Advertisement

Food Analytical Methods

, Volume 7, Issue 8, pp 1719–1728 | Cite as

Inter-laboratory Testing of GMO Detection by Combinatory SYBR®Green PCR Screening (CoSYPS)

  • Elodie Barbau-Piednoir
  • Pieter Stragier
  • Nancy Roosens
  • Marco Mazzara
  • Cristian Savini
  • Guy Van den Eede
  • Marc Van den Bulcke
Open Access
Article

Abstract

Combinatory SYBR®Green real-time PCR Screening (CoSYPS) is an efficient, sensitive approach for detecting complex targets such as genetically modified organisms (GMOs) in food and feed products. GMO analysis for legal purposes has become increasingly complex and costly due to the diversity in recombinant targets present in the different GMOs. For this reason, screening for the presence of GMOs is in general the first step in the detection of GM material in a product. CoSYPS allows detecting the large majority of globally commercial GMOs using SYBR®Green real-time PCR methods for six GM targets (P35S, Tnos, CryIAb, CP4-EPSPS, PAT and BAR) combined with species-specific PCR methods (e.g., maize, soy, rapeseed). Here, the results of an inter-laboratory trial on seven samples with different GMO mixtures at different levels are presented. In total, 13 laboratories participated in the trial and the currently most frequently used PCR analysis platforms are represented. The inter-laboratory study clearly demonstrates that PCR methods used in CoSYPS form a very robust GMO screening system. Sensitivity, specificity, positive and negative predictive values are for all PCR methods higher than 95 % for all samples. Together, these results show that the SYBR®Green real-time PCR methods used in CoSYPS are effectively applicable to different PCR platforms and amendable to configuration into a sensitive high-throughput GMO screening and decision support tool.

Keywords

Combinatory SYBR®Green PCR Screening GMO detection Genetically modified food 

Introduction

Genetically modified (GM) food and feed products have become a reality today (James 2012). Since their initial commercialization in 1994, about 150 genetically modified plants have received approval for use as food or feed (http://cera-gmc.org). In most countries, such use is strictly regulated and compliance measures have been implemented. An essential element in the enforcement of this legislation is the efficient detection of genetically modified organisms (GMOs) or traces thereof in food and feed products (http://gmo-crl.jrc.ec.europa.eu/gmomethods).

Considering their large diversity (different GM elements in various species), the application of a generic screening for the presence of GM material is most often the first step in GMO analysis (Holst-Jensen et al. 2012). Such qualitative screening methods provide a presence/absence response (taxon-specific sequence or a GM element) and help reducing the number of subsequent identification analyses.

For this reason, screening for the presence of GMOs is in general the first step in the detection of GM material.

In most cases, GMO screening approaches apply quantitative PCR (qPCR) methods for detecting the presence of GM material in food and feed samples. More complex strategies employing microarrays (e.g., GMOchips (Leimanis et al. 2006), NAIMA (Morisset et al. 2008), and PADLOCK probes ligation (Ujhelyi et al. 2012)) have been attempted, but are, however, still under validation (NAIMA, padlock approach) or received limited distribution within the enforcement world (GMOchips). Thus, today, qPCR screening is by far the most commonly applied approach in screening for GM materials in food and feed products. A set of inter-laboratory trial validated GMO detection methods have been compiled into a Compendium (Van den Eede 2010). The most common qualitative qPCR methods used for screening are probe-based qPCR methods (Fernandez et al. 2005; Lee et al. 2007; Waiblinger et al. 2010). The PCR methods described in the Compendium were designed by many different groups and, as a consequence often, do not apply the same PCR conditions and protocols. As such, a combined use of these methods in a single run in a 96 well plate format may not be feasible without prior verification of the performance and efficacy in such usage (Morisset et al. 2008).

To meet the need of a sensitive, reliable but also cost-effective and flexible GMO screening strategy, a set of simplex SYBR®Green qPCR GMO detection methods which all can be run together in a single analytical PCR plate has been developed at the Scientific Institute of Public Health (Barbau-Piednoir et al. 2010; Barbau-Piednoir et al. 2012; Van den Bulcke et al. 2010; Mbongolo Mbella et al. 2011). Six GM-element SYBR®Green qPCR screening methods were newly developed for this purpose: four trait-specific methods (CP4-EPSPS, CryIAb, PAT, BAR), and two methods detecting the major GM generic recombinant elements, the cauliflower mosaic virus (CaMV) 35S promoter (p35S) and the Agrobacterium Nos terminator (tNos) (Barbau-Piednoir et al. 2010; Barbau-Piednoir et al. 2012). These GM-element-specific methods were to be combined with SYBR®Green taxon/species-specific qPCR methods targeting e.g.; a lectin, an alcohol dehydrogenase, and a cruciferin target DNA sequence for determining the presence of material derived from soybean, maize, and rapeseed, respectively (Mbongolo Mbella et al. 2011). In addition, a method detecting a generic plant marker, the chloroplastic RbCl element (Mbongolo Mbella et al. 2011) and a CaMV-specific method to demonstrate the possible presence of CaMV material in a sample, were included. The combined use of both sets of qPCR methods and a decision support system (DSS) allows performing a GMO screening analysis within one run on a single PCR plate and was designated “Combinatory SYBR®Green qPCR Screening” (CoSYPS) (Van den Bulcke et al. 2010). In the DSS, the outcome of analyses using the SYBR®Green qPCR methods is interpreted using a DSS that applies a “prime number”-based algorithm to indicate which GMOs are possibly present in the sample (Van den Bulcke et al. 2010).

It was shown in-house that all these SYBR®Green qPCR GMO screening methods performed equally well under the same reaction conditions with high specificity and sensitivity on different qPCR platforms (own unpublished results). To complete the validation of CoSYPS in GMO analysis, an inter-laboratory trial was organized to evaluate the performance in different laboratories and evaluate the general applicability of SYBRGreen technology considering the lesser experience of most laboratories with this type of qPCR. Multiple test samples were prepared including various GMOs (soy GTS40-3-2, maize MON 810 and Bt11, and rapeseed RF3 and RT73) at different levels (0–2 %) in different backgrounds (soy, maize and rapeseed). Participants were provided with the test samples, the test protocol, all reagents and the CoSYPS DSS. Here, the results of the CoSYPS inter-laboratory trial are presented. The significance, practical bottlenecks and future opportunities of the use of CoSYPS in GMO analysis are discussed.

