To develop a qPCR approach for the detection of Pseudomonas aeruginosa in soil and manure and explore its efficacy and limitations compared with that of a classical culture-dependent approach.
To develop a qPCR approach for the detection of Pseudomonas aeruginosa in soil and manure and explore its efficacy and limitations compared with that of a classical culture-dependent approach.
A Ps. aeruginosa ecfX qPCR assay was developed. This assay was optimized for soils of contrasting physico-chemical properties and evidenced a three-log dynamic range of detection [5 × 104 – 5 × 106 cells (g drywt soil)−1] in inoculated microcosms. Sensitivity was determined to be around 5 × 104 cells (g drywt soil)−1. In parallel, the minimum detection limit was estimated in the range of 10–100 CFU (g drywt soil)−1 using a culture-dependent approach based on the use of a selective medium (cetrimide agar base medium supplemented with nalidixic acid), coupled to ecfX gene amplification to confirm isolate identity. These soil samples led to the growth of abundant non-Ps. aeruginosa colonies mainly belonging to other Pseudomonas species but also some beta-Proteobacteria. These bacteria strongly impacted the detection threshold of this approach. Efficacy of these approaches was compared for Ps. aeruginosa enumeration among manure and agricultural soil samples from various sites in France, Tunisia and Burkina Faso.
The developed qPCR assay enabled a specific detection of Ps. aeruginosa in soil and manure samples. The culture-based approach was usually found more sensitive than the qPCR assay. However, abundance of non-Ps. aeruginosa species among the indigenous communities able to grow on the selective medium affected the sensitivity of this latter approach.
This study describes the first specific and sensitive qPCR assay for the detection and enumeration of Ps. aeruginosa in soil and manure and shows its complementarity with a culture-based approach.
Pseudomonas aeruginosa is an opportunistic pathogen frequently responsible for nosocomial infections. It is the main pathogen associated with respiratory tract infection in cystic fibrosis patients (Gilligan 1991; Nazaret et al. 2009), and it can cause a wide variety of infections among compromised hosts (Richard et al. 1994). Infections in healthy individuals can also occur as keratitis (Song et al. 2000), otitis (Heslop and Ovesen 2006) and others. Pseudomonas aeruginosa can be detected in human and animal faecal samples (Lavenir et al. 2008) and has been identified as an animal pathogen responsible for ocular infections in dogs (Ledbetter et al. 2007) and as an occasional cause of bovine mastitis (Daly et al. 1999). It is a metabolically versatile Gram-negative bacterium described as a ubiquitous micro-organism able to colonize a wide variety of environments. It has been frequently isolated from water sources including rivers (Pirnay et al. 2005), sea water (Kimata et al. 2004), bottled and tap waters (Hunter 1993) and wastewaters (Filali et al. 2000). Its isolation from ornamental plants (Cho et al. 1975) or vegetables (Wright et al. 1976) as well as its detection in hydrocarbon-contaminated environments (Kaszab et al. 2010) or agricultural lands (Green et al. 1974; Marques et al. 1979) have also been reported.
Detection of Ps. aeruginosa within clinical samples mostly relies on the use of culture-dependent methods since they allow antibiotic susceptibility testing and help defining proper treatments. These methods involve morphological and biochemical tests after isolation and culture of single colonies grown on selective media. In the context of low levels of Ps. aeruginosa cells, an enrichment step can be included into the isolation procedure (Green et al. 1974). However, the culture-dependent approach may not be appropriate for microbiologically complex environmental samples such as soil due to the presence of a large species diversity that might result in the growth of other species even on selective agar. The need for colony identification using biochemical and/or genotypic typing makes this approach lengthy for high-throughput ecological studies that require the investigation of a large number of samples dispatched on wide temporal and/or spatial scales.
Advances in molecular assays, such as qPCR, have become important tools for detecting more rapidly and accurately bacterial pathogens in various environments (i.e. Karpowicz et al. 2010; Ma et al. 2011; Pontiroli et al. 2011; Rogers et al. 2011). qPCR has been used to detect and quantify Ps. aeruginosa in clinical DNAs extracted from sputum samples (Xu et al. 2004), wound biopsy samples (Pirnay et al. 2000) or blood cultures (Jaffe et al. 2001) using the oprL gene. This approach has also been used to detect Ps. aeruginosa from various wastewater systems, targeting the 23S rDNA sequence (Schwartz et al. 2006) or the regA gene (Shannon et al. 2007). To our knowledge, such approaches were never applied in the context of studies aiming at monitoring Ps. aeruginosa in terrestrial samples and their performances for a quick and sensitive detection of Ps. aeruginosa in soil samples were not yet investigated and validated.
The objectives of this study were to develop a quantitative PCR (qPCR) assay for Ps. aeruginosa detection in soil and manure and to evaluate its complementarity to a classical culture-dependent method. A SYBR Green I qPCR assay based on the use of the ecfX gene as a species-specific target was developed. The ecfX gene was chosen since it was previously identified as a reliable genetic marker of Ps. aeruginosa (Lavenir et al. 2007; Anuj et al. 2009; Hillenbrand et al. 2011). As a first step, we confirmed primer specificity on a collection of strains of Pseudomonas spp. and of Ps. aeruginosa from clinical and environmental sources. The qPCR assay was then optimized for soil analysis and applied on soil samples inoculated or not with Ps. aeruginosa to estimate the detection limit. In parallel, we evaluated the specificity and sensitivity of a culture-dependent approach using a nalidixic acid-supplemented cetrimide agar base (CAB) as a selective medium, coupled to ecfX gene amplification to confirm isolate identity, for the detection and enumeration of Ps. aeruginosa on soil and manure samples. Both approaches were further applied on samples collected from various agricultural sites in France, Tunisia and Burkina Faso.
