Samples
Water (n = 112) and soil (n = 27) samples were taken from August to October 2013 in Flanders, Belgium (Additional file 1). For the water samples, an autoclaved 1 L Duran bottle was opened and filled 10 cm below the water surface. Per sample, three times 150 ml was filtered using a Nalgene vacuum filter funnel and cellulose nitrate membrane filters with 0.45 μm pore size and 47 mm diameter (Thermo Scientific). For the soil samples, samples were taken 2 cm below the soil surface using a sterile spoon, and collected in sterile falcon tubes. Per sample, 1 g of soil was homogenized in 9 ml phosphate-buffered resuspension buffer (0.15 M NaCl, 10 mM EDTA, 0.1 M phosphate buffer, pH 8.0) using a Stomacher blender for 30 s at 230 rpm. Three times 1 ml of soil suspension was transferred to an Eppendorf tube and centrifuged for 5 min at 13,000 rpm (17,949 g) before removing the supernatant. Filters and soil pellets were stored at −20 °C until DNA extraction.
For each sample, sampling date, address, region (i.e., West-Vlaanderen [WV], Oost-Vlaanderen [OV], Limburg [L], Vlaams-Brabant [VB] or Antwerpen [A]), and class (i.e., swimming or recreational water [SRW], canal-river-stream [CRS] or other [O]) were recorded (Additional file 2). Swimming and recreational waters were those under surveillance of the Flemish Environment Agency (www.kwaliteitzwemwater.be). For water samples, pH and temperature were measured on site. For soil samples, pH was measured after dissolving 10 g of soil in 50 ml distilled water and magnetic stirring for 10 min.
DNA extraction from environmental samples and quality assessment
Prior to DNA extraction, filters with biological material from water samples were cut into smaller pieces with sterilized scissors. Total DNA was extracted in triplicate from the filters and soil pellets (three per sample) following the protocol for Gram-negative bacteria of Pitcher et al. [34]. DNA pellets were dissolved by adding 50 or 100 μl TE buffer depending on the size of DNA pellet, and left to dissolve overnight at 4 °C. RNA was degraded by adding 2.5 or 5 μl RNase (2 mg/ml) for pellets dissolved in 50 or 100 μl TE, respectively, and incubating at 37 °C for 1 h.
The quality and quantity of the extracted DNA were examined by measuring optical densities (OD) at 234 nm, 260 nm, 280 nm and 320 nm [24, 25] with a SpectraMax Plus 384 spectrophotometer. DNA was considered of acceptable quality if the OD260/280 ratio was higher than 1.7, the OD234/260 ratio was smaller than 1 and the OD320/260 ratio was smaller than 0.15. If both quality and quantity of the three DNA extraction replicates per sample were similar, these replicates were pooled. DNA fragmentation and RNA contamination was assessed by agarose (1 %) gel electrophoresis and EtBr staining.
To test for the presence of PCR inhibitors, DNA extracts were subjected to a 16S rRNA amplification PCR with universal primers ARI C/T (5′-CTG GCT CAG GAY GAA CGC TG-3′) and pH (5′-AAG GAG GTG ATC CAG CCG CA-3′). The PCR mix contained 1× CorelLoad PCR buffer (Qiagen), 0.2 mM dNTP (Applied Biosystems), 0.5 U AmpliTaq (Applied Biosystems) 0.1 μM of both primers and 200 ng/μl BSA (Roche). For each sample, 2 μl DNA was added to 23 μl PCR master mix. B. multivorans R-20526 DNA and sterile MQ were used as positive and negative control, respectively. PCR was performed using a MJ Research PTC-100 thermal cycler. Initial denaturation for 5 min at 95 °C was followed by 3 cycles of 1 min at 95 °C, 2 min 15 s at 55 °C and 1 min 15 s at 72 °C, another 30 cycles of 35 s at 95 °C, 1 min 15 s at 55 °C and 1 min 15 s at 72 °C, and a final elongation for 7 min at 72 °C. The presence of amplicons was verified via agarose (1 %) gel electrophoresis with SmartLadder (Eurogentec) as molecular size marker and EtBr staining.
If the OD320/260 ratio was higher than 0.15 and/or no universal 16S rRNA amplicon could be obtained, an extra purification step using agarose plugs was performed to remove humic acids and other PCR inhibiting contaminants [23]. After purification using plugs, DNA extracts were diluted 5× in TE buffer. Only DNA extracts for which a universal 16S rRNA amplicon could be obtained, were subjected to the B. multivorans PCR assay.
