Eur Food Res Technol (2003) 216:347354 DOI 10.

1007/s00217-002-0653-4

ORIGINAL PAPER

Sissel B. Rnning Marc Vatilingom Knut G. Berdal Arne Holst-Jensen

Event specific real-time quantitative PCR for genetically modified Bt11 maize (Zea mays)
Received: 8 September 2002 / Revised: 4 November 2002 / Published online: 16 January 2003  Springer-Verlag 2003

Abstract We report the characterisation of the 3integration junction between host plant DNA and integrated gene-construct of the genetically modified Bt11 maize, and the successive development and first validation of a transformation event-specific quantitative TaqMan 5-nuclease PCR assay for detection and quantitation of Bt11 maize on the LightCycler. The transgenic DNA in Bt11 is inserted in a tandem repeated DNA sequence motif of approximately 180 bp, and we discuss some of the problems this creates for design, development and validation of event-specific PCR assays. The limit of detection of the PCR assay was estimated to 10 initial template copies. The limit of quantitation in pure maize materials was estimated to approximately 40 template copies. In food samples containing high levels of exogenous DNA, the impact of this DNA was estimated to increase the limit of quantitation to approximately 100 initial template copies. Because of the low number of template copies required for detection and quantification, in combination with small size of the amplicon (~100 bp), the assay may be applied to analyses of raw materials as well as processed food and feed. Keywords Event-specific PCR Real-time PCR Genetically modified organism GMO Bt11 maize Detection and quantitation limit Integration site

Introduction
At the present time, 19% of the worlds planted maize (Zea mays) acreage is genetically modified (GM) [1].
S. B. Rnning K. G. Berdal A. Holst-Jensen ()) Section of Food and Feed Microbiology, National Veterinary Institute, Ullevlsveien 68, P.O. Box 8156 Dep., 0033 Oslo, Norway e-mail: arne.holst-jensen@vetinst.no Tel.: +47 23 21 62 43 Fax: +47 23 21 62 02 M. Vatilingom TEPRAL, 68 route dOberhausbergen, 67200 Strasbourg, France

Consumers are concerned about the use of genetically modified plants in food production, and as a consequence, labelling of GM food has become an important issue. The principles of European legislation on GM foods, as laid down in EC regulations: 258/97 [2], 1139/98 [3], 49/2000 [4] and 50/2000 [5], is that a product shall be mandatory labelled if it consists of a GM organism (GMO) or if it is derived from a GMO but no longer contains the GMO and there is still modified DNA or protein present. The legal threshold value for labelling, set down by the European Commissions Standing Committee in 1999, is 1%. Methods to detect GMO-derived DNA can be divided into four categories; screening, gene-, construct- and event-specific, and their ability to discriminate between GM- and non-GM-derived DNA varies accordingly [6]. The screening methods are associated with a particular risk of yielding false positives [7]. The methods for detection of the modified genes are somewhat more specific, although the modified genes may be used in several GMO and with variable copy number, hence reducing the specificity and accuracy for quantitation. Construct-specific methods are normally specific for GMO, and the risk of reporting false positives is significantly lower than with screening and gene-specific methods. However, construct-specific methods cannot distinguish between different GMO if the same construct has been integrated. Furthermore, the number of integrated copies of the PCR target sequence may vary between different GMO containing the same target sequence. The introduction of new GMO in the future, containing many of the same genetic elements, will reduce the specificity of the construct-specific methods. For example, Mon809 and Mon810 maize have been transformed with the same plasmid, PV-ZMBK07, containing a PE35Shsp70intronCryIA(b)Tnos construct [8]. To overcome specificity problems, a line or transformation event-specific (hereafter termed event-specific) PCR may be performed. Since, with currently used transformation techniques, the integration of an introduced gene-construct into the genome of a plant is unlikely to occur twice at the same genomic locus, the

348 Table 1 Primers and probes used for characterisation of the 3integration- junction and for real-time PCR quantitation of Bt11 content PCR assay Primer/probea Name Sequence of oligonucleotide 53 CCA GTT AGG CCA GTT ACC CAG AT AAA CCT TGC GCC TCC ATA GAC TT TGA TTA GAG TCC CGC AAT TAT ACA TT AAC CTT TGA TGC CTA TGT G CAC TCC ATC GTG GAG AGC TT GGC GTT GTT GAA GAG GAA GA TAC CCC ACA CGA GCC ATC TAC GAC T GCG GAA CCC CTA TTT GTT TA TCC AAG AAT CCC TCC ATG AG AAA TAC ATT CAA ATA TGT ATC CGC TCA TAT CAT CGA CTT CCA TGA CCA CTT TGA TCT TTT CTA CGG GGT CT TTT TCG TTC CAC TGA GCG TCA GAC C Size (bp)