Materials and Methods

Materials

Plant gDNA was extracted from leaf tissue of in-house grown plants (WT maize, soy and rapeseed; Bt11 and MON 810 GM Maize; GTS40-3-2 soy) or from CRM powders of the Institute for Reference Materials and Measurements (IRMM) (MON 810 GM maize) or as purified DNA of the American Oil Chemists’ Society (AOCS) (GT73 and MS8/RF3 rapeseed)

For the in-house plant tissue production, standard growth conditions were at 16/8-h day/night regime at 25 °C with chamber humidity at 80 % in a Schneijder Scientific Plant growth chamber (S1084).

All Sybricon plasmids used as positive controls were constructed in-house, isolated using Qiagen mini/midi plasmid preparations and verified by DNA sequence analysis (Barbau-Piednoir et al. 2010; Barbau-Piednoir et al. 2012; Mbongolo Mbella et al. 2011). Plasmid DNA concentration was measured by Spectrophometry (Smart Spec Plus; Bio-Rad), diluted in two steps at 5 ng/μl in DNAse/RNAse free water (Acros). This dilution was requantified with Picogreen (Invitrogen). Then a second dilution at 5 fg/μl was prepared, in two steps, from the dilution at 5 ng/μl by dilution in carrier DNA (Calf Thymus DNA 4 ng/μl). Plasmid dilutions were verified by qPCR using the appropriate SYBR®Green qPCR methods corresponding to each of the targets (data not shown).

As negative control, DNase/RNase free water (Acros) was used and indicated as “No Template Control” (NTC).

General Methods

Genomic DNA Isolation and Quantification

Total DNA extraction of the different materials (seeds, powders, leaves) was performed as described in (Barbau-Piednoir et al. 2010). In short, 250 mg to 1 g homogenized material was extracted by a CTAB DNA extraction protocol. The final genomic DNA (gDNA) pellet was resuspended in 200 μl DNase/RNase-free water (ACROS) and stored at −20 °C.

The extracted gDNA was quantified using the PicoGreen dsDNA quantitation kit (Invitrogen) on a VersaFluor™ Fluorimeter (Bio-Rad).

Primers, Probes and PCR Reagents

All primers applied in the SYBR®Green qPCR methods were as described (Barbau-Piednoir et al. 2010, 2012; Mbongolo Mbella et al. 2011) (see Table 1). Endogene Taqman® qPCR methods applied in the EURL-GMFF validated GMO detection methods were used to identify possible traces of contamination by other crops: for maize, the ADH reference gene PCR detection method from the Bt11 quantitative method (http://gmo-crl.jrc.ec.europa.eu/summaries/Bt11-protocol.pdf); for soybean, the lectin reference gene PCR detection method from the GTS40-3-2 quantitative method (http://gmo-crl.jrc.ec.europa.eu/summaries/40-3-2_validated_Method.pdf), for oilseed rape, the MDB reference gene PCR detection method from the Gt73 quantitative method (http://gmo-crl.jrc.ec.europa.eu/summaries/RT73_validated_Method.pdf), for rice, the KVM reference gene PCR detection method from the LL62 quantitative method (http://gmo-crl.jrc.ec.europa.eu/summaries/LLRICE62_validated_Protocol.pdf), and for cotton, the acp1 reference gene PCR detection method from MON 531 quantitative method (http://gmo-crl.jrc.ec.europa.eu/summaries/MON531_validated_method.pdf).
Table 1

Description of the SYBR®Green qPCR methods applied in the GMO screening

PCR method

Name

Amplified gene target

Amplicon length (bp)

Target class

Reference

Plant

VPRBCP1

Ribulose-1,5-bisphophate craboxylase/oxygenase

95

Plant kingdom

Mbongolo Mbella et al. 2011

VPRBCP1

STLM (lectin)

SLTM 1

Lectine

81

Soy specific

Mbongolo Mbella et al. 2011

SLTM 2

ADH alt

ADH-alt Fwd

Alcohol dehydrogenase 1 promotor

83

Maize specific

Mbongolo Mbella et al. 2011

ADH-alt Rev

CRU

Cru770 F

Cruciferine

85

OSR specific

Mbongolo Mbella et al. 2011

Cru770 R

P35S short

35S_N3Fwd

35S RNA promoter (CaMV)

75

Generic recombinant

Barbau-Piednoir et al. 2010

35S_N3Rev

Tnos

tNOS_NN_Fwd

Terminator nopaline synthase (Agrobacterium)

69

Generic recombinant

Barbau-Piednoir et al. 2010

tNOS D Rev

CP4 epsps

CP4 Synthetic F

EPSPS CP4

108

Trait specific

Barbau-Piednoir et al. 2012

CP4 Synthetic R

EPSPS RRS

CP4 synth Rbis

Gt73 EPSPS

CryIAb

CryIAb_Bt.Cott_Fwd

Endotoxine delta (Bacillus thuringiensis)

73

Trait specific

Barbau-Piednoir et al. 2012

CryIAb_Bt.Cott_Rev

PAT

Pat-Pat Fwd

Phosphinothricin-N-acetyl transferase (Strepomyces viridochromogenes)

109

Trait specific

Barbau-Piednoir et al. 2012

Pat-Pat Rev

BAR

Pat-Bar Fwd

Phosphinothricin-N-acetyl transferase (Strepomyces hygroscopicus)

69

Trait specific

Barbau-Piednoir et al. 2012

Pat-Bar Rev

CRT 2

CRT-F

Reverse transcriptase of CaMV

94

Donor Organism marker

Adapted from Wolf et al. 2000

CRT-R2

All primers and probes were purchased at Eurogentec. Mastermixes were purchased at Diagenode.

