OUP user menu

Evaluation of VNTR typing for the identification of Mycobacterium ulcerans in environmental samples from Victoria, Australia

Caroline J. Lavender , Timothy P. Stinear , Paul D.R. Johnson , Joseph Azuolas , Mark Eric Benbow , John R. Wallace , Janet A.M. Fyfe
DOI: http://dx.doi.org/10.1111/j.1574-6968.2008.01328.x 250-255 First published online: 1 October 2008

Abstract

Reliable molecular detection of Mycobacterium ulcerans in environmental samples is essential to study the ecology and transmission of this important human pathogen. Variable number tandem repeat (VNTR) typing is a valuable method for distinguishing M. ulcerans isolates from different geographic regions and for distinguishing M. ulcerans from other members of the Mycobacterium marinum/M. ulcerans complex, but its application to environmental samples has not yet been evaluated systematically. This study compares the sensitivity and specificity of PCR detection of 13 VNTR loci to determine the best loci for the analysis of environmental samples. This study demonstrates that VNTR typing using selected loci can be a useful addition to established molecular methods for detecting M. ulcerans in the environment and highlights some of the issues encountered when using molecular methods to detect microorganisms in environmental samples. When applied to environmental samples collected from an endemic region in Victoria, Australia, VNTR typing confirmed that the strain of M. ulcerans being detected was indistinguishable from the strain causing disease in humans in that region.

Keywords
  • Mycobacterium ulcerans
  • Buruli ulcer
  • VNTR typing
  • environmental

Introduction

Mycobacterium ulcerans is the causative agent of Buruli ulcer (BU), a debilitating skin disease that has been reported in 30 countries (Johnson et al., 2005), predominantly in riverine areas with tropical and subtropical climates, but also in temperate climates such as south-eastern Australia (Johnson et al., 1996; Veitch et al., 1997). In this part of Australia, there is currently an outbreak in the small coastal town of Point Lonsdale, c. 60 km southwest of Melbourne in the state of Victoria (Johnson et al., 2007). The precise mode of transmission and environmental reservoir(s) of M. ulcerans remain unknown, but these are ongoing subjects of intense research (Merritt et al., 2005).

Essential to studies on the ecology and transmission of M. ulcerans is the reliable molecular detection of M. ulcerans in environmental samples, as culturing M. ulcerans directly from such samples remains problematic (Portaels et al., 2008). We recently described two multiplex real-time PCR assays targeting the M. ulcerans insertion sequences IS2404 and IS2606 and a sequence encoding the ketoreductase B domain (KR). These enabled us to distinguish between M. ulcerans and other closely related mycolactone-producing mycobacteria (MPM) that also harbour IS2404 and IS2606, based on the relative number of copies of these target sequences (Fyfe et al., 2007). However, these assays do not reveal whether the DNA being detected originated from the same strain of M. ulcerans as that causing disease in humans in Victoria.

Variable number tandem repeat (VNTR) and mycobacterial interspersed repetitive unit (MIRU) typing are valuable methods for distinguishing M. ulcerans isolates from different geographic regions and for distinguishing between M. ulcerans and the other MPM (Ablordey et al., 2005a, b, 2007; Stragier et al., 2007). However, their application to DNA extracts from environmental samples, which often contain low concentrations of M. ulcerans DNA, PCR inhibitors and DNA from many other sources that may generate nonspecific PCR products, has not been evaluated systematically. The aim of the present study was to compare the sensitivity and specificity of PCR detection of VNTR loci 1, 4, 6, 8, 9, 14, 15, 18, 19 (Ablordey et al., 2005a, b) and ST1 (Hilty et al., 2006), and MIRU loci 1, 9 and 33 (Stragier et al., 2005) using the published reaction conditions, and to compare the ability of each locus to differentiate between M. ulcerans and the other MPM on the basis of PCR product size and/or nucleotide sequence, in order to identify the best loci for the analysis of environmental samples.