Primers for characterisation of Bt11 Inverse PCR Forward Pat/Bar3 First PCR Reverse Pat/Bar5 Inverse PCR Forward Nos2 Nested PCR Reverse Pat/Bar1 Reference gene primers and probe for real-time PCR Maize invertase gene Forward ZmayInvFor Reverse ZmayInvRev Probe ZmayInvFT Bt11 specific primers and probes for real-time PCR Bt11 3 junction Forward Bt113JFor Reverse Bt113JRev Probe Bt113JFT Bt11 5 junction Forward Bt111b Reverse Bt115JRev Probe Bt115JFT
a The b

~1 kb2 kb

111

70

82

probes were labelled with 5-FAM and 3-TAMRA With exception for primer Bt111 developed by Zimmermann et al [13], the primers and probes were developed as part of this study

junctions between the integrated DNA sequence and the host DNA are unique for a single transformation event. The use of a PCR system targeting this integration border will be event-specific because the assay is specific for both the transgenic construct and the integration site. Furthermore, since a single locus is targeted, eventspecific methods (in contrast to other methods) will have few copy number variation problems. Until now, eventspecific quantitative methods have been published for Roundup Ready soybean (RRS) [9, 10, 11] and Mon810 maize [12] using real-time PCR, and for Bt11 maize [13] using competitive PCR. A major portion of the transgenic crops will be marketed as processed products (e.g. meat mixed with soy protein from GM soybeans). The applicability of the PCR methods for GMO detection in processed foods is determined by the quality of the DNA still present in the final product. Physical and chemical parameters may lead to random cleavage and/or removal of the genomic DNA. With increased processing, the size of the genomic DNA-fragments decreases. Consequently, the amplicon size in an analysis should be minimised to obtain the highest achievable detectability. This will make it more difficult to detect the presence of GMO with conventional agarose gel electrophoresis. Furthermore, quantitation of the amplified DNA is difficult if image analysis is combined with agarose gel visualisation and small amplicons. To overcome the size problem, a DNA hybridisation technique is commonly applied for detection, identification and quantitation [9, 10, 11, 12, 14, 15]. This paper describes the characterisation of the 3event-specific integration junction of Bt11 and the successive development and first validation of primers and TaqMan probes [16] for quantitative real-time PCR assays targeting the event-specific upstream and downstream integration junctions of Bt11 on a LightCycler

(Roche) real-time thermal cycler. The maize endogenous invertase gene is used as a reference gene.

Materials and methods

Characterisation of the 3-integration junction The 3-integration junction was characterised using a modified version of the inverse PCR protocol originally described by Hartl et al [17]. Digestion and ligation of genomic DNA Consejo Superior de Investigaciones Cientficas (CSIC, Barcelona, Spain) provided CsCl2 purified and ethanol precipitated DNA from 100% Bt11 homozygous leaf material for the characterisation of the 3-integration junction. Genomic Bt11 DNA (1.5 g) was digested using 20 units of EcoRI, XhoI and HindIII (Promega) for 3 h at 37 C in a total volume of 50 L in three separate tubes, and successively purified on a silica gel column (Qiagen). One volume of digested DNA was added to 5 volumes of PB buffer (Qiagen), transferred to the spin column and centrifuged for 1 min at 15 000g. After washing twice with 700 L of PE buffer, the DNA was eluted with 100 L of EB buffer (10 mM Tris-HCl, pH 8.0). The digested DNA was self-ligated [17] in order to get small circularised genomic DNA as follows: ten units of T4 DNA ligase (Promega) was used to ligate 500 ng of digested DNA in a total volume of 500 L for 24 h at 4 C. The DNA was purified as described above using a Qiaquick DNA cleanup kit (Qiagen), and eluted in a final volume of 50 L. Inverse PCR The ligated product was amplified using a modified inverse PCR protocol [13] to obtain enough material for sequencing. Inverse PCR reactions were performed on a GeneAmp 2400 thermal cycler (Applied Biosystems), in 50 L final volume, containing 1x reaction buffer (Qiagen), 1.5 mM MgCl2, 0.2 mM dNTP, 0.2 M of each primer and 2.5 units of Taq DNA polymerase (Qiagen) in a nested approach. The first amplification contained 2 L of previously purified ligated DNA and the primers Pat/Bar5 and

