Assessment of the specificity of Burkholderia and Pseudomonas qPCR assays for detection of these genera in soil using 454 pyrosequencing

Authors


Correspondence: Søren J. Sørensen, Molecular Microbial Ecology Group, Department of Biology, University of Copenhagen, Sølvgade 83H DK 1307K, Copenhagen, Denmark. Tel.: +45 35 32 20 59; fax: +45 35 32 20 40; e-mail: sjs@bio.ku.dk

Abstract

In this study, two highly specific quantitative PCR assays targeting the bacterial genera Burkholderia and Pseudomonas were developed and evaluated on soil samples. The primers were targeting different multivariate regions of the 16S rRNA gene and designed to be compatible with quantitative PCR and the high throughput 454 pyrosequencing technique. The developed assays were validated using the standard methods. All tests with the new developed assays showed very high specificity. Pyrosequencing was used for direct analysis of the PCR product and applied as a specificity measurement of the primers. The Pseudomonas primers showed a 99% primer specificity, which covered 200 different Pseudomonas sequence clusters in 0.5 g of soil. In contrast to that the same approach using the genus-specific Burkholderia primers showed only 8% primer specificity. This discrepancy in primer specificity between the normal procedures compared with pyrosequencing illustrates that the common validation procedures for quantitative PCR primers may be misleading. Our results exemplify the fact that current 16S RNA gene sequence databases might lack resolution within many taxonomic groups and emphasize the necessity for a standardized and functional primer validation protocol. A possible solution to this could be to supplement the normal verification of quantitative PCR assays with a pyrosequencing approach.

Introduction

The number of species and the diversity of the microbial community in ecosystems are immense (Torsvik et al., 1990; Gans et al., 2005; Singh et al., 2009). In recent years, a tremendous effort has been put into identification of the earth's microbiome (Vogel et al., 2009; Editorial, 2011). In soil, the vast majority of bacteria are nonculturable, and the knowledge of bacterial community organization and its importance for ecosystem function is poorly understood (Singh et al., 2009). It is of fundamental importance to identify the bacterial community for a better understanding of nutrient cycling and energy flow in the ecosystem (Saleh-Lakha et al., 2005; van der Heijden et al., 2008).

During the last few years, quantitative PCR (qPCR), also known as real-time PCR, has emerged as a powerful tool for examination of microbial communities (Zhang & Fang, 2006). The use of qPCR allows quantification of specific microbial populations more accurately than other methods (Zhang & Fang, 2006; Sharma et al., 2007).

Presently, there is no clear consensus on how to optimize and verify the specificity of new qPCR assays. The general approach has been to verify the primers in silico using various primer-design software and then when possible test them on single-species isolates for a positive or negative reaction (Widmer et al., 1998; Salles et al., 2002; Fierer et al., 2005; Lloyd-Jones et al., 2005; Yu et al., 2005). This approach in not perfect but acceptable and useful on single-species samples; however, it is not sufficient when working with samples from complex environments such as soil, where the majority of the bacteria are unculturable and not represented in gene sequence databases such as the Ribosomal Database Project (http://rdp.cme.msu.edu/ ; Bustin et al., 2009; Cole et al., 2009; Morales & Holben, 2009). In complex samples, the specificity of the qPCR assay needs to be very accurately confirmed to avoid over- or underestimation, so far a useful method has been missing.

Some species of Burkholderia and Pseudomonas are known pathogens of plants and animals, including humans; they are also active in the nitrogen cycle and some produce metabolites suitable for the biotechnology industry. They are some of the most ubiquitous genera worldwide and have been found in many different habitats such as water, humans/animals, plants, fungi, clouds and soil, as well as in extreme environments like arctic and desert soil (Palleroni, 2005; Peix et al., 2009). The detection of Burkholderia and Pseudomonas species in the environment may help us gain a more complete understanding of their ecological significance (Peix et al., 2009).

The aim of this study was to develop updated Burkholderia- and Pseudomonas-specific qPCR assays for quantification of both genera in soil, and to test the specificity of the these assays using the classical method of single-species amplification compared with pyrosequencing of soil samples using the new primers.

Materials and methods

Soil samples and extractions

Topsoil samples were collected in triplicate from an agricultural test site in Tåstrup, Denmark. The soils had been treated with elevated level of household compost or sludge, details in Magid et al. (2006) and Poulsen et al. (2012). All soil samples were sieved through a 2-mm sieve to remove stones and roots and provide homogenous samples that were stored field moist below 4 °C to minimize changes in microbial population.