In-house Quality Verification and Unknown Preparation

The quality of the DNA extraction was estimated on the basis of two criteria: 1) the extracted DNA quantity and 2) the purity of the gDNA extracted (no contamination by unexpected targets). Extracted DNA quantity needed to be concentrated enough to be able to prepare the samples for the 13 laboratories.

qPCRs were carried out on ABI 7300 PCR System (Applied Biosystems) using the SYBR®Green qPCR Mastermix (Diagenode).

Taqman® qPCR contamination checks were performed on an ABI 7300 PCR System (Applied Biosystems) using the TaqMan® PCR Mastermix (Diagenode).

The purity of the gDNA extract was checked as described above. Trace amounts of cross-contaminating GM material could not be avoided 100 % in all preparations though. Neither the nature nor the source of the contaminating GM material was determined in further detail but late positive signals (C t >35) were obtained for some GM markers (p35S, tNos, CP4-EPSPS, CryIAb and PAT) and for the maize and soybean plant species markers (data not shown). For this trial, a gDNA extract was considered acceptable when the C t observed from the impurities was above 35 C t. All acceptable gDNA fractions were pooled, mixed overnight on a rotating wheel, diluted at 2 ng/μl and used for preparing the respective 7 ml of unknown in v/v ratios.

Statistical Data Analysis

Sensitivity, specificity, false positive and negative rates were calculated for all methods and samples along with the positive and negative predictive values (http://www.cebm.net/index.aspx?o=1042). Confidence intervals (P (95 %), α = 0.05) were calculated according to Newcombe (1998). Pearson chi-square analysis (α = 0.05) was performed using STATA Statistics/Data Analysis version 10.1.

Inter-laboratory Trial Methodology

General Organization

A large-scale (LS) collaborative trial was organized jointly by WIV-ISP and the EURL-GMFF. The study was launched on 17 June 2009 and all experimental work by the participating laboratories was to be finalized within 4 weeks after receipt of the materials. During this period, the laboratories could contact the WIV-ISP GMOLab for technical assistance. An official invitation was sent to all laboratories of the European Network of GMO Laboratories (ENGL) (n = 96) by the EU-RL GMFF. Twenty-three laboratories, belonging to 17 countries of the EU, responded positively and from this group 13 laboratories were selected at random. All participating laboratories were requested to sign a Material Transfer Agreement prior to sending off of the test materials. All selected laboratories received ten unknowns (U1 to U10), the CoSYPS inter-laboratory protocol, the Excel interpretation matrix files (DSS), and the Deviation Report Form. After the experimental work, all participating laboratories were requested to complete an information sheet to document the technical infrastructure of the laboratory and their experience with fulfilling the CoSYPS system protocol.

Aim and Design of the Large-Scale CoSYPS Inter-laboratory Trial

The aim of this study was dual: first, to demonstrate a “proof of concept” of CoSYPS to detect various levels of GMOs in different backgrounds and the evaluation of pre-fixed “cut-off values” and the CoSYPS DSS; and second, to demonstrate the general applicability of CoSYPS to estimate the impact of introduction of SYBR®Green qPCR analysis and the potential adaptive actions to take in the context of implementation of the CoSYPS methodology in a new laboratory.

For this, three control (maize, soy or rapeseed material) and seven GM test samples, the latter further indicated as GM unknowns, were prepared in-house by the WIV-ISP (see Table 2). As positive control, a plasmid DNA mix was prepared with ±40 copies/μl of each of the so-called “Sybricon” plasmids. Each Sybricon contains the amplicon produced by the corresponding SYBR®Green real-time qPCR method (Barbau-Piednoir et al. 2010).
Table 2

Composition of the test samples (U1 to U10) used in the inter-laboratory studies

 

Plant materials (mass%)

Plant

Adh alt

STLM (lectine)

CRU

P35S short

T-nos

CP4- EPSPS

CryIAb

PAT

BAR

CRT 2

U1

100 % maize

+

+

         

U2

100 % OSR

+

  

+

       

U3

100 % soy

+

 

+

        

U4

0,1 % GTS40-3-2 soy in 100 % soy

+

 

+

 

+

+

+

    

U5

2 % Bt11 maize in 100 % maize

+

+

  

+

+

 

+

+

  

U6

0,5 % MON 810 maize in 100 % maize

+

+

  

+

  

+

   

U7

0,1 % GT73 OSR in 100 % OSR

+

  

+

  

+

    

U8

0,9 % RF3 OSR in 100 % OSR

+

  

+

 

+

   

+

 

U9

2 % RF3 OSR, 1 % GTS40-3-2 soy in 100 % maize

+

+

+

+

+

+

+

  

+

 

U10

0,1 % GT73 OSR, 0,5 % MON 810 in 40 % OSR and 60 % maize

+

+

 

+

+

 

+

+

   

The markers are as described in Table 1. Plus symbol (+) stands for “present in the GMO”; minus symbol (−) denotes “not present in the GMO”

The proof of concept for CoSYPS analysis was tested by comparing the outcomes of the laboratory tests with the expected true values. Evaluation of the inter-laboratory data was performed per GM unknown (n = 7), per qPCR method (n = 11) and combined for all seven unknowns.

Preparation of Sample Panel and Quality Assessment

For the preparation of the GM unknowns, a set of GMOs was chosen to cover every GM marker in CoSYPS at least once (GM markers present in each particular GMO are indicated between brackets; see also Table 2): GM maize Bt11 (P35S, Tnos, CryIAb, PAT) and MON810 (P35S and CryIAb), GM soy GTS40/3/2 (P35S, Tnos, CP4 epsps), and GM rapeseed GT73 (CP4 epsps) and RF3 (Tnos, BAR).

An aliquot of each unknown sample was tested by the CoSYPS qPCR method for further characterization and verification at the WIV-ISP, prior to send off (data not shown). These outcomes served as a reference in the statistical/heuristic assessments and were used to assess the Excel DSS file.