349 Pat/Bar3 (Table 1), using a PCR program consisting of a first 95 C denaturation step for 2 min followed by 25 cycles of 30 s denaturation at 95 C, 30 s annealing at 56 C and 1 min 30 s extension at 72 C. The second (nested) amplification contained 2 L of the amplification product from the first PCR reaction and the primers Pat/Bar1 and Nos2 (Table 1), using a PCR program consisting of a first 95 C denaturation step for 2 min followed by 35 cycles of 30 s denaturation at 95 C, 30 s annealing at 52 C and 1 min 30 s extension at 72 C. All the PCR products were separated by electrophoresis on 1.5% agarose gel (TBE buffer). The fragments were successively isolated from the gel by cutting the gel under UV-light and purifying using Qiaquick Gel extraction kit (Qiagen). The fragments were cycle sequenced with Big Dye terminator v.3 (Applied Biosystems) on an ABIprism 310 (Applied Biosystems). Design and validation of a detection assays Maize DNA samples and extraction DNA was extracted from certified reference materials (CRMs) produced by the Institute for Reference Materials and Measurements (IRMM, Geel, Belgium) and commercialised by Fluka, using a CTAB-based protocol [18]. The Bt11 CRMs (IRMM-412) are a mixture of dried heterozygous Bt11 grain and non-GM maize, produced on a weight:weight basis. CSIC provided CsCl2 purified and ethanol precipitated DNA from 100% Bt11 homozygous leaf material for the construction of calibration curves. In addition, DNAs extracted from CRMs of Bt11 (IRMM-412), Bt176 (IRMM-411), Mon810 (IRMM-413) and RRS (IRMM-410R), and from the cloning vector pUC18 (Amersham Biotech) were used as templates in order to validate the TaqMan assays. DNA concentrations were measured in ng/L using a SYBR Green method [9]. Primers and probes Different primer sets with corresponding probes were designed for the 3-integration junction between the host DNA and the inserted gene construct. In addition, the event-specific detection system for the 5-junction described by Zimmermann et al. [13] was modified to fit a real-time PCR assay by designing a TaqMan probe and new reverse primers. The primers and probes were designed using the software program OLIGO-Applet [19], specifying an optimal melting temperature (Tm) of 60 C for the primers, and 70 C for the probes. The full-length maize endogenous invertase gene [20] (EMBL/GenBank accession number U16123) was used as input sequence for the development of primers and a corresponding probe for a reference gene. The probes were labelled with 5-FAM and 3-TAMRA. All the primers and probes were purchased from DNA Technology (rhus, Denmark). PCR conditions The PCR reaction volume of 30 L contained 5 L template DNA, 0.75 M of each primer, 0.25 M probe, 1.5 units Taq DNA polymerase (AmplitaqGold, Applied Biosystems), 0.3 units uracilN-glycosylase (AmpErase UNG, Applied Biosystems), 0.1 mg/mL BSA, 0.2 mM dATP, dCTP and dGTP (Applied Biosystems) and 0.4 mM dUTP (Applied Biosystems), 4 mM MgCl2 and 1x PCR buffer (100 mM Tris-HCl, pH 8.3, 500 mM KCl). The eventspecific target sequence and the maize endogenous invertase gene were amplified in separate capillaries, and the fluorescence was measured in channel 1 of the LightCycler and analysed with software version 3.5. The PCR-program included an initial decontamination step for 2 min at 50 C to allow optimal UNG enzymatic activity, followed by 10 min at 95 C in order to activate the DNA polymerase, to deactivate the UNG and to denature the double-stranded template; and successively 50 cycles of 15 s To determine the LOD and LOQ, 2% Bt11 CRM (IRMM-412) was used. Due to heterozygosity this material contained approximately 100 copies of the maize genome per copy of the Bt11 event-specific target sequence. The DNA was diluted in a fourfold dilution series and each concentration was analysed in four parallels. In order to estimate the impact of exogenous background DNA and to mimic the situation in real food samples, a background of 20 000 soybean genome equivalents (IRMM-410R) per 5 mL template DNA was added to each of the dilutions of the 2% Bt11 CRM. denaturation at 95 C and 30 s synthesis at 60 C. The same reaction conditions and solutions were used for the maize endogenous invertase gene. Calibration curves The haploid genome size of maize has been estimated to be within the range of 22922716 Mbp, corresponding to a molecular weight of ~2.5 pg for the haploid genome [21]. Calibration curves for the amplicons were established with fourfold dilutions of DNA from 100% Bt11 material (containing approximately 20 000 to 5 template copies per PCR reaction). Estimation of the limit of detection (LOD) and the limit of quantitation (LOQ)

