolated from cases and three types of food. The outbreak strain and the food source were rapidly identified by mPCR/RLB and predicted the phage type involved more than a week before the phage typing results were available.
Had it been performed routinely on all S. Typhimurium isolates, the mPCR/RLB typing alone would have provided early warning of the first two outbreaks, before the phage typing or epidemiological data were available. For the third outbreak, which was immediately apparent because of the large number of related cases presenting within a short period, it rapidly defined the outbreak strain and confirmed the food source. It is more sensitive and specific than phage typing and results are available much sooner.
MLVA is also highly sensitive for strain typing, especially in outbreaks. Three of the loci tested, STTR5, STTR6 and STTR10pl, displayed a high degree of polymorphism indicated by indices of 0.91, 0.92 and 0.84, respectively, which were comparable with published work (Lindstedt et al., 2004). One of the major drawbacks of this typing strategy is that the generation of variation in some MLVA loci is under selection pressure such that two identical clones may appear to differ if isolates have originated from different niches or from hosts with altered immune responses (Lindstedt, 2005). This may be manifested as too high a level of discrimination to correctly cluster isolates from a common source in an outbreak investigation. The PT 135 isolates from the first outbreak were divided into four MLVA types, of which one was predominant and the others differed by two to three alleles. It is not clear whether the three isolates belonging to minor MLVA types were sporadic or variants of the outbreak strain. Further experience with both RLB and MLVA is required to assist interpretation of these results.
Both mPCR/RLB assay and MLVA demonstrated similar discriminatory power in this study. For some phage types, the strains identified were the same between the two methods. For example, both methods separated PT 30, PT 141 and RDNC into different molecular types for each isolate tested (Fig. 1). Some phage types contained the same number of molecular types identified by both methods; for example, four isolates of PT 102 contained three molecular types and seven isolates of PT 9 contained four. Both methods demonstrated a high level of similarity between isolates of some groups of phage types such as PT 135, PT 102 and PT 170 (Fig. 1).
Based on these results, mPCR/RLB is as sensitive and specific as MLVA. Both methods can provide informative strain typing data and can presumptively identify many common phage types within 24 h, compared with at least 3 weeks for phage typing at a reference laboratory. This difference represents a highly significant advantage for prompt identification of an outbreak source. An advantage of mPCR/RLB over MLVA is that the PCR amplicons are detected by hybridization rather than fragment size, which overcomes the problem of reproducibility between different platforms. Based on our results, mPCR/RLB showed less variation than MLVA among outbreak isolates, which made interpretation easier. MLVA is less labour intensive than mPCR/RLB and apparently more sensitive for outbreak investigations, but further experience with both methods is required to determine which more accurately reflects the epidemiology.
mPCR/RLB is a promising tool that can be developed further by the introduction of additional gene markers, including serotype-specific markers, to increase its sensitivity and discriminatory ability for Salmonella species other than S. Typhimurium. It has the potential to replace phage typing in the future.
|
ACKNOWLEDGEMENTS |
|---|
|
REFERENCES |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. Cai, F. Kong, Q. Wang, H. Wang, M. Xiao, V. Sintchenko, and G. L. Gilbert A new multiplex PCR-based reverse line-blot hybridization (mPCR/RLB) assay for rapid staphylococcal cassette chromosome mec (SCCmec) typing J. Med. Microbiol., August 1, 2009; 58(8): 1045 - 1057. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| INT J SYST EVOL MICROBIOL | J MED MICROBIOL | MICROBIOLOGY | J GEN VIROL | ALL SGM JOURNALS |
Isolates 17, 19, 26 and 35 (RLB 41) were from the drug and alcohol rehabilitation centre outbreak in New South Wales in 2004; *RLB types unique for specific phage types.