Materials and methods

VNTR/MIRU PCRs were performed on three groups of DNA extracts: (1) 10-fold serial dilutions of known amounts of purified genomic M. ulcerans DNA from a Victorian patient isolate, (2) 10-fold serial dilutions of known amounts of purified genomic DNA prepared in four types of IS2404-negative environmental DNA extracts (referred to as spiked extracts) and (3) four types of IS2404-positive environmental DNA extracts from Point Lonsdale with matching IS2404-negative control extracts (referred to as true extracts) (Table 1). The four diverse sample types were chosen to represent the kinds of samples tested routinely by researchers in the BU community: aquatic plant biofilm, mosquitoes, water filtrand (residue remaining in the filter paper following filtration) and marsupial faeces (Order: Diprotodontia).

View this table:
1

Samples used in the study showing the results of IS2404, IS2606 and KR real-time PCR and VNTR/MIRU PCR

PCR templateApproximate no. of genomes/PCRReal-time PCR CT valuesVNTR/MIRU locus and expected size (bp) of PCR product
IS2404IS2606KR1468914151819ST1MIRU 1MIRU 9MIRU 33
179431510659486634547387344424404442802
M. ulcerans genomic DNA100023.3326.1424.80+++++++++++++
10026.4429.3128.09+++++++++++
1028.8233.2631.17++++++++#
132.3135.8033.85+#
0NDNDND#
Spiked biofilm DNA extract100023.0926.0124.72+++++++/#++++×
10026.3129.0627.78++++++#+++×
1029.2732.7431.22++×#+××
132.1535.9034.12#×#××
0NDNDND×#×
Spiked mosquito DNA extract100023.1526.2924.85++++++++++××
10026.1229.2328.00++++++++++××
1029.7832.5131.63+++++××
132.2735.3534.56×××
0NDNDND××
Spiked water filtrand DNA extract100023.2525.9924.83++++++++++×##
10026.2329.1227.98++++++++++×##
1029.4332.5130.79++++×##
132.2635.5936.52+×##
0NDNDND××##
Spiked faecal DNA extract100023.3026.2324.78++++++++++##
10026.3929.3427.95+++++++++##
1029.3032.8231.48++++++##
132.8036.3135.64×++##
0NDNDND##
True biofilm DNA extracts1–1031.6033.2933.78+
0NDNDND
True mosquito DNA extracts≤133.9536.1236.82#
0NDNDND###
True water filtrand DNA extracts1–1032.0033.6133.64#+/#×#
0NDNDND###×#
True marsupial faeces DNA extracts103–10422.0024.0023.32+++++++++++#
100–100024.1026.0125.74++++++++++++
100–100024.0026.5125.39++++++++++#
100–100025.5128.2226.97++++++++++#
10–10027.6230.0729.33+/##×#
1–1030.5433.332.90××#
1–1031.5533.1532.93#
≤134.3638.1838.61##
≤137.5439.69###
0#
0#
  • Legend: +, size and sequence of PCR product identical to Mycobacterium ulcerans Victorian isolate; ×, similar size to expected PCR product but sequence did not match M. ulcerans; #, size of PCR product different to Victorian isolate; −, no visible PCR product.

  • * For purified genomic DNA and spiked environmental samples, DNA concentration was determined by a spectrophotometer and calculated using the predicted mass of a single copy of the M. ulcerans genome (5.8 fg) (Stinear et al., 2007); for true environmental samples, DNA concentration was estimated using a standard curve generated by the analysis of known amounts of genomic DNA (Fyfe et al., 2007); ND, not detected.

  • CT, cycle threshold.

  • Expected size based on M. ulcerans genomic DNA from Victorian patient isolate.

  • § Patient isolate from Point Lonsdale, Victoria, diluted in water.

  • Sequence of PCR product was c. 80% similar to Mycobacterium avium 104 (GenBank accession number CP000479.1).

  • Sequence of PCR product was c. 90% similar to Stenotrophomonas maltophilia (GenBank accession number AY956411.1).

  • ** Sequence of PCR product was c. 80% similar to Xanthomonas axonopodis (GenBank accession number AE012008.1).

  • †† Sequence of PCR product was c. 80% similar to Methylibium petroleiphilum (GenBank accession number CP000555.1).

  • ‡‡ Sequence of PCR product was c. 90% similar to Methylobacterium sp. (GenBank accession number CP000943.1).