Preliminary experiment
A preliminary experiment compared the specificity of the B. multivorans-specific primers that were based on the recA and 16S rRNA gene and were available from previous studies [15, 16]. Sampling of two water samples, DNA extraction and quality assessment of the DNA extracts was performed as described above. For the nested B. multivorans 16S rRNA PCR assay, PCR products of the first, universal 16S rRNA PCR were used as template in a second PCR in which B. multivorans-specific primers BC-GII (5′- AGG CGG TCT GTT AAG ACA -3′) and BC-R (5′- AGC ACT CCC GAA TCT CTT -3′) were used [15]. The second PCR mix was identical to the first PCR mix, except for a lower BSA concentration (50 ng/μl). For each sample, 2 μl PCR product of the first PCR was added to 23 μl PCR master mix. The positive (B. multivorans R-20526) and negative (blank) control of the first PCR (5 μl) were also transferred as template into the second PCR. The thermal cycling program was identical to that of the first PCR. The presence of amplicons was verified via agarose (1 %) gel electrophoresis with SmartLadder (Eurogentec) as molecular size marker and EtBr staining. If an amplicon (445 bp) was visible in the second PCR, it was sequenced using the BC-GII and BC-R primers as described previously [35] to exclude false positive results. The recA based B. multivorans PCR assay was performed as described below.
Nested recA PCR assay for B. multivorans
In a first PCR, Bcc-specific recA primers recA-01-F (5′-GAT AGC AAG AAG GGC TCC-3′) and recA-02-R (5′-CTC TTC TTC GTC CAT CGC CTC-3′) were used [36]. The PCR mix contained 1× CorelLoad PCR buffer (Qiagen), 0.25 mM dNTP (Applied Biosystems), 1 U Taq (Qiagen), 0.5 μM of both primers, 1× Q-solution (Qiagen) and 200 ng/μl BSA (Roche). For each sample, 2 μl DNA was added to 23 μl PCR master mix. B. multivorans R-20526 DNA and sterile MQ were used as positive and negative control, respectively. This PCR reaction was setup in duplicate for each sample to enable pooling and to increase sensitivity. PCR was performed using a MJ Research PTC-100 thermal cycler. Initial denaturation for 2 min at 94 °C was followed by 35 cycles of 30 s at 94 °C, 45 s at 58 °C and 1 min at 72 °C, and a final elongation for 10 min at 72 °C. PCR product of the duplicate reactions for each sample were pooled using filter tips and used as template in the second PCR, in which B. multivorans-specific primers BCRBM1 (5′-CGG CGT CAA CGT GCC GGA T-3′) and BCRBM2 (5′-TCC ATC GCC TCG GCT TCG T-3′) were used [16]. The second PCR mix was identical to the first PCR mix, except for a lower BSA concentration (50 ng/μl) and a higher primer concentration (1 μM). For each sample, 5 μl of pooled PCR product from the first PCR was added to 20 μl PCR master mix. The positive (B. multivorans R-20526) and negative (blank) control of the first PCR (5 μl) were also transferred as template into the second PCR. Thermal cycling was identical to the first PCR, except that the annealing temperature was 64 °C instead of 58 °C [21]. The presence of amplicons was verified via agarose (1 %) gel electrophoresis with SmartLadder (Eurogentec) as molecular size marker and EtBr staining. The nested recA PCR assay was performed twice for each sample. If the results for the two runs were not the same, the assay was performed a third time. If an amplicon (714 bp) was visible in the second B. multivorans-specific PCR, it was sequenced using the BCRBM1 and BCRBM2 primers as described previously [35] to exclude false positive results. Only if at least for one of the replicate runs an amplicon from the second PCR was sequenced that showed at least 97 % similarity to the recA sequence of B. multivorans ATCC 17616, the sample was considered a true positive for the detection of B. multivorans.
To determine the detection limit of this PCR assay, it was applied on serial dilutions of genomic DNA from B. multivorans R-20526, a strain for which whole-genome sequencing data is available (BioProject PRJNA234537). The mean weight for an AT and GC base pair is 615.3830 Da and 616.3711 Da, respectively [37]. Given the GC content of this genome of about 68 %, and ignoring the presence of modified nucleotides, the mean relative weight of one base pair of R-20526 is 616.0549 Da or 1.023 × 10−9 pg (1 Da = 1.660539 × 10−24 g). Given the genome size of 6.5 Mb, one genome of R-20526 contains 6.65 × 10−6 ng of DNA, or 1 ng of DNA contains 1.50 × 105 genome equivalents. The undiluted genomic DNA stock of strain R-20526 contained 225 ng/μl DNA, as measured with the Promega QuantiFluor ONE dsDNA system, or 3.38 × 107 genomic DNA equivalents per μl. The PCR assay was first applied on a 50× dilution series (in TE buffer) to find the approximate fading range. The highest 50× dilution for which an amplicon could be obtained was then used to make a twofold dilution series to find the limit of detection. The PCR assay was performed three times for each dilution and the detection limit was defined as the highest twofold dilution that tested positive in all three runs. Considering the recA gene is a single-copy gene for B. multivorans R-20526, the detection limit was calculated from the measured DNA concentration and calculated DNA content.
Statistical data analysis
Statistical data analysis was performed using R version 3.1.2 in RStudio (version 0.98.1091), with the following packages: car, ggplot2, and MASS. Binomial logistic regression was used to test which variables (type, region, class, pH, temperature) were significant predictor variables for the outcome variable, i.e., B. multivorans detection (Bm). Type, class, region, and Bm were coded as factors, and the level with the most cases was chosen as reference category. Temperature and pH were coded as numeric variables, and centered around the mean to reduce standard error (SE). Backward stepwise model selection was applied to select the best fitting models.