Results and discussion

Bt11 transgene 3-integration junction characterisation In order to obtain the Bt11 integration 3-junction fragment, a modified inverse PCR method based on the 5-junction characterisation method described by Zimmermann et al. [13] was used. The inverse PCR primers annealed between the Pat/Bar gene and the NOS terminator (Fig. 1). After the first amplification reaction with the primers Pat/bar3 and Pat/bar5, no PCR products were detectable with agarose gel electrophoresis. Consequently, a nested PCR with 35 cycles was performed using the primers Pat/bar1 and Nos2. Two PCR products of approximately 1 and 2 Kb length were amplified with EcoRI and XhoI digested DNA, respectively (Fig. 2). With HindIII digested DNA, no amplification product was obtained. The two fragments were sequenced and were 100% homologous. A part of the consensus alignment including the transgene 3-integration junction is shown in Fig. 3. The consensus sequence has been submitted to EMBL/GenBank under accession number AY123624. Development of real-time PCR assays Different primer sets for both the 5- and 3-junctions were subjected to a first qualitative PCR test using DNA extracted from 2% Bt11 CRM. For each junction, the primer pairs that produced the single band with highest intensity were selected to develop a quantitative Bt11 real-time PCR assay (Table 1). A map of the Bt11 5- and 3-junctions and the location of the primers and corresponding probes is shown in Fig. 3. We selected primer-

350

Fig. 1 Schematic representation of the Bt11 transgenic map. The primers used for the 3 junction characterisation are indicated with arrows. Primers for the first PCR (Pat/bar3 and Pat/bar5) are shown

as open arrows and primers for the second (nested) PCR (Pat/bar1 and Nos2) are shown as closed arrows. Reverse primers restriction sites for EcoRI, HindIII and XhoI are indicated

Fig. 2 Inverse PCR amplification results with Bt11 DNA digested with EcoRI (E1 to E6, black arrow) and XhoI (X1 to X6, white arrow). Six independent parallel experiments were performed for each enzyme (E1-E6 and X1-X6). L Low mass ladder (Gibco BRL), N PCR negative control

sets yielding amplicons of 111 bp for the endogenous invertase gene, and 82 bp and 70 bp for the 5- and 3event-specific junctions, respectively. For all the PCR reactions, the same thermal PCR program was used. In order to optimise the PCR conditions we varied the MgCl2 concentration within the range of 26 mM. For both the 5- and 3-junction assays, 4 mM MgCl2 yielded the best amplification (lowest Ct values). Further optimisation and testing indicated that the 5-assay was less specific than the 3-assay. The 3-assay was therefore chosen as our primary target for further design and validation of a real-time quantitative assay. Calibration curves Two parallels of each fourfold dilution (corresponding to approximately 20 000 to 5 genome equivalents of 100% Bt11 DNA per PCR) were used to establish calibration curves for the invertase and the 3-assay (Fig. 4). The square regression coefficients (R2) were 0.9973 and 0.9986 for the invertase and the Bt11 amplicons, respectively. Good linearity between copy number and fluorescence values (Ct) visualised in the calibration curves in Fig. 4, i.e. high agreement between the amount of

Fig. 3 Sequence of the 5- and 3-integration junction regions and location of the primers and the TaqMan probes for the eventspecific amplification and detection of Bt11 maize. The asterisks and arrows indicate the exact junctions between the host-plant genomic DNA and the inserted DNA, and the amplicons are presented as double-stranded sequences in bold letters. The names and the direction of the primers (normal types) and probes (italics) are indicated. The maize endogenous DNA sequences are shown in small italics, while the insert DNA is shown in capitals. Top: The 5-junction amplicon and the location of the primers and probe. The upstream primer (Bt111) spans the event-specific 5-junction and is partially derived from the plant DNA (11 nt) and partially from the insert sequence. Bottom: The 3-integration junction and location of the primers and probe. The downstream primer (Bt113JRev) spans the event-specific 3-junction and is partially derived from plant DNA (9 nt) and partially from insert sequence