A set of bacterial strains belonging to Ps. aeruginosa and other Pseudomonas species (Table 1) was used to optimize the design of the qPCR primers and ensure their specificity for the detection of Ps. aeruginosa isolates only and their universality, that is, the detection of all isolates belonging to that species. Thirty-six bacterial strains belonging to Ps. aeruginosa and 19 other Pseudomonas species were obtained from public international collections. Others were isolated and included in our team collection (‘Bactéries Pathogènes Opportunistes et Environnement’ collection, Villeurbanne, France; available from the EML Biological Resource Center at http://www.eml-brc.org/). Nine clinical strains previously isolated from cystic fibrosis patients (Nazaret et al. 2009), 26 strains collected from snakes (Colinon et al. 2010), 14 strains from domestic waters (Deredjian et al. 2011) were included in this study. The bacterial strains were routinely grown aerobically at 28 or 37°C in Luria-Bertani (LB) or tryptic soy medium. All bacteria were stored at −80°C in LB broth containing 20% glycerol.
7NSK2a, CFBP5036, CFBP5037, DSMZ6195, CFBP5036, CFBP5037, ATCC21776, ATCC31479, CIP104590, LMD5034, LMD68.7, LMG15153
From natural waters
From domestic waters
bpoe1026, bpoe1028, bpoe1033, bpoe1045, bpoe1062, bpoe1081, bpoe1091, bpoe1108, bpoe1150, bpoe1302, bpoe1312, bpoe1314, bpoe1372, bpoe1436
ATCC14425, LMG1272, CFBP2466, CFBP5031, CFBP5032, CFBP5033, CFBP5034, CFBP5035, LMG1272, LMG5031, LMG5032, LMG5033, LMG6855
bpoe600, bpoe602, bpoe603, bpoe605, bpoe610, bpoe612, bpoe616, bpoe618, bpoe626, bpoe636, bpoe639, bpoe642, bpoe649, bpoe652, bpoe654, bpoe657, bpoe662, bpoe667, bpoe674, bpoe688, bpoe690, bpoe692, bpoe700, bpoe701, bpoe742, bpoe753
CFBP5035, CFBP2466f, LMG15153, ATCC27014
|Pseudomonas agaraci CFBP2063|
|Pseudomonas alcaligenes CFBP2437f|
|Pseudomonas asplenii CFBP3279|
|Pseudomonas balearica CIP105297f|
|Pseudomonas chlororaphis CFBP2132f|
|Pseudomonas cichorii CFBP2101|
|Pseudomonas citronellolis CIP104381f, CFBP5585|
|Pseudomonas flavescens CIP104204f|
|Pseudomonas fluorescens CFBP2102e|
|Pseudomonas fragi CFBP4556f|
|Pseudomonas oleovorans CIP59.11|
|Pseudomonas pseudoalcaligenes CFBP2435f|
|Pseudomonas putida CFBP2066|
|Pseudomonas resinovorans CIP61.9|
|Pseudomonas stutzeri CFBP2443e|
|Pseudomonas syringae pathovar pisi BD13-2|
|Pseudomonas tolaasii CFBP2068|
|Pseudomonas viridiflava CFBP2107f|
Development and optimization of the qPCR approach were performed with one agricultural soil from Feucherolles, Ile-de France [experimental site of INRA (Institut National de Recherche Agronomique) of Versailles] and nine soils of the French soil quality monitoring network [Réseau de Mesures de la Qualtité des Sols (RMQS)] (INFOSOL, INRA of Orléans) (Arrouays et al. 2002). The nine RMQS soils were chosen based on their contrasting structural and physico-chemical properties (Table 2) that were expected to have an influence on soil DNA quantity and purity and then on qPCR sensitivity. These soils did not exhibit Ps. aeruginosa based on the culture approach described below. Comparisons between the culture approach and the developed qPCR approach were further performed on soils sampled from various agricultural lands in France including Burgundy (12 fields from 12 sites), Ile-de-France (five fields from four sites) and Rhône-Alpes (one field from one site) regions. To increase the diversity of soil types, we included samples from two Mediterranean fields planted with orange tree in Tunisia (INRGREF for Institut National de Recherche en Génie Rural Eaux et Forêts, Nabeul, Tunisia), and Sahelian fields planted with sorghum from three sites (Tabtenga, Toudoubwéogo and Yagma) in the vicinity of Ouagadougou, Burkina Faso. In each, field samples, that is, five to 10 sampling per field composed one sample, were collected from the upper layer (0–5, 0–10 cm or 0–20 cm), sieved (2 mm) and stored at room temperature for no longer than 2 weeks. They were collected during various campaigns between 2006 and 2011.
|Site||Clay %||Silt %||Sand %||Total carbon %||Total nitrogen %||pH H2O||Land use|
|Réseau de Mesures de la Qualtité des Sols soils|
Farmyard manure and wastewater samples known to harbour or not Ps. aeruginosa were added in the study. The wastewater samples were collected in March 2006 from a wastewater treatment lagoon (WWTL) in Montracol (Rhône-Alpes, France). Samples were collected from the entry of the lagoon (one sample), several surface points at the surface of the first pond (10 samples) and exit of the lagoon (one sample). Two freshwater samples, collected downstream the nearby river, were added. Bovine manure samples were obtained from a farm in Feucherolles (Ile-de-France) in 2006 and 2007, and a farm in Versailleux (Rhône-Alpes). Horse manure was sampled in a farm at Saint Olive (Rhône-Alpes). About 3 kg was sampled and homogenized manually before been used for bacterial counts and DNA extraction.