For the true IS2404-positive biofilm, mosquito and water filtrand DNA extracts, we selected those extracts with the highest concentrations of M. ulcerans DNA in our collection as determined using IS2404 quantitative PCR (Fyfe et al., 2007). The M. ulcerans DNA concentration of the biofilm and water filtrand was estimated to be 1–10 genomes μL−1 DNA extract and the concentration of M. ulcerans DNA in the mosquito sample was estimated to be ≤1 genome μL−1 DNA extract (Table 1). For the true IS2404-positive faecal DNA extracts, a subset of nine extracts were selected with estimated M. ulcerans DNA concentrations ranging from 103–104 genomes μL−1 DNA extract to ≤1 genome μL−1 DNA extract (Table 1).

All DNA preparations were assayed for IS2404, IS2606 and KR as described previously (Fyfe et al., 2007) (Table 1). These three qPCRs served to measure the relative amounts of M. ulcerans DNA in the samples, test for the presence of PCR inhibitors (because the IS2404 assay is multiplexed with an assay for an internal positive control) and determine the IS2404 : IS2606 copy number ratio (which differentiates M. ulcerans from the other MPM). In all cases where real-time PCR cycle threshold (CT) values were obtained for both IS2404 and IS2606, the ΔCT (IS2606–IS2404) was in the expected range for M. ulcerans, suggesting that M. ulcerans was being detected and not another MPM. According to Fyfe (2007), the average ΔCT (IS2606–IS2404) for M. ulcerans is 2.37 [95% confidence interval (CI) 2.17–2.79], while for the other MPMs it is 7.60 (95% CI 6.94–8.07).

PCRs for the 13 VNTR/MIRU loci were performed using the conditions described previously (Ablordey et al., 2005a, b; Stragier et al., 2005; Hilty et al., 2006) in 25 μL reactions using 1 μL of DNA template. PCR products of the expected size were sequenced and analysed as described previously (Fyfe et al., 2007). Where more than one PCR product was generated in a single reaction, the predominant band was sequenced where possible.

Results and discussion

The results of the VNTR/MIRU PCRs for the three groups of DNA extracts are shown in Table 1. Using purified genomic DNA as a template, PCRs for VNTR loci 1, 4, 6, 8, 9, 19, ST1 and MIRU locus 1 were the most sensitive of the 13 reactions tested, each amplifying the target sequences from DNA preparations with ≥10 genomes μL−1 DNA extract. PCRs for VNTR locus 18 and MIRU locus 9 were the least sensitive and MIRU locus 33 was the only locus for which nonspecific PCR products were generated. Using IS2404-negative environmental DNA extracts spiked with genomic DNA as a template, PCRs for VNTR loci 1, 4, 6, 19 and ST1 were the most sensitive, each giving consistent amplification of the target sequences from DNA preparations with ≥10 genomes μL−1 DNA extract and generating few, if any, nonspecific PCR products. PCRs for VNTR loci 8, 9 14, 15 and 18 and MIRU loci 1, 9 and 33 demonstrated comparatively lower sensitivity and/or specificity, with MIRU loci 1 and 33 generating nonspecific amplicons in every reaction. Thus, as expected, the major difference between the PCRs using purified genomic DNA compared with spiked environmental samples was the number of nonspecific PCR products generated. In particular, the nonspecific amplicons of the same size as the expected products generated from some samples highlight the likelihood of generating false positives when testing environmental samples and the importance of sequencing VNTR PCR products to confirm their identity (Fig. 1).

1

Agarose gel electrophoresis photo showing PCR products generated for the VNTR locus 14. Lane 1, 100-bp DNA ladder; lanes 2–5, decreasing 10-fold serial dilutions of Mycobacterium ulcerans genomic DNA prepared in water (1000, 100, 10, 1 genomes); lanes 6–9, 10-fold serial dilutions of M. ulcerans genomic DNA prepared in an IS2404-negative biofilm DNA extract (1000, 100, 10 and 1 genomes). Despite being the expected size, the sequences of the circled PCR products were most similar to Mycobacterium avium 104 (GenBank accession number CP000479.1) with c. 80% nucleotide identity. Although not included in this gel photo, a PCR product was also amplified from the nonspiked biofilm extract, which, on sequence analysis, was identified as being derived from M. avium.