template (ln input DNA) and amount of product (Ct values), indicates that this assay is well suited for quantitative measurements. Accuracy and specificity of the method In order to determine the accuracy (precision and trueness) of the quantitative 3-assay we used DNA extracted from CRMs of relative concentrations of 0, 1.0 and 2.0% Bt11 as template in the PCR reactions. Each template was analysed in five parallels, and the quantitative estimates were computed using standard curves made from dilution series of known amounts of invertase and Bt11. No amplification signal was observed with DNA from 0% Bt11, while estimates from 1% Bt11 CRM ranged from 0.71% to 1.21% with a mean of 0.99% and a S.D. of 0.22%, and estimates from 2% Bt11 CRM ranged

351 Fig. 4AD Amplification and calibration curves for the maize endogenous invertase gene and the Bt11 event-specific 3junction were established using diluted DNA from 100% Bt11 material. The dilution series contained approximately 20 000, 5000, 1250, 312, 78, 19 and 4.9 initial template copies in the PCR reactions. A Amplification curves for the maize endogenous invertase gene, analysed using Fit Points on the LightCycler. B Calibration curves plotting ln (input DNA) versus Ct. C Amplification curves for the Bt11 event-specific 3junction, analysed using Fit Points on the LightCycler. D Calibration curves for Bt11 plotting ln (input DNA) versus Ct

352

from 1.13% to 2.22% with a mean of 1.56% and a S.D. of 0.48%. The precision (RSDr=0.22) and trueness (error=1.0%) were better for 1% Bt11 CRM than for 2% Bt11 CRM (RSDr=0.31 and error 22%) The DNA from the 2% Bt11 we used was extracted from recently purchased CRM (June 2001), while the DNA from the 1% material was extracted from material purchased in April 2001. The IRMM has recently performed stability testing of the CRMs. With Bt11 and Bt176 maize, degradation problems have been discovered on the material from May 2001 to February 2002. Consequently, the IRMM-411 and IRMM-412 series were withdrawn in February 2002. DNA degradation, although not sequencespecific, leads to reduction in the amount of extracted DNA and consequently to a reduction in available amplifiable DNA. Low concentration of the target DNA leads to higher variation among parallel runs than with high concentration. This may explain some of the relatively low precision and trueness we observed with the 2% CRM. DNA from soybean, Mon810 and Bt176, was used in order to determine the specificity of both the qualitative and quantitative 3-assay. With these template DNAs no signal was detected on the LightCycler after 50 cycles. However, when the reaction products from maize DNA were analysed on agarose gel, amplification products were occasionally observed with variable signal intensity. Unspecific amplification products were never observed when the PCR was limited to 35 cycles. A homology search (blastn) against the EMBL/GenBank database, (http://www.ncbi.nlm.nih.gov/BLAST/), using the 3event-specific junction sequence of Bt11 as probe, retrieved a maize-specific 180 bp knob-specific repeat with high homology (EMBL/GenBank accession number AF030935) [22]. The reason for the observation of false positives on agarose gel may be that the Bt11 eventspecific primers occasionally amplify parts of this repeated region. However, with a TaqMan assay this problem is minimised due to the high specificity of the TaqMan probe. The use of a PCR system covering the integration border, amplifying parts of both the host DNA and the transgenic construct, enables the best and most specific PCR system for identification and detection of GMO. Ideally, an event-specific detection system shall consist of one primer located in the transgenic construct and one located in the host DNA and an event-specific probe that spans the junction. The Bt11 gene-construct is integrated into a maize-specific 180 bp knob-specific repeat. Consequently, we expected and observed multiple amplification products with gel electrophoresis when a primer was entirely located in the repeat motif due to priming within several repeats. To overcome these problems, the primers designed in our detection systems span the junctions, while the probes are placed inside the transgenic construct. Although, this is not ideal, we believe this was the best way to obtain high specificity of the Bt11 event-specific assay. The amplicon for the 3-junction assay, with the exception of 9 nucleotides (nt) at the 5-end of the reverse