The strains' genomic DNA was extracted by gentle alkaline lysis (using sarkosyl and NaCl) and purified according to Johnson (1994). For water samples, 75 ml (influent and pond), 150 ml (lagoon) and 225 ml (river) were filtered using 0·22-μm HA membrane filters (Millipore, Molsheim, France). Filters were then ground in liquid nitrogen, and DNA was extracted as described previously (Ranjard et al. 2000). DNA extraction and purification from soil samples of Burkina Faso and France sites except those from the RMQS library were performed using the FastDNA® SPIN Kit for Soil (MP Biomedicals, Solon, OH, USA) and the S-400-HR mini-columns, respectively (Pharmacia, St Quentin Yvelines, France), following the manufacturer's instructions. DNAs from soils of the RMQS library, Tunisian sites and spiked microcosms were extracted by the GenoSol platform according to a single procedure optimized by Ranjard et al. (2003). All soil samples were extracted in triplicates. The DNA extracts were analysed by electrophoresis in 0·8% agarose gels, stained with ethidium bromide and photographed using a Gel Doc 1000 camera (Bio-Rad, Ivry sur Seine, France). DNA concentrations in the crude and purified soil extracts were determined by electrophoretic comparison with a standard curve as previously described (Ranjard et al. 2003). The strains' genomic DNA concentration was estimated using a Nanodrop® ND-1000 spectrophotometer (Labtech International, Paris, France) at a 260 nm wavelength.
Pseudomonas aeruginosa-specific qPCR was set up using the ECF5 (5′-AAGCGTTCGTCCTGCACAA-3′) and ECF2 (5′-TCATCCTTCGCCTCCCTG-3′) primers. These primers amplify a 146-bp-long fragment of the ecfX gene. The Eva Green SMX 1000R (Bio-Rad), which is recommended for high GC sequences, was used, according to the manufacturer's instructions. The reactions were carried out in a 20 μl reaction mix containing 500 nmol l−1 of each primer. Five nanograms of soil DNA, or 100 pg of strain genomic DNA was taken as templates (5 μl each per reaction). Nontemplate controls, including the reaction mixture with sterile water instead of DNA template, were added to each run. Quantitative PCR was performed in a LightCycler 480 system (Roche Diagnostics, Meylan, France). The following PCR protocol was applied: initial denaturation at 95°C for 5 min, followed by 50 cycles with denaturation at 98°C for 10 s, annealing and elongation at 63°C for 20 s. Subsequently, a melting curve was recorded by increasing the temperature from 65 to 98°C (+1°C every 10 s). For reproducibility, all qPCRs were triplicated on independent runs. To determine the detection sensitivity of the qPCR, quantification was performed by comparison with a two- to five-fold diluted standard of genomic DNA (1 fg to 100 pg, 5 μl per reaction) from UCBPP-PA14 or PAO1 strains. Data analysis was performed using the LightCycler® 480 software (Roche Diagnostics). PCR product specificity was checked by melting curve analysis and agarose gel electrophoresis. Real-time PCR data were expressed in genome-equivalent (GE) or cell (one cell corresponding to one GE), and as the means of triplicate determinations and one standard deviation. To mimic the complexity of the soil's DNA sequences in terms of its diversity and evaluate its influence on the efficiency and sensitivity of our PCR, we added 50 or 10 ng of complex DNA (calf thymus DNA or soil DNA in which no Ps. aeruginosa had been detected from culturing data, that is, soil DNA from the Feucherolles site) to the standard curve.
To confirm the amplification of target genes corresponding to ecfX-like sequences from environmental samples, qPCR products were migrated on agarose gel and the expected ecfX fragments were purified using the QiaexII gel extraction kit (Qiagen, Courtaboeuf, France). The purified products were ligated in a pGEM-T easy vector (Promega, Madison, WI, USA) and used to transform Escherichia coli DH5α (Invitrogen, Carlsbad, France), according to the manufacturer's instructions. Correct insertion into the plasmid was verified by PCR amplification using the vector-specific M13 primers (M13-R: 5′- CAGGAAACAGCTATGAC-3′; M13-F: 5′-GTAAAACGACGGCCAG-3′). Amplified fragments with M13 primers (containing the insert) were sequenced with Sp6/T7 primers (Sp6: 5′- ATTTAGGTGACACTATAGAA-3′; T7: 5′- TAATACGAC-TCACTATAGGG-3′). The identity of the amplified fragments was confirmed by comparing their nucleotide sequences to the GenBank database using the BLASTN program (http://blast.ncbi.nlm.nih.gov/).
To check for PCR inhibitors in the soil DNAs and evaluate the most appropriate concentration of DNA templates, we performed an assay based on the detection of a plasmid among the RMQS DNAs. This assay involved adding 106 copies of the circularized pGEM-T Easy plasmid DNA (2·5 μl per 20 μl of PCR) to various amounts (0·5, 1, 2·5, 5 ng, 12, and 25 ng) of the nine RMQS DNA extracts added. The plasmid DNA was quantified afterwards by comparison with a 10-fold diluted standard of pGEM-T Easy plasmid DNA (101–107 copies, 2·5 μl each per reaction). qPCRs and cycling conditions were as stated previously, except for the primers that were the SP6 and T7 universal primers. The annealing temperature was set at 55°C. The PCR kit LightCycler® 480 SYBR Green I Master kit (Roche Diagnostics) was used. PCR product specificity was checked by melting curve analysis. The prevention of inhibiting effects was tested by adding different amounts of T4 gene 32 protein (25, 50, 75 or 100 ng μl−1; Roche Diagnostics) in the qPCR (Tebbe and Vahjen 1993).
To determine the detection limit of the qPCR assay in the soil samples, microcosms were set up with 1·5 g of sieved soil (dry weight) from the set of nine RMQS soils mentioned earlier. Each soil was inoculated with six different concentrations, that is, 5 × 101 to 5 × 106 CFU of Ps. aeruginosa strain PAO1 per g of soil (dry weight). Concentrations of 5 × 103 and 5 × 104 were inoculated twice. Briefly, a bacterial suspension for inoculation was prepared in cold saline solution (0·8% NaCl) from a 10 ml fresh culture and washed three times in saline solution. The bacterial density of the inoculum was estimated by optical density measurement at 600 nm and controlled by enumeration on tryptic soy agar plates. A volume of 100 μl of serially diluted bacterial suspensions was spread over the surface of each microcosm. The same procedure was applied to the control using bacterium-free saline solution. The microcosms were incubated at room temperature for 24 h and immediately processed for DNA extraction as described previously. All qPCRs were triplicated on independent assays and performed as described previously.