For the true environmental DNA extracts, four marsupial faeces with the highest concentrations of M. ulcerans DNA generated PCR amplicons identical to the Victorian outbreak strain for at least 10 of the 13 loci tested (VNTR loci 1, 4, 6, 8, 9, 14, 15, 19, ST1 and MIRU locus 1) (Table 1). Three DNA extracts with lower M. ulcerans DNA concentrations also generated expected PCR products for one locus only: biofilm (VNTR locus 6), marsupial faeces (VNTR locus 6) and water filtrand (VNTR locus 14). No PCR products with the expected nucleotide sequence were amplified for any of the loci from the remaining IS2404-positive (which contained lower concentrations of M. ulcerans DNA) or IS2404-negative environmental samples, indicating that these loci are linked to the presence of IS2404 and could not be amplified independently. The ability to detect M. ulcerans with the real-time PCR assay targeting IS2404, but not with PCR targeting any of the VNTR loci is expected, given the number of copies of IS2404 in the M. ulcerans genome (205–209 copies per genome) relative to the number of copies of the VNTR loci (present as single copies per genome) (Stinear et al., 2007).

The results of this study indicate that when purified genomic DNA from M. ulcerans isolates is used as a template, PCR amplification of most VNTR/MIRU loci is possible when the concentration of M. ulcerans DNA in the sample is ≥10 genomes μL−1 DNA extract. When applied to DNA extracted directly from environmental samples, the sensitivities of the PCRs were c. 10-fold lower, with PCRs for most loci requiring an M. ulcerans DNA concentration of ≥100 genomes μL−1 DNA extract. This may be due to one or more factors including the method of DNA extraction used for environmental samples, the presence of PCR inhibitors, the integrity of the DNA in the samples and the presence of DNA from other environmental organisms that may be preferentially amplified. To date, only a small proportion of IS2404-positive environmental samples collected in Victoria have contained sufficiently high concentrations of M. ulcerans DNA [estimated to be equivalent to ≥105 organisms g−1 sample material (Fyfe et al., 2007)] to permit amplification of most VNTR loci and, with a few exceptions, these have all been from marsupial faeces. Nucleotide sequencing of the products confirmed that they were identical to the VNTR sequences obtained for the Victorian strain of M. ulcerans currently causing disease in humans. The significance of such high bacterial loads in marsupial faeces samples compared with other sample types is the subject of another report (manuscript in preparation). For the remaining IS2404-positive DNA extracts for which VNTR/MIRU loci could not be amplified, it is likely that M. ulcerans is being detected and not another MPM based on the IS2404 : IS2606 copy number ratio (Fyfe et al., 2007). It should be noted that, in contrast to the findings of a study conducted in Ghana, which found a mixture of M. ulcerans and other MPM in the environmental samples tested (Williamson et al., 2008), no other MPM were detected in the environmental samples from Victoria included in this study.

According to this study, the best VNTR loci for the analysis of environmental samples in terms of PCR sensitivity and specificity are VNTR loci 1, 4, 6, 8, 9, 19 and ST1 (Table 2). However, in addition to understanding the performance characteristics of each VNTR PCR, it is also important to consider the ability of each locus to differentiate between M. ulcerans and the other MPMs (Ablordey et al., 2005a, b; Stragier et al., 2007) and, if applicable, different genotypes within a particular setting (Hilty et al., 2006). Therefore, deciding which particular VNTR loci should be part of an environmental testing protocol will depend not only on the type of samples being tested but also the geographic region from which the samples are collected and whether there are other MPM present in the environment. We evaluated the seven most sensitive and specific loci (VNTR loci 1, 4, 6, 8, 9, 19 and ST1) with respect to their ability to discriminate the Victorian outbreak strain of M. ulcerans from other M. ulcerans strains in the region (Papua New Guinea, Malaysia, Western Australia, Northern Territory and Queensland), and M. ulcerans from other MPM [Mycobacterium liflandii, Mycobacterium pseudoshotsii, M. marinum DL240490 and M. marinum DL045 (Ucko & Colorni, 2005)] by comparing both the size and the nucleotide sequence of the PCR products generated (Table 2). The results revealed that the PCR products obtained for VNTR loci 1, 6, 8, 9, 19 and ST1 enabled a clear distinction between the M. ulcerans tested and the MPM strains tested. However, only VNTR loci 8, 9 and 19 were able to distinguish the Victorian M. ulcerans genotype from other strains in the region.