primer, is 100% similar to pUC18 (EMBL/GenBank accession number L08752). From this, we predicted that pUC18-derived amplification products would be generated at least under sub-optimal reaction conditions. In order to determine the specificity of the assay with respect to the pUC18 sequence, we used the pUC18 vector as a template with the Bt11 event-specific 3-junction assay. As we expected, a positive signal was detected on the LightCycler when high concentrations of pUC18 were used as template. There is a significant risk of reporting false positives in the presence of concentrations of pUC18-derived DNA corresponding to the limit of detection of pUC18 (1109 template copies with a Ctvalue above 40). Such concentrations could be due to the presence of another GMO containing pUC18 vector sequences, bacteria transformed with pUC18, or by vector contamination of the polymerase. It is normally not possible to obtain such concentrations of pUC18 in a food or feed sample, unless it is highly contaminated with e.g. genetically modified microorganisms. Consequently, we believe this is an unlikely situation in routine analysis. It is important to realise that a Ct-value >40 usually means that the PCR reaction is inhibited or that the target is absent. Determination of LOD and LOQ There are at least three different ways of expressing detection and quantitation limits, although they all refer to the lowest quantity of the target that reliably can be detected and quantified with a probability of 95% [9]. The absolute limit is the lowest number of initial template copies that can be detected and quantified, the relative limit refers to the lowest percentage of GMO relative to the species (e.g. maize) that can be detected and quantified, and the practical limit is the functional limit of the sample during an analysis. In order to determine the LOD and LOQ of the 3-assay, a fourfold dilution series of 2% Bt11 CRM DNA containing an estimated average of 320, 80, 20, 5 and 1.25 copies of the Bt11 genome per PCR were analysed in four parallel real-time PCR analyses (Table 2 and Fig 5). As expected, the ability to detect Bt11 decreased with decreasing copy numbers. We were able to detect Bt11 in all the four parallels down to 20 copies, while only three of the parallels were positive when we used an estimated average of 5 copies, and only one of the parallels was positive when we used an estimated average of 1.25 copies in the PCR. These results are within the range predicted by a Poisson distribution model and an LOD of approximately 10 copies. The lowest number of copies we were able to detect was estimated to 12 initial template copies. From the results, we estimate the absolute LOD to be approximately 10 copies of the Bt11 3-junction [9]. The experiments show that the Ct variations among parallels of the same template concentration increase with decreasing copy number. To obtain reliable quantitation results under ideal conditions, approximately 40 initial

353

Fig. 5 Estimation of limit of detection (LOD) and quantitation (LOQ). Amplification curves for Bt11 based on fourfold dilutionseries with an estimated average of 320, 80, 20, 5 and 1.25 initial template copies of Bt11 per PCR, analysed using second derivate

maximum on the LightCycler. The Bt11 is detected in all the four parallels down to 20 template copies, but only in three of four parallels at 5 copies and one of four parallels at 1.25 copies

Table 2 Amplification dataa used to determine the absolute LOD and LOQ Estimated template copies 320 80 20 5 1.25
a

Signal rate (#positive signals) 4/4 4/4 4/4 3/4 1/4

Mean Ct-values 33.22 34.30 36.33 37.92 38.63

SD of observed Ct-values 0.32 0.45 0.73 1.20 n.a.

Fourfold dilutions series of 2% Bt11, based on initial template copy numbers n.a. Not applicable

Table 3 Quantitation limit with 20 000 soybean genome equivalents as background in the PCR reaction, using 2% Bt11 and 2% RRS as template in the reaction Initial template copies 80 80+Soybean DNA 60 60+Soybean DNA 40 40+Soybean DNA
a

A 33.23 33.44 34.3 34.56 34.31 35.19

B 33.52 33.78 33.84 34.78 34.83 36.05

C 33.71 33.67 34.04 35.09 34.92 34.30

D 33.60 33.85 34.66 33.84 34.77 35.56

Mean 33.52 33.69 34.21 34.57 34.71 35.28

SD 0.21 0.18 0.35 0.53 0.27 0.74

Influence of background DNA (D Ct)a 0.17 0.36 0.57

Relative to the Ct value observed without soybean DNA in the reaction

template copies are required. Consequently, we conclude that the absolute LOQ with two parallel PCR reactions is 40 initial template copies. The estimated LOQ value is, however, only valid under exceptionally optimal conditions, i.e. with high molecular size DNA in pure maize samples. For real food and feed samples, the LOQ value may differ from the estimated LOQ using high concentration and high purity DNA, e.g. because of large background of

non-target DNA. In order to mimic a situation more similar to that in many food and feed samples and to estimate the impact of a high exogenous DNA background on the LOQ value, we made a new dilution series by mixing ~20 000 soybean genome equivalents (~20 ng) with approximately 80, 60 and 40 copies of 2% Bt11 CRM (Table 3). The large background of soybean-derived DNA had a major influence when few copies of Bt11 were present in the sample (observed values roughly