Pseudomonas aeruginosa enumeration was performed using CAB medium (Oxoïd, Cambridge, UK) supplemented with nalidixic acid (15 mg l−1) and cycloheximide (200 mg l−1). Cell recovery from water, soil and manure samples was as described previously (Lavenir et al. 2007). Three plates were inoculated per dilution and incubated at 28°C for up to 3 days for CAB plates. One hundred microlitres was usually spread per 90-mm diameter plate. To improve Ps. aeruginosa detection sensitivity, 1 ml of suspension was used with the soil samples from Burkina Faso as these soils contained high quantities of sand, low quantities of organic matter, and therefore, yielded clear soil suspensions. In addition, enrichment assays were performed by transferring 2 g of soil into 20 ml of a salt solution supplemented with acetamide, as described previously (Green et al. 1974). Inoculated enrichment broths were incubated for 3 days at 28°C with shaking at 180 rev min−1. Tenfold serial dilutions were performed and plated on CAB agar medium. Counts from Burgundy soils were performed on one sample, whereas counts from other soils and manures were performed at least on three samples.
All the greenish and clearly yellowish fluorescent colonies were collected from CAB plates and confirmed as Ps. aeruginosa by PCR screening with the ecfX gene encoding an extracytoplasmic function (ECF) sigma factor, as previously described (Lavenir et al. 2007) and by an oxidase test. All ecfX-positive isolates were analysed using biochemical tests (VITEK2; bioMérieux, Marcy l'Etoile, France) for taxonomic identification. When VITEK2 test gave excellent identification of Ps. aeruginosa, we selected 2–5 isolates per sample for further 16S rDNA sequencing (up to 1500 bp) to confirm their identification. Depending on the abundance of colonies on CAB plates up to 10 colonies that were slightly fluorescent under UV light and up to 10 truly nonfluorescent colonies were randomly chosen and further tested for the presence of the ecfX gene. All colonies from the Tunisian soil samples were submitted to VITEK2 for identification. We selected, at random, colonies from manure and soil samples from four sites in France (Feucherolles, Pierrelaye, Brienon and Versailleux), and Burkina Faso for identification based on 16S rDNA sequencing. All BLAST analyses were run at NCBI (http://www.ncbi.nlm.nih.gov). The sequences of the identified bacterial isolates were deposited in the GenBank nucleotide sequence database under Accession Numbers KC195874 to KC195911.
Various sets of primers were designed based on the sequence of the ecfX gene using Oligo 6.65 software (data not shown). In the qPCR assays that yielded a positive signal for all strains of the Ps. aeruginosa panel, one set of primers was found highly sensitive, that is, annealed at the expected Tm and showed only one fragment of the expected size after agarose gel electrophoresis and also showed a good specificity, that is, gave no signal for the 18 strains representative of other well-characterized Pseudomonas species (data not shown). The Tm of the ecfX target of the various Ps. aeruginosa strains was found to vary between 89·5 and 90°C. DNA from five-fold dilutions of quantified PA14 or PAO1 Ps. aeruginosa strains were analysed to determine reaction efficiency. The standard curves had a linear range of quantification from 5000 pg to 50 fg. Down to 50 fg, we observed a very good reproducibility but detection was not always possible with 10 fg. The detection limit was then 10–50 fg of genomic DNA, which corresponds to approximately 1·4–7 Ps. aeruginosa GEs based on a PAO1 genome size of 6·26 Mb and a GC content of 66·6%. Adding 50 or 10 ng of calf thymus DNA or soil DNA did not change the sensitivity: the ecfX target was still detected in standard curve tubes containing 50 fg of genomic PAO1 DNA. However, several nonspecific peaks (with unexpected melting temperatures) or bands (of unexpected sizes) were observed at the lowest concentrations of the standard curve for calf thymus DNA (Fig. 1c) and at all concentrations for soil DNA (Fig. 1d).
We evaluated the quality of genomic DNA with respect to coisolated PCR inhibitors by adding about 106 copies of the pGEM-T Easy plasmid to the qPCR together with 0·5–5 ng of DNA extracts from the nine RMQS soils. Figure 2 reports on the number of pGEM-T Easy copies detected in the various qPCR conditions. The DNA extracts from soils 1, 3, and 5 were found the most inhibiting for qPCR amplification. No plasmid copies or about 104 out of 106 added targets were detected in the presence of either 5 or 2·5 ng of these soil DNAs (total inhibition or >98% inhibition) (Fig. 2a). In contrast, DNA extracts from soils 4, 6, and 7 were found the less inhibiting ones of the qPCR amplification (detection of about 106 plasmid copies even in the presence of 5 ng of DNA) followed by soil DNAs 8 and 9 (12–77% inhibition with 5 ng of DNA, and 0–6% inhibition with 1 ng). Soil DNA 2 showed a moderate inhibiting effect. No relationship was observed between the level of inhibition and the physico-chemical soil characteristics. To remove inhibition, the influence of adding T4 gene 32 protein to the qPCR was assayed with 1–5 ng of DNA from the most inhibiting extracts (3 and 5) (Fig. 2b). A concentration of T4 gene 32 protein as low as 25 ng μl−1 was shown appropriate to enable an efficient and sensitive amplification (Fig. 2c) even with 5 ng of DNA. In these conditions, the numbers of amplified targets was about 106 plasmid copies. We tested adding higher amounts of soil DNA, that is, 12 or 25 ng of DNA per reaction mixture but we could not achieved an optimal detection of the target for all soil types whatever the amount of added T4 gene 32 protein (data not shown).