During this quality assessment performed by the WIV-ISP, the presence of GMO traces in some wild type maize and soybean materials was observed, but at C t levels above the usual in house cut-off value for routine testing (cut off set at C t = 35). This cut-off represents a rounded average C t value (minus 1 C t to cover a 2-fold PCR efficiency decrease due to, e.g., matrix inhibition effects). This average C t value was calculated prior to the inter-laboratory trial from the C t values corresponding to the lowest number of copies detectable by each of the GM methods applied in screening analysis, wherein “lowest number of copies detectable” was defined as obtaining one out of two positives when two repeats are applied in a serial dilution analysis of genomic certified reference material with known zygosity for the respective targets (data not shown).

In addition, presence of CaMV was observed in all the samples containing AOCS rapeseed material (positive results with the CaMV-specific target qPCR method) whereas the samples prepared with in-house produced rapeseed material were negative (data not shown). Due to this difference with regard to the origin of the reference material, it was decided not to include the results obtained with the CaMV-specific method.

Procedure for qPCR

All participants received a full analytical procedure together with the samples. No training in advance was given to the participants. A standard 25 μl reaction volume was applied containing 5 μl of template (wherein or 2 ng/μl gDNA or 40 copies plasmid DNA/μl were included), 1× SYBR®Green qPCR Mastermix, and 250 nM of each primer (except RbCl method which used final primer concentration of 1.6 μM for VPRBCP1 and 1.2 μM for VPRBCP2). The thermal program consisted of a single cycle of DNA polymerase activation for 10 min at 95 °C, followed by 40 amplification cycles of 15 s at 95 °C (denaturing step) and 1 min at 60 °C (annealing-extension step). After completion of the run, a melting point analysis was performed by stepwise temperature increase (±1.75 °C/min) from 60 °C to 95 °C over 20 min.

Acceptance Criteria for Positive qPCR Test

A qPCR test was considered “positive” when the observed C t value was lower than 35 in combination with an acceptable T m value related to the T m value of the positive control for the particular qPCR test (see also below). The acceptance of the T m value from the positive controls were compared to the outcome of the in-house qPCR device testing and accepted with a maximum deviation of 1 °C. For the T m values of the sample, the T m value was set as positive when it differed less than 1 °C from the T m of the positive control, except for two targets. For the CryIAb target, two different GM sequences have been engineered giving rise to two distinct melting peaks for the same target region. For this reason, the acceptable T m range for CryIAb was set to ±2 °C. For CP4 epsps, also two different GM sequences have been engineered, three primers (two forward and one reverse) were designed, amplifying the respective engineered GM targets also resulting in two different melting peaks. Here, the acceptable T m range was set as −1 °C from the lowest T m value to +1 °C of the highest T m value of the two positive controls.

Decision Support on Interpretation of the qPCR Results

Each qPCR analysis was performed in six folds on each sample. In-house CoSYPS LOD levels were determined in a six-fold repeat analysis of a target dilution series (Barbau-Piednoir et al. 2010). Here, three decision levels were distinguished: “Above LOD” (at least five out of six positives), “At LOD” (four out of six positives) and “Below LOD” (less than four out of six positives). The outcome for each method is then combined into a Gödel Prime number Product (Van den Bulcke et al. 2010) and interpreted using the prime-number GMO matrix table included in the CoSYPS DSS using a factorization algorithm (Van den Bulcke et al. 2010).

Results

General Evaluation of the CoSYPS Inter-laboratory Trial

All participants finalized the analyses within the requested 4-week period. Six laboratories requested additional information and three required supplementary material. The main questions related to the trial organisation were essentially due to a lack of familiarity with the SYBR®Green technology and issues with DSS when using other qPCR devices platforms than an ABI-type. Technical issues resulting in partly missing data were observed with eight of 130 tested plates.

At the end of the trial, a survey was performed to record the setup used in the different laboratories (qPCR device and software, PCR plates, sealing type) and to evaluate the general appreciation by the participants of the use of the CoSYPS. The majority of the laboratories (69 %) used an ABI-type qPCR machine. Most laboratories (92 %) experienced the CoSYPS system as easy to understand/perform and user-friendly (77 %). All laboratories were satisfied with the included protocols, although 31 % of the participants considered the transfer of the analytical results into the CoSYPS Decision Support System as “tedious”.

It can be concluded that in general the CoSYPS trial was well executed and all participants were satisfied with the protocol and the CoSYPS Decision Support System, except for the data transfer into the COSYPS Excel DSS file.

Description of the Applied Adaptative Actions on the LS Inter-laboratory Raw Data Sets

As the results were communicated and analyzed in an Excel format, a first trouble shoot analysis was performed based on these datasheets. The data obtained with the ABI 7500–7500 software V2.0 could not be interpreted using the provided formats and an adapted version had to be developed (data not shown). For the other platforms applied in the LS trial, namely, ABI 7300, 7700 and 7900, the LC480 Roche, the Stratagene X3000 and the Bio-Rad iQ5, the provided Excel DSS were functioning appropriately.

A second adaptation was done in the context of the inappropriate setting of the threshold value. The results obtained for each unknown by each laboratory were then regrouped in tables and ranked per unknown and per method (data not shown). From this analysis, it was observed that 1) in some samples from the maize and soy wild material, traces of GMOs (but above C t = 35) were present (source of false positives and 2) in a number of GM unknown samples, some GM markers were detected at levels just above the cut-off value (resulting in false negatives). The scoring by some laboratories of some positive GM markers just above the cut-off value was re-analyzed as described above. In total 63 analyses (=4.4 %; 63/1430) were re-evaluated by the same person at IPH and for 22 runs (=1.85 %) an adaptation was introduced. In all these cases, the threshold was indeed not set as explained in the protocol.

Finally, one setup failed to correctly detect both melting peaks for the CP4-EPSPS positive plasmid controls. In their analysis the CP4 epsps positive control yielded only one of the two expected peaks in melting analysis (resulting in a “below LOD” conclusion for the second melting peak, if present). A re-analysis was tried to be performed but the raw data were not readable, this probably due to incompatibility with the software versions available. For this, signals with T m values falling within the limits of the PCR device type were accepted as positives.