DNA extraction

DNA extraction from soil was carried out by FastDNA® SPIN for Soil kit (MP Biomedicals, Solon, OH) according to manufacturer's instructions. The DNA extracted from soil was stored at −20 °C. The DNA concentration from each sample was determined using Quant-iT dsDNA HS Assay Kit and the Qubit fluorometer (Invitrogen, Palsley, UK).

Primers and probes design

A total of 116 Pseudomonas and 55 Burkholderia type strains 16S rRNA gene sequences available at RDP (Cole et al., 2009) on December 2010 were downloaded. The 16S rRNA gene sequences from each group were aligned using clc Workbench 4.2 (CLC bio, Aarhus, Denmark). The Pseudomonas and Burkholderia 16S rRNA gene sequence contains three hyper variable regions (HVR) and several minor variable regions (Moore et al., 1996; Baker et al., 2003). The HVR is the candidate spot to detect sequence variation from genus to species level, whereas conserved regions flanking the variable regions as well as inside the alignments for the two microbial groups were manually checked to locate the optimal sequences for primers and probes. The specificity of all possible primer and probe sequences was tested in the RDP probe match software. Furthermore, in silico validation of selected primers and probes was carried out in clc 4.2 and Amplify 3X software. The dual-labelled probes were designed with a fluorophore (6-carboxyfluorescein/FAM) and a quencher (Black Hole Quencher BHQ I) linked to the 5′ and the 3′ ends, respectively. The characteristics of the two qPCR assays developed in this study are summarized in Table 1.

Table 1. Characteristics of primers and probes
NameFunctionTargetSequence (5′→3′)PositionTarget region in 16SMatching efficiency (%)aTm (°C)Annealing temp (°C)Amplicon size (bp)Ref.
  1. a

    Ratio (%) of number of organisms matched with the corresponding primer (or probe) of the target to number of all type specific organisms of the group.

Pse435FF PrimerPseudomonasACTTTAAGTTGGGAGGAAGGG435–455V3–V49157.960251This study
Pse449ProbePseudomonasFam-ACAGAATAAGCACCGGCTAAC-BHQ491–511V3–V49557.960 This study
Pse686RR PrimerPseudomonasACACAGGAAATTCCACCACCC666–686V3–V48559.860 This study
Tit-B-Pse435F1CTATGCGCCTTGCCAGCCCGCTCAGACGAGTGCGTACTTTAAGTTGGGAGGAAGGGSequencing primerThis study
Tit-A-Pse686RCGTATCGCCTCCCTCGCGCCATCAGACACAGGAAATTCCACCACCCSequencing primerThis study
BKH812FF PrimerBurkholderiaCCCTAAACGATGTCAACTAGTTG812–831V5–V77858.959438This study
BKH872ProbeBurkholderiaFam-AAGTTGACCGCCTGGGGAGT-BHQ872–891V5–V710061.459 This study
BKH1249RR PrimerBurkholderiaACCCTCTGTTCCGACCAT1232–1249V5–V7915659 This study
Tit-B-BKH812F8CTATGCGCCTTGCCAGCCCGCTCAGCTCGCGTGTCCCCTAAACGATGTCAACTAGTTGSequencing primerThis study
Tit-A-BKH1249RCGTATCGCCTCCCTCGCGCCATCAGACCCTCTGTTCCGACCATSequencing primerThis study
Eub338FF PrimerAll bacteriaACTCCTACGGGAGGCAGCAG338–357V39157.960180Fierer et al. (2005)
Eub518RR PrimerAll bacteriaATTACCGCGGCTGCTGG502–518V38959.760  

To verify that the primers were suitable for studies of intra-genus diversity, an in silico analysis was performed in which the internal sequence variation between the forward and reverse primers was tested. The regions between the primers (possible amplicons) were recovered from alignment of the entire 16S RNA gene (for all 116 and 55 type sequences), and partial alignments were conducted (clc 4.2.). The partial alignments were checked for suitable internal base variation, and phylogenetic neighbour-joining trees were constructed [SplitsTree (Huson & Bryant, 2006)] to verify possible species separation.