The ability to quantify Ps. aeruginosa in soil samples was tested by spiking soil microcosms with known concentrations of Ps. aeruginosa PAO1 cells (Fig. 3). The assay was performed on the nine RMQS soils using the previously optimized conditions (5 ng of DNA per reaction and 25 ng μl−1 of T4 gene 32 protein). We previously tested the DNA from these nine soils in the absence of inoculation and evidenced the lack of detection of Ps. aeruginosa using these qPCR conditions, as also shown from the culture-dependent method (data not shown). Pseudomonas aeruginosa was always detected in microcosms inoculated with 5 × 106, and 5 × 105 cells (g drywt soil)−1, whatever the soil type. In microcosms inoculated with 5 × 104 cells (g drywt soil)−1, detection was achieved in the nine soils, but the results varied, some replicates sometimes yielding no detection. The percentages of absolute cell numbers detected for all the spiked microcosms of a soil type varied from 6 to 31%. The lowest recoveries were from soil 1. Detection consistently failed in microcosms inoculated with 5 × 101, 5 × 102 and 5 × 103 cells (g drywt soil)−1 whatever the soil type.
It should be noted that inhibition as assessed by adding pGEM-T Easy plasmid DNA and tested on DNAs extracted from inoculated microcosms was found higher than the inhibition observed with the DNA extracted from the same soil samples provided by the GenoSol platform (data not shown) and that variations were found between the various spiked microcosms of a soil type. This observation and variability in DNA extraction efficiency could explain the discrepancies between the number of inoculated cells and the estimated number based on qPCR detection as well as the variability of the results observed between soils.
The culture-dependent approach with the use of the CAB medium supplemented with nalidixic acid was applied on bulk soil samples taken from various agricultural fields in France (43 samples from 18 sites), Tunisia (15 samples from two sites) and Burkina Faso (15 samples from three sites). Eight samples from farmyard manure were also studied. We observed that the numbers of colonies able to grow on the CAB media strongly differ between samples (Table 3). The highest amount was found from manure sample and some soil samples, that is, Gilly sur Loire, La Guiche and Salornay, with levels of 8 × 105 CFU (g drywt)−1 and about 2 × 105 CFU (g drywt)−1 for manure and soils, respectively. On the opposite, some soil samples never led to colony development whatever the diluted suspension, that is, samples from Burgundy (Cudot and Palinges sites) and Ile-de-France (Chavenay and Fontenay le Fleury sites), or to a very low level of colonies (sites from Burkina Faso).
|Country, region and site||Number of treated samples||Description at sampling time||Total colonies on CAB CFU × 103 (g drywt soil)−1 (±SD)|
|Manure samples from France|
|Feucherolles – 2006||3||Bovine manure||29 (±3·5)|
|Feucherolles 2007||3||Bovine manure||9·2 (±1·0)|
|Versailleux||3||Bovine manure||87 (±38)|
|Saint Olive||3||Horse manure||800 (±101)|
|Soil samples from France|
|Brienon sur Armançon||1||Maize||76 (±22)|
|Merry sur Yonne||1||Cereal||14 (±6·0)|
|L'Hôpital Le Mercier||1||Grassland||21 (±20)|
|Gilly sur Loire||1||Grassland||200 (±90)|
|Rigny sur Arroux||1||Grassland||25 (±17)|
|La Guiche||1||Grassland||120 (±18)|
|Salornay sur Guye||1||Fallow||100 (±39)|
|Pierrelaye 1||10||Maize||27 (±4·2)|
|Pierrelaye 2||10||Miscanthus||47 (±5·7)|
|Fontenay le Fleury||3||Wheat||0|
|Soil samples from Burkina Faso|
|Soil samples from Tunisia|
|Nabeul 1||9||Orange tree||22 (±5·4)|
|Nabeul 2||6||Orange tree||15 (±2·7)|
All the greenish fluorescent colonies and the yellowish fluorescent colonies recovered from the plates were identified as Ps. aeruginosa-like isolates based on ecfX-specific amplification and oxydase test. Forty-eight of these fluorescent, oxydase-positive and ecfX-positive colonies were then selected from the soil samples from Yagma site and from the manure samples and analysed on the VITEK2. They all showed an excellent identification as Ps. aeruginosa (99% probability). Sequencing about 1000 bp of the 16S rDNA of 23 strains from soil (five from Yagma) and manure (five, six and seven from Versailleux, Saint Olive and Feucherolles sites, respectively) samples consistently confirmed these isolates as belonging to the Ps. aeruginosa species (100% homology).
About 350 colonies were selected from the plates spread with soil and manure suspensions among slightly fluorescent colonies (i.e. colonies that were slightly fluorescent under UV light), and nonfluorescent colonies as some natural Ps. aeruginosa isolates had been reported to be unpigmented (Pirnay et al. 2005). These colonies never led to a positive signal with ecfX amplification. Based on VITEK2, colonies from the Tunisian sites led to the identification of Pseudomonas putida (99·7% of confidence) for 28 isolates, and Pseudomonas fluorescens (99·7% of confidence), for 12 isolates. Isolates from the other sites were also identified as belonging to the Pseudomonas genus (Ps. putida, Ps. fluorescens, Ps. psychrotolerans, Ps. poae, Ps. chlororaphis, Ps. migulae and Pseudomonas sp.) based on 16S rDNA sequencing (Table 4). The dominant species among the isolates from French soil samples were Pseudomonas sp., Ps. putida and Ps. fluorescens that represented 43, 16 and six of the 69 sequences, respectively. In soil samples from Burkina Faso, most colonies on CAB media did not belong to the Pseudomonas genus and were identified as Burkholderia species, Ralstonia sp., Achromobacter xylosoxidans, Bordetella sp. and Bordetella avium. Only three of the 25 sequences were identified as Ps. putida.
|Strain origin||Number of isolates with similar sequences||Size (bp)a||Identification based on sequence producing significant alignment with the maximum score||Accession number||Maximum identity %|
|Soil from France|
|Brienon sur Armançon||6||1059||Pseudomonas sp.||EU681021-1||99|
|Soil from Burkina Faso||1||1528||Ps. putida||JQ012750-1||99|
|2||1518||Ps. putida b||CP002290-1||99|
|4||1449||Burkholderia cepacia b||EU597838-1||99|
|3||1540||Burkholderia multivorans b||AP009387-1||98|
In manure samples, isolates were also found to belong to Pseudomonas genus with 15 isolates of 25 closely related to Pseudomonas sp., and 10 closely related to Ps. putida.