Sensitivity, Specificity, and the Positive and Negative Predictive Value of the CoSYPS SYBR®Green qPCR Methods in the LS Inter-laboratory Trial

The outcomes of the CoSYPS analyses of ten markers of the seven GM unknowns are summarized in Table 3. The sensitivity or specificity for each method in each sample separately was calculated. In case a marker was expected to be positive, the sensitivity was reported, in case the marker was expected to be absent, the specificity was reported. The results show that all separate SYBR®Green qPCR methods performed appropriately in detecting their respective target(s) as in quasi all samples levels >95 % are obtained for sensitivity or specificity. The lower specificity of the soy marker (specificity at 94 % in some samples) is due to the presence of a small soy contamination in some of the samples (at the cut-off value), detected due to the high sensitivity of the method.
Table 3

Evaluation of all PCR analyses of the GM unknowns of the inter-laboratory trial (after re-reading)

Sample

Test outcome

Plant

ADH alt

STLM (lectin)

CRU

P35S short

Tnos

CP4 epsps

CryIAb

PAT

BAR

Unknown 4 (0.1 % GTS-40-3 soy)

Expected outcome

+

+

+

+

+

Pos

72

0

66a

1

72

70

1

0

0

0

Neg

0

72

0

71

0

2

71

72

72

72

 

Spec/sens

100 %

100 %

100 %

99 %

100 %

99 %

99 %

100 %

100 %

100 %

Unknown 5 (2 % Bt11 maize)

Expected outcome

+

+

+

+

+

+

Pos

71

71

4

1

72

71

4

70

71

0

Neg

1

1

68

71

0

1

68

2

1

72

 

Spec/sens

99 %

99 %

94 %

99 %

100 %

99 %

94 %

97 %

99 %

100 %

Unknown 6 (0.1 % Bt11)

Expected outcome

+

+

+

+

+

+

Pos

72

72

1

0

72

71

2

71

69

1

Neg

0

0

71

72

0

1

70

1

3

71

 

Spec/sens

100 %

100 %

99 %

100 %

100 %

99 %

97 %

99 %

96 %

99 %

Unknown 7 (0.1 % GT73 rape)

Expected outcome

+

+

+

Pos

72

0

4

72

0

4

72

4

0

2

Neg

0

72

68

0

72

68

0

68

72

70

 

Spec/sens

100 %

100 %

94 %

100 %

100 %

94 %

100 %

94 %

100 %

97 %

Unknown 8 (0.9 % RF3 rape)

Expected outcome

+

+

+

+

Pos

65

1

4

66

1

66

1

2

1

66

Neg

1

65

62a

0

65

0

65

64

65

0

 

Spec/sens

98 %

98 %

94 %

100 %

98 %

100 %

98 %

97 %

98 %

100 %

Unknown 9 (2 % RF3 rape, 1 % GTS-40-3 soy)

Expected outcome

+

+

+

+

+

+

+

+

Pos

72

72

72

0

72

72

1

71

72

0

Neg

0

0

0

72

0

0

71

1

0

72

 

Spec/sens

100 %

100 %

100 %

100 %

100 %

100 %

99 %

99 %

100 %

100 %

Unknown 10 (0.1 % GT73 rape, 0.5 % MON810 maize)

Expected outcome

+

+

+

+

+

+

Pos

72

72

5

72

72

1

71

69

1

3

Neg

0

0

61a

0

0

71

1

3

71

69

 

Spec/sens

100 %

100 %

92 %

100 %

100 %

99 %

99 %

96 %

99 %

96 %

For each unknown, the GMO is indicated between brackets and the expected outcome is indicated as “+” when the marker is present, as “−” when absent. A sensitivity measure was calculated when a marker is expected to be present; a specificity measure when the marker is expected absent. Abbreviations for the markers are as described in Table 1

Applied formulae: sensitivity (%) = (Pos/(Pos + Neg)) × 100; specificity (%) = (1 − (Neg/(Pos + Neg)) × 100

aMissing date due to failure of the analysis or lack of materials

In Table 4, four parameters were calculated for the ten markers of the seven GM unknowns: sensitivity, specificity, positive and negative predictive value. In the case of GMO screening applications, especially the sensitivity and the negative predictive value are important. Indeed, as enforcement is including a positive GMO identification step, false positives due to high sensitivity will/can be corroborated in the second round of analyses where event-specific GMO detection methods are applied. The negative predictive value is crucial in assuring that no positives are being missed in the screening.
Table 4

Sensitivity, specificity, positive and negative predictive value of the PCR methods applied in CoSYPS based on the analysis of the GM unknowns of the inter-laboratory trial (after re-reading)

Expected outcome

Test outcome

Formula

ADH alt

STLM (lectin)

CRU

P35S short

Tnos

CP4 epsps

CryIAb

PAT

BAR

Overall

Pos

Pos

a

287

138

210

352

350

214

281

212

66

2,110

Pos

Neg

b

1

0

0

8

4

2

5

3

0

23

Neg

Pos

c

1

16

1

1

5

8

6

3

5

46

Neg

Neg

d

209

332

287

137

139

274

204

351

355

2,288

Sensitivity (%)

(a/(a + c)) × 100

99.7 %

100.0 %

100.0 %

97.8 %

98.9 %

99.1 %

98.3 %

98.6 %

100.0 %

98.9 %

Specificity (%)

(1 − (b/(b + d))) × 100

99.5 %

95.4 %

99.7 %

99.3 %

96.5 %

97.2 %

97.1 %

99.2 %

98.6 %

98.0 %

Positive Predictive Value (%)

(a/(a + b)) × 100

99.7 %

100.0 %

100.0 %

97.8 %

98.9 %

99.1 %

98.3 %

98.6 %

100.0 %

98.9 %

Negative Predictive value (%)

(1 − (c/(c + d))) × 100

99.5 %

95.4 %

99.7 %

99.3 %

96.5 %

97.2 %

97.1 %

99.2 %

98.6 %

98.0 %

Abbreviations for the markers are as described in Table 1

a denotes the number of analyses that are expected to contain the marker and scored positively, b is the number of analyses that are expected to contain the marker but scored negatively, c is the number of analysis that do not contain the marker but scored positively, and d number of analysis that do not contain the marker and scored negatively

It can thus be concluded that these SYBR®Green qPCR methods, when considered separately, perform appropriately for screening of GM events at levels as low as 0.1 % or estimated target copies of GM material present in a sample.