View this table:
2

Relative sensitivities and specificities of VNTR/MIRU PCRs used in this study and appropriateness for use on environmental samples

PCR targetPCR sensitivityPCR specificityDistinguishes VIC strainsDistinguishes MU from MPMRecommended
VNTR-1HighHighNoYesYes
VNTR-4ModerateHighNoNoNo
VNTR-6HighModerateNoYesYes
VNTR-8ModerateHighYesYesYes
VNTR-9ModerateHighYesYesYes
VNTR-14ModerateModerateNDNDNo
VNTR-15LowHighNDNDNo
VNTR-18LowHighNDNDNo
VNTR-19HighHighYesYesYes
ST1HighHighNoYesYes
MIRU-1LowLowNDNDNo
MIRU-9LowHighNDNDNo
MIRU-33LowLowNDNDNo
  • * PCR products generated for at least three of the four types of spiked environmental samples with 10 genomes μL−1 of DNA extract (high sensitivity), 100 genomes μL−1 of DNA extract (moderate sensitivity) or 1000 genomes μL−1 of DNA extract (low sensitivity).

  • Nonspecific PCR products were generated for one or less of the four types of spiked environmental samples (high specificity), for two to three of the spiked environmental samples (moderate specificity) or all four types of environmental samples (low specificity).

  • Locus distinguishes Victorian Mycobacterium ulcerans from other strains in the Australasian region; ND, not done.

  • § Locus distinguishes M. ulcerans from the other MPM.

  • Decision based on sensitivity, specificity, differentiation of Victorian strains from other Australasian strains and differentiation of M. ulcerans from the other MPM.

  • Different PCR product size.

  • ** Two single nucleotide polymorphisms (SNPs) distinguish M. ulcerans from the MPMs at this locus. The first SNP corresponds to nucleotide 4981666 of M. ulcerans Agy99 (accession number CP000325)/nucleotide 5159589 of Mycobacterium marinum M (accession number CP000325). Mycobacterium ulcerans strains have an adenine residue while the MPMs and M. marinum have a guanine residue. The second SNP corresponds to nucleotide 4981652 of M. ulcerans Agy99/nucleotide 5159603 of M. marinum M. Mycobacterium ulcerans strains have a thymine residue while the MPMs and M. marinum have a cytosine residue.

  • †† One SNP corresponding to nucleotide 1782097 of M. ulcerans Agy99/nucleotide 2969763 of M. marinum M distinguishes M. ulcerans from the MPMs. Mycobacterium ulcerans strains have a cytosine residue at this position while the MPMs and M. marinum have a thymine residue.

This study highlights the trade-off between sensitivity and specificity when using PCR to detect M. ulcerans DNA in environmental samples. Although real-time PCR targeting a high copy number sequence such as IS2404 is a rapid, highly sensitive method for the detection of M. ulcerans, there is a limit to which one can be confident that the strain of M. ulcerans being detected is the same as that causing disease in humans. In contrast, gel-based VNTR typing combined with the necessary DNA sequencing, while less sensitive and more labour intensive, can provide a high level of confidence that the strain of M. ulcerans being detected is the same as the outbreak strain. Nevertheless, the tendency to generate nonspecific amplicons demonstrates that VNTR data generated from environmental samples without the rigour outlined in this report should be interpreted with a degree of caution. In light of this study, our strategy for testing large numbers of environmental samples for the presence of M. ulcerans now consists of: (1) real-time PCR screening for IS2404, followed by (2) confirmatory real-time PCR for IS2606 and KR and, finally, (3) PCR for the six VNTR loci recommended here only on those samples with sufficiently high concentrations of M. ulcerans DNA. This study has shown that VNTR typing using selected loci is a useful addition to the established suite of molecular methods used to detect this important human pathogen in the environment.

Acknowledgements

This work was supported by a Victorian Government Department of Human Services Public Health Research Grant (2004–2008). DHS Victoria was not involved in study design, study conduct or manuscript preparation. The research conducted by M.E.B. and J.R.W. was supported, in part, by Grant Number R01TW007550 from the Fogarty International Center through the NIH/NSF Ecology of Infectious Diseases Program, and grant number R03AI062719 awarded to Michigan State University. Collection of biofilm and water filtrand samples was supported by PASSHE Faculty Professional Development Council Grant in Pennsylvania, USA.

Footnotes

  • Editor: Roger Buxton

References

View Abstract