Then, the detection limit using cultivation-based measurements varied between samples. For instance, we estimated the detection limit of soil samples from the Burkina Faso to be about 10 CFU (g drywt soil)−1 since we were able to spread 1 ml of the undiluted soil suspension on plate and that few non-Ps. aeruginosa colonies grew on that media. On the opposite, the highest estimated detection limit which was observed for the manure samples and some Burgundy soils had a value of about 105–106 CFU (g drywt sample)−1. To lower the detection limit of Ps. aeruginosa from environmental samples, an enrichment step can be performed before plating on CAB media. However, we also evidenced with soil samples from Burkina Faso the growth of non-Ps. aeruginosa isolates belonging to Pseudomonas, Burkholderia and Ralstonia genera (Table 4).
We first evaluated the qPCR efficiency of the ecfX gene target on water samples known to harbour or not Ps. aeruginosa cells based on culture detection (Lavenir et al. 2007). The qPCR amplifying the ecfX target led to a negative signal with the samples from the river and water at the exit of the lagoon (Table 5). These observations confirmed the lack of detection of Ps. aeruginosa in these samples as shown previously by the culture approach. On the opposite, we were able to detect Ps. aeruginosa at level of 0·419 (±0·162) × 103 CFU ml−1 and 0·130 (±0·056) × 103 CFU ml−1 of water in samples from the entry influent and the surface of the pond, respectively. These numbers were in the range of these found with the culture approach (Table 5). Cloning and sequencing confirmed that the amplified target from the pond were ecfX-like gene (data not shown). The 20 sequences obtained showed 100% identity with the ecfX sequences from the genomes of PAO1, PA14, M18, LESB58, NCGM2·51 and DK2, while they showed 96·6% identity (five mismatches over 146 bp) with the ecfX sequence from the genome of PA14.
|Samples||CFU × 103 ml−1 or g−1 (±SD)||Target × 103 ml−1 or g−1 (±SD)|
|Entry influent||0·312 (±0·058)||0·419 (±0·162)|
|Surface pond 1||0·064 (±0·0081)||0·130 (±0·056)|
|Feucherolles 2006||15 (±2·2)||23 (±3·5)|
|Feucherolles 2007||5·4 (±1·4)||76 (±15)|
|Saint Olive||5·5 (±1·1)||nd|
|Brienon sur Armançon||0a||0|
|Merry sur Yonne||0a||0|
|L'Hôpital Le Mercier||0a||0|
|Gilly sur Loire||0a||0|
|Rigny sur Arroux||0a||0|
|Salornay sur Guye||0a||0|
|Fontenay le Fleury||0a||0|
To ensure that lack of detection within soil DNA was not due to PCR inhibition, we previously performed the inhibitory test by adding 106 copies of the pGEM-T Easy plasmid to the reactions with 5 ng of DNA extracts from each soil or manure samples. No significant inhibition was observed since samples showed at least detection of 95% of the targets whether the DNA was extracted with the GenoSol procedure or the FastDNA® SPIN Kit for Soil (MP Biomedicals, Solon, OH, USA) and the S-400-HR mini-columns (data not shown).
Considering ecfX target detection, none of the DNA samples from agricultural soil samples emitted a noteworthy positive fluorescence signal. The results of the culture-dependent approach confirmed the absence of Ps. aeruginosa even after an enrichment step in all agricultural soils of France and Tunisia (Table 5). Only one of the three sites of Burkina Faso, that is, the Yagma site, led to the isolation of Ps. aeruginosa. This successful isolation was achieved after an enrichment step. We also included samples from farmyard manure and evidenced the presence of Ps. aeruginosa in samples from bovine manure and horse manure. Bovine manure from the samples collected at the Feucherolles farm showed 15 (±4·2) × 103 CFU (g drywt manure)−1 and 5·4 (±1·4) × 103 CFU (g drywt manure)−1 of Ps. aeruginosa based on the culture approach and levels of 23 (±3·45) × 103 CFU (g drywt manure)−1 and 76 (±15) × 103 CFU (g drywt manure)−1 of Ps. aeruginosa based on qPCR. It has to be noticed that five replicates of the qPCR were performed for manure samples since some replicates gave negative amplification. Culture approach allowed the detection of Ps. aeruginosa at level of 19 (±3·8) × 103 CFU (g drywt manure)−1 and 5·5 (±1·1) × 103 CFU (g drywt manure)−1 for the bovine manure from Versailleux and the horse manure from Saint Olive, respectively. In these samples, the Ps. aeruginosa cells represented 0·68–59% of the colonies growing on CAB media.
The opportunistic pathogen Ps. aeruginosa is described as a ubiquitous micro-organism colonizing a wide range of habitats. It is frequently detected in aquatic environments and the risk to human health associated with its presence has been evaluated (Mena and Gerba 2009). Pseudomonas aeruginosa has also been isolated from soil, however, its abundance, persistence and dissemination, have been poorly addressed. Pseudomonas aeruginosa can enter soil through human activities as irrigation or organic amendment application for fertilization. Contaminated water or amendment can then contribute to the spread of Ps. aeruginosa in terrestrial environments, which in return can lead to river or underground water contamination through run off or vertical transfer, respectively. To monitor Ps. aeruginosa in soil, efficient detection and quantification methods are then needed. In this study, we investigated how qPCR can facilitate the detection of Ps. aeruginosa in complex environmental samples as soil or manure ones compared with classical culture-dependent approaches.