Evaluation of the Performance of the CoSYPS Decision Support System

The CoSYPS DSS provided to each participant laboratory generates the list of possible GM events present in the analyzed sample based on the combined outcome of the qPCR analyses. In Table 5, the DDS outcomes of the repetitions for each marker are listed. Upon correcting the analytical results as described above, the correct list of possible GM events present in the GM unknowns was obtained in 100 % of the analyses. Note that only in one case (Tnos for unknown U4), the combined analysis resulted in indicating the marker to be present “at LOD”, instead of “below LOD”. In such cases, follow-up analyses will be required to determine the origin of these minor signals.
Table 5

Overview of the analysis of the outcome obtained by the combined CoSYPS decision support per marker (n = 6) on all samples in the inter-laboratory trial

PER method

  

Plant

ADH alt

SLTM (lectin)

CRU

P35S short

Tnos

CP4 epsps

CryIAb

PAT

BAR

 
 

Total

83

83

83

83

83

83

83

83

83

83

 
 

Correct

83

83

83

83

83

82

83

83

83

83

 
 

wrong

0

0

0

0

0

1

0

0

0

0

 
 

%

100 %

100 %

100 %

100 %

100 %

99 %

100 %

100 %

100 %

100 %

 

PER sample

 

# Runs

Plant

ADH alt

SLTM (lectin)

CRU

P35S short

Tnos

CP4 epsps

CryIAb

PAT

BAR

% correct

Unknown 4

12

12

12

12

12

12

11

12

12

12

12

99 %

Unknown 5

12

12

12

12

12

12

12

12

12

12

12

100 %

Unknown 6

12

12

12

12

12

12

12

12

12

12

12

100 %

Unknown 7

12

12

12

12

12

12

12

12

12

12

12

100 %

Unknown 8

11

11

11

11

11

11

11

11

11

11

11

100 %

Unknown 9

12

12

12

12

12

12

12

12

12

12

12

100 %

Unknown 10

12

12

12

12

12

12

12

12

12

12

12

100 %

“Correct” means that the markers included in the unknown were indicated correctly as “above LOD” when present, and “at or below LOD” when absent, applying the decision criteria set in Materials and Methods. Abbreviations for the markers are as described in Table 1

Consequently, the outcomes of the DSS analysis of the CoSYPS screening results matched very well the expected results. Indeed, due to its strictly mathematical interpretation, based on prime-number factorization, this fully compliant screening analysis will give the predicted outcome at the GMO level in the CoSYPS decision support (Van den Bulcke et al. 2010).

Discussion

The results in the inter-laboratory trial demonstrate that the SYBR®Green qPCR methods used in CoSYPS system represent very sensitive GMO detection methods. All SYBR®Green qPCR methods used in the trial indeed detect their respective targets correctly (positive predicting values >97 %) while giving only very few to no false negative results (negative predicting values >95 %). When combined with a DSS, these SYBR®Green qPCR methods represent a versatile, highly performant and reliable GMO detection approach.

Although all participants scored the CoSYPS system as a simple, easy-to-use system and the provided protocols as very clear, a considerable number (about 5 %) of the analytical runs were not analyzed exactly as described in the protocol. Re-reading of the runs using the same software applied by the laboratories showed that the main deviations were 1) not setting the threshold according to the instructions in the provided protocol and 2) not setting the threshold for every method separately. While the first deviation mainly leads to too high C t values, the second one may also result in erroneous data. However, both deviating practices resulted in false positive and false negative outcomes. In a very few cases, no results were obtained due to a technical failure (not further specified).

After re-reading, the large majority of the false positives and negatives could be resolved. This indicates that CoSYPS, though considered an easy system, merits experimental and theoretical training before being applied in GMO analysis. Indeed, in a separate small-scale trial with only five laboratories who had received a theoretical and practical CoSYPS training previously, the obtained results were close to perfect (data not shown).

Overall, the SYBR®Green qPCR methods allowed to correctly detect three major crop markers for soy, maize and rapeseed, six main GMO markers (P35S, Tnos, CP4 epsps, CryIAb, PAT and BAR) and a generic plant marker (RbCL).

Both in the large- and the small-scale trials, the targets of some GMOs present at 0.1 % levels were reliably detected. Recalculated in absolute estimated copy numbers, in these trials the SYBR®Green qPCR methods detected as little as eight (CP4 epsps), 17 (BAR), 23 (P35S, Tnos and CryIAb) and 95 (PAT) copies in a total of 25 ng DNA extract. In separate reports, all methods were previously shown to efficiently detect less than 20 copies of their respective targets (Barbau-Piednoir et al. 2010; Barbau-Piednoir et al. 2012; Mbongolo Mbella et al. 2011). Here, in these inter-laboratory trials, several GMOs present at a 0.1 % mass level in the GM unknowns (e.g., MON810 maize and GT73 rapeseed) were efficiently detected. Thus, CoSYPS is highly suitable to screen for the presence of GMOs complying with the recent so-called “Low level Presence” EU regulation EU/619/2011 (European 2011).