Quantitative PCR conditions

All qPCRs were performed using 25 μL reactions on the Mx3000 (Stratagene, Cedar Creek, TX). The qPCR program and the reagents concentrations were identical in all SYBR Green I assay reactions consisting of 1× of Brilliant SYBR Green QPCR Master Mix (Stratagene), 385 nM of forward primer and reverse primer and 2 μL sample DNA. The qPCR conditions were 10 min at 95 °C followed by 40 cycles of 95 °C for 30 s and 1 min at 60 °C ended by a dissociation curve segment. Fluorescent measurements were taken at the end of every merged annealing/extension steps.

In the hydrolysis probe assay, the reactions contained the following: 1× TaqMan Environmental Master Mix 2.0 (Applied Biosystems, Warrington, UK), 770 nM forward primer and reverse primer, 100 nM probe and 2 μL sample DNA. The qPCR program consisted of 10 min at 95 °C, followed by 45 cycles at 95 °C for 30 s, and 1 min at 60 °C (merged annealing/extension steps).

For validation, the data trend from the developed qPCR assays was compared with a 16S eubacterial qPCR assay (see Table 1 for primer details; Fierer et al., 2005).

A negative control was included in all qPCR assays, and all experiments were performed in triplicates. Tenfold serial dilutions of extracted genomic DNA from pure cultures of Pseudomonas putida kt2440 and Burkholderia cepacia were used as standard curves. Standard curve calculations as described in Park and Crowley 2005.

Data analyses

All statistical data analysis was conducted in sas Enterprise Guided 4.2. One-way anova with Tukey's studentized range distribution was used to detect differences. A < 0.05 level of significance was used.

Pyrosequencing

To validate the specificity of the Pseudomonas primers, DNA extracted from the soil sample treated with sludge was amplified using Pseudomonas primers and sequenced on a standard plate using the GS FLX system (Roche, Basel, Switzerland) as previously described (Poulsen et al., 2012). Briefly, DNA extracted from soil was amplified with the Pseudomonas 16S rRNA gene primers Pse435F and Pse686R as described above. The amplified products were purified from gel using The Montage DNA Gel Extraction Kit (Millipore). Addition of adapter and tags necessary for pyrosequencing was performed using the fusion primers (primer Pse435F with Adapter A and tag and primer Pse686R with Adapter B. The amplified fragments with adapters and tags were quantified as mentioned above. Sequencing was performed using a modified version of the GS FLX amplicon sequencing protocol (Roche).

A similar approach with tagged primers was used to test the specificity of the Burkholderia primers (BKH812F and BKH1249R), sequencing on a Titanium plate using the GS FLX system (Roche, Basel, Switzerland).

Analysis of pyrosequencing results

The Pyrosequencing Pipeline Initial Process at RDP was used for quality filtering and trimming of sequences with a minimum length of 150 bp. The RDP pipeline was also used to generate rarefaction curves.

Operational Taxonomic Unit (OTU) picking was carried out using the uclust/usearch software (http://www.drive5.com/usearch/). The OTUs were picked by clustering the reads at ≥ 97% sequence identity, with the ‘optimal’ option enabled. Taxonomic classification was made on OTU representatives with the RDP classifier (ver. 2.1) software, which was run locally using the Training Data 5 set as a reference. A confidence threshold of ≥ 50% was chosen as the requirement for accurate genus-level determination, because of the reads length < 250 bp. Accordingly, sequences assigned to a genus with lower than 50% confidences were deemed as unclassified. Further species-level classification was made using usearch against a locally curated database of c. 45 000 nonredundant (nr100) 16S rRNA gene sequences from the RDP (release 10.20) and NCBI RefSeq databases. The reference set was truncated to only include the V3–V4 HVR.

Results

Primer validation

Based on the percentage of sequences that matched the primer and probes, an in silico analysis showed a specificity between 78% and 100%. Based on the type strains sequences in the RDP database, the primer and probe sets matched all Burkholderia and Pseudomonas with only one base mismatch (Table 1), and only very few nontarget organisms (0–0.7% of all type species) were detected.

In addition, the qPCR assays were validated by testing the specificity of the primers on the following 12 closely related species: Klebsiella pneumoniae, Klebsiella oxytoca, Acinetobacter calcoaceticuc, Burkholderia cepacia, Burkholderia sp., Ralstonia eutrophus. Brevundimonas sp., Stenotrophomonas maltophilia, Pseudomonas putida, Pseudomonas fluorescens, Pseudomonas aeruginosa and Pseudomonas stutzeri. The assays were specific for their targets and gave no or very high Ct values for the nontarget groups equal to the nontemplate control (data not shown), which furthermore confirmed the specificity of the primers.