In contrast to the latter approaches that require confirmation of colonies identity, qPCR is time saving since a well-defined target enables one to specifically detect and quantify a requested species in a mix of complex microbial genomes. However, it is associated with several technical difficulties when used on soil samples. One problem to be faced is the high species diversity, leading to a highly complex DNA extract in terms of sequence diversity that favours false-positive amplification. This problem is emphasized with soil samples since the soil harbours about 10 billion microbes of possibly thousands of different species (Rosselló-Mora and Amann 2001) and a total diversity of species estimated at 103–105 per g soil sample (Torsvik et al. 1990; Curtis et al. 2002). Consequently, a potentially high amount of close relatives to Ps. aeruginosa including also undescribed relative species may be present in our soil samples. It is then necessary to ensure the high specificity of the detection methodology. To our knowledge, the only study dealing with qPCR detection of Ps. aeruginosa in outdoor environments was conducted in wastewater using a TaqMan probe strategy (Schwartz et al. 2006; Shannon et al. 2007). Our choice of a less costly technique than hybridization probe PCR and previous demonstrations of the ecfX gene as a good specific species marker (Lavenir et al. 2007; Anuj et al. 2009; Hillenbrand et al. 2011) led us to develop a SYBR Green I qPCR with the ecfX gene as the gene target. We first tried to use the same primers as those we had previously defined (Lavenir et al. 2007) but failed to achieve good sensitivity PCRs (data not shown). The length of the target and the need to add DMSO due to the high GC content of the target probably accounted for these unsuccessful tests. New primers were then defined with the requirement of a lower fragment size. These primers were used together with the Eva Green kit that is recommended for high GC sequences. The screening of Ps. aeruginosa and non-Ps. aeruginosa strains confirmed the high specificity of the new ecfX primers. PCR sensitivity was also found appropriate as 7 Ps. aeruginosa GEs were successfully detected in a standard curve.
Another limitation of the qPCR is that the target species can be in low copy numbers and diluted in a complex DNA extract that renders it nondetectable. Using calf thymus DNA or soil DNA extract, we showed that an efficient qPCR is more difficult to achieve if the target is diluted in a high amount of DNA background (Fig. 1). We also observed that the lower the number of targets the higher the risk of unspecific amplification. Using a higher amount of total DNA extract in PCRs can increase detection efficiency. However, the presence of potential inhibitors (i.e. humic acids, polyphenols, polysaccharides and metals) co-extracted with the DNA can hamper optimal target detection. In qPCR, humic acid has been demonstrated to inhibit PCRs through sequence-specific DNA-binding, which limits the amount of available template (Opel et al. 2010) and decreases Taq polymerase catalytic activity (Kermekchiev et al. 2009). In our study, tests were then performed on a panel of nine DNA extracts selected from nine French soils chosen for their contrasting structural and physico-chemical properties, expected to influence the DNA extraction efficiency and DNA purity and consequently qPCR efficiency. Using pGEMT plasmid as a target, we showed that three of nine extracts were not pure enough to allow target amplification in the reaction mix with the highest DNA amounts (i.e. 5 ng). Our assays evidenced the benefit of adding T4 gene 32 protein and showed that a concentration of 25 ng μl−1 prevented inhibition with all DNA extracts when added at a level of 5 ng in the reaction mixture (Fig. 2). However, inhibition was not totally removed in the presence of 12 or 25 ng with the most inhibiting sample despite the purification step with PVPP in our protocol. Consequently, we estimated that 5 ng of DNA was appropriate to apply the ecfX qPCR assay to soils showing contrasting inhibition properties. We then evaluated the detection limit of the optimized qPCR with soil samples inoculated with various amount of Ps. aeruginosa cells. Our experiment evidenced that detection up to 5 × 104 cells (g drywt soil)−1 could be achieved whatever soil characteristics (Fig. 3). In agreement with several reports from the literature (i.e. Henry et al. 2006; Travis et al. 2011; Trung et al. 2012), qPCR detection limit is about 104–105 cells (g drywt soil)−1 when the gene used as a target is present in only one copy per genome. Whether the observed detection limit totally reflected real conditions cannot be ensure since DNA extraction efficiency and DNA purity achieved with introduced cells might differ from those obtained with indigenous populations. Surprisingly, we observed a higher inhibition with the DNA extracts from the inoculated microcosm. It could be explained by the re-wetting of soil that led to a higher amount of co-isolated PCR inhibitors in soil DNA as evidenced by the more brownish colour of the microcosm extract even after purification compared with that of the extract from noninoculated soil.