Recently a number of reports have documented the inter-laboratory testing of a several probe-based multitarget qPCR systems, both simplex qPCR methods and multiplex qPCR methods (e.g., see Reiting et al. 2007; Waiblinger et al. 2008; Grohmann et al. 2009). These methods allow detecting GMOs as efficiently in test samples as the SYBR®Green qPCR methods used in CoSYPS. Due to the inherent need of a fluorescently labeled probe as the means of detection, the combination of three nucleotides in one test does in most cases not allow detecting multiple related GM targets within a single test. Although such a one-to-one correlation increases specific recognition at the DNA sequence level, it hampers broad-range screening approaches wherein related sequences can be detected by the same test (see, e.g., the SYBR®Green qPCR methods for CryIAb and CP4 epsps used in this study). In the latter, the number of assays to be included in a screening can be reduced. In the advent of new events comprising recombinant sequences with little to no homology with previously registered GMOs, the success of an “open” screening strategy may well be the preferred approach in GMO analysis. The GM marker set used in the CoSYPS setup applied in this 2009 inter-laboratory trial covers merely 42/48 (=88 %) of the currently EU authorized GMOs (according to the information in the EU Register of authorised GMOs at 27/02/2014; http://ec.europa.eu/food/dyna/gm_register/index_en.cfm). To cover the whole spectrum of GMOs falling under the current EU legislation for commercial applications, 18 screening methods would be required and more are expected to be needed in the future (Broeders et al. 2013).

Considering the variety in sequence nature of the targets of all SYBR®Green qPCR methods presented in this study (generic plant targets, plant species, recombinant sequences), the CoSYPS method development criteria and performance assessment (Barbau-Piednoir et al. 2010; Barbau-Piednoir et al. 2012; Van den Bulcke et al. 2010; Mbongolo Mbella et al. 2011) provides without any doubt a valuable framework for further extending the set of GMO screening qPCR methods operating under the same experimental conditions (see, e.g., Block et al. 2013). It is likely though that in the future larger PCR formats may have to be introduced, simplex PCR methods will have to be combined by multiplexing, or GMO screening applying other technologies such as digital PCR or Luminex may need to be established taking also cost into consideration (Randhawa et al. 2014; Shao et al. 2014; Brod et al. 2014; Huber et al. 2013; Cottenet et al. 2013). Finally, genome-wide-sequencing/exome-sequencing approaches may ultimately become feasible when cost and reliability of this technology are at the level affordable and required by enforcement measures (Wahler D 2999). For the coming years though, real-time PCR matrix-based screening will still play a central role in GMO enforcement testing. The methods and perhaps even more the molecular targets applied in CoSYPS could represent a starting point in the development as well as the application of GMO screening approaches with these new technologies.

Notes

Acknowledgments

We thank all members of the laboratories that participated at either of the two inter-laboratory trials in this study: Voedsel en Waren Autoriteit (Nederland), Biogeves-France, Finnish Customs Laboratory (Finland), Institut fur Hygiene und Umwelt (Germany), Centro Nacional de Alimentacion (Spain), Danish Plant Directorate (Denmark), Istituto Superiore di Sanità (Italy), German National Reference Laboratory for GMO (Germany), Federal Office of Public Health (Switzerland), Talinn University of Technology (Estonia), Institute for Diagnosis and Animal Health (Roumania), National Center of Public Health Protection (Bulgaria) and Instituto Nacional de Engenharia, Tecnologia e Inovação (Portugal). We also thank our colleagues at the Institute of Public Health (Brussels, Belgium) and at the Walloon Agricultural Research Centre (Gembloux, Belgium) for the use of some of their equipment in the testing of the feasibility of the CoSYPS on different PCR devices. We thank Els Vandermassen and Dirk Van Geel for their valuable help in preparing and testing some of the reference materials.

Conflict of Interest

Elodie Barbau-Piednoir has no conflict of interest. Pieter Stragier has no conflict of interest. Nancy Roosens has no conflict of interest. Marco Mazzara has no conflict of interest. Cristian Savini has no conflict of interest. Guy Van den Eede has no conflict of interest. Marc Van den Bulcke has no conflict of interest. This article does not contain any studies with human or animal subjects.