354

corresponding to the effect of removing approximately 1020 copies of Bt11, which with <80 copies corresponds to >12,5% of the Bt11 copies). By extrapolation of the observed effect on Ct-values of adding the soybean DNA, we estimate the effect to be insignificant with 100 copies or more (cf. Table 2). The fluctuation of the curves was also altered by the addition of soybean DNA, yielding a significant increase in standard deviation with lower copy numbers and decreasing influence with increased copy numbers. With 80 copies or more, we could no longer observe this effect. These results correspond to observations we have made with an RRS detection assay [9] under similar conditions. This means that an absolute LOQ of 40 initial template copies is only representative with DNA from CRMs or materials of equal or better quality and with low concentrations of DNA from pure maize materials. However, with a large background of other DNA, as is typically the case with foodstuffs, higher initial template copy numbers are required (LOQ is estimated to be approximately 100 copies). We conclude that the Bt11 maize transformation event-specific quantitative PCR assay presented here is highly specific, despite the difficulties we had because of the repeated elements at the integration site and the remaining vector sequences in the transgenic DNA. Due to the small size of the PCR products, this method is also suitable for processed food and feed matrices. The absolute detection limits and quantitation limits were estimated to 10 and 100 initial template copies, respectively, indicating that the system is highly sensitive.
Acknowledgements We wish to thank Martha Hernandez Perez, CSIC, Barcelona, Spain for providing 100% Bt11 material. This study was financially supported by the EC FP5 Quality of Life and Management of Living Resources program within the project Reliable, standardised, specific, quantitative detection of genetically modified food (contract no. QLK1199901301, acronym Qpcrgmofood), and by a grant from the Research Council of Norway (136430/140). This is gratefully acknowledged.

Reference List
1. James C (2001) ISAAA Briefs No. 24 2. Regulation EC (1997) Official Journal of the European Communities No. L 43:15 3. Regulation EC (1998) Official Journal of the European Communities No. L 159/4:47 4. Commission Regulation EC (2000) Official Journal of the European Communities No. L6:1314 5. Commission Regulation EC (2000) Official Journal of the European Communities No. L 6:1517 6. Anklam E, Gadani F, Heinze P, Pijnenburg H, Van den Eede G (2002) Eur Food Res Technol 214:326 7. Wolf C, Scherzinger M, Wurz A, Pauli U, Hubner P, Luthy J (2000) Eur Food Res Technol 210:367372 8. Agbios database (2002) http://www.agbios.com/default.asp 9. Berdal KG, Holst-Jensen A (2001) Eur Food Res Technol 213:432438 10. Terry CF, Harris N (2001) Eur Food Res Technol 213:425431 11. Taverniers I, Windels P, Van Bockstaele E, De Loose M (2001) Eur Food Res Technol 213:417424 12. Holck A, Vatilingom M, Didierjean L, Rudi K (2002) Eur Food Res Technol 214:449454 13. Zimmermann A, Luthy J, Pauli U (2000) Lebensm Wiss Technol 33:210216 14. Vatilingom M, Pijnenburg H, Gendre F, Brignon P (1999) J Agric Food Chem 47:52615266 15. Wurz A, Bluth A, Zeltz P, Pfeifer C, Willmund R (1999) Food Contr 10:385389 16. Heid CA, Stevens J, Livak KJ, Williams PM (1996) Genome Res 6:986994 17. Hartl D, Ochman H (1996) Meth Mol Biol 58:293301 18. Van den Eede G, Lipp M, Eyquem F, Anklam E (2000) Report from the EC-JRC, Ispra, Italy, EUR 19677 19. Landt O, Peters M (1998) http://www.lynet.de/TIB-MOLBIOL/ oligo_ag.html 20. Xu J, Pemberton GH, Almira EC, McCarty DR, Koch KE (1995) Plant Physiol 108:12931294 21. Arumuganathan K, Earle ED (1991) Plant Mol Biol Rep 9:211 215 22. Dennis ES, Peacock WJ (1984) J Mol Evol 20:341350