A culture-dependent approach involving the use of selective media, combined to a genotypic analysis, is a powerful tool for detecting Ps. aeruginosa in the context of clinical studies. We then evaluated the potentiality of this approach for assessing its prevalence in soil and manure. Most media designed to isolate Ps. aeruginosa rely on its ability to synthesize the pyocyanin and fluorescein pigments (Brown and Lowbury 1965; Brodsky and Nixon 1973; Green et al. 1974). We used a CAB medium, containing cetrimide, a quaternary ammonium compound that inhibits the growth of other micro-organisms, and magnesium chloride and potassium sulfate that stimulate the synthesis of pigments. This medium was supplemented with nalidixic acid to facilitate Ps. aeruginosa recovery and identification from soil samples (Ringen and Drake 1952). This medium generally shows a high selectivity when used to screen clinical samples. Kodaka and collaborators screened about 1000 urine, pus and sputum samples and 98% of the colonies obtained on this medium were confirmed as Ps. aeruginosa (Kodaka et al. 2003). Conversely, our results highlighted that the same medium was less effective when used to isolate Ps. aeruginosa from soil samples. Many nonfluorescent colonies were observed on the selective agar plates and only the greenish fluorescent colonies and clearly yellowish fluorescent colonies were confirmed as Ps. aeruginosa. The non-Ps. aeruginosa colonies were mostly identified as belonging to the Pseudomonas genus (Table 4). These species as Ps. fluorescens and Ps. putida share with Ps. aeruginosa the characteristic production of pigments that fluoresce under ultraviolet light. They are known for their ability to grow on cetrimide-based media, and this selective agent is used to select for Pseudomonas spp. (Cho and Tiedje 2000). Some are also known for their ability to resist to nalidixic acid thanks to a RND efflux pump (Hearn et al. 2006). Therefore, the isolation of numerous non-Ps. aeruginosa cells did not really come as a surprise as Pseudomonas spp. are among the most abundant species within soil bacterial communities, contributing 1·6% of the cloned sequences from soils and up to 10% among plated colonies (Janssen 2006). In our study, biochemical tests with VITEK2 and 16S rDNA sequencing confirmed that these species, especially Ps. fluorescens and Ps. putida, were frequently encountered on plates whatever the soil origin. Plates from soils collected in Burkina Faso also showed the presence of nonfluorescent colonies that were mostly found related to beta-proteobacterial genera as Ralstonia, Burkholderia, or Bordetella. Some of these genera were previously detected on that medium (Cho and Tiedje 2000; Khan et al. 2007; Kaszab et al. 2010). As these species and Pseudomonas spp. are common indigenous soil populations, their presence at a high level may hamper the detection of Ps. aeruginosa if that species is a minor soil population. Such limited efficiency in isolating Ps. aeruginosa from environmental samples was previously reported for water samples. As an extreme example, Khan et al. (2007) and collaborators screened 1670 colonies on cetrimide-nalidixic acid agar from 17 water samples (open ocean, bay…) and only 62 were confirmed as Ps. aeruginosa. In our study, the detection limit of the culture-dependent approach was found to be about 10–100 CFU (g drywt soil)−1 at best. Our lowest detection limit was achieved with sandy soil with low organic matter (i.e. soil samples collected from Burkina Faso) and a low amount of total heterotrophic bacteria (data not shown). On the other hand, in several soils, we estimated that detection of Ps. aeruginosa cannot be achieved if that species is present at a level below 105 CFU (g drywt soil)−1 since soil suspension has to be diluted due to the presence in high amount of non-Ps. aeruginosa species able to grow on this selective medium. Furthermore, soil suspensions sometimes contained many organic particles that strongly interfered with colony detection on plates and sometimes allowed the growth of cells that grow on particles but cannot grow on selective medium in the second re-isolation step. Our study clearly evidenced that the culture-dependent approach varies in sensitivity and is time-consuming due to the need to confirm isolate identity.
The ecfX qPCR approach was found efficient for the detection of Ps. aeruginosa in water samples and in manure samples. The ranges of abundance were in agreement between both the qPCR and the culture-dependent approaches. Similarly, soil samples that did not lead to Ps. aeruginosa isolation with the culture approach did not lead to positive amplification with the qPCR. However, we were sometimes able to detect Ps. aeruginosa using the selective media, whereas the qPCR did not allow it. For instance, thanks to the enrichment procedure and due to the low amount of non-Ps. aeruginosa species within the indigenous community of soil from Burkina Faso able to grow on the selective media, we evidenced the presence of Ps. aeruginosa based on the culture approach, whereas the qPCR assay could not achieve it. Consequently, in our study, qPCR detection limit, that is, about 5 × 104 cells (g drywt soil)−1, is above the detection limit of the culture approach with CAB medium. Furthermore, qPCR detection limit may vary depending on soil samples due to the difficulties in ensuring equal efficiency in DNA extraction and purification. However, in some instance, the detection limit of the qPCR is as low as the detection limit of the culture approach. We showed that this limit was comparable with the detection limit of the culture approach for manure samples and some soil ones due to the high numbers of Pseudomonas species that were found able to grow on the CAB media. Both methods then showed complementarity to monitor Ps. aeruginosa in soil samples. Our preliminary screening of Ps. aeruginosa in soil showed that it was not present in the French or Tunisian agricultural soils but was detectable thank to an enrichment procedure at a very low level in some soils of Burkina Faso. These data then suggested that human populations will be rarely expose to Ps. aeruginosa through contact with soil and that risk of human health are low due to low level of bacterial cells. Our study also showed that whatever the method used Ps. aeruginosa was successfully detected and quantified in manures. As organic amendment with farmyard manure is a widespread agricultural practice for soil organic fertilization, we intend to apply the culture-dependent and culture-independent approaches described herein to investigate the impact of such practice on Ps. aeruginosa dissemination and the influence of soil characteristics and soil selective pressure on Ps. aeruginosa survival and genetic evolution (i.e. virulence properties and antimicrobial susceptibilities).
We thank all the soil surveyors and technical assistants involved in sampling the sites in Tunisia and Burkina Faso. We thank Mélanie Lelièvre (Plateforme Genosol, Dijon, France), and the technical staff of the French soil library (Unité INFOSOL, INRA, Orléans, France) and of the platform DTAMB (IFR 41 - Université de Lyon, France) for their dedicated technical assistance in DNA extraction, qPCR optimization and soil preparation. This study was funded by the ‘Agence Nationale de la Recherche’ (ANR) (programmes 07 SEST project 018-01, 05 SEST project 009-01). Céline Colinon was funded with a postdoctoral grant from the ANR 07 SEST project 018-01 and Amélie Deredjian with a PhD grant from the National Centre for Scientific Research (CNRS). We thank the Ministry of Foreign Affairs (CORUS programme) for supporting part of this study financially.