References

  1. Barbau-Piednoir E, Lievens A, Mbongolo-Mbella G, Roosens N, Sneyers M, Leunda-Casi A et al (2010) SYBR®Green qPCR screening methods for the presence of "35S promoter" and "NOS terminator" elements in food and feed products. Eur Food Res Technol 230(3):383–393CrossRefGoogle Scholar
  2. Barbau-Piednoir E, Lievens A, Vandermassen E, Mbongolo-Mbella EG, Leunda-Casi A, Roosens N et al (2012) Four new SYBR(r)Green qPCR screening methods for the detection of Roundup Ready(r), LibertyLink(r), and CryIAb traits in genetically modified products. Eur Food Res Technol 234(1):13–23CrossRefGoogle Scholar
  3. Block A, Debode F, Grohmann L, Hulin J, Taverniers I, Kluga L, Barbau-Piednoir E, Broeders S, Huber I, Van den Bulcke M, Heinze P, Berben G, Busch U, Roosens N (2013) The GMOseek matrix: a decision support tool for optimizing the detection of genetically modified plants. BMC Bioinforma 14:256. doi: 10.1186/1471-2105-14-256 CrossRefGoogle Scholar
  4. Brod FC, van Dijk JP, Voorhuijzen MM, Dinon AZ, Guimarães LH, Scholtens IM, Arisi AC, Kok EJ (2014) A high-throughput method for GMO multi-detection using a microfluidic dynamic array. Anal Bioanal Chem 406(5):1397–1410. doi: 10.1007/s00216-013-7562-1 CrossRefGoogle Scholar
  5. Broeders S, Barbau-Piednoir E, Vandermassen E, Debode F, Mazzara M, Roosens N (2013) New SYBRGreen methods targeting promoter sequences used for screening of several GM events pending for authorisation in Europe. EFRT 236 537Google Scholar
  6. Cottenet G, Blancpain C, Sonnard V, Chuah PF (2013) Development and validation of a multiplex real-time PCR method to simultaneously detect 47 targets for the identification of genetically modified organisms. Anal Bioanal Chem 405(21):6831–6844. doi: 10.1007/s00216-013-7125-5 CrossRefGoogle Scholar
  7. European Parliament (2011) Commision Regulation (EU) 619/2011 of 24 June 2011 laying down the methods of sampling and analysis for the official control of feed as regards presence of genetically modified material for which an authorisation procedure is pending or the authorisation of which has expired. Official Journal of the European UnionGoogle Scholar
  8. Fernandez S, Charles-Delobel C, Geldreich A, Berthier G, Boyer F, Collonnier C et al (2005) Quantification of the 35S promoter in DNA extracts from genetically modified organisms using real-time polymerase chain reaction and specificity assessment on various genetically modified organisms: Part I. Operating procedure. J AOAC Int 88(2):547–557Google Scholar
  9. Grohmann L, Brunen-Nieweler C, Nemeth A, Waiblinger HU (2009) Collaborative trial validation studies of real-time PCR-based GMO screening methods for detection of the bar gene and the ctp2-cp4epsps construct. J Agric Food Chem 57:8913CrossRefGoogle Scholar
  10. Holst-Jensen A, Bertheau Y, De Loose M, Grohmann L, Hamels S, Hougs L et al (2012) Detecting un-authorized genetically modified organisms (GMOs) and derived materials. Biotechnol Adv 30:1318–1335CrossRefGoogle Scholar
  11. Huber I, Block A, Sebah D, Debode F, Morisset D, Grohmann L, Berben G, Stebih D, Milavec M, Zel J, Busch U (2013) Development and validation of duplex, triplex, and pentaplex real-time PCR screening assays for the detection of genetically modified organisms in food and feed. J Agric Food Chem 61(43):10293–10301. doi: 10.1021/jf402448y CrossRefGoogle Scholar
  12. James C (2012) ISAAA Brief 44–2012 Global Status of Commercialized Biotech/GM CropsGoogle Scholar
  13. Lee SH, Kim JK, Yi BY (2007) Detection methods for biotech cotton MON 15985 and MON 88913 by PCR. J Agric Food Chem 55(9):3351–3357CrossRefGoogle Scholar
  14. Leimanis S, Hernandez M, Fernandez S, Boyer F, Burns M, Bruderer S et al (2006) A microarray-based detection system for genetically modified (GM) food ingredients. Plant Mol Biol 61(1–2):123–139CrossRefGoogle Scholar
  15. Mbongolo Mbella EG, Lievens A, Barbau-Piednoir E, Sneyers M, Leunda-Casi A, Roosens N et al (2011) SYBR®Green qPCR methods for detection of endogenous reference genes in commodity crops: a step ahead in combinatory screening of Genetically Modified Crops in food and feed products. Eur Food Res Technol 232(3):485–496CrossRefGoogle Scholar
  16. Morisset D, Dobnik D, Hamels S, Zel J, Gruden K (2008) NAIMA: target amplification strategy allowing quantitative on-chip detection of GMOs. Nucleic Acids Res 36(18):e118CrossRefGoogle Scholar
  17. Newcombe RG (1998) Two-sided confidence intervals for the single proportion: comparison of seven methods. Stat Med 17(8):857–872CrossRefGoogle Scholar
  18. Randhawa GJ, Morisset D, Singh M, Žel J (2014) GMO matrix: a cost-effective approach for screening unauthorized genetically modified events in India. Food Control 38:124–129CrossRefGoogle Scholar
  19. Reiting R, Broll H, Waiblinger HU, Grohmann L (2007) Collaborative study of tNOS rtPCR method. J Verbr Lebensm 2:116CrossRefGoogle Scholar
  20. Shao N, Jiang SM, Zhang M, Wang J, Guo SJ, Li Y, Jiang HW, Liu CX, Zhang DB, Yang LT, Tao SC (2014) MACRO: a combined microchip-PCR and microarray system for high-throughput monitoring of genetically modified organisms. Anal Chem 86(2):1269–1276. doi: 10.1021/ac403630a CrossRefGoogle Scholar
  21. Ujhelyi G, Dijk JP, Prins TW, Voorhuijzen MM, Hoef AM, Beenen HG et al (2012) Comparison and transfer testing of multiplex ligation detection methods for GM plants. BMC Biotechnol 12:4CrossRefGoogle Scholar
  22. Van den Bulcke M, Lievens A, Barbau-Piednoir E, Mbongolo Mbella G, Roosens N, Sneyers M et al (2010) A theoretical introduction to "Combinatory SYBR®Green qPCR Screening", a matrix-based approach for the detection of materials derived from genetically modified plants. Anal Bioanal Chem 396(6):2113–2123CrossRefGoogle Scholar
  23. Van den Eede G (2010) Compendium of reference methods for GMO analysis, EUR 24526 EN ISBN 978-92-79-15627-4 ISSN 1018–5593 doi  10.2788/16745 Luxembourg, OPOCE
  24. Wahler D, Schauser L, Bendiek J, Grohmann L Next-generation sequencing as a tool for detailed molecular characterisation of genomic insertions and flanking regions in genetically modified plants: a pilot study using a rice event unauthorised in the EU. Food Anal Methods DOI  10.1007/s12161-013-9673-x
  25. Waiblinger HU, Ernst B, Anderson A, Pietsch K (2008) Validation and collaborative study of P35S and tNOS duplex rtPCR. Eur Food Res Technol 226:1221CrossRefGoogle Scholar
  26. Waiblinger HU, Grohmann L, Mankertz J, Engelbert D, Pietsch K (2010) A practical approach to screen for authorised and unauthorised genetically modified plants. Anal Bioanal Chem 396(6):2065–2072CrossRefGoogle Scholar
  27. Wolf C, Scherzinger M, Wurz A, Pauli U, Hübner P, Lüthy J (2000) Detection of cauliflower mosaic virus by the polymerase chain reaction: testing of food components for false-positive 35S-promoter screening results. Eur Food Res Technol 210:367–372CrossRefGoogle Scholar

Copyright information

© The Author(s) 2014

Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.

Authors and Affiliations

  • Elodie Barbau-Piednoir
    • 1
  • Pieter Stragier
    • 1
  • Nancy Roosens
    • 1
  • Marco Mazzara
    • 2
  • Cristian Savini
    • 2
  • Guy Van den Eede
    • 2
  • Marc Van den Bulcke
    • 1
    • 2
  1. 1.Scientific Institute of Public HealthBrusselsBelgium
  2. 2.European Commission Joint Research Centre, Molecular Biology and Genomics UnitInstitute for Health and Consumer ProtectionIspraItaly

Personalised recommendations