Pyrosequencing analysis

Pyrosequencing of PCR products amplified from the sludge soil sample with the Burkholderia primers resulted in 24 890 sequences longer than 250 bp. RDP classification of these sequences showed that 99% of the sequences belonged to Betaproteobacteria and of these only 8% to Burkholderia (Fig. 1). Based on these results, the Burkholderia primer specificity is 8%. Because of the low primer specificity, no further data treatment was carried out.

Figure 1.

Specificity of Burkholderia primers on soil. Relative distribution of 454 pyrosequences obtained by using the Burkholderia primers on soil. All genera and unclassified bacteria representing < 0.5% of the total sequences were removed in this figure.

Pyrosequencing of PCR products amplified from the same soil sample with the Pseudomonas-specific primers generated a total of 24 354 sequences longer than 150 bp. RDP classification of these sequences showed that 98.76% belonged to Pseudomonas (Fig. 2), 0.56% to unclassified bacteria, 0.40% to unclassified Pseudomonadacea, and the last 0.28% belonged to closely related bacteria. Based on these numbers, we estimated that the Pseudomonas primers have the specificity close to 99%.

Figure 2.

Specificity of Pseudomonas primers on soil. Relative distribution of 454 pyrosequences obtained by using the Pseudomonas primers on soil.

Using the RDP Pyrosequencing pipeline, the rarefaction curves estimated that 0.5 g of soil contains c. 200 different Pseudomonas OTUs at 3% maximum cluster distance (Fig. 3).

Figure 3.

Pseudomonas rarefaction curve. Rarefaction curves at different dissimilarity levels indicate the number of Pseudomonas OTU's in 0.5 g of soil.

To assess the distribution of the Pseudomonas community in soil, clusters containing more than 50 identical copies were blasted against the full RDP database to identify the species level. In most cases, a high identity score on a single species was possible, but in a few blasts several species appeared with identical similarity scores. Where several hits were shown with identical similarity score, the number of sequences in the cluster was distributed evenly between the different hits. The different clusters and the number of species and sequences they represent are illustrated in Fig. 4. Using this method, the most dominant Pseudomonas groups in the soil are clearly uncultured Pseudomonas and P. putida followed by P. flourescens and Pseudomonas sp. The figure also shows that there is a rather diverse mixture of Pseudomonas species present in the soil.

Figure 4.

Diversity and distribution of the Pseudomonas community in soil. Pseudomonas clusters containing more then 50 identical copies were blasted against the full RDP database to identify species level and the distribution of Pseudomonas in soil.

Enumeration of Pseudomonas in soil

Pseudomonas was quantified in two different soils: one treated with household compost and the other with sewage sludge. The two assays, SYBR Green I and hydrolysis probes detection format, were validated and compared. All qPCR runs showed high efficiency c. 100% and R2-value in the average range between 0.981 and 0.999 (data not shown).

Based on qPCR data, the number of Pseudomonas in the soil samples was between 4.93 × 105 and 3.90× 106 cells g−1 of soil. In addition, qPCR results showed that the lowest number of Pseudomonas was in the soil treated with sludge (Table 2).

Table 2. Bacterial values detected in two different soils using qPCR
Detection method (cells per gram of soil)Soil fertilizer treatment
SludgeCompost
  1. ± values represent standard errors of the means from the replicate samples. The two soils are amended with two different fertilizers; sludge and household compost.

16S eubacterial
SYBR Green assay3.43 × 108 ± 3.76 × 1074.24 × 108 ± 2.11 × 107
Hydrolysis Probe Assay3.14 × 107 ± 2.98 × 1063.83 × 107 ± 3.22 × 106
Pseudomonas
SYBR Green assay2.42 × 106 ± 1.86 × 1053.90 × 106 ± 2.14 × 105
Hydrolysis Probe Assay4.93 × 105 ± 8.88 × 1047.33 × 105 ± 6.50 × 104

The total number of bacteria in the two soils was estimated to be in the range of 3.43 × 108 and 4.24× 108 cells g−1 of soil using a general qPCR assay targeting the eubacterial 16S rRNA gene (Fierer et al., 2005). Similar to the Pseudomonas data, the total number of bacteria was lowest in the sludge-treated soil.

Discussion

The quantification of Pseudomonas cells in the soils with qPCR (Table 2) showed a significantly higher number of bacteria in the compost-treated soil (P < 0.0001). Detecting 106 Pseudomonas cells g−1 soil is in accordance with previously published data on Pseudomonas in soil (Pallud et al., 2001; Lloyd-Jones et al., 2005).

Results from the eubacterial qPCR assay showed the same differences between the soil types as with the genus-specific protocols, highest bacterial counts in the compost-treated soil and a lower in the sludge-treated soil.

The sequencing data showed a high diversity of Pseudomonas, identifying c. 200 different OTUs and more than 20 different species at a 3% maximum cluster distance. If the length of the PCR fragments is taken into consideration, the observed diversity in the Pseudomonas genus is rather high, especially because it is well-documented that the 16S rRNA gene does not contain enough genetic variation to identify all Pseudomonas species to species level (Peix et al., 2009). However, in this study, c. 200 different Pseudomonas OTUs, many to species level, were detected by pyrosequencing.

Analysis of the Pseudomonas primers using pyrosequencing showed that 99% of the sequences belonged to the genus Pseudomonas. However, only 8% of the PCR products amplified with Burkholderia primers belonged to the genus Burkholderia and 36% of the sequences were defined as unclassified betaproteobacteria and the remaining divided primarily between Methylotenera, Methylovorus and Thiobacillus.

In the Burkholderia sequencing data, several nontarget bacteria were detected. Bacteria like Pseudomonas, Sinobacteraceae, Legionella, Alcaligenaceae, Methylophilaceae and Rhodocyclaceac should not be present. The primer target sequences in all bacteria in NCBI from these groups have a 1–2 bp mismatch to our Burkholderia primers. The most likely explanation is that we used a too low Tm value. The Tm for the Burkholderia primers was set to 60 °C based on a temperature gradient PCR, above 60 °C the bands began to fade. Another explanation could be presence in the soil of bacteria other than Burkholderia with exact match to the primer sequence and that these bacteria are absent from current sequence databases.

Whether this unspecificity is because of inaccurate Tm determination, bad quality of the primers, malfunction in the PCR machine or incomplete databases, the low specific of the primers would not have been detected without the pyrosequencing specificity test.

Because both primer pairs were designed to be highly specific, and performed very well when tested in silico and against selected cultured strains, the surprisingly low specificity of the Burkholderia primers compared with the high specificity of the Pseudomonas primers clearly illustrates the usefulness of pyrosequencing as a tool for validation of new primers.

The last years' rapid development of fully sequenced bacteria and changing phylogenetic trees has called for a revision of the previously used Burkholderia and Pseudomonas primers, because they were designed using a limited number of sequences, which makes these genus-specific primers unspecific or too specific not covering the entire genera of Pseudomonas and Burkholderia (Widmer et al., 1998; Johnsen et al., 1999; LiPuma et al., 1999; Khan & Yadav, 2004; Lloyd-Jones et al., 2005). Furthermore, some of the published primers for Burkholderia and Pseudomonas are based on the use of one specific primer and one general primer, which increases the possibility of false positives. In a study by Morales & Holben (2009), it was shown that even specific primers exhibit a high degree of unspecificity, stressing the importance of proper primer validation.

It is important to use the MIQE guidelines when running and designing a qPCR experiment (Bustin et al., 2009); there are no such minimum guidelines when designing primers and testing the specificity of the primers for qPCR assays.

As an addition to the verification of qPCR primer specificity by in silico analysis and screening on single bacterial isolates, we propose to sequence DNA amplified from a high diversity sample such as soil as an additional way to verify the primers specificity. Next generation sequencing is becoming cheaper, and several thousand species in a single sample can be identified; therefore, we recommend using this approach as a time efficient way of verifying the specificity of new primers. Thereby scientific arguments about the primer specificity could be avoided and time used on numerous tests on single culture bacteria, clones and isolates could be saved.

In conclusion, the data presented in this study showed that with the designed primer and probe set, it is possible to detect and quantify Pseudomonas in soil samples with high specificity, and to identify variations in the bacterial soil community. The designed qPCR assay holds great application potentials and is without modifications, compatible with the high throughput pyrosequencing techniques. Thereby it is possible to detect and quantify Pseudomonas to species level, increasing our knowledge and understanding of, for example, some opportunistic pathogenic bacteria.

The data also stress the importance of proper qPCR assay validation using pyrosequencing, exemplified via the Burkholderia primers, two supposedly highly specific and thoroughly tested primers with only 8% specificity.

Acknowledgement

Annelise Kjøller, Karin Vestberg and Julie Backe Bergmark are thanked for their assistance.

Ancillary